(a) General considerations

The recognition by toxicologists of the extreme toxicity and biological activity of some PCDD/PCDF congeners has generated the requirement for highly sensitive and specific analytical methods. Method development has been directed primarily at the quantitative determination of the seven PCDDs and 10 PCDFs with chlorine at the 2, 3, 7 and 8 positions on the aromatic rings (see Table 2 in this monograph as well as Table 2 in PCDF monograph). The remaining 68 PCDDs and 125 PCDFs can seriously interfere with the determination of the 2,3,7,8-substituted congeners. In addition, the pattern of congeners in an environmental sample can provide clues as to the source of the PCDDs/-PCDFs. The larger number of isomers with four or five chlorines (see Table 1) makes isomer-specific analysis more difficult in the tetra- and penta-substituted series (Rappe, 1984a; Buser, 1991; Clement, 1991).

Ultra-trace PCDD/PCDF analyses can require sample enrichment by a few thousandfold to a million-fold or more before GC-MS determination. Co-extracted, interfering compounds may be present in the sample at much higher levels than the PCDDs and PCDFs, necessitating sophisticated matrix-specific clean-up techniques as well as highly selective separation and detection methods (Clement & Tosine, 1988; Buser, 1991; de Jong & Liem, 1993).

Two different strategies have been applied in PCDD/PCDF analyses. In the first approach, the objective is to recover all PCDDs and PCDFs in a single fraction by a containment-enrichment procedure, and then to analyse this fraction for all congeners by high-resolution GC-MS. In this approach, congener distribution patterns can be obtained that may help to identify sources of the PCDDs/PCDFs. The second approach focuses on separation of isomers during sample preparation and purification (e.g., by including reversed-phase and normal-phase high-performance liquid chromatography (HPLC) steps in the clean-up procedures) and results in multiple fractions that may be analysed for one or more congeners. This approach has often been applied to the determination of a specific congener, such as 2,3/7,8-TCDD (Rappe, 1984a; Buser, 1991).

Analyte standards for calibration are essential in ultra-trace analysis, and by the mid-1980s an adequate set of labelled and unlabelled PCDDs and PCDFs had become commercially available. Although not all of the 210 PCDD and PCDF congeners are available, all of the 2,3,7,8-substituted congeners can be purchased in crystalline form or in solution. These congeners are also available fully ¹³C-labelled, and some are available fully labelled with ³⁷Cl. Analysis of environmental and biological samples requires the use of these standards for method development, to monitor recoveries and for isotope-dilution or other GC-MS analyses. It would be desirable to have available natural-matrix certified reference materials to verify spiked-sample recoveries based on calibration standards, but few are currently available (Rappe, 1984a; Clement & Tosine, 1988; Alvarez, 1991; Clement, 1991; Schimmel et al., 1994; Maier et al., 1995).

PCDD and PCDF determinations are frequently required for incinerator emissions (flue gases, fly ash, bottom ash and aqueous effluents), soils, sediments and sludges, air (vapour and particulates), water, biological samples of all types and chemical products. At present, typical detection limits of methods used for biological samples are in the 1 ng/kg range; lower detection limits are possible for some media, such as ambient air (in the low femtogram [10⁻¹⁵ g]/m³ range) and drinking-water (as low as 0.01 pg/L), and higher detection limits (µg/kg range) are often adequate for chemical products (Buser, 1991; de Jong & Liem, 1993).

The analysis of PCDDs and PCDFs in environmental and biological samples can be considered to proceed in five stages, each of which must be carefully controlled and optimized to ensure reliable data; (1) sampling, (2) extraction, (3) clean-up, (4) separation and (5) quantification. Each of these stages is discussed briefly in the following sections. Reviews and guidelines have been published for various environmental and biological matrices (Buser, 1991; Maier et al., 1994). The following comments are based on these reviews and the other references cited. In the subsequent text and tables, methods are indicated by the system presented in Table 5. In the text, this information is given as ‘(analytical method…)’, using these abbreviations.

Table 5

Abbreviations for descriptions of analytical methods.

(b) Sampling

The sampling protocols for PCDD and PCDF analyses depend on the type of sample, the level of PCDDs/PCDFs and potential interferences in the sample, the detection limit of the method, the requirements of the analysis, and the specific situations encountered. As with most ultra-trace analyses, sample size, homogeneity, storage and handling and the avoidance of contamination are important considerations.

Samples should be protected from light and heat. Although PCDDs and PCDFs are generally quite stable, they are prone to photolysis, especially in solution. The less chlorinated congeners tend to photolyse more rapidly than those with more chlorine substituents (ECETOC, 1992).

Contamination of samples is a serious problem. Potential sources of contamination include sampling equipment and containers, solvents and reagents, adsorbents, glassware, other samples and even laboratory tissue wipes, floor-cleaning solutions and cigarette smoke (Albro, 1979; Patterson et al., 1990).

Samples to characterize incinerator emissions may include grab samples of fly ash (from electrostatic precipitators) and bottom ash (or slag), aqueous effluents from gas-scrubbing equipment and flue gases and particulates. Stack sampling is carried out isokinetically, collecting particulates on filters (e.g., glass fibres) and volatiles by cooling and trapping in impingers or on adsorbent resins. Collection efficiencies are checked by introducing isotope-labelled PCDDs and PCDFs into the sampling train (Ozvacic, 1986; ECETOC, 1992; United States Environmental Protection Agency, 1995). For measurement of flue gases of municipal waste incinerators operating with emissions below 0.1 ng I-TEQ/m³ (European Union Directive; see Section 1.5), certain modifications in the sampling and analytical procedures are necessary, as described by Ball and Düwel (1996) and Bröker (1996).

PCDD and PCDF levels in air (particles and gases) are normally determined by collecting high-volume samples (up to 1000 m³ over a 24-h period) on glass fibre/poly-urethane foam filters (Tondeur et al., 1991). For samples of, say, 1500 m³ collected over several days, the filters should be spiked before sampling with ¹³C-labelled standards to demonstrate possible sampling losses (Tysklind et al., 1993).

PCDDs and PCDFs have very low solubilities in water, with solubility decreasing with increasing number of chlorine substituents. However, they tend to adsorb strongly onto fine particles suspended in water, so water samples must include suspended particulates and cannot be subsampled after collection unless they are first micro-filtered and the water and fine particulates are analysed separately. Small to medium sample volumes (1–5 L) can be extracted directly, but larger volumes needed for ultra-trace analysis are preconcentrated on sorbent resins (e.g., XAD-2) or a polyurethane filter (Ryan, 1991; Luksemburg, 1991).

Soils and sediments are sampled with core samplers at the surface (top 5–10 cm) and sometimes at lower depths. Often, several core samples are pooled for composite analysis (500–1000 g), based on a sampling grid for investigation of soil contamination in an area. Samples are air-dried, sieved (to remove debris), mixed and homogenized before extraction (Kleopfer et al., 1985; Solch et al., 1985; de Jong et al., 1993; Fortunati étal, 1994).

Biological samples should be deep frozen (−20 °C or lower) until analysed to prevent enzymatic or microbiological alterations. Choice of sampling procedures depends on the tissue and species; for example, usually only small samples of human fat are available, whereas larger samples of ecological species (e.g., fish) or foods normally can be obtained. Tissue samples may be ground with anhydrous sodium sulfate or silica gel and homogenized before extraction (Stanley et al., 1985; Norström & Simon, 1991; Patterson et al., 1991; Olsson, 1994).

In discussing problems associated with the variation in biological samples, Bignert et al. (1994) noted that large numbers of samples are generally required to define spatial or temporal trends or differences between various ecological matrices with respect to the concentrations of PCDDs and PCDFs.

(c) Extraction

¹³C-Labelled internal standards are added before extraction and clean-up to determine recoveries and to allow correction for losses during work-up. Homogenized samples are sometimes digested to destroy the sample matrix and free any trapped PCDDs/PCDFs before extraction (e.g., hydrochloric acid treatment of fly ash and some soil/sediment samples). However, alkaline saponification at elevated temperatures (which has sometimes been used with fatty samples) is not recommended, as it can cause decomposition of PCDDs and PCDFs (Albro, 1979; Ryan et al., 1989a). For chlorophenols and chloro-phenoxyacetic acid herbicides, initial separation of the (neutral) PCDDs and PCDFs is often accomplished by partitioning with alkali (Buser, 1991; United States Environmental Protection Agency, 1996a).

Extraction procedures and solvents vary widely, depending on the type of sample and method of clean-up and analysis. Extraction of PCDDs/PCDFs into solvent may be accomplished by simple dissolution, shaking, blending, ultrasonic treatment or Soxhlet extraction. For example, fats and oils may be dissolved directly in dichloromethane, and water samples may be extracted with dichloromethane, toluene or hexane. Particulates filtered from water or resin-sorbed water samples may require Soxhlet extraction with toluene or benzene. Incinerator flue gases (adsorbed on polyurethane foam) and fly or bottom ash are usually Soxhlet-extracted with toluene or benzene. For soils, sediments and sludges, sequential Soxhlet extractions with different and/or mixed solvents (acetone/hexane, toluene, benzene, dichloromethane) are sometimes required.

Many different extraction procedures and solvents have been used with biological samples, such as fish and animal tissues, human tissues and vegetation. Multi-laboratory comparison studies with fish tissues and porcine fat, human blood and adipose tissue and human and cow's milk suggest that acceptable recoveries can be obtained with a variety of extraction methods (Clement & Tosine, 1988; Patterson et al., 1990; Stephens et al., 1992; Schimmel et al., 1994).

Supercritical fluid extraction (SFE) using carbon dioxide is a new technique, which has been used for extraction of PCDDs and PCDFs. Alexandrou et al. (1992) have used it for extraction of solid samples such as fly ash and paper pulp, and also for aqueous matrices such as pulp mill effluents. More recently van Bavel et al. (1996) have demonstrated the use of SFE in the extraction of biological samples, including human adipose tissue.

(d) Clean-up

The objective of the clean-up procedures is to purify and prepare the extract for final separation and quantification. Such procedures remove co-extracted compounds that may interfere in the GC-MS analysis. The extent of clean-up required is determined by the analytical objectives (number of congeners to be quantified), the matrix and the sophistication of the GC-MS system.

Clean-up is normally accomplished by column chromatography through a series of columns, sometimes followed by HPLC. Pretreatment to remove large quantities of co-extractives may be necessary for some samples. Pretreatment may include acid or base washing; elution through multilayer columns containing silica gel impregnated with sulfuric acid, sodium hydroxide and silver nitrate (de Jong et al., 1993); or gel permeation chromatography to remove lipids and other compounds of high molecular weight (Norström & Simon, 1991).

Column chromatography of extracts on alumina removes chlorinated benzenes, poly-chlorinated biphenyls (PCBs) and terphenyls, and higher chlorinated diphenyl ethers in a first fraction (2% dichloromethane in n-hexane); PCDDs and PCDFs are then recovered from the column with 50% dichloromethane in n-hexane. This treatment also removes the polychlorinated 2-phenoxyphenols (predioxins) that can undergo thermal ring closure in the gas Chromatograph to form PCDDs (Buser, 1991).

Since the mid-1980s, a two-column procedure (Smith et al., 1984a) has come into extensive use. Extracts are first chromatographed on activated carbon with dichloromethane to separate planar compounds (including PCDDs and PCDFs, which are retained) from non-planar compounds. The planar compounds are removed from the carbon column by reverse elution with toluene and then chromatographed on alumina to separate the PCDDs and PCDFs from other planar contaminants (e.g., non-ortho-substituted PCBs, polychloronaphthalenes). The method has been adapted and validated for complex biological samples in a semi-automated format (with additional clean-up steps) (Patterson et al., 1990; Turner et al., 1991).

HPLC also has found extensive application in PCDD/PCDF analysis in recent years (Clement & Tosine, 1988). It has been used principally for isomer-specific quantification of 2,3,7,8-TCDD and/or the other toxic PCDD/PCDF congeners. It is also used to supplement other clean-up methods for difficult samples.

Bergqvist et al. (1993) have described the use of a polyethylene semipermeable membrane in a nondestructive method for the reduction of lipid in analyses of PCDDs and PCDFs in environmental biological samples. This allows the analysis of samples with total size of 500 g.

Many other clean-up methods have been reported for environmental and biological matrices. The extent of clean-up required for a given sample depends on a number of factors, as noted above. Even with elaborate clean-up procedures, the final fractions may still contain chlorinated contaminants that need to be considered in the separation and quantification steps.

(e) Separation

The final separation of PCDDs and PCDFs from residual contaminants and into the individual congeners is almost invariably performed by high-resolution GC (HRGC). In the 1960s and 1970s, GC with packed columns (low-resolution GC (LRGC)) allowed separation into groups of isomers (e.g., separation of the TCDDs as a group from the PeCDDs), but the resolution was inadequate for separation of the isomers within a group (e.g., separation of the 22 isomeric TCDDs). In the mid-1970s, glass capillary columns were first used with PCDDs and PCDFs and were found to offer much better separation of isomers and improved sensitivity. The special skills required to prepare these columns, however, initially restricted their use to a limited number of laboratories. Thus, the commercial development of prepared glass capillary columns, especially the introduction of flexible, stable, reproducible fused silica columns around 1980, was one of the most important advances in environmental PCDD/PCDF analysis (Clement & Tosine, 1988; Clement, 1991).

The HRGC columns used for PCDD/PCDF analysis range from 15 to 60 m in length with inner diameters in the range of 0.22–0.35 mm. They are often referred to as wall-coated open-tubular (WCOT) columns, as the inner wall of the column is uniformly coated with a thin film (0.15–0.25 µm) of a silicone (stationary phase) that accomplishes the separations. Many different stationary phases have been used in PCDD/PCDF analyses. In general, non-polar stationary phases (e.g., alkyl/aryl siloxanes) efficiently separate PCDD/PCDF mixtures into groups with the same numbers of chlorines (all TCDDs and TCDFs, all PeCDDs and PeCDFs, etc.), while polar stationary phases (e.g., cyanosilicones) distinguish between the isomers within a group. Frequently, separate analyses using more than one column are required to ensure adequate separation of congeners, for example, a short to medium-length, non-polar (SE-54 or DB-5) column and a longer polar (Silar 10C or SP 2330) column (Clement & Tosine, 1988). A workshop convened by the Community Bureau of Reference of the Commission of the European Union recently recommended the use of a single, nonpolar column (e.g., DB-5) for samples containing only 2,3,7,8-substituted PCDDs and PCDFs, and both a non-polar and a polar column (e.g., SP 2331, CPSEL 88) for samples which also contain PCDDs/-PCDFs with chlorine at other positions (Maier et al., 1994). Since higher terrestrial vertebrates tend to accumulate selectively the 2,3,7,8-substituted congeners, satisfactory analyses of biological samples from farm animals and humans are often achieved on non-polar columns alone (de Jong et al., 1993).

Gas chromatographic retention times are compared with those of authentic standards of the various PCDD and PCDF isomers (¹³C-labelled or unlabelled) for tentative isomer identification. Confirmation of identity and quantification are accomplished by mass spectrometry.

(f) Quantification

Although a few early studies used GC with ECD to quantify PCDDs and PCDFs, modern practice depends almost exclusively on mass spectrometric detection with selected ion monitoring (SIM) of two or more ions from the isotopic group of molecular ions. This allows the use of stable-isotope-labelled internal standards, provides selectivity against coextracted endogenous compounds and many other contaminants and allows congeners with different degrees of chlorination to be quantified separately (Clement & Tosine, 1988). Monitoring of the exact masses of the ions at high mass spectometer resolution provides additional selectivity (Lamparski et al., 1991; Tondeur & Beckert, 1991).

The advantages and disadvantages of various mass spectrometric methods for PCDDs and PCDFs have been extensively reviewed and discussed in recent years (Clement & Tosine, 1988; Buser, 1991; de Jong & Liem, 1993) and are therefore not reviewed here. Either low-resolution mass spectrometry (LRMS) or high-resolution mass spectrometry (HRMS) can provide adequate data, if it has been demonstrated by the analyst that ‘the entire analytical procedure exhibits a sufficient sensitivity and specificity with regard to concentration levels of the PCDD/F and the matrix composition’ (Maier et al., 1994). Most analysts agree that HRMS, although not essential, is preferable. The monitoring of additional fragment ions or use of collision-induced dissociation and tandem MS can provide additional selectivity for confirmation or resolution of specific analytical difficulties but, in practice, are seldom used.

Electron impact ionization is most commonly used for PCDD/PCDF analyses. Negative-ion chemical ionization is very sensitive for the highly chlorinated congeners but is not suitable for lower congeners such as 2,3,7,8-TCDD and may be difficult to use quantitatively because of its marked dependence on operating conditions (Buser, 1991; Maier et al., 1994).

Some differences in the relative abundances of ion fragments from different isomers have been observed (Buser & Rappe, 1978), but do not provide a means of unambiguously differentiating between isomers; this requires chromatographic separation by GC and/or by prior fractionation (Hagenmaier et al., 1986; Lamparski et al., 1991).

Quantification of PCDDs and PCDFs is accomplished by SIM comparisons of the responses for sample components with those of internal standards, usually ¹³C-labelled 2,3,7,8-substituted PCDDs. Calibration standards are used to determine detector response for the various congeners and to confirm its linearity in the concentration range of the samples. Careful attention to quality control and quality assurance procedures is essential for the successful analysis of PCDDs and PCDFs at ultra-trace levels (Mitchum & Donnelly, 1991).

(a) Chemical reactions

(i) Chlorophenoxyacetic acid herbicides (referred to hereafter as phenoxy herbicides)

The phenoxy herbicide 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) (see IARC, 1977b, 1986a, 1987b) was introduced in the late 1940s, and its use was maximal in the 1960s and 1970s. After that time, it was phased out in most European countries and the United States of America. 2,4,5-Trichlorophenol (TCP) (see IARC, 1979a, 1986b, 1987c) is the key intermediate in the production of 2,4,5-T. PCDDs, primarily 2,3,7,8-TCDD, were formed during the production of TCP from 1,2,4,5-tetrachlorobenzene.

Depending on the temperature control and purification efficiency, levels of 2,3,7,8-TCDD in commercial products vary greatly. For example, levels of 2,3,7,8-TCDD in drums of the herbicide Agent Orange (a 1 : 1 mixture of the n-butyl esters of 2,4,5-T and 2,4-dichlorophenoxyacetic acid (2,4-D); see IARC, 1977c) stored in the United States and in the Pacific before 1970 were between 0.02 and 47 mg/kg (analytical method N). Nearly 500 samples were analysed and the mean value was 1.98 mg/kg (Young et al., 1978; Young, 1983). Since Agent Orange was formulated as a 1 : 1 mixture of the butyl esters of 2,4,5-T and 2,4-D, the levels of 2,3,7,8-TCDD in individual preparations of 2,4,5-T manufactured and used in the 1960s could have been as high as 100 mg/kg (Rappe & Buser, 1981).

Rappe et al. (1978a) reported that in other samples of Agent Orange (as well as in 2,4,5-T formulations produced in Europe and the United States in the 1950s and 1960s), 2,3,7,8-TCDD was the dominant PCDD/PCDF contaminant. Only minor amounts of other PCDDs were found, primarily lower chlorinated PCDDs, in samples of Agent Orange (analytical method AC). The concentrations of 2,3,7,8-TCDD have been reported for 2,4,5-T formulations used in Scandinavia (Table 6) and New Zealand (Table 7).

Table 6

Concentrations of 2,3,7,8-TCDD (mg/kg) in Scandinavian 2,4,5-trichloro-phenoxyacetic acid and ester formulations.

Table 7

Average concentrations of 2,3,7,8-TCDD (µg/kg) in 2,4,5-T produced in New Zealand.

As a result of governmental regulations, efforts were made by producers during the 1970s to minimize the formation of 2,3,7,8-TCDD during the production of 2,4,5-T. In the 1980s, all producers claimed that their products contained less than 0.1 mg/kg 2,3,7,8-TCDD (Rappe & Buser, 1981). During production, TCP is separated from 2,3,7,8-TCDD by one or two distillations, which results in 2,3,7,8-TCDD being concentrated in the still-bottom residues. Up to 1 g/kg 2,3,7,8-TCDD in such residues has been reported (Kimbrough et al., 1984; analytical method N).

Sixteen samples of 2,4-D esters and amine salts from Canada were analysed for the presence of PCDDs. Eight of nine esters and four of seven amine salts were found to be contaminated, with esters showing significantly higher levels (210–1752 µg/kg) than salts (20–278 µg/kg) (analytical method AC). The TCDD isomer observed was the 1,3,6,8-isomer, verified using a synthetically prepared standard (Cochrane et al., 1982). On the other hand, Schecter et al. (1993a) found no 2,3,7,8-TCDD at a detection limit of 0.02 µg/kg in one 2,4-D sample of Russian origin. Higher chlorinated PCDDs were found in this sample at < 1 µg/kg (I-TEQ for PCDDs/PCDFs, 0.2 µg/kg). Although 2,3,7,8-TCDD is not expected to be a contaminant in 2,4-D, Hagenmaier (1986) reported that one German 2,4-D formulation contained 6.8 µg/kg 2,3,7,8-TCDD (analytical method ACS).

No data were available to the Working Group on the analysis of samples of the herbicide 4-chloro-2-methylphenoxyacetic acid (MCPA) for 2,3,7,8-TCDD or other PCDDs. [However, it is not expected that PCDDs would be formed during the production of MCPA, based on the starting materials and production route.]

(iii) Chlorophenols

Due to occupational and environmental risks, the use of chlorophenols has now been phased out in most European countries and in a few countries outside Europe. Chlorophenols have been used extensively since the 1950s as insecticides, fungicides, mould inhibitors, antiseptics and disinfectants. In 1978, the annual world production was estimated to be approximately 150 000 tonnes (Rappe et al., 1979a). The most important use of 2,4,6-tri-, 2,3,4,6-tetra- and pentachlorophenol (PCP) and their salts is for wood preservation. PCP is also used as a fungicide for slime control and in the manufacture of paper pulp, and for a variety of other purposes such as in cutting oils and fluids, for tanning leather and in paint, glues and outdoor textiles (Rappe et al., 1978b). Table 8 summarizes a number of analyses of the levels of PCDDs in commercial chlorophenol formulations.

Table 8

Levels of PCDDs in commercial chlorophenols (mg/kg).

Buser and Bosshardt (1976) reported the results of a survey of the PCDD and PCDF content of PCP and its sodium salt from commercial sources in Switzerland (analytical method AC). On the basis of the results, the samples could be divided into two groups: a first series with generally low levels (HxCDD, < 1 mg/kg) and a second series with much higher levels (HxCDD, > 1 mg/kg) of PCDDs. Samples with high PCDD levels also had high levels of PCDFs. The ranges of the combined levels of PCDDs and PCDFs were 2–16 and 1–26 mg/kg, respectively, for the first series of samples and 120–500 and 85–570 mg/kg, respectively, for the second series of samples. The levels of OCDD were as high as 370 mg/kg; this congener dominated the PCDD content of the samples.

Miles et al. (1985) analysed HxCDDs in PCP samples from five different manufacturers in Canada using an isomer-specific analytical method (analytical method ACI). The study included both free PCP and sodium salts. Total HxCDDs in PCP samples ranged from 0.66 to 38.5 mg/kg while, in the sodium salts, levels between 1.55 and 16.3 mg/kg were found. The most abundant HxCDD isomer in free PCP was the 1,2,3,6,7,8-isomer; however, in the sodium salts, the 1,2,3,6,7,9- and 1,2,3,6,8,9-HxCDD pair was the most abundant, probably due to different routes of synthesis.

Hagenmaier (1986) reported that 2,3,7,8-TCDD could be detected in PCP formulations from Germany and the United States at concentrations of 0.21–0.56 µg/kg (analytical method ABS). 1,2,3,7,8-PeCDD was found in these formulations at higher concentrations (0.9–18 µg/kg; analytical method ABS) (Hagenmaier & Brunner, 1987).

(b) Thermal reactions

(i) Incineration of municipal waste

Olie et al. (1977) reported the occurrence of PCDDs and PCDFs in fly ash from three municipal incinerators in the Netherlands. Their results indicated the presence of up to 17 PCDD peaks, but isomer identification and quantification were not possible due to the lack of synthetic standards. Buser et al. (1978a) studied fly ash from a municipal incinerator and an industrial heating facility in Switzerland. In the former, the total level of PCDDs was 0.2 mg/kg. In the industrial incinerator fly ash, the level was 0.6 mg/kg.

In 1986, a working group of experts convened by the World Health Organization Regional Office for Europe (WHO/EURO, 1987) reviewed the available data on emissions of PCDDs from municipal solid-waste incinerators. They concluded that, because of their high thermal stability, PCDDs were destroyed only after adequate residence times (> 2 s) at temperatures above 800 °C Total emissions of PCDDs from tests on municipal solid-waste incinerators were reported to range between a few and several thousand ng/m³ dry gas at 10% carbon dioxide. The WHO working group prepared a table giving a range of estimated isomer-specific emissions for those isomers of major concern with respect to municipal solid-waste incinerators operating under various conditions (Table 9).

Table 9

Estimated range of emissions of PCDDs from municipal solid-waste (MSW) and municipal sewage sludge (MSS) incinerators.

The emissions tabulated in column 1 are those which the working group considered to be achievable in modern, highly controlled and carefully operated plants in use in 1986. The results given in column 1 are not representative of emissions that might be expected from such plants during start-up or during occasional abnormal conditions. Emission levels listed in column 2 were considered by the working group to be indicative of the upper limit of emissions from modern municipal solid-waste incinerators. These plants might experience such emissions during start-up or during occasional abnormal conditions, although some of the data reviewed have shown that the figures in column 2 should not be considered to be an absolute maximum. However, most plants existing in 1986, if carefully operated, will have had PCDD emissions in the range between columns 1 and 2. The highest values for municipal solid-waste incinerators (column 3) were obtained by multiplying the values in column 2 by a factor of 5. Emission levels that were reported to the working group from all tests and under all circumstances were no greater than these values. Generally, these emission levels are associated with irregular or unstable operating conditions, high moisture content of the municipal solid waste, or low combustion or afterburner temperatures. Of special importance is the observation that the emission of 1,2,3,7,8-PeCDD normally exceeds the emission of 2,3,7,8-TCDD by a factor of 3–10.

During the second half of the 1980s and 1990, regulatory agencies in several countries, such as Sweden, Germany and the Netherlands, announced strict regulations for municipal solid-waste incinerators. The European Union value is 0.1 ng I-TEQ per m³ (European Union, 1994) (see column 4, Table 9). This directive has resulted in the introduction of modern air pollution control devices and, together with improved burning conditions, has led to a decrease in PCDD emissions from municipal solid-waste incinerators, which had been considered to be major sources.

(v) Combustion of wood

Schatowitz et al. (1994) measured PCDD/PCDF emissions from small-scale laboratory studies of combustion of wood and household waste (analytical method ABS). The results (in ng I-TEQ/m³) are summarized in Table 10. Data on PCDDs and PCDFs are not separated.

Table 10

PCDD/PCDF emissions from wood and household waste combustion.

(vi) Automobile emissions

Hagenmaier et al. (1990a) reported on the automobile exhaust emissions of PCDDs/-PCDFs and the lower chlorinated congeners and brominated analogues. The results of four representative experiments using leaded gasoline, unleaded gasoline with or without catalytic converters, and diesel fuel (analytical method ABS) are summarized in Table 11.

Table 11

PCDDs from automobile emissions (pg/m³).

(c) Photochemical reactions

It was reported by Nilsson et al. (1974) that chlorinated 2-phenoxyphenols (predioxins) could be converted to PCDDs by a photochemical cyclization reaction. More recently, Vollmuth et al. (1994) reported on the photochemical dimerization of chloro-phenols yielding OCDD. The most important photochemical reaction, however, is photochemical dechlorination resulting in transformation of PCDDs to lower chlorinated compounds. Kieaitwong et al. (1990) reported that 2,3,7,8-TCDD could be identified among the products of photochemical dechlorination of OCDD.

(d) Biochemical reactions

Öberg et al. (1990) showed that chlorinated phenols could be transformed in vitro to PCDDs by peroxidase-catalysed oxidation. It was also demonstrated, using ¹³C-labelled PCP, that HpCDDs and OCDDs could be formed in sewage sludge (Öberg et al., 1993).

(a) Occupational exposures

(i) Exposure during production of TCP, 2,4,5-T and PCP

Germany: In November 1953, an uncontrolled decomposition reaction occurred in an enclosed TCP production facility of BASF at Ludwigshafen, Germany. 2,3,7,8-TCDD was formed during the reaction and contaminated the autoclave section of the building. Employees were exposed to 2,3,7,8-TCDD contaminant during subsequent clean-up and repair activities (Zober et al., 1990; Zober & Päpke, 1993). Analysis of a sample of adipose tissue collected in 1984 from one of the exposed workers showed a 2,3,7,8-TCDD concentration of 100 ng/kg wet weight adipose tissue (Nygren et al., 1986). Zober and Päpke (1993) later analysed samples of autopsy tissues collected from four cases 35–39 years after the same accident. In all samples, the concentrations of most PCDDs were in the normal range, with the exception of 2,3,7,8-TCDD, the concentrations of which were much higher (see Table 12; analytical method ACS). Measurements of the blood concentration of 2,3,7,8-TCDD were performed in 1988 in 138 men whose first contact with PCDDs occurred within one year of the accident (Zober et al., 1994). Current values were extrapolated back to the time of exposure. The geometric mean concentrations ranged from 1118 ng/kg in group with severe chloracne to 148 ng/kg in group with no chloracne. Background blood levels of 2,3,7,8-TCDD in the plant were ≤ 5 ng/kg. Ott et al. (1993) reported on the concentrations of 2,3,7,8-TCDD in blood lipids of these workers sampled between 1988 and 1992, 35–39 years after the accident. The geometric mean value was 15.4 ng/kg. [Back-calculation to the time of the accident, using a 2,3,7,8-TCDD half-life of seven years, gives an estimated blood lipid concentration of 480 ng/kg. If the normal German background concentration in 1990 is first subtracted, the estimate is approximately 400 ng/kg.]

Table 12

Post-mortem concentrations of 2,3,7,8-TCDD (ng/kg fat) in autopsy samples from four individuals involved in the 1953 BASF accident.

Päpke et al. (1992) analysed in 1988–91 the blood of 12 workers exposed during production of TCP. Concentrations of 2,3,7,8-TCDD were approximately 100 times higher than those in a group of background controls, while the concentrations of the higher chlorinated PCDDs were lower in exposed workers than in controls, but not significantly. In blood from 20 workers exposed to PCDDs during production of PCP, the concentrations of OCDD, 1,2,3,4,6,7,8-HpCDD and 2,3,7,8-derived HxCDDs were much higher in the exposed workers than in controls (see Table 13).

Table 13

Concentrations of PCDDs in human blood (ng/kg fat) during production of chlorophenols in Germany.

Flesch-Janys et al. (1996a) studied 48 workers (45 men and 3 women) from a Boehringer-Ingelheim plant in Hamburg manufacturing a range of herbicides. The blood of these workers was sampled twice (for 43 workers) or three times (for 5 workers). The mean time between the end of employment and the first blood sampling was 5.4 years (median, 2.0 years), and the mean time between first and last blood sampling was 5.6 years (median, 6.3 years) (analytical method AB). The results are given in Table 14. In the first sampling, the median concentration of 2,3,7,8-TCDD was 84.1 ng/kg (range, 15.6–300.2 ng/kg). The corresponding values for the second sampling are: median, 48.9 ng/kg (range, 7.7–277.9 ng/kg). More measurements from the Boehringer plant and two other German plants (the BASF plant in Ludwigshafen (this cohort was different from the accident-exposed group) and a Bayer plant in Ingelheim (Becher et al., 1996)) were also performed (summarized in Kogevinas et al., 1997). The 190 subjects from the Boehringer plant showed mean estimated concentrations at the end of employment of 141 ng/kg 2,3,7,8-TCDD (range, 3–2252 ng/kg). In the BASF plant, 20 subjects showed mean 2,3,7,8-TCDD values of 401.7 ng/kg (range, 23–1935 ng/kg), while 19 workers at the Bayer plant had a mean level of 3.2 ng/kg (range, 1.3–6.5 ng/kg).

Table 14

Concentrations of PCDDs (ng/kg fat) in blood of workers at a German herbicide plant (Boehringer-Ingelheim plant).

United States: The largest study in the United States of workers assigned to production of substances contaminated with 2,3,7,8-TCDD was conducted by the United States National Institute for Occupational Safety and Health (Fingerhut et al., 1991a). A more detailed technical report of this study is also available (Fingerhut et al., 1991b). This 12-plant study included the Nitro, WV, plant of Monsanto and the Midland, MI, plant of Dow. The cohort was constructed after a review of personnel records at plants producing chemicals known to be contaminated with 2,3,7,8-TCDD (principally TCP and 2,4,5-T). The cohort included most workers in the United States likely to have been exposed to 2,3,7,8-TCDD in chemical manufacturing; these were 5000 men with work records showing assignment to a production or maintenance job in a process involving 2,3,7,8-TCDD contamination, as well as an additional 172 men without work history records but known to have been exposed based upon inclusion in a prior cross-sectional medical study by Suskind and Hertzberg (1984) at the Nitro, WV, plant. These latter 172 men and an additional 30 men in the United States National Institute for Occupational Safety and Health study lacked sufficient work history information to permit their inclusion in more detailed analyses by duration of exposure. Follow-up was conducted through 1987. Serum 2,3,7,8-TCDD levels, available for 253 cohort members at two plants (different from the Nitro and Midland plants), measured in 1987 averaged a mean of 233 ng/kg, compared with 7 ng/kg for a group of 79 unexposed workers. The mean level was 418 ng/kg for 119 workers exposed for more than one year. All workers had last been exposed 15–37 years earlier. Extrapolation to the date of last employment of these workers, assuming a 7.1-year half-life for 2,3,7,8-TCDD elimination, indicated a mean serum level at that time of 2000 ng/kg (highest level, 32 000 ng/kg). The correlation between extrapolated serum 2,3,7,8-TCDD levels and duration of exposure was 0.72 (Fingerhut et al., 1991b).

Sweden: Littorin et al. (1994) analysed blood from five workers at a factory in Sweden where 2,4-D, MCPA, 2-(4-chloro-2-methylphenoxy)propanoic acid (MCPP) and 2,4,6-trichlorophenol (Saracci et al., 1991) had been formulated, but where production had ceased in 1979, and that of five referents. [2,4,5-T was not produced at this site.] The results are given in Table 15. The concentrations of all PCDDs were high, especially that of 1,2,3,7,8-PeCDD, which was unexpectedly found to be higher than that of 2,3,7,8-TCDD. However, a leachate sample from the factory also contained a higher level of 1,2,3,7,8-PeCDD.

Table 15

Blood plasma concentrations of PCDDs in five phenoxy herbicide/chlorophenol production workers and five referents in Sweden.

Austria: Neuberger et al. (1991) reported mean blood levels of 389 ng/kg 2,3,7,8-TCDD (range, 98–659 ng/kg) in TCP and 2,4-D production workers with chloracne exposed about 17 years earlier.

Netherlands: Hooiveld et al. (1996a) took blood samples in 1993 from 48 persons occupationally exposed to phenoxy herbicides, chlorophenols (and the contaminant 2,3,7,8-TCDD) in 1955–85 in a chemical factory where an accident occurred in 1963. The geometric mean levels of 2,3,7,8-TCDD were 22.9 ng/kg in men who worked in the main production unit and 87.2 in subjects exposed as a result of the accident.

(ii) Exposure during handling and spraying of 2,4,5-T

Nygren et al. (1986) reported analyses of adipose tissue from 31 persons in Sweden, 13 of whom were exposed to phenoxy herbicides and 18 of whom were unexposed controls. Of the exposed persons, 11 had neoplastic diseases; in controls, six persons had cancer, giving a total of 17 cancer patients and 14 persons without cancer. No difference was found between exposed and unexposed or between cancer patients and persons without cancer (see Table 16; analytical method ACS).

Table 16

Concentrations of PCDDs in human adipose tissue (ng/kg wet weight) from persons exposed to phenoxy herbicides in Sweden.

In another study in Sweden including 20 non-Hodgkin lymphoma patients and 17 controls, Hardell et al. (1996) reported no significant differences in the concentrations of PCDDs and PCDFs in adipose tissue (see also Section 2.2.2(b)).

Professional pesticide applicators involved in ground-level spraying of 2,4,5-T in New Zealand are claimed to be the group most heavily exposed to agricultural use of 2,4,5-T in the world (Smith et al., 1992a). Many of the applicators sprayed for more than six months per year and some were spraying for more than 20 years. Measurements of 2,3,7,8-TCDD in blood serum of nine of these workers (Table 17) gave arithmetic means of 53.3 ng/kg for the exposed group and 5.6 ng/kg for a group of matched controls (ratio, 9.5). For all other congeners, the ratio between the levels in the exposed group and the controls was below 1.5.

Table 17

Levels of PCDDs in serum of nine 2,4,5-T applicators and nine matched control subjects in New Zealand.

Military personnel in Viet Nam: The Viet Nam Experience Study is described by its authors as a ‘multidimensional assessment of the health of Viet Nam veterans (Centers for Disease Control Vietnam Experience Study, 1988a,b,c,d). This study was designed to examine effects among men who served in the United States armed forces in Viet Nam, where Agent Orange (see Section 1.2.1(a)(i)) was widely sprayed as a herbicide and defoliant (Operation Ranch Hand). The study population was a random sample of men who enlisted in the United States Army from 1965 through 1971, whose military occupational status was other than ‘duty soldier’, who enlisted for a single term with a minimum of 16 weeks' active duty and who were discharged at pay grades of E-1 to E-5. The controls were selected from among veterans enlisting during the same period but whose duty station was the United States, Germany or Korea. Participation involved completion of a telephone survey of current and past health status by 7924 veterans who served in Viet Nam and 7364 veterans who served outside Viet Nam. A random subsample of 2940 Viet Nam and 1972 non-Viet Nam veterans participated in the health evaluation component.

Investigators in the Centers for Disease Control Veterans Health Studies (1988; see Table 18) studied serum levels of 2,3,7,8-TCDD among 646 Viet Nam veterans who were ground combat troops during 1967–68 in areas heavily sprayed with Agent Orange. Also studied were 97 veterans who did not serve in Viet Nam. The mean serum level was 4 ng/kg for each group, as measured in 1987. There was no correlation between serum levels of 2,3,7,8-TCDD and service in areas of Viet Nam ranked by level of presumed intensity of spraying based on military records, or between serum levels and self-reported levels of exposure to Agent Orange. Only two Viet Nam veterans had levels above 20 ng/kg (25 and 45 ng/kg). This study did not include Viet Nam veterans who served in the Air Force spraying of Agent Orange (the Operation Ranch Hand veterans), nor members of the Army Chemical Corps which also sprayed Agent Orange. There are data indicating that these specific groups (especially Ranch Hand veterans) had significant levels of exposure to PCDDs above background.

Table 18

Serum levels of 2,3,7,8-TCDD in United States veterans who served in Viet Nam and elsewhere in 1987.

One of the largest epidemiological studies of United States military personnel stationed in Viet Nam is being conducted by the United States Air Force. The study population consists of Air Force personnel who served in Operation Ranch Hand units in Viet Nam from 1962 to 1971 and who were employed in the dissemination of Agent Orange through aerial spraying. Comparisons included Air Force personnel who flew or maintained C-130 aircraft in south-east Asia during the same period. The study design includes a series of cross-sectional medical studies conducted at five-year intervals beginning with the baseline study in 1982 (1045 exposed, 1224 unexposed). Two follow-up evaluations were conducted in 1985 (1016 exposed, 1293 unexposed) and 1987 (995 exposed, 1299 unexposed). Each cross-sectional study included comprehensive physical and psychological evaluations. In the 1982 baseline and 1985 and 1987 follow-up studies, exposure was based on the comparison of the Ranch Hand group versus the comparison group. An additional analysis estimated the approximate exposure (low, medium or high) for the Ranch Hand group by using historical military data and herbicide procurement and usage records (Roegner et al., 1991).

The Air Force Ranch Hand veterans (Wolfe et al., 1990) (see Table 19) tested in 1987 and exposed during 1962–71 (n = 888) had a median of 12.4 ng/kg serum 2,3,7,8-TCDD compared with a median of 4.2 ng/kg for comparison subjects (n = 856) who were Air Force personnel not exposed to herbicides. The highest levels were for non-flying enlisted personnel (n = 407), with a median of 23.6 ng/kg. Air Force Ranch Hand veterans sprayed approximately 88–90% (Thomas & Kang, 1990) of the Agent Orange applied in Viet Nam, with application from airplanes. The remaining 10–12% was sprayed by the Army Chemical Corps which applied the herbicide around the perimeter of bases either from the ground or from helicopters. Kahn et al. (1988) studied 2,3,7,8-TCDD serum levels in a sample of 10 heavily exposed Viet Nam veterans and matched controls. The two highest levels were found in Army Chemical Corps personnel (approximately 75 and 180 ng/kg. It should be noted that both Ranch Hand veterans and Army Chemical Corps veterans sprayed other herbicides (Agent White or Agent Blue) besides Agent Orange, although few other herbicides were contaminated with PCDDs (Thomas & Kang, 1990). These data suggest that, although some Ranch Hand subjects were exposed to very high levels of 2,3,7,8-TCDD, most had lower exposures.

Table 19

Serum 2,3,7,8-TCDD levels (ng/kg) in Ranch Hand and control veterans.

Nygren et al. (1988) analysed adipose tissue and blood samples collected 15–20 years after military service from 27 men, 10 of whom were heavily exposed during their service in Viet Nam, 10 of whom had marginal exposure during service and served as Viet Nam controls and seven veterans who did not serve in Viet Nam and were used as ‘era’ controls. The results are summarized in Table 20. The only difference in the mean values was for 2,3,7,8-TCDD, for which the arithmetic mean for the heavily exposed group is approximately 8–10 times higher than that for controls. The highest level found was 213 ng/kg. In another report from the same study group, Gochfeld et al. (1989) found a good correlation (r = +0.89) between the concentrations of 2,3,7,8-TCDD in blood and adipose tissue.

Table 20

Concentrations of PCDDs in blood plasma of exposed and unexposed groups of United States Army veterans (ng/kg fat).

(b) Population exposure due to industrial accident

In an accident on 10 July 1976 at the ICMESA plant at Seveso, Italy, a runaway reaction led to a blow-out of a TCP production reactor. The chemical cloud that emerged from the reactor entrained nearly 2900 kg of organic matter, including at least 600 kg of sodium trichlorophenate and an amount of 2,3,7,8-TCDD which is still being evaluated [probably of the order of several kilograms]. The visible part of the cloud rose up to about 50 m, subsequently subsided and fell back to the earth, but was wind-driven over a wide area. Within less than 2 h after the accident, chemicals settled on the ground as far as 6 km south of the factory, or were dispersed by wind streams. Plants, domestic animals and birds were seriously affected, many dying within a few days of the accident [probably due to chemicals other than 2,3,7,8-TCDD]. About 10 days after the accident, 2,3,7,8-TCDD was found in samples of various types collected near the factory (Bertazzi & di Domenico, 1994). As a first step, information on the location of toxic and pathological events was used to draw an approximate diagram of the contaminated area. This was confirmed by chemical monitoring of 2,3,7,8-TCDD in the soil carried out under emergency conditions. Within five weeks of the accident, the area was subdivided into Zones A, B and R in descending order of 2,3,7,8-TCDD contamination and toxicological risk levels. The borderline between Zones A and B was set at average 2,3,7,8-TCDD concentrations in the soil of < 50 µg/m²; the boundaries between Zones A and B and Zone R were fixed where the average contamination was ≤ 5 µg/m². Zone R included the remaining territory where detectable levels of 2,3,7,8-TCDD (formally ≥ 0.75 µg/m²) were found. Soil levels are further described in Section 1.3.2.

From Zone A, over 730 inhabitants were evacuated. Zones B and R were subjected to area-specific hygiene regulations including prohibition of farming, consumption of local agricultural products and keeping poultry and other animals. All persons residing in these zones at the time of the accident, as well as all the newborn and new residents in the subsequent 10-year period, were considered to have been exposed. Three exposure categories were formed, corresponding to the zone of residence of the subjects at the time of the accident or later entry into the area. As a reference, the population of 11 municipalities surrounding the contaminated area was adopted (Bertazzi et al., 1993).

Mocarelli et al. (1990) reported analyses of 19 non-randomly selected samples of blood collected in 1976 from persons living in Zone A (Table 21) (analytical method ACS). Of these persons, 10 had chloracne of type 3 or 4 (see Section 4.2.1(a)) and nine had no chloracne. The persons with chloracne in general had higher concentrations than persons without chloracne, but high concentrations were also found in the healthy group. The 2,3,7,8-TCDD level of 56 000 ng/kg fat is the highest ever reported.

Table 21

Concentrations of 2,3,7,8-TCDD in the blood of individuals from Zone A, Seveso.

Recently, blood measurements of 2,3,7,8-TCDD were performed in persons randomly sampled from the most contaminated areas (Zones A and B), and in members of the population adopted as reference in the epidemiological investigations (Landi et al., 1996). Preliminary results were reported. In the six persons from Zone A, the median value was 71.5 ng/kg (extrapolation back to 1976 assuming a half-life of 7.1 years gave a value of 388.7 ng/kg); in the 52 inhabitants in Zone B, the corresponding values were 12.5 ng/kg and 77.6 ng/kg, while in the 52 subjects from the reference population, the median value was 5.5 ng/kg. Women consistently had higher levels of 2,3,7,8-TCDD than men in all three areas.

(c) Summary table

Table 22 summarizes data from the major cohorts of populations highly exposed to 2,3,7,8-TCDD occupationally or as a result of accidents. Estimates of mean or median exposures, or in one case a range of individual exposures, are presented as reported in the cited publications. When tissue concentrations were measured, concentrations at the time of exposure estimated in the cited publications, or back-calculated by the Working Group based on a seven-year half-life, are also given.

Table 22

Estimated exposures to 2,3,7,8-TCDD (ng/kg blood lipid) of selected study populations.

(a) Air

PCDD/PCDF levels in air have been reported for many countries during the last 10 years. Due to the enormous amount of information, only a selection of representative data are summarized below. For detailed information, see Appendix 1 (Table 1).

(b) Water (see Appendix 1, Table 2)

Because of the complex cycling and partitioning of PCDDs/PCDFs in the aquatic environment and the difficulty of analysis, adequate sampling and analysis methods have become available only over the past decade.

PCDDs and PCDFs are highly lipophilic and thus have an affinity for particulate organic carbon (Webster et al., 1986). Knowledge of their partitioning between the sedimentary, colloidal, dissolved, organic and even vapour phases remains limited, largely because of the difficulty of measuring these compounds in solution.

PCDDs partition between particulate and dissolved fractions. Material that passes through a filter of 0.45 µm or 0.2 µm pore size is commonly defined as dissolved. The major problem with this definition is that colloidal particles (suspended solid particles of < 0.2 µm diameter) and other macromolecules pass through such filters and are included in the ‘dissolved’ phase. However, colloids and ‘dissolved’ organics may function as ‘small particles’ in terms of the mechanisms of PCDD/PCDF uptake by biological systems (Broman et al., 1991, 1992).

(c) Soil (see Appendix 1, Table 3)

(ii) Austria

Around a densely populated industrial urban area (Linz), soils from grassland and forest areas were analysed for PCDDs/PCDFs and PCBs (Weiss et al., 1993, 1994). The concentrations of PCDDs/PCDFs were higher in soils near chemical and/or steel plants. The highest PCDD/PCDF concentration in grassland soil (I-TEQ, 14.4 ng/kg) was measured near a hospital refuse incineration plant.

Riss et al. (1990) compared PCDD/PCDF levels in soil, grass, cow's milk (see Section 1.3.2(d)(i)), human blood and spruce needles in an area of PCDD/PCDF contamination through emissions from a metal reclamation plant (Brixlegg, Tyrol). Soil concentrations were highest near the plant and in the main wind direction, and decreased in all directions with increasing distance. The highest concentration found corresponds to 420 ng German TEQ/kg. The level of 40 ng/kg TEQ (which has been proposed as an upper limit for agricultural use) was exceeded in an area reaching 600 m from the plant in the main wind directions. Milk samples from farms with meadows in the contaminated area contained TEQ levels of PCDDs/PCDFs ranging from 13.5 to 37.0 ng/kg fat, which were considerably higher than that of 3.6 ng/kg in the control sample from Kossen/Tyrol. About 50% of the TEQ of the contaminated milk was accounted for by 2,3,7,8-TCDD (normally about 10%). Because of the elevated PCDD/PCDF levels in cow's milk, blood plasma from two farmers who drank milk from their own cows was analysed for PCDDs/PCDFs. One had significantly elevated blood levels (see Table 25).

Table 25

Concentrations of PCDDs in human samples from the general population.

Determinations of PCDDs and PCDFs in soil samples from rural, urban and industrial sites in Salzburg state (Boos et al., 1992) gave I-TEQ values generally in the low ng/kg range. Six samples contained PCDDs/PCDFs. Levels exceeding 5 ng German TEQ/kg, above which limit the cultivation of certain vegetables is restricted, but none exceeded 40 ng TEQ/kg, which would exclude the cultivation of any plants. Including in the calculation undetected congeners at 50% of the detection limit would lead to seven samples exceeding the lower limit. The lowest level found was 0.1 ng TEQ/kg.

(d) Food

It is estimated that intake from food consumption accounts for well over 90% of the body-burden of PCDDs and PCDFs in the general human population (Gilman et al., 1991; Travis & Hattemer-Frey, 1991).

Over the last decade a number of studies of foods have appeared, but several are of only limited usefulness because of low sensitivity. The results from Canadian studies of meat collected in 1980 were published by Ryan et al. (1985a), but the lower chlorinated congeners were not detected. Firestone et al. (1986) published the results of analyses for the higher chlorinated PCDDs in various foods collected in the United States by the Food and Drug Administration in a five-year period beginning in 1979; lower chlorinated PCDDs including 2,3,7,8-TCDD were not determined. Stanley and Bauer (1989) analysed composites of selected foods collected from San Francisco and Los Angeles, CA, including salt-water fish, freshwater fish, beef chicken, pork, cow's milk and eggs. The analytical methods and quality control were rigorous but the limits of detection were somewhat higher than those obtained in some other recent studies and few measurable residues were found. Birmingham et al. (1989) analysed samples of beef, pork, chicken, eggs, milk, apples, peaches, potatoes, tomatoes and wheat from Ontario (Canada). Some samples were found to contain detectable concentrations of the higher chlorinated congeners but no Cl₄ or Cl₅ PCDDs or PCDFs were detected. The data from this study are available only on a whole sample rather than a lipid-adjusted basis.

Studies by Beck et al. (1987, 1989a) of foods available in Berlin, Germany, were based on analysis of 12 random purchases of foods (chicken, eggs, butter, pork, red fish, cod, herring, vegetable oil, cauliflower, lettuce, cherries and apples) and of eight samples of cow's milk. The analysis used high-resolution mass spectrometry and achieved very high sensitivity, so that most of the congeners with a 2,3,7,8-substitution pattern were detected in most of the samples. Fürst et al. (1990) concentrated on fatty foods and foods of animal origin in a study involving over 100 individual samples from the North-Rhine Westphalia area of Germany, but with slightly lower sensitivity. Fürst et al. (1992a) subsequently reported further analysis of milk and dairy products which made use of more sensitive high-resolution mass spectrometry, but detailed congener-specific results were not published.

A comprehensive assessment of Dutch foodstuffs was reported by Liem et al. (1991a,b). The sampling scheme was designed after considering the individual consumption data of approximately 6000 individuals from 2200 families over a two-day period. Animal fat and liver from six different types of animal were collected from slaughter houses, and cereal products, cow's milk, dairy products, meat products, nuts, eggs, fish and game from retail sources. In each of these groups, duplicate collection of samples was carried out in each of four Dutch regions and proportional pooling was performed to provide a pair of duplicate samples for each group and region, and one national composite sample from the four regional pools. Analysis was by high-resolution mass spectrometry and achieved high sensitivity. Liem at aL (1991c) also reported the analysis of 200 samples of cow's milk obtained from the vicinity of municipal solid-waste incinerators and other potential PCDD/PCDF sources in the Netherlands.

Results are available from a study of composite samples from the UK Total Diet Study (TDS) (Ministry of Agriculture, Fisheries and Food, 1995; Wright & Startin, 1995; Wearne et al., 1996; see Table 23). In this scheme, retail samples of 115 specified food items are purchased at two-week intervals from different locations in the United Kingdom, prepared as for consumption, then combined into one of 20 food groups in proportions representing the relative importance of the retail foods in the average British diet. Concentrations of PCDDs and PCDFs were determined in archived TDS samples of fatty foods and bread collected in 1982 and 1992. For each food group, the material analysed was a composite of samples from all 24 locations included in the TDS that year. Fruit, vegetables and other non-fatty foods were not analysed. The determinations used high-resolution mass spectrometry and met appropriate acceptance criteria (Ambidge et al., 1990).

Table 23

Concentrations of PCDDs/PCDFs (ng I-TEQ/kg whole food) in Total Diet Study samples collected in 1982 and 1992 in the United Kingdom.

Several studies of food items from Viet Nam, Russian Federation and the United States have been reported by Schecter et al. (1989a, 1990a, 1994a), who worked with various analytical laboratories of good repute. Most of the data relate to individual samples. In the study of food in the United States, 18 samples of dairy products, meat and fish from a supermarket in upstate New York were analysed (Schecter et al., 1994a), A more recent study (Schecter et al., 1996a) of 100 food items combined by type has not been fully reported, but led to estimates of intake close to those of the earlier study. A more comprehensive investigation of concentrations in beef in the United States has been reported by Ferrario et al. (1996).

Several studies have been conducted of foods in the Japanese diet, but some doubt must be attached to the results. Ono et al. (1987) reported the analysis of vegetables, cooking oils, cereals, fish, pork, beef, poultry and eggs collected in Matsuyame in 1986. In contrast to other studies, the dominant congeners in the fish and animal samples were generally not those with 2,3,7,8-substitution, suggesting that the results do not reflect biologically incurred residues but some other form of contamination. OCDD (which is normally the congener of highest concentration) was not consistently found and it is possible that the alkaline saponification procedure used for some samples resulted in dechlorination of OCDD and distortion of the congener profile. Other studies by Takizawa and Muto (1987) and by Ogaki et al. (1987) also employed alkaline saponification and the reports do not include congener-specific results.

Relevant data are summarized in Appendix 1 (Tables 9–18). The data included have been selected to meet a number of criteria, including relevance to dietary intake rather than environmental monitoring, and adequate detection limits and appropriate analytical methodology. Some exceptions have, however, been made and some data that are available only as summed I-TEQs have been included where they seem to be of value. Additionally, the results included were either available on a lipid basis or could be so converted using either reported fat contents or reasonable assumptions. Some additional results omitted from the tables are discussed below. All concentrations quoted are lipid-adjusted unless otherwise stated.

Most of the summed I-TEQ concentrations included in the tables have been recalculated using I-TEF values and assuming that congeners that were not detected were present at the full value of the detection limit. Obviously, however, this was not possible where detection limits were not reported (indicated as ‘ND’) or where the original author's calculations were used.

(ii) Foods from contaminated areas

Vegetables

The outer surface of root crops can obviously become contaminated by soil contact; low concentrations of PCDDs and PCDFs have been measured in root crops such as carrots and potatoes grown in contaminated soils but were largely absent if the vegetables were peeled (Facchetti et al., 1986; Hülster & Marschener, 1993).

Surface-borne PCDD and PCDF contamination of foliage and fruits may include contributions from direct deposition of airborne particulates and from absorption of vapour-phase contaminants from the air, including those which are attributable to evaporation from the soil (Reischl et al., 1989). In apples and pears grown on highly contaminated soil, Müller et al. (1993) found total concentrations of PCDDs and PCDFs (including non-2,3,7,8-substituted congeners) in the range of 1–4 ng/kg. Peeling removed most of the PCDD and PCDF contamination, although washing was not effective.

In studies of field-grown vegetables, measurable amounts of PCDDs/PCDFs have generally been found only in areas where a specific contamination problem was known to exist. Concentrations of 2,3,7,8-TCDD of 100 ng/kg were detected in the peel of fruits grown on soils contaminated in the Seveso incident, but not in the flesh (Wipf et al., 1982). Investigations in the vicinity of a wire reclamation incinerator showed contamination of leaf vegetables at concentrations of 5–10 ng I-TEQ/kg and rather less in fruits (Prinz et al., 1990).

Cows' milk

A number of studies have demonstrated the localized influence of incinerators and other sources on PCDD/PCDF concentrations in cow's milk (Table 24). Thus, Rappe et al. (1987b) found between about 8 and 12 ng I-TEQ/kg in the milk of individual cows grazing near municipal solid-waste incinerators and a chlorinated chemical production site in Switzerland.

Table 24

Concentrations of PCDDs reported in cow's milk from contaminated areas.

More recently, in surveillance around incinerators in the Netherlands, levels up to 13.5 ng I-TEQ/kg in cow's milk were found (Liem et al., 1991c). The highest PCDD concentrations were usually found within about 2 km of the source.

Schmid and Schlatter (1992) analysed milk from sites near waste incineration, metal recycling and other industrial facilities in Switzerland, where the PCDD/PCDF levels were reported to be two- to four-fold greater than in commercial samples.

In the United Kingdom, Startin et al. (1990) found between 3 and 6.2 ng I-TEQ/kg for PCDDs/PCDFs in milk from farms near an incinerator and close to a densely populated and industrialized area, compared with the mean for rural background areas of 1.1 ng I-TEQ/kg. In 1990, concentrations of 40 and 42 ng I-TEQ/kg were found in milk from two farms near Bolsover in Derbyshire, while milk from 30 other farms in the region contained an average of 4.3 ng I-TEQ/kg (range, 1.8–12.5) (Harrison et al., 1996). In contrast to the results discussed above where the proportions of different congeners were fairly similar to those in background samples, milk from Bolsover showed a distinctive pattern dominated by 2,3,7,8-TCDD, 1,2,3,7,8-PeCDD and 1,2,3,4,7,8-/1,2,3,6/7,8-HxCDD.

Riss et al. (1990) investigated contamination caused by a metal reclamation plant at Brixlegg in Austria and found PCDD/PCDF levels in two samples of cow's milk giving 55 and 69 ng I-TEQ/kg.

(a) Austria

For comparison with an exposed group of 2,4,5-T production workers, Neuberger et al. (1991) reported 2,3,7,8-TCDD blood concentrations in men with low or no occupational exposure to 2,3,7,8-TCDD. The range and median for the group were < 5–23 and 13 ng/kg, respectively.

Samples of milk from cows grazing in the vicinity of a metal reclamation plant showed significantly higher PCDD/PCDF levels than control samples. In the blood of two farmers in the same area, increased levels of certain isomers were found. The highest value (one sample) was found for 1,2,3,7,8-PeCDD at 780 ng/kg (Riss et al., 1990).

(b) Canada

Ryan et al. (1985b) reported that adipose tissue from 23 older subjects (> 60 years old) who had died in Ontario hospitals in 1979–81 contained an average of 11 ng/kg 2,3,7,8-TCDD.

Ryan et al. (1985c) analysed 46 and 10 adipose tissue samples originating from people who had died accidentally in 1976 and 1972, respectively, in Canada for various reasons, as well as 10 others from deceased hospital patients. Total PCDD levels were about an order of magnitude higher than total PCDFs. PCDD levels increased with increasing chlorination from 2,3,7,8-TCDD (average, 5–13 ng/kg) to OCDD (average, 528–1355 ng/kg). Only the penta-, hexa- and hepta-CDF congeners were detected at these levels (10–60 ng/kg); TCDF and OCDF were not detected.

Human adipose tissue samples obtained during autopsies in five Canadian municipalities within the Great Lakes basin (Cornwall, London, St Catherine's, Welland, Windsor) were analysed for PCDDs and PCDFs (LeBel et al., 1990). The mean congener levels for male and female donors in each municipality were similar to those previously reported. The ages of the 40 men ranged from 29 to 83 years (mean, 63 years) and that of the 36 women from 12 to 88 years (mean, 69 years). No significant differences in congener levels between male and female donors or between municipalities were detected. However, in four cases, the values for women were 15–45% higher than the corresponding values for men. Additionally, a positive correlation with age was observed for the levels of several congeners as well as for the I-TEQ.

In connection with the forestry industry, there has been concern that by-products of pulp and paper production and chlorophenol fungicides used in sawmills may result in residents being environmentally or occupationally exposed to PCDDs/PCDFs. To examine this possibility, PCDD and PCDF levels were measured in the adipose tissue of 41 British Columbians selected to match the age and sex distribution of the exposed population. The group consisted of 18 men and 23 women. The mean age of the subjects was 45 years (range, 18–77 years) and the mean weight was 96 kg (range, 50–193 kg). The mean of the 41 samples was 29.1 ng I-TEQ/kg (range, 8.4–56.4). The highest 2,3,7,8-TCDD level measured was 9.2 ng/kg (lipid-based) (Teschke et al., 1992).

Ryan (1986) examined the relationship between the PCDD/PCDF level and age (between 14 and 76 years) in 46 Canadian individuals.

(c) China

Human adipose tissue from seven patients (four men, three women; mean age, 54 years) undergoing general surgery in Shanghai was analysed by Ryan et al. (1987). Compared with data from other countries, the values were low. At a detection limit of 2.0 ng/kg (based on wet weight), no 2,3,7,8-TCDD was found.

In connection with a study to examine the exposure of agricultural workers to PCP (for snail control) and PCDDs/PCDFs, Schecter (1994) reported data on two pooled age-matched control groups. For the two groups from the general population, very low blood concentrations of PCDDs/PCDFs were found at 4.8 ng I-TEQ/kg (age, 15–19 years) and 5.7 ng/kg (over 40 years), respectively. These are consistent with values found in mothers' milk (see Section 1.4(b)).

(d) Finland

In a study of pulp and paper mill workers in Finland (Rosenberg et al., 1995), a comparison group with no known exposure was analysed. This control group consisted of 14 persons with a mean age of 41 years (range, 28–60 years). The mean total I-TEQ level in plasma was 49 ng/kg (range, 20–99 ng/kg) (see Section 1.3.1(a)(iv)).

(e) France

The levels of PCDDs and PCDFs in adipose tissue from eight persons living in Paris were reported by Huteau et al. (1990a). Most of the 2,3,7,8-substituted isomers were found, in some cases at unexpectedly high values (2,3,7,8-TCDF, 1,2,3,7,8-PeCDF and 2,3,4,7,8-PeCDF). Surprisingly, non-2,3,7,8-substituted isomers were also reported at relatively high values (TCDFs, TCDDs, HpCDFs). [Sample contamination cannot be excluded.]

(f) Germany

The first German data on background levels of PCDDs in human adipose tissue were reported by Beck et al. (1989b). Concentrations in 20 unexposed individuals (mean age, 50 years) were used for comparison in a study of 45 occupationally exposed employees at a chemical plant in Hamburg. The values for the comparison group ranged between 18 and 122 ng I-TEQ/kg (mean, 56 ng/kg).

No correlation was seen between adipose tissue or liver concentrations and age or sex in 28 subjects aged between 26 and 80 years (Thoma et al., 1989, 1990). Large differences in the concentrations of PCDDs and PCDFs between adipose and liver tissues were demonstrated for most of the isomers (see Table 26). Thoma et al. (1990) also reported concentrations of PCDDs and PCDFs in adipose tissue from eight infants (age 2–12 months). The levels were lower than in adults for nearly all isomers.

Table 26

Concentrations of PCDD isomers in adipose and liver tissues of German adults and adipose tissue of infants.

Background data on PCDDs and PCDFs in human blood from Germany published by Päpke et al. (1989b) have been updated since 1991 by various authors (see Table 27). The results suggest a decrease in PCDD/PCDF blood levels in Germany over the past decade.

Table 27

Time trend in background data on concentrations of PCDDs/PCDFs in human blood.

Age-related increases in blood levels of most of the PCDD congeners and the I-TEQ have been reported (Sagunski et al., 1993; Schrey et al., 1992; Päpke, 1996).

The Kieselrotstudie (Wittsiepe et al., 1993) was designed to assess the degree of exposure to PCDDs and PCDFs in 56 persons living in the vicinity of a former copper smelter located in Marsberg (see monograph on PCDFs in this volume; Section 1.3.2(c)). The copper smelter was in operation until 1945. In 1991, high levels of PCDDs/PCDFs were found in materials from the slag dumps (10–100 µg I-TEQ/kg). The median I-TEQ values of the Marsberg group (43.2 ng/kg blood lipid) and the reference group from Steinfurt (43.0 ng/kg blood lipid) were similar, whereas the mean of the Marsberg group (52.7 ng/kg) was higher than that of the control group (44.4 ng/kg).

Near a metal reclamation plant in Baden-Württemberg, Rastatt, PCDD/PCDF contamination of soil, dust from homes, indoor air and vegetables was investigated in 1987. Blood samples from 22 volunteers living in the vicinity of the plant were analysed for PCDDs and PCDFs. Levels of certain Pe-, Hx- and HpCDF isomers were increased, in a similar pattern to the contamination throughout the area. The increase in PCDD/PCDF levels was traced to occupational exposure in the case of workers and to food intake in the other cases. For children (four samples), soil and/or dust ingestion may be a pathway of special importance (Wuthe et al., 1990).

In Rheinfelden in southern Germany, soil concentrations of PCDDs/PCDFs of up to approximately 1000 ng German TEQ/kg were found. The source of contamination of the soils was identified as residues from a PCP production process. Locally produced food, such as eggs, chicken and vegetables, with high PCDD/PCDF levels were reported to be an important source of elevated human levels. Good agreement between human blood and milk concentrations was demonstrated in one woman (Wuthe et al., 1993).

PCDD/PCDF levels were determined in the venous blood of 21 allotment gardeners from Duisburg (Ewers et al., 1994). Soil analysis showed elevated levels of PCDDs/-PCDFs in garden soil (range, 16.4–77.6 ng I-TEQ/kg). Vegetable plants also had elevated levels (up to 65.6 ng I-TEQ/kg). The mean I-TEQ of 44.3 ng/kg (range, 29.2– 81.1 ng/kg) in blood fat of the gardeners was within the range of a control group.

In seven breast cancer patients without known occupational exposure to PCDDs/-PCDFs, the levels of PCDD/PCDF congeners in mammary carcinoma tissue were not elevated above the concentrations in tumour-free tissue (Körner et al., 1994).

PCDD/PCDF levels in various organs from infants were measured by Beck et al. (1990, 1994). Adipose tissue, liver, spleen, thymus and brain from eight infants (age, 3–23 months) who had mostly died from sudden infant death syndrome were analysed. Five of the infants had been breast-fed exclusively for between 21 and 91 days. The lipid-based PCDD/PCDF levels in liver, thymus and spleen were higher than those in adipose tissue. Very often the concentrations were below the detection limits. The lowest levels were found in brain.

Elimination of PCDDs/PCDFs through the faeces was studied in a three-month-old breast-fed infant (Jödicke et al., 1992). Little excretion of C1₄−C1₆ congeners was found. The data suggest that more than 90% had been absorbed by the infant. However, 1,2,3,4,6,7,8-HpCDD as well as OCDD were found to be highly concentrated in stool fat (by a factor of 5–12 over the concentration in milk fat).

(g) Guam

Schecter et al. (1992) reported a mean PCDD/PCDF level of 28 ng I-TEQ/kg in blood samples from 10 residents of the Pacific island of Guam. This was suggested to be due to consumption of food mainly imported from the United States and Japan. [A local source cannot be excluded.]

(h) Japan

In adipose tissue from six individuals 20–70 years old in Japan collected in 1984 (Ryan, 1986), the mean 2,3,7,8-TCDD concentration was 6.6 ng/kg and that of OCDD 1360 ng/kg.

Thirteen samples of human adipose tissue from cancer patients were analysed for Cl₄− Cl₈ PCDDs and PCDFs (Ono et al., 1986). These compounds were identified in all of the analysed samples. 2,3,7,8-TCDD concentrations ranged from 6 to 18 ng/kg.

In a large study, more than 500 human milk and 17 adipose tissue samples were analysed for PCDDs (Ogaki et al., 1987). Adipose tissues from inhabitants of a large city and its suburbs had higher levels of PCDDs that those of rural towns.

Muto et al. (1991) investigated the tissue distribution of 2,3,7,8-substituted PCDDs in humans who died of cancer. The values for OCDD were the highest in each organ and tissue. All data were wet weight-based, and no lipid concentrations for the tissues were reported.

(i) The Netherlands

Intake and faecal excretion of PCDDs/PCDFs in breast-fed infants at a range of ages were studied (Pluim et al., 1993a). In three totally breast-fed infants, the amount of PCDDs/PCDFs consumed via breast milk and excreted in the stools was measured at the ages of four, eight and 12 weeks. Intake was high, especially at the age of four weeks (mean, 257 pg/kg bw). A strong decline was observed in the second month, mainly due to a reduction in PCDD/PCDF concentrations in whole breast milk as a result of reduced fat content. With the exception of OCDD, faecal excretion of the congeners was below 5% of their intake, indicating a bioavailability of more than 95% from breast milk. No influence of age on faecal excretion of PCDDs/PCDFs during the first three months of life was observed.

(j) New Zealand

In a study to determine whether blood serum levels of 2,3,7,8-TCDD in a group of professional 2,4,5-T applicators in New Zealand were greater than those of a matched control group not involved in 2,4,5-T spraying, the mean level of 2,3,7,8-TCDD in the control group was 5.6 ng/kg ±1.1 (standard error) (Smith et al., 1992a) (see Section 1.3.1 (a)(ii) and Table 17).

(k) Norway

PCDDs/PCDFs have been analysed in human blood in relation to consumption of crabs from a contaminated fjord area in Norway (Johansen et al., 1996). The analyses were performed in three different groups: reference, moderate crab intake and high crab intake (age between 40 and 54 years). A significant increase in blood levels of many PCDD and PCDF congeners was found in crab consumers; the differences were greatest for several of the PCDFs that are characteristic of the contamination of marine biota in the fjord caused by a magnesium-producing plant. Almost all subjects in the high-intake group exceeded the tolerable weekly intake of 35 pg 2,3,7,8-TCDD/kg bw per week proposed by the Nordic Expert Group (see Section 1.5).

(l) Poland

Changes over time in PCDD/PCDF levels in the adipose tissue of a person with long-term exposure to PCP were studied (Gorski et al., 1984). The analyses, performed by GC-ECD, yielded half-lives of 1,2,3,6,7,8-HxCDD, 3.5 years; 1,2,3,4,6,7,8-HpCDD, 3.2 years; OCDD, 5.7 years; 1,2,3,4,6,7,8-HpCDF, < 1.7 years; OCDF, 1.8 years.

(m) Russian Federation

Schecter et al. (1992) found PCDD/PCDF concentrations in pooled blood samples from Baikalsk (eastern Siberia) and St Petersburg to be 18 and 17 ng I-TEQ/kg, respectively. Levels of OCDD were very low (57 and 89 ng/kg, respectively).

(n) Spain

Jiménez et al. (1995) measured the blood levels of PCDDs and PCDFs from 11 individuals (age, 19–55 years) living in Madrid. The mean total level of PCDDs was 5153 ng/kg, of which 1.5 ng/kg was 2,3,7,8-TCDD.

(o) Sweden

After Rappe et al. (1983) found elevated PCDD/PCDF levels in the blood of occupationally exposed workers (chlorophenol workers, textile workers), samples from the general population in Sweden were analysed. Levels of all higher chlorinated (C1₄−C1₈) PCDDs and PCDFs in adipose tissue were measured (Rappe, 1984b). The samples originated from patients undergoing cancer or gall bladder surgery. It was concluded that the low ng/kg levels for tetra- to octa-CDDs/CDFs represented typical contamination of the general population.

In a continuation of this study (Nygren et al., 1986; Rappe et al., 1986a), PCDD/-PCDF patterns in adipose tissue from a group of 17 cancer patients and a group of 14 controls were compared. No difference was seen in the pattern of congeners between the two groups.

In order to study the influence of diet on the body burden of PCDDs/PCDFs, blood from three different Swedish groups was analysed (Svensson et al., 1991; Rappe, 1992): Group 1, with no fish consumption (persons suffering from an allergy); Group 2, with normal fish consumption (around 50 g/day); and Group 3, with high fish consumption (> 100 g/day). Group 3 had a body burden, calculated as TEQ, approximately three times higher than Group 2. Group 1, however, had only slightly lower blood levels of PCDDs/PCDFs than Group 2. The dominant congener among the tetra- and penta-chlori-nated congeners was 2,3,4,7,8-PeCDF. The difference between the three groups was also highest for this particular congener, which is also the major congener in fish from the Baltic Sea (see Section 1.3.2(d)). The mean PCDD/PCDF values for Groups 1, 2 and 3 were 17.5, 25.8 and 63.5 ng I-TEQ/kg, respectively.

This study was extended to cover the main fishing areas in Sweden (Svensson et al., 1995a) and assessed dietary habits and exposure to selenium, persistent organochlorine compounds, including PCDDs/PCDFs, methylmercury and methylamines among Swedish fishermen. The interview data showed that 250 fishermen ate almost twice as much fish as the 250 referents from the general population. Fishermen from the Baltic Coast ate more fatty fish than fishermen from the Atlantic Coast, and they also had higher blood levels of persistent organochlorine compounds such as PCDDs/PCDFs than both the Atlantic Coast fishermen and the referents.

Measurements of PCDD/PCDF concentrations in adipose tissue from seven patients with malignant lymphoproliferative diseases and 12 surgical patients without malignant disease showed significantly higher concentrations of 1,2,3,7,8-PeCDD, 1,2,3,6,7,8-HxCDD and 2,3,4,7,8-PeCDF in the cases with lymphoproliferative diseases than in the patients in the other group (Hardell et al., 1995). Nordic TEQ values also were significantly higher in the first group (mean, 64.7 ng/kg; range, 19.9–187) than in the second group (mean, 29.7 ng/kg; range, 12.9–53.4).

(p) Switzerland

Wacker et al. (1990) analysed liver and adipose tissue samples of 21 Swiss inhabitants (age, 15–85 years; mean, 47 years) for PCDDs/PCDFs. All values were reported on a wet weight basis. Table 28 shows the concentration ratios between adipose tissue and liver. No correlation between age or weight and tissue levels was detected.

Table 28

Concentration ratios (wet weight-based) of PCDDs in adipose and liver tissue of Swiss inhabitants.

(q) United Kingdom

PCDD/PCDF background levels were measured in pooled human adipose tissue samples from five areas in Wales (Duarte-Davidson et al. 1993). With the exception of OCDF, which was found at unexpectedly high values in all pooled samples (36–62 ng/kg), the concentrations were similar to those in other industrialized countries. 2,3,7,8-TCDD and 2,3,7,8-TCDF were not detected at detection limits of 10 ng/kg.

(r) United States

Six samples from both biopsy and autopsy fat taken in 1983–84 from New York State residents were analysed (Ryan et al., 1985c). PCDDs and PCDFs were found in all samples with total (C1₄−C1₈) PCDD levels about an order of magnitude higher than total (C1₅−C1₇) PCDFs. PCDD levels increased with increasing chlorination from tetra- (mean, 6.4 ng/kg) to octa-CDD (mean, 585 ng/kg). Only the penta-, hexa- and hepta-PCDF congeners were detected, at levels that were of the same order of magnitude. TCDF and OCDF were absent.

The tissue distribution of PCDDs and PCDFs was studied in three autopsy subjects from the general population of New York State (Ryan et al., 1986). These were the first reports to show that several 2,3,7,8-chlorine-substituted PCDDs/PCDFs are present not only in adipose tissues from the general population, but also in all other tissues assayed. The ratios of the PCDD/PCDF congeners to each other were similar in each tissue, with overall levels on a wet weight basis decreasing in the order fat, adrenal, bone marrow, liver, muscle, spleen, kidney and lung. If the levels are expressed on a lipid basis rather than on a wet weight basis, liver had the highest value and the variation between tissues showed only a two- to four-fold difference.

Schecter et al. (1986a) reported PCDD/PCDF levels in adipose tissue of eight control samples from Binghamton, NY. All values were based on wet weight. The 2,3,7,8-TCDD levels ranged between 1.4 and 17.7 ng/kg. HpCDDs and OCDD were found at mean values of 164 and 707 ng/kg, respectively, with maximum values of 691 and 1931 ng/kg, respectively.

Patterson et al. (1986a) reported 2,3,7,8-TCDD levels in 61 adipose tissue samples from 35 autopsy cases from Georgia and Utah. The geometric mean for these samples on a wet-weight basis for 2,3,7,8-TCDD was 7.1 ng/kg. The geometric mean of values for 2,3,7,8-TCDD in 31 of these samples on a lipid basis was 9.6 ng/kg. On the basis of the wet-weight adipose concentrations, the authors concluded that concentrations of 2,3,7,8-TCDD increased with age in both sexes. There was no significant difference in the concentration of 2,3,7,8-TCDD between blacks and whites, and women had a slightly higher (2 ng/kg) mean concentration than men.

Analysis for C1₄−C1₈ PCDDs/PCDFs was performed for 46 adipose tissue samples prepared from the United States Environmental Protection Agency National Human Adipose Tissue Survey (NHATS) as composites from over 900 specimens to represent the nine United States census divisions and three age groups (0–14, 15–44 and ≥ 45 years) (Stanley et al., 1986). The results demonstrate that PCDDs/PCDFs are prevalent in the general United States population and that differences exist with age. Only means and ranges of all data were reported.

Stanley et al. (1990) also found a decrease in 2,3,7,8-TCDD with time in a subset of the NHATS adipose tissue samples (age group, 15–44 years) collected between 1971 and 1987 (see Table 29).

Table 29

Differences in 2,3,7,8-TCDD levels in adipose tissues by year of sample collection in the United States.

In an interim report, Patterson et al. (1986b) described high concentrations of 2,3,7,8-TCDD in adipose tissue from persons (mean age, 52.6 years) exposed recreationally, residentially and occupationally in a highly contaminated area in Missouri, compared with 57 controls. The geometric and arithmetic means were 6.4 and 7.4 ng/kg (wet weight-based), respectively, with a range of 1.4–20.2 ng/kg.

The Missouri study control group (Patterson et al., 1986b) was expanded in 1989 by 128 persons with no known exposure (Andrews et al., 1989): adipose 2,3,7,8-TCDD levels ranged from non-detectable to 20.2 ng/kg, with 95% of the levels at 16.6 ng/kg or below.

Patterson et al. (1994) analysed adipose samples originating from Atlanta, GA, in 1984–86. They found mean 2,3,7,8-TCDD concentrations for men and women of 7.4 ng/kg and 11.6 ng/kg, respectively. Further adipose tissue samples were analysed for all isomers.

Piacitelli et al. (1992) compared serum 2,3,7,8-TCDD levels in 280 chemical plant workers with the levels in 99 unexposed referents. The mean serum level for the control group was 7 ng/kg (lipid-based).

A comparison of PCDD/PCDF levels in whole blood, plasma and adipose tissue was performed by Schecter et al. (1990b) and Päpke et al. (1992). There were few differences in PCDD/PCDF levels between blood plasma and adipose and also between whole blood and adipose tissue when reported on lipid basis. The difference was most striking for OCDD between plasma and adipose tissue, with a ratio of 2 : 1. Total PCDDs/PCDFs appeared higher in plasma than in adipose tissue, if reported by actual measurement, because OCDD is usually the most abundant congener. Comparing whole blood with adipose tissue, the values were more similar. When TEQs were used, however, the values were almost identical in the two series. Wendling et al. (1990a) found 2,3,7,8-TCDD levels in samples of faeces from laboratory workers of between 0.61 and 0.74 ng/kg, dry weight-based. The fat content of the faeces was not measured. With some assumptions, a faecal fat 2,3,7,8-TCDD concentration of 3.7 ng/kg was estimated.

PCDDs and PCDFs in adipose tissue of United States Viet Nam veterans and controls were determined by Kang et al. (1991). The samples were collected in 1978. The geometric mean (± SD) 2,3,7,8-TCDD levels in adipose tissue for Viet Nam veterans, non-Viet Nam veterans and civilian controls were 11.7 (± 1.7), 10.9 (± 1.7) and 12.4 (± 1.9) ng/kg on a lipid weight basis, respectively. The mean levels for these groups were not significantly different from each other.

The effect of fasting on blood PCDD/PCDF levels was tested in 13 Viet Nam veterans (Hansson et al., 1989). Although 2,3,7,8-TCDD levels increased slightly, for no congener including 2,3,7,8-TCDD was the change statistically significant.

In connection with analysis of blood samples from various geographical locations for PCDDs/PCDFs, Schecter et al. (1992) reported a total I-TEQ of 41 ng/kg (lipid-based) in a pooled blood sample (n = 100) from the United States. [The Working Group noted that the dates of collection of these samples were not given.]

In a study by Schecter et al. (1994b; 1996b), levels of PCDDs/PCDFs in placenta, blood and fetal tissue were measured. The highest semen levels, reported on wet weight, were found for OCDD, OCDF and HpCDF. The highest I-TEQ values (lipid-based) were found in blood, followed by placenta. The fetal tissue contained approximately one third of the I-TEQ of the adult values.

In a further study of partitioning of PCDDs/PCDFs in human maternal tissues, including blood, milk, adipose tissue and placenta, Schecter et al. (1996c) collected samples from five American women (mean age, 21.6 years; range, 21–34 years) residing in upstate New York and undergoing caesarean section deliveries between September 1995 and January 1996. Blood, placenta and fat were collected at the time of delivery. The milk and second blood were collected about four to eight weeks later. The lowest concentrations were found in the cord blood, at about one half of the maternal adipose and blood levels. A reduction in PCDD/PCDF levels was observed in the ‘second’ blood samples after a breast-feeding period of between four and eight weeks.

PCDD/PCDF levels in two pools of whole blood and serum (n = 100) collected in 1996 were compared with older blood data; a decrease was not clearly shown. The mean age of the blood donors was not specified (Schecter et al., 1996d).

(s) Viet Nam

2,3,7,8-TCDD was not detected in seven adipose tissue samples from north Viet Nam (no exposure to Agent Orange) but was found in 10/13 samples from south Viet Nam (mean level, 34 (range, 9–103) ng/kg). Most of the other chlorinated PCDDs and PCDFs were found in samples from south Viet Nam (Schecter et al., 1986b). A further 27 individual and 10 pooled human adipose tissue specimens, collected from persons in south and north Viet Nam, respectively, were analysed for 2,3,7,8-TCDD and 2,3,7,8-TCDF (Schecter et al., 1989c). The mean values were 19 ng/kg 2,3,7,8-TCDD and 7 ng/kg 2,3,7,8-TCDF in the samples from persons in the south; no 2,3,7,8-TCDD or 2,3,7,8-TCDF was detected in samples from persons in the north. Differences in 2,3,7,8-TCDD body burden continued to be substantial between the populations of south and north Viet Nam.

In 1984–85, adipose tissues were taken from nine patients at a hospital in south Viet Nam (Nguyen et al., 1989) and analysed for PCDDs/PCDFs. Eight patients had detectable levels of 2,3,7,8-TCDD ranging from 4 to 103 ng/kg, with a mean of 23 ng/kg.

Effects of geographical conditions and other variables on the distribution of 2,3,7,8-TCDD levels in adipose tissues from Viet Nam have been studied (Hoang et al., 1989). No 2,3,7,8-TCDD was detected in 10/11 samples from the north (the positive sample contained 5 ng/kg). In the south, 2,3,7,8-TCDD was not detected in 4/44 samples and the mean level was 20.5 ng/kg in the four positive samples.

PCDD/PCDF levels in 27 adipose tissue samples from south Viet Nam were reported by Huteau et al. (1990b). Besides the usual 2,3,7,8-substituted isomers, they found non-2,3,7,8-substituted isomers in many samples. [Sample contamination cannot be excluded.]

In connection with analysis of blood samples from various geographical locations for PCDDs/PCDFs, Schecter et al. (1992) reported results for pooled samples from north Viet Nam (two analyses with a total of 82 persons) and south Viet Nam (nine analyses totalling 383 persons). The 2,3,7,8-TCDD level and I-TEQ value for samples from north and south Viet Nam were 2.2 and 14.6 ng/kg and 15 and 36 ng I-TEQ/kg, respectively.

PCDD levels in persons living at various localities in Viet Nam were summarized (Le et al., 1995) using data from pooled blood samples. Each sample was obtained from 30–100 adults over 40 years old, who had been living for more than five years in a given area. As from 1992, some samples were taken from younger age groups between 18–40 years. The results of 43 analyses (containing blood of 2722 individuals) indicated that most of the samples from the south had higher levels of 2,3,7,8-TCDD as well as I-TEQ than the ones from the north.

In a study of the persistence of elevated PCDD levels in human tissues from Viet Nam, Schecter et al. (1995) performed 160 analyses on tissues from 3243 persons. 2,3,7,8-TCDD levels of up to 103 ng/kg were found in adipose tissue collected in the 1980s. Pooled blood collected from south Viet Nam in 1991–92 also showed 2,3,7,8-TCDD levels up to 33 ng/kg, whereas tissues from north Viet Nam had 2,3,7,8-TCDD levels at or below 2.9 ng/kg.

(a) Two United States plants

(b) Comprehensive United States study

The largest study of United States production workers exposed to PCDDs was conducted by the United States National Institute for Occupational Safety and Health (NIOSH) and published by Fingerhut et al. (1991a). A more detailed technical report of this study is also available (Fingerhut et al., 1991b). This 12-plant study included the Nitro, WV, plant of Monsanto and the Midland, MI, plant of Dow discussed above, which represented 9% and 40% of the cohort, respectively. In order to be included into the cohort, all workers had to have had presumed exposure to 2,3,7,8-TCDD. The cohort was constructed by NIOSH after a review of personnel records at 12 United States plants producing chemicals known to be contaminated with 2,3,7,8-TCDD (principally TCP and 2,4,5-T). The cohort included most workers in the United States likely to have been exposed to 2,3,7,8-TCDD in chemical manufacturing, comprising 5000 men with work records showing assignment to a production or maintenance job in a process involving 2,3,7,8-TCDD contamination, as well as an additional 172 men without work history records but known to have been exposed at the Nitro, WV, plant based upon inclusion in a prior cross-sectional medical study by Suskind & Hertzberg (1984). These latter 172 men and an additional 30 men in the NIOSH study lacked sufficient work history information for their inclusion in more detailed analyses by duration of exposure. Follow-up was conducted through 1987 and United States referent rates were used. Serum levels of 2,3,7,8-TCDD in 253 cohort members at two plants measured in 1987 averaged 233 ng/kg lipid, compared with 7 ng/kg lipid in a group of 79 unexposed workers. Levels increased to 418 ng/kg for 119 workers exposed for more than one year (Fingerhut et al., 1991a). Extrapolation to the date when these workers were employed, assuming a half life of 7.1 years, indicated a mean serum level at that time of 2000 ng/kg lipid (highest level, 32 000 ng/kg). Workers had last been exposed 15–37 years earlier. The correlation between serum 2,3,7,8-TCDD level and duration of exposure was 0.72 (Fingerhut et al., 1991b). There were 1052 deaths in this cohort, with 116 748 person-years. The SMR for all causes was 1.0 (95% CI, 0.9–1.1). Mortality from all cancers (265 deaths) was slightly but significantly elevated (SMR, 1.2; 95% CI, 1.0–1.3). Soft-tissue sarcoma mortality was elevated based on four deaths (SMR, 3.4; 95% CI, 0.9–8.7). Two of the deaths from soft-tissue sarcoma, upon further review of medical records, were found to be misclassified (false positives), but three deaths from other causes, upon review, were found to be soft-tissue sarcomas (false negatives). Other causes of death of interest in this study were not remarkable. There were 10 deaths from non-Hodgkin lymphoma (SMR, 1.4; 95% CI, 0.7–2.5); stomach cancer mortality was not elevated (10 deaths observed; SMR, 1.0; 95% CI, 0.5–1.9); lung cancer showed a slight increase (89 observed; SMR, 1.1; 95% CI, 0.9–1.4). Sub-cohort analyses focused on those workers with more than one year's duration of 2,3,7,8-TCDD exposure and at least 20 years' potential latency (1520 workers, 29% of the cohort; mean duration of employment, 19 years; mean duration of exposure to 2,3,7,8-TCDD, 7 years). In this sub-cohort, mortality from all cancers combined was significantly elevated (114 observed; SMR, 1.5; 95% CI, L2–1.8). A wide variety of cancer sites showed some excess, but only soft-tissue sarcoma was significantly elevated (3 observed; SMR, 9.2; 95% CI, 1.9–27). The SMR for lung cancer was 1.4 (95% CI, 1.0–1.9) based on 40 deaths. Conversely, the SMR for non-Hodgkin lymphoma was not elevated in this sub-cohort with presumed higher exposure (2 deaths observed; SMR, 0.9; 95% CI, 0.1–3.4). Internal analyses comparing longer duration of exposure to 2,3,7,8-TCDD to a short duration (< 1 year) referent category found nonsignificant positive trends for all cancers and lung cancer (p = 0.3 and p = 0.2, respectively). Rate ratios also increased with latency (all cancer SMRs, 0.7, 1.1 and 1.3; lung cancer SMRs, 0.8, 1.0 and 1.2 with latency of < 10 years, 10–20 years and ≥20 years, respectively). The authors considered that smoking was not likely to be responsible for the excess of lung cancer for several reasons including that (a) there was no increase in non-malignant respiratory disease, which is strongly related to smoking; (b) an indirect adjustment for smoking based on known smoking habits for a sample of the cohort did not account for the observed increase; and (c) there was no increase in lung cancer among the group with 20 years' potential latency but short duration (< 1 year) of exposure to 2,3,7,8-TCDD (17 observed; SMR, 1.0). [The Working Group noted that these are indirect ways of controlling for confounding.]

(c) German accident cohort

In the 1953 accident at the BASF TCP production unit at Ludwigshafen, Germany, the total number of employees identified as being involved directly or in the subsequent clean-up, repair or maintenance activities was 247 (243 men, 4 women). Analyses of adipose tissue and blood from groups of these workers are described in Section 1.3.1(a)(i). Part of the cohort was first studied by Thiess et al. (1982). It was completed by Zober et al. (1990) and further studied by Zober et al. (1994) and Ott and Zober (1996). Out of the 247, 69 (1 woman) were identified by a company physician early after the accident as the most directly involved (cohort CI); 84 (2 women) were identified by August 1983 as probably involved (cohort C2); and 94 (1 woman) were recognized by the end of 1987 as possibly exposed during participation in demolition and toxicology investigations, or because they were members of the safety department and plant management at the time of the accident (cohort C3). Their mortality was investigated through 1987, with no loss to follow-up (Zober et al., 1990). Death certificates were obtained for 67 deceased subjects, and 11 additional deaths were ascertained through other means (information from physicians, autopsy reports). Expected deaths were calculated from national rates. All members of cohort C1 were affected by chloracne and their mortality from malignancies was moderately higher than expectation (9 deaths; SMR, 1.3; 95% CI, 0,7–2.3). Other nonsignificant increases were seen for stomach cancer (3 deaths; SMR, 3.0; 95% CI, 0.8–7.7), colon and rectum (2 deaths; SMR, 2.5; 95% CI, 0.4–7.8) and lung cancer (4 deaths; SMR, 2.0; 95% CI, 0.7–4.6). In cohort C2, 17 workers were affected by chloracne and four had other exposure-related skin lesions. Their mortality from all cancers combined after 20 years since first exposure was significantly increased (8 deaths; SMR, 2.4; 95% CI, 1.2–4.3), as was mortality from other and unspecified cancer sites (5 deaths; SMR, 3.2; 95% CI, 1.3–6.8) (whole cohort). A nonsignificant increase was noted for colorectal cancer (2 deaths; SMR, 2.7; 95% CI, 0.5–8.5) (whole cohort). Suicides were elevated in both cohorts C1 and C2. In cohort C3, where 28 persons had experienced mild forms of chloracne, no significant increase for deaths from any cause was seen, although one single leukaemia death represented a greater than five-fold elevated risk. Among all 127 workers affected by chloracne or other skin lesions (erythema), the mortality from all cancers combined was significantly elevated after 20 years since exposure (14 deaths; SMR, 2.0; 95% CI, 1.2–3.2). Nonsignificant increases were seen for cancer of the stomach (3 deaths; SMR, 1.8; 95% CI; 0.5–4.7), colon and rectum (3 deaths; SMR, 2.2; 95% CI, 0.6–5.8) and lung (6 deaths; SMR, 1.8; 95% CI, 0.8–3.6). In 1986, blood concentrations of 2,3,7,8-TCDD were measured in 28 workers. The median values were 24.5 ng/kg in cohort C1 (10 subjects), 9.5 ng/kg in C2 (7 subjects) and 8.4 ng/kg in C3 (11 subjects). The median value for workers with chloracne and other skin lesion was 15 ng/kg versus 5.8 ng/kg in those without skin manifestations. The small size of the study population precludes definitive conclusions.

The cancer incidence and the updated mortality, through 1992, have been reported (Ott & Zober, 1996). The study population comprised the 243 exposed men. One death from a motor vehicle accident had occurred among the four exposed women, and they were not further considered in the analysis. Incident cases were ascertained from available medical and necropsy data and from survey results (the area was not covered by a cancer registry). Death certificates were obtained for all but one of the deceased workers. Expected numbers of incident cases were calculated from the cancer statistics of the state of Saarland and expected deaths were calculated on the basis of national rates. At that time, serum measurements of 2,3,7,8-TCDD were available for 138 cohort members (Ott et al., 1993) (see Section 1.3.1(a)(i)). Model-based estimates of the cumulative dose of 2,3,7,8-TCDD (µg/kg bw) were calculated for each study subject. Chloracne status was also adopted as an indicator of past exposure to 2,3,7,8-TCDD; 113 workers had a diagnosis of chloracne, and 55 were classified as severe. Standardized mortality and incidence ratios and their 95% CIs were estimated using standard techniques. Internal dose-response analyses were performed with the Cox's proportional hazard model, taking into account — among other variables — cigarette smoking and other potentially confounding exposures. Results of the models were reported in terms of conditional risk ratios (CRRs). The conditional risk represents the risk per unit increase (1 µg/kg bw) in 2,3,7,8-TCDD dose. In the mortality analysis, an increased cancer mortality with increasing dose was apparent: 2,3,7,8-TCDD level < 0.1 (µg/kg bw, eight deaths (SMR, 0.8; 95% CI, 0.4–1.6); 0.1–0.99 µg/kg bw, eight deaths (SMR, 1.2; 95% CI, 0.5–2.3); 1.0–1.99 µg/kg bw, eight deaths (SMR, 1.4; 95% CI, 0.6–2.7); ≥2 µg/kg bw, seven deaths (SMR, 2.0; 95% CI, 0.8–4.0) [p for trend = 0.07]. Digestive system and respiratory tract cancers also tended to increase with increasing dose. In the 2,3,7,8-TCDD category > 1 µg/kg bw, after 20 or more years since initial exposure, all-cancer mortality (13 deaths; SMR, 2.0; 95% CI, 1.1–3.4) and respiratory cancer mortality (6 deaths; SMR, 3.1; 95% CI, 1.1–6.7) were significantly increased. In the sub-cohort with chloracne, 20 or more years after first exposure, 18 cancer deaths were observed (SMR, 1.9; 95% CI, 1.1–3.0). An increase was also seen for digestive tract cancer (6 deaths; SMR, 1.8; 95% CI, 0.7–4.0) and respiratory tract cancer (7 deaths; SMR, 2.4; 95% CI, 1.0–5.0). In the incidence analysis, a slight, nonsignificant increase in the risk for all cancers combined with increasing 2,3,7,8-TCDD dose was obtained. Respiratory tract cancer incidence was elevated in the high-dose group (8 cases; SMR, 2.0; 95% CI, 0.9–3.9). With the Cox's regression model, after controlling for other relevant variables, a significant association between 2,3,7,8-TCDD dose and increase in digestive system cancer was obtained, as both cause of death (CRR, 1.5; 95% CI, 1.1–1.9) and incident cases (CRR, 1.4; 95% CI, 1.1–1.7). Joint analysis of 2,3,7,8-TCDD dose and cigarette smoking showed a positive relationship between 2,3,7,8-TCDD dose and all cancers combined among current cigarette smokers (2,3,7,8-TCDD dose, SMR (95% CI) in current smokers: < 0.1, 1.2 (0.3–3.1); 0.1–0.99, 1.4 (0.3–4.2); 1.0–1.99, 3.0 (1.1–6.5); ≥ 2 µg/kg bw, 4.0 (1.5–8.6)) but not among nonsmokers.

(d) Other German plants

Several reports have considered workers from a chemical plant operated by Boehringer-Ingelheim, in Hamburg, Germany, that produced herbicides heavily contaminated with 2,3,7,8-TCDD and other PCDDs/PCDFs (Manz et al., 1991; Nagel et al., 1994; Flesch-Janys et al., 1995; Becher et al., 1996). In the latter study, workers from three other German plants were also considered.

Manz et al. (1991) reported the mortality experience of workers at the Hamburg plant since late 1951. An outbreak of chloracne in 1954 led to a halt in the production of TCP and 2,4,5-T. In 1957, production was resumed using a new process that reduced 2,3,7,8-TCDD formation. The vital status of all permanent employees at the plant for at least three months between 1 January 1952 and 31 December 1984 (1184 men, 399 women) was investigated through 1989. Causes of death were derived from medical records or, when not available, from death certificates. 2,3,7,8-TCDD concentrations were measured in workers from various production departments, mainly after the plant had closed in 1984. Workers were classified in exposure groups according to their work history: high (496 subjects), medium (901) and low (186). Mean concentrations of 2,3,7,8-TCDD in adipose tissue in 1985 from 48 volunteer surviving members of the cohort were 296 ng/kg (SD, 479; median, 137) in 37 workers in the high-exposure group and 83 ng/kg (SD, 73; median, 60) in 11 workers in the combined medium-/low-exposure groups. For comparison, national death rates and rates from a cohort of some 3500 workers at the Hamburg gas supply company were used. In the latter case, the follow-up period was restricted to 1985. SMRs and 95% CIs were estimated using standard techniques assuming a Poisson distribution. Follow-up was 97% successful. In men, mortality due to all cancers combined was increased in comparison with both the general population (93 deaths; SMR, 1.2; 95% CI, 1.0–1.5) and the gas workers (75 deaths; SMR, 1.4; 95% CI, 1.1–1.8). The increase was highest among those who entered the plant before 1954 (who had the highest exposure to 2,3,7,8-TCDD based on the history of the plant and subsequent serum measurement) and remained employed for 20 or more years. In men, in comparison with gas workers, deaths from all cancers combined were: high exposure, 29 deaths (SMR, 1.8; 95% CI, 1.2–2.6); medium exposure, 39 deaths (SMR, 1.2; 95% CI, 0.9–1.6); and low exposure, seven deaths (SMR, 1.5; 95% CI, 0.6–3.0) [p for trend = 0.24]. Among specific sites, significant increases were seen for lung cancer (26 deaths; SMR, 1.7; 95% CI, 1.1–2.4) and malignancies of the haematopoietic system (9 deaths; SMR, 2.7; 95% CI, 1.2–5.0). Smoking prevalence was similar in chemical (73%) and gas (76%) workers. Suicides were also significantly increased in comparison with both reference populations. Among women, mortality from all cancers combined was close to expectation; an increase was seen for breast cancer (9 deaths; SMR, 2.2; 95% CI, 1.0–4.1).

Mortality of the female workers in the cohort was further investigated by Nagel et al. (1994). Compared with the results of Manz et al. (1991), one additional breast cancer death was identified (10 cases; SMR, 2.4; 95% CI, 1.1–1.4). Cox's regression analysis yielded increasing breast cancer risk with duration of employment. 2,3,7,8-TCDD levels in blood or adipose tissue were measured for 26 women (22 for blood and 4 for adipose tissue), with an arithmetic mean of 109.7 ng/kg (median, 23 ng/kg; range, 7–1439 ng/kg).

The mortality experience of 1189 male workers employed for at least three months between 1952 and 1984 in the same plant in Hamburg was investigated through 1992 (Flesch-Janys et al., 1995). A quantitative exposure index was constructed. Fourteen relevant production departments were identified and levels of PCDDs/PCDFs were determined in the buildings, products and wastes. A detailed work history for each worker ever employed in these departments was constructed. In a sample of 190 workers, concentrations of each PCDD/PCDF congener in adipose tissue or whole blood were determined, and their levels at the end of workers' exposure were calculated assuming a one-compartment first-order kinetic model; the contribution of working time in each production department to the end-of-exposure levels in these 190 workers was estimated. Based on this model, an estimated total end-of-exposure level was then calculated for each worker employed in these departments. The strongest association with 2,3,7,8-TCDD levels was found for duration of employment in 2,4,5-T (estimated yearly increase, 75.6 ng/kg blood) and TCP (estimated yearly increase < 1957, 292 ng/kg) production departments. For the entire cohort, the mean estimated 2,3,7,8-TCDD level was 141.4 ng/kg (median, 38.2 ng/kg). Blood measurements of PCDDs are presented in Table 14. Exposure to other carcinogens (e.g., dimethyl sulfate (see IARC, 1987g) and benzene (see IARC, 1987h) had occurred in some departments. As a reference, a cohort of 2528 workers from a gas supply company was adopted. Causes of death were ascertained from hospital and medical records whenever possible, or from death certificates, insurance data or next-of-kin interview. The validity of cause of death determination was successfully tested. Relative risks at seven exposure levels (reference, the first four quintiles and the ninth and tenth deciles of the estimated 2,3,7,8-TCDD levels) were estimated using Cox's regression techniques. An internal comparison group was also adopted composed of the chemical workers in the two lowest quintiles of the exposure levels distribution. Follow-up was 99% successful. Total mortality was elevated in all 2,3,7,8-TCDD exposure categories and showed a significant trend (p = 0.01) with increasing exposure level Updated relative risks (RR) for total TEQ levels (not 2,3,7,8-TCDD levels) were published in an erratum (Flesch-Janys, 1996) and were as follows for all cancers combined (124 cases): 1.0–12.2 ng TEQ/kg, RR, 1.0 (95% CI, 0.6–1.8); 12.3–39.5 ng TEQ/kg, RR, 1.3 (95% CI, 0.8–2.1); 39.6–98.9 ng TEQ/kg, RR, 1.2 (95% CI, 0.7–1.9); 99.0–278/5 ng TEQ/kg, RR, 1.2 (95% CI, 0.8–1.9); 278.6–545.0 ng TEQ/kg, RR, 1.3 (95% CI, 0.7–2.3); 545.1–1362 ng TEQ/kg, RR, 2.7 (95% CI, 1.7–4.4). A significant pattern of increasing risk with increasing TEQ levels was also obtained for cardiovascular and ischaemic heart diseases, but not for other causes of death. The mortality pattern for all causes, all cancers, cardiovascular diseases and ischaemic heart diseases remained virtually unchanged when 2,3,7,8-TCDD exposure levels were used. When the internal comparison group of chemical workers was adopted, the relative risk estimates were lower and the confidence intervals wider. Nevertheless, total mortality (RR, 1.4; 95% CI, 1.0–2.1) and all-cancer mortality (RR, 2.0; 95% CI, 1.1–3.8) remained clearly elevated in the highest exposure category. When the 149 workers with exposure to one of the possible confounders (dimethyl sulfate) were excluded from analysis, the all-cancer mortality remained elevated, especially in the highest exposure category (RR, 2.3; 95% CI, 1.1–4.6) and the dose-response pattern was significant (p > 0.01). Smoking habits and socioeconomic status were similar in the chemical and gas worker cohorts. No data on specific cancer sites were reported.

Becher et al. (1996) reported the mortality of 2479 male workers employed in four German plants who were involved in the production of phenoxy herbicides and chloro-phenols or who were likely to have been in contact with these substances and their contaminants, PCDDs (often including 2,3,7,8-TCDD) and PCDFs. Only workers with German nationality and at least one month of employment were included. The follow-up was performed through 1989 (1992 in one plant). Cause of death was recorded from death certificates or, in some cases, hospital or physician reports. Expected figures were calculated from national rates. Blood concentrations of 2,3,7,8-TCDD were measured in workers from three of the four plants (Kogevinas et al., 1997). Workers in the Hamburg plant (plant I), including a group of females not considered here, had already been under investigation (see above; Manz et al., 1991; Nagel et al., 1994; Flesch-Janys et al., 1995; Flesch-Janys et al., 1996b). In this plant (1144 male workers), concentrations of 2,3,7,8-TCDD in the order of 5–10 mg/kg were measured in 1957–70 in some products (chloro-phenols and 2,4,5-T and its esters) and of 10–50 mg/kg in some process and waste streams. After the early 1970s, the concentrations were below 1 mg/kg. Numerous cases of chloracne occurred in 1954, at the early stage of development of herbicide manufacture. Measurements of blood concentrations from 190 workers in 1985–94 ranged between 3 and 2252 ng/kg (mean, 141 ng/kg). In plant II (135 workers), TCP was manufactured with 2,3,7,8-TCDD levels on average about 10 µg/kg. Eleven cases of chloracne were reported among maintenance workers in the early 1970s and measurements in 1989–92 from 20 workers ranged between 23 and 1935 ng/kg fat (mean, 401.7 ng.kg). In plant III (520 workers), a variety of phenoxy herbicides were produced. 2,3,7,8-TCDD levels in the products were reported to be in the ‘sub-ppm range’. In plant IV (680 workers) (BASF, but the cohort is different from that of the accident study), phenoxy herbicides were synthesized in a large building where many other chemicals were also produced. Measurements in 1996 from 19 workers ranged between 1.3 and 6.49 ng/kg fat (mean, 3.2 ng/kg). In this plant, slightly elevated concentrations of higher chlorinated PCDDs were found in synthesis workers. No cases of chloracne were reported in plants III or IV. In the total cohort, mortality increases were seen for all cancers combined (138 deaths; SMR, 1.2; 95% CI, 1.0–1.4), cancer of the buccal cavity and pharynx (9 deaths; SMR, 3.0; 95% CI, 1.4–5.6), lung cancer (47 deaths; SMR, 1.4; 95% CI, 1.1–1.9), lymphatic and haematopoietic neoplasms (13 deaths; SMR, 1.7; 95% CI, 0.9–2.9) and non-Hodgkin lymphoma (6 deaths; SMR, 3.3; 95% CI, 1.2–7.1). Examination of trends with increasing time since first exposure showed: for all cancers combined (0–10 years: 15 deaths; SMR, 0.9; 95% CI, 0.5–1.5; 10-< 20 years: 46 deaths; SMR, 1.3; 95% CI, 0.9–1.7; ≥20 years: 77 deaths; SMR, 1.2; 95% CI, 1.0–1.5 [p for trend = 0.67]), for lymphatic and haematopoietic neoplasms (0–10 years: 2 deaths; SMR, 1.1; 95% CI, 0.1– 4.0; 10–< 20 years: 4 deaths; SMR, 1.7; 95% CI, 0.5–4.4; ≥20 years: 7 deaths; SMR, 1.9; 95% CI, 0.8–4.0 [p for trend = 0.51]) and for non-Hodgkin lymphoma (0–10 years: 0 deaths; 10-< 20 years: 2 deaths; SMR, 3.6; 95% CI, 0.4–13.1; ≥ 20 years: 4 deaths; SMR, 4.3; 95% CI, 1.2–10.9; [p for trend = 0.24]). All-cause (345 deaths; SMR, 1.2; 95% CI, 1.0–1.3) and all-cancer mortality (97 deaths; SMR, 1.3; 95% CI, 1.1–1.6) was elevated only for plant I, where statistically significant increases were also seen for lung cancer (31 deaths; SMR, 1.5; 95% CI, 1.0–2.1), lymphatic and haematopoietic neoplasms (11 deaths; SMR, 2.4; 95% CI, 1.2–4.3), non-Hodgkin lymphoma (4 deaths; SMR, 3.8; 95% CI, 1.0–9.6) and multiple myeloma (3 deaths; SMR, 5.4; 95% CI, 1.1–15.9). Mortality from accidents and suicide was also increased. An increase in the incidence of non-Hodgkin lymphoma for plant II (2 deaths; SMR, 12.0; 95% CI, 1.5–43.5), and in that of cancer of the buccal cavity and pharynx for plant IV (6 deaths; SMR, 8.2; 95% CI, 3.0– 17.9) were seen.

(e) British plants

Four cohorts of workers employed between 1963 and 1985 in British factories producing or formulating — among other chemicals — phenoxy herbicides, including 2,4,5-T (Kauppinen et al., 1993) and chlorophenols (some of which may have been contaminated by PCDDs) were investigated (Coggon et al., 1991). In one of the plants (factory A), 2,4,5-T was synthesized between 1968 and 1978. In the other plants, 2,4,5-T was formulated only. The study was restricted to men. A total of 2239 subjects (employed during 1963–85) met the inclusion criteria. Job histories were used to classify workers according to potential exposure to phenoxy herbicides and chlorophenols. Approximately 50% of employees had directly worked with phenoxy herbicides. No environmental or personal monitoring had been carried out. Subjects were traced up to December 1987. Information was obtained on cause of death for deceased persons and on any cancers registered among living study subjects. Expected figures were obtained from rates for England and Wales, and in some analyses adjustment for local differences in mortality was applied. In the combined cohort, all-cause mortality was slightly higher than expected (152 deaths; SMR, 1.1; 95% CI, 0.9–1.3), and this was largely due to excesses of circulatory diseases and deaths from violent causes. Mortality from all-cancers combined was as expected (37 deaths; SMR, 1.0; 95% CI, 0.7–1.4). Statistically nonsignificant excesses were present for lung cancer (19 deaths; SMR, 1.3; 95% CI, 0.8– 2.1) and non-Hodgkin lymphoma (2 deaths; SMR, 2.3; 95% CI, 0.3–8.3), even after local adjustment. When analysis was restricted to subjects with greater than background exposure to phenoxy compounds or chlorophenols, the SMR for lung cancer was 1.2 (14 observed deaths; [95% CI, 0.7–2.1]). Both non-Hodgkin lymphoma deaths occurred among these workers (SMR, 2.8; [95% CI, 0.3–10.2]). In factory A, 12 cancer deaths were observed (12.3 expected) and six were from lung cancer (SMR, 1.3 [95% CI, 0.5– 2.8]).

(f) Dutch plants

The mortality of two cohorts of workers employed between 1955 and 1986 in the synthesis and formulation of phenoxy herbicides and chlorophenols in the Netherlands was examined (Bueno de Mesquita et al., 1993). In one of the plants (A), where the main production was 2,4,5-T and derivatives, an accident in 1963 caused a release of PCDDs, including 2,3,7,8-TCDD. In factory B, production included MCPA, 4-chloro-2-methyl-phenoxypropionic acid (MCPP) and 2,4-D. The study enrolled 2074 manufacturing male workers from the two plants (963 exposed to phenoxy herbicides and 1111 not exposed). In addition, 145 workers probably exposed to 2,3,7,8-TCDD during the industrial accident and the clean-up operations were examined. Definition of individual exposure to phenoxy herbicides, chlorophenols or contaminants (PCDDs/PCDFs) was based on occupational history derived from job records and personal interviews, including periods of employment in different departments and positions held. Follow-up was 97% complete. The 190 female workers were excluded from the analysis. Of the accident workers, only 139 had sufficient data for analysis. Expected numbers of deaths were calculated from national rates. In addition, mortality rates of exposed and non-exposed workers were internally compared by Poisson regression analysis. Among subjects in factory A only, in comparison with the general population, all-cancer mortality was increased (26 deaths; SMR, 1.2; 95% CI, 0.8–1.7). Statistically nonsignificant increases were also seen for cancers of the pancreas (3 deaths; SMR, 2.9; 95% CI, 0.6–8.4), large intestine (3 deaths; SMR, 2.4; 95% CI, 0.5–7.0) and prostate (2 deaths; SMR, 2.2; 95% CI, 0.3–7.8), lymphosarcoma (non-Hodgkin lymphoma) (1 death; SMR, 2.0; 95% CI, 0.1–11.4) and myeloid leukaemia (1 death; SMR, 2.9; 95% CI, 0.1–15.9). No death due to soft-tissue sarcoma was reported. In comparison with the non-exposed workers, exposed subjects in both plants exhibited increased mortality from all cancers combined (31 exposed deaths; rate ratio (RR), 1.7; 95% CI, 0.9–3.4) and respiratory tract cancer (9 deaths; RR, 1.7; 95% CI, 0.5–6.3). Six cancers of the urogenital organs were observed among the exposed and none among the non-exposed. A nonsignificant increase in deaths from lymphatic and haematopoietic neoplasms was also noted (4 deaths; RR, 2.6; 95% CI, 0.3–125). Increases in all cancers combined (RR, 2.0; 95% CI, 0.8–4.9) and lung cancer (RR, 3.9; 95% CI, 0.5–31.1) were confined to factory A, where the accident occurred and where the opportunity for exposure to 2,3,7,8-TCDD was highest. In the small group of accident-exposed workers, 10 cancer deaths were observed (SMR, 1.4; 95% CI, 0.7–2.5).

The study was later extended in time (1955–91) and enlarged in size (2298 subjects including 191 females) (Hooiveld et al., 1996a; Kogevinas et al., 1997). More accurate and elaborate proxies of exposure were used in the analysis, based on modelled 2,3,7,8-TCDD levels in serum, measured in 1993 in a subset of 31 surviving exposed (mean concentration, 53 ng/kg; range, 1.9–194) and 16 unexposed (mean concentration, 8 ng/kg) cohort members. Fourteen subjects exposed during the accident in factory A in 1963 had the highest levels (mean concentration, 96 ng/kg; range, 15.8–194). In this factory, both all-cause (139 observed; SMR, 1.3; 95% CI, 1.1–1.5) and all-cancer mortality (51 observed; SMR, 1.5; 95% CI, 1.1–1.9) were significantly increased. Excesses at specific sites were seen for urinary bladder (4 observed; SMR, 3.7; 95% CI, 1.0–9.5), kidney cancer (4 observed; SMR, 4.1; 95% CI, 1.1–10.4) and non-Hodgkin lymphoma (3 observed; SMR, 3.8; 95% CI, 0.8–11.0). Non-Hodgkin lymphoma was also increased in factory B (1 case only). Age- and time-adjusted relative risks comparing exposed and unexposed workers in factory A showed significant increases in mortality from all causes (139 observed; RR, 1.8; 95% CI, 1.2–2.5) and all cancers (51 observed; RR, 3.9; 95% CI, 1.8–8.8). Lung and urinary tract cancers showed numerically higher, but statistically nonsignificant increases. Three non-Hodgkin lymphomas were seen among the exposed and one among the unexposed (a nonsignificant increase). When workers were subdivided into three categories (low, medium and high exposure) according to model-predicted serum levels of 2,3,7,8-TCDD, relative risks for all causes, all cancers and lung cancer were significantly elevated in both middle and high categories, and were highest in the highest exposure group.

(g) IARC multi-country study

An international cohort of workers exposed to phenoxy herbicides and chlorophenols was set up by the International Agency for Research on Cancer in association with the United States National Institute of Environmental Health Sciences (Saracci et al., 1991). The cohort included 16 863 men and 1527 women employed in production or spraying, distributed among 20 cohorts from 10 countries, including the British and Dutch cohorts described above. Their mortality from 1955 onwards was examined, and follow-up was successful for 95% of the cohort members. National mortality rates were used for reference. Exposure assessment was based on work histories collected in each factory through questionnaires with the assistance of industrial hygienists, workers and factory personnel. A total of 13 482 workers were classified as ‘exposed’ to phenoxy herbicides, 416 were classified as ‘possibly exposed’, 541 had ‘unknown’ exposure and 3951 were classified as ‘non-exposed’. In the entire cohort, all-cause mortality was lower than expected (SMR, 0.95; 95% CI, 0.9–1.0). Among men exposed to phenoxy herbicides or chlorophenols, mortality from all cancers combined was close to expectation (499 observed; SMR, L0; 95% CI, 0.9–1.1). Significant increases were seen for thyroid cancer (4 observed; SMR, 3.7; 95% CI, 1.0–9.4) and benign and unspecified neoplasms (12 observed; SMR, 2.0; 95% CI, 1.0–3.5); significant deficits were observed for skin (3 observed; SMR, 0.3; 95% CI, 0.1–0.9) and brain cancer (6 observed; SMR, 0.4; 95% CI, 0.1–0.8). Elevated risks were also seen for cancers of the nose and nasal cavities (3 observed; SMR, 2.9; 95% CI, 0.6–8.5), testis (7 observed; SMR, 2.3; 95% CI, 0.9–4.6) and other endocrine glands (3 observed; SMR, 4.6; 95% CI, 1.0–13.5) and for soft-tissue sarcomas (4 observed; SMR, 2.0; 95% CI, 0.6–5.2). In the probably exposed category, lung cancer mortality was significantly increased (11 observed; SMR, 2.2; 95% CI, 1.1– 4.0). Significant increases were also seen among non-exposed workers, for unspecified digestive organs and for benign and unspecified neoplasms. The increase in soft-tissue sarcoma concerned workers after 10–19 years since first exposure (4 observed; SMR, 6.1; 95% CI, 1.7–15.5); no differentiation in risk was noted in relation to duration of exposure or probable 2,3,7,8-TCDD exposure. The SMR estimates for testicular and thyroid cancer were highest among workers probably exposed to 2,3,7,8-TCDD (SMR, 3.0 and 4.3, respectively) compared with those probably not exposed (SMR, 1.6 and 3.1, respectively).

Two nested case–control studies of soft-tissue sarcoma (11 incident cases, 55 controls) and non-Hodgkin lymphoma (32 incident cases, 158 controls) were conducted by Kogevinas et al. (1995) within the IARC cohort studied by Saracci et al. (1991). Exposures to 21 chemicals or mixtures were estimated by a panel of three industrial hygienists. Levels of exposure were evaluated using a relative scale, since few actual measurements of past exposure were available. A cumulative exposure score was calculated for each subject and chemical, on the basis of estimated level of exposure and duration of exposure (in years). The model that was used as the conceptual framework in deriving levels of exposure included variables related to department/job, emission of chemicals, contact with chemicals, personal protection, and other relevant determinants of exposure. Excess risk for soft-tissue sarcoma was associated with exposure to any phenoxy herbicide (odds ratio, 10.3; 95% CI, 1.2–91.0) and to each of the three major classes of phenoxy herbicides (2,4-D, 2,4,5-T and MCP A). Soft-tissue sarcoma was also associated with exposure to any PCDD/PCDF (odds ratio, 5.6; 95% CI, 1.1–28.0) and with exposure to 2,3,7,8-TCDD (odds ratio, 5.2; 95% CI, 0.9–32.0). Associations between non-Hodgkin lymphoma and phenoxy herbicides were generally weaker. The odds ratio between non-Hodgkin lymphoma and 2,3,7,8-TCDD exposure was 1.9 (95% CI, 0.7–5.1). A monotonie increase in risk was observed for cumulative exposure (categorized in four categories, non-exposed, low, medium, high exposure) to 2,4-D (odds ratio for highest category, 13.7; 95% CI, 0.9–309; p-value for trend = 0.01), 2,4,5-T (odds ratio for highest category, 7.7; 95% CI, 0.5–477; p-value for trend = 0.07), any PCDD/PCDF (odds ratio for highest category, 19.0; 95% CI, 1.3–1236; p-value for trend = 0.008) and 2,3,7,8-TCDD (odds ratio for highest category, 10.6; 95% CI, 0.6–671; p-value for trend = 0.04). [The Working Group noted that analysis for specific exposures was complicated by the exposure of most workers to a multitude of herbicides.]

While Saracci et al. (1991) studied the men in the IARC cohort, 701 women were studied by Kogevinas et al. (1993) for both cancer incidence and mortality. Among 169 women probably exposed to 2,3,7,8-TCDD, excess incidence was observed for all cancers combined (9 cases; standardized incidence ratio (SIR), 2.2; 95% CI, 1.0–4.2). There was one case of breast cancer (SIR, 0.9; 95% CI, 0.0–4.8). For 532 women probably not exposed to 2,3,7,8-TCDD, the rate ratio for incidence of all cancers combined was 0.8 (20 cases; 95% CI, 0.5–1.2). Cause-specific analyses were based on small numbers. Mortality results paralleled those for incidence.

The international cohort studied by Saracci et al. (1991) was updated and expanded with the data of Fingerhut et al. (1991a,b) and Becher et al. (1996) (Kogevinas et al., 1997). Follow-up differed by plant, but most European plants were followed through 1991–92, while the United States plants were followed through 1987. Each of the 21 863 male and female workers exposed to phenoxy herbicides or chlorophenols was placed in one of three categories: (1) those exposed to 2,3,7,8-TCDD or higher chlorinated PCDDs (n = 13 831); (2) those not exposed to 2,3,7,8-TCDD or higher chlorinated PCDDs (n = 7553); and (3) those of unknown exposure to 2,3,7,8-TCDD or higher chlorinated PCDDs (n = 479). The latter category included all workers in one British cohort for which production history, particularly for 2,4,5-T, was incomplete. Three criteria were used to classify workers as exposed to 2,3,7,8-TCDD or higher chlorinated PCDDs: (i) employment during the period of production, formulation or spraying of 2,4,5-T, 2,4,5-trichlorophenoxypropionic acid, TCP, hexachlorophene, Erbon, Ronnel, PCP or 2,3,4,6-tetrachlorophenol; and (ii) employment in plants with documented (through serum, adipose tissue or environmental measurements) exposure to 2,3,7,8-TCDD or higher chlorinated PCDDs at levels above background; or (iii) in the absence of PCDD measurements, employment in plants or companies with documented large-scale production, formulation or spraying of the above-mentioned phenoxy herbicides and chlorophenols. An average production of these chemicals of 10 tonnes per year was chosen a priori, below which it was considered that the probability of contamination and significant exposure to 2,3,7,8-TCDD and higher chlorinated PCDDs would be minimal for most workers in a cohort. Current mean levels of 2,3,7,8-TCDD, measured in 574 workers from 10 companies in 7 countries, ranged from 3 to 389 ng/kg lipid. Among workers exposed to 2,3,7,8-TCDD or higher chlorinated PCDDs, mortality was elevated for soft-tissue sarcoma (6 deaths; SMR, 2.0; 95% CI, 0.8–4.4). Mortality from all cancers combined (710 deaths; SMR, 1.1; 95% CI, 1.0–1.2), non-Hodgkin lymphoma (24 deaths; SMR, 1.3; 0.9–2.1) and lung cancer (225 deaths; SMR, 1.1; 95% CI, 1.0–1.3) was slightly elevated. Risks for all cancers combined and for soft-tissue sarcomas and lymphomas increased with time since first exposure. Workers not exposed to 2,3,7,8-TCDD or higher chlorinated PCDDs had SMRs of 1.0 for all cancers, for non-Hodgkin lymphoma and for lung cancer; soft-tissue sarcoma was slightly elevated, based on two deaths (SMR, 1.4; 95% CI, 0.2–4.9). In a direct comparison between those exposed to higher chlorinated PCDDs versus lower ones or none, a rate ratio of 1.3 (95% CI, 1.0– 1.8) for all cancers combined was found. This study represents the largest overall cohort of 2,3,7,8-TCDD-exposed workers.

(a) Commercial

A cohort study of 348 male railroad workers in Sweden exposed to herbicides during the period 1957–72 was initially followed through 1972 (Axelson & Sundell, 1974) and subsequently through 1978 (Axelson et al., 1980). The workers had been exposed to amitrole (see IARC, 1987i) and to phenoxy herbicides, particularly 2,4-D and 2,4,5-T. Only workers with a minimum of 45 days of exposure, not necessarily continuous, were included in the cohort. Workers were classified into three sub-cohorts depending on the type of herbicide used: those with exposure to amitrole, those with exposure to phenoxyherbicides and those with exposure to both. Workers had also been exposed to some less commonly used herbicides such as monuron, diuron, or atrazine (see IARC, 1991). None of the 348 individuals was lost to follow-up. Underlying cause of death was recorded on the basis of information from death certificates as classified by the National Central Bureau of Statistics. Expected numbers of deaths were calculated using cause-, calendar period- and age-specific national mortality rates. Mortality rates for 1975 were used for the period 1975–78 since national rates were not available at the time of the study for the last three years. In the total cohort, all-cause mortality was approximately that expected (45 deaths; SMR, 0.9, [95% CI, 0.7–1.2]) and mortality from all cancers combined was slightly elevated (17 deaths; SMR, 1.4 [95% CI, 0.8–2.3]). The SMRs for all cancers in the sub-cohort exposed to phenoxy herbicides was 1.1 (6 deaths [95% CI, 0.4–2.3]), with an excess observed for stomach cancer (2 deaths; SMR, 3.1 [95% CI, 0.4–11.1]). The SMR for all cancers combined in the sub-cohort exposed to both phenoxy herbicides and amitrole was 2.1 (6 deaths [95% CI, 0.8–4.5]) with an excess for stomach cancer based on only one death (0.3 expected) and lung cancer (one death; 0.5 expected). When a 10-year latency period was applied, the SMR for all cancers combined for the phenoxy herbicides sub-cohort was 1.9 (6 deaths [95% CI, 0.7–4.2]) and that for the sub-cohort exposed to both phenoxy herbicides and amitrole was 3.4 (6 deaths [95% CI, 1.2–7.3]).

A follow-up study was conducted among forestry workers in Sweden (Hogstedt & Westerlund, 1980). One hundred and forty-five male forestry workers registered as having worked with phenoxy herbicides (2,4-D and 2,4,5-T) for at least five days, 16 foremen (claimed to be highly exposed) and 251 unexposed foresters were identified from pay lists of a timber company in Hälsingland between 1954 and 1967. Persons who worked for more than five days with other herbicides were not included. Vital status and cause of death were assessed for the period 1954–78 from national health registers, except for three exposed and seven unexposed foresters. Incident cancer cases between 1958 and 1978 were identified through cancer registries. National statistics were used as a reference. The average duration of exposure was 30 days (range, 6–114) among forest workers and 176 days (range, 16–317) among foremen. In the exposed cohort, 29 deaths occurred versus 28.0 expected; 47 deaths were observed in the unexposed versus 64.4 expected. Tumour deaths were three (versus 1.0 expected) among foremen, two (versus 5.4 expected) among exposed workers and five (versus 6.4 expected) among unexposed workers. Incident cancer cases were in excess among foremen (5 versus 1.4 expected; p > 0.02) and less than expected in the other exposed workers (3 versus 8.4 expected). No specific cancer site was remarkable (stomach (1), pancreas (2), lung (1), skin (1), prostate (2), urinary bladder (1)).

A cohort of male phenoxy herbicide applicators in Finland was followed prospectively from 1972 to 1980 (Riihimäki et al., 1982). The herbicides 2,4-D and 2,4,5-T had been used in Finland since the mid-1950s, with the peak consumption (about 50–70 tonnes of active ingredient per year) occurring in the late 1960s. Analysis of some old herbicide preparations used in Finland in 1962–67 suggested that the 2,3,7,8-TCDD content ranged between 0.1 and 0.9 mg/kg. After 1972, the use of these chemicals declined. The cohort of herbicide applicators was identified in 1972 from the personnel records of the four main Finnish employers involved in chemical control of brushwood. The cohort included 1971 men exposed to 2,4-D and 2,4,5-T for at least two weeks during 1955–71. The data were collected by office personnel in the companies from various sources such as payrolls. In some cases, supplementary information was sought through interviews with foremen and clerks. About a quarter of the population had been exposed for more than eight weeks, as of 1971 During 1955–71, 45 individuals died, leaving 1926 subjects alive followed during 1972–80; fifteen subjects were not located in the population data register and were excluded from the cohort. The underlying cause of death for all deceased subjects was retrieved from death certificates registered at the Central Statistical Office. National age- and sex-specific mortality rates for 1975 were used to calculate expected numbers of deaths. Mortality from all cancers during the period 1972–80 was lower than expected from national rates (26 deaths [SMR, 0.7; 95% CI, 0.5–1.0]), with a slightly higher rate observed during the last years of follow-up, (1977–80; 17 deaths [SMR, 0.9; 95% CI, 0.5–1.5]). Similar results were obtained when allowing for a 10-year or a 15-year latency period, or when the analysis was restricted to subjects with the longest duration of exposure. The SMR for lung cancer, allowing for a 10-year latency period, was 1.1 (12 deaths [95% CI, 0.6–1.9]). There were no deaths registered from malignant lymphomas or soft-tissue sarcoma.

Cancer incidence was studied in a cohort of 20 245 licensed pesticide applicators in Sweden (Wiklund et al., 1987, 1988, 1989). Since 1965, a licence has been mandatory for using the most acutely toxic pesticides. The workers in the study had been issued a licence between 1965 and 1976 and were followed up for cancer incidence to 31 December 1982. A survey of a random sample of 273 workers showed that 72% had been exposed to phenoxy herbicides for one day or more during the 1950s, 1960s and 1970s. Phenoxy herbicides had been used in Sweden since the late 1940s. The main compound used in agriculture was MCPA and, since the mid-1960s, MCPP and dichlor-prop (2-(2,4-dichlorophenoxy)propionic acid). 2,4-D and 2,4,5-T were used to a lesser extent. Cancer incidence rates for the Swedish population were used for comparison. A total of 558 cancer cases were observed in the cohort (SIR, 0.86; 95% CI, 0.79–0.93). Excess risks were observed for lip cancer (14 observed; SIR, 1.8; 95% CI, 1.0–2.9), testicular cancer (18 observed; SIR, 1.6; 95% CI, 0.9–2.5) and Hodgkin's disease (11 observed; SIR, 1.2; 95% CI, 0.6–2.2) but not for lung cancer (38 observed; SIR, 0.5; 95% CI, 0.4–0.7), non-Hodgkin lymphoma (21 observed; SIR, 1.0; 95% CI, 0.6–1.5) or cancer at any other site. The SIR for testicular cancer increased with time since licence (0–4 years, 0.9; 5–9 years, 1.4; ≥ 10 years, 2.5; based on four, six and eight cases, respectively). The SIR for lung cancer increased with years since first employment from 0.3 (0–4 years) to 0.5 (5–9 years) to 0.6 (≥10 years). The authors provided data showing that smoking was less prevalent among pesticide applicators than among other occupational categories in Sweden, strongly suggesting that the observed deficit of lung cancer was due to lower cigarette consumption. A follow-up from the date of licence until 31 December 1984 indicated no excess risk for soft-tissue sarcoma (7 cases observed; SIR, 0.9; 95% CI, 0.4–1.9) (Wiklund et al., 1988).

A historical cohort study was conducted among Danish gardeners heavily exposed to pesticides (Hansen et al., 1992). The cohort comprised 859 women and 3156 men and included all employed persons who, in May 1975, were members of one of the 10 local trade unions of gardeners associated with the Danish Union of General Workers. Subjects were followed from 1 May 1975 to 1 January 1985 through the records of the Danish Central Population Register, with 100% of the subjects being traced. Cancer incidence was recorded through the Danish Cancer Registry. The analysis was limited to subjects 30–79 years of age. The gardeners constituted three fairly separate groups, namely workers in greenhouses (nearly all female), a small group of nursery gardeners and gardeners in public parks, gardens and cemeteries (the majority of workers). Gardeners working outdoors had been exposed to phenoxy herbicides (2,4-D, MCPA, 2,4,5-T) and amitrole. This exposure took place regularly throughout the growing season. Data on individual job histories were not obtained, except for a limited subset of cancer cases. A total of 219 cancer cases were identified in the study, 217 among subjects aged 30–79 years. There was no elevation in incidence of all cancers combined (SIR, 1.0; 95% CI, 0.9–1.2) or of most common neoplasms. An increased risk was seen for soft-tissue sarcoma (code 197 in ICD 7th revision), with an SIR of 4.6 (3 cases; 95% CI, 0.9–13.3) which was statistically significant among men (3 cases; SIR, 5.3; 95% CI, 1.1–15.4). Elevated risks were seen for non-Hodgkin lymphoma (8 cases; SIR, 2.0; 95% CI, 0.9– 3.9) and for chronic lymphatic leukaemia (6 cases; SIR, 2.5; 95% CI, 0.9–5.5). Two individuals with soft-tissue sarcoma, six with non-Hodgkin lymphoma and three with chronic lymphatic leukaemia had worked as gardeners for more than 10 years. No additional soft-tissue sarcoma cases were identified in this cohort when the records of the cancer registry were searched for sarcomas which might have occurred in parenchymal organs and consequently not been included in the ICD code (code 197, 7th rev.) used for the main analysis.

(b) Military

United States Viet Nam veterans as a group, with the exception of those known to have directly handled Agent Orange, show no evidence of exposure to PCDDs or PCDFs above that of other veterans who did not serve in Viet Nam (i.e., beyond background levels), on the basis of serum measurements. However, veterans actually involved in spraying, as members of the Air Force Operation Ranch Hand or of the Army Chemical Corps, have shown elevated serum levels of 2,3,7,8-TCDD (see Section 1.3.1(a)(ii)).

The mortality of 1261 Ranch Hand veterans has been compared with that of 19 101 other Air Force Viet Nam veterans who were not exposed to Agent Orange (Michalek et al., 1990). Person-time at risk began when the tour of duty in Viet Nam began. The most recent mortality report extends follow-up through 1993 (Ketchum & Akhtar, 1996), with 31394 person-years among Ranch Hand personnel and 490 792 person-years among the comparison group. The SMR for all causes of death was 1.0 (95% CI, 0.8– 1.2), based on 118 deaths. There were 30 cancer deaths (SMR, 0.9; 95% CI, 0.6–1.3) and 12 lung cancer deaths (13.0 expected). There was one death from soft-tissue sarcoma (a fibroma; ICD 171.3) versus 0.4 expected for the category cancer of the connective tissue (ICD 171.0–171.9), and one from non-Hodgkin lymphoma (0.7 expected). No case of chloracne has been reported among Ranch Hand veterans.

The mortality of 894 members of the Army Chemical Corps was studied by Thomas and Kang (1990). Follow-up extended through 1987, and the United States population was used as the referent. Person-time at risk was taken to begin when service ended in Viet Nam. There were 53 deaths (SMR, 1.1) in this cohort, with about 16 000 person-years and an average follow-up of 18 years. There were six cancer deaths (SMR, 0.9): two lung cancer deaths versus 1.8 expected, two leukaemias versus 0.5 expected and two brain cancers versus 0.4 expected. Review of medical records found two incident cases of Hodgkin's disease, with approximately 0.7 expected.

(a) Soft-tissue sarcoma (see Table 35)

Table 35

Case–control studies on soft-tissue sarcoma containing information on exposure to phenoxy herbicides, chlorophenols or PCDDs.

A series of studies on soft-tissue sarcoma were conducted in Sweden applying similar methodology (Hardell & Sändstrom, 1979; Erickson et al., 1981; Hardell & Erickson, 1988; Erickson et al., 1990). Relevant exposure measurements are presented in Tables 16 and 17.

The first case–control study associating phenoxy herbicides and chlorophenols with soft-tissue sarcoma was conducted in 1978 in the population of the region of Umeå, in northern Sweden (Hardell & Sändstrom, 1979). The study was initiated following a case report of three soft-tissue sarcoma patients who had been exposed to phenoxy herbicides (Hardell, 1977). The study included 21 living and 31 deceased pathologically verified soft-tissue sarcoma male cases, diagnosed in one hospital during 1970–77. Four population controls matched for age, sex, place of residence and vital status were selected per case. Living controls were selected from the national population registry and deceased controls from the national registry of causes of death. The response rate was 100% for cases and 99% for the controls. A self-administered questionnaire was sent to living subjects and to the next-of-kin of deceased subjects (approximately 60% of all questionnaires). Information was requested about a variety of exposures and the questionnaire was supplemented by telephone interviews of selected subjects. Any subject exposed for more than one day was characterized as exposed. Exposure to phenoxy herbicides and/or chlorophenols was associated with a six-fold increased risk. In the matched analysis, the odds ratio was 6.2 [confidence intervals not provided]. In an unmatched analysis, the odds ratio was 5.7 (95% CI, 2.9–11.3). The odds ratio among living subjects was 9.9 and that among deceased subjects was 3.8. The odds ratio for exposure only to phenoxy herbicides was 5.3 (95% CI, 2.4–11.5). After excluding the three index cases [but not four additional cases included in a pilot study], this odds ratio became 4.7 (95% CI, 2.0– 10.7). Tobacco smoking, exposure to DDT, exhaust fumes and emulsion agents did not appear to be associated with risk for soft-tissue sarcoma.

Responding to criticisms that selection and information bias could have affected the results of the previous study (Hardell & Sändstrom, 1979) and of a study on malignant lymphomas (Hardell et al., 1981), Hardell (1981) conducted a further analysis using colon cancer patients as controls. The study is fully described in Section 2.2.4(d). Results obtained when using colon cancer patients as controls were very similar to the original results obtained when using population controls.

A case–control study of the association between soft-tissue sarcoma and phenoxy herbicides and chlorophenols was conducted among the population of the five southernmost counties of Sweden (Erickson et al., 1981). The study included 110 pathologically verified soft-tissue sarcoma cases (including 38 deceased cases) reported to the cancer registry during 1974–78. Two population controls matched for age, place of residence and vital status were selected per case. Living controls were selected from the national population registry and deceased controls from the national registry of causes of death. A self-administered questionnaire was sent to living subjects and to the next-of-kin (approximately 35% of all questionnaires) of deceased subjects. Information was requested about a variety of exposures. The questionnaire was supplemented by telephone interviews for subjects with incomplete or obscure replies to questions on exposure to solvents or pesticides, and also for all subjects reporting work in agriculture, forestry or horticulture. Any subject exposed to phenoxy herbicides for more than one day was characterized as exposed. The predominant exposure in this area was to MCPA, considered to be free of 2,3,7,8-TCDD contamination, to 2,4-D and the analogous phenoxypropionic acids, MCPP and dichlorprop, possibly contaminated by PCDDs other than 2,3,7,8-TCDD. A high level of exposure to chlorophenols used as wood preservatives was defined as one week of continuous or one month of discontinuous exposure. The odds ratio for exposure to phenoxy herbicides and/or chlorophenols was 5.1 in the matched analysis (95% CI, 2.5–10.4). The odds ratio for exposure to any phenoxy herbicide was 6.8 (95% CI, 2.6–17.3). A higher relative risk was observed for exposure to phenoxy herbicides considered to be contaminated with 2,3,7,8-TCDD (odds ratio, 17.0) compared with herbicides uncontaminated with 2,3,7,8-TCDD (odds ratio, 4.2). Longer duration of exposure to phenoxy herbicides (more or less than 30 days) was not associated with an increased risk. High-level exposure to chlorophenols was associated with an increased risk (odds ratio, 3.3; 95% CI, 1.3–8.1), while low-level exposure was not associated with an increased risk. Tobacco smoking or exposure to organic solvents, various pesticides or other chemical agents did not appear to be associated with risk for soft-tissue sarcoma.

A case–control study on soft-tissue sarcoma was conducted in the population of three northern counties of Sweden (Hardell & Erickson, 1988). The study included 55 male histopathologically confirmed soft-tissue sarcoma cases (18 alive, 37 deceased) aged 25–80 years. Cases were diagnosed during 1978–83 and were identified through the regional cancer registry in Umeå. In eight cases, the tumour was classified as probable soft-tissue sarcoma but another malignancy could not be excluded. Three groups of controls were selected. A first group of 220 living population controls was matched with cases by age and residence at the time of diagnosis. The second group consisted of 110 dead controls. The third group consisted of 190 cancer cases (112 alive) (except for malignant lymphomas and nasopharyngeal cancer) drawn at random from the population cancer registry. A self-administered postal questionnaire was used and information was requested on lifetime occupational history, exposure to specific agents at work or leisure and lifestyle factors. The questionnaire was supplemented by telephone interviews for subjects with incomplete or obscure replies, and also for all subjects reporting work in agriculture, forestry, horticulture, carpentry and sawmills. The participation rate among both cases and controls was about 95%. The odds ratios for at least one day of exposure to phenoxy herbicides were 3.3 (95% CI, 1.4–8.1) when the two combined population control groups were used and 2.2 when cancer controls were used (95% CI, 0.9–5.3). Exclusion of the eight soft-tissue sarcoma cases with uncertain diagnosis gave a higher estimate of the risk (odds ratio, 3.7 (95% CI, 1.5–9.1) for population controls and 2.4 (95% CI, 1.0–5.9) for cancer controls). No association with exposure to chlorophenols was found. Exposure to 2,3,7,8-TCDD gave a crude rate ratio of 3.5 when population-based referents were used and 3.1 with cancer referents. A higher risk associated with exposure to DDT was present only among subjects with concomitant exposure to phenoxy herbicides. No statistically significant results were observed for other occupational exposures or for smoking.

A case–control study on soft-tissue sarcoma was conducted among the population of seven counties of central Sweden covered by the population cancer registry of Uppsala (Erickson et al., 1990). The study included 218 male histopathologically confirmed soft-tissue sarcoma cases (78 alive, 140 deceased) aged 25–80 years. Cases were diagnosed during 1978–86 and were identified through the regional cancer registry. One group of population controls was selected, matched for age, gender, county of residence and vital status. A self-administered postal questionnaire was used and information was requested on lifetime occupational history (including details of 16 specific occupations), exposure to specific agents at work or leisure and lifestyle factors. The questionnaire was supplemented by telephone interviews for subjects with incomplete or obscure replies, and also for all subjects reporting work in agriculture, forestry, horticulture, carpentry and sawmills. The participation rate for the cases was 92% (218/237 originally identified cases) and that for controls was 89% (212/237 originally identified controls). The odds ratio for exposure (at least one day) to phenoxy herbicides was 1.3 (95% CI, 0.7–2.6). The risk associated with use of 2,4,5-T was higher (odds ratio, 1.8; 95% CI, 0.9–3.9), especially for use before the 1950s (odds ratio, 2.9; 95% CI, 1.1–8.0) when contamination with PCDDs could be expected to be highest. A high risk was associated with exposure to chlorophenols (odds ratio, 5.3; 95% CI, 1.7–16.3), particularly PCP (odds ratio, 3.9; 95% CI, 1.2–12.9) among subjects characterized as having a high level of exposure (continuous exposure for more than one week or discontinuous exposure for more than one month). No clear dose-response relationship was observed for exposure to phenoxy herbicides. Exposure to phenoxy herbicides not contaminated with 2,3,7,8-TCDD was not associated with an increased risk. No statistically significant associations were observed for other occupational exposures or for smoking.

The data from the four Swedish case–control studies on soft-tissue sarcoma were aggregated (for a total of 434 cases and 948 controls) and re-analysed according to duration of exposure, latency and type of PCDD exposure, reported in a letter to the editor (Hardell et al., 1991). For exposure to PCDDs of any type, less than one year of exposure gave an odds ratio of 2.4 (58 exposed cases; 90% CI, 1.7–3.4), and more than one year gave an odds ratio of 6.4 (24 exposed cases; 90% CI, 3.5–12). Forty-six cases were exposed to 2,4,5-T, presumed to be contaminated with 2,3,7,8-TCDD, yielding odds ratios of 3.0 (40 exposed cases; 90% CI, 2.0–4.5) for less than one year of exposure and 7.2 (6 exposed cases; 90% CI, 2.6–20) for more than one year of exposure. However, even exposure to other PCDDs was associated with an increased risk for soft-tissue sarcoma (< 1 year: 18 exposed cases; odds ratio, 1.7; 90% CI, 1.0–2.9;>1 year: 18 exposed cases; odds ratio, 6.2; 90% CI, 2.9–13). It was concluded that PCDDs other than 2,3,7,8-TCDD might have contributed to the noted effect.

A case–control study evaluating the association between exposure to phenoxy herbicides and chlorophenols with the occurrence of soft-tissue sarcoma was carried out in New Zealand (Smith et al., 1984b). Preliminary results had been reported earlier (Smith et al., 1982a, 1983). The study included 82 living or deceased male cases with histo-pathologically verified soft-tissue sarcoma reported to the national cancer registry between 1976 and 1980. One control per case was selected randomly from other cancer patients in the national cancer registry matched for age, gender and year of registration. The participation rate was 84% among cases (82/98 eligible) and 83% among controls (92/111 eligible). Interviews were conducted with the subjects or the next-of-kin (43% of cases, 34% of controls). Subjects were interviewed by telephone concerning work in particular occupations with potential exposure to phenoxy herbicides or chlorophenols. Exposure to phenoxy herbicides and chlorophenols was ascertained through a combination of information on occupation, industry, type of cultivation sprayed and self-reported exposure to specific agents. In a separate study, PCDD levels were determined in serum of nine workers first employed before 1960 and having sprayed 2,4,5-T for a minimum of 180 months (Smith et al., 1992a). The mean value of 2,3,7,8-TCDD was 53.3 ng/kg (range, 3–131 ng/kg) (see Table 17). Any potential exposure to a phenoxy herbicide was associated with a small non-statistically significant excess risk (odds ratio, 1.3; 90% CI, 0.7–2.5). A slightly higher risk was observed in subjects with more than one day of probable or definite exposure, after excluding the last five years of exposure before registration (odds ratio, 1.6; 90% CI, 0.7–3.3). Potential exposure to chlorophenols was associated with a small increased risk (odds ratio, 1.3; 90% CI, 0.5–3.6), particularly for those subjects who were potentially exposed for more than five days (odds ratio, 1.6; 90% CI, 0.5–5.2). However, a review of the working histories of those subjects through additional interviews and contacts with their employers indicated that only two cases, out of the seven characterized as potentially exposed, actually had potential for exposure to TCP.

An update was conducted by Smith and Pearce (1986) combining new data, including a general population control group, with the study population of Smith et al. (1984b). New cases diagnosed from 1980 to 1982 (n = 51) were compared with population controls (n = 315). The odds ratio estimate was 0.7 (90% CI, 0.3–1.5) for probable or definite exposure to phenoxy herbicides for more than one day not in the five years before cancer registration. The odds ratio estimate for the combined studies was 1.1 (90% CI, 0.7–1.8). There was no evidence in either study of an increase in risk with longer duration of exposure or longer latency since first exposure.

A population-based case–control study was conducted in three provinces of northern Italy where rice growing is the predominant agricultural activity and phenoxy herbicides have been used since 1950 (Vineis et al., 1986). During 1950–55, rice weeding was still done manually, mostly by a seasonal work force of women, who were presumed to have had the highest exposure to phenoxy herbicides, predominantly 2,4,5-T. Incident cases of soft-tissue sarcoma aged 20 years or older, with a proven or suspected histological diagnosis in 1981–83 were identified through all the pathology departments of the three provinces, and six pathology departments of the city of Turin and the National Cancer Institute in Milan. Out of 135 cases initially identified, 37 were excluded because the diagnosis was not confirmed or because the cases were prevalent or were visceral sarcomas. Of the remaining 98 eligible cases, interviews were obtained from 68 patients or next-of-kin (44 living, 24 deceased). The participation rate was 69%. A random sample of living controls was drawn from electoral registers. Deceased controls (any cause of death apart from suicide) matched by age, sex, year of death and municipality were selected through demographic offices of the relevant municipalities. Interviews were obtained from 122/168 eligible controls (participation rate 73%) and 36/40 relatives of deceased subjects (participation rate 90%). Direct interviews were carried out with subjects, who were asked for information on lifetime occupational history, and additional information on jobs with titles indicating potential exposure to phenoxy herbicides or chlorophenols. In addition, supplementary questionnaires were administered to subjects working in agriculture, particularly cultivation of rice and other crops. Two experts assessed exposure to phenoxy herbicides and classified subjects into three categories: (1) those not exposed to phenoxy herbicides; (2) those for whom exposure could not be ruled out; and (3) those ‘definitely’ exposed, based on occupational histories primarily as rice weeders. Among living subjects, no men were ‘definitely’ exposed. Among living women, four cases and five controls were ‘definitely’ exposed (odds ratio, 2.7; 90% CI, 0.6–12.4). The odds ratio for the combined categories 2 and 3 was 2.4 (5 exposed cases; 90% CI, 0.6–10.3). Among living women continuously exposed during 1950–55, the crude odds ratio was [9.9] (3 exposed cases, 1 exposed control). No excess risk was observed when analyses were limited to deceased subjects. [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

A population-based case–control study was conducted in western Washington State, United States, to evaluate the relationship between occupational exposure to phenoxy herbicides and chlorophenols and risk for soft-tissue sarcoma and non-Hodgkin lymphoma (Woods et al., 1987). Living and deceased cases, aged 20–79 years and diagnosed during 1981–84, were identified from the population-based tumour registry that covers 13 counties of the state. Enrolled in the study were 128 pathologically reviewed soft-tissue sarcoma cases (97 alive, 31 dead). Population controls were selected using random digit dialling or from social security records (for older cases) or non-cancer death certificates (for deceased cases). Controls were matched for age and vital status with a parallel series of non-Hodgkin lymphoma cases. The participation rate was 62% for soft-tissue sarcoma cases (128/206) and 76% for controls. Subjects were interviewed in person and information was requested on occupational exposures to phenoxy herbicides and chlorophenols and to other potential risk factors. Additional information was requested for specific occupations and job activities involving potential exposure to the chemicals of interest. The risk for soft-tissue sarcoma associated with all occupations involving potential exposure to phenoxy herbicides was 0.8 (95% CI, 0.5–1.2). Exposure to phenoxy herbicides or chlorophenols was not associated with increased risk for soft-tissue sarcoma. No association was seen with estimated intensity of exposure to either phenoxy herbicides or chlorophenols. [The Working Group noted that no information was provided on use of specific herbicides.]

A case–control study on patients with soft-tissue sarcoma was undertaken in Victoria, Australia, during 1982–88 (Smith & Christophers, 1992). The study included 30 male cases with soft-tissue sarcoma registered in the Victoria Cancer Registry. Cases were first diagnosed between 1976 and 1987. For each case, one cancer control was selected from the cancer registry, matched for age, sex and residence. A second set of population controls was selected from the electoral register using the same matching criteria. The response rates were 70% for cases, 56% for cancer controls and 70% for population controls. Cases were interviewed mostly before 1986, while both series of controls were mostly interviewed after 1986. A comprehensive occupational history was obtained through a personal interview. Exposures to phenoxy herbicides and to chlorophenols were coded by expert assessment as none, possible or definite/probable. Exposures within the five-year period before the year of diagnosis of a case were ignored, for both cases and their matched controls. The main chlorinated herbicides used in Victoria were 2,4-D, 2,4,5-T and MCPA. There were no significant differences between population and cancer controls with respect to definite exposure and these two groups were combined for the analysis. For soft-tissue sarcoma, the odds ratio was 1.0 (95% CI, 0.3–3.1) for at least one day's exposure to phenoxy herbicides or chlorophenols and 1.3 (95% CI, 0.4–4.1) for exposure only to phenoxy herbicides. The odds ratio for more than 30 days of exposure to phenoxy herbicides or chlorophenols was 2.0 (95% CI, 0.5–8.0). [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

(b) Malignant lymphomas (see Table 36)

Table 36

Case–control studies on malignant lymphoma containing information on exposure to phenoxy herbicides, chlo-rophenols or dioxins.

A case–control study followed the report in Sweden of 17 cases of histiocytic malignant lymphoma, 11 of whom had been exposed to phenoxy herbicides or chlorophenols (Hardell, 1979). All men aged 25–85 years admitted between 1974 and 1978 to the oncology department of the University Hospital in Umeå, Sweden, with histologically verified malignant lymphoma were included in the study (Hardell et al., 1981). For every living case (in total 107), eight controls matched for sex, age and place of residence were extracted from the national population registry and the two closest in age were used in the analysis. For the 62 deceased cases, 10 controls per case were selected from the national registry of causes of death among those who had died from causes other than malignant tumour, matched for sex, age, municipality and year of death. The two deceased controls closest in age to the cases were used in the analysis. Exposures were reconstructed by means of a self-administered questionnaire. The data concerning deceased subjects were obtained by contact with next-of-kin. For exposure to phenoxy herbicides or chlorophenols, an odds ratio of 6.0 (95% CI, 3.7–9.7) was obtained. For phenoxy herbicides alone (41 exposed cases), the odds ratios were 4.8 (95% CI, 2.9–8.1) and 7.0 for those exposed for more than 90 days. Five cases and no controls were exposed only to MCPA; 7 cases and 1 control only to 2,4-D. These herbicides were not likely to be contaminated with PCDDs or PCDFs. Fifty cases were exposed to chlorophenols, giving an odds ratio of 2.9 (95% CI, 1.6–5,2) for those with low-level exposure and 8.4 (95% CI, 4.2–16.9) for high-level exposure. High-level exposure to organic solvents was also associated with a significantly increased risk. Combined exposure to solvents and phenoxy herbicides or chlorophenols further increased the odds ratio estimate. Separate analysis for Hodgkin's disease (60 cases) and non-Hodgkin lymphoma (109 cases) yielded similarly increased risks. [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

A population-based case–control study on lymphoma was conducted in Kansas, a major wheat-producing state in the United States (Hoar et al., 1986). All newly diagnosed cases of Hodgkin's disease and non-Hodgkin lymphoma among white male residents in Kansas aged 21 years or older, diagnosed from 1976 through 1982, were identified through the population-based cancer registry covering the state of Kansas. All cases included in the study were histologically confirmed. Interviews were conducted with 121 cases with Hodgkin's disease and 170 with non-Hodgkin lymphoma (96% participation rate). One-half of the patients with non-Hodgkin lymphoma and one-third of those with Hodgkin's disease had died before the study began. Controls were 948 white men resident in Kansas matched to cases for age and vital status. The response rate for controls was 94%. Subjects were interviewed by telephone and detailed information was obtained concerning farming practices, including years lived or worked in farms, crops grown or livestock raised, herbicides and insecticides used, numbers of years and acres of treatment and protective equipment used. Corroborative evidence was sought for a sample of 130 subjects by contacting their suppliers. Farm herbicide use was associated with an increased risk for non-Hodgkin lymphoma (odds ratio, 1.6; 95% CI, 0.9–2.6). The relative risk for non-Hodgkin lymphoma increased significantly with the number of days of herbicide exposure per year and with latency. Men exposed to herbicides for more than 20 days per year had a six-fold increased risk for non-Hodgkin lymphoma (odds ratio, 6.0; 95% CI, 1.9–19.5). Frequent users (> 20 years per year) who mixed or applied the herbicides themselves had an odds ratio of 8.0 (95% CI, 2.3–27.9). Use of phenoxy herbicides (odds ratio, 2.2; 95% CI, 1.2–4.1) essentially indicated use of 2,4-D, since only three cases and 18 controls had used 2,4,5-T, and all but two of these controls had also used 2,4-D. Farm herbicide use was not associated with risk for Hodgkin's disease (odds ratio, 0.9; 95% CI, 0.5–1.5). For this neoplasm, no consistent pattern of excess risk was observed either with duration or frequency of exposure. The authors noted that the observed excess risk for non-Hodgkin lymphoma was associated with exposure to phenoxy herbicides (2,4-D) which were not contaminated with 2,3,7,8-TCDD, although they might be contaminated with other PCDD congeners.

A population-based case–control study from western Washington State (Woods et al., 1987), described above, included 576 non-Hodgkin lymphoma cases (402 alive, 174 dead). Population controls were selected using random digit dialling or from social security records (for older cases) or non-cancer death certificates (for deceased cases). Controls were frequency matched for age and vital status with the non-Hodgkin lymphoma cases. The participation rate was 77% for non-Hodgkin lymphoma cases and 76% for controls. The risk for non-Hodgkin lymphoma associated with all occupations involving potential exposure to phenoxy herbicides was 1.1 (95% CI, 0.8–1.4). Exposure to phenoxy herbicides or chlorophenols was not associated with increased risk for non-Hodgkin lymphoma. No association was seen with estimated intensity of exposure to either phenoxy herbicides or chlorophenols. A statistically significant excess risk for non-Hodgkin lymphoma was observed for specific occupational groups, including farmers (odds ratio, 1.3; 95% CI, 1.0–1.7) and subjects spraying forests with herbicides (odds ratio, 4.8; 95% CI, 1.2–19.4). All forestry sprayers reported combined use of 2,4-D and 2,4,5-T. [The Working Group noted that no information was provided on use of specific herbicides.]

Following the same methodology, a population-based case–control study was conducted in western Washington State on non-Hodgkin lymphoma and phenoxy herbicide exposure in farmers (Woods & Polissar, 1989). Cases of non-Hodgkin lymphoma occurring between 1983 and 1985 were identified from the cancer surveillance system. A total of 694 control men without cancer selected randomly from the same geographical area were matched with the 576 cases by five-year age group and vital status. Among them, 181 non-Hodgkin lymphoma cases and 196 controls reported having worked as farmers. Information on exposure to phenoxy herbicides was obtained through personal interviews. Regular use was defined as ‘more than just once or twice per year’. Mantel-Haenszel odds ratios were calculated for farmers, for all subjects for 2,4,5-T and for ‘phenoxy per se’ exposures, among others. No excess risks were observed in all subjects nor in the group of farmers for the above exposures: for 2,4,5-T, the odds ratios were 0.7 (95% CI, 0.3–2.1) in farmers and 1.0 (95% CI, 0.4–2.0) for all subjects; for ‘phenoxy per se’ the odds ratios were 0.7 (95% CI, 0.3–1.5) and 0.9 (95% CI, 0.5–1.5), respectively.

Pearce et al. (1987) expanded a previously reported case–control study on malignant lymphomas and farming in New Zealand (Pearce et al., 1986). The study included male public hospital patients registered under ICD codes 200 or 202 during the period 1977–81, who were under the age of 70 years. All cases were pathologically verified. Out of a total of 215 eligible cases, 183 cases were enrolled in the study (participation rate 85%). For each of the cases, two cancer controls were randomly selected from the cancer registry files, matched for age and year of cancer registration. In total, 338 controls were enrolled (81% participation rate). Interviews were conducted by telephone with the patients or their next-of-kin [no information was given on the proportion of interviews with next-of-kin]. Information was requested concerning work in particular occupations with potential exposure to phenoxy herbicides or chlorophenols. If the response to a stem question was affirmative, then a series of subsidiary questions were asked to clarify the work done and the actual potential for exposure to specific chemicals. Exposure to phenoxy herbicides and chlorophenols was ascertained through a combination of information on occupation, industry, type of cultivation sprayed and self-reported exposure to specific agents. The proportions of cases and controls who had worked in the occupations examined in this study were very similar. The odds ratio for any potential exposure to phenoxy herbicides was 1.0 (95% CI, 0.7–1.5) and none of the odds ratios relating to specific phenoxy herbicides was elevated. The odds ratio for any potential exposure to chlorophenols was slightly elevated (odds ratio, 1.4; 95% CI, 0.8–2.3), largely due to the results for meat workers, which included workers potentially exposed to 2,4,6-TCP in pelt departments. In a re-analysis of the data (Pearce, 1989), little evidence was found of an association of non-Hodgkin lymphoma either with duration or with frequency of phenoxy herbicide use. Findings were similar in the earlier publication using general population controls (Pearce et al., 1986). [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

Olsson and Brandt (1988) examined the lifetime work history of 167 incident cases of non-Hodgkin lymphoma diagnosed at a hospital in Lund (Sweden) between 1978 and 1981 in adult men and in 130 age-matched population controls, 50 from the same geographical area as the cases and 80 from different parts of Sweden. Exposure to organic solvents, phenoxy herbicides and chlorophenols was given special attention. Interviewers were not blind to subject status (case or control). One day of handling was taken to constitute exposure to phenoxy acids or chlorophenols. Odds ratios were estimated using logistic models, adjusting for age. The roles of different exposures, of their interaction and of length of exposure were examined using logistic regression analysis. In the multivariate analysis, the odds ratio for exposure to phenoxy herbicides was 1.3 (95% CI, 0.8–2.1) and that for chlorophenols was 1.2 (95% CI, 0.7–2.0). Separate analysis for localized ‘skin lymphoma’ yielded a significantly increased risk (odds ratio, 10.0; 95% CI, 2.7–37.1) for exposure to phenoxy herbicides. No interaction between exposures was detected. Risk for non-Hodgkin lymphoma was significantly associated with length of exposure to solvents but not to phenoxy herbicides and chlorophenols. [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

Concurrent population-based case–control interview studies of leukaemia, non-Hodgkin lymphoma and multiple myeloma in Iowa, United States, and leukaemia and non-Hodgkin lymphoma in Minnesota, United States, were conducted during 1981–84. The studies used the same questionnaire and the same controls. Results from the case–control study on non-Hodgkin lymphoma were reported by Cantor et al. (1992). Results for the studies on leukaemia (Morris-Brown et al., 1990) and multiple myeloma (Morris-Brown et al., 1993) are described in Section 2.2.2(c). All male cases of non-Hodgkin lymphoma aged 30 years or older and newly diagnosed during 1980–82 were ascertained from Iowa State Health Registry records and a special surveillance of Minnesota hospitals and pathology laboratory records. Residents of Iowa and of selected areas of Minnesota were eligible. A review panel of four pathologists confirmed diagnoses. Out of 780 reported non-Hodgkin lymphoma cases, 622 (438 living, 184 deceased) were confirmed as non-Hodgkin lymphoma and interviewed. The participation rate was slightly above 80% [exact proportion not estimable]. A population control group of 1245 white men without haematopoietic or lymphatic cancer was randomly selected and frequency matched with the non-Hodgkin lymphoma and leukaemia cases by five-year age group, vital status at time of interview and state of residence. Random digit dialling was used to select controls for living cases under 65 years of age. A 1% random listing of Medicare files was used for the selection of controls for living cases over 65 years of age. State death certificates were used for deceased cases. The participation rate was around 77% in all three groups of controls. Direct structured interviews were conducted during 1981–84. Detailed information on farming and pesticide use was requested for all subjects who had worked on a farm for at least six months since the age of 18 years. The information recorded included years of farming activity, total acreage, crops grown and detailed history of pesticide use. There was a small increase in risk (odds ratio, 1.2; 95% CI, 1.0–1.5) associated with ever living or working on a farm. The odds ratio for use of one or more herbicides was 1.3 (95% CI, 1.0–1.6), but no single family of herbicides was significantly associated with risk for non-Hodgkin lymphoma. No significant risk elevations were observed for ever handling, mixing or applying specific phenoxy herbicides. The odds ratio for 2,4-D was 1.2 (95%, 0.9–1.6) and that for 2,4,5-T was 1.2 (95% CI, 0.7–1.9). The risk for 2,4,5-T was slightly higher among farmers who did not use protective gear (odds ratio, 1.4; 95% CI, 0.7–2.5). The odds ratios for handling these herbicides before 1965 (assuming a latency of about 15 years) were 1.3 (95% CI, 0.9–1.8) for 2,4-D and 1.7 (95% CI, 0.8–3.6) for 2,4,5-T. The authors stated that there was minimal confounding of results for any single pesticide by exposure to pesticides belonging to other chemical families.

The case–control study conducted in Victoria, Australia, described in Section 2.2.2(a) (Smith & Christophers, 1992), also included 52 male cases with malignant lymphoma. The odds ratios for malignant lymphoma were 1.5 (95% CI, 0.6–3.7) for at least one day's exposure to phenoxy herbicides or chlorophenols, 1.1 (95% CI, 0.4–3.0) for exposure only to phenoxy herbicides and 1.4 (95% CI, 0.3–6.1) for exposure only to chlorophenols. The odds ratio for more than 30 days of exposure to phenoxy herbicides or chlorophenols was 2.7 (95% CI, 0.7–9.6). [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

(c) Other haematopoietic malignancies (see Table 37)

Table 37

Case–control studies on other tumour sites containing information on exposure to phenoxy herbicides, chlorophenols or dioxins.

A case–control study on leukaemia was conducted in Iowa and Minnesota, United States (Morris-Brown et al., 1990) using the same methodology and same control group as that described in Section 2.2.2(b) (Cantor et al., 1992). All newly diagnosed cases of leukaemia among white men aged 30 years or older were ascertained from tumour registry or hospital records retrospectively (one year before the start of the study) or prospectively (two years after the start of the study). Interviews were completed with 86% of the cases (or next-of-kin). The final study population consisted of 578 cases (340 living, 238 deceased) and 1245 population controls. Apart from the main interview (described above), a supplementary interview was conducted including 92 cases and 211 controls (or their next-of-kin) from Iowa who had reported agricultural use of pesticides in their initial interview. There was a small risk for all leukaemias among persons who had lived or worked on a farm as an adult (odds ratio, 1.2; 95% CI, 1.0–1.5). A similar risk was seen for farmers reporting ever having used herbicides (odds ratio, 1.2; 95% CI, 0.9–1.6) or phenoxy herbicides (odds ratio, 1.2; 95% CI, 0.9–1.6). Risks for all leukaemias were not significantly increased among subjects who personally mixed, handled or applied specific herbicides. When analyses were restricted to persons first exposed to specific herbicides more than 20 years before interview, increased risks were observed for exposure to MCPA (odds ratio, 2.4; 95% CI, 0.7–8.2) and 2,4,5-T (odds ratio, 1.8; 95% CI, 0.8–4.0). Among specific cell types, the highest risk for those who handled 2,4-D was seen for chronic myelogenous leukaemia (odds ratio, 1.9; 95% CI, 0.9–3.9). The odds ratios for those who handled 2,4,5-T were 2.1 (95% CI, 0.9–4.9) for acute non-lymphocytic leukaemia and 1.6 (95% CI, 0.7–3.4) for chronic lymphocytic leukaemia. The risk for those who handled 2,4,5-T at least 20 years before interview was significantly elevated for chronic lymphocytic leukaemia (odds ratio, 3.3; 95% CI, 1.2– 8.9). No consistent dose-response pattern in terms of days of handling per year was seen for any of the herbicides used.

A case–control study on multiple myeloma was conducted in Iowa, United States (Morris-Brown et al., 1993), using the same methodology and same control group as that described above in Section 2.2.2(b) (Cantor et al., 1992). All cases of multiple myeloma among adult white men aged 30 years or older and diagnosed during 1981–84 were identified from the Iowa State Health Registry. Pathological material and laboratory reports were reviewed by an expert pathologist. Included the study were 173 cases (101 alive, 72 deceased) and 650 controls (452 alive, 198 deceased). Interviews were completed for 84% of multiple myeloma cases and 78% of controls. Some farming activity was reported by 64% of the cases and 58% of the controls (odds ratio, 1.2; 95% CI, 0.8–1.7). Risks were not elevated for subjects who handled the phenoxy herbicides 2,4-D (odds ratio, 1.0; 95% CI, 0.6–1.6) or 2,4,5-T (odds ratio, 0.9; 95% CI, 0.4–2.1).

(d) Other solid tumours (Table 37)

Hardell (1981) conducted a case–control study on colon cancer patients following the same methodology as that described earlier (see Section 2.2.2(a); Hardell & Sändstrom, 1979; Erickson et al., 1981; Hardell et al., 1981). Cases were 157 men with colon cancer (response rate, 98.1%) aged 25–85 years who were residents of the region of Umeå, Sweden, and who had been reported to the Swedish Cancer Registry in 1978–79. All had a histopathological diagnosis of adenocarcinoma. Sixty-five cases (41%) were deceased. The 541 controls were derived from two earlier studies (Hardell & Sändstrom, 1979; Hardell et al., 1981). A low excess risk was observed for exposure to phenoxy herbicides (odds ratio, 1.3; 95% CI, 0.6–2.8) and for exposure to chlorophenols (odds ratio, 1.8; 95% CI, 0.6–5.3). [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

In a case–control study on nasal and nasopharyngeal cancer in the region of Umeå, Sweden, Hardell et al. (1982) followed the same methodology as that described above (Hardell & Sändstrom, 1979; Hardell et al., 1981). The cases comprised all male patients aged 25–85 years with histopathologically confirmed nasopharyngeal cancer (n = 27) and cancer of the nose and nasal sinuses (n = 44). All cases had been reported to the Swedish Cancer Registry in 1970–79 and were residents in the three most northern counties of Sweden. The 541 controls were derived from two earlier studies in the same region (Hardell & Sändstrom, 1979; Hardell et al., 1981). Fifty (70.4%) of the cases were deceased, compared with 245 (45.3%) of the controls. [No information was provided on response rates.] Exposure to phenoxy herbicides was associated with a two-fold risk for the combination of nasopharyngeal and nasal cavity cancer (odds ratio, 2.1; 95% CI, 0.9–4.7). A high risk was found for high-level exposure to chlorophenol, defined as a cumulative exposure of one week of continuous or one month of discontinuous exposure (odds ratio, 6.7; 95% CI, 2.8–16.2). No obvious difference was reported between cases and controls for low-level exposure to chlorophenols. The risk associated with use of chlorophenol exposure was higher among wood workers (odds ratio, 8.4) than in other occupations (odds ratio, 2.7) but, in a stratified analysis, it was shown that occupation as a wood worker was not a confounding factor for exposure to chlorophenols. [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

Men aged 25–80 who had been diagnosed with liver cancer between 1974 and 1981 and reported to the Department of Oncology, Umeå, Sweden, were included in another case–control study (Hardell et al., 1984). Microscope slides were reviewed for the 166 assembled cases, and 103 cases of primary liver cancer were retained for the study; 206 population-based controls were matched with cases for age and residence. Information on exposure was obtained as in previous studies (see Hardell & Sandström, 1979); responses were obtained for 102 cases and 200 controls. The analyses were restricted to the 98 cases of hepatocellular and cholangiocellular carcinoma. Odds ratios, without controlling for other agricultural exposures, were 1.8 (95% CI, 0.9–4.0) for exposure to phenoxy herbicides and chlorophenols, 1.7 (95% CI, 0.7–4.4) for exposure to phenoxy herbicides only and 2.2 (95% CI, 0.7–7.3) for high-level exposure to chlorophenols only. [The Working Group noted that no distinction was made between herbicides likely to be contaminated with PCDDs and others.]

A case–control study conducted in two hospitals in Hanoi (Viet Nam) between 1989 and 1992 included 152 male cases of hepatocellular carcinoma and 241 hospital controls, admitted mainly in abdominal surgery departments, matched for age, sex and area of residence at the time of admission (Cordier et al., 1993). Exposure to Agent Orange may have occurred during stays in the south of Viet Nam after 1960 (date of the beginning of spraying missions) especially for military purposes. Duration of military service in the south, for North Vietnamese soldiers, was considered as a proxy for exposure to Agent Orange. This is justified by the fact that sprayings were principally aimed at uncovering trails used by the Vietnamese combatants, who stayed one or two years on average in these areas and consumed locally grown (and contaminated) foodstuffs. The overall odds ratio associated with military service in the south was 1.3 (95% CI, 0.8–2.1). The risk rose with increasing duration of stay in the south, reaching 8.8 (95% CI, 1.9–41) after 10 years. Odds ratios were adjusted for matching variables (hospital, age, area of residence), hepatitis B virus surface antigen, anti-hepatitis C virus status and alcohol consumption.

3.1.1. Mouse

Groups of 45 male outbred Swiss/H/Riop mice, 10 weeks of age, were given gastric instillations of 0.007, 0.7 or 7.0 µg/kg bw 2,3,7,8-TCDD [purity unspecified] dissolved in sunflower oil once a week for one year. A control group received vehicle only. Animals were followed for the rest of their life span. Survival was significantly decreased in the high-dose animals (average life span, 424 days compared with 588 days in controls). It was reported that all organs [unspecified] were examined histologically. Treatment with 2,3,7,8-TCDD caused severe chronic, ulcerous skin lesions, followed by generalized lethal amyloidosis and was associated with an increased incidence of liver tumours (hepatocellular adenomas and carcinomas combined) [the two histological types were not reported separately]: control, 7/38; low-dose, 13/44; mid-dose, 21/44 (p > 0.01, χ² test) and high-dose, 13/43. No statistically significant increase in the incidence of lung tumours or lymphomas was observed (Töth et al., 1979). [The Working Group noted that mortality-adjusted analysis was not performed and, therefore, the tumour incidence in the high-dose group may be underestimated.]

Groups of 50 male and 50 female B6C3F1 mice, six weeks of age, were given gastric instillations of 0.01, 0.05 or 0.5 µg/kg bw (males) and 0.04, 0.2 or 2.0 µg/kg bw (females) 2,3,7,8-TCDD (purity, 99.4%) in a vehicle of 9 : 1 corn oil-acetone twice a week for 104 weeks. One control group of 75 males and 75 females received vehicle alone and another group of 50 males and 50 females was untreated. Mean body weights of the treated groups were comparable with those of the vehicle-control group. Treatment did not affect survival: 30/50 untreated control, 38/74 vehicle-control, 30/50 low-dose, 31/50 mid-dose and 31/50 high-dose males were still alive at 105–107 weeks and 34/50 untreated control, 58/75 vehicle-control, 37/50 low-dose, 36/50 mid-dose and 32/50 high-dose females were still alive at 106–107 weeks. Treatment-related increases in hepatotoxicity were found in high-dose animals of both sexes. The incidences of hepatocellular carcinoma were dose-related, with significantly higher incidence in the high-dose groups (males, p = 0.002; females, p = 0.014, Fisher's exact test) than in the vehicle-control groups (males: vehicle-control, 8/73; low-dose, 9/49; mid-dose, 8/49; and high-dose, 17/50; females: vehicle-control, 1/73; low-dose, 2/50; mid-dose, 2/48; and high-dose, 6/47; males, p = 0.002; females, p = 0.008 Cochran-Armitage test for trend). Dose-related increases in the incidence of follicular-cell adenomas of the thyroid were observed in female mice, with significantly higher incidence (p = 0.009) in the high-dose group than in the vehicle-control group (vehicle-control, 0/69; low-dose, 3/50; mid-dose, 1/47; and high-dose, 5/46; p = 0.016 for trend). In female mice, there was a significant increase in the incidence of lymphomas at the high dose (18/74 controls, 11/50 low-dose, 13/48 mid-dose and 20/47 high-dose; p = 0.029). There was also a significant increase in the incidence of subcutaneous fibrosarcomas in high-dose females (1/74 controls, 1/50 low-dose, 1/48 mid-dose and 5/47 high-dose; p = 0.032). A dose-related increase (p = 0.04, Cochran-Armitage test for trend) in lung tumours (alveolar/bronchiolar adenomas or carcinomas) was observed in male mice (10/71 controls, 2/48 low-dose, 4/48 mid-dose and 13/50 high-dose) (United States National Toxicology Program, 1982a).

Groups of 45–55 male and female (C57BL/6J × C3Hf)F1 mice, six weeks of age, were given gastric instillations of 0, 2.5 or 5.0 µg/kg bw 2,3,7,8-TCDD [laboratory grade; purity unspecified] in 0.01 mL/kg bw corn oil vehicle (containing 1.2% acetone) once a week for 52 weeks. At 31–39 weeks of age, 41 males and 32 females in the 2.5-µg/kg bw group were erroneously treated once with a dose of 25 µg/kg bw 2,3,7,8-TCDD. The treatment of these mice was interrupted for five weeks and then continued until week 57, as for the other treated mice. At the end of treatment, all groups were kept under observation until 110 weeks of age, when all survivors were killed. Histopathological examination was carried out on Harderian glands, pituitary, thyroid, tongue, oesophagus, trachea, lungs, liver, pancreas, mesenteric lymph nodes, small intestine, spleen, kidney, adrenal glands, testis or ovaries, uterus, urinary bladder and all other organs with apparent or suspected pathological alterations. Treatment with 2,3,7,8-TCDD at both dose levels caused a marked depression in mean body weight in both males and females. Survival was significantly reduced in male and female 2,3,7,8-TCDD-treated mice (p > 0.001). Long-term administration of 2,3,7,8-TCDD increased the incidence of some non-neoplastic lesions, including liver necrosis, amyloidosis of multiple tissues and nephrosclerosis. Dermatitis developed in most 2,3,7,8-TCDD-treated mice and regressed after cessation of treatment. Tumour incidence was evaluated statistically adjusting for intercurrent mortality. [The authors report that, due to the toxicity of the treatment, it was not possible to distinguish between fatal and incidental tumours and both statistical methods were used.] Hepatocellular adenomas occurred in 10/45 control, 11/51 low-dose and 10/50 high-dose males (p > 0.05, incidental tumour test). However, when fatal tumours were considered, there was a significant increase (p < 0.001, fatal tumour test). In females, the incidence of hepatocellular adenomas was 2/49 control, 4/42 low-dose and 11/48 high-dose animals (p < 0.01, fatal tumour test; p < 0.001, incidental tumour test). The incidences of hepatocellular carcinomas were: 5/43 control, 15/51 low-dose and 33/50 high-dose males (p > 0.001 for both fatal and incidental tumour tests); and 1/49 control, 12/42 low-dose and 9/48 high-dose females (p > 0.01, fatal tumour test; p > 0.05, incidental tumour test). The incidence of any other tumour type was not associated with treatment (Della Porta et al., 1987).

3.1.2. Rat

Groups of 10 male Sprague-Dawley rats, weighing approximately 60 g, were fed diets containing 1, 5, 50, 500 ng/kg diet (parts per trillion; ppt), 1, 5, 50, 500 or 1000 µg/kg diet (parts per billion; ppb) 2,3,7,8-TCDD [purity unspecified]. One control group was maintained on basal diet. Animals were maintained on the diets for 78 weeks, at which time the treated animals were changed to the basal diet. At 65 weeks of the experiment, laparotomies were performed on all surviving animals and at 95 weeks all surviving animals were killed. All animals treated with the three highest doses (50, 500 and 1000 ppb) died between the second and fourth week of the experiment. At the end of the experiment, four, eight, six, six and five, zero and zero animals were alive in the control, 1-, 5-, 50- and 500-ppt and 1- and 5-ppb groups, respectively. Various types of neoplasm were observed in the treated rats and none in the control rats. The numbers of animals with neoplasms were zero, zero, five, three, four, four and seven in the control, 1-, 5-, 50-and 500-ppt and 1- and 5-ppb groups, respectively. In the 5-ppb group, four squamous-cell tumours of the lung, four neoplastic nodules of the liver and two cholangiocarcinomas of the liver were observed. One cholangiocarcinoma was observed in the 1-ppb group (Van Miller et al., 1977). [The Working Group noted the small number of animals per group, but that cholangiocarcinomas are relatively rare in rats and may be related to treatment.]

Groups of 50 male and 50 female Sprague-Dawley rats, six to seven weeks of age, were fed 22, 210 or 2200 ppt (ng/kg) 2,3,7,8-TCDD (purity, > 99%) in the diet, corresponding to 0.001, 0.01 or 0.1 µg/kg bw daily for two years. Control groups of 86 male and 86 female rats received a basal diet containing the vehicle (acetone) only. Terminal necropsy was performed at 105 weeks. Representative portions of most organs and any additional gross lesions were preserved in formalin fixative. Histological examination was conducted on an extensive list of tissues from control and high-dose rats. All rats in the low- and mid-dose groups were subjected to histological examination of tissues identified as possible target organs and all gross lesions. Reduced survival was observed in high-dose females and in mid- and high-dose males. Mean body weights of high-dose males and mid- and high-dose females were below those of the control animals throughout the major portion of the study [details not reported]. Non-neoplastic, treatment-related pathological changes were reported, especially in the liver. High- and mid-dose rats had multiple hepatocellular necrosis and inflammatory changes. In males, the incidences of hepatocellular hyperplastic nodules (6/85 controls, 0/50 low-dose, 3/50 mid-dose and 2/50 high-dose) and hepatocellular carcinomas (2/85 controls, 0/50 low-dose, 0/50 mid-dose and 1/50 high-dose) were not increased. In female rats, the incidence of hepatocellular hyperplastic nodules was 8/86 control, 3/50 low-dose, 18/50 mid-dose and 23/49 high-dose animals (p > 0.05, Fisher's exact test) and that of hepatocellular carcinomas was 1/86 control, 0/50 low-dose, 2/50 mid-dose and 11/49 high-dose animals (p > 0.05). The lipid concentrations of 2,3,7,8-TCDD in the low-, mid- and high-dose rats were 540, 1700 and 8100 ng/kg. Squamous-cell carcinomas of the hard palate or nasal turbinates occurred in 4/49 (p > 0.05) high-dose and 1/50 mid-dose females, squamous-cell carcinomas of the lung were observed in 7/49 (p > 0.05) high-dose females and squamous-cell carcinomas of the tongue occurred in 2/49 high-dose females; tumours at these sites were not observed in the other groups. In high-dose males, 4/50 rats developed squamous-cell carcinomas of the hard palate or nasal turbinates (p > 0.05), one developed a squamous-cell carcinoma of the lung and three squamous-cell carcinomas of the tongue (p > 0.05). In comparison with the high incidence of endocrine-related tumours (pituitary adenomas, phaeochromocytomas and pancreatic islet-cell tumours) in controls, reduced incidences were observed in the high-dose group (Kociba et al., 1978).

A re-evaluation of the slides of liver specimens from the female animals studied by Kociba et al. (1978) was performed by a panel of pathologists (Keenan et al., 1991). In contrast to the original findings, they found about two-thirds fewer tumours present in the livers of female Sprague-Dawley rats. On the basis of their results, they established a no-observed-adverse-effect level for hepatocellular carcinomas of 0.01 µg/kg bw per day 2,3,7,8-TCDD. [The Working Group noted that the results of this re-evaluation do not change substantially the positive findings of Kociba et al. (1978) on the liver carcinogenicity of 2,3,7,8-TCDD in female rats.]

Liver slides from the study by Kociba et al. (1978) were further reviewed by a pathology working group with the aim of evaluating proliferative lesions in the livers of female rats. The results of the working group substantially confirmed the previous results on dose-related increased tumour incidence (hepatocellular adenomas: 2/81 control, 1/50 low-dose, 9/50 mide-dose and 14/45 high-dose; hepatocellular carcinomas: 0/86, 0/50, 0/50 and 4/45). There was a dose-related increase in the incidence of hepatic eosinophilic foci (31/86 control, 23/50 low-dose, 37/50 mid-dose and 40/45 high-dose) (Goodman & Sauer, 1992).

Groups of 50 male and 50 female Osborne-Mendel rats, six weeks of age, were given gastric instillations of 0.01, 0.05 or 0.5 µg/kg bw 2,3,7,8-TCDD (purity, > 99.4%) in a vehicle of 9 : 1 corn oil-acetone twice a week for 104 weeks. One control group of 75 male and 75 female rats received vehicle alone and another control group of 50 males and 50 females was untreated. Mean body weights of the high-dose groups were lower than those of the corresponding controls after week 55 in males and after week 45 in females. Treatment did not affect survival: 23/50 untreated control, 29/75 vehicle-control, 17/50 low-dose, 20/50 mid-dose and 19/50 high-dose males were alive at the end of the experiment at 105–108 weeks; 29/50 untreated control, 39/75 vehicle-treated control, 29/50 low-dose, 34/50 mid-dose and 32/50 high-dose females were alive at the end of the experiment at 105–107 weeks. Treatment-related increased hepatotoxicity was observed in high-dose males and females. Treatment-related increased incidences of follicular-cell adenomas of the thyroid were seen in males and were significantly higher (p = 0.001; Fisher's exact test) in the high-dose group than in the vehicle controls (1/69 vehicle-control, 5/48 low-dose, 6/50 mid-dose and 10/50 high-dose; p = 0.006, Cochran-Armitage test for trend) and in high-dose female rats (3/73 vehicle-control, 2/45 low-dose, 1/49 mid-dose and 6/47 high-dose; p = 0.022, Cochran-Armitage test for trend). The incidence of neoplastic nodules of the liver was significantly higher (p = 0.006) in high-dose females than in vehicle-control females (5/75 vehicle-control, 1/49 low-dose, 3/50 mid-dose and 12/49 high-dose). In males, a dose-related positive trend (p = 0.005) was seen (0/74, 0/50, 0/50 and 3/50, respectively) (United States National Toxicology Program, 1982a).

3.6.1. Skin

Mouse: Groups of 30 female CD1 mice, seven to nine weeks of age, received skin applications of 2 µg 2,3,7,8-TCDD or 2.56 µg 7,12-dimethylbenz[a]anthracene (DMBA) per animal or both chemicals together in 0.2 mL acetone solvent. Starting one week later, the mice received thrice weekly applications of 12-O-tetradecanoylphorbol-13-acetate (TPA) (5 jug/animal) in 0.2 mL acetone for 32 weeks. The 2,3,7,8-TCDD/TPA-treated mice had 0.1 papillomas/mouse (14% incidence), the 2,3,7,8-TCDD + DMBA/TPA-treated mice had 2.2 papillomas/mouse (63% incidence) and the DMBA/TPA-treated mice had approximately 1.8 papillomas/mouse (40% incidence) (DiGiovanni et al., 1977). [The Working Group noted that adequate control groups were not available, precluding an evaluation.]

Three groups of 30 female CD1 mice, six to eight weeks of age, received single skin applications of 0.2 µmol/animal (60 µg) DMBA in 0.2 mL acetone or solvent alone. Starting one week later, mice were treated with an acetone solution of 0.1 µg 2,3,7,8-TCDD or 2 µg TPA per animal (positive control) or acetone solvent alone twice weekly for 30 weeks, at which time the experiment was terminated. The percentages of mice with skin papillomas were DMBA/2,3,7,8-TCDD, 0%; DMBA/TPA, 92% and acetone/-TPA, 0% (Berry et al., 1978).

In groups of female HRS/J haired (hr/+) mice given single skin applications of DMBA (0.2 µmol in acetone) followed by twice weekly applications of 20 ng/animal 2,3,7,8-TCDD for eight weeks then 50 ng/animal for 17 weeks, no skin tumours were found. In contrast, in HRS/J hairless (hr/hr) mice treated with 0.2 µmol/animal DMBA and the same regimen of 2,3,7,8-TCDD (25 weeks), 15/19 surviving mice developed skin tumours (1.4 tumours per mouse). In hairless animals treated with DMBA alone, one skin papilloma was found and, in the absence of DMBA, 2,3,7,8-TCDD produced no skin tumour (Poland et al., 1982).

In a further experiment, groups of 20 female HRS/J hairless (hr/hr) mice, seven weeks of age, were given single skin applications of 5 µmol (0.75 mg)/animal N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) in 50 µL acetone followed by twice weekly doses of 0 or 3.75µ30 ng/animal 2,3,7,8-TCDD or 1 µg or 3 µg/animal TPA as positive controls. The numbers of mice with skin tumours at 20 weeks are shown in Table 42 (Poland et al., 1982).

Table 42

Induction of skin tumours in female HRS/J hairless mice.

Groups of 26 female HRS/J hairless (hr/hr) mice, eight weeks of age, were given single skin applications of 0 or 5 µmol/animal MNNG in 50 µL acetone followed by 50 ng/animal 2,3,7,8-TCDD for five weeks then 20 ng/animal for 15 weeks, both in 50 µL acetone twice weekly during the 20-week period. Skin tumours developed in 16/19 (1.6 tumours per mouse) mice treated with MNNG plus 2,3,7,8-TCDD compared with 0/18 in mice treated with 2,3,7,8-TCDD alone (Poland et al., 1982).

Three groups of 20 female HRS/J hairless (hr/hr) mice, five to eight weeks of age, were given single skin applications of 5 µmol/animal MNNG in 50 µL acetone. Seven days later, the mice were treated with 2.5, 5 or 10 ng/animal 2,3,7,8-TCDD in 25 µL acetone twice weekly for 20 weeks. A control group of 20 mice received acetone followed by 10 ng/animal 2,3,7,8-TCDD. The numbers of surviving mice with papillomas, carcinomas or hyperproliferative nodules of the skin were 8/20, 8/19 and 7/18 after treatment with MNNG and 2.5, 5 or 10 ng/animal 2,3,7,8-TCDD compared with 0/18 mice treated with 2,3,7,8-TCDD alone (Hébert et al., 1990a).

In two experiments with female Senear mice, it was shown that prior skin application of 2,3,7,8-TCDD reduced tumorigenesis of DMBA (or monofluoro derivatives of DMBA) and benzo[a]pyrene (Cohen et al., 1979; DiGiovanni et al., 1983).

3.6.2. Lung

3.6.3. Liver

(b) Rat

Groups of 4–7 female Sprague-Dawley rats, weighing 200–250 g, were subjected to 70% partial hepatectomy and 24 h later treated once with saline or 10 mg/kg bw NDEA in saline by gastric instillation. Starting seven days later, the rats were given 0, 0.14 or 1.4 jug/kg bw 2,3,7,8-TCDD by subcutaneous injection every two weeks for 28 weeks at which time all surviving rats were killed. There was an increase in the number and size of focal and nodular hepatic lesions with 2,3,7,8-TCDD in the low- and high-dose groups. Focal and nodular hepatic lesions were identified by y-glutamyl transpeptidase (GGT), canalicular adenosine triphosphatase (ATPase) and glucose-6-phosphatase (G6Pase). The number of focal lesions was 309 ± 98 (NDEA alone), 34 ± 17 (low-dose 2,3,7,8-TCDD), 25 ± 7 (high-dose 2,3,7,8-TCDD), 1068 ± 166 (NDEA followed by low-dose 2,3,7,8-TCDD) and 871 ± 66 (NDEA followed by high-dose 2,3,7,8-TCDD). The corresponding volume fractions of liver occupied by these lesions were 0.7, 0.2, 0.1, 9.0 and 43.0% (Pitot et al., 1980).

Female Fischer 344 rats (150–220 g) [number not stated] were subjected to a 70% partial hepatectomy and given 10 mg/kg bw NDEA orally 24 h after the surgery. Starting two weeks after surgery, the rats were given intramuscular injections of 0, 0.0014, 0.014, 0.14 or 1.4 µg/kg bw 2,3,7,8-TCDD in corn oil every two weeks for six months. Treatment with 2,3,7,8-TCDD alone had virtually no effect on the number or volume fraction of altered hepatic foci. No effect on the number or volume fraction of focal hepatic lesions was observed at the three lowest doses. Only the highest dose (0.1 µg/kg bw per day) significantly increased the number and volume fraction of hepatic foci (detected by GGT, ATPase and G6Pase) (approximately 10 000 foci/liver with NDEA + 2,3,7,8-TCDD compared with 4000 foci/liver with NDEA alone; volume fraction approximately 2.8% with NDEA + 2,3,7,8-TCDD compared with 0.8% with NDEA alone) (Pitot et al., 1987).

Groups of 10 female Sprague-Dawley rats (140–160 g) were subjected to a 70% partial hepatectomy and 24 h later were given a single intraperitoneal injection of 30 mg/kg bw NDEA. Starting one week later, the rats were given weekly subcutaneous injections of the corn oil solvent or 2,3,7,8-TCDD (0.7 µg/kg bw) for 14 or 26 weeks. Alternatively, some groups of rats were given a single loading dose (3.5 µg/kg bw) of 2,3,7,8-TCDD at five weeks after NDEA treatment or solvent in order to attain the same total dose. These latter groups were then given a weekly maintenance dose of 0.7 µg/kg bw 2,3,7,8-TCDD for nine or 21 weeks until killing. Focal hepatic lesions (detected by GGT) were significantly increased with weekly administration of 0.7 µg/kg bw 2,3,7,8-TCDD for 26 weeks. In the group receiving a loading dose of 2,3,7,8-TCDD, the percentage of liver occupied by hepatic foci was even greater than in the group treated with the same cumulative dose (Flodström & Ahlborg, 1989). Using the same experimental design, the influence of vitamin A deficiency upon the 2,3,7,8-TCDD response was investigated. Vitamin A deficiency increased the mean volume fraction of foci following NDEA + 2,3,7,8-TCDD treatment (Flodström et al., 1991).

Female Fischer 344 rats (130–220 g) were subjected to a 70% partial hepatectomy and 24 h later were given a single gastric instillation of 10 mg/kg bw NDEA. Starting two weeks later, nine rats were administered oil and four rats were given 0.14 µg/kg bw 2,3,7,8-TCDD by subcutaneous injection every two weeks for six months. The number and volume fraction of the liver occupied by focal hepatic lesions (detected by glutathione S-transferase P (GSTP), GGT, ATPase, G6Pase) in the 2,3,7,8-TCDD-treated group was significantly increased compared with controls. The numbers of focal hepatic lesions per liver were 90 ± 30 (NDEA alone) and 14 220 ± 1340 (NDEA followed by 2,3,7,8-TCDD), while the volume fractions of liver were 0.14 ± 0.05% (NDEA alone) and 1.23 ±0.11% (NDEA followed by 2,3,7,8-TCDD) (Dragan et al., 1991).

Groups of 10–20 female Sprague-Dawley rats were subjected to a 70% partial hepatectomy followed by a single intraperitoneal injection of 30 mg/kg bw NDEA. Starting five weeks later, rats were given a loading dose of 0.22, 0.88 or 3.5 µg/kg bw 2,3,7,8-TCDD by subcutaneous injection followed by a weekly maintenance dose of one fifth of the loading dose of 2,3,7,8-TCDD. One group received no 2,3,7,8-TCDD. The rats were killed after 20 weeks of treatment. A significant increase in the percentage of liver occupied by GGT-positive focal hepatic lesions was observed for all doses compared with NDEA controls (approximately 0.15% in controls, 0.6% in the low-dose, 0.59% in the mid-dose and 1.15% in the high-dose rats). A significant increase was seen in the number of foci at all doses (approximately: low-dose, 5000; mid-dose, 4250, high-dose, 7500) as compared to NDEA alone (2000) (Waern et al., 1991).

Groups of nine female Sprague-Dawley rats, 70 days of age, were ovariectomized or sham-operated and were given an intraperitoneal injection of saline or 200 mg/kg bw NDEA. One week later, the rats were administered 2,3,7,8-TCDD in corn oil by intragastric instillation biweekly to provide a daily dose of approximately 100 ng/kg bw per day. The rats were killed after 30 weeks of 2,3,7,8-TCDD administration and GSTP-and GGT-positive foci were examined. For the GGT-positive foci, the percentages of the liver occupied by foci were 0.01 ± 0.01 (control), 0.01 ± 0.01 (2,3,7,8-TCDD alone), 0.3 ± 0.1 (NDEA alone) and 0.87 ± 0.06 (NDEA followed by 2,3,7,8-TCDD) in the intact animals. In the ovariectomized animals, the corresponding GGT-positive volume fractions were 0, 0, 0.03 ± 0.01 and 0.08 ± 0.04. The percentages of liver occupied by GSTP-positive foci were 0.04 ± 0.02 (control), 0.02 ± 0.01 (2,3,7,8-TCDD alone), 0.35 ± 0.11 (NDEA alone) and 1.17 ± 0.26 (NDEA followed by 2,3,7,8-TCDD) in the intact animals and 0.01 ± 0.01, 0.03 ± 0.01, 0.16 ± 0.05 and 0.57 ± 0.1 in the corresponding groups of ovariectomized animals. There was a reduction in the volume fraction of liver occupied by GSTP lesions in ovariectomized rats treated with NDEA alone and NDEA followed by 2,3,7,8-TCDD (Lucier et al., 1991).

Groups of 5–10 female Sprague-Dawley rats were subjected to a 70% partial hepatectomy and administered oil vehicle or 10 mg/kg bw NDEA 24 h later by gastric instillation. Starting one week after the surgery, the rats were given biweekly subcutaneous injections of 0.14 µg/kg bw 2,3,7,8-TCDD for one, three or five months and either killed or maintained for a further six months before killing. The percentages of liver occupied by focal hepatic lesions (as identified by GSTP, GGT, ATPase and G6Pase) were 0.17 ± 0.03 (NDEA alone) and 1.47 ± 0.19 (NDEA and 2,3,7,8-TCDD) at the end of the first five months. The corresponding values in the group observed for an additional six months were 2.00 ±0.27 (NDEA followed by 2,3,7,8-TCDD) and 1.35 ±0.16 (NDEA alone, killed at the 11-month time point) (Dragan et al., 1992).

Groups of 8–10 female Sprague-Dawley rats (70 days of age) were given a single intraperitoneal dose of saline or 175 mg/kg bw NDEA. Starting two weeks later, these rats were then administered an average of 0, 3.5, 10.7, 35.7 and 125 ng/kg bw 2,3,7,8-TCDD biweekly in corn oil by gastric instillation for 30 weeks and the number and volume fraction of GSTP-positive foci were determined. Dose-related increases in the number of foci per liver and the volume percentage of liver occupied by foci (as identified by GSTP) were found (see Table 43) (Maronpot et al., 1993).

Table 43

Induction of focal hepatic lesions in rats treated with 2,3,7,8-TCDD and/or NDEA.

Groups of 20 female Wistar rats (100 g) were administered doses of 10 mg/kg bw NDEA in water by gastric instillation on five consecutive days. Starting two weeks after NDEA administration, rats were treated with either corn oil or 1.4 µg/kg bw 2,3,7,8-TCDD biweekly by subcutaneous injection. Focal hepatic lesions (detected by ATPase deficiency) were quantitated at 9, 13 and 17 weeks for 4–8 rats per treatment group. In the control rats, the percentage of the liver occupied by foci was zero at all time points. In the group treated with NDEA alone, the percentages of liver occupied by foci were 0.025 ± 0.007 (at 9 weeks), 0.045 ± 0.017 (at 13 weeks) and 0.048 ± 0.018 (at 17 weeks). In the rats treated with 2,3,7,8-TCDD alone, the percentages of the liver occupied by ATPase-deficient hepatic foci were 0.030 ± 0.021 (at 9 weeks), 0.020 ± 0.011 (at 13 weeks) and 0.283 ± 0.211 (at 17 weeks). In the NDEA/2,3,7,8-TCDD-treated rats, the percentages of liver occupied by foci were 0.043 ± 0.02 (at 9 weeks), 0.211 ± 0.065 (at 13 weeks) and 0.313 ± 0.215 (at 17 weeks) (Buchmann et al., 1994).

Groups of five female Wistar rats (190–210 g) were given N-nitrosomorpholine (80 mg/L) in the drinking-water for 25 days. Starting two weeks later, the rats were given biweekly subcutaneous injections of corn oil or 2,3,7,8-TCDD for 13 weeks. The calculated average daily dose was 2, 20 or 200 ng/kg bw. The number and volume fraction of hepatic focal lesions (ATPase-deficient) were increased following 13 weeks of 2,3,7,8-TCDD administration at all doses. The numbers of hepatic focal lesions were approximately 125/cm³ (control), 250/cm³ (low-dose 2,3,7,8-TCDD), 250/cm³ (mid-dose 2,3,7,8-TCDD) and 500/cm³ (high-dose 2,3,7,8-TCDD), while the corresponding volume fractions were approximately 0.125% (control), 0.10% (low-dose 2,3,7,8-TCDD), 0.125% (mid-dose 2,3,7,8-TCDD) and 0.6% (high-dose 2,3,7,8-TCDD) (Schrenk et al., 1994a).

Groups of 6–15 female Sprague-Dawley rats, 21 days of age, were treated with 10 mg/kg bw NDEA by intraperitoneal injection. After 30 days, rats were given either a basal diet or the basal diet supplemented with 2,3,7,8-TCDD (150 ppt [ng/kg diet], equivalent to a daily dose of about 0.007 µg/kg bw). Hepatic lesions were scored as ATPase-deficient foci. A time-dependent increase was noted in the volume fraction of hepatic foci in NDEA + 2,3,7,8-TCDD-treated rats (170 days after NDEA, 2%; 240 days, 4%; 450 days, 17%) as compared to NDEA alone (170 days, 0.6%; 240 days, 1.0%; 450 days, 4.0%). No hepatotoxicity was detected in any of the groups. Other rats were treated with NDEA (as above) followed by phenobarbital (PB) in the diet (500 ppm) for 30–170 days, at which time the rats were either returned to the basal diet or exposed to dietary 2,3,7,8-TCDD (150 ppt; 0.007 µg/kg bw per day). There was a time-dependent increase in the volume fraction of focal hepatic lesions, which was enhanced in the NDEA + PB 4–2,3,7,8-TCDD group (240 days, 2%; 450 days, 13%) as compared to NDEA + PB (240 days, 1%; 450 days, 4%). However, when comparing the groups given NDEA + PB + 2,3,7,8-TCDD and NDEA + 2,3,7,8-TCDD (240 days, 4%; 450 days, 17%), no difference was seen. This suggests that phenobarbital does not enhance the ATPase-negative foci caused by NDEA + 2,3,7,8-TCDD. [The Working Group noted that loss of ATPase does not identify all possible preneoplastic lesions.] Hepatic tumours were examined at 450 days after NDEA treatment. They were primarily eosinophilic and expressed ras. The number of combined hepatocellular adenomas and carcinomas was 1/12 in the NDEA group, 5/15 in the NDEA + 2,3,7,8-TCDD group, 5/15 in the NDEA 4-PB group and 6/15 in the NDEA + PB + 2,3,7,8-TCDD group (Sills et al., 1994).

The single and combined effects of 2,3,7,8-TCDD and 3,4,5,3′,4′-pentachlorobiphenyl (PCB 126) were examined in groups of 10–15 female Sprague-Dawley rats. Female rats (120–140 g) were subjected to a 70% partial hepatectomy and 24 h later received an intraperitoneal injection of corn oil or 30 mg/kg bw NDEA. Five weeks later, the rats were administered either PCB 126, 2,3,7,8-TCDD or a combination of the two, as an initial loading dose (equivalent to five single doses) followed by weekly injections for 19 weeks, at which time the rats were killed. Weekly doses of PCB 126 were 0, 0.316, 1, 3.16 or 10 µg/kg bw, while those of 2,3,7,8-TCDD were 0.1, 0.316 or 1 µg/kg bw. The doses of the combinations were 1 µg/kg bw PCB 126 plus 0.1 µg/kg bw 2,3,7,8-TCDD, 3.16 µg/kg bw PCB 126 plus 0.316 µg/kg bw 2,3,7,8-TCDD and 10 µg/kg bw PCB 126 plus 1 µg/kg bw 2,3,7,8-TCDD. The number and volume fraction of the liver occupied by altered hepatic foci expressing GGT were increased with the highest dose of PCB 126 (10 µg/kg bw per week), with the highest dose of 2,3,7,8-TCDD (1 µg/kg bw per week), and with two of the combinations of PCB 126 with 2,3,7,8-TCDD (3.16 + 0.316 and 10+ 1 µg/kg bw per week of PCB 126 and 2,3,7,8-TCDD, respectively) (Hemming et al., 1995).

Groups of four female Wistar rats (120 g) were given gastric instillations of 10 mg/kg bw NDEA in water daily for 10 days and were then allowed to recover for eight weeks. One group of rats was given a single subcutaneous injection of 1.4 µg/kg bw 2,3,7,8-TCDD (‘acute’), while a second group received biweekly subcutaneous injections of 1.4 µg/kg bw 2,3,7,8-TCDD for 115 days (‘chronic’). All rats were killed 26 weeks after the start of NDEA treatment. The volume fraction of altered hepatic foci, identified as GSTP-positive, was significantly increased in the ‘chronic’ 2,3,7,8-TCDD + NDEA group (3.4%) as compared to NDEA alone (1.2%). There was no difference between the ‘acute’ 2,3,7,8-TCDD + NDEA group (1.7%) and the NDEA-alone group (1.2%) (Stinchcombe et al., 1995).

Groups of 7–11 female Sprague-Dawley rats given 0 or 175 mg/kg bw NDEA in saline by intraperitoneal injection followed by biweekly treatment with 1.75 µg/kg bw 2,3,7,8-TCDD in corn oil by gastric instillation for 30 weeks. The percentages of the liver occupied by focal hepatic lesions (detected by GSTP) were 0.01 ± 0.01 (non-NDEA-treated controls at 30 weeks), 2.23 ± 1.47 (NDEA followed by 2,3,7,8-TCDD at 30 weeks), 0.32 ± 0.97 (non-NDEA-treated controls at 62 weeks) and 6.05 ± 4.3 (NDEA followed by 2,3,7,8-TCDD for 30 weeks and a 32-week observation period). [The Working Group noted that this study did not compare the NDEA/2,3,7,8-TCDD group to an NDEA group not given 2,3,7,8-TCDD.] The incidence of liver neoplasms was increased in the NDEA/2,3,7,8-TCDD group relative to the controls (5/7 and 0/11) (Tritscher et al., 1995).

Mouse

Groups of 50 male and 50 female B6C3F1 mice, five weeks of age, were administered 0 (control), 5000 (low-dose) or 10 000 (high-dose) mg/kg of diet (ppm) dibenzo-para-dioxin (purity, 99.5%) for 87 (high-dose males) or 90 (low-dose males and females, high-dose females) weeks. All surviving male mice were killed at 92–97 weeks, and all surviving female mice at 91–93 weeks. Mean body weights of the treated male and female mice were slightly lower than those of the corresponding controls. Survival of high-dose females was lower than that of the control and low-dose groups (p > 0.001, Tarone test). At week 90 of the study, 48/50 (96%), 50/50 (100%) and 46/50 (92%) control, low-dose and high-dose males, respectively, were still alive; and 44/50 (88%), 44/50 (88%) and 27/50 (54%) control, low-dose and high-dose females, respectively, were still alive. Tumours were not induced in mice of either sex at significantly higher incidence in the treated groups than in the corresponding control groups (United States National Toxicology Program, 1979a).

Rat

Groups of 35 male and 35 female Osborne-Mendel rats, five weeks of age, were administered 0 (control), 5000 (low-dose) or 10 000 (high-dose) mg/kg of diet (ppm) dibenzo-para-dioxin (purity 99.5%) in the diet for 110 weeks. Mean body weights of the treated male and female rats were lower that those of the corresponding controls. In male rats, survival in the control group was lower than in the treated groups (p = 0.011, Tarone test). Survival of high-dose female rats was lower than that of the control and low-dose groups (p = 0.007, Tarone test). At week 90 of the study, 24/35 control and 29/35 rats in each treated group of males were still alive, and 32/35, 31/35 and 20/35 control, low-dose and high-dose female rats, respectively, were still alive. Tumours were not induced in rats of either sex at significantly higher incidence in the treated groups than in the corresponding control groups (United States National Toxicology Program, 1979a).

3.1.1. Mouse

Groups of 50 male and 50 female B6C3F1 mice, five weeks of age, were given 0 (control), 5000 (low-dose) or 10 000 (high-dose) mg/kg of diet (ppm) 2,7-DCDD for 90 weeks. Three impurities with peak areas 3–6% of the major peak were detected by gas chromatography. One was identified as trichlorodibenzo-para-dioxin. No 2,3,7,8-TCDD was detected. All surviving male mice were killed at 92–101 weeks and all surviving female mice at 91–98 weeks. Mean body weights of the male mice were unaffected by administration of the test chemical, whereas those of female treated mice were slightly lower than those of the corresponding controls. Survival in male mice was unaffected by treatment. Survival in high-dose female mice was significantly lower than that in control and low-dose groups (p > 0.001, Tarone test). At the end of the study, 48/50, 36/50 and 38/50 control, low-dose and high-dose males, respectively, were still alive; and 45/50, 46/50 and 28/50 control, low-dose and high-dose females, respectively, were still alive. Treated male and female mice showed an increased incidence of focal necrosis of the liver. The incidence of hepatocellular carcinomas in male mice was unaffected by treatment: 4/49, 5/50 and 5/42 control, low-dose and high-dose mice, respectively. However, the incidence of hepatocellular adenomas was 4/49, 15/50 and 12/42 in control, low-dose and high-dose males, respectively. The incidence of hepatocellular adenomas and carcinomas combined (8/49, 20/50 and 17/42) was significantly higher in the treated groups than in the control group (p = 0.008, Cochran-Armitage test for trend). The incidences in the low- and high-dose groups were both significantly higher than that in the control group (p = 0.008 and p = 0.010, respectively, Fisher's exact test). In male mice, the incidence of lymphoma or leukaemia was higher in the low-dose (6/50) than in the control group (0/50) (p = 0.006, Fisher's exact test). However, the incidence in the high-dose group (3/50) was not statistically different from that of controls. In the female mice, no tumours were observed at significantly higher incidence in the dosed groups than in the corresponding controls (United States National Toxicology Program, 1979b). [The Working Group noted that the presence of three impurities makes it impossible to ascribe the observed carcinogenic effect specifically to the 2,7-dichloro congener.]

3.1.2. Rat

Groups of 35 male and 35 female Osborne-Mendel rats, five weeks of age, were given 0 (control), 5000 (low-dose) or 10 000 (high-dose) ppm 2,7-DCDD in the diet for 110 weeks. Three impurities with peak areas 3–6% of the major peak were detected by gas chromatography. All surviving male rats were killed at 110–112 weeks, and all surviving female rats at 110–117 weeks. Mean body weights of the dosed groups were lower that those in the corresponding controls. Survival was not significantly affected by administration of 2,7-DCDD. At week 78 of the study, 30/35, 26/35 and 29/35 control, low-dose and high-dose males, respectively, were still alive; and 33/35, 30/35 and 28/35 control, low-dose and high-dose females, respectively, were still alive. Toxic hepatic lesions characterized by centrilobular fatty metamorphosis (33–48%) and/or necrosis (6–20%) were observed in both low- and high-dose rats. Tumours were not induced in rats of either sex at significantly higher incidence in the dosed groups than in the corresponding control groups (United States National Toxicology Program, 1979b)

Skin

Mouse: Groups of 20 female HRS/J hairless (hr/hr) mice, five to eight weeks of age, were treated with single skin applications of 0 or 5 µmol (0.75 mg) per animal MNNG in 50 µL acetone followed by 20 µg/animal 2,7-DCDD in 50 µL acetone twice weekly for 20 weeks. A control group of 20 mice was given acetone followed by 10 ng/animal 2,3,7,8-TCDD. Skin tumours developed in 0/19 mice treated with MNNG plus 2,7-DCDD and in 0/20 mice treated with 2,7-DCDD alone (Poland et al., 1982).

3.1.1. Mouse

Groups of 50 male and 50 female B6C3F1 mice, six weeks of age, were administered HxCDD (purity, 98.6%; 31% 1,2,3,6,7,8-HxCDD and 67% 1,2,3,7,8,9-HxCDD of the total HxCDD content), suspended in a vehicle of 9 : 1 corn oil-acetone, by gastric instillation twice a week for 104 weeks at doses of 1.25 (low), 2.5 (mid) or 5 (high) µg/kg per week for male mice and 2.5 (low), 5 (mid) or 10 (high) µg/kg per week for female mice. Groups of 75 mice of each sex served as vehicle controls and 50 mice of each sex as untreated controls. All surviving animals were killed at 105–108 weeks. The mean body weights and survival in the treated groups were similar to those of the vehicle-control groups. At the end of the study, 32/50, 38/75, 29/50, 26/50 and 23/50 untreated control, vehicle-control, low-, mid- and high-dose male mice, respectively, were still alive; for females, the corresponding numbers were 36/50, 58/75, 31/50, 33/50 and 36/50 animals. In males, hepatocellular adenomas occurred at increased incidence in the high-dose group (7/73, 5/50, 9/49 and 15/48 vehicle-control, low-, mid- and high-dose mice, respectively; p = 0.003, Fisher's exact test). There was no significant increase in the incidence of hepatocellular carcinomas (8/73, 9/50, 5/49 and 9/48). In females, hepatocellular adenomas occurred at incidences that were dose-related (2/73, 4/48, 4/47 and 9/47 in the vehicle-control, low-, mid- and high-dose mice, respectively; p = 0.002, Cochran-Armitage test for trend); the incidence in the high-dose group was significantly higher (p = 0.003, Fisher's exact test) than that in the corresponding vehicle-control group. Hepatocellular carcinomas were found in 1/73, 0/48, 2/47 and 2/47 females, respectively (United States National Toxicology Program, 1980a). [The Working Group noted that it is unlikely that the impurities consisting of various polyhalogenated dibenzo-para-dioxins were responsible for the observed carcinogenic effects.]

3.1.2. Rat

Groups of 50 male and 50 female Osborne-Mendel rats, six weeks of age, were administered HxCDD (purity, 98.6%; 31% 1,2,3,6,7,8-HxCDD and 67% 1,2,3,7,8,9-HxCDD of the total HxCDD content), suspended in a vehicle of 9 : 1 corn oil-acetone, by gastric instillation twice a week for 104 weeks at doses of 1.25 (low), 2.5 (mid) or 5 (high) µg/kg per week. Groups of 75 rats of each sex served as vehicle controls and 50 rats of each sex as untreated controls. All surviving animals were killed at 105–108 weeks. A dose-related depression in mean body weight was seen in both sexes. At the end of the study, 24/50, 29/75, 18/50, 19/50 and 19/5Q untreated control, vehicle control, low-, mid- and high-dose rats, respectively, were still alive; for females, the corresponding numbers were 33/50, 39/75, 36/50, 36/50 and 37/50 animals. In male rats, hepatic neoplastic nodules occurred at incidences that were dose-related (0/74, 0/49, 1/50 and 4/48 in the vehicle-control, low-, mid- and high-dose groups, respectively (p = 0.003, Cochran-Armitage test for trend). No hepatocellular carcinomas occurred in males. In female rats, hepatic neoplastic nodules occurred at incidences that were dose-related (5/75, 10/50, 12/50 and 30/50 in the vehicle-control, low-, mid- and high-dose groups, respectively; p > 0.001, Cochran-Armitage trend test); the incidences in the mid- and high-dose groups were significantly higher (p = 0.006 and p > 0.001, respectively, Fisher's exact test) than that in the corresponding vehicle-control group. In addition, four high-dose females had hepatocellular carcinomas (United States National Toxicology Program, 1980a). [The Working Group noted that it is unlikely that the impurities consisting of various polyhalogenated dibenzo-para-dioxins were responsible for the observed carcinogenic effects.]

Liver

Rat: Waern et al. (1991) analysed the GGT-positive foci in groups of 10–20 female Sprague-Dawley rats subjected to a 70% partial hepatectomy followed by single intraperitoneal injection of 30 mg/kg bw NDEA and 1,2,3,7,8-PeCDD. Five weeks after NDEA administration, a loading dose of 1,2,3,7,8-PeCDD consisting of five times the weekly maintenance dose in each treatment group was given by subcutaneous injection. Weekly maintenance doses of 0.088, 0.35 and 1.4 µg/kg bw 1,2,3,7,8-PeCDD per week were then given for 20 weeks. The rats were killed at the end of the treatment. A significant increase in the percentage of liver occupied by GGT-positive focal hepatic lesions was observed for all doses of 1,2,3,7,8-PeCDD (approximately: low-dose, 0.35%; mid-dose, 0.7%; high-dose, 1.6%) compared with NDEA alone (0.15%). A significant increase in the number of foci per liver was also seen in the two highest-dose groups (approximately: low-dose, 2750; mid-dose, 5000; high-dose, 10 250) compared with NDEA alone (2000).

Liver

Rat: Groups of 4–8 female Wistar rats (100 g) were given five doses of 10 mg/kg bw NDEA in water by gastric instillation. Starting two weeks later, rats were treated with either corn oil solvent or 70 µg/kg bw 1,2,3,4,6,7,8-HpCDD biweekly by subcutaneous injection. The development of focal hepatic lesions was determined at 9, 13 and 17 weeks. ATPase-deficient lesions were quantitated and the GSTP-positive focal population was used to determine focal proliferation rates from pulse labelling with bromo-deoxyuridine (BrdU) before sacrifice. The numbers of ATPase-deficient lesions were 619 ± 113 (at 9 weeks), 551 ± 158 (at 13 weeks) and 1691 ± 652 (at 17 weeks) per cm³ in the NDEA-treated rats given 1,2,3,4,6,7,8-HpCDD. The corresponding numbers in the groups given 1,2,3,4,6,7,8-HpCDD but no NDEA were 63 ± 35 (at 9 weeks), 87 ± 16 (at 13 weeks) and 120 ± 30 (at 17 weeks) per cm³. The percentage volume fractions of ATPase-deficient lesions were 0.093 ± 0.38 (at 9 weeks), 0.091 ± 0.037 (at 13 weeks) and 0.76 ± 0.40 (at 17 weeks) (Buchmann et al., 1994).

Groups of five female Wistar rats (190–210 g) were given N-nitrosomorpholine (80 mg/L) in the drinking-water for 25 days. Starting two weeks later, the rats were given biweekly subcutaneous injections of corn oil or 1,2,3,4,6,7,8-HpCDD for 13 weeks. The calculated average daily doses were 100, 1000 or 10 000 ng/kg bw. The number and volume fraction of ATPase deficient lesions were assessed following 13 weeks of admi-nistration. The numbers of ATPase-deficient hepatic foci were approximately 125/cm³ (controls), 450/cm³ (low-dose), 250/cm³ (mid-dose) and 500/cm³ (high-dose), while the corresponding percentage volume fractions were approximately 0.125 (control), 0.25 (low-dose), 0.125 (mid-dose) and 0.9 (high-dose) (Schrenk et al., 1994a).

Liver

Rat: Groups of five female Wistar rats (190–210 g) were given N-nitrosomorpholine (80 mg/L) in the drinking-water for 25 days. Starting two weeks later, the rats were given biweekly subcutaneous injections of corn oil or a defined mixture of 49 PCDDs for 13 weeks at 200 ng, 2000 ng or 20 000 ng PCDD/kg. The number and volume fraction of ATPase-deficient lesions were assessed following 13 weeks of administration. The numbers of ATPase-deficient hepatic foci were approximately 125/cm³ (controls), 125/cm³ (200 ng/kg), 300/cm³ (2000 ng/kg) and 800/cm³ (20 000 ng/kg), while the corresponding percentage volume fractions were 0.125 (control), 0.2 (200 ng/kg), 0.25 (2000 ng/kg) and 0.75 (20 000 ng/kg) (Schrenk et al., 1994a).

(a) Absorption

Absorption across the gastrointestinal mucosa depends on the vehicle and on the molecular size and solubility of the congener. These factors seem to be most significant for the hepta- and octa-CDDs. There appear to be only minor differences in absorption between rodent species (Van den Berg et al., 1994). Single-dose exposure studies with 2,3,7,8-TCDD showed 70–90% absorption from the gastrointestinal tract in rats, hamsters and guinea-pigs (Piper et al., 1973; Allen et al., 1975; Rose et al., 1976; Olson et al., 1985; Decad et al., 1990), whereas less than 50% gastrointestinal absorption was found in female C57BL/6J and male ICR/Ha Swiss mouse strains (Koshakji et al., 1984; Curtis et al., 1990). When 2,3,7,8-TCDD was administered in the diet to rats for up to six weeks, absorption was only 50–60% (Fries & Marrow, 1975) compared with 70% following a single dose in corn oil (Piper et al., 1973). No significant influence of age was found for gastrointestinal absorption of 2,3,7,8-TCDD in rats (Hébert & Birnbaum, 1987). Absorption of 1,2,3,7,8-PeCDD is in the same range as that of 2,3,7,8-TCDD (Yoshimura et al., 1986; Brewster & Birnbaum, 1987). The only other PCDD congener of which the gastrointestinal absorption has been studied in some detail is OCDD; only 2–15% of a single oral dose was taken up by rats (Birnbaum & Couture, 1988). For metabolites of 2,3,7,8-TCDD, it has been shown that enterohepatic circulation is not significant in the rat (Ramsey et al., 1982).

The uptake of PCDDs by dermal permeation and pulmonary absorption is much more limited than uptake after oral ingestion (Nessel et al., 1992; Diliberto et al., 1996). For 2,3,7,8-TCDD, dermal permeation is strongly dose-dependent in rats (Banks et al., 1990). Uptake from the application site is slow but also depends on the vehicle used and shows a good inverse correlation with the octanol-water partition coefficient (Brewster et al., 1989; Banks & Birnbaum, 1991; Jackson et al., 1993). The bioavailability after dermal exposure is probably less than 1% for all congeners (Van den Berg et al., 1994).

The adsorption of PCDDs on various environmental matrices can lead to a significant reduction in bioavailability, which depends strongly on the properties of the particles (Nessel et al., 1992; Van den Berg et al., 1994). On the basis of a number of studies with rodents, a bioavailability for PCDDs from soil of 25–50% was suggested for the C1₄−C1₆ congeners, while 10% was proposed for the Cl₇ and Cl₈ congeners. For combustion particles (fly ash), a bioavailability of 5–20% was proposed as a realistic estimate for PCDDs, this being high for the hepta- and octa-chlorinated PCDDs (Van den Berg et al., 1994).

(b) Body distribution

In all rodent species, the liver and adipose tissue are the major storage sites for PCDDs. However, in certain species, the skin can also act as an important storage site and high concentrations can be found in adrenals. Most detailed information is available for 2,3,7,8-TCDD, but reasonable quantities of data are available to support this primary tissue distribution in rodents for the higher chlorinated congeners. Long retention in tissues is caused primarily by the steric hindrance towards cytochrome P450 activity caused by chlorine atoms on the 2,3,7,8 positions, resulting in limited metabolism (Van den Berg et al., 1994). Some tissue retention has been observed for certain non-2,3,7,8-substituted PCDDs in rodents and monkeys (Abraham et al., 1989; Neubert et al., 1990a), but the levels should not be considered as toxicologically relevant. Transport in the blood occurs through binding to plasma lipids and lipoproteins, with increasing affinity found for the higher chlorinated congeners in the plasma proteins (Patterson et al., 1989; Schecter et al., 1990h). The majority of 2,3,7,8-TCDD is bound to very low-density lipoprotein, followed by low-density lipoprotein and high-density lipoprotein (Marinovich et al., 1983).

Depending on the experimental conditions, 25–70% of the administered dose of 2,3,7,8-TCDD is stored in the liver of rats, mice, hamsters and guinea-pigs approximately one day after exposure. Studies with mixtures of PCDDs showed that in rodents the Cl₅ and Cl₆ congeners have higher hepatic retention than 2,3,7,8-TCDD (Van den Berg et al., 1994). As a result, the liver to adipose tissue concentration ratio strongly increases with increasing chlorination. Measurement of tritiated 2,3,7,8-TCDD administered in corn oil by gastric instillation to adult female rhesus monkeys, infant male rhesus monkeys and young adult male Sprague-Dawley rats indicated that liver retention of the dose was > 40% in rats, but < 10% in the rhesus monkeys. A large percentage of the dose was located in monkey tissues with a high lipid content, especially the skin, muscle and adipose tissue. In the rats, these tissues had much lower levels of 2,3,7,8-TCDD (Van Miller et al., 1976). Differences in liver to adipose tissue ratio between rats and marmosets (Callithrix jacchus) have been observed (see Figure 2) (Olson, 1994).

Figure 2

The liver-to-adipose tissue ratios for 2,3,7,8-substituted PCDDs and PCDFs in marmosets and rats

Hepatic disposition of 2,3,7,8-TCDD is dose-dependent (Kociba et al., 1978; Abraham et al., 1988); this has been suggested to be caused by the presence in the liver of inducible protein binding sites, at least in mice (Poland et al., 1989a). As a possible mechanism, binding to CYP1A2 in the liver has been proposed (Voorman & Aust, 1987; Poland et al., 1989; Voorman & Aust, 1989). In rats, 2,3,7,8-TCDD was equally distributed between hepatic P9 (mitochondrial, lysosomal, nuclear) and S9 (cytosol and microsomal) fractions, localization in the S9 being in microsomes. In extrahepatic tissues (lung/kidneys), 2,3,7,8-TCDD was concentrated in the P9 fraction, while in the S9 fraction localization was in the cytosol. Use of CYP1A1 and CYP1A2 immunoreactive proteins and measurement of marker enzyme activities, ethoxyresorufin O-deethylase and methoxyresorufin O-demethylase, respectively, showed that CYP1A1 was induced in microsomes of all three organs, while CYP1A2 was induced only in liver (Santostefano et al., 1996). However, the role of other dioxin (inducible) hepatic binding proteins cannot be excluded (Landers et al., 1990; Buckley-Kedderis et al., 1992; Tritscher et al., 1992; Van den Berg et al., 1994).

Pre- and postnatal transfer of PCDDs has been studied in rats and mice. Lactational transfer was found to be one or two orders a magnitude higher than placental transfer in these rodents (Nau & Bass, 1981; Weber & Birnbaum, 1985; Nau et al., 1986; Van den Berg et al., 1987; Li et al., 1995a). The level of placental transfer depends on the molecular size, with the highest fetal retention observed for 2,3,7,8-TCDD (Van den Berg et al., 1987). In the mouse embryo, the liver is the major storage site, with two to five times higher concentrations than those found in other tissues (Nau & Bass, 1981; Krowke & Neubert, 1990). In contrast, there is no evidence for hepatic concentration in the fetal rat liver. From a quantitative point of view, lactational transfer in marmosets and rhesus monkeys is also more important than placental transport, but placental transport in marmosets can result in significant accumulation in fetal tissues (Hagenmaier et al., 1990b; Krowke et al., 1990). On the basis of studies with complex mixtures of PCDDs and PCDFs, it appears that these compounds are mobilized primarily from adipose tissue and not from the liver in rats (Van den Berg et al., 1987). In marmosets, lactational transfer can lead to concentrations in the tissues of offspring that are equal to or higher than those found in maternal tissues. In both rats and marmosets, lactational transfer decreases with increasing number of chlorine atoms in the PCDD molecule (Van den Berg et al., 1987; Krowke et al., 1990).

(c) Metabolism

The metabolism of PCDDs has been studied primarily in rodents. In rats, oxidation by P450 occurs preferentially at the lateral, 2, 3, 7 and 8 positions, yielding primarily mono-and dihydroxylated metabolites. Furthermore, sulfur-containing metabolites have been identified, which probably arose through glutathione conjugation (Tulp & Hutzinger, 1978). Studies with 2,3,7,8-TCDF in rats suggest that CYP1A1 is directly involved in phase I metabolism of these compounds and not CYP1A2 (Tai et al., 1993). If all of the 2, 3, 7 and 8-positions are substituted with chlorine atoms, metabolic conversion of the molecule is strongly hindered. Recent quantum-mechanical calculations determining highest occupied molecular orbital (HOMO) energy levels for the dibenzo-para-dioxin molecule have confirmed the higher susceptibility for epoxidation of the 2–3 or 7–8 positions in the molecule (Weber et al., 1996). Although 2,3,7,8-TCDD is highly resistant towards biotransformation, small amounts of metabolites have been identified in rats and dogs. Metabolic transformation included oxidative and reductive dechlorination, including NIH shifts (shift of a chlorine substituent). In addition, oxygen bridge cleavage was found to be another important pathway (Poiger & Schlatter, 1979; Ramsey et al., 1982; Poiger & Buser, 1984; Huwe et al., 1996). Conjugation appears to be important for the elimination of PCDDs from the body, as many 2,3,7,8-TCDD metabolites were present as glucuronide conjugates in the bile of the rat (Poiger & Buser, 1984; Wroblewski & Olson, 1985; Huwe et al., 1996). From a qualitative point of view, metabolic pathways are fairly similar among rodent species, but large quantitative differences can be observed (Van den Berg et al., 1994; Larsen et al., 1996). Guinea-pigs show a lower metabolic capacity than rats, Syrian hamsters and mice (Wroblewski & Olson, 1985). Limited information is available about the metabolic conversion of higher chlorinated 2,3,7,8-substituted PCDDs, but it can be expected that pathways will be similar to those for 2,3,7,8-TCDD, provided that susceptible positions are not sterically hindered by additional chlorines. In Figure 3, a generalized scheme of metabolic pathways for PCDDs is given (Van den Berg et al., 1994).

Figure 3

A generalized scheme of pathways for the biotransformation of PCDDs based on the information from mammalian studies in vivo

(d) Excretion

The elimination of PCDDs has been studied in several laboratory species under a variety of experimental conditions (Van den Berg et al., 1994). After exposure, the compounds are usually more rapidly eliminated from blood and muscle tissue than from liver and adipose tissue. In some species, such as primates, elimination from the skin is similar to that from the adipose tissue (Birnbaum et al., 1980; Brewster & Birnbaum, 1987; Brewster et al., 1988, 1989). For most body compartments, the elimination of 2,3,7,8-substituted PCDDs can be described by a one-compartment open model (Hiles & Bruce, 1976; Rose et al., 1976), but models with two- or three-phase elimination have also been applied. In view of the non-linear distribution of PCDDs in many experimental systems, the use of physiologically based pharmacokinetic models has been successfully applied (Carrier et al., 1995a,b). In almost all laboratory species which have been studied, elimination takes place through bile and faeces in the form of hydroxylated or conjugated metabolites that are rapidly eliminated from the body (Van den Berg et al., 1994). Guinea-pigs seem to be an exception among rodents, as this species eliminates 2,3,7,8-TCDD more than its polar metabolites (Olson, 1986).

For 2,3,7,8-TCDD, whole body half-lives in the rat have been reported to range between 17 and 31 days, depending on the strain and experimental conditions used. Most studies with rodents have found similar elimination rates from the liver and adipose tissue for 2,3,7,8-TCDD (Piper et al., 1973; Allen et al., 1975; Fries & Marrow, 1975; Rose et al., 1976; Abraham et al., 1988, 1989; Pohjanvirta et al., 1990a). In rats, lactation is a very effective route of elimination of 2,3,7,8-TCDD, leading to a 50% reduction of the half-life from the liver (Abraham et al., 1988; Korte et al., 1990). The half-life of 1,2,3,7,8-PeCDD elimination in the rat is approximately 30 days, which is similar to that of 2,3,7,8-TCDD (Wacker et al., 1986). However, with more chlorine atoms present in a 2,3,7,8-substituted PCDD, elimination is much slower. For the Cl₆−Cl₈ congeners, half-lives between 75 days and seven years have been calculated for the rat (Birnbaum & Couture, 1988; Van den Berg et al., 1989a, 1994). In the mouse, half-lives for 2,3,7,8-TCDD are in the same range as those in the rat, but there are distinct strain differences (Gasiewicz et al., 1983; Birnbaum, 1986). Thus, the half-life of 2,3,7,8-TCDD in the non-responsive DBA strain is approximately twice as long as that in the responsive C57BL strain (Gasiewicz et al., 1983). In the Syrian hamster, the half-life of 2,3,7,8-TCDD is two- to three-fold lower than in the rat (Olson et al., 1980). Although this faster elimination of 2,3,7,8-TCDD in the Syrian hamster might contribute to the relative insensitivity of this species towards acute toxicity, it can be assumed that the 100-fold difference in sensitivity between Syrian hamsters and other rodents is dominated more by genetic background than kinetics (Van den Berg et al., 1994). The half-life of 2,3,7,8-TCDD elimination in guinea-pigs has been estimated as 30 days (i.e., the same as in rats) (Gasiewicz & Neal, 1979) and 94 days (Olson, 1986). No clear reason for the discrepancy was discovered.

The half-life of 2,3,7,8-TCDD is much longer in primate species than in rodent species. In seven adult female rhesus monkeys, an average half-life of about 391 days was determined after four years of dietary exposure to 25 ng/kg diet (0.67 ng/kg bw/day). Lactation during four months resulted in a 21% decrease of the maternal body burden. During the first year after birth, the half-life of 2,3,7,8-TCDD in breast-fed rhesus monkeys was approximately 181 days, which was significantly lower than that observed in the mothers (Bowman et al., 1989a). In marmosets, longer half-lives for the 2,3,7,8-substituted PCDDs than in rodents were also observed. In liver and adipose tissue, half-lives from eight weeks up to several years were measured (Neubert et al., 1990a).

During pregnancy, elimination from the dam is approximately twice as fast as in the non-pregnant female (Weber & Birnbaum, 1985).

(e) Kinetics and toxicity

Kinetics to some extent influence the toxicity of individual PCDD and PCDF coners. This was illustrated in experiments with B6C3F1 mice dosed with 2,3,7,8-TCDD or 2,3,7,8-TCDF, in which CYP1A1 activity was found to depend on the time period to acquire a steady-state situation (DeVito & Birnbaum, 1995). In addition, total body fat content in various species and strains of laboratory animals may contribute to some of the observed differences in species sensitivity for 2,3,7,8-TCDD (Geyer et al., 1990). However, the available data seem to imply that kinetics are not a governing factor influencing the toxicity of these compounds, but genetic factors seem to predominate (Van den Berg et al., 1994).

(a) Chloracne and other effects on the skin

Chloracne is a persistent acneiform condition characterized by comedones, keratin cysts and inflamed papules with hyperpigmentation and an anatomical distribution frequently involving the skin under the eyes and behind the ears. It occurs after acute or chronic exposure to a variety of chlorinated aromatic compounds by skin contact, ingestion or inhalation (Crow, 1978; Moses & Prioleau, 1985). This acne-like condition is reported to have occurred with or without other effects in at least a few workers after all reported accidents at TCP production facilities (Ashe & Suskind, 1950; Goldman, 1972; May, 1973; Zober et al., 1990), among individuals involved in daily production of 2,3,7,8-TCDD-contaminated products (Bleiberg et al., 1964; Poland et al., 1971; Pazderova-Vejlupkova et al., 1981; Moses et al., 1984; Suskind & Hertzberg, 1984; Moses & Prioleau, 1985; Bond et al., 1989b), among three laboratory workers exposed to pure 2,3,7,8-TCDD (Oliver, 1975) and among at least 193 (0.6%) Seveso residents, mostly children (at least 20% among children aged 0–14 years from Zone A) (Reggiani, 1978; Caramaschi et al., 1981; Ideo et al., 1985; Mocarelli et al., 1986; Assennato et al.,1989). Chloracne was not found among Missouri residents (Hoffman et al., 1986; Webb et al., 1989) examined 10 years after exposure or among Ranch Hand personnel (Roegner et al., 1991). In United States Army Viet Nam veterans, chloracne-like skin lesions were rarely observed on examination (0.9% in Viet Nam veterans versus 0.8% in non-Viet Nam veterans; odds ratio, 1.4; 95% CI, 0.7–2.9) (Centers for Disease Control Vietnam Experience Study, 1988a).

Chloracne appears shortly after exposure to 2,3,7,8-TCDD-contaminated chemicals. In Seveso, the eruption of comedones, usually accompanied by cysts, was observed between two weeks and two months after the reactor release (Reggiani, 1980), and within six months of the explosion, 34 cases of chloracne were identified among children (Caramaschi et al., 1981). In chemical workers involved in the TCP reactor release at the BASF plant in Ludwigshafen, Germany, most cases of chloracne developed within two days after first exposure (Zober et al., 1990; Ott et al., 1994). One case of chloracne developed only two years after the accident, but the authors suggest that the etiology of this case is unclear.

Among Seveso residents, despite high serum 2,3,7,8-TCDD levels, the chloracne resolved in all but one person by 1983 (Assennato et al., 1989; Mocarelli et al., 1991). However, for TCP workers at the Nitro, WV, USA, plant, Moses et al. (1984) reported that the mean duration of chloracne was 26.1 ±5.9 years.

Positive associations between serum and adipose tissue levels of 2,3,7,8-TCDD and other PCDDs and the risk of chloracne among chemical production workers have been reported (Ott et al., 1987; Beck et al., 1989b; Bond et al., 1989b); the risk was also shown to be greater for subjects of young age at exposure, long exposure duration and a history of employment in production areas of high potential exposure.

Mocarelli et al. (1991) described chloracne in people from Seveso Zone A who had 2,3,7,8-TCDD levels in serum lipids ranging from 828 to 56 000 ng/kg (sampled within one year of the reactor release) (see Table 21). The study also included other individuals from Zone A, but without chloracne, who had serum 2,3,7,8-TCDD levels that ranged from 1770 to 10 400 ng/kg. With the exception of one person with chloracne who was 16 years old at the time of the accident, all of the cases were in children under the age of 11 years. This contrasts with higher serum levels measured in some adult chloracne subjects among German production workers (Table 44). Cases had estimated adipose levels of greater than 200 ng/kg 2,3,7,8-TCDD and over 2000 ng/kg lipid HxCDD at the time of diagnosis. However, a threshold level above which chloracne occurs has not been established.

Table 44

Chloracne and adipose tissue levels of 2,3,7,8-TCDD and HxCDD in German chemical workers.

Other dermatological alterations, including hypertrichosis and hyperpigmentation, have been reported among workers exposed to 2,3,7,8-TCDD in the United States (West Virginia and New Jersey) (Ashe & Suskind, 1950; Bleiberg et al., 1964; Poland et al., 1971), Germany (Bauer et al., 1961; Goldman, 1972) and Czechoslovakia (Jirasek et al., 1974). Increased prevalence of actinic or solar elastosis was observed only in West Virginia TCP workers (59.1% exposed versus 30.1% unexposed; p < 0.01). In the same exposed population, three cases of Peyronie's disease (a rare progressive scarring of the penile membrane) were identified (Suskind & Hertzberg, 1984).

(b) Hepatic effects

Lipid levels

A number of case reports and epidemiological studies have described increases in levels of serum lipid fractions, particularly total cholesterol and triglycerides, in TCP production workers, laboratory workers, Seveso and Missouri residents and Viet Nam veterans. Others report no differences between subject and reference levels. A summary of the reported levels is presented in Tables 45 and 46.

Table 45

Mean total cholesterol levels among Seveso and Missouri residents, TCP production workers, BASF accident cohort and Viet Nam veterans.

Table 46

Mean triglyceride levels among Seveso children and Missouri residents, TCP production workers, BASF accident cohort and Viet Nam veterans.

(c) Other gastrointestinal effects

A variety of gastrointestinal disorders other than liver conditions have been reported following heavy, acute or chronic exposure of chemical workers (Ashe & Suskind, 1950; Baader & Bauer, 1951; Bauer et al., 1961; Jirasek et al., 1974). The most consistently reported symptoms were transient episodes of right upper quadrant pain, loss of appetite and nausea. None of the reports suggest an etiology for these symptoms and the symptoms were not reported in later follow-up studies of any cohort (Pazderova-Vejlupkova et al., 1981; Moses et al., 1984; Suskind & Hertzberg, 1984).

Three investigations of TCP production workers reported an increased prevalence of a history of upper gastrointestinal tract ulcer across all age strata of West Virginia workers (exposed, 20.7% versus unexposed, 5.5%) (Suskind & Hertzberg, 1984) and all digestive system diseases [not specified] among workers employed in a plant in Midland, MI (prevalence: exposed, 1.5% versus unexposed, 0.5%) (Bond et al., 1983). The factors contributing to these conditions have not been examined fully. Neither the Ranch Hand study (Roegner et al., 1991; Grubbs et al., 1995) nor the NIOSH study (Calvert et al., 1992) found an increased risk of upper gastrointestinal tract ulcers with increasing serum 2,3,7,8-TCDD level.

(d) Thyroid function

Effects in adults

Little or no information has been reported on the effects of 2,3,7,8-TCDD specifically on thyroid function in production workers or Seveso residents, two groups with documented high serum 2,3,7,8-TCDD levels.

No difference was found between West Virginia TCP production workers and controls for thyroxine (T4) and thyroxine-binding globulin (TBG), although quantitative results were not presented (Suskind & Hertzberg, 1984). Thyroid-stimulating hormone (TSH), T4 and TBG levels were within the normal range in workers exposed in the BASF Ludwigshafen (Germany) accident (Ott et al., 1994). The 1987 Ranch Hand study indicated a nonsignificant reduction in the percentage of triiodothyronine (T3) uptake (Roegner et al., 1991). A slight increase in the mean level of TSH with increasing serum 2,3,7,8-TCDD level was noted in both 1987 and 1991, but did not reach statistical significance (Roegner et al., 1991; Grubb et al., 1995). Among Army Viet Nam veterans, mean TSH levels, but not mean free thyroxine index (FTI) (biologically active) levels, were statistically significantly higher than among non-Viet Nam veterans, after adjustment for the six entry characteristics of age and year of enlistment, race, enlistment status, general technical test score and primary military occupation (Table 47) (Centers for Disease Control Vietnam Experience Study, 1988a).

Table 47

Levels of triidothyronine percentage (T3%) uptake and free thyroxine index in Viet Nam veterans.

Effects in infants

Two studies in the Netherlands examined thyroid function in infants and related this to PCDD, PCDF and/or PCB levels in breast milk, cord blood or third trimester maternal serum samples.

Pluim et al. (1992, 1993c) examined thyroid function among 38 full-term breast-fed infants in relation to the total I-TEQ per kg of breast milk fat of seven PCDDs (see Table 30) and 10 PCDFs, Total T4, TBG and TSH levels were measured sequentially in cord blood and in the blood of infants at one week of age and at 11 weeks of age (Table 48). Total T3 was measured in plasma at 11 weeks. Infants were classified as ‘high’ or ‘low’ with respect to the median of the range of total I-TEQ. At one week and 11 weeks postnatally, mean total T4 and total T4 : TBG ratios were significantly higher among infants in the ‘high’ group. At 11 weeks, TSH was also significantly higher in the ‘high’ group. Total T3 was unchanged.

Table 48

Levels of thyroxine-binding globulin (TBG), thyroxine (T4), free thyroxine (FT4), T4/TBG ratio and thyroid-stimulating hormone (TSH) in nursing infants and workers of the BASF accident cohort.

Koopman-Esseboom et al. (1994b) examined thyroid function in 78 mother–infant pairs in relation to the I-TEQ levels for PCDDs, planar PCBs, non-planar PCBs and total PCBs–PCDDs in breast milk collected in the first and second weeks after delivery. Total T4, total T3, free T4 and TSH levels were measured in the mother during the last month of pregnancy and 9–14 days after delivery, in cord blood and in the blood of infants at 9–14 days and three months after birth (Table 48). All the I-TEQs (PCDDs, co-planar PCBs, non-planar PCBs and total PCDDs/PCBs) were significantly correlated with infant plasma levels of TSH at the second week and third month, and inversely correlated with total T3 pre-delivery, and with total T3 and total T4 post-delivery for the mothers. The non-planar PCB TEQ was not significantly correlated with the mothers' total T4 after delivery and the infants' third month TSH. The measurements in infants with higher PCDD TEQ (based on median TEQ) during their second week of life showed a significant increase in TSH and significant decreases for total T4 and free T4.

These two studies of nursing infants suggest that ingestion of breast milk with elevated levels of PCDDs may alter thyroid function. Both studies covered a short observation period which limits the examination of persistent or long-term changes in thyroid status and the analyses did not control for other factors which might affect thyroid status.

(e) Diabetes

Cross-sectional studies of workers from Nitro, West Virginia, found no difference in glucose levels between the exposed and control populations, although no quantitative values were presented in either report (Moses et al., 1984; Suskind & Hertzberg, 1984). Similarly, the adjusted odds ratio for out-of-range fasting glucose levels comparing Viet Nam veterans with non-Viet Nam veterans was not statistically significant (odds ratio, 1.0; 95% CI, 0.4–2.2) (Centers for Disease Control Vietnam Experience Study, 1988a). Mean fasting glucose levels in the workers exposed in the BASF Ludwigshafen (Germany) accident were elevated compared with the control population and were associated with current levels of 2,3,7,8-TCDD (p = 0.062) but not with the levels back-extrapolated to the time of exposure (Ott et al., 1994).

In the Ranch Hand study, diabetic status was assessed by measuring fasting serum glucose and 2-h postprandial glucose and by using a case definition of diabetes. Diabetes was defined as having a verified history of diabetes or an oral glucose tolerance test of ≥ 11.1 mmol/L (200 mg/dL) (Roegner et al., 1991). The analyses of all three parameters suggested a consistent association between serum 2,3,7,8-TCDD levels above 33.3 ng/kg and an increased risk of diabetes. Adjusted relative risks in Ranch Hand veterans with serum 2,3,7,8-TCDD above 33.3 ng/kg compared with an unexposed group were statistically significantly elevated for fasting serum glucose levels and diabetes (glucose RR, 3.0; p < 0.001; diabetes RR, 2.5; p < 0.001) (Table 49) and for the 2-h postprandial glucose test (RR, 2.4; p= 0.035). In addition, Ranch Hand personnel meeting the case definition for diabetes were also more likely to have earlier onset of diabetes than the unexposed group (Wolfe et al., 1992).

Table 49

Adjusted relative risk (RR) for fasting serum glucose levels, cases of diabetes and mean 2-h postprandial glucose levels by category of lipid-adjusted serum 2,3,7,8-TCDD level in Ranch Hand veterans.

(f) Immunological effects (Tables 50–57)

Table 51

Total lymphocytes in 2,4,5-T production workers, Missouri residents, Viet Nam veterans and BASF accident cohort.

Table 52

B1 Lymphocytes in 2,4,5-T production workers, Missouri residents, Viet Nam veterans, BASF accident cohort and extruder personnel.

Table 53

CD4 Lymphocytes in production workers, Missouri residents, Viet Nam veterans and BASF accident cohort.

Table 54

CD8 Lymphocytes in production workers, Missouri residents, Viet Nam veterans and BASF accident cohort.

Table 55

IgG levels in Missouri residents, Viet Nam veterans and BASF accident cohort.

Table 56

IgM levels in Missouri residents, Viet Nam veterans and BASF accident cohort.

Nine epidemiological studies and one case report have assessed immunological function in populations exposed to 2,3,7,8-TCDD (Reggiani, 1978; Hoffman et al., 1986; Centers for Disease Control Vietnam Experience Study, 1988a; Evans et al., 1988; Jennings et al., 1988; Webb et al., 1989; Roegner et al., 1991; Ott et al., 1994; Tonn et al., 1996). With the exception of the Tonn et al. (1996) study, none has found a clear relationship between exposure and impaired immunological status. Among the Seveso children resident in the area of highest 2,3,7,8-TCDD contamination, immunoglobulins, complement levels, lymphocyte subpopulations and lymphocyte activity analyses were within the normal range, with no differences from those of unexposed controls (Reggiani, 1978).

In Missouri residents with potential exposure to 2,3,7,8-TCDD-contaminated soil (see Section 1.3.2(c)), depression in cell-mediated immunity (delayed hypersensitivity) was reported by Hoffman et al. (1986). However, a follow-up study did not confirm the presence of anergy (Evans et al., 1988). In a later study of a different cohort of Missouri residents, Webb et al. (1989) found no clinical evidence of immunosuppression in 40 individuals whose adipose 2,3,7,8-TCDD levels ranged from below 20 ng/kg to over 430 ng/kg [top of range not given].

The effect of past occupational exposure on parameters of the immune system was examined in 18 British workers who were evaluated 17 years after accidental industrial exposure to chemicals contaminated with 2,3,7,8-TCDD (Jennings et al., 1988). There were no significant differences in the levels of immunoglobulins, T and B lymphocytes, responsiveness to phytohaemagglutinin A or the CD4 and CD8 counts. Three measurements were different in workers compared with controls: antinuclear antibodies (8 workers versus 0 controls, p < 0.01), immune complexes (11 workers versus 3 controls, p < 0.05) and natural killer cells (workers, 0.40 × 10⁶/L versus controls, 0.19 × 10⁶/L; p < 0.002).

Levels of IgA, IgG, IgM and complement C4 and C3 were higher in exposed workers than in the unexposed control population in the BASF accident study (Ott et al., 1994).

Flow cytometric analysis of lymphocytes from workers with moderately increased serum 2,3,7,8-TCDD (25–140 ng/kg fat) and other PCDDs/PCDFs (maximum, 522 ng I-TEQ/kg fat) did not indicate any decrease of specific cellular components of the immune system (Neubert et al., 1993a, 1995). Moderate increases in the 2,3,7,8-TCDD body burden did not induce any medically significant change in the capacity for proliferation of lymphocytes, measured as [³H]thymidine incorporation (Neubert et al., 1995).

Tonn et al. (1996) examined lymphocyte function among 11 workers employed between 1966 and 1976 in TCP production. Lipid-adjusted serum 2,3,7,8-TCDD levels, measured in 1989–92, ranged from 43 to 874 ng/kg. Although the numbers of lymphocyte subsets were normal, there was a small but significant difference in the response to alloantigen challenge of T-cells and their proliferative response to interleukin (IL)-2. In addition, compared with unexposed controls, the lymphocytes of the exposed workers displayed a suppressive activity that inhibited an on-going allo-response of HLA-unrelated lymphocytes.

Two studies extensively evaluated parameters of the immune system in Viet Nam veterans. No significant differences were detected between United States Army ground troops and the comparison population in lymphocyte subset populations, T-cell populations or serum immunoglobulins (Tables 50–57) (Centers for Disease Control Vietnam Experience Study, 1988a). In the United States Air Force Ranch Hand Study (Roegner et al., 1991), significant positive associations were found between IgA and serum 2,3,7,8-TCDD levels. The authors suggest that the rise in IgA is consistent with a subclinical inflammatory response of unspecified origin.

Table 50

CD4/CD8 ratios in Missouri residents, Viet Nam veterans and BASF accident cohort.

Table 57

Levels of natural killer cells in Missouri residents, Viet Nam veterans and extruder personnel.

One report (Weisglas-Kuperus et al., 1995) examined direct and surrogate measures of immune status in 207 babies in Rotterdam. The surrogate measures were derived from questionnaires given to the mothers, that covered incidence of rhinitis, bronchitis, tonsillitis and otitis in children up to 18 months of age. Almost all of the children (205) were vaccinated against measles, rubella and mumps; the children's antibody levels were subsequently measured to assess humoral immunity. No relationship was found between pre- and postnatal PCB/PCDD exposure and respiratory tract symptoms (i.e., number of periods with rhinitis, bronchitis, tonsillitis and otitis) or humoral antibody production. However, high prenatal exposure, estimated by cord blood levels, was associated with alterations in T-cell subsets. All values were within clinically normal ranges, and the small changes observed do not necessarily mirror alterations in the cell composition of lymphoid and non-lymphoid organs, nor do they necessarily reflect functional defects. The long-term effects of such subtle shifts in the distribution of outcome measures remain unknown.

(g) Neurological effects

Adults: Several studies have evaluated the relationship between neurobehavioural and neurological function among chemical production workers and community residents exposed to 2,3,7,8-TCDD (Sweeney et al., 1989). Two studies have reported significant neuropathy in Seveso residents (Filippini et al., 1981) and in TCP production workers (Moses et al., 1984), but these findings have not been reproduced in subsequent studies of the same populations. Studies of TCP production workers and Viet Nam veterans have found no neurological disorder associated with PCDD exposure (Suskind & Hertzberg, 1984; Assennato et al., 1989; Hoffman et al., 1986; Centers for Disease Control Vietnam Experience Study, 1988a,b; Assennato et al., 1989; Webb et al., 1989; Sweeney et al., 1993).

Infants: A series of studies of infants born in either Rotterdam or Groningen in the Netherlands evaluated their neurological and behavioural development in a series of tests conducted at two weeks and three, seven and 18 months, in relation to breast milk levels of PCBs, PCDDs and PCDFs (see Table 31). Exposed infants were breast-fed; the controls were formula-fed. Neurological status of 418 neonates (10–21 days) measured using the Prechtl neonatal neurological examination (Huisman et al., 1995a) was related to total I-TEQ (odds ratio, 3.2; 95% CI, 1.4–7.5) and specifically to higher chlorinated PCDDs, 2,3,7,8-TCDF and 2,3,4,7,8-PeCDF. Multiple tests were conducted in the 207 Groningen children at three, seven and 18 months to evaluate psychomotor development (Bayley Scales of Infant Development) (Koopman-Esseboom et al., 1996) and visual recognition memory (Fagan Test of Infant Intelligence) at three and seven months (Koopman-Esseboom et al., 1995a,b). Results of the Fagan Tests and the psychomotor and mental developmental indices were no different in breast-fed or formula-fed infants. Similarly, the neurological status of children at 18 months was not significantly affected by PCB/PCDD exposure through breast milk (Huisman et al., 1995b).

(h) Circulatory system

Several studies have described mortality from diseases affecting the circulatory system among populations exposed to 2,3,7,8-TCDD (Bond et al., 1987; Bertazzi et al., 1989; Zober et al., 1990; Coggon et al., 1991; Fingerhut et al., 1991b; Bertazzi et al., 1992; Bueno de Mesquita et al., 1993; Collins et al., 1993; Flesch-Janys et al., 1995) (Table 58).

Table 58

Mortality from diseases of the circulatory system in populations exposed to 2,3,7,8-TCDD.

Mortality from all diseases of the circulatory system was similar to that of general populations in studies of workers from the Netherlands (Plant A) (Bueno de Mesquita et al., 1993), the United States (Nitro, West Virginia) (Collins et al., 1993) and the United Kingdom (Coggon et al., 1991). In two studies of workers with chloracne, mortality was not significantly different from that of the national comparison groups (Bond et al., 1987; Zober et al., 1990). However, in a study of German chemical workers who manufactured TCP and 2,4,5-T in addition to chemicals contaminated with higher chlorinated PCDDs and PCDFs (Flesch-Janys et al., 1995), mortality from all circulatory diseases was positively related to estimated 2,3,7,8-TCDD levels and significantly related to estimated total I-TEQ concentrations above 39 ng/kg (lipid-adjusted).

Mortality from more specific endpoints, including ischaemic heart disease and cerebrovascular disease, has been reported in some studies of TCP production workers (Bond et al., 1987; Bueno de Mesquita et al., 1993; Fingerhut et al., 1991b; Flesch-Janys et al., 1995). With the exception of an increase in risk for ischaemic heart disease with levels of 2,3,7,8-TCDD exposure in one study (Flesch-Janys et al., 1995), ischaemic heart disease mortality was consistent with background population rates.

The SMRs for circulatory system diseases reported in the various mortality studies are close to 1.00, suggesting that the ‘healthy worker effect’ is not seen in these studies. Because employed workers are usually healthier than the general population, the SMR for cardiovascular disease in employed populations tends to be lower than 1.00 (Fox & Collier, 1976; McMichael, 1976). The absence of a healthy worker effect, in the light of the positive results of experiments with animals, suggests that more detailed analyses should be conducted for cardiovascular outcomes in the exposed populations.

Studies of Viet Nam veterans have reported nonsignificant increases in the relative mortality ratio for circulatory diseases (Fett et al., 1987b; Michalek et al., 1990; Wolfe et al., 1994).

Bertazzi et al. (1989, 1992) reported increased circulatory disease mortality among residents of Seveso Zone A (the most highly contaminated region) in both men (RR, 1.8; 95% CI, 1.0–3.2) and women (RR, 1.9; 95% CI, 0.8–4.2). The study was limited by the small number of subjects and the crude measure of 2,3,7,8-TCDD exposure.

Several other effects of 2,3,7,8-TCDD on the cardiovascular system have been reported (Bond et al., 1983; Moses et al., 1984; Suskind & Hertzberg, 1984; Centers for Disease Control Vietnam Experience Study, 1988a; Roegner et al., 1991; Grubbs et al., 1995). Statistically significant associations with 2,3,7,8-TCDD exposure were found only in the Ranch Hand study for diastolic blood pressure, arrhythmias detected on the electrocardiogram and peripheral pulse abnormalities (Roegner et al., 1991). Significant increases in blood pressure were found in subjects with serum 2,3,7,8-TCDD levels from 15 to 33.3 ng/kg, but not in those with higher levels.

(i) Pulmonary effects

There is conflicting evidence from controlled epidemiological studies regarding an association between chronic respiratory system effects and human exposure to substances contaminated with 2,3,7,8-TCDD. One study of workers involved in production of TCP and 2,4,5-T suggested that 2,3,7,8-TCDD exposure increases the risk for abnormal ventilatory function (Suskind & Hertzberg, 1984). This study found statistically significant decreases in pulmonary function as measured by Spirometric evaluation. However, the exposed workers were, on average, 10 years older than controls and therefore had greater potential exposure to 2,4,5-T formulated in powder form, as used in earlier periods of production, which may have presented a greater risk of impaired lung function than the liquid form used later.

No association between ventilatory function and serum 2,3,7,8-TCDD was found in the NIOSH study (Calvert et al., 1991). The Ranch Hand study found significant declines in the mean FEV₁ and the mean FVC for subjects with serum 2,3,7,8-TCDD levels above 33.3 ng/kg (adjusted mean FEV₁, 91.3%; mean FVC, 87.4%) compared with an unexposed comparison group (adjusted mean FEV₁, 93.5%; mean FVC, 91.7%) (Roegner et al., 1991). Smoking appeared to have a greater influence on lung function than 2,3,7,8-TCDD exposure. In a follow-up examination conducted in 1992, no consistent relationship was found between serum 2,3,7,8-TCDD concentration and respiratory parameters (Grubbs et al., 1995).

(j) Renal effects

There is little evidence in the human data to suggest that exposure to 2,3,7,8-TCDD is related to renal or bladder dysfunction. No major renal or bladder dysfunction was noted among Ranch Hand veterans (Roegner et al., 1991) or among TCP production workers from West Virginia (Suskind & Hertzberg, 1984) or New Jersey (Poland et al., 1971).

(a) Species comparisons of toxic effects

(i) Lethality and other major effects

2,3,7,8-TCDD-induced mortality does not occur immediately after exposure, but only after several days or several weeks, and it is therefore not reasonable to perform animal experiments studying specific organ functions during the lag phase between exposure to potentially lethal doses and death.

A very large number of studies on the acute, subchronic and chronic toxicity of PCDDs and PCDFs has revealed that the toxic outcome of treatment with a certain congener or with a mixture of congeners strongly depends on the species, strain and toxi-cological endpoint investigated. Furthermore, there are remarkable differences in the potency of individual PCDD or PCDF congeners. The situation is further complicated by the fact that the relative potency, and probably the dose–response relationship, of a given congener varies with the experimental system and the parameters of toxicity measured. Finally, the large differences in doses or dose ranges used contribute to the sometimes considerable uncertainty when an ‘effect’ of PCDDs/PCDFs on a certain biological parameter is discussed.

However, there is no doubt that potent PCDDs/PCDFs have a number of characteristic toxic effects in common, which make them an almost unique example for a distinct class of toxic substances. Among the toxic responses consistently observed in all mammalian species tested so far are a progressive loss of body weight, a reduced intake of food, atrophy of the thymus, gastrointestinal haemorrhage and delayed lethality (Safe, 1986; Vanden Heuvel & Lucier, 1993). Other characteristic signs of toxicity are frequently found in the liver, skin and organs of the endocrine system (see Tables 59 and 60 for 2,3,7,8-TCDD data). The most sensitive endpoints for 2,3,7,8-TCDD effects are listed in Table 61.

Table 59

Acute lethality of 2,3,7,8-TCDD in various animal species and strains.

Table 60

Acute toxic responses following exposure to 2,3,7,8-TCDD: species differences.

Table 61

No observed-effect and lowest-observed-effect levels (NOEL and LOEL) of 2,3,7,8-TCDD for mammalian species.

An early report on the acute toxicity of 2,3,7,8-TCDD, 2,7-DCDD, an unspecified HxCDD and OCDD in animals was published by Schwetz et al. (1973). The authors reported the much higher acute toxicity of 2,3,7,8-TCDD compared with HxCDD or OCDD in guinea-pigs and rabbits. Furthermore, the acnegenic, teratogenic and hepatotoxic properties of 2,3,7,8-TCDD and HxCDD were described.

Guinea-pig

Guinea-pigs are among the most sensitive species towards 2,3,7,8-TCDD and other PCDDs and PCDFs, with regard to lethality (Plüss et al., 1988a), with 90% dying from a single 3 µg/kg dose (Harris et al., 1973). The mean time interval until death was 18 days, and a large weight loss over a period of days preceded death. Thymic atrophy was observed after treatment with weekly doses of 0.008 µg/kg 2,3,7,8-TCDD over four to five weeks (Gupta et al., 1973). Other lesions were haemorrhages, hyperplasia of the urinary bladder mucosa and atrophy of the adrenal zona glomerulosa. In a subchronic study, DeCaprio et al. (1986) treated guinea-pigs with various doses of 2,3,7,8-TCDD in the diet for up to 90 days. At a total dose of 0.44 µg/kg, animals exhibited a decreased rate of body weight gain and increased relative liver weights. Male animals also displayed a reduction in relative thymus weights and elevated serum levels of triglycerides, while females showed hepatocellular cytoplasmic inclusion bodies and lowered serum ALT activities. A no-observed-effect level of approximately 0.65 ng/kg per day for prolonged exposure was calculated for these lesions.

After a single lethal dose of ≥ 10 µg/kg 2,3,7,8-TCDD, there was a marked reduction in the size of the thymus, loss of body fat and reduction of muscle mass in guinea-pigs. Histopathological examination revealed epithelial hyperplasia in the renal pelvis, ureter and urinary bladder, hypocellularity of the bone marrow and seminiferous tubules and loss of lymphocytes in thymic cortex, spleen and Peyer's patches, whereas no marked pathological alterations of the liver were evident (Moore et al., 1979). After treatment of guinea-pigs with 20 µg/kg 2,3,7,8-TCDD, hepatocellular hypertrophy, steatosis, focal necrosis and hyalin-like cytoplasmic bodies as well as hepatic focal necrosis were observed in a study by Turner and Collins (1983).

Rat

The single oral LD₅₀ value of four PCDDs in male Sprague-Dawley rats was determined to be 43 µg/kg for 2,3,7,8-TCDD, 206 µg/kg for 1,2,3,7,8-PeCDD, 887 µg/kg for 1,2,3,4,7,8-HxCDD and 6325 µg/kg for 1,2,3,4,6,7,8-HpCDD (Stahl et al., 1992a). A mixture containing all four congeners in a predicted equitoxic amount showed strictly additive toxic effects.

The discovery of a more than 300-fold difference in acute LD₅₀ values for 2,3,7,8-TCDD between the Long-Evans and Han/Wistar rat strains led to a number of studies aimed at elucidating the biochemical basis of this difference. Han/Wistar rats are extraordinarily resistant towards 2,3,7,8-TCDD (Pohjanvirta & Tuomisto, 1987). Doses of 125–1400 µg/kg resulted in marked decreases in food consumption, and in a decrease of 20–25% in body weight. However, the lethality to the animals was low (two of 40 died). A similar strain difference exists for 1,2,3,7,8-PeCDD, whereas it was much less pronounced for 1,2,3,4,7,8-HxCDD (Pohjanvirta et al., 1993). Genetic crossings between Long-Evans and Han/Wistar rats suggest that 2,3,7,8-TCDD resistance is the dominant trait in the rat. Two (or possibly three) genes seem to regulate resistance (Pohjanvirta, 1990). However, even within a given rat strain, such as Long-Evans or Han/Wistar, substrains with markedly differing sensitivity towards 2,3,7,8-TCDD may exist.

Therefore, a critical analysis of 2,3,7,8-TCDD sensitivity in the used animals is required (Pohjanvirta & Tuomisto, 1990a). Most toxic effects are similar in both strains, the Long-Evans strain, however, being more sensitive. In the livers of Long-Evans rats, a (lethal) dose of 50 µg/kg bw 2,3,7,8-TCDD caused marked hepatocyte swelling, vacuoles probably containing fat in the cytoplasm, multinuclear cells (up to eight nuclei) and accumulation of inflammatory cells in hepatic sinusoids (Pohjanvirta et al., 1989a). At the same dose, Han/Wistar rats showed slight cytoplasmic swelling and irregular haematoxylin–eosin staining. Thymic atrophy was seen in both strains, the Long-Evans strain being about 10-fold more sensitive than the Han/Wistar strain. Furthermore, pronounced reduction in serum T4 and increases in TSH, corticosterone and free fatty acids were observed in Long-Evans rats.

In CD rats, moderate thymic atrophy was observed after treatment with 1 µg/kg per day over four to five weeks (Gupta et al., 1973). Other effects were haemorrhages, degenerative changes in kidney and thyroid, increase in megakaryocytes in spleen, lymphoid depletion of the spleen and lymph nodes and severe liver lesions. Female CD rats were slightly more sensitive than males (Harris et al., 1973). A more sensitive indicator of 2,3,7,8-TCDD exposure than body weight reduction was a decrease in thymus weight. Blood coagulation was not significantly affected in five-week-old rats that had received 1 µg/kg 2,3,7,8-TCDD (Bouwman et al., 1992).

In a 13-week oral toxicity study in adult Sprague-Dawley rats (Kociba et al., 1976; Goodman & Sauer, 1992), 1 µg/kg 2,3,7,8-TCDD per day caused some mortality, inactivity, decreased body weights and food consumption, icterus, pathomorphological changes in the liver, lymphoid depletion of the thymus and other lymphoid organs and minimal alterations of some haematopoietic components. Doses of 0.1 µg/kg 2,3,7,8-TCDD per day also caused decreased body weights and food consumption and lymphoid depletion. Effects seen only in males given this dose level included a depression in packed blood cell volume, red blood cells and haemoglobin. Lower doses (0.01 or 0.001 µg/kg 2,3,7,8-TCDD per day) did not affect any of the measured parameters, except for a slight increase in the mean liver-to-body weight ratio in rats given 0.01 µg/kg 2,3,7,8-TCDD per day. A single dose of 1 or 10 µg/kg 2,3,7,8-TCDD led to a pronounced reduction in blood glucose. In female Sprague-Dawley rats, chronic daily intake of 47 ng/kg 2,3,7,8-TCDD resulted in a decrease in plasma concentration of thyroid hormone (total T4) and body weight. More sensitive parameters influenced at 14 ng/kg per day were a decrease in relative thymus weight, loss of hepatic retinoids and induction of CYP1A1 and CYP1A2 activities in the liver (van Birgelen et al., 1995a). Hepatic CYP1 A1 and CYP1A2 were induced at comparable doses of 2,3,7,8-TCDD in both mice and rats (Smialowicz et al., 1994).

Hamster

In the Syrian hamster, 2,3,7,8-TCDD treatment led to loss of body weight and thymic atrophy, while no histopathological changes were seen in the liver, spleen, kidney adrenals or heart (Olson et al., 1980). Oral treatment resulted in moderate to severe ileitis and peritonitis in many animals, associated with marked hyperplasia with mild to severe haemorrhage and necrosis of the mucosal epithelium.

Inter-species rodent comparisons

In a comparative study in rats, mice and guinea-pigs (Zinkl et al., 1973), 2,3,7,8-TCDD was found to cause acute/subacute increases in serum levels of alanine aminotransferase (ALT, previously SGPT) and aspartate aminotransferase (AST, previously SGOT), cholesterol and bilirubin in rats after a single dose of 1 or 10 µg/kg bw. Histopathological examination also suggested that the liver is highly sensitive to 2,3,7,8-TCDD in rats but not in guinea-pigs and mice. Furthermore, significant reductions in numbers of platelets in the peripheral blood of rats and guinea-pigs, and lymphopenia in mice and guinea-pigs were evident.

In male C57BL/6 mice (Gasiewicz et al., 1980) and guinea-pigs, various PCDD congeners were administered as a single oral dose (McConnell et al., 1978a). LD₅₀s of 2 µg/kg in guinea-pigs and 283.7 µg/kg in mice were determined for 2,3,7,8-TCDD. The acute toxicity of 2,3,7,8-substituted PCDD was in the following rank order: 1,2,3,4,6,7,8 < 1,2,3,7,8,9 = 1,2,3,6,7,8 = 1,2,3,4,7,8 < 1,2,3,7,8 < 2,3,7,8. In both species, 2,3,7,8-TCDD strongly reduced the thymus weight due to a reduction in cortical lymphocytes. Significant macroscopic or histopathological hepatic effects including porphyria were observed only in the mouse. Hyperplasia of the transitional epithelium in the urinary tract was found in guinea-pigs.

The relative toxicity of four different congeners was found by Rozman et al. (1993) to be similar in guinea-pigs and Sprague-Dawley rats following acute, subchronic or chronic dosing. Furthermore, the dose–response relationships for 2,3,7,8-TCDD concerning the endpoints mortality, porphyria and carcinogenicity were very similar. Therefore, the authors suggested that the product of average tissue concentration and time can describe the toxicity for a given congener.

Rhesus monkey

In rhesus monkeys (Macaca mulatta) fed toxic fat containing 2,3,7,8-TCDD, gastric hyperplasia and ulceration, hydropericardium, ascites, reduced spermatogenesis, focal liver necrosis, decreased haematopoiesis, skin lesions and eventual mortality were observed (Allen et al., 1977). In an earlier study (Norback & Allen, 1973), atrophy of the lymph tissue, liver enlargement and a reduction in the number of spermatocytes in the testes were also reported.

Following administration of a single dose of 70 µg/kg 2,3,7,8-TCDD by gastric instillation to rhesus monkeys, McConnell et al. (1978b) observed weight loss, blepharitis, loss of fingernails and eyelashes, facial alopecia with acneform eruptions, mild anaemia, neutrophilia, lymphopenia and a decrease in serum cholesterol with an increase in serum triglyceride concentrations. The liver, adrenal gland and kidney showed increased relative organ weight, whereas the thymus was dramatically reduced in size due to a loss of cortical lymphocytes. Histopathological examination revealed hyperplastic and metaplastic changes in sebaceous glands, especially in the eyelid (meibomian glands) and ear canal, together with epithelial hyperplasia in the renal pelvis, stomach, gall-bladder and bile duct, whereas the histology of the liver parenchyme was relatively normal.

Cachexia

One of the most striking effects of 2,3,7,8-TCDD in an mammalian species tested so far is body weight loss combined with a reduced intake of food. From a pair-feeding experiment, hypophagia was suggested to be the major factor leading to wasting in guinea-pigs, Fischer 344 rats and C57BL/6 mice (Kelling et al., 1985). However, lethality was only partially due to body weight loss, since the lethality, in particular in rats and mice, was significantly lower among pair-fed controls than among 2,3,7,8-TCDD-treated animals. Similarly, total parenteral nutrition, preventing the body weight loss associated with 2,3,7,8-TCDD toxicity in rats after a single dose of 100 µg/kg bw, did not prevent death (Gasiewicz et al., 1980).

Seefeld et al. (1984a) reported that 2,3,7,8-TCDD-treated rats maintained their body weight at a lower, nearly constant percentage of that of control rats fed ad libitum. The authors concluded from this and a number of additional experiments that the target level for body weight is reduced under the influence of 2,3,7,8-TCDD. Hypophagia relative to untreated animals was suggested to be the key symptom of this down-regulation of body weight, whereas malabsorption or less efficient feed utilization did not play a decisive role (Seefeld et al., 1984b). The role of hypophagia in weight loss, and the changes in carbohydrate and lipid metabolism observed after 2,3,7,8-TCDD treatment, were questioned by Chapman and Schiller (1985), who found no influence of 2,3,7,8-TCDD on the cumulative amount of feed consumption by C57BL/6 mice until the animals became moribund. In addition, the alterations in serum glucose and lipid concentrations in fasted mice were very different from those induced by 2,3,7,8-TCDD exposure. In DBA/2J mice bearing a low-affinity Ah receptor, dose–response experiments showed comparable changes in glucose and lipid parameters when the animals were exposed to 10-fold higher doses of 2,3,7,8-TCDD than C57BL/6 mice.

In a series of experiments, Pohjanvirta and Tuomisto (1990b) administered sodium chloride, 2-deoxy-D-glucose, sodium mercaptoacetate, insulin, naloxone, glucose or fructose to 2,3,7,8-TCDD-treated or control rats. 2,3,7,8-TCDD caused an attenuation of feeding responses to metabolic deficits, resulting, for example, in severe hypoglycaemia (Pohjanvirta et al., 1991a), but induced hypersensitivity to peripheral satiety signals. The central nervous system was suggested to play a crucial role in these effects. Transfusion of blood from rats with 2,3,7,8-TCDD-induced appetite suppression into untreated rats did not affect their feed intake, whereas transfusion of blood from normally sated rats did (Rozman et al., 1991). This indicated that 2,3,7,8-TCDD treatment does not increase a satiety-signalling factor in blood but may rather suppress the formation of hunger-related signals. Evidence for the involvement of a serotonergic mechanism was derived from the finding that 2,3,7,8-TCDD increased the levels of tryptophan and its metabolites serotonin and 5-hydroxyindoleacetic acid in blood and brain.

2,3,7,8-TCDD treatment also produced a longer-lasting suppression of feed intake after oral treatment with glucose than in untreated rats (Pohjanvirta et al., 1991b) and induced aversion to eating energy-providing food, irrespective of its type (Pohjanvirta & Tuomisto, 1990c). From another series of experiments in rats, including behavioural, biochemical and antiemetic approaches, it was concluded that nausea cannot explain the lethal wasting syndrome (Pohjanvirta et al., 1994a). Food intake and body weight loss after 2,3,7,8-TCDD treatment of Long-Evans rats did not change after either vagotomy or portacaval anastomosis resulting in a circumvention of the liver and a direct blood flow from the intestine to extrahepatic organs (Tuomisto et al., 1995).

Cachexia observed after 2,3,7,8-TCDD treatment can also be induced with certain cytokines such as the endotoxin-responsive tumour necrosis factor α (TNFα) (Döhr et al., 1994), which led to the suggestion that TNFα might mediate the acute toxicity of 2,3,7,8-TCDD. In fact, mortality following a single 2,3,7,8-TCDD dose of 300 µg/kg in male C57BL/6J mice was significantly reduced by treatment with the anti-inflammatory corticoid dexamethasone or with anti-TNFα antibodies (Taylor et al., 1992). In endo-toxin-non-responsive C3H/HeJ mice, Clar and Taylor (1994) found no trend in body weight loss after a single 2,3,7,8-TCDD dose of 350 µg/kg, while endotoxin-responsive C57BL/6 mice demonstrated a statistically significant decline in body weight. However, pathological effects, such as peritoneal infiltration and hepatotoxicity, were also observable in C3H/HeJ mice.

Mechanistic studies of toxicity

Early attempts to investigate the mechanism of action of 2,3,7,8-TCDD in cell culture were made by Knutson and Poland (1980). In 23 cultured cell types including rat hepatoma cells and primary hepatocytes, mouse fibroblasts and human epithelial cell lines, no morphological abnormalities were observed with 10⁻⁷–10⁻¹¹ M 2,3,7,8-TCDD. Analysis of trypan blue exclusion or cell proliferation in a number of cell types also failed to show any toxicity. Also, aryl hydrocarbon hydroxylase activity was inducible only in a subgroup of cell types including most hepatoma cell lines, some mouse embryo cell lines and primary rat and chicken hepatocytes.

The fundamental role of the Ah receptor for the toxicity of 2,3,7,8-TCDD (see Section 4.3) was demonstrated by Poland and Glover (1979). DBA/2J mice, which have a low-affinity receptor, were approximately 10-fold less sensitive to thymic involution after 2,3,7,8-TCDD treatment than C57BL/6J mice, a result that precisely matches the 10-fold difference in affinity of 2,3,7,8-TCDD for the Ah receptor (Ema et al., 1994). Furthermore, the capacity of other halogenated aromatic hydrocarbons to produce thymic atrophy corresponded to their capacity to bind to the receptor. The same dependence on receptor affinity applied to the production of cleft palate and to the reduction of the adipose-type (type 4) glucose transporter (GLUT) in adipose tissue and of the brain-type transporter GLUTI in brain (Liu & Matsumura, 1995).

Comparison of low-affinity Ah receptor (Ah^d/Ah^d) with congenic wild-type high-affinity (Ah^b/Ah^b) Ah receptor mice revealed LD₅₀ values for 2,3,7,8-TCDD of 3351 and 159 µg/kg, respectively. A similar difference in dose–response was found for the decrease in body weight, increase in liver weight and decreases in the weight of thymus, spleen, testes and epididymal fat pad (Birnbaum et al., 1990). Okey et al. (1995) pointed out the strong genetic link between the Ah receptor and the sensitivity of different mouse strains towards a wide variety of toxic outcomes in vivo such as thymic atrophy, induction of cleft palate and porphyria (Jones & Sweeney, 1980; Poland & Knudson, 1982).

It has been claimed, however, that the discrepancy between the Han/Wistar and Long-Evans rats cannot be explained on the basis of Ah receptor concentration, and that differences in the acute toxicity of 2,3,7,8-TCDD between Long-Evans and Sprague-Dawley rats are not related to Ah receptor-mediated induction of ethoxyrufin-O-deethylase activity (Fan & Rozman, 1994). In a reply, Okey et al. (1995) pointed out that the very diversity of tissue-selective and species-selective responses elicited by 2,3,7,8-TCDD almost requires that the receptor is part of a multicomponent system. Therefore, it is unlikely that the differences in dose–response are related solely to differences in Ah receptor concentrations or affinities in various species or tissues. In particular, conclusions on the relative susceptibility of different species cannot be drawn on the basis of receptor data only.

Fernandez-Salguero et al. (1995) demonstrated that Ah receptor-deficient mice are relatively unaffected by doses of 2,3,7,8-TCDD (2000 µg/kg) 10-fold higher than those found to induce severe toxic and pathological effects in litter mates expressing a functional Ah receptor.

(b) Immunological responses

Immunological responses induced by PCDDs and PCDFs in mammals have been observed for the last 25 years with doses varying over many orders of magnitude. Initial studies were performed with doses in the mg/kg and µg/kg range in mice and rats, but it is now known that alterations of some immune responses can be observed after exposure to ng/kg doses of 2,3,7,8-TCDD in mice and non-human primates.

Immunological responses observed after treatment with doses which cause overt toxicity or even mortality are without relevance to the human situation in relation to environmental or occupational exposures.

Studies in mice and rats performed at doses which induce thymic atrophy should be interpreted with great caution.

(i) Effects of 2,3,7,8-TCDD

The effects of 2,3,7,8-TCDD and other polyhalogenated dibenzo-para-dioxins and polyhalogenated dibenzofurans on the mammalian immune system have been reviewed several times. 2,3,7,8-TCDD causes suppression of both cell-mediated and humoral immunity, but little is known about the underlying mechanisms (Vos & Luster, 1989; Holsapple et al., 1991a,b; Vos et al., 1991; Kerkvliet, 1994; Kerkvliet & Burleson, 1994; Holsapple, 1995).

Effects on thymus and role of Ah receptor

Poland and Glover (1979) studied the dose–response relationship for thymic atrophy produced by 2,3,7,8-TCDD in two strains of mice. C57BL/6 mice, which have a high-affinity Ah receptor, were approximately 10-fold more sensitive to thymic involution than DBA/2 mice, which have a lower-affinity receptor.

Germolec et al. (1996) have described CYP1A1 induction in lymphoid tissues from Fischer 344 rats exposed to a single oral dose of 100 µg/kg bw 2,3,7,8-TCDD. Immunohistochemical localization of CYP1A1 in immune tissues indicated that cells other than the lymphoid populations are responsible for the increased CYP1A1 expression.

The presence of the Ah receptor and the protein Arnt was demonstrated in T cells, and a combined exposure to 2,3,7,8-TCDD and the T-cell activator anti-CD3 caused the Ah receptor to translocate to the nucleus, but DNA binding activity of the murine T-cell Ah receptor was not detected (Lawrence et al., 1996).

Ah receptor-deficient mice showed decreased accumulation of lymphocytes in the spleen and lymph nodes, but not in the thymus (Fernandez-Salguero et al., 1995). However, corresponding results were not observed in Ahr⁻Ahr⁻ mice generated by another group (Schmidt et al., 1996).

Kerkvliet and Brauner (1990) showed that, in C57BL/6 mice treated with 2 µg/kg bw 2,3,7,8-TCDD, the percentage of double positive CD4⁺CD8⁺ (DP) thymocytes was decreased, whereas the percentage of double negative CD4⁻ CD8⁻(DN) thymocytes was increased.

In C57BL/6 mice treated with a single intraperitoneal injection of 50 µg/kg bw 2,3,7,8-TCDD, a decrease was seen in cell number in the thymus mainly of the DP and DN populations (Lundberg et al., 1990a; Lundberg, 1991).

Silverstone et al. (1994a) showed that thymic atrophy after a single intraperitoneal dose of 30 µg/kg bw 2,3,7,8-TCDD in BALB/c mice resulted from a proportional loss of all classes of thymocytes. There was no significant relative reduction in terminal deoxynucleotidyl transferase (TdT)⁺ recombinase activating gene (RAG-1)⁺ cells in the thymus, but a slow and persistent reduction of TdT and RAG-1 in bone marrow [TdT and RAG-1 are markers for lymphocyte stem cells].

A single intraperitoneal 2,3,7,8-TCDD dose of 30 µg/kg bw to sham-operated or neonatally thymectomized BALB/c mice reduced the bone marrow levels of mRNA for TdT and RAG-1. Thus, neonatal thymectomy had no effect on the 2,3,7,8-TCDD-elicited reduction of TdT or RAG-1 mRNAs. Corresponding effects of 2,3,7,8-TCDD were also observed in athymic nu/nu mice (Frazier et al., 1994a).

Severe combined immunodeficient C.B-17 scid/scid (SCID) mice engrafted with human fetal thymus and liver tissue fragments (SCID-hu mice) were used to assess the sensitivity of the human thymus to 2,3,7,8-TCDD. The relative size of the cortex showed a dose-dependent decrease in grafted human thymus as well as in rat thymus, which was significant after exposure to 25 µg/kg 2,3,7,8-TCDD. A dose-dependent increase in keratinization of Hassal's corpuscles in the medullary areas of the thymus grafts was observed (de Heer et al., 1995)

Muralidhara et al. (1994) observed decreased activity of the enzyme adenosine deaminase in thymic tissue of BALB/c mice (but not in DBA/2 mice) after treatment with a single intraperitoneal injection of 28.8, 57.5 or 115 µg/ kg bw 2,3,7,8-TCDD or more. The lowest dose tested (11.5 µg/kg) did not cause a significant reduction in the enzyme activity.

Thymocytes from 15-day-old C57BL/6 mice, treated with 50 µg/kg bw 2,3,7,8-TCDD four days before sacrifice, showed an earlier response and a higher maximal cell proliferation than thymocytes from control mice upon stimulation with concanavalin A in vitro (Lundberg et al., 1990b).

The effects of 2,3,7,8-TCDD on the murine fetal thymus were studied in an organ culture system. A concentration of 5 × 10⁻¹⁰ M caused a 50% inhibition of lymphoid development. At 10⁻⁹ M, reduction of DP cells was most pronounced (Dencker et al., 1985; d' Argy et al., 1989; Lundberg et al., 1990a).

Greenlee et al. (1985) studied the effects of 2,3,7,8-TCDD on primary cultures of thymic epithelial cells from C57BL/6 mice. Treatment of the cultures with 10 nM 2,3,7,8-TCDD resulted in altered maturation of cocultured thymocytes as judged by the suppression (40% of control) of thymic epithelium-dependent responsiveness of thymocytes to the mitogens concanavalin A and phytohaemagglutinin.

With human thymic epithelial cells, marked differences were seen in the sensitivity of the cells from different donors to 2,3,7,8-TCDD. Cytochrome P450 activities were inducible in these cells in vitro (EC₅₀ value, approximately 1 nM), with maximal increases in 7-ethoxycoumarin-O-deethylase (ECOD) and EROD activities of 3–00-fold and 1–21-fold, respectively (Cook et al., 1987).

Thymic atrophy in rats following exposure to 2,3,7,8-TCDD (at doses of 1 and 10 mg/kg bw) was first described in the Wistar strain by Buu-Hoi et al. (1972).

In Fischer 344 rats, Rice et al. (1995) observed a significant reduction in the relative thymus weight (thymus weight/body weight) after a single intraperitoneal injection of 0.3 µg/kg bw 2,3,7,8-TCDD. de Heer et al. (1994a,b) reported a significant reduction in the number of immature CD4⁺CD8⁺ thymocytes after single oral doses of 1 µg/kg bw 2,3,7,8-TCDD or more in Wistar rats. Numbers of mature CD3 medullary thymocytes were not affected at doses of up to 25 µg/kg bw 2,3,7,8-TCDD. A detailed study of the time course of the effect after treatment with 25 µg/kg 2,3,7,8-TCDD revealed a recovery in proliferative activity in the thymic cortex (after six days) and an increase in cellularity after day 13.

Kurl et al. (1993) studied the time course of events which precede 2,3,7,8-TCDD-induced thymic apoptosis. They showed that, in thymocytes from immature rats incubated in vitro, nuclear accumulation of 2,3,7,8-TCDD reaches maximal levels within 60 min, paralleling 2,3,7,8-TCDD-induced increases in RNA synthesis and poly(A)poly-merase activity.

Pronounced thymic atrophy was induced in PVG rats by treatment with a single dose of 50 µg/kg bw 2,3,7,8-TCDD. Thymus lobes were transplanted into control rats and evaluated 20 days later, when they did not differ from controls, indicating that the 2,3,7,8-TCDD-induced damage in rat thymus is rapidly reversible (van Loveren et al., 1991).

de Waal et al. (1992, 1993) investigated rat thymus by immunohistochemistry and electron microscopy after treatment with single doses of 50 and 150 µg/kg bw 2,3,7,8-TCDD. They observed changes which mainly affected the cortical epithelial cells, but, because both dose levels used in this study were lethal to the animals, the relevance of the results is doubtful.

Cytotoxic T-cells

Doses of 4 ng/kg bw 2,3,7,8-TCDD altered the ability of adult male C57BL/6 mice to generate cytotoxic T-lymphocytes (CTL) in response to alloantigen challenge. The cellular basis of the 2,3,7,8-TCDD-induced suppression was shown to be an enhanced suppressor T-cell activity of CTL responses, whereas CTL precursers and IL-2 production appeared to be intact in 2,3,7,8-TCDD-exposed mice. CTL activity generated in vitro following allogenic stimulation was not impaired when spleen cells from 2,3,7,8-TCDD-treated DBA/2 mice were used (Clar et al., 1981, 1983; Nagarkatti et al., 1984). However, these results at very low dose levels could not be corroborated by other investigators.

Hanson and Smialowicz (1994) specifically designed a study to re-evaluate the effect of 2,3,7,8-TCDD on CTL response as described by Clar et al. (1981). Neither the in-vivo- nor the in-vitro-generated CTL response was altered following a single intraperitoneal injection of 2,3,7,8-TCDD at doses ranging from 0.24 to 7.2 µg/kg bw. Also, no effect was observed on the in-vivo-generated CTL response following four weekly exposures to 2,3,7,8-TCDD at doses of 0.01 to 3.0 µg/kg bw. Similarly, 2,3,7,8-TCDD did not alter the in-vitro-generated CTL response at these dose levels. [The Working Group noted that the sex of the C57BL/6 mice was different (female instead of male), but it is unlikely that this accounts for the discrepancy in the results.]

Kerkvliet et al. (1990a) observed a significant suppression of CTL response in C57BL/6 mice (Ah^b/Ah^b) at doses of 5–20 µg/kg bw 2,3,7,8-TCDD. Ah^d/Ah^d mice were less susceptible.

Single oral doses of 2.5–40 µg/kg 2,3,7,8-TCDD suppressed the activity of CTL in C57BL/6 mice with a calculated ID₅₀ (50% immunosuppressive dose) of 7.2 µg/kg. Glucocorticoid levels (corticosterone) were not altered at doses below 40 µg/kg, indicating that 2,3,7,8-TCDD-induced CTL suppression is not dependent on elevated glucocorticoid levels (de Krey & Kerkvliet, 1995).

Rice et al. (1995) treated Fischer 344 rats with single oral doses of 2,3,7,8-TCDD up to 30 µg/kg bw and examined CTL activities 24 days following treatment. Syngenic in-vivo tumour-specific CTL were generated that model cell-mediated immune reactions against neoplastically transformed self antigens. Under these conditions, CTL activity showed no significant dose-dependent alteration due to 2,3,7,8-TCDD exposure, but relative thymus weight was significantly decreased at the lowest dose studied (0.3 µg/kg bw).

Effects on lymphocytes in vivo

Oughton et al. (1995) studied the effects of 2,3,7,8-TCDD on lymphocytes by flow cytometry. Female C57BL/6 mice were treated weekly with 200 ng/kg bw 2,3,7,8-TCDD for 14–15 months. Besides an age-matched vehicle control, a group of four-month-old mice was evaluated to assess alterations associated with ageing. In the thymus of the 2,3,7,8-TCDD-treated mice, the proportion of CD4⁻CD8⁻ cells was increased, as was the proportion of gamma-delta thymocytes. The most definite change in 2,3,7,8-TCDD-treated mice was a decrease in the frequency of memory T helper cells, defined as CD4⁺Pgp-1^hiCD45RB^lo, with a concomitant increase in the proportion of naive T helper cells identified as CD4⁺Pgp-1^loCD45RB^hi. [Pgp-1 and CD45RB are the murine analogues to the human markers CD29 and CD45RA. The results of this paper are consistent with the findings in marmosets described below.]

Following a single subcutaneous injection of 10 ng/kg bw 2,3,7,8-TCDD into four mature female marmosets, a pronounced decrease was observed in the number of cells from a defined lymphocyte subpopulation carrying both the CD4 and the CD29 epitopes (‘helper-inducer’ or memory cells). Also, the percentage of pan-B cells (B1, CD20⁺) was clearly lower than that in the controls (Neubert et al., 1990b).

In a 42-week study with lower doses, an opposite effect was initially observed, whereas the results of the first study were confirmed after an increase of the weekly dose. The study consisted of three parts: (a) weekly subcutaneous doses of 0.3 ng/kg bw 2,3,7,8-TCDD were given to seven marmosets for 24 weeks; (b) the weekly dose was then raised to 1.5 ng/kg bw 2,3,7,8-TCDD for another six weeks; (c) finally, this was followed by a dose-free period of 12 weeks. An increase in the percentage as well as the absolute number of CD4⁺CD29⁺ cells was observed after the 24-week treatment period with weekly injections of 0.3 ng/kg bw 2,3,7,8-TCDD. In the second part of the study, this lymphocyte subpopulation decreased after weekly injections of 1.5 ng/kg bw 2,3,7,8-TCDD. The effect was completely reversed during the 12-week recovery period. In addition, several changes were observed in other lymphocyte subpopulations: there was a change in CD4⁺CD45R⁺ cells contrary to that of the helper-inducer cells. Furthermore, a transient increase in CD8⁺CD56⁺ cells (‘cytotoxic T-cells’) was found, and a decrease in pan-B cells (CD20⁺), which were again reversible after discontinuation of the dosing (Neubert et al., 1992a,b).

After subcutaneous injection of a single dose of 300 ng/kg bw 2,3,7,8-TCDD into three marmosets, the number of total lymphocytes decreased and at doses of 1 µg/kg or more the number of total leukocytes in peripheral blood of two marmosets was significantly reduced (Neubert et al., 1993b).

Effects on lymphocytes in vitro

Neubert et al. (1991) studied the effects of 2,3,7,8-TCDD on pokeweed mitogen-stimulated proliferation and differentiation of peripheral lymphocytes in vitro with cells from a marmoset and from two healthy human donors. They observed a pronounced decrease in the percentage of B-cells (CD20⁺) and CD4⁺ cells and a concomitant increase in the percentage of CD8⁺ cells at dose levels as low as 10⁻¹²-10⁻¹⁴ M 2,3,7,8-TCDD.

Using a similar experimental approach, Lang et al. (1994) exposed human lymphocytes from peripheral blood in vitro to concentrations of 10⁻⁷-10⁻¹⁴ M 2,3,7,8-TCDD. Cells were stimulated by pokeweed mitogen or anti-CD3 monoclonal antibody. Analysis of surface markers (e.g., CD4⁺, CD29⁺, CD19⁺) by flow cytometry showed no significant alterations under the given conditions.

Effects on antibody-producing cells in vivo

Intraperitoneal injections of single 2,3,7,8-TCDD doses ranging from 1.2 to 30 µg/kg bw suppressed primary antibody production to SRBC plaque-forming cell (PFC) response in C57BL/6J and in less sensitive DBA/2 mice (Vecchi et al., 1980, 1983).

A single oral 2,3,7,8-TCDD dose ranging from 0.2 to 5 µg/kg bw given two days before SRBC sensitization produced a suppression of anti-SRBC PFC response on day 5 in C57BL/6 mice; the ID₅₀ was 0.74 µg/kg (Kerkvliet & Brauner, 1990).

Narasimhan et al. (1994) determined the splenic antibody PFC response to SRBC nine days after administration of 2,3,7,8-TCDD to B6C3F1 mice. Suppression of PFC response expressed per spleen or per 10⁶ cells was observed at 100 ng/kg bw 2,3,7,8-TCDD and higher doses (results with 50 ng/kg bw 2,3,7,8-TCDD were not significantly different from controls).

Kerkvliet et al. (1990b) compared the effects of 2,3,7,8-TCDD in two congenic strains of C57BL/6 mice that differed at the Ah locus. Using the haemolytic antibody isotope release assay, both Ah^b/Ah^b and Ah^d/Ah^d mice exhibited dose-dependent suppression of the anti-trinitrophenyl (TNP) response following 2,3,7,8-TCDD exposure (T-cell-independent). The ID₅₀ values were 7.0 µg/kg (Ah^b/Ah^b mice) and 30.8 µg/kg (Ah^d/Ah^d mice). Suppression of the antibody response to the T-cell-dependent antigen SRBC occurred at lower doses (0.6 µg/kg bw 2,3,7,8-TCDD in Ah^b/Ah^b mice) and the reaction in Ah^d/Ah^d mice showed an apparent biphasic dose–response relationship.

When spleen cells from mice treated two days previously with 5 µg/kg bw 2,3,7,8-TCDD were transferred to an irradiated host, no suppression of the PFC response to SRBC was observed, while a similar dose in a non-irradiated animal produced profound suppression of the PFC response. Cultures of spleen cells from SRBC-primed 2,3,7,8-TCDD-treated (5 µg/kg) C57BL/6 mice produced fewer anti-TNP PFC when immunized with TNP-SRBC, as compared to cells from primed vehicle-treated controls (Tomar & Kerkvliet, 1991).

Dooley and Holsapple (1988) showed that the B lymphocyte is the primary target for suppression of the T-dependent antibody response in B6C3F1 mice by demonstrating that the suppression of several humoral responses (the polyclonal response to lipopolysaccharide (LPS), the T-cell-independent response to dinitrophenyl (DNP)-Ficoll and the T-cell-dependent response to SRBC) was characterized by dose–response-effect curves that were approximately parallel. They used séparation/reconstitution assay techniques to show that the suppression of antibody responses to 2,3,7,8-TCDD is the predominant result of a specific effect on the functional capacities of the B-cell.

Dooley et al. (1990) treated B6C3F1 mice orally on five consecutive days with daily doses of 1 µg/kg bw 2,3,7,8-TCDD. They observed no effect of 2,3,7,8-TCDD exposure on [³H]thymidine uptake in splenocytes of these mice 24 h after in-vitro stimulation with the T-cell mitogen concanavalin A. Titration of T-cells from 2,3,7,8-TCDD-treated mice into naïve splenocyte cultures did not suppress the humoral response to either a T-cell-dependent (SRBC) or -independent (DNP-Ficoll) antigen. Data suggest that an alteration of T-cell function following 2,3,7,8-TCDD exposure does not play a role in the suppression of the antibody response elicited by antigen stimulation of splenocytes from B6C3F1 mice.

Morris et al. (1992) showed that the sensitivity of DBA/2 mice to suppression of the antibody response increased significantly when 2,3,7,8-TCDD was administered daily over two weeks rather than as an acute exposure. This change in sensitivity was not paralleled by a shift in the sensitivity to other effects of 2,3,7,8-TCDD, including thymic atrophy and liver enzyme induction.

Smialowicz et al. (1994) studied the effects of 2,3,7,8-TCDD on the antibody PFC response to SRBC comparatively in B6C3F1 mice and in Fischer 344 rats. Their data for mice were in agreement with the results from other laboratories (ED₅₀, 0.7 µg/kg bw 2,3,7,8-TCDD). In contrast, 2,3,7,8-TCDD failed to suppress, and in fact enhanced, the PFC response to SRBC in rats at doses of 3 and 30 µg/kg bw. Flow cytometric analysis of thymocytes and splenic lymphocytes from 2,3,7,8-TCDD-dosed (3, 10 and 30 µg/kg bw) and SRBC-immunized mice and rats revealed that CD4⁻CD8⁺ splenocytes were reduced and IgM⁺ splenocytes were increased in a dose-related manner in rats only.

Corresponding experiments were performed with a T-cell-independent antigen TNP-LPS in B6C3F1 mice and in Fischer 344 rats. The dose of 2,3,7,8-TCDD required to suppress the immune response in rats was higher (30 µg/kg) than in mice (10 and 30 µg/kg). Thus, species differences were demonstrable also under these conditions; however, they were not as pronounced as with the T-cell-dependent antigen SRBC (Smialowicz et al., 1996).

Harper et al. (1993) studied the effects of a single intraperitoneal injection of 2,3,7,8-TCDD on the suppression of the splenic PFC response to the T-cell-independent antigen TNP-LPS in C57BL/6 and DBA/2 mice. The ED₅₀ values (PFC/10⁶ viable cells) were 1.5 µg/kg bw (C57BL/6 mice) and 9.7 µg/kg bw (DBA/2 mice).

Effects on antibody-producing cells in vitro

Tucker et al. (1986) observed a direct effect of 2,3,7,8-TCDD on cultured lymphcytes resulting in a selective inhibition of the differentiation of B-cells into antibody-secreting cells. Using lymphocytes from congenic mice differing only at the Ah locus, it was determined that the Ah^b/Ah^b-derived cells were inhibited by 2,3,7,8-TCDD in vitro, whereas the Ah^d/Ah^d-derived cells were not.

Holsapple et al. (1986a) reported a reduction in the number of antibody-producing murine spleen cells which developed in response to LPS, DNP-Ficoll and SRBC with 2,3/7,8-TCCD at concentrations ranging from 5 to 20 nM. Direct addition of 2,3/7,8-TCDD had no effect on mitogen-induced proliferation. There was no suppression when 2,3,7,8-TCDD was added to the medium 3 h after LPS (200 µg/mL). 2,3,7,8-TCDD suppressed the antibody response of cells from DBA/2 mice at concentrations similar to those required to suppress the cells from B6C3F1 mice, suggesting that the observed effect is independent of the Ah locus.

In agreement with these results, Davis and Safe (1991) reported that 2,3,7,8-TCDD and other congeners produced a similar concentration-dependent suppression of the in-vitro anti-SRBC response using cells from either C57BL/6 mice or DBA/2 mice.

Morris et al. (1991) showed that, with spleen cells from B6C3F1 mice, the suppression of the in-vitro T-cell-dependent humoral immune response by 2,3,7,8-TCDD is dependent on the lot of serum used. Only three of 23 commercial lots supported a full dose-responsive suppression.

Morris et al. (1994) compared the influence of calf serum and mouse serum on the results of in-vitro splenocyte antibody response experiments with cells derived from B6C3F1 or DBA/2 mice. With calf serum, a similar degree of suppression of antibody responses by 2,3,7,8-TCDD in splenocytes from both strains was found. In contrast, responses in the presence of mouse serum showed an Ah receptor-dependence that was characterized by a dose-related suppression of antibody responses by B6C3F1 splenocytes only.

Morris and Holsapple (1991) observed an increase in proliferation of dense resting B-cells in the presence of 30 and 60 nM 2,3,7,8-TCDD in the medium.

When 2,3,7,8-TCDD was added at concentrations between 0.3 and 30 nM to activated low-density B-cells isolated from whole spleen cell suspensions from B6C3F1 mice, suppression of cell proliferation and antibody response was observed. In contrast, neither the proliferation nor the antibody response of high-density B-cells (small resting cells) stimulated with LPS was affected by 2,3,7,8-TCDD (Morris et al., 1993).

Wood et al. (1992) studied the effects of 2,3,7,8-TCDD on cultured murine splenocytes ((C57BL/6 × C3H)F1 mice) and compared the effects with those seen with human tonsillar lymphocytes. 2,3,7,8-TCDD at concentrations between 0.3 and 30 nM caused a significant reduction in the proliferation of both human and murine cells; however, the substance had no effect on pokeweed mitogen-induced proliferation or antibody production.

In low-density human B-cells (isolated from human tonsils), 2,3,7,8-TCDD suppressed background proliferation and IgM secretion at concentrations between 0.3 and 30 nM. 2,3,7,8-TCDD-induced suppression was less pronounced when cells were stimulated with LPS or T-cell replacing factor. 2,3,7,8-TCDD did not alter background or stimulated proliferation of high-density human B-cells (Wood et al., 1993).

Wood and Holsapple (1993) examined the effects of 2,3,7,8-TCDD upon human tonsillar lymphocytes stimulated with toxic shock syndrome toxin-1. 2,3,7,8-TCDD at concentrations < 30 nM had no effect upon T-cell or B-cell proliferation, but B-cell differentiation, as manifested by IgM secretion, was significantly suppressed at all concentrations tested (0.3–30 nM). The sensitivity to 2,3,7,8-TCDD varied considerably, with IC₅₀ values ranging from < 0.3 nM to 25 nM with cells from four different donors.

Karras and Holsapple (1994a,b) observed inhibitory effects of 2,3,7,8-TCDD on proliferative responses of murine low-density B-cells to activation by anti-IgM.

In another study, Karras et al. (1996) reported that 2,3,7,8-TCDD suppressed murine B-cell IgM secretion induced by either soluble or insolubilized anti-IgM plus lymphokines, but did not affect IgM secretion stimulated by activated T_H-cells and lymphognes. Their data indicate that 2,3,7,8-TCDD elevates resting intracellular calcium levels in murine B-cells and may selectively inhibit calcium-dependent signalling pathways linked to surface immunoglobulin.

Karras et al. (1995) investigated whether the immunosuppression mediated by direct exposure of murine B-cells to 2,3,7,8-TCDD in vitro is due to an IL-4-like biological activity. However, 2,3,7,8-TCDD failed to demonstrate any of the activities of IL-4 observed in parallel cultures.

Masten and Shiverick (1995) found that 25 nM 2,3,7,8-TCDD decreased steady-state levels of CD19 mRNA by 67% in a human B-lymphocyte cell line (IM-9). They identified a DNA-binding complex in nuclear extracts that appeared to be the Ah receptor. The Ah receptor complex recognized a DNA-binding site for B-cell lineage-specific activator protein in the promoter region of the human CD19 gene that was similar to the Ah receptor DNA-binding site.

Effects on macrophages, neutrophils and natural killer (NK) cells

Mantovani et al. (1980) found no alteration in macrophage-mediated or NK cell-mediated cytotoxicity in C57BL/6J mice after treatment with single doses of up to 30 µg/kg bw 2,3,7,8-TCDD.

Kerkvliet and Oughton (1993) showed that, in C57BL/6 mice, the inflammatory response following intraperitoneal injection of SRBC was aggravated when mice were treated with 5 µg/kg bw 2,3,7,8-TCDD compared with vehicle-treated controls. The increased number of peritoneal cells reflected significant increases in both neutrophils and macrophages. 2,3,7,8-TCDD treatment did not significantly alter expression of the macrophage activation markers F4/80 or I-A on Mac-1⁺ cells. The antigen-presenting function of the peritoneal exudate cells was unaltered by 2,3,7,8-TCDD.

Ackermann et al. (1989) observed a reduced cytolytic and cytostatic activity of polymorphonuclear neutrophils (PMN) from B6C3F1 mice (but not those from DBA/2 mice) after a single oral exposure to 5 or 10 µg/kg bw 2,3,7,8-TCDD. Supernatants recovered from PMN cell cultures of B6C3F1 mice (but not those of DBA/2 mice) showed reduced killing capacity for actinomycin D-treated L929 tumour cells.

Funseth and Ilbäck (1992) treated male A/J mice with a loading dose of 5 µg/kg bw 2,3,7,8-TCDD and three weekly maintenance doses of 1.42 µg/kg, given intraperitoneally. NK cell activity increased significantly in blood and spleen (3.4-fold and 2.2-fold, respectively). The effects were still present on day 120 after the treatment.

Effects on popliteal lymph nodes after local stimulation

In C57BL/6 mice, a single intraperitoneal injection of 50 µg/kg bw 2,3,7,8-TCDD suppressed the normal immune response in popliteal and inguinal lymph nodes to ovalbumin injected into the hind foot pads four days after 2,3,7,8-TCDD treatment. Increase of the lymph node cell number was inhibited and a reduced frequency of antigen-specific B-cells was observed. The antigen-specific T-cell proliferation and IL-2 production in response to ovalbumin were significantly suppressed by 2,3,7,8-TCDD (Lundberg et al., 1991, 1992).

An anti-CD3 monoclonal antibody was injected into both rear footpads of female C57BL/6 mice and the draining popliteal and inguinal lymph nodes were removed 24 h later. 2,3,7,8-TCDD enhanced anti-CD3-induced [³H]thymidine incorporation and increased the percentage of both CD4⁺ and CD8⁺ cells cycling in S and G₂M phases. The authors concluded that 2,3,7,8-TCDD appeared to be targeting T-cells undergoing activation rather than resting cells (Neumann et al., 1993).

Schmidt et al. (1992) treated NMRI mice with single subcutaneous injections of up to 3 µg/kg bw 2,3,7,8-TCDD. One week later they injected streptozotocin or other stimulants into one hind foot pad of the animals. Lymph node enlargement (weight or cell number) in 2,3,7,8-TCDD-treated mice was not significantly different from that of controls.

Korte et al. (1991a) and Stroh et al. (1992) used the popliteal lymph node assay to investigate effects of 2,3,7,8-TCDD on immune reactions in Wistar rats. Animals were treated with single subcutaneous injections of 2,3,7,8-TCDD at doses of up to 600 ng/kg or 3 µg/kg bw, respectively. Seven days later, either erythrocytes or streptozotocin were injected into one hind foot pad and the weight and cell number of the lymph node were determined after a further seven days. 2,3,7,8-TCDD had no significant effect on the results of this test under the given conditions.

Fan et al. (1995) injected 2,3,7,8-TCDD directly into one foot pad of Sprague-Dawley rats (approximately 10 µg/kg bw) and observed a slightly increased weight index (1.59 ± 0.2 versus 1.07 ± 0.2 in controls; mean ± SEM; n = 4 and 6).

Theobald et al. (1983) first reported that the paw oedema formation after subplantar injection of carrageenan was enhanced in Sprague-Dawley rats treated with 2,3,7,8-TCDD. The ED₅₀ of 2,3,7,8-TCDD for this effect was 6 µg/kg bw.

Katz et al. (1984) injected various irritants into the subplantar surface of one hind paw of Sprague-Dawley rats or C57BL/6 mice pretreated with 10 µg/kg bw 2,3,7,8-TCDD. 2,3,7,8-TCDD increased the oedemagenic potency of carrageenan, dextran, bradykinin and histamine, but not that of prostaglandin E2 or serotonin.

Effects on complement

Treatment of B6C3F1 mice with daily doses of 0.01–2.0 µg/kg bw 2,3,7,8-TCDD for two weeks suppressed serum total haemolytic complement activity (CH50). Serum levels of complement component C3 were decreased at daily doses of 0.5 µg/kg or more. Immediately after treatment with single doses of 10–40 µg/kg bw 2,3,7,8-TCDD, a rapid, but transient, dose-dependent increase in liver intracellular C3 levels was observed. No inhibitory effect of 2,3,7,8-TCDD on C3 production was detected when cells of a hepatoma cell line (Hepa 1c1c7) were exposed to 2,3,7,8-TCDD (White et al., 1986; Lin & White, 1993a,b,c).

Host resistance assays

Host resistance assays are often used for assessing effects of xenobiotics on the mammalian immune system, because more than one aspect of the specific and non-specific defence mechanisms are included. Several authors have described the potential of PCDDs to suppress resistance to bacterial, viral, parasitic and neoplastic challenges in mice.

Acute exposure of adult mice to 2,3,7,8-TCDD results in hypersensitivity to endotoxin (Vos et al., 1978; Rosenthal et al., 1989). Clar et al. (1991b) investigated the effects of 2,3,7,8-TCDD on the endotoxin-induced release of TNFoc. They observed a significant increase in TNF± in the serum of endotoxin-exposed mice (Ah^b/Ah^b) after exposure to a single oral dose of 10 µg/kg bw 2,3,7,8-TCDD.

Studies of effects of 2,3,7,8-TCDD on host resistance in mice are summarized in Table 62. To date, the finding of Burleson et al. (1996), that a single dose of 2,3,7,8-TCDD at 0.1, 0.05 or 0.01 µg/kg bw resulted in an increase in mortality from Hong Kong influenza virus when B6C3F1 mice were challenged seven days after 2,3,7,8-TCDD administration, represents the most sensitive adverse effect yet reported for 2,3,7,8-TCDD.

Table 62

Effects of 2,3,7,8-TCDD on host resistance in mice.

Only a few host resistance studies have been performed with PCDDs in rats.

Virus-augmented NK activity assessed 48 h after infection in the lung was significantly suppressed in Fischer 344 rats treated with 3, 10 or 30 µg/kg bw 2,3,7,8-TCDD. Significantly higher virus titres were observed on days 2, 3 and 4 after infection in the lungs of rats treated with 10 µg/kg bw (Yang et al., 1994).

Using the parasite Trichinella spiralis in five-week-old Wistar rats that were exposed perinatally to 2,3,7,8-TCDD (one subcutaneous injection of 0.3 or 3 µg/kg bw on day 19 of gestation), no differences were found in the antibody titres or the number of T. spiralis larvae in muscle (Korte et al., 1991b).

In Fischer 344 rats treated with single doses of 30 µg/kg bw 2,3,7,8-TCDD, no evidence for an immunosuppressive effect was seen in the T. spiralis host resistance assay, but proliferative responses of lymphocytes cultured with parasite antigen were enhanced (Luebke et al., 1995). This is in contrast to results obtained with the T. spiralis test in mice from the same laboratory (Table 62). It was shown in this study that infection delayed elimination from the host: infected mice had higher 2,3,7,8-TCDD levels than non-infected mice (Luebke et al., 1994).

A/J mice infected with Coxsackievirus B3 (CB3) had increased tissue concentrations of 2,3,7,8-TCDD in the brain, pancreas, heart, spleen and liver compared with uninfected controls (Funseth & Ilbäck, 1994).

Direct exposure of isolated human erythrocytes to 10 nM 2,3,7,8-TCDD caused a two-fold increase in the infectivity of Plasmodium falciparum after 48 h, when the parasites were in a synchronized state of growth. Additional treatment with sodium orthovanadate, an inhibitor of plasma membrane Ca-ATPase and phosphotyrosine phosphatase, completely blocked the 2,3,7,8-TCDD-induced increase in parasitaemia (Kim et al., 1994).

Pre- and perinatal exposure

Vos and Moore (1974) treated female Fischer 344 rats and C57BL/6 mice with doses of up to 5 µg/kg bw 2,3,7,8-TCDD pre- and postnatally (rats: gestation days 11 and 18, postnatally on days 4, 11 and 18; mice: gestation days 14 and 17, postnatally on days 1, 8 and 15). The high-dose regimen was lethal to 31/34 rat offspring. The authors observed a depletion of lymphocytes in the thymic cortex of the offspring which, histologically evaluated, was not due to lymphocyte destruction. Cellular immunity was impaired. Allograft rejection times were prolonged in rats and mice. Graft-versus-host activity of spleen cells was reduced, as well as the response of rat thymus and spleen cells to phytohaemagglutinin. In four-month-old mice treated with six weekly doses of 25 µg/kg bw 2,3,7,8-TCDD, these effects were not observed.

Fine et al. (1988, 1990a,b) studied the effects of 2,3,7,8-TCDD on lymphocyte stem cell function in BALB/c mice following direct or perinatal exposure (maternal treatment with 10 or 15 µg/kg bw 2,3,7,8-TCDD on day 14 of gestation). In the fetus and the neonate, significant reductions were found in both the biosynthesis and mRNA levels of the lymphocyte stem cell-specific DNA polymerase TdT. Thymic biosynthesis was relatively unaffected, suggesting that alterations of early events of T-cell lymphopoiesis (prothymocytes) can contribute to 2,3,7,8-TCDD-induced thymic atrophy. A slight reduction in the expression of the thymocyte surface marker Lyt-2⁺L3T4⁺ was demonstrated by flow cytometry.

Holladay et al. (1991) and Blaylock et al. (1992) treated pregnant C57BL/6 mice on nine consecutive days (gestation days 6–14) with daily doses of 1.5 or 3.0 µg/kg bw 2,3,7,8-TCDD by gastric instillation (total doses of 13.5 and 27 µg/kg bw 2,3,7,8-TCDD). 2,3,7,8-TCDD treatment (both groups) resulted in significant decreases in fetal thymic weight and the percentage of CD4⁺CD8⁺ fetal thymocytes (DP), as well as significantly increased CD4⁻CD8⁻ (DN) and CD4⁻CD8⁺ thymocytes on gestation day 18 as analysed by flow cytometry. 2,3,7,8-TCDD induced a significant shift in T-cell receptor (TCR) expression of thymocytes with a decrease in alpha-beta-TCR and a concomitant increase in gamma-delta-TCR expression. On postnatal day 6, no significant differences were detectable by flow cytometry; however, a significantly depressed CTL activity was demonstrable until eight weeks postnatally (offspring had been cross-fostered to avoid exposure via milk). There were no significant differences at postnatal weeks 7–8 between controls and 2,3,7,8-TCDD-exposed mice in lymphocyte proliferation to mitogens or antibody PFC response.

Faith and Moore (1977) showed that when Fischer 344 rats were exposed to 2,3,7,8-TCDD prenatally (day 18 of gestation) and postnatally (days 0, 7, 14 of lactation), body weight and thymus/body weight ratios were suppressed up to 145 days. The effects were less pronounced when the rats were exposed only postnatally (days 0, 7, 14 of lactation).

Badesha et al. (1995) fed a 2,3,7,8-TCDD-containing diet to lactating Leeds rats starting on postnatal day 1. Total doses of 0.2, 1.0 or 5.0 µg/kg were administered over a period of 18 days. At the age of 130 days, body weights remained significantly reduced in all three groups of female offspring and the two highest-dose groups of males. Immunocompetence of the offspring was affected: in-vitro T-cell-dependent and T-cell-independent responses and mitogen-induced in-vitro production of IL-1 and IL-2 were suppressed at postnatal day 130.

(a) Endocrine and gonadal effects

Total serum testosterone, LH and follicle-stimulating hormone (FSH) were measured in 248 TCP production workers and in 231 non-exposed neighbourhood controls matched for age and race (Egeland et al., 1994). In linear regression analyses, current serum levels of 2,3,7,8-TCDD were positively related to LH and FSH and inversely related to total testosterone levels after adjustment for potential confounders (age, body mass index, diabetes mellitus, current alcohol consumption, race and smoking status). Similar results were obtained by multiple logistic regression, showing stronger adverse effects in higher-exposure groups. The presence of both low testosterone and high LH was not observed in the same individuals, therefore, the authors interpreted their findings as being suggestive of more subtle alterations in gonadotropins and testosterone than of primary gonadal failure due to PCDD exposure.

In a random subsample of 571 men, from a total of 4462 who were examined, Viet Nam veterans (n = 324) had significantly lower sperm concentrations than the 247 non-Viet Nam veterans (64.8 million/mL versus 79.8 million/mL) and a lower proportion of ‘normal’ sperm heads (57.9% versus 60.8%) (Centers for Disease Control Vietnam Experience Study, 1988a). There was also a doubling of the proportion of men with sperm characteristics (concentration, percentage of motile cells, percentage of morphologically ‘normal’ cells) below the normal range. Ability to father children was not affected. Differences in semen concentration and morphology could not be related to a particular subgroup of veterans or military occupational specialty, to self-reported combat experiences or to herbicide exposure.

In the 1982 examination, 474 Ranch Hand veterans out of 1045 and 532 of the 1224 comparison veterans were included in the analysis of association between serum 2,3,7,8-TCDD and serum testosterone, FSH, LH, testicular abnormality, sperm count, percentage of abnormal sperm and testicular volume. No pattern of consistent or meaningful associations was seen between 2,3,7,8-TCDD body burden and any of the variables studied, when either categorized (% abnormalities) (Henriksen et al., 1996) or analysed as continuous variables (Henriksen & Michalek, 1996). The authors explain that their study and the previous study by Egeland et al. (1994) differ in terms of the exposure circumstances and demographics and conclude that ‘the Ranch Hand exposure was not sufficient to exhibit the associations similar to those seen in the industrial workers, who have higher dioxin body burdens and were exposed over a longer period of time than the Ranch Hand veterans.’

(b) Effects on pregnancy (Table 63)

Table 63

Results of studies examining the effect of 2,3,7,8-TCDD on pregnancy outcomes in humans.

(a) Developmental effects

(b) Functional developmental toxicity

(c) Reproductive effects

Reproductive effects have been reviewed by Allen et al. (1979), Morrissey and Schwetz (1989) and Theobald and Peterson (1994).

Multi-generation studies have been used to assess reproductive effects in both sexes. In a three-generation study in Sprague-Dawley rats, Murray et al. (1979) noted that exposure to a diet providing doses of up to 0.1 µg/kg bw per day 2,3,7,8-TCDD resulted in impaired reproductive capacity in rats as indicated by reductions in fertility and litter size. While they reported a maximal no-effect level of 1 ng/kg bw per day, a different method of statistical analysis indicated a lowest observed effect level of 1 ng/kg bw per day (Nisbet & Paxton, 1982). [The Working Group noted that no sex ratio data were provided.]

Mutations in tumours

DNA was extracted and analysed for activating mutations in H-ras codon 61 arising in hepatocellular adenomas and carcinomas of male and female B6C3F1 and C57BL/6 mice treated with a single intraperitoneal dose of vinyl carbamate (0.005 εM/g bw in saline) or vehicle and then gastric instillation doses of 2,3,7,8-TCDD (2.5 µg/kg bw in corn oil) every two weeks for 52 weeks. Another group received only the vinyl carbamate treatment. Of 45 tumours from B6C3F1 mice treated with 2,3,7,8-TCDD alone, only 23 (51%) had H-ras codon 61 mutations and 70% of these were C→A transitions in the first base. The pattern was similar to that found in spontaneous tumours of this strain and contrasted with the combined vinyl carbamate plus 2,3,7,8-TCDD and the vinyl carbamate alone-treated groups. In these two groups, respectively, there were A→T transitions in the second base in 39/53 (74%) and 17/27 (63%) of the tumours containing activating mutations. Similar results were obtained for the C57BL/6 mouse strain. Thus, 2,3,7,8-TCDD treatment did not change the mutational spectrum arising within H-ras codon 61 of tumours arising either without treatment or as a result of treatment with vinyl carbamate (Watson et al., 1995).

Mixture of P CDD s and/or PCDFs

Mixtures of PCDDs did not enhance mutagenic or recombinogenic effects, but in combination with N-ethyl-N-nitrosourea, positive effects were obtained in the spot test with mice.

Mixtures of PCDDs, PCDFs and coplanar PCBs increased the frequency of sister chromatid exchange formation in human lymphocytes in vitro in the presence or absence of α-naphthoflavone.

(a) Carcinogenesis is a multistep process

Fundamentally there are three ways in which a compound such as a PCDD can influence the process of carcinogenesis (Barrett, 1993):

1. it may induce a heritable mutation (initiation);
2. it may induce a heritable epigenetic change in a critical gene(s);
3. it may increase the clonal expansion of a cell possessing a heritable alteration in a critical gene(s).

(b) Genotoxicity

2,3,7,8-TCDD is considered a ‘non-genotoxic’ substance (see Section 4.5).

(c) Ah receptor

It has been proposed that the broad spectrum of biological responses associated with exposure to 2,3,7,8-TCDD is due to alteration in expression of 2,3,7,8-TCDD-regulated genes mediated by the Ah receptor (Section 4.3). However, even with the same receptor and the same ligand, there are both qualitative and quantitative differences between species. Even though Ah receptor activation is likely to be required for the carcinogenicity of 2,3,7,8-TCDD, its precise role in this process remains unclear.

In analogy to the mouse, two forms of Ah receptor exhibiting about a 4–5-fold difference in binding affinity for 2,3,7,8-TCDD (K_d, ∼ 0.4 nM and ∼ 2 nM, respectively) have been described in humans (Ema et al., 1994).

(d) Effects of 2,3,7,8-TCDD on gene expression

Induction of expression of genes driven by the XRE element (recognized by the Ah receptor), e.g., CYP1A1 and CYP1A2, represents a useful marker for 2,3,7,8-TCDD effects. The broad spectrum of effects of 2,3,7,8-TCDD on hormone and growth factor systems, cytokines and other signal transduction pathways indicates that this substance is a powerful growth dysregulator (see Sections 4.3 and 4.4).

(e) Oxidative damage

In a series of studies by Stohs and colleagues, single treatment of rats or mice with high doses (≥ 50 µg/kg) resulted in increased superoxide anion production by peritoneal lavage cells, lipid peroxidation and DNA single-strand breaks (Stohs et al., 1990; Alsharif et al., 1994b). The relevance of these high dose studies to the carcinogenicity of 2,3,7,8-TCDD is questionable.

In-vitro studies showed that the promotion of transformation of C3H mouse fibroblasts by low non-cytotoxic concentrations of 2,3,7,8-TCDD (1.5 pM) was inhibited by the antioxidants mannitol and vitamins C and A (Wölfle & Marquardt, 1996). Promotion of cellular transformation is a well documented effect of 2,3,7,8-TCDD, as discussed below, and this study provides evidence for an oxidative-stress mechanism for this effect. Another study showed an Ah receptor-dependent formation of 8-hydroxydeoxyguanine (8-OH-dG) adducts in DNA following treatment of the mouse Hepa1c1c7 cell line with 500 pM 2,3,7,8-TCDD for 48 h (Park et al., 1996).

Production of oxidative damage by 2,3,7,8-TCDD is consistent across several different experimental systems, both in vivo and in vitro. The requirement in rats for ovarian hormones in the mechanism of tumour promotion by 2,3,7,8-TCDD (Lucier et al., 1991) is associated with a 2–3-fold higher level of 8-OH-dG DNA adduct formation in intact compared with ovariectomized rats (Tritscher et al., 1996). It has been suggested that this increase in 8-OH-dG DNA adducts is a result of a production of genotoxic metabolites via redox cycling of catechol oestrogens, although there may be alternative sources of reactive oxygen species. The role of these adducts in carcinogenicity has yet to be established.

(f) Cell transformation

In the mouse C3H 10T1/2 embryonic fibroblast cell transformation system (which contains a functional Ah receptor (Okey et al., 1983)), significant increases in foci formation in cells pre-initiated with MNNG were observed following exposure to non-cytotoxic doses of 2,3,7,8-TCDD (4–4000 pM 2,3,7,8-TCDD) in a dose-dependent pattern (Abernethy et al., 1985). 2,3,7,8-TCDD also gave positive results in an in-vitro transformation assay using rat tracheal epithelial cells only after prior initiation of the cells with MNNG and at concentrations of at least 300 pM continuously for seven days (Tanaka et al., 1989). These data support the observations that 2,3,7,8-TCDD acts as a tumour promoter in vivo.

Virally immortalized human foreskin epidermal keratinocytes treated with 100 pM 2,3,7,8-TCDD were transformed, as shown by colony formation in soft agar, foci formation, an increased maximal cell density and a 100% incidence of squamous-cell carcinomas in nude mice when injected with 1 × 10⁷ cells compared with an incidence of zero with cells exposed to 0.1% dimethyl sulfoxide (Yang et al., 1992). Maximal induction of neoplastic transformation occurred at 1 nM, whereas induction of aryl hydrocarbon hydroxylase activity was maximal at 30 nM. Neoplastic transformation of human cells occurs at a concentration similar to that needed for rodent cell transformation.

(g) Cell proliferation and tumour promotion

Since 2,3,7,8-TCDD is not directly genotoxic, it either could be acting to ‘promote’ the development of tumours from previously initiated cells and/or may be causing mutations via an indirect mechanism.

In a rat tumour initiation-promotion protocol experiment, hepatic cell proliferation as measured by BrdU incorporation (labelling index) was decreased at a dose of 3.5 ng/kg 2,3,7,8-TCDD per day and increased an average of three-fold at 125 ng/kg 2,3,7,8-TCDD per day, after 30 weeks of 2,3,7,8-TCDD treatment (Maronpot et al., 1993). Significant interindividual variation was observed in labelling index, with approximately half of the animals exhibiting significantly higher labelling indices than similarly treated animals. Preneoplastic GSTP-positive foci were elevated only at 125 ng/kg per day. Cell proliferation was significantly stronger in initiated than in non-initiated rats. In another study by Stinchcombe et al. (1995), 2,3,7,8-TCDD only marginally affected DNA synthesis in GSTP-positive liver foci after treatment with a dose 100 ng/kg 2,3,7,8-TCDD per day for 115 days. However, in this study, apoptosis in foci was markedly reduced.

(h) Suppression of immune surveillance

Effects on the immune system can have significant effects on the disease process and the manifestation of toxicity in other organ systems. This is particularly relevant to cancer, where a primary effect of 2,3,7,8-TCDD on immune function could secondarily aid in the progression and development of malignancy by allowing genetically altered cells to escape immune surveillance. There is a wealth of information concerning immune suppression in laboratory animals, but results on immune function in human studies are inconsistent (see Sections 4.2.1 and 4.2.2(b)).

(a) Liver

(b) Other target tissues

2,3,7,8-TCDD-induced carcinogenesis has been reported in a number of other tissues including lung, nasal ethnoturbinates, thyroid, lymphoid tissues, skin and tongue (see Section 3). Ah receptor expression and receptor-dependent responses have been observed in many of these tissues (see Section 4.3) and may play a role in carcinogenesis. A study in congenic strains of mice and using a number of isomers has indicated a role for the Ah receptor in PCDD-induced skin papilloma (Poland et al., 1982).

In the case of thyroid carcinogenesis, an indirect mechanism has been proposed involving enhanced metabolism of thyroid hormones in the liver (see Section 4.3) (Hill et al., 1989; Kohn et al., 1996).

(a) Ah receptor

The Ah receptor is a ubiquitous transcription factor found in both rodents and humans. PCDDs bind to human and rodent Ah receptors with very similar structure–activity relationships; 2,3,7,8-TCDD has the highest affinity of the PCDDs for both rodent and human receptors.

Both in humans and in mice, two forms of Ah receptor have been identified which exhibit a 5–10-fold difference in binding affinity for 2,3,7,8-TCDD. In humans, one form of the Ah receptor exhibits a K_d (a measure of binding affinity) for 2,3,7,8-TCDD of 0.4 nM, whereas the other form binds 2,3,7,8-TCDD with a K_d of about 2 nM.

In congenic mouse strains, expression of the lower or higher affinity forms of receptor has been extensively demonstrated to result in proportional differences in sensitivity to 2,3,7,8-TCDD with regard to biochemical changes and toxic effects. Thus, congenic mice expressing the lower-affinity form of receptor require higher doses of 2,3,7,8-TCDD to elicit these effects than strains expressing higher-affinity forms. A similar difference in sensitivity to PCDDs has also been demonstrated in tumour promotion studies in skin of congenic mouse strains. In these studies, PCDDs show the same rank order of potency in Ah receptor binding in vitro and tumour induction in vivo. Taken together, these data strongly support a receptor-mediated mechanism of mouse skin carcinogenesis.

(b) Gene expression

The best studied 2,3,7,8-TCDD-dependent gene expression response is the induction of CYP1A1 and CYP1A2. In both rodent and human cells, this response is mediated by the Ah receptor. In rodent and human cells, PCDDs show very similar potencies in inducing CYP1A1 and CYP1A2 expression in rodent and human cells. The role, if any, of the induction of these genes in carcinogenesis by 2,3,7,8-TCDD is unclear. 2,3,7,8-TCDD-induced gene regulatory responses and biochemical effects documented in rodent tissues and/or cells have also been observed in human tissues or cells.

(c) Comparison of tissue concentrations in humans and animals

Four epidemiological studies of high-exposure industrial cohorts in Germany, the Netherlands and the United States found an increase in overall cancer mortality.

In these cohorts, the blood lipid 2,3,7,8-TCDD levels estimated to the last time of exposure were 2000 ng/kg (mean) (up to 32 000 ng/kg) in the United States cohort, 1434 ng/kg geometric mean (range, 301–3683 ng/kg) among accident workers in the Dutch cohort, 1008 ng/kg geometric mean in the group of workers with severe chloracne in the BASF accident cohort in Germany and measurements up to 2252 ng/kg in the Boehringer cohort in Germany. These calculated blood 2,3,7,8-TCDD levels in workers at time of exposure were in the same range as the estimated blood levels in a two-year rat carcinogenicity study. In rats exposed to 100 ng/kg bw 2,3,7,8-TCDD per day, hepatocellular carcinomas and squamous-cell carcinomas of the lung were observed. Estimated blood levels were 5000–10 000 ng/kg 2,3,7,8-TCDD. In the same study, in rats exposed to 10 ng/kg bw 2,3,7,8-TCDD per day, hepatocellular nodules and focal alveolar hyperplasia were observed. Estimated blood levels were 1500–2000 ng/kg 2,3,7,8-TCDD. These results indicate parallel tumorigenic responses to high exposure to 2,3,7,8-TCDD in both humans and rats.

In view of the results mentioned above, it should be noted that the present background levels of 2,3,7,8-TCDD in human populations (2–3 ng/kg) are 100 to 1000 times lower than those observed in this rat carcinogenicity study. Evaluation of the relationship between the magnitude of the exposure in experimental systems and the magnitude of the response (i.e., dose–response relationships) do not permit conclusions to be drawn on the human health risks from background exposures to 2,3,7,8-TCDD.