Use of this information is subject to copyright laws and may require the permission of the owner of the information, as described in the ECHA Legal Notice.
EC number: 200-001-8
CAS number: 50-00-0
Relevant dose-descriptor for the endpoint concerned: NOAEC systemic 15 ppm (18 mg/m³), chronic inhalation study in mice and rats (CIIT, 1981; Kerns et al., 1983; 6 h/d, 5 d/week for 24 months; cf. Section 7.5.3).The systemic NOAEC corresponds to a total body burden in rats of 18 mg/m³ x 0.29 m³/kg (for 6 h) = 5.2 mg/kg bw/d. This body burden is more than one order of magnitude lower than the systemic NOAEL of the drinking water study (82 mg/kg bw/d) in which systemic renal effects were related secondarily to reduced water intake. Thus, in the inhalation study any possible systemic effects would still be related to formaldehyde per se.
Correction of starting point:
For the total respiratory tract bioavailability in humans and rats is 100%; rapid metabolism in both species (see endpoint summary in Section 7.1). No correction necessary for absorption.
Same route of exposure, no correction.
Correction of exposure conditions (human exposure duration 8 h/day versus 6 h/day in the rat study). Factor 6/8 = 0.75.
NOAEC 18 mg/m³ x 0.75 = 13.5mg/m³ (corrected NOAEC).
Differences in respiratory volumes between experimental animals (at rest) and humans (light activity). In human at rest the respiratory volume is 6.7 m³ in 8 h and in workers 10 m³ in 8 h: Factor 6.7/10= 0.67. Corrected NOAEC 13.5 mg/m³ x 0.67= 9 mg/m³
Relevant dose-descriptor for the endpoint concerned: NOAEL systemic 82 mg/kg bw/day with systemic renal toxicity attributed to reduced water intake, chronic drinking water study in rats (Til et al., 1989).
Route of exposure (oral versus dermal exposure): the oral NOAEL of 82 mg/kg bw/day is converted to a dermal NOAEL by a factor of 100/4 (oral absorption rat 100 %/ dermal absorption monkey 4%) = 82 x 25 mg/kg bw/day = 2050 mg/kg bw/day.
Although different exposure conditions were given (occupational exposure during 8 h shift in humans versus continuous exposure in rats for 24 h) no correction is necessary for daily exposure duration since in both species the effects are related to the daily body dose. In the drinking water study rats were exposed 7 d/week leading to a correction by 7/5 for the workplace: 2050 x 7/5 = 2870 mg/kg bw/d.
Formaldehyde is a normal endogenous metabolite and has been found in all tissues investigated. Human exhaled air contains formaldehyde in concentrations in the order of 0.001–0.01 mg/m3, with an average value of about 0.005 mg/m3.Formaldehyde reacts at the site of first contact and/or is eliminated rapidly as formic acid in the urine or as CO2 in the expired air or it enters the carbon-1 pool in the body. Dermal absorption should differentiate between penetration through the skin possibly leading to systemic effects and penetration through and into the skin possibly leading to local effects. For monkeys, penetration through the skin was 4% and through + into skin 15%. In rats and guinea pigs, higher dermal absorptions were found. In humans 76% absorption was reported after inhalation exposure for the nasal passages (90% in rats) and almost complete absorption in the total respiratory tract of humans and rats (Overton et al., 2001). Therefore 100% systemic absorption is used for respiratory tract of rats and humans. Almost complete absorption was reported after oral exposure. Formaldehyde is rapidly metabolized (t1/2 ~ 1 min) by formaldehyde dehydrogenase (FAD). Since toxicity is a consequence of exposure to a very reactive parent compound that is not removed from the site of application allometric scaling is not appropriate (Guidance on information and chemical safety assessment; Chapter R.8; p. 31). FAD is highly conserved and is found in all tissues and species investigated as the most important scavenger for the highly toxic endogenous formaldehyde. In several investigations no polymorphism of FAD has been detected in humans. Therefore, no assessment factors for inter- and intraspecies variability are appropriate as long as formaldehyde is the toxic entity under consideration.
Skin and eye irritation/corrosion:
Aqueous solutions of formaldehyde like formalin (40%) induced skin corrosion in rabbits. Skin irritant effects are expected at concentrations > 3%. No studies according to current guidelines are available on eye irritation; however, formaldehyde has corrosive properties (no testing required). There is evidence that sensory eye irritation in humans due to exposure to gaseous formaldehyde is the most sensitive endpoint. From clinical studies with volunteers, it is concluded that the NOAEC for subjective and objective sensory eye irritation was 0.7 ppm in case of a constant exposure level or 0.4 ppm with four 15 min peaks of 0.8 ppm.
There is sufficient evidence for sensitizing properties of formaldehyde in the guinea pig maximisation test (GPMT), in the Buehler test in guinea pigs and in the mouse local lymph node assay (LLNA). The EC3 value (3-fold stimulation of proliferation as an index of the relative potency of a contact allergen) in the LLNA was 0.54% for an acetone solution of formaldehyde (corresponding to 135 μg/cm²; potency categorisation based on LLNA: strong sensitiser; for a solution in acetone the NOAEC for induction was calculated to be 0.06% corresponding to 15.3 µg/cm²). Formaldehyde is also a dermal allergen in humans. Anaphylaxis has been documented in case reports. The threshold in sensitized humans under exaggerated occluded test conditions was estimated to be 3 µg/cm². In non-sensitzised humans a threshold of 0.037% corresponding to 37 µg/cm² has been determined for induction of skin sensitization.A weight of evidence assessment does not indicate that formaldehyde may induce or exacerbate asthma or its related lung function.
Repeated dose toxicity:
There is evidence that formaldehyde induces toxic effects only at the site of first contact after oral, dermal or inhalation exposure. General signs of toxicity occur secondary to these local lesions. In chronic drinking water studies in rats, local effects in the forestomach and stomach were induced, the NOAEC is 0.020-0.026% formaldehyde in drinking water. For systemic effects the NOAEL is >= 82 mg/kg bw/day in males and 109 mg/kg bw/day in females. Studies on repeated dermal dose toxicity in compliance to current Guidelines are not available. Local effects in the upper respiratory tract were induced after repeated inhalation exposure in experimental animals. The most sensitive site in rodents and monkeys is the respiratory epithelium in the anterior part of the nasal cavity. At higher dose levels also the olfactory epithelium, larynx or trachea were affected, especially in monkeys. Rats are more sensitive than mice or hamsters. The LOAEC is 2 ppm in rats (2.4 mg/m³), 3 ppm in monkeys and 6 ppm in mice. The overall NOAEC in experimental animals for local effects is 1 ppm (1.2 mg/m³). The NOAEC for systemic effects in long-term inhalation studies in rats and mice is 15 ppm. As mentioned above the most sensitive endpoint in humans is sensory eye irritation by exposure to formaldehyde gas. In a controlled study with volunteers the NOAEC was 0.7 ppm at constant exposure conditions and 0.4 ppm with peaks of 0.8 ppm. These NOAECs pertained likewise to persons with high and low sensitivity for sensory irritation as determined by CO2 threshold.
Genotoxicity in vitro:
Gene and chromosome mutagenic activity of formaldehyde are well documented from in vitro studies and numerous studies on other endpoints suggested further evidence for genotoxicity of formaldehyde in vitro. Clastogenicity is the predominant mutagenic endpoint in mammalian cell systems. DNA-protein cross-links (DPC) as premutagenic lesion have been sufficiently investigated including threshold and repair. The threshold for DPC formation in cultured human lymphocytes is >10 μM (0.3 μg/mL), significant effects were reported at >=25 μM (0.75 μg/mL); DPC induced by concentrations up to 100 μM (3 μg/mL) are completely removed before lymphocytes start to replicate. There is some evidence that clastogenic effects are related to DPC formation.
Genotoxicity in vivo:
The available data in experimental animals demonstrated the genotoxic activity of formaldehyde at the site of first contact after oral exposure. Studies on local mutagenic effects in humans suggested increased micronucleus frequencies in nasal and buccal cells. However, these studies gave conflicting results and reliable test methods are not yet available. Therefore a final conclusion is not possible. Negative results with buccal cells were reported in a recent controlled clinical study after repeated exposure to <= 1ppm. The mechanism of clastogenicity might be related to DNA-protein cross-links and their repair. DNA-protein cross-links and DNA adducts at the site of first contact have been demonstrated after inhalation exposure in rats and monkeys. The most rigorous studies did not give evidence for systemic genotoxicity in experimental animals or in humans.
In a valid chronic oral study local effects were induced in the forestomach and stomach of rats at a concentration of 0.19% (NOAEC 0.02%) but no carcinogenic activity. In another drinking water study local carcinogenic effects were shown in rats, however, the validity was limited and the histopathological criteria for the papillomas described are unclear. Clear local carcinogenic effects were reported in the nasal cavity of rats after long-term inhalation exposure to ≥ 6ppm.
In workers exposed to formaldehyde statistically increased risk for nasopharyngeal cancer has been especially reported in one study. But the relationship to formaldehyde has been challenged and the new discovery of more than 1000 additional cancer deaths that were not previously identified by the NCI needs to be evaluated for its relevance/impact on cancer risk.In an update including about 10 additional years of exposure only one additional NPC was observed, but at the lowest exposure category. Nevertheless, there was still a statistically significant increase for the tumor type. But the conclusion of the authors that formaldehyde is a human carcinogen for NPC has been challenged. A correlation between formaldehyde exposure and leukaemia, especially myeloid leukaemia, was seen in some studies but not all. However, this tumour type is considered biologically not plausible. Again, the evidence that formaldehyde exposure is causally associated with leukemia has been challenged by several authors.Generally, there is evidence for a threshold effect concerning tumor induction in the upper respiratory tract, based on the highly non-linear dose response relationship for tumor induction and effects related to carcinogenicity, like histopathological lesions, cell proliferation, DNA protein cross links or effects on the mucus apparatus. On this basis the substance has been classified in Cat 4 by German MAK commission as for the carcinogenic potential a non-genotoxic mode of action is of prime importance and genotoxic effects play no or at most a minor part (Greim 2000). RAC (2012) proposed not to classify formaldehyde as a human carcinogen.
There is no evidence for effects on reproductive organs in experimental animals after oral or inhalation exposure. There is no evidence for adverse effects of formaldehyde on embryo and foetal development even at dose levels leading to local maternal toxicity.
Derivation of DNELs - worker
Derivation of DNELs for long-term exposure (additional information)
Worker-DNEL long-term for dermal route-local
Dermal exposure of workers via skin contact to formaldehyde solutions. No reliable dose descriptor available for repeated dermal exposure. Limited data indicate that formaldehyde does not act as a complete carcinogen or as a promoter or initiator on the skin after topical application.Most sensitive endpoint skin sensitisation in humans. The worker DNEL long term should provide protection against induction of sensitization.
There are three possible starting point for derivation of this DNEL.
a.EC3 in the LLNA
b.NOAEC in the LLNA
c.NOAEC in human volunteers (Marzulli and Maibach, 1974)
Ad a: starting point EC3 in the LLNA
four EC3 values with different solvents are reported:
- 0.54% in acetone (Hilton et al., 1988)
- 0.33% in dimethylformamide (Hilton et al., 1988)
- 0.35% in acetone/olive oil (4;1) (Basketter et al., 2001)
- 0.96% in acetone/olive oil (4;1) (De Jong et al., 2007)
Workers will (almost) exclusively be exposed to an aqueous solution, that was not tested in the studies above. Dimethylformamide is known to strongly enhance dermal penetration and therefore would lead to clearly exaggerated conditions and dimethylformamide gives the same EC3 as acetone/olive oil (4:1) in one experiment while in another one a much higher EC3 was found. The hydrophilic solvent acetone is considered to better simulate an aqueous solution as compared to dimethylformamide. Therefore the EC3 value in acetone is taken for DNEL derivation (further justification is given in the enspoint summary for sensitization):0.54% corresponding to 135 µg/cm².
Correction of starting point
Same bioavailability assumed in humans and animals, no correction. As the permeability of the skin of mice is higher than that of humans, this is a worst case assumption so no correction for absorptive differences between human and mouse skin is applicable.
Difference in exposure conditions: De Jong et al.(2007) have shown that prolonged exposure as compared to the standard LLNA will lead to an increase of cell proliferation in the lymph node by a factor of 3. Therefore the EC3 will be divided by 3.
Starting point: 0.54 : 3 = 0.18% corresponding to 135 µg/cm² : 3 = 45 µg/cm².
Interspecies differences: dermal sensitization is a consequence of local exposure to the very reactive parent chemical at the site of application. Therefore an AF for allometry is not appropriate. As the permeability of the skin of mice is higher than that of humans the AF for interspecies differences is 1.
Intraspecies differences: because dermal sensitization is governed by direct chemical reaction with macromolecules and because detoxification is widely conserved without an indication for polymorphism in humans, an AF of 1 is appropriate.
Differences in duration of exposure: this has been taken into consideration by correction of the starting point.
Issues related to dose-response: the starting point for the DNEL calculation is a LOAEL (EC3) (threshold concentration);assessment factor 3.
Overall assessment factor: 3
DNEL worker chronic dermal local: 0.18 : 3 = 0.06% corresponding to 45 : 3 = 15 μg/cm²
Ad b: starting point NOAEC in the LLNA
The NOAEC in the LLNA was 0.04% in acetone/olive oil (4:1); this solvent gave an EC3 of 0.35% (Basketter et al., 2001). As stated above the results obtained with acetone (EC3 0.54%; Hilton et al., 1998) should be preferred leading to a solvent adjusted NOAEC of 0.06% for acetone corresponding to 15 µg/cm².
Same bioavailability assumed in humans and animals, no correction. As the permeability of the skin of mice is higher than that of humans, this is a worst case assumption.
Starting point: 0.06 : 3 = 0.02% corresponding to 15 : 3 = 5 µg/cm².
Issues related to dose-response: the starting point is the NOAEC; assessment factor 1.
Overall assessment factor: 1
DNEL worker chronic dermal local: 0.02% corresponding to 5 μg/cm²
Ad a + b: selection of DNELs based on LLNA:
When comparing the results obtained for the DNELs derived from EC3 and NOAEC, preference should be given to the EC3 approach. EC3 values are more stable than NOAECs as the latter depend very much on the spacing of doses in each experiment. Specifically, the dose spacing in the Basketter et al. (2001) study between NOAEC (0.04%) and LOAEC (0.2% with a stimulation factor of 1.9) was quite high (factor of 5).
Therefore, based on animal LLNA studies the DNEL derived from EC3 should be used: 0.06% corresponding to 15 μg/cm²
Ad c: starting point NOAEC in human volunteers
The NOAEC for induction of sensitization in human volunteers (Marzulli and Maibach, 1974) is 0.037% corresponding to 37 µg/cm².
Route of exposure, no correction.
Difference in exposure conditions: human volunteers were subjected to severe exposure conditions: for induction occlusive dressing over 48 or 72 h and 10 repeated applications at the same site within 3.5 weeks; elicitation under occlusive dressing over 72 h with 0.37% formaldehyde. Therefore no correction.
Starting point: 0.037% corresponding to 37 µg/cm².
Interspecies differences: not applicable
Intraspecies differences: volunteers with normal healthy skin were selected. Workers possibly exposed to formaldehyde will also have healthy skin. Therefore an AF of 1 is appropriate.
Differences in duration of exposure: AF of 1 because the exposures were greater than in the standard HRIPT used for such volunteer studies (Basketter et al., 2008).
DNEL worker chronic dermal local: 0.037% corresponding to 37 μg/cm²
Ad a + b + c: selection of DNELs based on animal and human data:
If human and animal data are available, generally data from humans as the most relevant species should be selected for DNEL derivation provided a rigorous study design was used and reliable results were obtained for humans. Exposure conditions, number of subjects and experience of the investigators do fulfill these conditions. Furthermore, it has to be taken into consideration that the permeability of mouse skin is higher than that of human skin. As expected, the DNELs derived from the LLNA are lower than that based on the investigation in humans on a µg/cm² basis.
Therefore, the DNEL is based on human data:
Worker-DNEL long-term for dermal route-local: 0.037% corresponding to 37 μg/cm²
Worker-DNEL long-term for inhalation route-local
Formaldehyde is a carcinogen in the upper respiratory tract as has been shown by experiments in rats and mice, while the epidemiological evidence in this respect is not convincing (RAC, 2012). It is generally agreed in the science community that this effect is driven by local cytotoxicity followed by regenerative cell replication (RCP) (McGregor at al., 2006; MAK, 2000; SCOEL; 2015). RCP has been studied in the most sensitive rat and numerous studies have consistently shown a NOAEL of 2 ppm (Gelbke et al., 2014). Interspecies extrapolations have to take into account respiratory physiology which shows major differences when comparing rats on the one side, with non-human primates and humans on the other (DeSesso, 1993). Therefore, ideally RCP caused by cytotoxic irritation should be studied in humans. As this is not possible because of ethical reasons, sensory irritation should be taken as the more sensitive surrogate as compared to cytotoxic irritation. Numerous studies are available for sensory irritation and all have shown that eye irritation is a more sensitive parameter than irritation to the nose or the throat, the potential targets for formaldehyde carcinogenicity. Therefore, using sensory eye irritation as the starting point for risk assessment has a high, albeit not quantifiable, inbuilt margin of safety because of sensory instead of cytotoxic irritation and irritation to the eye instead of the upper respiratory tract. Studies in humans may either measure subjective ratings of sensory irritation, or objective parameters like eye blinking frequency or conjunctival redness. Subjective ratings depend on many potential confounders (like anxiety, expectation, or other psychological factors) and therefore preference must be given to studies measuring objective parameters.
Such studies are now available (Lang et al., 2008; Mueller et al., 2013) showing no increase of objective signs of eye irritation at concentrations of 0.7 ppm over 4 h or at 0.4 ppm superimposed by 4 peaks of 0.8 ppm. For sensory irritation the is strong evidence that after exposure periods of 4 h an effect plateau has been reached (Brüning et al., 2014). This NOEL is well in line with former studies relying mainly on subjective reporting of sensory irritation. In total, a database for about 400 exposed volunteers has been published. As sensory irritation is a physiological effect (although taken here a surrogate for the adverse cytotoxic irritation) these concentrations represent a NOEL and not a NOAEL.
For assessment of carcinogenicity, genotoxicity has also to be taken into consideration, although this effect is not considered to be the driver of carcinogenicity. The genotoxic effects of formaldehyde are formation of DPX as known since decades and DNA-adducts in the form of the N2-hydroxymethyl-dG adduct as more recently identified. NOAELs for these effects have not been established. Apart from species differences due to respiratory physiology, also the genotoxicity by endogenous formaldehyde, being present in every cell, has to be taken into account. A differentiation between genotoxicity caused by endogenous and exogenous formaldehyde has been achieved by inhalation of [14CD2]formaldehyde. After a single 6 h exposure to 0.7 ppm only 1% of the total DNA-adduct load is caused by the exogenous formaldehyde, the rest of 99% stem from endogenous formaldehyde. A highly non-linear dose response relation was observed, for example a 21.7-fold increase in exposure leads to a 286-fold increase in exogenous adducts. But these adducts have a biological half-life of about 7 days. At steady state after prolonged exposure the exogenous adducts will be by a factor of 5.5 higher than after a single exposure, as shown by inhalation at 2 ppm over 28 days. Extrapolation to the lowest experimental dose shows that after a prolonged exposure to 0.7 ppm (the lowest dose tested in this series of experiments) endogenous adducts will exceed the exogenous by a factor of 14 to 22. And due to the highly non-linear dose response relationship this factor will even be higher at lower concentrations and endogenous adducts will be far prevail. In addition, it was shown that the ratio of exogenous/endogenous adducts in primates is by a factor of ~5 lower than in rats, because of the different respiratory physiology and higher endogenous adducts in the primates. This finding strongly supports the former observation of Casanova et al. (1991) that DPX formation in rats is much higher than in monkeys. These data demonstrate that genotoxicity in rats is more pronounced than in primates.
As cytotoxic irritation is the driver for carcinogenicity, the starting point for a carcinogenic risk assessment should be the NOEL of 0.7 ppm observed in controlled exposure studies with humans and the question of an appropriate assessment factor (AF) for intra-human variability has to be addressed. No guidance is given by ECHA (2012) for an AF to be applied to a NOEL for physiological sensory effects, but ECETOC (2010) proposes an AF of 1. This is supported by Brüning et al. (2014) who also proposed an AF of 1 for sensory irritation in controlled human exposure studies based on a detailed assessment of a recent database. The question remains whether the sensitivity of the volunteers in the studies of Lang et al. (2008) and Mueller et al. (2013) might have been lower than that of the target population, i.e. the workforce. Firstly, it was shown by Mueller that the NOEL was the same for subjects hypo- and hypersensitive for nasal irritation to CO2. In addition, the age of the subjects studied was <~40 years and is therefore at the lower end of that of the general workforce. It is known (Doty et al., 2004) that sensory irritation decreases with age. And finally, all volunteers were non-smokers (what will not be the case for the workforce) and smokers are less sensitive to nasal irritation (Doty et al., 2004). As both of these studies fulfill the quality criteria given by Brüning et al. (2014) an AF of 1 for the NOEL is appropriate. This AF of 1 would lead to a DNEL of 0.4 ppm with superimposed peaks of 0.8 ppm reflecting the variable workplace exposure situations.
Finally the NOAELs observed in rats have to be discussed in relation to derivation of the DNEL. As the inhalation route is under consideration for rats and humans, the AF of 4 for correction of differences in metabolic rate is not to be applied. This leaves according to ECHA (2012) default AFs of 2.5 for toxicokinetic intra-species differences not related to metabolic rate (small part) and toxicodynamic differences (larger part) and for intra-human variability of 5 for workers. Both of these default AFs can be reduced considerably for 3 reasons: 1. The toxic action of formaldehyde depends on the molecule itself by simple adduction to DNA and proteins which is basically identical in all species and therefore without inter-species variability. 2. Detoxification is governed by ADH5, an enzyme highly efficient and conserved in all species. Especially for humans, it has been shown that polymorphism is not likely to play a role for detoxification of exogenous or endogenous formaldehyde (apart from Fanconi Anemia patients) provided that the endogenous pool is not depleted. This will not be the case at exposure concentrations <2 ppm (Andersen et al., 2010; Casanova et al. 1989). 3. For high quality sensory irritation studies or when based on a broad database, an AF of 1 is appropriate to address intraspecies variability for sensory irritation (Brüning et al., 2014).For such a situation with good evidence that the default AFs are over-conservative, no guidance is given by ECHA (2012). But it is admitted by ECHA that the database they considered is very scarce. This limitation is overcome by the systematic analysis of Brüning et al. (2014) proposing a total factor of 3 for extrapolating animal data to humans.
This AF of 3 should then be applied to the NOAEL for RCP. An analysis of all related publications has consistently shown a NOAEL of 2 ppm (Gelbke et al., 2014). This would lead to a DNEL of 0.7 ppm.
For histopathological lesions a NOAEL of 1 ppm was found for rats and monkeys. Monkeys were continuously exposed over 26 weeks (22 h/d, 7 d/week) (Rusch et al., 1983) without any exposure free time for repair of lesions in contrast to exposures at the workplace. Therefore, workplace exposure conditions would most probably lead to a higher NOAEL in monkeys. For the NOAEL of 1 ppm in rats it has to be taken into consideration that the lesions observed at the LOAEL (2 ppm) cannot be regarded as prestages to tumor development, neither by their histopathological features nor by their location (Gelbke et al., 2014). Applying the AF of 3 to the histopathological NOAEL would lead to a DNEL of 0.3 ppm. In summary, by application of the AFs proposed here, the following DNELs can be derived:
- Sensory irritation in humans: 0.4 ppm (AF of 1)
- RCP in rats: 0.7 ppm (AF of 3)
- Histopathological lesion in rats and monkeys: 0.3 ppm (AF of 3)
These alternatives for derivation of a DNEL all lead to very similar values. The scientifically most defensible DNEL is that derived from sensory irritation in humans because
- Of the inbuilt margin of safety between sensory eye irritation and cytotoxic irritation to the upper respiratory tract
- Such data obviates application of AFs, be it default or arbitrary AFs
- The volunteers exposed were more sensitive to sensory irritation than the general workforce because of age and smoking status
The AF of 1 is supported by the database analyzed by Brüning et al. (2014) and several published studies with about 400 exposed volunteers in total. Other factors supporting that this DNEL is sufficiently conservative are
- The higher sensitivity of rats as compared to nonhuman primates with regard to formation of DNA adducts and DPX
- The steep upward bent of the dose response curve at low concentrations for all decisive parameters (tumor incidence, cell proliferation, DPX formation, DNA adducts).
On this basis, 0.3 ppm is taken as DNEL long-term, local, inhalation, workers.
This DNEL will also protect the workforce from undue nuisance of odor and slight subjective irritation as has been reported by some of the volunteers, especially with a high PANAS score for negative affectivity and that have also been reported under control condition (0 ppm formaldehyde).This DNEL is supported by mathematical risk extrapolations from experimental animals to humans (Conolly et al., 2004; Andersen et al., 2101; Starr and Swenberg, 2013). It also corresponds to the OELs developed by MAK (2000), SCOEL (2015)., and the German AGS (2015), all using slightly different approaches.
Relevant dose-descriptor for the endpoint concerned: NOAEC systemic 15 ppm (18 mg/m³), chronic inhalation study in mice and rats (CIIT, 1981; Kerns et al., 1983; 6 h/d, 5 d/week for 24 months; cf. Section 7.5.3). The systemic NOAEC corresponds to a total body burden in rats of 18 mg/m³ x 0.29 m³/kg (for 6 h) = 5.2 mg/kg bw/d. This body burden is more than one magnitude lower than the systemic NOAEL of the drinking water study in which systemic renal effects might have been caused by metabolites of formaldehyde. Thus, in the inhalation study any possible effects would still be related to formaldehyde per se.
For the total respiratory tract bioavailability in humans and rats is 100%; rapid metabolism in both species (see endpoint summary in Section 7.1); no correction necessary for absorption.
Correction of exposure conditions (human exposure duration 24 h/day and 7 d/week versus 6 h/day and 5 d/week in the rat study). Factor 6/24 x 5/7. NOAEC 18 mg/m³ x 0.25 x 5/7 = 3.2 mg/m³ (corrected NOAEC).
No differences in respiratory volumes between experimental animals (at rest) and humans (at rest), no correction.
Relevant dose-descriptor for the endpoint concerned: NOAEL systemic 82 mg/kg bw/day with secondary renal toxicity due to reduced water intake, chronic drinking water study in rats (Til et al., 1989).
Systemic bioavailability in humans via the dermal route is derived from monkey data with 4%. Route of exposure (oral versus dermal exposure): the oral NOAEL of 82 mg/kg bw/day is converted to a dermal NOAEL by a factor of 100/4 (oral absorption rat 100 %/ dermal absorption monkey 4%) = 82 x 25 mg/kg bw/day = 2050 mg/kg bw/day.
No different exposure conditions, no difference in physical activity, no corrections.
Relevant dose-descriptor for the endpoint concerned: NOAEL systemic >= 82 mg/kg bw/day with systemic renal toxicity at higher doses, chronic drinking water study in rats (Til et al., 1989).
- Bioavailability in humans versus rat concerning oral route: no data are available for humans (rat: nearly complete absorption) but the same bioavailability is assumed, no correction.
- Same route of exposure, no correction.
- Similar exposure conditions in rats (drinking water) and humans (drinking water/food stuff). No correction necessary.
Endogenous sources of formaldehyde:
In humans, as in other animals, formaldehyde is an essential metabolic intermediate in all cells. It is produced endogenously from serine, glycine, methionine and choline, and it is generated in the demethylation of N-, O- and S-methyl compounds. It is an essential intermediate in the biosynthesis of purines, thymidine and certain amino acids. Owing to its high reactivity at the site of contact and rapid metabolism, exposure of humans, monkeys or rats to formaldehyde by inhalation does not alter the concentration of formaldehyde in the blood from that endogenously present, which is about 2–3 mg/l for each of the three species. This concentration represents the total concentration of both free and reversibly bound endogenous formaldehyde in the blood. The absence of an increase after inhalation is explained by the fact that formaldehyde reacts rapidly at the site of contact. As a consequence of the endogenous formation, human exhaled air contains formaldehyde in concentrations in the order of 0.001–0.01 mg/m3, with an average value of about 0.005 mg/m3.
Derivation of general population DNEL for long term inhalation (local):
Using the worker DNEL (long-term, local) of 0.3 ppm and a time extrapolation factor of 3 to account for 24 hour exposure of the general population, the DNEL for the general population for long term inhalation can be calculated to
0.3 ppm/3 = 0.1 ppm
After a very comprehensive review, the World Health Organization (WHO) established an indoor air quality guideline for short- and long-term exposures to FA of 0.1 mg/m3 (0.08 ppm) for all 30-min periods at lifelong exposure. This value was mainly based on the results of the Lang et al. (2008) study using an assessment factor of 5 and five other supporting studies on sensory irritation. Nielsen et al. reviewed the WHO guideline value in 2013 and 2017 (Nielsen et al. 2013, 2017) an found no recent and additional data or evidence questioning this value.
Thus, the DNEL for the general population for long term inhalation exposure to formaldehyde is set to 0.1 mg/m3 with the provisions of the WHO guidance value.
In the context of a restriction of formaldehyde and formaldehyde releasers, the sensory irritation studies of Lang et al. (2008) and Müller et al. (2014) were reevaluated (RAC & SEAC, Opinion on an Annex XV dossier proposing restrictions on formaldehyde and formaldehyde releasers, compiled version, adopted 13 March 2020 by RAC and 17 September 2020 by SEAC). In the view of RAC, RAC considers that the absence of a sensory irritation effect at formaldehyde concentrations below 1.24 mg/m³ (1 ppm) is uncertain. Due to high variability of the measured effect (large ranges, high standard deviations), the low numbers of volunteers (yielding low statistical power) and the additional uncertainties identified, false negative results at 0.62 mg/m³ (0.5 ppm) or lower cannot be excluded. The studies of Lang et al. (2008) and Mueller et al. (2013), thus, cannot be used to derive a DNEL, because the uncertainties are too high and the lack of observed effects cannot be considered as evidence for the absence of dose-related effects. As a consequence, RAC focused on endpoints other than sensory irritation to derive a DNEL for long term exposure of the general population.
In the RAC & SEAC opinion (2020), RAC derived an indoor derived no effect level (DNEL) of 50 µg/m³ for formaldehyde and considered the WHO indoor guideline value of 100 µg/m³ as not sufficiently protective for the consumers. The RAC DNEL was based on studies in animals, i.e. rats and monkeys. Human data from sensory irritation studies were evaluated by RAC but not taken into account for derivation of the DNEL due to a lack of reliability and robustness of these data.
RAC derived DNELs using different points of departure in rats and monkeys considered as precursor events to malignant tumour responses and the tumour response itself. Finally, RAC selected the DNEL derived from metaplasia/hyperplasia in monkeys, i.e. 50 µg/m³, as the most relevant.
The following table displays DNELs based on different points of departure considered as precursor events to malignant tumour responses in comparison to the malignant tumour response:
DNA adduct formation
Cell proliferation in rats
Metaplasia/ hyperplasia in monkeys
Cytotoxicity, metaplasia/ hyperplasia and benign tumours in rats
Malignant tumours in rats for comparison
# if corrected for sub-chronic exposure duration
* POD NOAEC
** POD LOAEC
DNELs were calculated from the no adverse effect concentration (NOAEC) from the respective endpoint with use of the default assessment factors (AF) as given by the ECHA Guidance R8 which were partly modified. As the genotoxic potential of formaldehyde is not expected to give rise to mutagenicity at low doses and the effects of DPX formation cannot be regarded as adverse per se, consequently, a DNEL derived based solely on genotoxicity results was considered inappropriate. The DNELs on the other endpoints in rats and monkeys include an- AF of 2.5 for inter- or intraspecies differences in sensitivity- AF of 3.16 for intraspecies variability which was reduced from the default AF of 10 due to the assumed lower variability of these endpoint in the general population.- AF of 2 to correct for exposure duration (only used for cell proliferation and cytotoxicity/metaplasia in rats)
Although the rat DNELs on DNA adduct formation and cancer precursor effects were lower and may therefore be considered to be taken forward, RAC decided to take preference of a long-term DNEL and proposed a DNEL of 0.05 mg/m³ based on the LOAEC for hyperplasia/metaplasia in nasal turbinates in the study with monkeys (Rusch et al., 1983) and taking into account all additional DNELs derived based on studies on precursor events observed in rats, which were in a similar range.
In the same document (RAC & SEAC 2020), as part of the socio-economic assessment SEAC estimated that the EU had a housing stock of about 250 million dwellings in 2015 with an estimated 0.7 % of the EU’s housing stock, or 1.75 million, coming from newly built/completed dwellings. In addition, according to section 2.5.2 of the Background Document, the average household in the EU had 2.3 members in 2016. Based on these numbers, SEAC assumed that 4 million individuals live in newly built dwellings each year (= 1.75 million dwellings built per year with 2.3 individuals per dwelling). This means that for the RAC proposal to break even, 6 522 nasopharyngeal cancer cases would have to be avoided among these 4 million individuals. In other words, 1 631 in 1 million (= 6 522 nasopharyngeal cancer cases in 4 million individuals) would have to suffer nasopharyngeal cancer for the RAC proposal to break even. However, according to IARC (2018) data, in 2018 the incidence rate of nasopharyngeal cancer in the EU was only 7.5 in 1 million (crude rate). This means that one would need to see a more than 217 times higher incidence rate than is actually observed for the RAC proposal to break even clearly implying that the RAC evaluations and DNELs are by far overconservative.
Thus, RAC ended up with DNELs from rats in the range of 6 – 10 µg/m³. Interestingly, exhaled air of humans contains formaldehyde in concentrations in the order of 1 – 10 µg/m³, with an average of about 5 µg/m³, i.e. the formaldehyde concentration in human exhaled air is in the same concentration range as the RAC rat DNELs. Given the fact, that indoor concentrations of formaldehyde under typical living conditions in Europe are between 11 and 42 µg/m³ and usually below 30 µg/m³, these DNELs imply that we would all be at risk of developing effects in the respiratory system or tumors from indoor formaldehyde exposure. This is clearly not the case. This simple comparison and ‘reality check’ clearly shows that – though considering a threshold mode of action - the RAC DNELs from rats are clearly overconservative due to the formal use of assessment factors. Even the final DNEL of 50 µg/m³ from the monkey study is overconservative in that way, that the same AF of 2.5 was taken for inter- and intraspecies differences in sensitivity in rats and monkeys. However, it is evident that the sensitivity of monkeys better reflects human sensitivity than rats and therefore a lower AF in monkeys would be justified with which one would end up more or less with the WHO value.
In summary, the RAC DNEL for indoor formaldehyde concentrations is non-realistically low and overconservative due to the formal use of mostly default assessment factors.
In contrast, the WHO indoor guideline value was set to 100 µg/m³ in 2010 in a less formal but rather more pragmatic approach considering and evaluating over 150 publications on various endpoints. The WHO value is a 30 min time-weighted average concentration for every 30 min interval during the day lifelong. It will prevent effects on the respiratory tract as well as long term health such as nasopharyngeal cancer. The WHO indoor guideline value was reviewed again in 2013 and 2017 considering all relevant endpoints and all available publications. In neither of the reviews, there were any findings or effects, either acute (e.g. eye irritation) or long-term (e.g. cancer) in nature, published that questioned or contradicted the WHO guideline value. The recent studies included once more confirmed the non-linear dose response relationship of formaldehyde and that no health effects have been observed below this value.
In addition, the German Committee on Indoor Air Guide Values of the German UBA established an indoor air guidance value for long term exposure in 2016 which is identical to the WHO guideline value, i.e. 100 µg/m³. A particular focus during the discussions in the German committee was the potential association between formaldehyde exposure and asthma in children. Therefore,
“UBA (2016) performed a review of epidemiological studies investigating the association between formaldehyde exposure and the induction or exacerbation of asthma in children. On the basis of the available data, UBA concluded that there is no clear association between formaldehyde exposure in the indoor environment and asthma in children. It was stated that the above mentioned epidemiological studies (e.g. Krzyzanowski et al., 1990) suffer from small sample sizes (which was much larger than in the studies by Lang et al. (2008) and Mueller et al. (2013)), from implausible formaldehyde concentrations, and the fact that other substances or factors initiating asthma and asthma-like complaints were not adequately considered. Results derived from controlled human exposure studies as well as animal experiments support their opinion.”
Thus, both the reviews of the WHO guidance value as well as the German guidance value clearly confirmed that the WHO value reflects a highly precautionary and conservative approach. There weren’t any findings or effects, either acute (e.g. eye irritation) or long-term (e.g. cancer) in nature, published that questioned or contradicted the WHO guideline value of 100 µg/m³ and thus, there would be no health benefits at all by lowering the indoor air guidance value to the RAC DNEL level of 50 µg/m³. Due to the formal use of mostly default assessment factors, the RAC DNEL for indoor formaldehyde concentrations is non-realistically low and overconservative and does not provide any health benefits or other added value as compared to the WHO guidance value.
Information on Registered Substances comes from registration dossiers which have been assigned a registration number. The assignment of a registration number does however not guarantee that the information in the dossier is correct or that the dossier is compliant with Regulation (EC) No 1907/2006 (the REACH Regulation). This information has not been reviewed or verified by the Agency or any other authority. The content is subject to change without prior notice.Reproduction or further distribution of this information may be subject to copyright protection. Use of the information without obtaining the permission from the owner(s) of the respective information might violate the rights of the owner.
Welcome to the ECHA website. This site is not fully supported in Internet Explorer 7 (and earlier versions). Please upgrade your Internet Explorer to a newer version.
Do not show this message again