(i) Railroad workers

Diesel locomotives were introduced on railroads in Canada and the USA in 1928 and in Germany in 1932. In the USA, the change-over to diesel engines was 95% complete by 1959 (Garshick et al., 1988).

Hobbs et al. (1977) reviewed the earlier literature on air contaminants in the environment of train crews. In addition, measurements of air contaminants in locomotives and cabooses were made during their passage through tunnels and during freighting and switch-yard operations. These authors estimated 8-h time-weighted averages (TWAs) for combined tunnel and freight operations, and Heino et al. (1978) evaluated levels of diesel exhaust components in locomotive cabs and round-houses in Finland (Table 12).

Table 12.

Levels of air contaminants to which railroad workers are exposed.

As part of a large epidemiological study on railroad workers, Hammond et al. (1984) presented data on components of diesel exhaust. The respirable particles collected had a dichloromethane extractable fraction of 46% (liquid chromatography fractionation), which was found to be composed of 45% aliphatic hydrocarbons, 33% olefinic and aromatic hydrocarbons and 23% polar compounds. The aromatic fraction included phenanthrene and alkylated phenanthrenes.

Woskie et al. (1988a) conducted an industrial hygiene survey of the US railroad industry as a part of epidemiological studies reported by Garshick et al. (1987, 1988). Personal exposure to respirable particles was measured and then corrected for the estimated contribution of cigarette smoke particulates. These data are presented in Table 13. Corrections for cigarette smoke were made by analysing composited respirable particulate samples for nicotine content; an adjusted respirable particulate concentration was then calculated for each job group, and the applicable average fraction of cigarette smoke was subtracted from the average respirable particulate concentration.

Table 13.

Personal exposures to respirable particulate matter, and adjusted respirable particulate matter concentration, among railroad workers by job group.

(ii) Mine workers

The first diesel engine-powered vehicles in underground mines were used in Germany in 1927 (Kaplan, 1959), and they are now used widely throughout the world (Daniel, 1984). Other sources of exposure in mines include activities that produce large quantities of airborne particles, and blasting, which produces particles and gases such as methane and sulfur dioxide. In addition, these gases may be released spontaneously from the ore bed or from surrounding geological formations. Some exposures that may occur in mines, depending on the ores present, were evaluated by previous IARC working groups; these include radon (IARC, 1988), silica (IARC, 1987b), nickel (IARC, 1987a), chromium (IARC, 1987a) and asbestos (IARC, 1987a).

Lassiter and Milby (1978) gave examples of the levels of carbon monoxide and nitrogen dioxide that can be found at the diesel operator's position in an underground mine. On the basis of 2977 samples taken in 1963–72, the average concentration of carbon monoxide was 8.5 ppm (9.7 mg/m³), 5% of the samples containing >50 ppm (>57 mg/m³); 1504 samples contained an average concentration of 0.2 ppm (0.4 mg/m³) nitrogen dioxide, with 0.75% above 3 ppm (>6 mg/m³) and one sample >5 ppm p(>10 mg/m³).

A study conducted for the US Bureau of Mines on levels of diesel exhaust components in 24 mines included two coal mines (Holland, 1978); the results are shown in Table 14. Anthracene and phenanthrene were found at measurable levels, but five other PAHs (benz[a]anthracene, benzo[a]pyrene, benzo[e]pyrene, chrysene and pyrene) were not.

Table 14.

Levels of diesel exhaust components (mg/m³) in 24 US mines.

Levels of air contaminants due to diesel emissions in other coal mines are summarized in Table 15.

Table 15.

Levels of air contaminants (mg/m³, unless otherwise specified) in coal mines.

The environment of six potash mines in New Mexico, USA, was investigated in 1976 (Attfield et al., 1982). The use of diesel equipment in these mines had begun between 1950 and 1966, and seven to 57 diesel-powered units were used. Environmental concentrations in production jobs and other areas in the six mines (based on 25–34 samples) were 6–10 mg/m³ carbon monoxide, 0.2–6.6 mg/m³ nitrogen dioxide and 0.1–4.0 ppm aldehydes.

Cornwell (1982) evaluated employee exposure to diesel emissions at a molybdenum mine in Colorado, USA. Diesel-powered equipment used in the mine included drills, five-yard load haul-dumps and two-yard load haul-dumps. Personal and area sampling was conducted for oxides of carbon, nitrogen and sulfur, formaldehyde, respirable particulate matter, PAHs and cyclohexane-soluble material (sum of particulate and gaseous samples). The results are shown in Table 16.

Table 16.

Air contaminant levels in a molybdenum mine.

Daniel (1984) reported 0.2–1.3 ppm (0.4–2.6 mg/m³) nitrogen dioxide, 3.1–8.7 ppm (3.8–10.7 mg/m³) nitric oxide and 0.5–2.1 ppm (0.6–2.4 mg/m³) carbon monoxide in a South Dakota, USA, gold mine when samples were taken during the operation of a diesel mine vehicle. He found 0.3–0.6 ppm (0.8–1.6 mg/m³) sulfur dioxide, 0.1–0.3 mg/m³ sulfate, 0.4–1.7 mg/m³ respirable combustible dust and 1.1–4.4 mg/m³ total respirable dust.

(iii) Bus garage and other bus workers

Exposures of bus garage and other bus workers to diesel exhaust emissions are listed in Table 17. Few studies addressed other exposures that may occur in bus garages, such as to metal fumes from welding and similar operations and to asbestos during brake servicing.

Table 17.

Levels of air contaminants in bus garages (mg/m³, unless otherwise specified).

The table includes the results of analyses of air during operations in two diesel bus garages in Egypt (El Batawi & Noweir, 1966) and measurements of airborne concentrations of diesel exhaust components in three bus repair facilities in Denver, CO, USA (Apol, 1983). In the latter study, sampling took place in March 1982 at several locations within each facility during peak dispatch and return times. A carbon monoxide level of 195 mg/m³ was recorded in one garage early in the morning near buses with gasoline engines that were starting up. The author stated that exposure of drivers to this and higher levels for 10–15 min is possible. Pryor (1983) also conducted an industrial hygiene survey of diesel exhaust at the garage of a bus company in Denver, CO, in 1982, where buses with gasoline engines parked nearby. Area samples were taken to measure carbon monoxide, sulfur dioxide, nitrogen dioxide, total particulate matter and formaldehyde in the terminal and in package-receiving areas (Table 17). The high concentrations of carbon monoxide in the terminal dropped to the background concentration within a few minutes of bus arrival or departure. The higher levels of carbon monoxide in the reservation room and at the inlet corresponded to peak car traffic in the parking area, and dropped to a background level within 10–15 min of the end of the peak traffic.

Waller et al. (1985) measured pollutant levels in two diesel bus garages in London, UK, in 1979. Sampling took place close to the buses in an area of only limited worker exposure, so the data are said by the authors to present extreme upper limits only. Table 17 gives data on levels of PAHs near the door of one garage during two different periods. del Piano et al. (1986), reporting in an abstract, found concentrations of several air contaminants in Italian bus garages surveyed over two years, as shown in the table. Gamble et al. (1987a) studied 232 workers in four diesel bus garages in the USA. Mean TWA concentrations of respirable particles and nitrogen dioxide in personal samples, combined over three shifts and for all four garages, were reported (Table 17). Ulfvarson et al. (1987) measured personal exposures of workers in a bus garage with both large and small diesel-powered vehicles as well as gasoline-powered ones. Storage, engine warm-up, refuelling, washing and repairs were all performed at the garage. Elevated concentrations, especially of diesel exhaust, accumulated in garage bays in the morning, afternoon and evening when most of the buses were coming or going. Specific exposures in the garage as measured by personal sampling during these periods of high activity are detailed in Table 17.

(iv) Truck drivers

Ziskind et al. (1978) measured the concentrations of several gases in the cabins of heavy-duty diesel trucks under a variety of conditions. Concentrations of carbon monoxide, nitric oxide and nitrogen dioxide in the air and in the cabins were measured continuously. The maximal total concentrations in cabins measured during idling and road testing were as follows: carbon monoxide, 30 ppm (34 mg/m³); nitric oxide, 2 ppm (2.5 mg/m³); nitrogen dioxide, 3 ppm (6 mg/m³). The maximal self-contamination concentrations (total cabin concentration minus ambient concentration) were: carbon monoxide, 10.5 ppm (12 mg/m³); nitric oxide, 1.55 ppm (1.9 mg/m³); nitrogen dioxide, 0.7 ppm (1.4 mg/m³). The authors found a correlation between vehicle-induced concentrations of specific gases in cabins and several testing parameters, including condition of windows, type of cabin configuration and the presence of exhaust leaks and underside cabin openings.

The results of a study of truck drivers on roll-on roll-off ships by personal sampling over an entire work shift (Ulfvarson et al., 1987) are given in Table 18.

Table 18.

Levels of airborne contaminants measured for truck drivers on Swedish roll-on roll-off ships (mg/m³).

(v) Fork-lift truck operators

Breathing zone exposures were measured in an army ammunition depot in the USA in the winter of 1983 during the use of diesel-powered fork-lift trucks. PAHs, sulfur dioxide and carbon monoxide levels were below the detection limits [not given], while the concentration of particulate matter ranged from <0.01 to 1.3 mg/m³, that of nitrogen dioxide from 0.1 to 3.2 ppm (<0.2–6.4 mg/m³) and that of total sulfates from <10 to 32 µg/m³. Area samples taken during the same test provided the following mean values: 1.1–2.6 ppm nitrogen oxides, 1.2–3.3 ppm (1.4–3.8 mg/m³) carbon monoxide, 0.2–0.4 ppm (0.4–0.7 mg/m³) sulfur dioxide and 7.4–8.5 ppm total hydrocarbons (Ungers, 1984). In a follow-up study in the summer of 1984 (Ungers, 1985), the level of PAHs was below the detection limit (3 ng/m³), while breathing zone values during warehouse operations were in the following ranges: particles, 0.5–5.0 mg/m³; sulfate, 0.02–0.6 mg/m³; sulfite, <0.02–0.09 mg/m³; nitric oxide, 1.6–13.6 mg/m³; and nitrogen dioxide, 0.7–2.5 mg/m³.

(vi) Fire-fighters

Fire-fighters are frequently and repeatedly exposed to diesel engine exhaust and other combustion products (Froines et al., 1987). Exposure to diesel exhaust occurs during response to an incident and in the fire station. When personal sampling was used to determine the exposures of fire-fighters in fire stations in three US cities — Boston, New York and Los Angeles — in 1985, total airborne particle levels (TWA) ranged from <0.1 to 0.48 mg/m³. The authors predicted an average total particulate exposure of roughly 0.3 mg/m³ on a typical day in Boston or New York. With an estimated 0.075 mg/m³ contributed by background and smoking, exposure would be to approximately 0.225 mg/m³ diesel exhaust particles and 0.054 mg/m³ of dichloromethane extractable material. Sampling during simulated ‘worst case’ exposures in Los Angeles fire stations gave an upper bound concentration of 0.748 mg/m³ particles.

(i) Toll-booth workers

Ayres et al. (1973) evaluated exposure to engine exhaust for perions who worked in both a tunnel and in an adjacent toll plaza in New York City, USA (Table 19). The authors reported a close correlation among the concentrations of various pollutants, such that the level of carbon monoxide could be considered to be indicative of total automotive pollution levels. Following the installation of ventilation systems in toll booths, carbon monoxide levels dropped to 16–18 mg/m³.

Table 19.

Levels of air contaminants (mg/m³, unless otherwise specified) to which toll-booth operators are exposed.

A study of exposure to carbon monoxide of toll collectors on an interstate highway near Louisville, KY, USA, was reported by Johnson et al. (1974). Testing was conducted over a 12-day period in April 1973, and ambient concentrations of carbon monoxide, lead and manganese at three booths were determined as an overall mean 8-h TWA (Table 19). Carboxyhaemoglobin (COHb) levels in the workers were measured before and after shifts; typical pre-shift levels were 0.8–1.5%; the post-shift mean was 3.9%, with a range of 1.6–11.7%.

Burgess et al. (1977) evaluated the exposure of toll-booth collectors in the Boston area, at a tunnel and at two interchanges during 1972–74 (Table 19). Airborne lead levels in the work environment were roughly four times ambient urban levels; correlated increases were found in the hair and blood of the workers.

(ii) Border-station inspectors

Cohen et al.(1971) studied a group of such workers at the San Ysidro, CA, USA, station in 1969. Ambient and expired carbon monoxide levels were measured over roughly a 24-h period encompassing three shifts. Average hourly carbon monoxide levels were 15 mg/m³ (8.00 to 16.00 h), 75 mg/m³ (16.00 to 0.00 h) and 131 mg/m³ (0.00 to 8.00 h); hourly levels ranged from 6 to 195 mg/m³. Taking data from all shifts combined, nonsmoking individuals had significantly higher expired carbon monoxide levels at the end of the shift than before the shift, corresponding to COHb levels of 3.6% after the shift and 1.5% before the shift. For smokers, the estimated pre- and post-shift COHb levels, estimated from carbon monoxide in expired air, were 4.8% and 6.4%, respectively.

Environmental sampling was conducted by the National Institute for Occupational Safety and Health at a number of US border crossing facilities in 1973–74, to investigate the exposure of Federal border inspectors (Kronoveter, 1976). Inspectors are rotated between different work locations through a shift, and the facilities are operated three shifts per day, seven days per week. Mean 8-h average carbon monoxide levels ranged from 2 to 54 ppm (2.3–62 mg/m³) with maximal 8-h average levels of 3–73 ppm (3.4–83.6 mg/m³). Sampling for total particulate matter revealed concentrations from <1.0 to 4.3 mg/m³. Ozone levels ranged from <0.01–0.08 ppm(<0.02–0.16 mg/m³), and concentrations of lead ranged from <10 to 20) µg/m³.

deBruin (1967) measured COHb levels in 13 nonsmoking Dutch customs officers stationed at four remote sites in 1965. The mean concentrations at the four locations studied ranged from 0.8 to 1.5% before work and 1.1 to 3.0% after work.

(iii) Traffic-control officers

deBruin (1967) measured COHb levels in nonsmoking and smoking traffic policemen in Rotterdam and Amsterdam, the Netherlands. Results are shown in Table 20. Ambient carbon monoxide levels in Rotterdam at the time of that study averaged 5–15 ppm (6–17 mg/m³) at crossings and crowded streets, ranging up to 50 ppm (60 mg/m³).

Table 20.

Carboxyhaemoglobin (COHb) levels in traffic control officers.

Göthe et al. (1969) measured COHb levels in 76 traffic policemen in three Swedish towns. Ambient carbon monoxide levels were not given, but in Stockholm nonsmoking officers who had controlled traffic in either the morning or afternoon rush hours had an average COHb of 1.2% ± 0.39; smokers had a level of 3.5% ± 1.17. In Malmö and Örebro, the levels were 0.8% ± 0.14 and 0.6% ± 0.38 for nonsmokers and 5.0% ± 2.44 and 2.4% ± 1.10 for smokers. The authors noted that unexposed persons in Sweden have an average COHb of 0.5%.

(iv) Professional drivers

deBruin (1967) determined COHb levels before and after work in nonsmoking taxi, delivery van and motor hearse drivers in Amsterdam, the Netherlands, in 1965. The mean COHb levels before and after work were 2.0 and 2.15 for 13 taxi drivers exposed for 7 h; 1.5 and 1.65 for six delivery men exposed for 8 h; and 2.0 and 2.45 for four drivers of motor hearses exposed for 6.5 h. The average difference was 0.25%, which was significant (p > 0.01).

Maruna and Maruna (1975) investigated δ-aminolaevulinic acid elimination in the urine of 200 taxi drivers in Vienna, Austria, as a means of measuring lead burden. The authors found that 26.5% of the taxi drivers had normal levels (<5.0 mg/1 urine), 27% had borderline levels (5.1–7.0 mg/l) and 46.5% had elevated levels (>7.1 mg/1). The authors concluded that the source of the lead was the atmosphere polluted by automobile engine emissions.

del Piano et al. (1986), reporting in an abstract, measured hydrocarbons in the breathing zone of the drivers of an Italian bus company. The concentration of aliphatic hydrocarbons ranged from below 0.01 to 51.8 ppm and that of aromatic hydrocarbons, including benzene (<0.01–9,6 ppm; <0.03–31 mg/m³), from below 0.01 to 36.7 ppm. [The Working Group was unable to determine if these workers were exposed mainly to gasoline or to diesel exhausts.]

(v) Ferry workers

Purdham et al. (1987) studied the potential exposure of stevedores employed in ferry operations to diesel and gasoline exhaust emissions. The constituents considered were total particles, PAHs, aldehydes, nitrogen oxides, sulfur dioxide and carbon monoxide. Exposures to particles averaged 0.50 mg/m³ (range, 0.06–1.72 mg/m³); carbon monoxide levels were detected, in the range of 20–100 ppm (23–115 mg/m³), only when gasoline-powered vehicles were being loaded onto the ferries. The levels of the other constituents did not differ from the background.

Ulfvarson et al. (1987) measured personal exposures to airborne contaminants on two types of car ferries during loading and unloading. On the first one, a 2-h route, loading and unloading of vehicles took an average of 20 min; on the second, a 20-min crossing, loading and unloading times were not specified. The results are shown in Table 21.

Table 21.

Levels of airborne contaminants measured on Swedish car ferries (mg/m³, unless otherwise specified).

(vi) Exhaust system mechanics

Chambers et al. (1984) measured the lead levels in airborne particles and deposited dusts in three centres for the replacement of passenger car exhaust systems. Airborne concentrations of lead collected with environmental and personal samplers ranged from 8.8 to 55.4 µg/m³, with peak concentrations ranging from 20.2 to 92.7 µg/m³ when floors were being swept during sampling. Dust concentrations in samples from floors and shelves ranged from 0.14 to 3.02% lead by weight. Dust in samples collected from inside exhaust systems ranged from 0.20 to 58.7% lead by weight, according to site.

(vii) Motor vehicle inspectors

An industrial hygiene survey of 38 motor vehicle inspection stations in New Jersey, USA, in 1973–74 indicated that inspectors were exposed to carbon monoxide at a TWA average of 10–24 ppm (11–28 mg/m³); measurements in semi-open and enclosed stations were 11–40 ppm (13–46 mg/m³) and those in outdoor stations, 4–21 ppm (5–24 mg/m³). Average COHb levels in nonsmokers were 2.1% before a shift and 3.7% after a shift (p > 0.01) (Stern et al., 1981).

(viii) Parking garage attendants

Ramsey (1967) determined both airborne carbon monoxide levels and blood COHb levels for 38 parking attendants in six garages in Dayton, OH, USA. These garages typically had four floors and a capacity of 300–500 cars. Hourly air sampling revealed a carbon monoxide concentration in the range of 8–275 mg/m³, with a mean of 67.4 ± 28.5 mg/m³ for 18 daily averages. COHb levels were determined in employees prior to and after the work day on Monday; a group of control students was also monitored. The 8:00 and 17.00 h levels were 1.5 ± 0.83 (SD) and 7.3 ± 3.46 for 14 nonsmokers; and 2.9 ± 1.88 and 9.3 ± 3.16 for 24 smokers. In controls, the levels were 0.81 ± 0.54 for ten nonsmokers and 3.9 ± 1.48 for 17 smokers. The differences between 8.00 and 17.00 h levels in garage workers, and between garage workers at 17.00 h and controls are highly significant (p < 0.0001).

(ix) Lumberjacks

Nilsson et al. (1987) studied the composition of exhaust emissions from two-stroke chain-saw engines run on gasoline and estimated operator exposure to chain-saw exhaust under snow-free conditions and with snow on the ground (Table 22). The presence of snow affects techniques used in cutting.

Table 22.

Estimated exposure of lumberjacks using chain saws during loggng under snow-free conditions and in the winter with snow on the ground (mg/m³).

(i) Dilution tube sampling

Vehicle exhaust generated by driving schedules under standard conditions on a chassis dynamometer is diluted with a well-defined quantity of filtered air in a dilution tunnel, from which samples can be taken isokinetically after they have been cooled to about 50°C (Hare & Barnes, 1979). It has been shown that this device (presented schematically in Figure 3) closely simulates the process of dilution (Lee & Schuetzle, 1983; Schuetzle, 1983) under actual atmospheric conditions. The residence time of exhaust in the tunnel before being trapped on filters has been estimated to be less than 5 sec. Particles may be collected on filters, such as Teflon, or by electrostatic precipitators (Lee & Schuetzle, 1983). Hallock et al. (1987) recommended a liquid electrostatic aerosol precipitator which preserves submicronic particle size. Semivolatile compounds in the gaseous phase can be trapped by adsorption filter systems containing inorganic silica or, more satisfactorily, organic materials (e.g., XAD-2, Tenax, Chromosorb 102, Porapak; Jones et al., 1976; Lee et al., 1979; Lee & Schuetzle, 1983; Schuetzle, 1983). Schuetzle (1983) recommended XAD-2, since this material, in contrast to Tenax, does not react with nitrogen oxides.

Fig. 3.

Dilution tube used for monitoring and collecting gas- and particulate-phase vehicle emissions^a.

Gases are sampled in bags either before entering the filter system or after having passed the adsorbent trap (see Figure 3). Stenberg et al. (1983) described a cryogradient technique which allows the separation of NO by cooling with a liquid nitrogen condenser.

(ii) Raw gas sampling

In most of these devices, the exhaust is partially condensed before separation by filter combinations. Initially, steel condensers were used (Stenburg et al., 1961); since that time, a collecting device consisting of a vertical glass cooler and a micron filter of impregnated glass fibre with a separation degree of over 99.9% for particle sizes of 0.3–0.5 µm has been described (Grimmer et al., 1973a; Kraft & Lies, 1981; VDI-Kommision, 1987; see Figure 4).

Fig. 4.

System for collecting undiluted vehicle exhaust^a.

Proportional raw gas sampling, resulting in samples of more manageable size, has been described (Stenberg et al., 1981; Johnson, 1988). Various filter materials, as such or in combination, have been used to separate particles and semivolatile compounds from exhausts. Quartz and glass fibre, cellulose acetate (Millipore), polyvinylidene fluoride (Duropore), polytetrafluoroethylene (Teflon, Zefluor, Fluoropore), polyvinyl chloride, polyethylene and polycarbonate (Nucleopore) can separate >99.9% of submicrometer particulates (John & Reischl, 1978; Lee, F.S.-C. et al., 1980; Lee & Schuetzle, 1983). Teflon has been cited as being superior to other materials in terms of separation efficiency and chemical inertness (Lee & Schuetzle, 1983; Schuetzle, 1983).

Semivolatile compounds in the gaseous phase can be separated out using polymer materials, such as Tenax, XAD-2 and Porapak, or with Chromosorb. Purification of these materials may be time-consuming since repeated washing (or Soxhlet extraction) with various solvents is required to remove organic impurities which interfere with analysis. Trapping of the gaseous phase can be carried out as in the case of dilution tube sampling, by (i) the cryogradient technique or (ii) gas bag collection.

One of the main problems described in the literature regarding the correct sampling of exhaust is avoidance of chemical conversion of sensitive compounds (e.g., PAHs) to artefacts (e.g., ketones, quinones, phenols, halides and nitroarenes). Losses of pyrene and perylene were studied after direct injection into exhaust near the end of the exhaust pipe, with recoveries of only 20–60% by gas chromatography and high-performance liquid chromatography (Lee et al., 1979). Schuetzle (1983) found that more than 90% of pyrene and benzo[a]pyrene was lost through oxidation under the same conditions.

PAHs can react with nitrogen oxides, ozone or sulfuric acid under sampling conditions. Photooxidation during sample analysis may result in the formation of endoperoxides which then undergo rearrangement to yield hydroquinones and subsequently quinones, or in the formation of exoperoxides which yield ketones (Schuetzle, 1983). In the presence of nitric acid, nitrogen dioxide reacts with PAHs to yield nitroarenes (Pitts et al. 1978; Brorström et al., 1983). On the assumption that nitroarene formation is acid-catalysed, diesel exhaust was collected with and without injection of ammonia during sampling; significantly higher amounts of 1-nitropyrene were found without ammonia injection at a sampling temperature of 100°C When passed through an exhaust-loaded filter, the particle-free gaseous phase enhanced the concentration of 1-nitropyrene. No such effect was observed when collecting filters were maintained at below 35°C (Grimmer et al., 1988).

The artificial formation of nitroarenes during sampling procedures is still a matter of controversy (Gaddo et al., 1984) and has been critically reviewed (Lee & Schuetzle, 1983).

It should be noted that the invariance of the PAH profile over the collection period is the essential precondition for correct sampling, since certain individual compounds are more sensitive to chemical reactions and/or evaporation effects.

(i) Thin-layer chromatography (TLC)

Two-dimensional cellulose TLC with fluorescence detection is the recommended TLC method for PAH fractions from vehicle exhaust and has given results in good agreement with those obtained by GC (Kraft & Lies, 1981). A simple TLC screening method for the determination of benzo[a]pyrene, which is also applicable to vehicle exhaust, has recently been recommended by the International Union of Pure and Applied Chemistry (Grimmer & Jacob, 1987). The use of TLC for the analysis of PAHs has been reviewed (Daisey, 1983).

(ii) High-performance liquid chromatography (HPLC)

The basic advantages and disadvantages of HPLC in comparison to other techniques such as GC have been reviewed (Wise et al., 1980; Wise, 1983). The method has been widely used for the detection and identification of organic constituents of vehicle exhaust, and results of determinations of PAHs in automotive exhaust condensate using HPLC have been compared with those obtained using capillary GC (Doran & McTaggart, 1974). A comparative study of different HPLC methods for the analysis of PAHs in diesel emissions has been carried out (Eisenberg & Cunningham, 1984). Using HPLC/GC-MS, 74 polycyclic aromatic compounds were identified or tentatively identified in diesel particulate extracts (Tong & Karasek, 1984). A normal-phase HPLC method using silica gel columns and n-hexane:benzene (3:1) as eluent has been developed to isolate PAHs and their nitro derivatives (Nielsen, 1983). A very precise, routine, on-line reverse-phase HPLC/fluorescence method has been reported for the analysis of nitroarenes in the picogram range by their reduction to highly fluorescent amines (Tejada et al., 1983). A ‘pyrenebutyric acid amide phase’ has been applied to a multidimensional HPLC method with on-line peak identification by ultra-violet-visible spectrometry, which allows the detection of 1-nitropyrene in the range of 3–100 ng per mg of soot collected on a filter (Lindner et al., 1985). HPLC has been used to compare the concentrations of some PAHs and their nitro derivatives in exhausts from four diesel cars (Gibson et al., 1981). A semipreparative HPLC analysis of a soluble organic fraction of diesel engine exhaust particulates has been reported, together with unsatisfactory results using HPLC/MS coupling (Levine et al., 1982).

(iii) Gas chromatography (GC) and gas chromatography /mass spectrometry (GC/MS)

GC methods for the determination of PAHs from exhaust condensates have been reviewed (Olufsen & Björseth, 1983), as have GC and GC/MS analyses of nitroarenes (White, 1985). Collaborative studies have been carried out to analyse PAHs in vehicle exhaust (Janssen, 1976; Metz et al., 1984) in which GC methods were compared with those of HPLC (Metz et al., 1984).

Packed-column GC has been largely replaced by high-resolution glass capillary column GC using fused silica columns and careful sample injection procedures (on-column injection at low temperatures of 50–80°C or thermal desorption cold trap injection procedure). Detection limits down to the picogram level have been reached with GC when single-ion monitoring mass spectrometry is used as the detection system (Ramdahl & Urdal, 1982).

Various classes of organic compounds from vehicle exhausts have been analysed by GC and GC/MS, including paraffins, olefins, PAHs, thia-arenes, aza-arenes, oxo-arenes and aldehyde, phenol, quinone and nitro derivatives of PAHs and their acid anhydrides (Hites et al., 1981; Schuetzle et al., 1981; Lee & Schuetzle, 1983; Alsberg et al., 1984; Ramdahl, 1984; White, 1985). More than 70 individual nitroarenes were found in diesel particulate extracts by means of fused-silica capillary column GC and a nitrogen-specific detector, with detection limits of 0.2–0.5 mg/kg (Paputa-Peck et al., 1983). The emission of PAHs and nitroarenes by diesel engines from PAH-containing and other well defined fuels (hexa-decane) were studied using GC/MS and tandem triple-quadrupole MS (Fulford et al., 1982; Henderson et al., 1983, 1984).

The rapid analysis of gaseous and other combustion-related compounds in hot gas streams by atmospheric-pressure chemical ionization/MS has been reported (Sakuma et al., 1981).

(iv) Other methods

Photometric, infra-red, colorimetric, electrochemical and chemiluminescence techniques have been used for the analysis of gases such as sulfur dioxide, carbon monoxide, carbon dioxide, nitrogen oxides and formaldehyde (Hare & Baines, 1979; Deutsche Forschungs-gemeinschaft, 1985). Test tubes in which various colour reactions are seen are available for the analysis of various constituents of vehicle exhaust such as carbon monoxide, carbon dioxide, sulfur dioxide and nitrogen dioxide (Leichnitz, 1986).

(a) Inhalation exposure

Mouse: Heinrich et al. (1986a) exposed two groups of 96 female NMRI mice, eight to ten weeks old, to filtered or unfiltered exhaust from a 1.6–1 displacement diesel engine operated according to the US-72 (FTP; see p. 80) test cycle to simulate average urban driving, or to clean air, for 19 h per day on five days per week for life.The unfiltered and filtered exhausts were diluted 1:17 with air and contained 4.24 mg/m³ particles. Levels of 1.5 ± 0.3 ppm (3 ± 0.6 (SD) mg/m³) nitrogen dioxide and 11.4 ± 2.1 ppm nitrogen oxides were found in whole exhaust and 1.2 ± 0.26 ppm (2.4 ±0.5 mg/m³) nitrogen dioxide and 9.9 ± 1.8 ppm nitrogen oxides in filtered exhaust. Exposure to total diesel exhaust and filtered diesel exhaust significantly increased the number of animals with lung tumours (adenomas and carcinomas) to 24/76 (32%) and 29/93 (31%), respectively, as compared to 11/84 (13%) in controls. When the incidences of adenomas and carcinomas were evaluated separately, significantly higher numbers of animals in both diesel exhaust-exposed groups had adenocarcinomas (13 (17%) and 18(19%), respectively) than in controls (2.4%); no increase was seen in the numbers of animals with adenomas. [The Working Group noted that the incidence of lung tumours in historical controls in this laboratory could reach 32% (Heinrich et al., 1986b).]

Groups of ICR and C57B1/6N mice (total number of treated and untreated animals combined alive at three months, 315 and 297, respectively) [initial numbers and sex distribution unspecified] were exposed to the exhaust from a small diesel engine (269 cm³ displacement, run at idling speed) used as an electric generator; the exhaust was diluted 1:2 to 1:4 in air (Takemoto et al., 1986). The mice were exposed within 24 h after birth for 4 h per day on four days per week (2–4 mg/m³ particles; size, 0.32 µm; 2–4 ppm, 4–8 mg/m³ nitrogen dioxide). Between months 13 and 28, lung tumours (adenomas and adenocarcinomas) were found in 14/56 exposed ICR mice and in 7/60 controls and in 17/150 treated C57B1/6N mice and 1/51 controls. The authors reported that the differences were not statistically significant. [The Working Group calculated that the difference in C57B1/6N mice was statistically significant at p < 0.05.]

Rat: Karagianes et al. (1981) exposed groups of male specific-pathogen-free Wistar rats [numbers unspecified], 18 weeks old, for 6 h per day for 20 months to one of five experimental atmospheres: clean air (controls); 8.3 ± 2.0 (SD) mg/m³ soot from diesel exhaust; 8.3 ± 2.0 mg/m³ soot from diesel exhaust plus 5.8 ± 3.5 mg/m³ coal dust; 6.6 ± 1.9 mg/m³ coal dust; or 14.9 ± 6.2 mg/m³ coal dust. The diesel exhaust was produced by a three-cylinder, 43-brake horse power diesel engine driving a 15 kW electric generator. The fuel injection system of the engine was modified to simulate operating patterns of such engines in mines and was operated on a variable duty cycle (dilution, approximately 35:1). Six rats per group were killed after four, eight, 16 or 20 months of exposure. Complete gross necropsy was performed, and respiratory tract tissues, oesophagus, stomach and other tissues with lesions were examined histopathologically. Significant non-neoplastic lesions were restricted primarily to the respiratory tract and increased in severity with duration of exposure. In the six rats examined from each group after 20 months of exposure, two bronchiolar adenomas were observed — one in the group exposed to diesel exhaust only and one in the group exposed to diesel exhaust and coal dust. None was observed in controls or in the two groups exposed to coal dust only. [The Working Group noted the limited number of animals studied at 20 months.]

Groups of 72 male and 72 female or 144 male Fischer 344 weanling rats were exposed for 7 h per day on five days per week for 24 months to either clean air (controls); 2 mg/m³ coal dust (<7 µm); 2 mg/m³ diesel exhaust particles, with specific limits on gaseous/vapour constituents; or 1 mg/m³ coal dust plus 1 mg/m³ diesel exhaust particles (Lewis et al., 1986). The nitrogen dioxide concentration in the diesel exhaust was 1.5 ± 0.5 ppm (3 ± 1 mg/m³); the exhaust was generated by a 7–1 displacement, four-cycle, water-cooled, ‘naturally aspirated’ (open-chamber) diesel engine. The exhaust was diluted by a factor of 27:1 before entering the exposure chambers. Following three, six, 12 and 24 months of exposure, at least ten male rats per group were removed for ancillary studies. After 24 months of exposure, all survivors were killed. The numbers of rats necropsied and examined histologically in each of the four groups were 120–121 males and 71 –72 females. No difference in survival was noted among treatment groups, chambers or sexes [data on survival unavailable]. No statistical difference in tumour incidence was noted among the four groups. [The Working Group noted that no detailed information on tumour incidence was available and that the animals were killed at 24 months, a shorter observation period than used in other inhalation studies with rats that gave positive results.]¹

Female specific-pathogen-free Fischer 344 rats [initial number unspecified], aged five weeks, were exposed to diesel exhaust from a small diesel engine (269 cm³ displacement) run at idling speed; rats were treated for 4 h per day on four days per week for 24 months, at which time they were killed or were left untreated (Takemoto et al., 1986). The exhaust was diluted 1:2 to 1:4 with air. The concentration of particulates (size, 0.32 µm) ranged from 2–4 mg/m³, and those of nitrogen dioxide were 2–4 ppm (3–8 mg/m³). No lung tumour was observed in either the 26 treated or 20 control rats; 15 and 12 rats in the two groups, respectively, survived 18–24 months. [The Working Group noted the small group sizes.]

Iwai et al. (1986) exposed two groups of 24 female specific-pathogen-free Fischer 344 rats, seven weeks of age, to either diluted diesel exhaust or diluted filtered diesel exhaust for 8 h per day on seven days a week for 24 months, at which time some rats were sacrificed and the remainder were returned to clean air for a further six months of observation. The diesel exhaust was produced by a 2.4–1 displacement small truck engine; it was diluted ten times with clean air and contained 4.9 ± 1.6 mg/m³ particles, 1.8 ± 1.8 ppm (3.6 ± 3.6 mg/m³) nitrogen dioxide and 30.9 ± 10.9 ppm nitrogen oxides. Another group of 24 rats was exposed to fresh air only for 30 months. Incidences of lung tumours, diagnosed as adenomas, adenocarcinomas, squamous-cell carcinomas and adenosquamous carcinomas, were significantly higher in the group exposed to whole diesel exhaust, with or without a subsequent observation period (in 8 /19 rats, including five with malignant tumours) than in the control group (one adenoma in 1 /22 rats; p < 0.01). No lung tumour was observed in the group exposed to filtered exhaust (0/16 rats). Incidences of malignant lymphomas and tumours at other sites did not differ among the three groups. [The Working Group noted the small group sizes.]

Ishinishi et al., (1986a) exposed groups of 64 male and 59 female specific-pathogen-free Fischer 344 rats, four weeks of age, to diesel exhaust from either a light-duty 1.8–1 displacement, four-cylinder engine (particle concentrations, 0.11,0.41,1.08 or 2.32 mg/m³; nitrogen dioxide concentrations, 0.08, 0.26, 0.70 or 1.41 ppm (0.2, 0.5, 1.4 or 2.8 mg/m³); nitrogen oxide concentrations, 1.24, 4.06, 10.14 or 20.34 ppm) or a heavy-duty 11–1 displacement, six-cylinder engine (particle concentrations, 0.46, 0.96, 1.84 or 3.72 mg/m³; nitrogen dioxide concentrations, 0.46, 1.02, 1.68 or 3.00 ppm (0.9, 2.1, 3.4 or 6 mg/m³); nitrogen oxide concentrations, 6.17, 13.13, 21.67 or 37.45 ppm). Exposure was for 16 h per day on six days per week for up to 30 months. The diesel emissions were diluted about 10–15 times (v/v) with air. Separate control groups for the light-duty and heavy-duty series were exposed to clean air. The incidence of lung tumours diagnosed as adenocarcinomas, squamous-cell carcinomas or adenosquamous carcinomas was significantly increased only in the highest-dose group (in 5/64 males and 3/60 females) of the heavy-duty diesel exhaust-exposed series compared to controls (in 0/64 males and 1/59 females; p<0.05). The incidences in the next highest-dose group in this series were 3/64 males and 1/59 females. [The Working Group noted that, although this incidence was not statistically different from that in the controls, it suggested an overall positive response for the two highest exposure levels.] No statistically significant increase in the incidence of lung tumours was noted in the groups exposed to light-duty diesel engine exhaust. [The Working Group noted that the highest level of exposure in the light-duty series was approximately one-half of the highest concentration used in the heavy-duty series, and that the incidence (3.3%) of lung tumours in the control animals of the light-duty diesel engine exhaust-exposed series was higher than that in the heavy-duty diesel controls (0.8%).]

Groups of 72 male and 72 female Fischer 344 rats, six to eight weeks old, were exposed to one of three concentrations of diesel engine exhaust or particle-filtered diesel engine exhaust from a 1.5–1 displacement engine operated according to the US-72 (FTP) driving cycle which simulates average urban driving; exposure was for 16 h per day on five days per week for two years (Brightwell et al., 1986). The exposure concentrations were reported as a dilution of the exhaust with a constant volume of 800 m³ of air (high dose), a further dilution of this mixture in air of 1:3 (medium dose) and a dilution of 1:9 (low dose). The particle concentrations in the unfiltered diesel exhaust atmosphere were 0.7, 2.2 and 6.6 mg/m³ for the low, medium and high doses (with 8 ± 2 ppm nitrogen oxides in the high dose), respectively. Two control groups of 144 rats of each sex were exposed to conditioned air. Following the exposure period, the animals were maintained for a further six months in clean air. An exposure concentration-related increase in the incidence of primary lung tumours [detailed histopathology unspecified] was reported only in groups exposed to unfiltered diesel exhaust. [The Working Group noted that no information of tumour incidence was given for rats exposed to filtered diesel exhaust.]¹

Heinrich et al. (1986a) exposed two groups of 96 female Wistar rats, eight to ten weeks old, to filtered or unfiltered exhaust, as described on p. 89. A significantly increased incidence of lung tumours (histologically identified as eight bronchioalveolar adenomas and nine squamous-cell tumours) was observed in rats exposed to unfiltered diesel exhaust (17/95(18%) versus 0/96 controls). No lung tumour was reported in rats exposed to filtered exhaust.

Mauderly et al. (1986, 1987) exposed groups of 221–230 male and female specific-pathogen-free Fischer 344 rats, 17 weeks old, to one of three concentrations of diesel engine exhaust generated by a 1980 model 5.7–1 V8 engine operated according to US FTP cycles; exposure was for 7 h per day on five days per week for up to 30 months. The exposure concentrations were reported as a dilution of the whole exhaust to measured soot concentrations of 0.35 (low), 3.5 (medium) or 7.0 (high dose) mg/m³. Levels of nitrogen dioxide were 0.1 ± 0.1 (0.2 ± 0.2), 0.3 ± 0.2 (0.6 ± 0.4) and 0.7 ± 0.5 ppm (1.4 ± 1 mg/m³), respectively. Sham-exposed controls received filtered air. The soot particles were approximately 0.25 µm mass median diameter, and approximately 12% of their mass was composed of solvent-extractable organics. Subgroups of animals were removed at six, 12, 18 and 24 months for ancillary studies; all rats surviving after 30 months of exposure were killed. All rats that died or were killed were necropsied and examined histologically for lung tumours. Exposures did not significantly affect the survival of animals of either sex. The median survival time ranged from 880 (low) to 897 (medium) days of age for males and from 923 (high) to 962 (low) days for females. A total of 901 rats were examined for lung tumours; four types were found: bronchoalveolar adenomas, adenocarcinomas, squamous cysts (mostly benign) and squamous-cell carcinomas. None of the tumours was found to have metastasized to other organs. The incidences of lung tumours in males and females combined were 0.9% in controls, 1.3% in low-dose, 3.6% in medium-dose and 12.8% in high-dose groups. The authors noted that the prevalences at the medium and high levels were significantly increased (p<0.05). A total of 42 rats developed 46 lung tumours; four females in the high-dose group had two lung tumours each. Lung tumours were found in two male controls, in one low-dose male, in four mid-dose males and in 13 high-dose males; in females, the respective incidences were zero, two, four and 20. Adenomas predominated in the medium-exposure group. Adenocarcinomas, squamous-cell carcinomas and squamous cysts were observed predominantly at the high dose. The tumours were observed late in the study: 81% after two years of exposure. The authors observed no exposure-related difference in cause of death; the tumours were found incidentally at death or at termination of the experiment.

Hamster: Groups of 48 female Syrian golden hamsters, eight weeks of age, were exposed to diluted (1:7 air) unfiltered diesel exhaust (mass median particle diameter, 0,1 µm; mean particle concentration, 3.9 ± 0.5 mg/m³; nitrogen dioxide, 1.2 ± 1.7 ppm (2.4 ± 3.4 mg/m³); nitrogen oxides, 18.6 ± 5.8 ppm) or filtered diesel exhaust (nitrogen dioxide, 1.0 ± 1.5 ppm (2 ± 3 mg/m³); nitrogen oxides, 19.2 ± 6.1 ppm); exposure was for 7–8 h per day on five days per week for life (Heinrich et al., 1982). The exhaust was generated by a 2.4–1 displacement engine operating at a steady state. A group of 48 hamsters inhaling clean air served as controls. There was no effect of diesel exhaust on survival; median lifespan was 72–74 weeks in all groups, and no lung tumour was reported in treated or control animals.

Groups of 48 female and 48 male Syrian golden hamsters, eight to ten weeks of age, were exposed to diluted (1:17 air) filtered or unfiltered exhaust as described on p. 89 (Heinrich et al., 1986a). A control group of 48 females and 48 males inhaled clean air. Median lifespan was not significantly influenced by diesel exposure and was 75–80 weeks for females and 80–90 weeks for males. No lung tumour was observed in treated or control animals.

Groups of 52 female and 52 male Syrian hamsters, six to eight weeks of age, were exposed to one of three concentrations of unfiltered or filtered exhaust as described on p. 91 (Brightwell et al., 1986). Two control groups of 104 hamsters of each sex were exposed to clean air only. The authors reported that there was no increase in the incidence of respiratory-tract tumours in treated hamsters. [The Working Group noted the incomplete reporting of tumour incidence and survival.]

Monkey: Groups of 15 male cynomolgus monkeys (Macacafascicularis) were exposed by Lewis et al. (1986) to coal dust and/or diesel exhaust particles for 7 h per day on five days per week for 24 months, as described on p. 90. Following the exposure period, all survivors (59/60) were necropsied and examined histologically. No significant difference in tumour incidence was reported among the four groups. [The Working Group noted the short duration and inadequate reporting of the study.]

(b) Intratracheal or intrapulmonary administration

Rat: Four groups of 31,59,27 and 53 female specific-pathogen-free Fischer 344 rats, six weeks of age, received ten weekly intrapulmonary instillations of 1 mg/animal activated carbon or 1 mg/animal diesel exhaust particles [source unspecified] in phosphate buffer with 0.05% Tween 80, 2 ml of buffer alone or were untreated (Kawabata et al., 1986). Rats surviving 18 months constituted the effective numbers. The experiment was terminated 30 months after instillation. The survival rate was 71–83%, with the lowest value in the diesel particle-treated group. The numbers of animals with malignant lung tumours [histological type unspecified] were significantly higher (p<0.01) in the groups treated with activated carbon (7/23) and with diesel particles (20/42) than in untreated (0/44) or vehicle controls (1/23). Similarly, the numbers of animals with benign and malignant lung tumours were also significantly increased in the groups treated with activated carbon (11/23) and diesel particles (31/42). [The Working Group noted the high incidence of pulmonary tumours observed after treatment with activated carbon, a material which is normally considered to be inert.]

Groups of 35 female inbred Osborne-Mendel rats, three months old, received lung implants of organic material from a diesel exhaust or a reconstituted hydrophobic fraction (Grimmer et al., 1987). The organic material was collected from a 3–1 diesel passenger car engine, operated under the first cycle of the European test cycle (see p. 80), and was separated by liquid-liquid distribution into a hydrophilic fraction (approximately 25% by weight of the total condensate) and a hydrophobic fraction (approximately 75% by weight). The hydrophobic fraction was separated by column chromatography into several further fractions: (i) nonaromatic compounds plus PAHs with two and three rings (72% by weight of the total condensate), (ii) PAHs with four or more rings (0.8% by weight), (iii) polar PAHs (1.1% by weight) and (iv) nitro-PAHs (0.7% by weight). Animals received 6.7 mg hydrophilic fraction, 20 mg hydrophobic fraction, 19.2, 0.2, 0.3 or 0.2 mg of the four hydrophobic fractions, respectively, or 19.9 mg of reconstituted hydrophobic fraction. Two groups of 35 animals were untreated or received implants of the vehicle (beeswax:tri-octanoin, 1:1) only. All animals were observed until spontaneous death (mean survival time, 24–140 weeks). Six lung tumours (squamous-cell carcinomas) were found in animals treated with the hydrophobic subfraction containing PAHs with four to seven rings. Similar carcinogenic potency was seen with the reconstituted hydrophobic subfractions (seven carcinomas) and with the hydrophobic fraction (five carcinomas). A low carcinogenic potential was observed with the subfraction of nitro-PAHs (one carcinoma); the polar PAH produced no tumour; and one bronchiolar-alveolar adenoma was observed in animals treated with the nonaromatic subfraction with two- and three-ring PAHs. One adenoma of the lung occurred in the vehicle control group.

Hamster: Shefner et al. (1982) gave groups of 50 male Syrian golden hamsters, 12–13 weeks of age, intratracheal instillations once a week for 15 weeks of 1.25, 2.5 or 5 mg diesel particles (obtained from US Environmental Protection Agency; 90% by mass <10 åm) or diesel particles plus the same amounts of ferric oxide in 0.2 ml propylene glycol/gelatine/-saline; or a dichloromethane extract of diesel particles plus ferric oxide in 0.2 ml propylene glycol/saline once a week for 15 weeks. Ten animals in each group were sacrificed at 12 months. At the time of reporting (61 weeks), one lung adenoma had been found in the group receiving the high dose of diesel particles and one in the group receiving the high dose of diesel particle extract plus ferric oxide. No lung tumour was reported in various untreated or solvent-treated controls. [The Working Group noted the short observation period and the preliminary reporting of the experiment.]

Three groups of 62 male Syrian golden hamsters, eight weeks of age, were given intratracheal instillations of 0.1 ml of a suspension of 0.1,0.5 or 1 mg of an exhaust extract in Tween 60:ethanol:phosphate buffer (1.5:2.5:30 v/v) from a heavy-duty diesel engine (V6 11–1) once a week for 15 weeks and observed for life (Kunitake et al., 1986). A control group of 59 animals received instillations of the vehicle only and a positive control group of 62 animals received 0.5 mg benzo[a]pyrene weekly for 15 weeks. Survival rates were 95%, 92%, 71% and 98% in the three treated groups and the vehicle controls, respectively. No significant difference in the incidences of tumours of the lung, trachea or larynx was observed between untreated control and treated groups; respiratory tumours occurred in 88% benzo[a]pyrene-treated hamsters. [The Working Group noted that length of survival was not reported.]

(c) Skin application

Mouse: A group of 12 male and 40 female C57B1 mice [age unspecified] received skin applications of 0.5 ml of an acetone extract of particles collected from a diesel engine [unspecified] running at zero load during the warm-up phase; treatment was given three times a week for life (Kotin et al., 1955). Groups of 50 male and 25 female strain A mice [age unspecified] received similar applications of an extract of particles derived from the warmed-up engine running at full load. Of the mice in the first group, 16 had died by ten weeks; 33 mice survived to the appearance of the first skin tumour (13 months), and two skin papillomas developed. Of the male strain A mice, eight survived to the appearance of the first skin tumour (16 months), and one papilloma and three squamous-cell carcinomas were observed. Of the female strain A mice, 20 survived to the appearance of the first skin tumour (13 months), and 17 skin tumours [unspecified] were observed between 13 and 17 months. Both experiments were terminated after 22–23 months. No skin tumour occurred in 69 C57B1 controls (37 alive after 13 months) or in 34 (24 female and 10 male) strain A controls.

In a study reported before completion (Depass et al., 1982), groups of 40 male C3H/HeJ mice [age unspecified] received skin applications of 0.25 ml of a 5 or 10% solution in acetone or 5,10,25 or 50% dichloromethane extracts of diesel particles collected from a [5.7–1] diesel engine; treatment was given three times per week for life. A positive control group received 0.2% benzo[a]pyrene in acetone, and a negative control group received acetone only. One squamous-cell carcinoma of the skin was observed in the group treated with the highest dose of dichloromethane extract after 714 days of treatment. All 38 mice receiving benzo[a]pyrene developed skin tumours. [The Working Group noted the inadequate reporting of the study.]

In a series of promotion-initiation studies (Depass et al., 1982), groups of 40 male C3H /HeJ mice received a single initiating dose of 0.025 ml 1.5% benzo[a]pyrene in acetone, followed one week later by repeated applications of the 10% solution of diesel particles in acetone described above, 50% dichloromethane extract, 25% dichloromethane extract, acetone only or 0.015 µg phorbol 12-myristyl 13-acetate (TPA) five times per week for life. An additional group received no further treatment after the initiating dose of benzo[a]pyrene. In initiation studies, a single initiating dose of 0.025 ml of the 10% solution of diesel particles in acetone, 50% dichloromethane extract, acetone or TPA was followed after one week by of 0.015 µg TPA three times per week. The concentration of TPA used in the initiation and promotion studies was changed after eight months to 1.5 µg. In the promotion study, one mouse receiving the 50% dichloromethane extract had a squamous-cell carcinoma and two mice receiving the 25% extract had one squamous-cell carcinoma and one papilloma. In the initiation study, three (two papillomas, one carcinoma), three (two papillomas, one fibrosarcoma), one (papilloma) and two (one carcinoma, one papilloma) tumours were observed in the groups that received diesel particles, dichloromethane extract, acetone and TPA, respectively. [The Working Group noted the preliminary reporting of the study.]

Nesnow et al. (1982a,b) gave skin applications to groups of 40 male and 40 female SENCAR mice, seven to nine weeks of age, of 0.1, 0.5, 1.0, 2 or 10 mg of dichloromethane extracts of particles obtained from the exhausts of five diesel engines, A, B, C, D and E (E being a heavy duty engine) in 0.2 ml acetone; the 10-mg dose was given in five daily doses. The benzo[a]pyrene content ranged from 1173 ng/mg in the exhaust from engine A to 2 ng/mg in that from engines B and E. One week later, all mice received 2 µg TPA in 0.2 ml acetone twice a week for 24–26 weeks. A control group was treated with TPA only. The sample from engine A produced a dose-related increase in the incidence of skin papillomas, with 5.5 and 5.7 papillomas/mouse, 31% of males and 36% of females at the highest dose having skin carcinomas. With samples from engines B, C and D, responses of 0.1–0.5 papilloma/mouse were observed compared to 0.05–0.08 papilloma/mouse in TPA controls. The sample from engine E produced a response similar to that in controls (0.05–0.2 papilloma/mouse).

Similar groups of 40 male and 40 female SENCAR mice received weekly skin applications of 0.1, 0.5, 1, 2 or 4 mg extracts of particles from the emissions of engines A, B and E for 50–52 weeks (Nesnow et al., 1982b, 1983). The high dose was given in two split doses. At that time, skin carcinomas had occurred in 3% of male and 5% of female mice given the 4-mg dose of the sample from engine A, in 3% of males given the 0.5-mg dose of the sample from engine B and in 3% of females given the 0.1-mg dose of the sample engine E. Doses of 12.6–202 µgper week benzo[a]pyrene produced skin carcinoma responses of 10–93%.

Groups of 50 female specific-pathogen-free ICR mice, aged eight to nine weeks, received skin applications of extracts of diesel particles collected from a V6 11–1 heavy-duty displacement diesel engine in 0.1 ml acetone onto shaved back skin every other day for 20 days (total doses, 5, 15 or 45 mg/animal; Kunitake et al., 1986). A further group of 50 mice treated with acetone only served as controls. Beginning one week after the last diesel extract treatment, each animal received applications of 2.5 µg TPA in 0.1 ml acetone three times a week for 25 weeks, at which time they were autopsied. No skin ‘cancer’ was found in either treated or control groups; skin papillomas were seen in 1/48 and 4/50 surviving animals in the 15- and 45-mg dose groups, respectively [The Working Group noted the short duration of both the treatment and observation time.]

(d) Subcutaneous administration

Mouse: Groups of 15–30 female specific-pathogen-free C57B1/6N mice, six weeks of age, received subcutaneous injections into the intrascapular region of suspensions in olive oil containing 5% dimethyl sulfoxide of 10, 25, 50, 100, 200 or 500 mg/kg bw of diesel particles collected from a V6 11–1 heavy-duty displacement diesel engine; the treatment was given once a week for five weeks (Kunitake et al., 1986). A control group of 38 mice received injections of the vehicle only. Animals were killed 18 months after the beginning of the experiment. The first tumours were palpated in week 47 (a total dose of 25 mg/kg bw), week 30 (50 mg/kg bw), week 27 (100 mg/kg bw) and week 39 (200 and 500 mg/kg bw) in the five treated groups, respectively. A significant increase in the incidence of subcutaneous tumours, diagnosed as malignant fibrous histiocytomas, was observed only in 5/22 mice receiving the 500-mg/kg bw dose (p < 0.05) in comparison with controls (0/38). [The Working Group noted the high dose required to produce a carcinogenic effect.]

(e) Administration with known carcinogens

Rat: Two groups of female specific-pathogen-free Fischer 344 rats [initial number unspecified], five weeks of age, were exposed to diesel exhaust, as described on p. 89 or to clean air for 4 h per day on four days per week for 24 months (Takemoto et al., 1986). One month after the beginning of treatment, both groups received three weekly intraperitoneal injections of 1 g/kg bw N-nitrosodipropanolamine. Rats were killed at six, 12, 18 and 24 months after the start of treatment. A slight but nonsignificant increase in the incidences of lung adenomas and adenocarcinomas was observed in rats exposed to both exhaust and the nitrosamine compared to those exposed to the nitrosamine alone. After 12–24 months of observation, 16 lung tumours (12 adenomas and four carcinomas) were observed in 29 N-nitrosodipropanolamine-treated rats and 34 tumours (24 adenomas and 10 carcinomas) were observed in 36 rats exposed to both exhaust and N-nitrosodipropanolamine. The authors interpreted this result as an ‘overadditive’ effect on lung tumour incidence.

Heinrich et al. (1986a) gave groups of 48 female specific-pathogen-free Wistar rats, eight to ten weeks of age, 25 weekly subcutaneous injections of 250 or 500 mg/kg bw N-nitrosopentylamine during the first 25 weeks of exposure by inhalation to unfiltered diesel engine exhaust, to filtered diesel engine exhaust or to clean air, as described on p. 89. Significant increases in the incidences of squamous-cell carcinomas of the lung were observed in animals treated with the nitrosamine and exposed to total exhaust (22/47 low-dose nitrosamine; 15/48 high-dose nitrosamine compared to 2/46 and 8/48 clean air controls, respectively), although overall lung tumour rates were comparable in the groups exposed to the nitrosamine and to engine exhaust or clean air. The incidence of benign tumours (papillomas) of the upper respiratory tract was significantly reduced in nitrosamine-treated rats exposed to unfiltered or filtered diesel exhaust compared to controls exposed to nitrosamine and clean air.

Hamster. Heinrich et al. (1982) gave groups of 48–72 female Syrian golden hamsters, eight weeks old, weekly intratracheal instillations of 0.1 or 0.3 mg dibenzo[a,h]anthracene for 20 weeks or a single subcutaneous injection of 1.5 or 4.5 mg/kg bw N-nitrosodiethyl-amine (NDEA) and exposed them concomitantly by inhalation to unfiltered or filtered diesel exhaust or clean air, as described on p. 92. The incidence of tumours in the larynx/trachea was increased in animals treated with the higher dose of NDEA and exposed concomitantly to total exhaust (70.2%) or filtered exhaust (66%) as compared to controls (44.7%). The lower dose of NDEA and treatment with dibenzo[a,h]anthracene resulted in a lower incidence of these tumours. Only two lung tumours were found: one with the high dose of dibenzo[a,h]anthracene and filtered exhaust, the other with the low dose of NDEA and total exhaust.

Groups of 48 male and 48 female Syrian golden hamsters, eight to ten weeks of age, received a single subcutaneous injection of 4.5 mg/kg bw NDEA or 20 intratracheal instillations of 0.25 mg benzo[a]pyrene with concomitant exposure by inhalation to filtered or unfiltered diesel engine exhaust or to clean air, as described on p. 89 (Heinrich et al., 1986a). Treatment with NDEA or benzo[a]pyrene produced respiratory tract tumour incidences of 10% or 2%, respectively, in animals exposed to clean air; rates were not significantly increased by concomitant exposure to filtered or unfiltered diesel engine exhaust.

Groups of 52 female and 52 male Syrian hamsters, six to eight weeks old, received a single subcutaneous injection of 4.5 mg/kg bw NDEA three days prior to exposure by inhalation to unfiltered or filtered diesel engine exhaust, as described on p. 91 (Brightwell et al., 1986). The authors reported a nonsignificantly increased incidence of tracheal papillomas. [The Working Group noted that no information on tumour incidence was given.]

(a) Inhalation exposure

Mouse: Campbell (1936) exposed two groups of 37 male and 38 female mice [strain unspecified], three months old, by inhalation for 7 h per day on five days per week for about two years to one of two gasoline engine exhaust emissions: A was from a four-cylinder, 23-horse power, ordinary gasoline engine and B from a six-cylinder, 24-horse power engine run on gasoline with tetraethyllead 1:1800. Exposure was to a dilution of 1:145 in air for 4 h in the morning and to a dilution of 1:83 for 3 h in the afternoon. [The total particulate content of the exhaust and the lead concentration were not specified.] Of the animals exposed to exhaust emissions from car A, 9/75 had primary lung tumours compared to 8/74 controls; of those exposed to emissions from car B, primary lung tumours were seen in 12/75 animals compared to 6/70 controls. [Survival data not given.] Other types of tumours observed included mammary tumours and skin cancers among both treated groups and controls. [The Working Group noted the inadequate reporting of the study.]

Two groups of female ICR mice [initial numbers and age unspecified] were either exposed by inhalation to 0.1 mg/m³ gasoline exhaust (1:250 dilution of emission from a small gasoline engine; carbon monoxide, 300 ± 50 ppm (350 ± 60 mg/m³); nitric oxide, 0.21 ppm (0.3 mg/m³); nitrogen dioxide, 0.08 ppm (0.16 mg/m³) [total particulate concentration unspecified]) for 2 h per day on three days per week for six to 12 months, or were administered urethane (0.01%) in the drinking-water until sacrifice (Yoshimura, 1983). No untreated control group was included. Lung adenomas were found in 2/19 exposed mice killed between seven and 12 months; the incidence of tumours (adenomas and adenocarcinomas) in the urethane-treated group was 21/25. [The Working Group noted the short period of treatment, the short observation time and the absence of a control group.]

Rat: Groups of 72 male and 72 female Fischer 344 rats, six to eight weeks old, were exposed to one of three dilutions of gasoline engine exhaust from a 1.6–1 displacement engine operated according to the US-72 (FTP) driving cycle; exposure was for 16 h per day on five days per week for two years (Brightwell et al., 1986). Further groups were exposed to exhaust from a gasoline engine fitted with a three-way catalytic converter. The exhaust was diluted by a constant volume of 800 m³ air or at further dilutions of 1:3 or 1:9 of this mixture in air; the particulate concentration was less than the detection limit of 0.2 mg/m³. The concentration of nitrogen oxides in the high dose of exhaust from the engine without a converter was 49 ± 5 ppm and that of carbon monoxide was 224 ± 32 ppm (260 ± 36 mg/m³). Two control groups of 144 rats of each sex were exposed to conditioned air. After the exposure period, animals were maintained for a further six months in clean air. No increase in lung tumour incidence was reported among rats exposed to gasoline engine exhaust as compared with controls. [The Working Group noted the inadequate reporting of the study.]

Three groups of 80–83 female Bor: WISW rats, ten to 12 weeks old, were exposed by inhalation to 1:61 or 1:27 dilutions with clean air of leaded gasoline engine exhaust generated by a 1.6–1 engine operated according to the US-72 (FTP) driving cycle or to clean air (Heinrich et al., 1986c). The lead content of the fuel was 0.3–0.56 g/l. Mean concentrations of exhaust components measured in the inhalation chambers were (high [low]): carbon monoxide, 350 ± 24 [177.5 ± 12.5] mg/m³; nitric oxide, 28 ± 3 [13.7 ± 1.5 ] mg/m³; nitrogen dioxide, 1.9 ± 0.4 [1.0 ± 0.2] mg/m³; particles, 95.8 ± 16.5 [47.9 ± 20.2] 25g/m³. About 35% of the particulate mass was lead. Exposure was for 18–19 h per day on five days per week for two years, followed by a maximal observation period of six months in clean air. Mean survival time of exposed and control animals was 105 weeks. Exposure to either concentration (1:61 or 1:27) of gasoline exhaust did not produce a significant increase in lung tumour incidence: 1/83 exposed to 1:61 had a squamous-cell carcinoma and 3/78 exposed to 1:27 had two squamous-cell carcinomas and one adenoma; 1/78 controls had an adenoma. In addition, one animal in each of the three exposure groups showed a tumour in the nasal cavities. [The Working Group noted that the nonlead particulate concentration was less than 1/20 the lowest level of particulates that produced an excess of lung tumours in the studies of diesel exhaust. The highest levels of gasoline engine exhaust that can be tested are limited by the toxicity of carbon monoxide.]

Hamster: Three groups of 80–83 female Syrian golden hamsters, ten to 12 weeks old, were exposed to gasoline engine exhaust, as described above but without the six-month observation period (Heinrich et al., 1986c). Median survival in treated and control groups was 70 weeks. One of 75 animals exposed to the high concentration of exhaust (1:27) and three of 80 exposed to the low concentration (1:61) had a tumour of the respiratory tract. No respiratory tract tumour occurred in the 83 controls. [The Working Group noted that the nonlead particulate concentration was less than 1/20 the lowest level of particulates that produced an excess of lung tumours in the studies of diesel exhaust. The highest levels of gasoline engine exhaust that can be tested are limited by the toxicity of carbon monoxide.]

Brightwell et al. (1986) exposed groups of 52 male and 52 female Syrian hamsters, six to eight weeks of age, to gasoline engine exhaust, as described on p. 98. Two control groups of 104 hamsters of each sex were exposed to conditioned air only. The authors reported that respiratory tract tumours in treated hamsters were rare and not related to treatment. [The Working Group noted the inadequate reporting of the data.]

Dog: Stara et al. (1980) exposed seven groups of 12 female beagle dogs, four months of age, to exhaust from a six-cylinder, 2.4–1 gasoline engine run on leaded fuel and operated to simulate urban driving, and to specific pollutants found in gasoline engine exhaust (dilution, 1:570 in air). The groups were exposed to nonirradiated exhaust, to exhaust irradiated with ultra-violet, to sulfur dioxide and sulfuric acid, to nonirradiated exhaust plus sulfur dioxide and sulfuric acid, to exhaust irradiated with ultra-violet plus sulfur dioxide and sulfuric acid, to nitrogen oxides with high nitrogen dioxide and to nitrogen oxides with high nitric oxide. A group of 20 dogs was exposed to clean air. The exhaust contained 100 ppm (115 mg/m³) carbon monoxide and 24–30 ppm hydrocarbon expressed as methane. The irradiated exhaust contained 0.5–1.0 ppm (1–2 mg/m³) nitrogen dioxide, 0.1 ppm (0.12 mg/m³) nitric oxide and 0.2–0.4 ppm oxygen expressed as 0₃. The concentration of lead measured in the different exposure atmospheres was 14–26 µg/m³. The dogs were exposed for 16 h per day for 68 months and then held in clean air for 29–36 months. Complete necropsies were performed on 85 dogs. No lung tumour was observed in the 40 exposed or 17 control dogs. [The Working Group noted that the concentrations of particles in the exposure atmospheres were not given.]

(b) Intratracheal or intrapulmonary administration

Rat: Groups of 34–35 inbred female Osborne-Mendel rats, three months old, received a single implantation of 5.0 or 10.0 mg/animal of gasoline engine exhaust condensate, 4.36, 8.73 or 17.45 mg/animal of a PAH-free fraction, 0.50, 0.99 or 1.98 mg/animal of a fraction of PAHs with two to three rings, 0.14, 0.28 or 0.56 mg/animal of a fraction of PAHs with more than three rings, or 0.03, 0.10 or 0.30 mg benzo[a]pyrene in beeswax:trioctanoin (1:1) into the left lobe of the lung and were observed until natural death (Grimmer et al., 1984). The exhaust was produced by a 1.5–1 passenger car engine operated on the European test cycle. One control group of 34 rats received an injection of the vehicle only, and another control group of 35 animals remained untreated. At death, animals were autopsied and lungs were examined histopathologically. Mean survival times in the treated groups and controls were similar, ranging from 80–111 weeks. Only the fraction containing PAHs with more than three rings produced lung tumour (carcinomas and sarcomas) incidences comparable to those induced by total exhaust condensate (4/35, 17/34 and 24/35 versus 7/35 and 20/35). No lung tumour was observed in the untreated or vehicle controls. A dose-response relationship was obtained with the total condensate and with the fraction of PAHs with more than three rings.

Hamster: In an experiment by Mohr et al., (1976) and Reznik-Schüller and Mohr (1977), two groups of six male Syrian golden hamsters, 12 weeks old, each received intratracheal instillations of 2.5 or 5 mg gasoline exhaust condensate, prepared from emissions of a common German passenger car operating according to the European test cycle and containing 340 µg/g benzo[a]pyrene, in Tris-HCI and EDTA solution. Treatment was every two weeks for life. Moribund animals were killed and their lungs examined histologically for tumours. A further group of six animals was treated with solvent only and were sacrificed after the last exhaust condensate-treated animal had died. Survival times ranged from 30–60 weeks, during which time animals had received 15–30 instillations of condensate. All condensate-treated animals developed pulmonary adenomas.

Groups of 30 male Syrian golden hamsters, 16 weeks of age, received intratracheal instillations of 0.2 ml of a gasoline exhaust condensate from a 1.5–1 engine, its fractions, including the methanol phase, the cyclohexane phase II and the nitromethane phase, a reconstitution product of these fractions, a synthetic mixture of pure carcinogenic PAHs or 40 µgbenzo[a]pyrene in Tris-buffer/saline; treatment was every two weeks until natural death (Künstler, 1983). One group of 30 untreated animals and one group of 30 solvent-treated animals served as controls. Tracheas and lungs of all hamsters were examined histologically by light microscopy. Survival time was 68–87 weeks. No lung tumour was found in animals treated with the condensate or its fractions. In the benzo[a]pyrene-treated group, one mucoepidermoidal carcinoma of the respiratory tract and one lung adenoma were found; one animal treated with cyclohexane phase II (0.13 mg/animal; 10.7 µg benzo[a]pyrene equivalents) had a lung adenoma.

(c) Skin application

Mouse: A group of 108 C57B1 mice [age and sex unspecified] received skin applications of a concentrated benzene extract of particles from a V8 gasoline engine [procedures unspecified] (Kotin et al., 1954). Among 86 mice surviving at the appearance of the first skin tumour (390 days), 38 developed 68 skin tumours, including 22 skin carcinomas. Among 69 benzene-treated controls, 42 survived to the time of appearance of the first skin tumour in treated mice; no skin tumour was reported.

Wynder and Hoffmann (1962) gave groups of 50 female Swiss (Millerton) mice, six weeks of age, skin applications of 5,10,25, 33 or 50% solutions in acetone of the ‘tar’ from a V8 gasoline engine (Hoffmann & Wynder, 1962b) exhaust extracted with benzene. Treatment was given three times a week for 15 months; the mice were observed for a further three months, at which time they were killed. Thirty mice painted with acetone served as controls. The numbers of mice with skin papillomas at 18 months were 0,4,50,60 and 60% in the control, 5, 10, 25 and 33% dose groups, respectively; the corresponding incidences of skin carcinomas were 0, 4, 32, 48 and 54, respectively. In the high-dose group, all mice had died by ten months; 70% had skin papillomas and 4% had skin carcinomas.

In similar studies by Hoffmann et al. (1965), the incidence of skin papillomas and carcinomas was higher in 20 Swiss ICR mice treated with extracts of exhaust from a V8 engine that used approximately 1 1 of engine oil/200 miles (0.3 1/100 km) than in those treated with exhausts from an engine that used approximately 1 1 of oil/1600 miles (0.04 1/100 km).

Brune et al. (1978) gave groups of 50 or 80 female random-bred CFLP mice, approximately 12 weeks of age, skin applications of an exhaust condensate produced from a 1.5–1 gasoline engine during a European test cycle, fractions of this condensate or benzo[a]pyrene in 0.1 ml dimethyl sulfoxide:acetone (3:1) twice a week for life. The groups treated with the total condensate received doses of 0.526, 1.579 or 4.737 mg/animal (0.15, 0.45 or 1.35 µg/animal benzo[a]pyrene equivalents) per treatment; the two groups treated with the methanol phase (66% of the total condensate) received doses of 1.389 or 4.168 mg/animal (0.60 or 1.80 µg/animal benzo[a]pyrene equivalents); those treated with the cyclohexane phases I and II (34% and 17% of the total condensate), the nitromethane phase (17% of the total condensate) and a reconstitution of the fractions received 0.30 and 0.90 µg/animal benzo[a]pyrene equivalents. Three further groups of 50 mice received applications of 1.92, 3.84 or 7.68 µg/animal benzo[a]pyrene. One control group received applications of the vehicle alone and another remained untreated. Animals with advanced malignant tumours were killed; all other animals were observed until natural death. Statistical analysis of the results revealed a linear relationship between the percentage of animals with local tumours (squamous-cell papillomas or carcinomas) and dose for the nitromethane phase (16.4 and 68.9%), the cyclohexane phase I (13.7 and 68.8%), the reconstitution (7.9 and 54.7%) and the total condensate (3.9, 35.1 and 76.9%). Local tumour rates in mice treated with total condensate were significantly higher than those in mice treated with benzo[a]pyrene (19.5, 15.2 and 60%) or the PAH-free fractions (methanol phase (2.6 and 5.9%) and cyclohexane phase II (2.8 and 1.5%)), which did not differ significantly from controls (1.3 and 0%). A second experiment by the same group using 40 mice per group gave similar results; however, local tumour incidences were significantly higher in the first experiment, probably due to minor differences in experimental techniques.

Grimmer et al. (1983a) gave groups of 65 or 80 female CFLP mice, seven weeks old, dermal applications of extracts of an exhaust condensate from a 1.5–1 gasoline engine run on the European test cycle, its fractions or benzo[a]pyrene in 0.1 ml dimethyl sulfoxide:acetone (1:3) solvent; treatment was given twice a week for 104 weeks. Doses administered were: total condensate — 0.292, 0.875 or 2.626 mg/animal (0.12, 0.36 or 1.09 µg/animal benzo[a]pyrene equivalents); benzo[a]pyrene, 0.0039, 0.0077 or 0.0154 mg/animal; the methanol phase (PAH-free fraction), 0.97 or 2.9 mg/animal (0.48 or 1.45 µg/animal benzo[a]pyrene equivalents); the PAH-fraction containing PAHs with two and three rings, 0.152 or 0.455 mg/animal (0.46 or 1.39 µg/animal benzo[a]pyrene equivalents); the PAH-fraction containing PAHs with more than three rings, 0.02 or 0.06 mg/animal (0.24 or 0.73 µg/animal benzo[a]pyrene equivalents); and a mixture of 15 PAHs in a ratio corresponding to that of the automobile exhaust, 0.003 or 0.009 mg/animal (0.24 or 0.73 µg/animal benzo[a]pyrene equivalents). One group treated with 0.1 ml of the solvent only and one untreated group served as controls. Animals with advanced tumours were killed; the remaining animals were observed until natural death. The PAH-free fraction (methanol phase) and the fraction of PAHs with two or three rings produced low rates of skin tumours (carcinomas and papillomas): 11 [13.9%] and one [1.3%] animals with local tumours, respectively, in the high-dose groups. Clear dose-response relationships were demonstrated for tumour incidence in the groups treated with total condensate (six [7.7%], 34 [44.3%] and 65 [83.3%]), in those given the fraction containing PAHs with more than three rings (seven [8.9%] and 50 [63.5%]), in those given the mixture of 15 PAHs (one [1.3%] and 29 [38.7%]) and in benzo[a]pyrene-treated animals (22 [34.4%], 39 [60.9%] and 56 [89.1%]). No local skin tumour was seen in controls. Similar results were obtained by Grimmer et al. (1983b).

Groups of 40 male and 40 female SENCAR mice, seven to nine weeks of age, received single skin applications in 0.2 ml acetone of 0.1,0.5,1,2 or 3 mg of dichloromethane extracts of particulates collected from the emission of an unleaded gasoline engine (of a 1977 model passenger car [engine volume unspecified]) with a catalytic converter (Nesnow et al., 1982a). One week later, all mice received 2 µg TPA in 0.2 ml acetone twice weekly for 24–26 weeks. At that time, the percentages of mice with papillomas and the numbers of papillomas/-mouse in TPA-treated controls were 8% and 0.08 in males and 5% and 0.05 in females, respectively. In the groups treated with both TPA and the gasoline extract, the respective percentages and numbers were: males — 5% and 0.05 (0.1 mg), 13% and 0.15 (0.5 mg), 18% and 0.18 (1 mg), 22% and 0.24 (2 mg) and 18% and 0.24 (3 mg); females — 13% and 0.23 (0.1 mg), 18% and 0.24 (0.5 mg), 10% and 0.13 (1 mg), 21% and 0.23 (2 mg) and 23% and 0.28 (3 mg).

(d) Subcutaneous administration

Mouse: Groups of 87 or 88 female NMRI mice [age unspecified] received a single subcutaneous injection in 0.5 ml tricaprylin of 20 or 60 mg exhaust condensate from a gasoline engine [unspecified] (Pott et al., 1977). A third group of 45 mice was injected three times with 60 mg condensate containing 0.163 µg/mg benzo[a]pyrene. A group of 89 mice that received 0.5 ml tricaprylin alone and a further group of 87 untreated mice served as controls. Animals that developed tumours up to 10 mm in diameter at the application site were killed. The mean survival time in the low- and medium-dose groups was in the range of that of the control groups (80–88 weeks), but was 57 weeks in the high-dose group. The numbers of animals with sarcomas at the injection site were 10/87 (11.5%), 6/88 (6.8%) and 5/45 (11.1%) in the condensate-treated groups and 3.4% in the tricaprylin-treated group.

(e) Administration with known carcinogens

Mouse: Groups of 60 female NMRI mice, eight to ten weeks old, received ten intratracheal instillations of 100 µg benzo[a]pyrene, 20 intratracheal instillations of 50 µg benzo[a]pyrene or ten intratracheal instillations of 50 µgdibenzo[a,h]anthracene, with concomitant exposure to gasoline engine exhaust, as described on p. 99, for 53 weeks only and were observed for a further 40 weeks (Heinrich et al., 1986c). Administration of benzo[a]pyrene or dibenzo[a,h]anthracene with clean air induced a high basic lung tumour rate of 70–90% (adenomas and adenocarcinomas). Mean survival times (75–85 weeks) of exhaust-exposed animals were clearly shorter, with the exception of the groups treated ten times with 100 µg benzo[a]pyrene, in which gasoline exhaust exposure induced a higher incidence of adenocarcinomas (22/38 and 28/40 in the 1:27 and 1:61 dilution groups) but a significantly reduced incidence of adenomas (4/38 and 3/40) compared to clean air controls (20/42 adenocarcinomas, 16/42 adenomas). The total numbers of tumour-bearing animals in clean air and exhaust-exposed groups were not, however, significantly different. In the groups exposed 20 times to 50 µg benzo[a]pyrene, adenocarcinoma induction by the exhaust was inhibited significantly (3/35, 5/36, 15/42 in the 1:27, 1:61 and control groups, respectively). Additional groups of 61–83 newborn NMRI mice received a single subcutaneous injection of 4 µg (females and males) or 10 µg (females only) dibenzo[a,h]anthracene followed by inhalation exposure to one of the two dilutions of gasoline exhaust for six months, after which they were killed; the number of lung tumours per animal was not significantly different from that in controls exposed simultaneously to clean air.

Groups of 86–90 female NMRI mice [age unspecified] were injected subcutaneously with 10,30 or 90 µgbenzo[a]pyrene alone or together with 6.6 or 20 mg exhaust condensate from a gasoline engine [unspecified] (Pott et al., 1977). The dose-response relationship for local sarcomas produced by benzo[a]pyrene (20%, 54%, 76%) was reduced significantly by the addition of both doses of the condensate. The difference was seen most clearly 30 weeks after treatment.

Rat: Two groups of female Sprague-Dawley rats [initial numbers unspecified] were either administered N-nitrosodiisopropanolamine in the drinking-water (0.01%) or were exposed concomitantly by inhalation for 2 h per day on three days per week to gasoline engine (generator EM300) exhaust diluted 1:250 in air for six to 12 months, at which time the animals were killed (Yoshimura, 1983). In animals killed between seven and 12 months, the number of lung tumours (11/37) in the combined treatment group (one adenoma and ten undifferentiated carcinomas, squamous-cell carcinomas, adenocarcinomas and mixed tumours) was significantly greater than that in the 24 nitrosamine controls (two carcinomas; p < 0.05).

Groups of 60 female Bor: WISW rats, ten to 12 weeks old, received 25 daily subcutaneous injections of 0.25 or 0.5 g/kg bw N-nitrosodipentylamine and were exposed to gasoline engine exhaust, as described on p. 99 (Heinrich et al. 1986c). The treatments induced significant increases in the incidences of benign tumours of the whole respiratory tract (in 9/47 and 14/48 rats given the 1:27 and 1:61 dilutions of exhaust and receiving 0.5 g/kg bw nitrosamine, and in 15/50 and 14/45 rats given the 1:27 and 1:61 dilutions and receiving0.25 g/kg bw nitrosamine, respectively) compared with clean air controls (5/48 and 4/46 rats), but decreases in the incidences of malignant tumours (33/47 and 34/48, respectively, compared to 43/48 controls; and 13/50 and 18/45 rats, compared to 29/46 in the groups receiving 0.5 and 0.25 g/kg bw nitrosamine). When lung tumour rates were evaluated separately, the incidences of malignant tumours (mostly squamous-cell carcinomas and adenocarcinomas) were also reduced in nitrosamine-treated rats by exposure to either concentration of exhaust (in 24/48, 25/49 and 40/49 rats in the 0.5 g/kg bw groups and in 11/54, 14/47 and 26/48 rats in the 0.25 g/kg bw groups exposed to 1:27 and 1:61 dilutions and clean air, respectively), whereas the incidence of benign tumours remained unchanged. Rats given the low dose of N-nitrosodipentylamine exposed to 1:61 or 1:27 dilutions of gasoline exhaust showed overall lung tumour rates of 15/47 and 13/54, respectively, versus 27/48 rats treated with nitrosamine but exposed to clean air. In animals given the high dose of N-nitrosodipentylamine, these rates were 33/49 and 28/48, respectively, versus 44/49 controls.

Hamster: Groups of 80–81 female Syrian golden hamsters, ten to 12 weeks old, received a single subcutaneous injection of 3 mg/kg bw N-nitrosodiethylamine (NDEA) or 20 intratracheal instillations of 0.25 mg benzo[a]pyrene and were exposed to gasoline engine exhaust, as described on p. 99 (Heinrich et al. 1986c). Administration of NDEA or benzo[a]pyrene to hamsters exposed to clean air resulted in basic rates of benign respiratory tract tumours of 12.8 and 6.5% of animals, respectively; one malignant tumour of the paranasal cavity was also seen in the group exposed to benzo[a]pyrene. The basic tumour rate was not significantly increased by exposure to either dilution of exhaust. Tumour rates in NDEA- and benzo[a]pyrene-treated animals inhaling the 1:27 dilution of exhaust were approximately 50% lower than those in treated animals inhaling the 1:61 dilution or clean air.

Groups of 52 male and 52 female Syrian hamsters, six to eight weeks old, received a single subcutaneous injection of 4.5 mg/kg bw NDEA three days prior to exposure by inhalation to gasoline engine exhaust, as described on p. 98 (Brightwell et al. 1986). The authors reported that NDEA-treated hamsters had a nonsignificantly increased incidence of tracheal papillomas. [The Working Group noted the inadequate reporting of the data.]

(i) Deposition, clearance, retention and metabolism

Engine exhaust contains material in gaseous, vapour and particulate phases, and the absorption, distribution and excretion of individual constituents is influenced by the phase in which they occur and by the properties of each compound. After inhalation, highly soluble compounds in the gaseous phase, such as sulfur dioxide, are absorbed in the upper airways and do not penetrate significantly beyond the level of the bronchioles. Compounds that interact biochemically with the body are also retained in significant quantities; thus, processes such as binding of carbon monoxide to haemoglobin normally occur in the gas-exchange (pulmonary) region of the lung. Retention characteristics of materials not associated with the particulate phase are highly compound-specific. The factors affecting the uptake of a wide variety of vapours and gases have been summarized (Davies, 1985).

As described on p. 47, a proportion of a compound in the vapour phase condenses onto the particulate material produced in the engine exhaust. The association of a compound with the particulate phase modifies the deposition pattern and affects its lung retention; the lung burden of a compound following continuous exposure to that compound coated on particles may be many times that of continuous exposure to the compound alone (Bond et al., 1986).

Deposition in the respiratory tract is a function of particle size. The median particle size in a variety of long-term exposure systems has been between 0.19 and 0.54 µm (Yu & Xu, 1986), representative of that in an urban environment (Cheng et al., 1984). However, some of the carbonaceous mass in environmental samples results from airborne suspension of material collected in automobile exhaust pipes and is >5 µmin size (Chamberlain et al., 1978); such particles are unlikely to be produced in a static exposure system. Dilution has little effect on the size distribution of particles used in long-term studies (0.3–7 mg/m³; Cheng et al., 1984), although rapid dilution (<1 sec) can lead to a smaller size (0.10–0.15 µm; Chan et al., 1981), The presence of sulfates in the particulate phase (Lies et al., 1986) may lead to enlargement of individual particles in the high humidity of the respiratory tract, thereby altering the deposition pattern (Pritchard, 1987).

Diesel engine exhaust

Deposition: Studies of the deposition of diesel engine exhaust, representative of fresh urban exhaust, are summarized in Table 24; the particle sizes used were in the lower part of the range found in long-term exposure chambers. Deposition following nose-only exposure was measured by radiotracer technique. Data are quoted as a proportion of the amount of inhaled aerosol, which is based on estimates of ventilation rates. [The Working Group noted that the data on deposition of diesel particles in rats are in broad agreement with data for other particulate materials of similar size (Raab et al., 1977; Wolff et al., 1984).]

Table 24.

Experimental deposition in the respiratory tract of diesel engine exhaust particles.

A model for the deposition of diesel exhaust particles predicts that, as the median size increases from 0.08 to 0.30 µm,total deposition in rats falls from 25 to 15%, tracheo-bronchial deposition from 5 to 2% and pulmonary deposition from 12 to 5%; upper respiratory tract deposition remains constant at 8% (Yu & Xu, 1986). The model predicts that pulmonary deposition will vary only with (body weight)^−0.14, since diffusion is the predominant mechanism (Xu & Yu, 1987). [The Working Group noted that this model is in good agreement with the observed deposition of other particles (e.g., Raab et al., 1977; Wolff et al., 1981, 1984)].

Following exposure of rats for six, 12, 18 and 24 months to 0.4, 3.5 and 7.1 mg/m³ diesel exhaust particles, there was no significant effect of length of exposure or exposure concentration on the deposition of 0.1 µm gallium oxide particles (Wolff et al., 1987).

Mucociliary clearance: The clearance of particles from the lung following a single exposure to radiolabeled diesel particles is summarized in Table 25. The fast phase of clearance is conventionally assumed to be due to mucociliary action, the remainder (slow phase) to pulmonary clearance. The variation in the fraction of the lung deposit cleared by mucociliary action (i.e., the tracheobronchial deposit) is linked to particle size and hence deposition pattern. [The Working Group noted that Gutwein et al. (1974) give no information on particle size and that, without this, the high tracheobronchial deposit cannot be accounted for.]

Table 25.

Clearance of diesel exhaust particles from rat lung following single exposures.

In rats exposed for short periods (4–100 h) to diesel exhaust with particulate concentrations in the range 0.9–17 mg/m³, a dose-dependent reduction in mucociliary clearance occurred, although the effect was less marked on exposure to the gas phase alone (Battigelli et al., 1966). No such effect occurred in sheep exposed for 30 min to concentrations of 0.4–0.5 mg/m³ of resuspended diesel particles, i.e., in the absence of the gas phase (Abraham et al., 1980). Exposure-related differences in tracheal mucociliary clearance have also been reported over 1–12 weeks in rats exposed to 1 and 4.4 mg/m³ particulates in diesel exhaust. However, in another study, there was no effect on tracheal mucociliary clearance of exposures of six to 24 months to particulate concentrations of 0.4–7.1 mg/m³ (Wolff et al., 1987). [The Working Group noted that there may be some impairment of mucociliary clearance, possibly caused by the gas phase of engine exhaust, but that its effect is of limited significance in the long term.]

Pulmonary (alveolar) clearance: The pulmonary clearance of diesel particles is very much slower than the mucociliary clearance (see Table 25). On the basis of these data, the lung burden of rats during protracted exposure should tend exponentially toward an equilibrium value at 12 months. In rats exposed to diesel exhaust with a particulate concentration of 0.3 mg/m³, there was evidence of equilibration after 12 months (only a 2.5-fold increase over 24 months); however, with exposures of 3.5 and 7.0 mg/m³, lung burdens increased steadily (five to 11 fold) over 24 months. This has been referred to as the ‘overload’ phenomenon (Wolff et al., 1987). The clearance rate of insoluble particles following prolonged exposure to diesel exhaust at a variety of concentrations and durations also indicates impaired long-term clearance (Wolff et al., 1984). Thus, it appears that the normal clearance mechanisms become seriously impaired, leading to very long-term retention of material in the lung, usually referred to as ‘sequestration’.

Results of studies on particulate clearance in rats following repeated exposures to diesel exhaust are summarized in Table 26. Lung clearance was estimated either by exposure to a pulse of ¹⁴C-labelled diesel exhaust particles at the end of the cumulative exposure (Chan et al., 1984; Lee et al., 1987) or by measuring the lung burden of soot spectrophotometrically (Griffis et al., 1983). Also included are data on the clearance of a pulse of radiolabeled fused aluminosilicate particles following exposure to diesel exhaust for two years at particulate concentrations between 0.4 and 7.0 mg/m³ (Wolff et al., 1987). [The Working Group noted that pulse techniques measure only the clearance of the material that has most recently entered the lung. Since there is no difference between this and total soot measurements, deposition, and hence ventilation, must continue to occur in those areas where clearance is impaired, confirming the findings of Wolff et al. (1987) that deposition is unaffected by prolonged exposure to diesel exhaust.]

Table 26.

Pulmonary clearance in rats of insoluble particles following exposure to diesel exhaust.

Pulmonary clearance as a function of contemporary lung burden has also been considered (McClellan, 1986). In an analysis of the published data, Wolff et al. (1986) concluded that in rats sequestration becomes a significant burden at a certain level [about 1 mg/lung] and is related to the rate of accumulation; i.e., short exposure to a high concentration produces an effect at a lower lung burden than more protracted exposure at a lower concentration. Thus, there is no strong relationship between the half-time of clearance and cumulative exposure.

[The relationship between half-times for pulmonary clearance of diesel exhaust particles and other insoluble particles in the rat following exposure to diesel exhaust and an ‘exposure rate’, calculated by the Working Group from cumulative exposure, mg/m³ × weeks × (h/week)/168, is plotted in Figure 8. The Working Group noted that there is an effect on clearance at ‘exposure rates’ above 10 mg/m³ × week (i.e., continuous exposure to 0.2 mg/m³ or exposure for 40 h per week to 0.8 mg/m³ for one year) and a strong suggestion that impaired clearance occurs over the whole range of ‘exposure rates’ studied.]

Fig. 8.

Pulmonary clearance in rats of diesel exhaust particles and other insoluble particles following exposure to diesel exhaust.

Studies in rats on the effect of exposure to diesel exhaust on the clearance of metal oxide particles containing a γ-emitting isotope are summarized in Table 27 (Bellmann et al., 1983; Heinrich et al., 1986a; Lewis et al., 1986; Wolff et al., 1987). The control animals cleared the metal oxide particles much faster than they did diesel particles or fused aluminosilicate particles (see Table 26; Wolff et al., 1987). [The Working Group noted that this suggests that clearance of metal oxides involves a significant soluble component.]

Table 27.

Pulmonary clearance in rats of metal oxide particles following exposure to diesel engine exhausts.

[The relationship between half-times for pulmonary clearance of metal oxide particles in rats following exposure to diesel exhaust and an ‘exposure rate’ calculated by the Working Group is plotted in Figure 9. The Working Group noted that impaired clearance of metal oxide particles does not become apparent until significantly higher values of ‘exposure rate’ than in the studies on diesel and fused aluminosilicate particles and considered that the differences in the results could be explained by continuing solubility masking an impairment in mechanical clearance, implying that sequestration is primarily a mechanical effect. For comparison, data for gasoline from Bellmann et al. (1983) have been added.]

Fig. 9.

Pulmonary clearance of metal oxide particles in rats following exposure to engine exhaust.

After only two months' exposure of rats to a diesel exhaust particulate concentration of 2 mg/m³, clearance of metal oxide particles was significantly faster than in controls, suggesting a stimulated lung response; no such effect was observed subsequently (Oberdoerster et al., 1984; Lewis et al., 1986). [The Working Group noted that overloading had probably occurred.]

The rate of clearance of ferric oxide in hamsters was slightly lower (75 ± 40 days) following one year's exposure to diesel exhaust particles (4 mg/m³) than that in clean-air controls (55 ± 17 days; Heinrich et al., 1986a). In another study, only 10% clearance of ¹⁴C-labelled diesel particles was observed 400 days after a single exposure of guinea-pigs (Lee et al. 1983). Six months after a three-month exposure of mice, rats and hamsters to diesel exhaust (particles, 1.5 mg/m³), the mice appeared to have a slower clearance than rats and hamsters (Kaplan et al., 1982).

The gas phase alone appears to have no effect on pulmonary clearance in rats or hamsters (Heinrich et al., 1986a). Clearance of diesel particles following prolonged exposure to a carbon black aerosol of similar size showed a pattern of impairment similar to that observed after diesel exposure (see Fig. 8), strongly suggesting that dust overloading per se impairs mechanical clearance (Lee et al., 1987). [The Working Group noted that the half-time lung clearance of carbon black is shorter than that of diesel exhaust at similar ‘exposure rates’. This may reflect a local effect of diesel particles on the alveolar macrophages which mediate mechanical clearance; diesel particles depress the phagocytic capacity of macrophages, whereas coal dust activates them (see below and Castranova et al., 1985).]

The majority of the particles that are cleared by macrophages from the pulmonary region leave via the ciliated epithelium and are excreted via the gut. However, a proportion penetrate the lymphatic system, borne by macrophages, and are filtered by the lymph nodes to form aggregates of particles (Vostal et al., 1981). It has been estimated that one-third of clearance occurred via this route during the first 28 days after exposure of rats to diesel exhaust (Chan et al., 1981). [The Working Group noted that there is no information on how this proportion changes with time or with prolonged exposure.]

Retention: The retention of the organic compounds associated with exhaust particles has been reviewed (McClellan et al., 1982; Vostal et al., 1982; Holmberg & Ahlborg, 1983; Vostal, 1983; Wolff et al., 1986). Organic compounds adsorbed on exhaust particles can be extracted by biological fluids, as has been observed in assays for mutagenesis (Claxton, 1983; Lewtas & Williams, 1986; see p. 121). The half-time of the slow phase of lung clearance for ¹⁴C derived from labelled diesel exhaust was 25 days in rats (Sun & McClellan, 1984), and that for ³H-benzo[a]pyrene coated on diesel particles was 18 days (Sun et al., 1984). The retention of 1-nitropyrene adsorbed onto diesel exhaust particles is described in the monograph on that compound.

No data were available on changes in the retention of individual compounds after prolonged exposure to diesel exhaust.

Metabolism: The metabolism of several components of engine exhausts has been reported previously: some polycyclic aromatic hydrocarbons (IARC, 1983), formaldehyde (IARC, 1982a), lead (IARC, 1980), nitroarenes (IARC, 1984) and benzene (IARC, 1982b). The metabolism of 1-nitropyrene associated with diesel exhaust particles is described in the monograph on that compound.

The metabolism of benzo[a]pyrene coated on diesel exhaust particles has been studied in different experimental systems. Fischer 344 rats were exposed for 30 min by nose-only inhalation to ³H-benzo[a]pyrene adsorbed onto diesel engine exhaust particles. The majority (65–76%) of the radioactivity retained in the lungs (as determined by high-performance liquid chromatography) 30 min and 20 days after exposure was associated with benzo[a]pyrene. Smaller amounts of benzo[a]pyrene-phenols (13–18%) and benzo[a]-pyrene-quinones (5–18%) were also detected. No other metabolite was found (Sun et al., 1984).

The pulmonary macrophages of dogs metabolized 1 µM ¹⁴C-benzo[a]pyrene, either in solution or coated on diesel particles, into benzo[a]pyrene-7,8-, -4,5- and -9,10-dihydrodiols (major metabolites) as well as into benzo[a]pyrene-phenols and benzo[a]pyrene-quinones (minor metabolites). The total quantity of metabolites did not differ when macrophages were incubated with either benzo[a]pyrene in solution or benzo[a]pyrene coated on diesel particles (Bond et al., 1984).

Fischer 344 rats were exposed to diesel engine exhaust (7.1 mg/m³ particles) for about 31 months. After sacrifice, DNA was extracted from the right lung lobe and analysed for adducts by ³²P-postlabelling: more DNA adducts were found in the exhaust-exposed group than in the unexposed group (Wong et al., 1986).

Fischer 344 rats and Syrian golden hamsters were exposed to different dilutions of diesel engine exhaust for six months to two years, when blood samples were analysed for levels of haemoglobin adducts (2-hydroxyethylvaline and 2-hydroxypropylvaline) by gas chromatography-mass spectrometry. A dose-dependent increase in the level of haemoglobin adducts was found, corresponding to the metabolic conversion of about 5–10% of inhaled ethylene and propylene to ethylene oxide and propylene oxide, respectively (Törnqvist et al., 1988).

Gasoline engine exhaust

Deposition: In a study on the deposition of particles from inhaled gasoline exhausts (mass median diameter, 0.5 µm) in rats, mean total deposition of particles was 30.5%. Most deposition occurred in the alveolar region and in the nasal passages (Morgan & Holmes, 1978). In this study, the concentration of carbon monoxide in the gasoline exhaust was reduced before inhalation, and the particles were larger than those of the diesel exhausts reported. [The Working Group noted that the greater deposition of gasoline exhaust particles is consistent with the larger size of the particles and does not imply any fundamental difference in deposition between diesel and gasoline exhausts particles.]

Clearance: The results of a study on pulmonary clearance of ferric oxide by rats and hamsters following exposure to gasoline engine particles (0.04 and 0.09 mg/m³) for two years are summarized in Table 28 (Bellmann et al., 1983). Clearance was similar to that in controls and in animals exposed to diesel exhaust (see Table 27). [The Working Group noted that, on the basis of the data concerning exposure to diesel exhaust, clearance of metal oxide particles would not be impaired by exposures to such low concentrations.]

Table 28.

Pulmonary clearance by rats and hamsters of ferric oxide particles following exposure to gasoline engine exhausts.

Metabolism: As reported in an abstract, crude extracts of gasoline exhaust were applied topically to male BALB/c mice over a period of one to two weeks and DNA was isolated from the treated skin for analysis by ³²P-postlabelling. The major DNA adduct derived from benzo[a]pyrene-7,8-dihydrodiol-9,10-epoxide was found in exposed mice(Randerath et al. 1985).

Fischer 344 rats and Syrian golden hamsters were exposed to different dilutions of gasoline engine exhaust for six months to two years, and blood samples were analysed for levels of 2-hydroxyethylvaline and 2-hydroxypropylvaline in haemoglobin by gas chromatography-mass spectrometry. A dose-dependent increase in the level of haemoglobin adducts was found, corresponding to the metabolic conversion of about 5–10% of inhaled ethylene and propylene to ethylene oxide and propylene oxide, respectively (Törnqvist et al., 1988).

(ii) Toxic effects

(iii) Effects on reproduction and prenatal toxicity

(iv) Genetic and related effects

The genetic and related effects of diesel and gasoline engine exhausts have been reviewed (Lewtas, 1982; Claxton, 1983; Holmberg & Ahlborg, 1983; Ishinishi et al., 1986b; Lewtas & Williams, 1986).

Since engine exhaust is difficult to administer in short-term tests, studies have been conducted on several components and fractions of exhausts. Early studies were conducted on exhaust condensates; recent dilution sampling methods have permitted the collection of soot particles. Biological studies have been conducted on collected particles and on various extracts of particles, primarily extractable or soluble organic matter. Several solvents are effective for extracting organic material from diesel and gasoline particles (Claxton, 1983); dichloromethane is that used most commonly. More volatile organic compounds are collected on adsorbent resins and extracted for bioassay. Only limited studies have been conducted on direct exposure to gaseous and whole exhausts.

Studies of genotoxicity are thus conducted on particles, particulate extracts, volatile organic condensates or whole emissions, and the results are expressed as activity per unit mass. In order to compare different emissions, genotoxicity is often expressed as emission rate or genotoxicity per unit distance driven or per mass of fuel consumed. Thus, for example, the mutagenic activity in Salmonella typhimurium TA98 of several gasoline particulate extracts is greater than that of diesel particulate extracts per unit mass of organic extract, while the mutagenic emission factor per kilometre driven for gasoline automobiles is less than that for diesel engines (Lewtas & Williams, 1986). The data on gasoline engine exhausts are considered together, whether or not the engine used was equipped with a catalyst and regardless of the type of fuel used (e.g., leaded or unleaded). When this information was available to the Working Group, however, it is noted in the text.

The genotoxic activity of diesel particulate extracts is generally decreased by the addition of a metabolic activation system (e.g., Aroclor 1254-induced or uninduced liver 9000 × g supernatant (S9), lung S9, microsomal preparations). In contrast, the genotoxicity of gasoline particulate extracts is generally increased by the addition of metabolic activation (Claxton, 1983; Lewtas & Williams, 1986).

(i) Deposition, clearance, retention and metabolism

The factors affecting the uptake of gases and vapours, including model calculations for their absorption in the different regions of the human respiratory tract, have been summarized (Davies, 1985).

Gasoline engine exhaust

The results of two laboratory experiments in which human volunteers inhaled the exhaust from an engine run on gasoline containing ²⁰³Pb-tetraethyllead are summarized in Table 29. In one of the experiments, the exhaust was contained in a 600–1 chamber; the concentrations of carbon monoxide and carbon dioxide were reduced using chemical traps; median particulate size was about 0.4 µum (Chamberlain et al., 1975) or 0.35 and 0.7 µm, resulting in an aerosol considered typical of urban environments (Chamberlain et al., 1978). In the other experiment, the exhaust was rapidly diluted in a wind tunnel which prevented coagulation of the primary exhaust particles and resulted in aerosols with median particulate sizes of 0.02–0,09 µm. Both experiments were conducted with a variety of breathing patterns, which were monitored but not controlled. Total deposition was relatively constant at 30% over a wide range of breathing patterns for sizes typical of urban aerosols (Chamberlain, 1985). However, as the size of the primary particles decreased (below 0.1 µm), deposition increased sharply, and the length of the respiratory cycle (time between the start of successive breaths) significantly affected deposition. [The Working Group noted that these data are in broad agreement with those for other particulate materials of similar size (Heyder et al., 1983; Schiller et al., 1986.]

Table 29.

Total deposition (%) of leaded gasoline particles as a function of size and breathing pattern.

In a separate analysis of the same data, deposition was shown to increase with respiratory cycle in an approximately linear fashion — ranging from 10% at 3 sec to 55% at 20 sec; the slope of this line was somewhat dependent on tidal volume. A small, but significant effect of expiratory reserve volume on deposition was observed: total deposition dropped by a factor of 1.2 for an increase in expiratory reserve volume of 2.5 1 (Wells et al., 1977).

In a third study, measurements of total deposition were performed in the field by comparing inhaled and exhaled airborne lead concentrations; the method was found to give results comparable to experimental measurements involving ²⁰³Pb. Total deposition was measured for inhalation at an average breathing pattern of 0.81 and a respiratory cycle of 5.2 sec in persons seated by a motorway (61 %), by a roundabout (64%), in an urban street (48%) and in a car park (48%). Median particulate sizes in the breath of persons near quickly moving traffic (0.04 µm) were found to be much smaller than those in persons in the urban environment or in a car park (0.3 µm), although the air near roundabouts also contained a large proportion by mass of adventitious particles (2 µm) (Chamberlain et al., 1978).

Lung clearance was best described by a four-component exponential clearance. The first two phases (half-times, 0.7 and 2.5 h) were similar for exhaust particles, lead nitrate (which is soluble) and lead oxide (which is insoluble), and therefore probably represent mucociliary clearance (Chamberlain et al., 1975, 1978). On average, 40% of lung deposition of 0.35-µm aerosols was in the pulmonary region and 60% in the tracheobronchial region. The removal of lead compounds from the pulmonary region was described by a two-component exponential with half-times of 9 and 44 h; one exception was the removal of lead from highly carbonaceous particles, which exhibited half times of 24 and 220 h (Chamberlain et al., 1978; Chamberlain, 1985).

No data on the metabolism in humans of gasoline engine exhaust were available to the Working Group.

(ii) Toxic effects

Early studies involving human volunteers showed that exposure to gasoline engine exhaust may cause headache, nausea and vomiting (Henderson et al., 1921), Sayers et al. (1929) monitored the carbon monoxide content of gasoline engine exhaust gas-air mixtures and found a relationship between increasing carbon monoxide concentration, carboxyhaemoglobin (COHb) level and reports of headache in six men exposed to atmospheres containing 229–458 mg/m³ carbon monoxide. In a more recent study of ten patients with angina (Aronow et al., 1972), significant increases in COHb levels and significant reductions in exercise performance until onset of angina symptoms were observed in persons driving for 90 min in heavy traffic, as compared with tests both before the experiment and after breathing purified air for 90 min.

Among six volunteers exposed for 3.7 h to diesel engine exhaust gases containing about 4 mg/m³ nitrogen dioxide, there was no increase in urinary thioether concentration (Ulfvarson et al., 1987).

Effects of exposure to diesel engine exhaust on the lung have been reviewed (Calabrese et al., 1981), Although bus garage and car ferry workers, exposed occupationally to mixtures of gasoline and diesel engine exhausts, had lower mean levels of respiratory function (forced respiratory volume in 1 sec (FEV₁) and forced vital capacity (FVC) than expected, they showed no change in these measures over working shifts (for exposure measurements, see Tables 17 and 21, respectively). In contrast, workers on roll-on roll-off ships, exposed mainly to diesel engine fumes, showed statistically significant reductions in FEV₁ and FVC during working shifts (for exposure measurements, see Table 18). These reductions were reversible, however, the levels returning to normal after a few days with no exposure. The work-shift concentrations of nitrogen dioxide and carbon monoxide in these three groups averaged 0.54 mg/m³ and 1.1 mg/m³, respectively (Ulfvarson et al., 1987). A small reduction in FEV₁/FVC and in FEF_25–75% (forced expiratory flow at 25–75% of forced vital capacity) was also observed at the end of a work shift among a group of chain-saw operators (Hagberg et al., 1983; for exposure measurements, see Table 22). Concentrations of diesel engine emissions in coal mines, involving, on average, 0.6 mg/m³ nitrogen dioxide and 13.7 mg/m³ carbon monoxide, were not associated with decrements in the miners' ventilatory function (Ames et al., 1982).

Studies in which changes in COHb levels were investigated over the course of a work shift are summarized in section 2 (pp. 69–73).

Possible effects on the lung of chronic occupational exposures to low levels of diesel engine exhaust emissions were studied cross-sectionally in railroad engine house workers (Battigelli et al., 1964), in iron ore miners (Jörgensen & Svensson, 1970), in potash miners (Attfield et al., 1982), in coal miners (Reger et al., 1982; for exposure measurements, see Table 15), in salt miners (Gamble et al., 1983), in coal miners exposed to oxides of nitrogen generated (in part) by diesel engine emissions underground (Robertson et al., 1984) and in bus garage workers (Gamble et al., 1987b). Effects of relatively high concentrations of automobile emissions have been described among bridge and road tunnel workers in two large cities (Speizer & Ferris, 1963; Ayres et al., 1973; for exposure measurements, see Table 19). Changes in lung function over a five-year period have also been studied longitudinally among coal miners working underground in mines with and without diesel engines (Ames et al., 1984). Some, but not all, of the results from these various studies showed decrements in lung function and increased prevalence of respiratory symptoms in subgroups exposed to engine emissions.

Exposure to engine exhaust has also been associated with irritation of the eyes (Waller et al., 1961; Battigelli, 1965; Hamming & MacPhee, 1967; Hagberg et al., 1983).

A 15-year follow-up of 34 156 members of a heavy construction equipment operators' union showed a highly significant overall excess of deaths certified as due to emphysema (116 observed, 70.2 expected), and this excess appeared to be higher among men with longer membership in the union (Wong et al., 1985). No data on smoking habits were included in the mortality analyses, and the authors noted that they were unable to estimate the degree to which exposure to diesel engine emissions (as distinct from other occupational factors, such as exposure to dust) might have contributed to the excess mortality from emphysema.

Another cohort study, of 1558 white motor vehicle examiners, yielded a slight excess of deaths from cardiovascular disease (124 observed, 118.4 expected) in a 29-year follow-up. The excess was more pronounced for deaths occurring during the first ten years of employment (28 observed, 20.9 expected; Stern et al., 1981). [The Working Group noted that the excesses observed are easily attributable to chance (p > 0.1).] A 32-year follow-up of 694 Swedish bus garage employees also showed a small, statistically nonsignificant, excess of deaths from cardiovascular disease (121 observed, 115.9 expected) which showed no pattern to indicate a relation to probable intensity or duration of exposure to diesel emissions (Edling et al., 1987). Moreover, Rushton et al. (1983) found no excess of deaths from cerebrovascular or ischaemic heart disease among maintenance workers in London bus garages. A 27-year follow-up of 3886 potash miners and millers also showed no excess mortality that could be attributed to the presence of diesel engines in some of the mines that were studied; in only two of eight mines had diesel engines been used (Waxweiler et al., 1973). None of these four analyses of mortality included adjustments for the men's smoking habits. However, the authors noted that the US potash workers whom they had studied included a greater proportion of cigarette smokers than among all US males.

(iii) Effects on reproduction and prenatal toxicity

No data were available to the Working Group.

(iv) Genetic and related effects

The frequency of chromosomal aberrations in cultured lymphocytes from 14 male miners exposed to diesel engine exhaust (five were smokers) was no greater than in 15 male office workers (five smokers; Nordenson et al., 1981). The incidence of chromosomal changes was also investigated in four groups of 12 men: drivers of diesel-engine trucks, drivers of gasoline-engine trucks, automobile inspectors and a reference group, matched with respect to age, smoking habits and length in the jobs. The frequencies of gaps, breaks and sister chromatid exchange in lymphocyte preparations were not significantly different in the four groups (Fredga et al., 1982). [The Working Group noted the small number of subjects in both of these studies.]

Among workers with relatively heavy exposure to diesel engine exhaust — in particular, crews of roll-on roll-off ships and car ferries and bus garage staff (the latter two groups also having exposure to gasoline engine exhausts) — no difference in mutagenicity to S. typhimurium TA98 or E. coli WP2 uvrA was observed between urine collected during exposed periods and that collected during unexposed periods. Similarly, no increase in urinary mutagenicity was found among six volunteers before and after an experimental exposure to diesel engine exhaust gases from an automobile run for 3.7 h at 60 km/h, 2580 revolutions/min (Ulfvarson et al., 1987).

(i) Railroad workers

Kaplan (1959) evaluated 6506 deaths among railroad workers from the medical records of the Baltimore and Ohio Railroad relief department between 1953 and 1958, 818 of which were due to cancer and 154 of which were lung cancer. The cases were categorized into three groups by exposure to diesel exhaust. In comparison with national death rates, none of the groups had an excess risk for lung cancer. [The Working Group noted that, since changeover to diesel engines began in 1935 and was 95% complete by 1959 (Garshick et al., 1988), few if any of the lung cancer deaths could have occurred in workers with more than ten years' exposure to diesel exhaust; in addition, smoking habits were not considered.]

Howe et al. (1983) studied a cohort of 43 826 male pensioners of the Canadian National Railway Company consisting of retired railroad workers who were known to be alive in 1965 plus those who retired between 1965 and 1977. Of the total of 17 838 deaths that occurred in 1965–77, 16 812 (94.4%) were successfully linked to a record in the Canadian mortality data base. The expected number of cancer deaths was estimated from that of the total Canadian population, adjusted for age and calendar period. Available information included birth date, province of residence, date of retirement and occupation at time of retirement. Occupational exposures were classified into three types: ‘diesel fumes’, coal dust and other. The two statistically significant results for the whole cohort were deficits in deaths from all causes (SMR, 95 [95% confidence interval (CI), 93–96]) and from leukaemia (SMR, 80 [95% CI, 65–97]). For exposure to diesel engine exhaust, the risk for cancer of the trachea, bronchus and lung increased with likelihood of exposure: the relative risks were 1.0 for unexposed, 1.2 [1.1–1.3] for ‘possibly exposed’ and 1.4 [1.2–1.5] for ‘probably exposed’ (p for trend < 0.001). The SMR for bladder cancer was 103 [88–119]. Similar results were found for the risk for cancer of the trachea, bronchus and lung from exposure to coal dust. Since there was considerable overlap in exposures to diesel fumes and coal dust, the risk was evaluated by calendar time during which one of these exposures predominated. The risk was largely accounted for by exposure to diesel exhaust. Since exposure to asbestos occurs during locomotive maintenance, workers thought to have had such exposure were removed from the analysis, with little effect on the risk associated with exposure to diesel engine exhaust. Exclusion of workers exposed to welding fumes did not alter the result. The authors noted that the data presented and the risks observed probably represent an underestimate of the true risk, for at least two reasons: exposure misclassification because of the use of job held last and failure to determine the cause of death for 5.6% of cases. [The Working Group noted that no data were available on duration of exposure, usual occupation or smoking habits and recognized the potential for competing biases in the way in which the cohort was composed.]

Garshick et al. (1988) studied a cohort of 55 407 white male railroad workers aged 40–64 in 1959 who had started railroad service ten to 20 years earlier. The cohort was traced from records of the pension scheme for US railway workers through to 1980; it was estimated that less than 2% left the industry during the period covered by the study. Death certificates were available for 88% of the 19 396 deaths, of which 1694 were from lung cancer; decedents for whom a death certificate was not obtained were classified as having died of unknown causes. Records of railroad jobs from 1959 through to death, retirement or 1980 were also available from the records of the pension scheme. Jobs were divided into regular exposure to diesel exhausts (train crews, workers in diesel repair shops) and no exposure (clerks, ticket and station agents, and signal maintenance workers). Job categories with recognized asbestos exposure, such as car repair and construction trades, were excluded from those selected for study. Information was available on duration of exposure. There was a significant excess risk for lung cancer in the groups exposed to diesel engine exhaust; this risk was highest in those who had the longest exposure: aged 40–44 (relative risk, 1.5; 95% CI, 1.1–1.9) and 45–49 (1.3; 1.0–1.7) and exposed to diesel exhaust in 1959. The groups aged 50–54 and 55–59 in 1959 also had excess risks, of 1.1 and 1.2, respectively, although these were not statistically significant. When workers with further potential asbestos exposure (shop workers) were excluded, similarly elevated lung cancer rates were observed. Although smoking habits were not considered directly, the authors pointed out that there was no difference in smoking habits by job title in comparison studies of current workers or in a case-control study in which smoking was assessed. [The Working Group noted that exclusion of shop workers would also have excluded men exposed to welding fumes.]

As part of this study, exposure was assessed on the basis of several hundred time-weighted samples of respirable dust taken in the early 1980s both at stationary sites in parts of four existing, smaller railroad yards and with personal samplers carried by railroad workers in different job categories (Woskie et al., 1988a). Samples were taken from workers in 39/155 Interstate Commerce Commission job codes, and the results were used to classify the jobs; these 39 categories were subsequently combined into 13 job groups, which could be further combined into five: clerks, signal maintenance, engineers/firers, brakers/conductors and shop workers. The nicotine content was used to adjust the extractable respirable particulate content of each sample to account for the portion contributed by cigarette smoking. Mean exposure levels by national career groups in the five major categories of exposure suggested a five-fold range of exposure to respirable particles between clerks and shop workers (Woskie et al. 1988b). These values confirmed the a-priori assignment of the categories of diesel exposure used in the cohort study (Garshick et al., 1988) and the assignment to appropriate exposure categories for the case-control study (Garshick et al., 1987; see p. 140).

(ii) Bus company employees

Raffle (1957) determined deaths, retirements and transfers due to lung cancer in London Transport employees aged 45–64 years in jobs with presumably different exposures to exhaust fumes in 1950–54 and compared the figures with those for lung cancer mortality for men in England and Wales or in Greater London. No relationship between presumed exposure and lung cancer incidence was noted. In a subgroup of bus and trolley bus engineering staff aged 55–64, 30 deaths from lung cancer occurred while 21.2 were expected (observed:expected, 1.4) on the basis of the experience of other London Transport employees. [The Working Group noted that no information on smoking habits was available, and that all the deaths occurred in men over 55 years of age.] Waller (1981) compared lung cancer deaths and retirements or transfers to alternative jobs due to lung cancer in men aged 45–64 employed within five job categories of London Transport (bus drivers, bus conductors, engineers (garages), engineers (central works) and motor men and guards) to lung cancer mortality (age- and calendar time-adjusted) for men in Greater London. The study covered 25 years, ending in 1974, thus including some of the data described by Raffle (1957). A total of 667 cases of lung cancer were observed; although the risk was not elevated for any of the five job categories, the highest SMR occurred in the group that was presumably most heavily exposed to diesel exhaust (bus garage workers). [The Working Group noted that no data on smoking habits were available, and neither duration nor latency was examined.]

Rushton et al. (1983) examined a cohort of 8684 men employed as maintenance workers in 71 bus garages in London for at least one year in 1967–75. Follow-up until 31 December 1975 was completed for 8490 (97.8%) workers, and cause of death was known for 701 of 705 who had died. The SMRs were 84 [95% CI, 78–91] for all causes and 95 [83–109] for all neoplasms, 101 [82–122] for lung and pleural cancer, 151 [60–307] for leukaemia, 121 [49–250] for central nervous system tumours and 139 [72–244] for bladder cancer. None of the rates for cancer at individual sites was statistically significantly increased. The authors noted the short follow-up period.

Edling et al. (1987) studied 694 men, five of whom (0.7%) were lost to follow-up, who had been employed as clerks, bus drivers or bus garage workers in five bus companies in south-eastern Sweden at any time between 1950 and 1959, and followed for 1951–83. The SMRs, based on age-, sex- and calendar time-adjusted national rates, were 80 (195 deaths observed; 95% CI, 70–90) for deaths from all causes and 70 (50–90) for deaths from malignancy. Dividing the data by exposure category, exposure time or latency did not appreciably change the risk ratios. The small sample size did not allow detailed examination of cancers at specific sites, although six lung cancer cases were observed compared to nine expected. [The Working Group noted that smoking habits were not addressed.]

(iii) Professional drivers and some other groups exposed to vehicle exhausts

Ahlberg et al. (1981) identified a cohort of Swedish drivers said by the authors to be exposed to diesel exhaust (1865 or 1856 [sic] fuel oil tanker drivers and 34 027 other truck drivers) from the national census of 1960. In this cohort, 1143 cancers were registered within the Swedish Cancer Registry in 1961–73. The reference population consisted of 686 708 blue-collar workers from the 1960 census who were thought to have had no exposure to petroleum products or chemicals. The data were adjusted for age and residence. The relative risk for lung cancer was elevated in the whole cohort. (1.3; 95% CI, 1.1–1.6) and in Stockholm truck drivers in particular (1.6; 1.2–2.3). From a questionnaire study of 470 professional drivers in Stockholm, it was noted that 78% of fuel truck drivers and 31% of other truck drivers smoked. The authors cited an unpublished study indicating that the comparable smoking rate in Stockholm was 40% and concluded that the results could not be explained by smoking.

Wong et al. (1985) studied a cohort of 34 156 male members of a heavy construction equipment operators' union in the USA with potential exposure to diesel exhaust. Cohort members had to have been a union member for at least one year between 1 January 1964 and 31 December 1978, by which time 3345 had died and 1765 (5.2%) could not be traced. Death certificates were obtained for all but 102 (3.1%) decedents. No information was available for jobs held before 1967 and limited information was available on jobs held between 1967 and 1978. The SMRs, based on national figures, adjusted for age, sex, race and calendar time, were 81 (95% CI, 79–84) for all causes, 93 (87–99.6) for all cancers, 99 (88–110) for lung cancer (ICD7 162–163) and 118 (78–172) for bladder cancer. The data were also analysed by duration of union membership, latent period, retirement status, job category and exposure status. Significant upward trends in risk were detected for lung cancer with duration of union membership, used as a surrogate for duration of potential exposure to diesel exhaust, with SMRs for lung cancer of 45 [22–83], 75 [49–111], 108 [81–141], 102 [78–132]and 107 [91–125] for workers with <5, 5–9, 10–14, 15–19 and ⩾ 20 years of union membership, respectively. A significant upward trend was also noted for lung cancer with latent period. Mortality from cancers of the digestive system (SMR, 142; 116–173) and respiratory system (SMR, 162; 138–190) and from lymphosarcoma and reticulosarcoma (SMR, 231; 111–425) was elevated in retirees. Exclusion of early retirees did not remove the risks for respiratory cancer or lymphatic cancer. In general, groups with jobs with presumed high exposure to diesel fumes did not show the excesses reported above. A random sample of union members was surveyed to determine smoking habits, and no significant difference between members and the general population was revealed.

In a review, Steenland (1986) presented data on a preliminary study of the mortality experience of about 10 000 teamsters (truck drivers, dock workers, mechanics and jobs outside the trucking industry) who had died in 1982–83 and had worked for at least ten years in a teamster job. Using occupational data on death certificates, proportionate mortality ratios were calculated for lung cancer for 255 mechanics (226; 95% CI, 162–309), 5834 truck drivers(154; 144–166), 490 dock workers(132; 99–175) and 1064 others(116; 95–142). [The Working Group noted that this was an interim report and that judgement should be reserved until the final results are available.]

Gustafsson et al. (1986) studied 6071 Swedish ‘dockers’ assumed by the authors to have been exposed to diesel exhaust and first employed before 1974 for at least six months. The group had been followed for death from 1 January 1961 or from the date of first employment (if this date occurred later) through to 1 January 1981. Age-, calendar time-and region-specific rates were used to generate expected numbers of deaths. The SMRs were 89 (95% CI, 84–94) for all causes, 103 for all cancers, 132for lung cancer (105–166) and 110 (85–142) for urogenital tract cancer. Cancer morbidity was determined among 6063 workers who had been alive and without cancer on 1 January 1961 and were followed through to 1 January 1980; a standard morbidity ratio of 110 (101–120; 452 cases) was seen for cancers at all sites and of 168 (136–207; 86 cases) for lung cancer. [The Working Group noted that there was no consideration of duration, intensity or latency of exposure or of smoking habits in this study.]

Stern et al. (1981) examined mortality patterns among 1558 white male vehicle examiners who had been employed in New Jersey, USA, for at least six months between 1944 and 1973. The vital status of all but eight (0.5%) of these was ascertained as of 31 August 1973; these eight were assumed to be alive. Approximately 63% of the cohort members had begun employment prior to 1957. A modified life-table analysis was used to generate the expected number of cause-specific deaths on the basis of national rates, adjusting for age and calendar time. There were 52 deaths from cancer (47.8 expected [SMR, 109; 95% CI, 81–143]). The SMRs for malignant disease increased significantly with latency: 0–9 years, 69 [25–151]; 10–19 years, 98 [56–159]; 20–29 years, 107 [62–171]; >30 years, 189 [101–323]. Cancer at no specific organ site accounted for this excess. The exposure of interest was carbon monoxide, but the authors speculated that other components of automobile exhaust might have been responsible. No information on smoking habits was available for deceased workers, but COHb levels in currently nonsmoking workers increased during the work shift, indicating exposure to exhaust.

In a cohort study of white men enlisted in the US Navy (Garland et al., 1988), 143 cases of testicular cancer were identified in the period 1974–79; age-specific incidence rates were similar to those for the US population, derived from the US National Cancer Institute Surveillance, Epidemiology and End Results (SEER) programme for 1973–77. Of 110 occupational groups in the Navy, three involving maintenance of gasoline and diesel engines and daily exposure to their exhaust emissions (aviation support equipment technicians, enginemen and construction mechanics) had significantly high standardized incidence ratios for testicular cancer: 3.4 (95% CI, 1.9–5.6) in comparison to SEER rates, and 3.8 (2.1–6.3) in comparison to men in the US Navy as a whole, based on 15 cases. The authors noted that this was a hypothesis-generating study and that the men also had potential daily exposure to solvents and other chemicals.

(iv) Miners

Although diesel engines have been used in many mines for a number of years, the Working Group decided not to consider all groups of miners because they may be exposed concurrently to other potential lung carcinogens such as radon decay products, heavy metals and silica, and there was no way that the possible confounding effects of such factors could be determined from the data available in published reports.

Waxweiler et al. (1973) studied potash miners and millers, who are exposed to no known carcinogens in the ore, who had been employed for at least one year between January 1940 and July 1967 by eight companies. The vital status of the cohort was identified to July 1967. Of a total of 3886 men, 31 could not be traced and were assumed to be alive. Causes of death were compared with those of the general US population, standardized for age, race, sex and calendar time. Of the cohort, 2743 men had worked at least one year underground and less than one year on the surface and 1143 men had worked at least one year on the surface and less than one year underground. In only two of the eight mines were diesel engines used; one mine changed to diesel in 1949 and the other in 1957. Death certificates were available for 433 of the 438 workers who had died. The effect of smoking was taken into account. No excess mortality from lung cancer was seen in either surface or underground miners. Mortality rates did not differ between the mines with diesel vehicles and those without. The authors noted the short follow-up, the small expected numbers of deaths and the broad classification of causes of death.

(i) Lung cancer

Williams et al. (1977) examined cancer incidence and its relationship to occupation and industry in a study based on the US Third National Cancer Survey. In this study, detailed personal interviews were sought for 13 179 cancer patients (a random 10% sample of all incident invasive tumours occurring in three years in eight areas in the USA) and obtained for 7518 (57%). The numbers of cases of cancer at various anatomical sites were compared with that of cases at all other sites combined. The interview included occupational history (main employment and recent employment), other demographic data and information on smoking and drinking habits; the analysis also controlled for age, sex, race and geographical location. A statistically nonsignificant lung cancer excess (odds ratio, 1.5; [CI could not be calculated]) was observed for truck drivers, which could not be accounted for by smoking. Intensity, duration of exposure and latency were not evaluated. [The Working Group noted the potential for bias due to the relatively low level of compliance with the questionnaire.]

In a population-based case-control study, Coggon et al. (1984) used the data on occupation on the death certificates of all men under the age of 40 years in England and Wales who had died of tracheobronchial carcinoma during the period 1975–79; 598 cases were detected, 582 of which were matched with two and the rest with one control who had died from any other cause, for sex, year of death, local authority district of residence and year of birth. Occupations were coded using the Office of Population Census and Surveys 1970 classification of occupations, and a job-exposure matrix was constructed by an occupational hygienist, in which the occupations were grouped according to likely exposure to each of nine known or putative carcinogens. All occupations entailing exposure to diesel fumes were associated with an elevated odds ratio for bronchial carcinoma (1.3; 95% CI, 1.0–1.6); however, for occupations with presumed high exposure, the odds ratio was 1.1 (0.7–1.8). [The Working Group noted the limited information on occupation from death certificates, the young age of the subjects and the consequent short times of exposure and latency, and the lack of information on smoking habits and on the possible confounding effects of other carcinogenic exposures.]

In a hospital-based case-control study (Hall & Wynder, 1984) in 18 hospitals in six US cities, 502 men with histologically confirmed primary lung cancer (20–80 years old) and 502 control patients, matched for age, race and hospital were identified. Patients were interviewed between December 1980 and November 1982. Half of the controls had cancer; patients with tobacco-related diseases were excluded. The questionnaire included items on smoking habits, demographic variables and usual occupation. Occupations were grouped either dichotomously as exposed to diesel exhaust (warehousemen, bus drivers, truck drivers, railroad workers and heavy equipment repairmen and operators) or nonexposed, or, in a separate evaluation, in three presumed categories of frequency of exposure in the job (high, moderate, little). Using the dichotomous division, the exposed group had a significantly elevated odds ratio (2.0; 95% CI, 1.2–3.2), which, however, decreased to 1.4 (0.8–2.4; not significant) when adjusted for smoking. The crude odds ratios were 1.7 (0.6–4.6) for a high probability of exposure to diesel exhaust and 0.7 (0.4–1.3) for a moderate probability of exposure. [The Working Group questioned the possible consequences on risk estimates of excluding patients with tobacco-related diseases from the control group.]

In a hypothesis-generating case-control study, Buiatti et al. (1985) investigated the occupational histories of histologically confirmed cases of primary lung cancer among residents of metropolitan Florence, Italy, diagnosed during 1981–83 in the regional general hospital and referral centre for lung cancers in the Province of Florence. For the 376 cases (340 men, 36 women), 892 controls (817 men, 75 women), matched by sex, age, date of admission and smoking status in seven categories, were selected from the medical service of the same hospital, excluding patients with lung cancer, attempted suicides and patients not resident in metropolitan Florence. Each case and control completed a structured questionnaire on demographic variables and on all jobs held for more than one year. The jobs were classified into 21 major classes and 251 subclasses, using the International Labour Office classification. Odds ratios for industries and occupations (ever versus never worked) were calculated using logistic regression, in which age and smoking status were included. Taxi drivers had an elevated relative risk for lung cancer after adjusting for tobacco smoking (1.8; 95% CI, 1.0–3.4). [The Working Group noted that multiple comparisons were made, increasing the probability that statistically significant results would be found.]

In a case-control study in northern Sweden, Damber and Larsson (1987) analysed the association between lung cancer and occupation. The cases were 604 male lung cancers reported to the Swedish Cancer Registry during 1972–77 and who had died before May 1979. For each case, a control was drawn from the National Registry for Causes of Deaths, and was matched for sex, year of death, age and municipality; cases of lung cancer and attempted suicide were excluded as controls. In addition, for each case, one living control (less than 80 years old) was drawn from the National Population Registry, matched for sex, year of birth and municipality. Information on residence, occupation, employment and smoking habits was collected by a questionnaire mailed to surviving relatives and to living controls; the response rates were 98% for cases and 96% and 97% for dead and living controls, respectively. Information was requested on all jobs held for at least one year and on lifetime smoking history. A linear logistic regression model, using three discrete levels of employment (<1 year, 1–20 years, and >20 years) and four levels of lifetime tobacco consumption, was used to calculate odds ratios. For professional drivers with more than 20 years' employment, the unmatched odds ratio was 1.5 (95% CI, 0.9–2.6) in comparison with dead controls; this was reduced to 1.2 (0.6–2.2) after adjustment for smoking. The figures obtained in comparison with living controls were 1.7 (0.9–3.2) and 1.1 (0.6–2.2), respectively.

Garshick et al. (1987) performed a case-control study on lung cancer deaths among employed and retired US male railroad workers with ten or more years of service, who had been born on 1 January 1900 or after and who had died between 1 March 1981 and February 1982. Cases of primary lung cancer (1256) were matched to two controls by age and date of death. Workers who had died from cancer, suicide, accident or unknown causes were not included among controls. Potential exposure to diesel exhaust was assigned on the basis of an industrial hygiene evaluation of the >150 railroad jobs and areas described by the US Interstate Commerce Commission. Job codes for each worker were available from the US Railroad Retirement Board starting in 1959 and ending with death or retirement. For workers who had retired between 1955 and 1959, the last railroad job held was available. Asbestos exposure prior to 1959 was categorized by job held in 1959 (end of steam locomotive era) or by the last job before retirement, if this was before 1959. Smoking history was obtained by questionnaire from the next-of-kin. Using multiple conditional logistic regression analysis to adjust for smoking and asbestos exposure, workers 64 years of age or younger at time of death who had worked in a diesel exhaust-exposed job for 20 years had a significantly elevated odds ratio for lung cancer (1.4; 95% CI, LI–1.9). No such effect was observed among older workers (0.91 ; 0.71–1.2), many of whom had retired shortly after the transition to diesel-powered locomotives and were therefore not exposed.

In a population-based case-control study (Lerchen et al., 1987), all white and Hispanic white residents of New Mexico, USA, aged 25–84 years, with primary lung cancer, excluding bronchioalveolar carcinoma, diagnosed between 1 January 1980 and 31 December 1982, were identified from the New Mexico Tumor Registry. The cases (333 men and 173 women) were frequency matched with controls selected randomly from the telephone directory or, for persons 65 years or older, from the roster of participants in a health insurance scheme, for sex, ethnic group and ten-year age band at a ratio of approximately 1.5 controls per case (449 men and 272 women). Detailed occupational and smoking histories were obtained by personal interview, with response rates of 89% for cases and 83% for controls. Next-of-kin provided interviews for 50% of the male and 43% of the female cases and for 2% of the controls; the authors recognized the possible bias introduced by this practice. The odds ratio for exposure to diesel exhaust fumes, adjusted for age, ethnic group and smoking, was 0.6 (95% CI, 0.2–1.6). [The Working Group noted the possible bias in choosing controls from the telephone directory when cases are not required to have a telephone or to be listed.]

In a case-control study of lung cancer in France (Benhamou et al., 1988), 1625 histologically confirmed cases and 3091 controls, matched for sex, age at diagnosis, hospital admission and interviewer, completed a questionnaire on residence, education, occupation, and smoking and drinking habits. All occupations held for more than one year were recorded and coded without knowledge of the case status of the patient, using the International Standard Classification of Occupations and according to chemical or physical exposures. The analysis was limited to men (1260 cases and 2084 controls); adjustment was made for age at starting smoking, amount smoked and duration of smoking. Several occupations were associated with increased odds ratios for lung cancer, including miners and quarry men (2.1; 95% CI, 1.1–4.3) and transport equipment operators (1.4; 1.1 – 1.8); the subcategory of motor vehicle drivers also had an increased risk (1.4; 1.1–1.9).

(ii) Bladder cancer

In a population-based case-control study in Canada (Howe et al., 1980), all patients with bladder cancer newly diagnosed in three Canadian provinces between April 1974 and June 1976 were identified; 77% of the patients were interviewed, and for each patient one neighbourhood control, individually matched for age and sex, was interviewed. In the analysis, 632 case-control pairs (480 male and 152 female) were included. Lifetime smoking and employment histories were obtained, and exposure to dusts and fumes was elucidated. Elevated odds ratios were observed for railroad workers [not further defined] (9.0; 95% CI, 1.2–394.5; nine exposed cases) and for exposure to diesel and traffic exhaust (2.8; 0.8–11.8; 11 exposed cases).

In a death certificate-based case-control study (Coggon et al., 1984; for details, see description on p. 139), the occupations of 291 bladder cancer cases and 578 hospital controls were compared. The odds ratio for all diesel fume-exposed occupations was 1.0 (95% CI, 0.7–1.3) and that for occupations with high exposure was 1.7 (0.9–3.3). [The Working Group had the same reservations about this study as expressed on p. 139.]

In a population-based case-control study, the relationship between truck driving and bladder cancer was investigated (Hoar & Hoover, 1985). Cases consisted of all white residents of New Hampshire and Vermont, USA, who had died from bladder cancer in 1975–79. One control per case was selected randomly from all other deaths among residents, excluding suicides, and matched for state, sex, age, race and year of death. A second control per case was selected with the additional matching criterion of county of residence. There were 230 and 210 eligible cases in the two states, respectively; the rate of response to interview was 87% for New Hampshire and 58% for Vermont, and the non-respondents were similar to the respondents with respect to case-control status, sex, age and county of residence. The odds ratio for ever having been a truck driver was 1.5 (95% CI, 0.9–2.6), and there was a significant trend between bladder cancer risk and number of years of truck driving: odds ratios, 1.4 (0.6–3.3), 2.9 (1.2–6.7) and 1.8 (0.8–4.1) for those employed as truck drivers for 1–4, 5–9 and >10 years, respectively. Additional adjustment for age, county, coffee drinking or cigarette smoking (six categories) did not alter these crude odds ratios. [The Working Group noted the nonlinearity of the trend.]

In a hospital-based case-control study in Turin, Italy (Vineis & Magnani, 1985), 512 male cases and 596 male controls randomly selected from among other patients in the main hospital of the city of Turin between 1978 and 1983 were interviewed for lifetime occupational and smoking histories. Occupations were coded using the International Labour Office classification, and associations between specific chemicals and bladder cancer were studied using a job exposure matrix. Adjusting for age and smoking, the odds ratio for bladder cancer for truck drivers was 1.2 (95% CI, 0.6–2.5).

In a hospital-based case-control study, Wynder et al. (1985) examined the occupational histories and life style factors (smoking, alcohol and coffee consumption, demographic factors) of 194 male cases of histologically confirmed bladder cancer, 20–80 years of age, diagnosed during two-and-a-half years (January 1981–May 1983) in 18 hospitals in six US cities, and of 582 controls, matched by age, race, year of interview and hospital of admission, hospitalized during the same period for diseases not related to tobacco use. The participation rate among eligible subjects was 75% among cases and 72% among controls. ‘Usual’ occupation was coded according to an abbreviated list of the US Bureau of Census codes. No significant association was detected between bladder cancer and occupations presumed to involve exposure to diesel exhaust: warehousemen and materials handlers, bus and truck drivers, railroad workers, heavy equipment operators and mechanics (odds ratio, 0.87; 95% CI, 0.47–1.6). [The Working Group questioned the possible consequences on risk estimates of excluding patients with tobacco-related diseases from the control group.]

Data from all ten areas of the US National Bladder Cancer Study were used to evaluate the association of motor exhausts with bladder cancer (Silverman et al., 1986). The study group comprised 1909 white male cases with histologically confirmed bladder carcinoma or papilloma not specified as benign and 3569 frequency-matched controls. Significantly elevated age- and smoking-adjusted odds ratios for bladder cancer were observed for truck drivers or delivery men, and for taxi drivers or chauffeurs: 1.5 (95% CI, 1.1–2.0) and 6.3 (1.6–29.3) for ‘usual’ occupation, 1.3 (1.1–1.4) and 1.6 (1.2–2.2) for ‘ever’ occupation. For bus drivers, the odds ratios did not reach significance (1.3, 0.9–1.9 and 1.5, 0.6–3.9 for ‘ever’ and ‘usual’, respectively). When allowance was made for a 50-year latency, a significant trend with increasing duration of employment as a truck driver was observed: 1.2, 1.4, 2.1 and 2.2 for a duration of employment of <5, 5–9, 10–24 and >25 years, respectively (p <0.0001). Information on subsets of this cohort has been published elsewhere (Silverman et al., 1983; Schoenberg et al., 1984; Smith et al., 1985). In the Detroit subset (Silverman et al., 1983), the adjusted odds ratio for bladder cancer for truck drivers who had never driven a vehicle with a diesel engine was 1.4 (0.7–2.9) and that for men who had ever driven a vehicle with a diesel engine was 11.9 (2.3–61.1).

Occupational risk factors were investigated as part of a population-based case-control study in Copenhagen, Denmark (Jensen et al., 1987). Between May 1979 and April 1981, a total of 412 live patients with bladder cancer (invasive tumours and papillomas) were reported in the study, 389 of whom were interviewed. Live controls were selected at random from the municipalities where the cases lived, and the sample was stratified to match the cases with regard to sex and age in five-year groups. Among the 1052 controls approached, the overall participation rate was 75%. Cases and controls were interviewed for information on occupational history coded according to the Danish version of the International Standard Industrial Classification. Cigarette smoking was adjusted for in the analysis by using two dichotomous variables (ever/never smoked, current/noncurrent smoker) and a continuous variable (logarithm of pack-years smoked). The adjusted odds ratio for bladder cancer was elevated in land transport workers (1.6; 95% CI, 1.1–2.3). The adjusted odds ratios for bladder cancer for bus, taxi and truck drivers were 0.7 (0.4–1.5), 1.6 (0.8–3.4), 3.5 (1.1–11.6) and 2.4 (0.9–6,6) for durations of employment of 1–9, 10–19, 20–29 and >30 years, respectively, representing a significant trend with duration of employment. The trend was not significant for land transport workers.

In a hospital-based case-control study in Argentina (Iscovich et al., 1987), 120 patients with histologically confirmed bladder carcinoma admitted to ten general hospitals in Greater La Plata between March 1983 and December 1985 were identified. The 117 patients who could be interviewed represented approximately 60% of all incident cases. For each case, a hospital control from the same establishment was selected (patients with diseases associated with tobacco smoking constituted 12% of the control group); a neighbourhood control, matched for age and sex, was also selected. Information on smoking and past and present occupations was collected by questionnaire. An exposure index based on a job-exposure matrix was generated. The adjusted odds ratio for truck and railway drivers was 4.3 [95% CI, 2.1–29.6].

Covering the period 1960–82, Steenland et al. (1987) identified 731 male bladder cancer (ICD-9 188) deaths in the Hamilton County, Ohio, region, where there is a known high bladder cancer rate. Six controls were matched to each case on sex and residence in the county at the time of death, year of death, age of death and race. Death certificates and city directories for all residents over 18 were used to identify job history. The first two controls that were listed in the directory within at least five years of the first listing of the cases were selected. Of the 648 cases (89%) listed in the directories, all but 21 had two controls; the remaining 21 had one control. A comparable analysis of all 731 cases and two controls per case was carried out using usual lifetime occupation from the death certificate. A significant increase in the frequency of bladder cancer was found for men with more than 20 years' duration of employment, identified through the city directories as truck drivers (odds ratio, 12.0 [95% CI, 2.3–62.9]; six cases, one control) and railroad workers (odds ratio, 2.2 [95% CI, 1.2–4.0]). Notably, those workers identified as ‘drivers not otherwise specified’ for ⩽20 years had an odds ratio of 0.15 [95% CI, 0–0.8]. In contrast, on the basis of job ever held identified from either the death certificate or the city directory (without taking duration into account), none of the above findings was significant. [The Working Group noted that this study involved application of a new methodology for exposure ascertainment, which requires further validation.]

In a case-control study of bladder cancer incidence in Edmonton, Calgary and Toronto and Kingston, Canada (Risch et al., 1988), 826 cases of histologically verified bladder cancer were compared with 792 population-based controls matched for age, sex and area of residence. Cases were aged 35–79 and had been ascertained between 1979 and 1982. Information was collected by questionnaire, administered by personal interview, covering family, medical, occupational, residential, smoking and dietary histories. Analysis of the occupational data included adjustment for lifetime smoking habits. Among other findings related to occupation and industry was that the 309 men who had had jobs with exposure to engine exhausts had an odds ratio of 1.5 (95% CI, 1.2–2.0) for ‘ever’ exposure and an odds ratio of 1.7 (1.2–2.3) for exposure during the period eight to 28 years prior to diagnosis. The authors also calculated that there was a significant increase in trend with duration of exposure for each ten years (1.2; 1.1–1.4). This relationship was not seen for women, but only 19 had been exposed. The relationship was also not seen when an analysis was undertaken by exposure to 18 categories of substances, including engine exhaust. [The Working Group found it difficult to interpret the differences in risk seen when exposure was defined in various ways.]

(iii) Other and multiple sites

In a hypothesis-generating, hospital-based case-control study in Sweden, Flodin et al. (1987) analysed the association between occupation and multiple myeloma. The cases were in persons diagnosed between 1973 and 1983 and still alive during 1981–83. From comparisons with cancer registry data, it was concluded that the cases represented one-third of all cases diagnosed in the area. Controls were drawn randomly from population registers. There were 131 cases and 431 controls for analysis. Information on occupational history, X-ray treatment and smoking habits were obtained by a mailed questionnaire. The crude odds ratio for occupational exposure to engine exhaust was 2.3 (95% CI, 1.4–3.7); this association remained significant after adjusting for confounding variables. In a study using the same set of controls (431) and source of cases, Flodin et al. (1988) investigated the association with occupational exposures for 111 cases of chronic lymphatic [lymphocytic] leukaemia. The crude odds ratio for occupational exposure to engine exhausts was 2.5 (95% CI, 1.5–4.0); the association remained significant after adjustment for confounding variables. [The Working Group noted that the study population and control of confounding were not clearly described, and that exposure to engine exhausts was self-reported and not further defined by the authors.]

In a large, hypothesis-generating, population-based case-control study in Canada (Siemiatycki et al., 1988), the associations between ten types of engine exhaust and combustion products and cancers at 12 different sites were evaluated. The 3726 cancer patients diagnosed in any of the 19 participating hospitals in Montreal were interviewed (rate of response, 82%). The patients were all men aged 35–70 years. For each cancer site, patients with cancers at other sites comprised the control group. The interview elicited a detailed job history, and a team of chemists and industrial hygienists translated each job into a list of potential exposures (Gérin et al., 1985). The probability of exposure ( ‘possible’, ‘probable’, ‘definite’), the frequency of exposure (<5, 5–30, >30% working time) and the level of exposure (low, medium, high) were estimated. Separate analyses were performed for oat-cell, squamous-cell, adenocarcinoma and other carcinomas of the lungs. After stratifying for age, socioeconomic status, ethnic group, cigarette smoking and blue-/white-collar job history, an elevated odds ratio was observed for squamous-cell cancer of the lung and exposure to gasoline engine exhaust (OR, 1.2; 90% CI, 1.0–1.4). In a detailed analysis in which all covariables that changed the estimate of the disease-exposure odds ratio by more than 10% were included as confounders, further associations were revealed: long-term high-level exposure to gasoline engine exhaust (1.4; 1.1–1.8) and short-term high-level exposure to diesel engine exhaust (1.5; 0.9–2.7) were associated with squamous-cell cancer of the lung. The odds ratio for squamous-cell cancer of the lung (1.5; 0.9–2.5) was also elevated for bus, truck and taxi drivers (classified as exposed to gasoline engine exhaust) and for mining and quarrying (classified as exposed to diesel engine exhaust; 2.8, 1.4–5.8), but analyses by duration and intensity of exposure did not support a causal association. Marginally elevated odds ratios were also seen for colon cancer and exposure to diesel engine exhaust (1.3; 1.1–1.6); for cancer of the rectum (1.6; 1.1–2.3) and kidney (1.4; 1.0–2.0) with long-term high-level exposure to gasoline engine exhaust; for colon cancer (1.7; 1.2–2.5) with long-term high-level exposure to diesel engine exhaust; and for rectal cancer (1.5; 1.0–2.2) in bus, truck and taxi drivers. [The Working Group noted that 90% CI were used and that, at the 95% level, most of the intervals would have included unity.]