1.3.1. Environmental occurrence and exposure

Formaldehyde is found as a natural product in most living systems and in the environment. It occurs naturally in fruits and some foods, and it is formed endogenously in mammals, including humans, as a consequence of oxidative metabolism. In addition to these natural sources, common non-occupational sources of exposure to formaldehyde include combustion processes, e.g. through emissions from motor vehicles, power plants, incinerators, refineries, wood stoves, and kerosene heaters. Formaldehyde may be released from particle boards and similar building materials, carpets, paints and varnishes, during cooking of some foods, and during its use as a disinfectant. It is also present in tobacco smoke. An indirect source of exposure to formaldehyde is its formation via photochemical oxidation of hydrocarbons, such as methane, and other precursors emitted from combustion processes (NTP, 2005; IARC, 2006). Formaldehyde has a short half-life in the environment, because it is removed from the air by photochemical processes and by precipitation and biodegradation (NTP, 2005).

Concentrations of formaldehyde in outdoor air are generally below 0.001 mg/m³ in remote areas and below 0.02 mg/m³ in urban settings. The levels of formaldehyde in indoor air of houses are typically 0.02–0.06 mg/m³; indoor combustion sources can significantly increase these levels. Cigarettes may contribute as much as 10–25% of the indoor exposure. Average concentrations of 0.5 mg/m³ or more have been measured in ‘mobile homes’, but these have declined since the late 1980s as a result of standards that require that building materials – e.g. particle boards – emit lower concentrations of formaldehyde. A recent study of emissions from mosquito coils found the average concentration of formaldehyde exceeded 100 μg/m³ (IARC, 2006, 2010; Lee & Wang, 2006). Data on formaldehyde concentrations in outdoor air in residential and public settings, and information on exposure to formaldehyde associated with household use of solid fuels and high-temperature frying, have been reviewed in IARC Monograph Volumes 88 and 95 (IARC, 2006, 2010).

Automobile exhaust is a major source of formaldehyde in ambient air. Recent reports suggest that formaldehyde emissions may be higher from vehicles powered by compressed natural gas compared with those running on ethanol or gasohol (Corrêa & Arbilla, 2005), and that these emissions may be decreased by substitution of an ethanol-biodiesel-diesel blend for diesel fuel (Shi et al., 2006). In addition, formaldehyde can be absorbed through the skin from cosmetics or via contact with other consumer products containing formaldehyde, such as unwashed permanent-press fabrics treated with formaldehyde-releasing resins (NTP, 2005).

1.3.2. Occupational exposure

Occupational exposure to formaldehyde occurs in a wide variety of occupations and industries. CAREX (CARcinogen EXposure) is an international information system on occupational exposure to known and suspected carcinogens based on data collected in the European Union (EU) from 1990 to 1993. The CAREX database provides selected exposure data and documented estimates of the number of exposed workers by country, carcinogen, and industry (Kauppinen et al., 2000). Table 1.1 presents the results for formaldehyde in the EU by industry (CAREX, 1999).

Table 1.1

Estimated numbers of workers exposed to formaldehyde above background levels in the European Union.

In IARC Monograph Volume 88 (IARC, 2006) data were reviewed on occupational exposure to formaldehyde by type of industry. The highest continuous exposures (2–5 ppm; 2.5–6.1 mg/m³) were measured in the past during varnishing of furniture and wooden floors, in the finishing of textiles, in the garment industry, in the treatment of fur, and in certain jobs within manufactured board mills and foundries. Short-term exposures to high levels (3 ppm and higher; ≥ 3.7 mg/m³) have been reported for embalmers, pathologists, and paper workers. Lower concentrations have usually been encountered during the manufacture of man-made vitreous fibres, abrasives and rubber, and in formaldehyde-production industries. A very wide range of exposure levels has been observed in the production of resins and plastic products. The development of resins that release less formaldehyde, and improved ventilation have resulted in lower exposure levels in many industrial settings in recent decades (IARC, 2006).

2.1.1. Cohort studies

In the most recent follow-up of the largest cohort study from the USA of industrial workers exposed to formaldehyde, a statistically significant excess of deaths from nasopharyngeal cancer was observed in comparison with the US national population, with statistically significant exposure–response relationships for peak exposure and cumulative exposure (Hauptmann et al., 2004; see Table 2.1 available at http://monographs.iarc.fr/ENG/Monographs/vol100F/100F-24-Table2.1.pdf). Based on eight cases, a significant excess mortality from nasopharyngeal cancer was observed among formaldehyde-exposed workers (SMR, 2.10; 95%CI: 1.05–4.21). A highly statistically significant (P_trend < 0.001) exposure–response relationship was observed between peak-exposure to formaldehyde and risk for nasopharyngeal cancer in a Poisson regression-analysis. All exposed cases were in the highest category of peak-exposure, and the relative risk was 1.83. This analysis excluded one case which, according to cancer registry data, had been misclassified as nasopharyngeal cancer. Weaker exposure–response relationships were observed between nasopharyngeal cancer and average or cumulative exposure, and duration of exposure (P_trend = 0.07, 0.03 and 0.15, respectively).

In the two other large cohort studies of industrial workers, cases of nasopharyngeal cancer were fewer than expected, but the power of these studies to detect an effect on nasopharyngeal cancer was low and the deficits were small. In the first study, of British chemical workers, one death was observed when 2.0 were expected (Coggon et al., 2003); in the second study, no deaths were observed among US garment-manufacturers, where 0.96 were expected (Pinkerton et al., 2004).

An excess of deaths from nasopharyngeal cancer was observed in a proportionate mortality analysis of the largest US cohort of embalmers (Hayes et al., 1990) and in a Danish study of proportionate cancer incidence among workers at companies that used or manufactured formaldehyde (Hansen & Olsen, 1995; see Table 2.2 available at http://monographs.iarc.fr/ENG/Monographs/vol100F/100F-24-Table2.2.pdf).

Marsh et al. (1996) conducted a cohort study in one of the plants considered in the NCI study (where five of the nine cases of nasopharyngeal cancer occurred). The cohort included earlier year of entry and was enumerated independently. Significantly increased mortality due to nasopharyngeal cancer was observed among formaldehyde-exposed workers compared with US and regional populations (Connecticut State and local county). In a recent follow-up through 2003, Marsh et al. (2007a) showed elevated SMRs when both national and local county rates were used. In addition, when conducting a case–control study nested within the cohort and including seven deaths from nasopharyngeal cancer, the authors obtained information on employment outside the formaldehyde industry and showed that five of these workers had been employed as a silversmith. However, while there was some evidence of effect modification by activities as a silversmith (based on small numbers), confounding alone did not explain the relatively high number of deaths from nasopharyngeal cancer in this plant (Marsh et al., 2007a).

Two analyses have been conducted to re-analyse the data from the most recent update of the NCI cohort, with a focus on solid tumours (Hauptmann et al., 2004). The first included an analysis of exposure category and SMR, as well as an analysis of Plant 1, where five of nine deaths from nasopharyngeal cancer occurred, compared with all other plants in the cohort (Marsh & Youk, 2005). Using their own cut-points of exposure, the authors concluded that their analysis lent uncertainty to the findings from the NCI cohort. In another re-analysis, the authors further controlled for the effect of plant for the peak-exposure metric and performed sensitivity analyses by imputing additional cases, which showed instability in the risk estimates (Marsh et al., 2007b). The authors concluded that an interaction between plant group and exposure makes generalization beyond Plant 1 difficult.

2.1.2. Case–control studies

The relationship between nasopharyngeal cancer and exposure to formaldehyde has also been investigated in seven case–control studies, five of which found elevated risks for overall exposure to formaldehyde or in higher exposure categories, although not all were statistically significant (see Table 2.3 available at http://monographs.iarc.fr/ENG/Monographs/vol100F/100F-24-Table2.3.pdf; Vaughan et al., 1986b; Roush et al., 1987; West et al., 1993; Vaughan et al., 2000; Hildesheim et al., 2001). One study found an elevation among women, but not men (Olsen et al., 1984) and one found no evidence of an association (Armstrong et al., 2000). Two case–control studies were considered as the most informative because of their size, their exposure assessment, and the evaluation of potential confounders. The first, a population-based case–control study in the USA, showed a significant association for the workers whose exposure duration had been the longest (OR = 2.1; 95%CI: 1.0–4.5, P_trend = 0.07), but not for maximum exposure (P_trend = 0.57) (Vaughan et al., 2000). When the analysis was limited to differentiated squamous-cell and epithelial NOS, there was a significant association in the highest exposure category for both duration and cumulative exposure with significant exposure-response trends (P_trend = 0.014 and 0.033, respectively). In the other study, conducted in Taiwan, China, an OR of 1.6 (95%CI: 0.91–2.9, P_trend = 0.08) was found in the category with the longest duration of exposure (Hildesheim et al., 2001). For cumulative exposure, there was a non-significant elevation in the highest exposure category and the trend test was not significant (P = 0.10). In subanalyses that were restricted to cases and controls who were seropositive for antibodies against Epstein-Barr virus, the association between exposure to formaldehyde and nasopharyngeal cancer appeared to be stronger, with an OR for ever exposure of 2.7 (95%CI: 1.2–6.2). However, no clear dose–response pattern was observed with increasing duration of exposure, or with estimated cumulative exposure.

2.1.3. Meta-analyses

A meta-analysis published in 1997 included some but not all of the above studies, and found an overall meta-relative risk for nasopharyngeal cancer of 1.3 (95%CI: 1.2–1.5) (Collins et al., 1997). From a pooled analysis including the three recently updated industrial cohorts (Coggon et al., 2003; Hauptmann et al., 2004; Pinkerton et al., 2004), Bosetti et al. (2008) reported an overall SMR of 1.33 (95%CI: 0.61–2.53). A recently published meta-analysis included both case–control studies (n = 6) and cohort studies (n = 7) (Bachand et al., 2010). For the case–control studies, the overall OR was 1.22 (95%CI: 1.00–1.50), with the meta-regression OR no longer significant when limited to studies that included adjustment for socioeconomic status, smoking or location. The risk estimate for cohort studies was 0.72 (95%CI: 0.40–1.29), including seven studies (Bachand et al., 2010). For the cohort studies, the authors used a re-analysis of the NCI cohort study from which Plant 1 was left out (Marsh & Youk, 2005).

2.2.1. Cohort studies

Excess mortality from leukaemia has been observed relatively consistently in studies of professional workers (i.e. embalmers, funeral parlour workers, pathologists and anatomists), with six mortality studies showing positive associations (Walrath & Fraumeni, 1983, 1984; Levine et al., 1984; Stroup et al., 1986; Hayes et al., 1990; Hall et al., 1991) and one not (Logue et al., 1986; see Table 2.2 online).

A weakness of the proportionate mortality studies among professionals has been the lack of exposure assessment. A recently published nested case–control study conducted among professionals in the funeral industry examined lifetime work practices and exposure in the funeral industry to develop metrics of exposure among this group, which included duration of jobs held while embalming, number of embalmings, average intensity of embalming and peak exposure (Hauptmann et al., 2009). Details of work practices were obtained by interviews with next of kin and co-workers. Positive associations were seen – at many levels of exposure and for multiple exposure metrics – for deaths from lymphohaematopoietic malignancies of non-lymphoid origin (n = 48). For myeloid leukaemia (n = 34) the OR was 13.6 (95%CI: 1.6–119.7; P_trend = 0.020) for the longest duration of work in jobs with embalming. Because only one case was reported to have never embalmed, additional analyses were conducted in which those who reported to have embalmed ≤ 500 times were taken as the referent group, to provide a more stable estimate. Results were attenuated, but still significant (OR = 3.9; 95%CI: 1.2–12.5). [There was a considerable amount of missing data that required imputation for analyses.]

The findings for leukaemia in studies of professional workers appeared to be contradicted by the lack of such findings among industrial workers. However, some evidence for an excess of deaths from leukaemia has been reported in the recent updates of two of the three major cohort studies of industrial workers. Since the previous evaluation (IARC, 2006), the NCI cohort of industrial workers in the USA has been updated with an additional ten years of mortality data resulting in 123 deaths from leukaemia, including 48 from myeloid leukaemia (Beane Freeman et al., 2009). This update extended the mortality follow-up through 2004 and included additional deaths before 1994 that had not been previously considered. Risk estimates from follow-up through 2004 were diminished for leukaemia and myeloid leukaemia compared with the follow-up through 1994 (Hauptmann et al., 2003), when both conditions had been significantly associated with increasing peak-exposure and average intensity of exposure to formaldehyde. As in the previous analysis of leukaemia, the association in the most recent update was stronger for myeloid leukaemia and peak exposure than for lymphatic leukaemia and for other metrics of exposure (Beane Freeman et al., 2009). However, because the last known exposure occurred in 1980 and median follow-up was over 40 years, the authors not only examined risks at the end of follow-up in 2004, but also assessed associations over time by extending follow-up in yearly increments. Risks appeared to be highest before 1980, but only achieved statistical significance in the mid-1990s, when a sufficient number of deaths had accrued. Additional analyses with time since first exposure and time since first high peak-exposure indicated that risks were highest during the first twenty-five years. Patterns were similar, but attenuated, for average intensity of exposure; no association was observed with cumulative exposure.

Mortality from leukaemia was also found to be in excess in an update of the study of US garment workers exposed to formaldehyde (Pinkerton et al., 2004). A small and statistically non-significant excess was observed for the entire cohort in comparison with rates among the general population (SMR = 1.09; 95%CI: 0.7–1.63). This excess was somewhat stronger for myeloid leukaemia (SMR = 1.44; 95%CI: 0.80–2.37), which is consistent with the findings from the study of industrial workers in the USA and several of the studies of medical professionals and embalmers. The excess was also stronger among workers with a longer duration of exposure and longer follow-up, and among those who had been employed early in the study period when exposures to formaldehyde were believed to be highest. The positive associations observed in the subgroup analyses presented in the study of US garment workers were based on a relatively small number of deaths, and were thus not statistically stable.

The updated study of British industrial workers found no excess mortality for leukaemia among all workers exposed to formaldehyde (SMR = 0.91; 95%CI: 0.62–1.29) or among those with the highest exposure (SMR = 0.71; 95%CI: 0.31–1.39) (Coggon et al., 2003). The lack of positive findings in this study is difficult to reconcile with the findings from the studies of garment workers and industrial workers in the USA, and with the results of studies on professionals exposed to formaldehyde. This British study is a relatively large, high-quality study with sufficiently long follow-up to have had a reasonable chance to detect an excess of deaths from leukaemia. It did not examine specifically the risk for myeloid leukaemia, which represented the strongest finding in the studies of garment workers and industrial workers in the USA and in several of the studies of medical professionals and funeral workers.

2.2.2. Case–control studies

Three case–control studies evaluated exposure to formaldehyde and risk for leukaemia (Linos et al., 1990; Partanen et al., 1993; Blair et al., 2001; Table 2.5 online). However, the numbers of exposed cases were few, and no significant elevations of risk were found.

2.2.3. Meta-analyses

A meta-analysis published in 2004 for ‘ever exposure’ to formaldehyde and leukaemia included eighteen studies and presented separate analyses by type of job: for industrial workers, the mRR was 0.9 (95%CI: 0.8–1.0); for embalmers 1.6 (95%CI: 1.2–2.0); and for pathologists and anatomists 1.4 (95%CI: 1.0–1.9), with an overall mRR of 1.1 (95%CI: 1.0–1.2) (Collins & Lineker, 2004). In another meta-analysis, analysis was restricted to 13 cohort or proportionate mortality studies and similar results were found, with a pooled RR based on the weighted average of the SMRs for leukaemia among industrial workers of 0.9 (95%CI: 0.75–1.07), based on 122 deaths, and of 1.39 (95%CI: 1.15–1.68) among professionals, based on 106 deaths (Bosetti et al., 2008). A further meta-analysis differed from these two previous ones by excluding all proportionate mortality studies and including the most recent update of the NCI cohort (Bachand et al., 2010). For leukaemia overall, a risk estimate of 1.05 (95%CI: 0.93–1.20) was calculated for ‘ever exposure’, based on 15 studies with the use of a fixed-effects model. For myeloid leukaemia, the calculated mRR was 1.09 (95%CI: 0.84–1.40, based on three studies) and for lymphatic leukaemia the mRR was 1.11 (95%CI: 0.81–1.52, based on two studies).

Zhang et al. (2009) published a meta-analysis that included 15 cohort or case–control studies. The authors selected only studies where it was clear that the workers had been exposed to formaldehyde. In contrast to the other meta-analyses, this one used one exposure metric from each study and considered the highest exposure category for calculating the mRR. For leukaemia, the mRR was 1.54 (95%CI: 1.18–2.00). In addition, a separate analysis of myeloid leukaemia – for the six studies that reported it – found an mRR of 1.90 (95%CI: 1.31–2.76).

2.3.1. Cohort studies

An analysis of proportionate cancer incidence among industrial workers in Denmark showed an increased risk for squamous-cell carcinomas (Hansen & Olsen, 1995, 1996). No excess of mortality from sinonasal cancer was observed in the three recently updated studies of industrial and garment workers in the USA, and of chemical workers in the United Kingdom (see Table 2.1 online; Coggon et al., 2003; Hauptmann et al., 2004; Pinkerton et al., 2004).

2.3.2. Case–control studies

The association between exposure to formaldehyde and the risk for sinonasal cancer has been evaluated in six case–control studies that primarily focused on formaldehyde (see Table 2.4 available at http://monographs.iarc.fr/ENG/Monographs/vol100F/100F-24-Table2.4.pdf; Olsen et al., 1984; Hayes et al., 1986; Olsen & Asnaes, 1986; Vaughan et al., 1986a; Roush et al., 1987; Luce et al., 1993; Pesch et al., 2008). Four of these six studies reported an increased risk (Olsen et al., 1984; Hayes et al., 1986; Vaughan et al., 1986a; Luce et al., 1993).

2.3.3. Pooled analysis

Four of the cohort studies contributed to a pooled analysis that collated occupational data from 12 case–control investigations (Luce et al., 2002). After adjustment for known occupational confounders, this analysis showed an increased risk for adenocarcinoma associated with high exposure (> 1 ppm) to formaldehyde in both men (OR, 3.0; 95%CI: 1.5–5.7) and women (OR, 6.3; 95%CI: 2.0–19.7). An exposure–response trend was observed in relation to an index of cumulative exposure. There was some evidence of an association with squamous-cell carcinoma.

[Most epidemiological studies of sinonasal cancer have not distinguished between tumours that arise in the nose and those that develop in the nasal sinuses. Thus, any effect on the risk for nasal cancer specifically would tend to be diluted if there were no corresponding effect on the risk for cancer in the sinuses and could mask its detection, particularly in cohort studies that have relatively low statistical power. However, the apparent discrepancy between the results of the case–control as compared with the cohort studies might also reflect residual confounding by wood dust in the former. Almost all of the formaldehyde-exposed cases in the case–control studies were also exposed to wood dust, which resulted in a high relative risk, particularly for adenocarcinomas.]

4.3.1. Humans

Micronucleus formation has been repeatedly reported to occur in cells of the nasal and oral mucosa of formaldehyde-exposed humans. The outcome of studies on induction of micronuclei, sister chromatid exchange and chromosomal aberrations in the lymphocytes of exposed humans – which is pertinent to the question concerning the potential of formaldehyde to cause lympho-haematopoietic cancer – has been less consistent (see Table 4.1).

Table 4.1

Cytogenetic studies on formaldehyde–exposed humans.

DNA–protein crosslinks in circulating white blood cells were found to be higher in 12 workers exposed to formaldehyde in an anatomy department and a pathology institute than in eight controls (P = 0.03) (Shaham et al., 1996). The number of crosslinks tended to be higher in workers who had been exposed longer (exposure duration, 2–31 years). Smoking had no effect. In a subsequent study, Shaham et al. (2003) reported an increase in the number of DNA–protein crosslinks in lymphocytes and in serum concentrations of the p53 protein in hospital pathology-department workers. The exposed subjects were assigned to a high-exposure subgroup (mean formaldehyde concentration in air, 2.24 ppm) and a low-exposure subgroup (mean, 0.4 ppm), based on personal sampling and field sampling for 15-minute periods on typical working days. The control group consisted of personnel of the administrative sections in the hospital. The amount of protein–cross-linked DNA was statistically significantly higher in the exposed group than in the controls (0.20 vs 0.14; these values are the ratios between protein-bound DNA – precipitable with sodium dodecyl sulfate – and total DNA) after controlling for age, smoking and other factors. Very little difference in DNA–protein crosslink levels was observed between the high- and low-exposure groups (0.20 vs 0.19); or between workers with > 16 years or < 16 years of exposure (0.20 vs 0.19). The percentage of formaldehyde-exposed male workers who had pantropic p53-protein (wild-type plus mutant p53) concentrations in serum higher than 150 pg/ml was statistically significantly greater than in the control group (54.8% vs 36.5%, P < 0.05; this difference was not seen in female workers). Formaldehyde-exposed workers with DNA–protein crosslink levels above the median had a significantly greater likelihood of having p53 concentrations in serum above 150 pg/ml. [The Working Group noted that the rationale for using a p53-protein level of 150 pg/ml as a cut-point was based upon previous experience with this assay; no reason is given for using 16 years as the cut-point between longer/shorter exposure. Questions have also been raised about the persistence of DNA–protein crosslinks, which are thought to be rapidly repaired within the cell (Schmid & Speit, 2007)]. In an earlier study, Casanova-Schmitz et al. (1984) failed to observe DNA–protein crosslinks in bone marrow of Fischer-344 rats exposed to [¹⁴C]- and [³H]-formaldehyde.

Compared with matched controls, pathology/anatomy workers from five hospitals, who were exposed to mean formaldehyde concentrations of 2.0 ppm (range, < 0.1 to 20.4 ppm) during 15 minutes, or to 0.1 ppm (range < 0.1 to 0.7 ppm) during 8 hours, showed a statistically significant increase in bi-nucleated cells and in mono-centromeric micronucleus formation in a cytokinesis-blocked micronucleus assay combined with fluorescence in situ hybridization (FISH) (Orsière et al., 2006).

A higher frequency of micronuclei was found in exfoliated nasal and oral cells of students with short-term exposure (3 hours per day; 3 days per week, for 8 weeks) to an average of 0.508 ± 0.299 mg/m³ formaldehyde, compared with controls (Ying et al., 1997, 1999). No increase in micronucleus formation or in the level of sister chromatid exchange (SCE) was observed in lymphocytes. Ye et al. (2005) reported a comparative analysis of 18 workers involved for various periods (mean, 8.5 years; range, 1–15 years) in a formaldehyde-manufacturing process, 16 waiters exposed for 12 weeks to an indoor source of formaldehyde during interior renovations, and a control group of 23 students; all were non-smokers. Average formaldehyde exposure-concentrations were 0.011 mg/m³ for the student controls; 0.107 mg/m³ for the waiters; and 0.99 mg/m³ for the formaldehyde-plant workers. There was a statistically significantly higher frequency of micronuclei in nasal mucosal cells and of SCE in peripheral lymphocytes in the workers at the formaldehyde-manufacturing plant, but not in the waiters, although both groups were exposed to formaldehyde at comparable concentrations (Table 4.1). The result is in line with the much longer exposure duration for the plant workers. The same authors (Ye et al., 2005) reported an increase in B-cells and changes in the ratios between lymphocyte subsets, similar to those reported by Madison et al. (1991) in an Alaskan community subject to acute formaldehyde exposure (estimated at 2–5 ppm) for a few days. Blood analyses were done three years after the accident. Total white blood cell counts and total lymphocyte counts did not differ from those in the control community in Alaska, also measured three years later (Madison et al., 1991).

An increase in SCE and other genotoxic effects were observed in the lymphocytes of workers with long-term exposure to formaldehyde (Costa et al., 2008). Thirty workers from four pathology/anatomy hospital units and 30 matched controls were included in the study. Compared with the control group, statistically significant effects were seen in SCE (6.13 ± 0.29 vs 4.49 ± 0.16 SCE/cell, P < 0.05), micronucleus frequency (in 1000 bi-nucleated cells: 5.47 ± 0.76 ‰ vs 3.27 ± 0.69 ‰, P = 0.003) and tail length in the comet assay (60.0 ± 2.31 μm vs 41.85 ± 1.97 μm, P < 0.05). A statistically significant positive correlation was found between formaldehyde exposure levels and micronucleus frequency and tail length. The mean formaldehyde exposure was 0.44 ppm (range, 0.04–1.58 ppm). None of the observed effects were related to the duration of exposure.

No genotoxic effects were observed in a study of 36 laboratory workers at a cancer-research institute who were exposed to 4.9–268.7 μg/m³ formaldehyde. There was a direct relationship between formaldehyde exposure levels and the presence of a formaldehyde human serum-albumin (FA-HSA) conjugate. The genotoxic endpoints measured were SCE, micronuclei and chromosome aberrations, but these did not show significantly elevated levels (Pala et al., 2008). Although a small study, its strength is its linkage to a biological marker of formaldehyde exposure.

Hayes et al. (1997) evaluated O⁶-Alkyl-guanine-DNA–alkyltransferase (AGT) activity as a measure of DNA-repair capacity in blood lymphocytes of 23 science students in a mortuary, before and after a nine-week period of classroom exposure to approximately 1.5 ppm formaldehyde. A statistically significant finding was that more students had a reduction in AGT activity than an increase. There was no clear link between the extent of exposure to formaldehyde and AGT activity.

Zhang et al. (2010) cultured myeloid progenitor cells from the peripheral blood of formaldehyde-exposed workers and controls and measured leukaemia-specific chromosomal changes. In a subset of ten of the most highly exposed subjects in their study, monosomy (loss) of chromosome 7 and trisomy (gain) of chromosome 8 were significantly elevated in the myeloid progenitor cells of formaldehyde-exposed workers compared with the same phenomena in 12 unexposed controls. The loss of chromosome 7 and gain of chromosome 8 were examined because they are among the most frequent cytogenetic changes observed in myeloid leukaemia and myelodysplastic syndromes; these events have been shown to be affected by exposure to the established human leukemogen, benzene. [The Working Group noted that the study is small and needs to be replicated].

4.3.2. Experimental systems

4.4.1. Cancer of the nasopharynx and nasal sinuses

Mechanistic evidence supporting a causal relation between inhalation of formaldehyde and induction of cancer of the nasopharynx and nasal sinuses is based on the chemical reactivity of formaldehyde in producing DNA–protein cross-links, and its genotoxicity in vitro and in vivo, including in the nasal cells of exposed humans. Computational fluid-dynamic models of formaldehyde in the nasal passages of rats, monkeys and humans have generally been accurate in predicting the area in the nose with the highest number of DNA–protein crosslinks (Georgieva et al., 2003). Local effects in the nasal passages, genotoxicity, and cell-proliferation rate appear to be the major determinants of nasal carcinogenicity after exposure to formaldehyde.

4.4.2. Leukaemia

The findings reviewed in IARC Monograph Volume 88 (IARC, 2006) pertaining to a potential mechanism for formaldehyde-induced leukaemogenesis were summarized as follows: “Based on the data available at this time, it was not possible to identify a mechanism for the induction of myeloid leukaemia in humans.” The Working Group further stated that “It is possible that formaldehyde itself can reach the bone marrow following inhalation, although the evidence is inconsistent.” Since that time, Zhang et al. (2009), reviewed potential pathways by which formaldehyde could act as a leukaemogen. Three mechanisms were suggested:

• by damaging stem cells in the bone marrow directly, as most other leukaemogens do;
• by damaging haematopoietic stem/progenitor cells circulating in the peripheral blood and
• by damaging the primitive pluri-potent stem cells present within the nasal turbinates and/or olfactory mucosa.

This subject was reviewed by Heck & Casanova (2004), Pyatt et al. (2008), and Goldstein (2011).