Registration Dossier

Diss Factsheets

Administrative data

Key value for chemical safety assessment

Genetic toxicity in vitro

Description of key information

 The test substance is part of a category approach of methylenediphenyl diisocyanates (MDI)  with existing data gaps filled according to ECHA guidance on Read Across (ECHA, 2017).  The read-across category justification document is attached in IUCLID section 13. In this category Substances of the MDI category all share similar chemical features namely that they a) all contain high levels of mMDI, and b) contain have at least two aromatic NCO groups that are electronically separated from other aromatic rings by at least a methylene bridge. It is the NCO value (driven by the bioaccessible NCO groups on relatively soluble mMDI and low molecular weight species (e.g. three-ring oligomer) which is responsible for chemical and physiological reactivity and subsequent toxicological profile. The substances 4,4’-MDI, 4,4’-MDI/DPG/HMWP and pMDI are identified as the boundary substances within this MDI category. These three substances represent the of key parameters (i.e. mMDI content and NCO value) within the MDI category that determine the hypothesized Mode of Action (MoA). Although NCO groups are present on the higher molecular weight constituents, they do not contribute to the toxicity profile because they are hindered due to their increased size and hydrophobicity.


Future in vitro Testing Plan

While broad coverage of MDI Category substances is availalbe for bacterial gene mutation, additional OECD 471 studies will be conducted on all remaining data-gaps to complete the data set.    Additional in vitro micronucleus studies (OECD 487) will be conducted on ALL category substance to assess potential effects on cytogenetics.  Combined with in vivo mutagenicity testing on the worst-case substance (4,4’-MDI) and hypothesized MoA, this complete data-set will confirm the lack of mutagenic potential for the entire MDI Substance category.  


Bacterial Systems

MDI substances are virtually insoluble in water and require a solvent to ensure homogeneous dispersion in in vitro genotoxicity assays. Dimethylsulphoxide (DMSO) has been used routinely as the solvent of choice for such assays and when used in an Ames test to assess MDI mutagenicity, positive results were often obtained (only in TA98 and TA100 strains) (Herbold, 1980c; Herbold, 1980a). However, further work demonstrated the selection of the solvent resulted in these positive results. Gahlmann et al. (1993) found there is a chemical conversion of MDI to MDA in the solvent which could explain a number of positive responses recorded in some genotoxicity assays in vitro.


A detailed evaluation of the stability of MDI substances in DMSO by Herbold et al. (1998) and Seel et al. (1999) showed that there was a rapid breakdown of MDI substance in DMSO with less than 40 % of the initial amount remaining after fifteen minutes with almost complete breakdown within two hours. A HPLC examination of the breakdown products showed the presence of MDA which is known to produce positive responses in various in vitro genotoxicity assays including mutations in Salmonella typhimurium. To determine if the positive results seen in in vitro genotoxicity assays in which a MDI substance was dissolved in DMSO were in fact a consequence of the chemical conversion of MDI to MDA, Herbold et al. (1998) (key study with reliability 2) and Seel et al. (1999) undertook a series of mutagenic investigations using dry ethyleneglycol dimethylether (EGDE) as the organic solvent, as investigations indicated the MDI substance was stable in this solvent with no formation of MDA.


The studies with Salmonella typhimurium (TA1535, TA1537, TA1538, TA100, TA98) showed quite clearly the absence of any mutagenic response when MDI (including 4,4’-MDI, 2,4’-MDI, MDI Mixed Isomers and pMDI) was dissolved in EGDE (Herbold et al., 1998). In contrast, when an MDI substance was dissolved in DMSO clear positive effects were seen in the presence of metabolic activation consistent with generation of MDA. Based on such evaluation it was concluded that the results of in vitro genotoxicity studies undertaken using solvents such as DMSO must be treated with caution as any positive response may very well be an artefact of the testing conditions caused by the breakdown of the isocyanate into the mutagenic amine.

Numerous other category substances have been tested in high quality bacterial mutagenicity studies are negative and includes (often multiple) representatives from all MDI subgroups. In these Ames studies with EGDE as the solvent, all were negative both with and without S9 incubation (TA98, TA100, TA1535, TA1537, WP2 uvrA). 


Mammalian Systems

Based on these observations the use of results from in vitro tests in aqueous cell systems are problematic because of physico-chemical characteristics of the MDI substances and interaction with the test system components. Those in vitro studies not addressing problems of solvent selection and hydrolysis of the substance (which represent the majority of in vitro investigations) are considered to be invalid, and not useful for determining the genotoxic potential of MDI substances. Because of these technical problems conducting mammalian cell gene mutation assays in vitro is challenging and interpretation of data problematic, thus an assessment relies on data from in vivo studies.

Mouse Lymphoma

In order to evaluate both point mutations and chromosomal (clastogenic) alterations, mouse lymphoma L5178Y specific locus mutation studies on both mMDI and pMDI dissolved in DMSO were conducted (McGregor et al. (1991), summarized in EC (2005)). The experiments were carried out over the concentration range 2.5 μg/mL to 250 μg/mL. Although 4,4’-MDI showed some evidence of mutagenicity in the presence of S9, it only occurred under excessive cytotoxic conditions (95 % toxicity) at concentrations ≥ 200 μg/mL (0.8 mM). As excessive cytotoxicity should be avoided in the mouse lymphoma assay, these results are considered invalid (OECD TG 490). Polymeric MDI showed no evidence of mutagenic activity in the mouse lymphoma forward mutation assay; pMDI was cytotoxic at ≥106 μg/mL. Further, the use of DMSO as the solvent may promote diamine formation and confound the results.  

Chromosomal Aberration

Maki-Paakkanen and Norppa (1987) exposed human whole-blood lymphocytes to pMDI (dissolved in acetone) for 24 hours. In the absence of S9, pMDI induced chromosome aberrations at all doses tested (0.54 - 4.30 μl/mL; 1.9 – 15 mM); in the presence of S9 (1.5 hours), aberrations were significantly increased only at the highest concentration. Polymeric MDI also increased sister chromatid exchanges (SCEs) at the highest dose tested (2.17 μl/mL; 7.7 mM) with and without S9. The effects did not exhibit a dose-response. On addition to culture medium, pMDI formed polymer-like fibers at all doses; at high doses, these polymers made metaphase analysis and toxicity determinations problematic. The lack of a clear dose-response in the induction of chromosome damage may be related to the solubility of pMDI or the reactive degradation products (e.g. MDA).


Zhong and Siegel (2000a) performed an in vitro micronucleus test with Chinese hamster lung fibroblasts (V79) exposed to 4,4’-MDI, MDA, bis-cysteine-MDI conjugate and bis-glutathione (GSH)-MDI conjugate. All were formulated in DMSO. While 4,4’-MDI (0.2 – 4 mM) either with or without S9 did not result in an increase in micronuclei, MDA (0.25 – 2.5 mM) significantly increased micronuclei in a dose-dependent fashion by equal to less than four-fold (without S9) to 6-fold (with S9). The MDI conjugates (0.1 – 1 mM) also produced a significant increase (equal or less than three-fold) in micronuclei. However, there was no clear dose-response. The bis-GSH-MDI conjugate also produced a significant (three to four-fold) increase in the number of cells in metaphase as well as intracellular precipitates. The authors suggested that the increases in intracellular precipitates and the percent of cells in mitosis following bis-GSH-MDI exposure may involve reaction of MDI with spindle components. In a further study, Zhong et al. (2001) used a fluorescent anti-kinetochore antibody assay to demonstrate that the micronuclei induced by the MDI-conjugates were due primarily to a disturbance of the mitotic spindle (70 % - 75 % kinetochore +), while the micronuclei induced by MDA were not (15 % kinetochore +). According to the authors, the latter effect is consistent with an interaction between MDA and DNA resulting in chromosome breakage. [Note: The induction of micronuclei by MDA and the two MDI-conjugates occurred in the absence of cytotoxicity as evidenced by no changes in the nuclear division index.] Given the high (mM) concentration of GSH in the alveolar lining fluid, the transfer of 4,4’-MDI to the spindle apparatus by GSH-MDI conjugates and the resulting aneuploidy is not implausible. However, as these experiments were conducted with DMSO as the solvent, precipitation was present in all concentration tested and a corroborating lack of aneugenic effects in a valid in vivo study, these results were deemed unreliable. 

As described for the bacterial studies, in vitro tests are often problematic due to interactions of the MDI substance with the test system buffer. In general, it can be concluded that the mammalian in vitro studies for gene mutations and chromosomal aberrations that are available are considered low quality (reliability 3) and do not contribute greatly to the assessment (Maki-Paakkanen and Norppa, 1987; McGregor et al., 1991; Zhong and Siegel, 2000a). Further, as in vivo cytogenicity and mammalian gene mutation test are available, then these studies do not represent a significant data gap and are waived according to Annex XI, Section 2 (ECHA, R.7.7-1). 

Endpoint conclusion
Endpoint conclusion:
no adverse effect observed (negative)

Genetic toxicity in vivo

Description of key information

The following information is taken into account for any hazard / risk assessment (genetic toxicity in vivo):

DNA Binding

In addition to the in vitro studies described above, the potential for MDI substances to react with DNA was investigated in vivo via dermally and inhalation. Vock et al. (1995) exposed the skin of rats to [14C]-mMDI at doses of approximately 3 mg and 7 mg for 24- and 48-hours, respectively, before isolation of epidermal nuclear protein and DNA. Epidermal nuclear protein exhibited very high specific radioactivity. While radioactivity associated with epidermal DNA at 48-hours post-exposure corresponded to 30 – 40 adducts per 108 DNA-nucleotides, 32P-postlabelling analysis did not reveal isocyanate-DNA adducts. The nuclear protein radioactivity in the liver, lung and kidney was >6,000-fold lower than in the epidermis. DNA reactivity in the liver was at the limit of detection. In view of the high radioactivity associated with epidermal protein, the authors indicated (a) that a substantial fraction of the DNA radioactivity was due to contamination by protein, and (b) that the results of their study were consistent with “the negative result seen with the 32P-postlabeling method adapted to detect isocyanate-DNA adducts (Vock et al., 1995)”.

In this earlier study, rats dermally exposed to 9 mg 4,4’-MDI for 90 minutes were found to have 7 adducts per 108 nucleotides, a level which at the time suggested to the authors that “minute fraction of MDI can reach DNA in vivo in a chemically reactive form” (Vock et al., 1995). Vock et al. (1996) also used the 32P-postlabeling method to detect DNA adducts after exposure via whole-body inhalation (17 hours/day, 5 days/week) to mMDI at concentrations of 0, 0.3, 0.7, or 2.0 mg/m3. After one year of exposure, neither MDI-DNA nor MDA-DNA adducts were detected in the lung, liver, bladder, kidney, peripheral lymphocytes, or respiratory epithelium. In contrast, low levels of MDA-DNA adducts were detected in the olfactory epithelium at 5 (at 0.3 mg/m3), 9 (at 0.7 mg/m3), and 10 (at 2 mg/m3) adducts per 1,010 nucleotides; the detection limit was 4 adducts per 1,010 nucleotides. Although the authors stated the MDA-DNA adduct co-chromatographed with the one formed in the liver of rats after oral gavage of MDA, the toxicological significance of this finding at a non-tumorigenic site is unclear and may reflect sample contamination with protein (Vock and Lutz, 1997). Taken together, these data support the conclusion that the DNA reactivity of MDI itself is extremely weak but that mMDI exposures might lead to the generation of MDA-DNA adducts, albeit at extremely low levels (near the detection limit of the method) in certain tissues, likely limited to the portal of entry. Interestingly, neither MDI nor MDA adducts were detected in the tumor target tissue (lungs).

Local mutagenicity

Two in vivo Comet Assays were employed to investigate the genotoxic potential of 4,4’-MDI at the site of contact (Suter, 2016; Randazzo, 2017). As the site of contact tissue, bronchoalveolar lavage (BAL) cells were selected to be analyzed. Although BAL cells do not represent a target cell population for lung carcinogenesis, they have been regularly used in the assessment of pulmonary genotoxicity after inhalation or instillation of various substances (Haney et al., 1999; Bornholdt et al., 2002; De Boeck et al., 2003; Zhao et al., 2004; Monleau et al., 2006; Micale et al., 2008; Neuss et al., 2010; Jackson et al., 2012; Wallin et al., 2017) and are commonly selected in the assessment of pulmonary genotoxicity after inhalation or instillation as mentioned in OECD TG489 (OECD, 2016). Pulmonary macrophages are predominant in the BAL fluid, usually one of the first cell types to come into contact with aerosols and thus represent the site of contact in lung. This is especially relevant for very reactive substances such as 4,4’-MDI. As described in Toxicokinetics, the first reaction of MDI substances is with macromolecules at site of contact (alveolar fluid) where formation of high molecular weight MDI-conjugates is immediate and complete (Wisnewski et al., 2013; Wisnewski et al., 2016). The predominant exposure of macrophages is shown in a number of studies, in which upon microscopic examination of lung tissue they were found to be filled with yellowish pigment (for instance Reuzel et al. (1994a) and Pauluhn et al. (1998); this pigment is exposure-related, and likely represents phagocytized polyurea. Also, foamy macrophages were observed, filled with large amounts of phospholipids, which is indicative of perturbation of surfactant homeostasis (Pauluhn et al., 1999b; Kilgour et al., 2002).

Further, the use of BAL cells, versus using minced lung tissue, reduces the dilution of effects from unexposed cells that occurs when for instance alveolar epithelial cells are isolated. It is technically difficult to harvest and sort different cell types from minced lung tissue. Harsh treatment often results in high background damage (e.g. Knaapen et al. (2002) and Dusinska et al. (1998)) and a major share of cells in minced lung tissue consist of haematopoetic cells. Compared to isolated rat lung macrophages, a five times higher background level of DNA damage was found in isolated rat lung epithelial cells, a difference that persisted for at least 48 h in culture (Dusinska et al., 1998).  This means that only a relatively large induction of DNA damage would be identified by the sensitive Comet Assay.

Importantly, during development of the historical control database in a preliminary study (Chappelle and Bruce, 2017), it was observed that the positive control substance EMS rapidly distributes from the lymph system to the BAL to induce DNA damage. This was confirmed during the main study and further demonstrates the value of BAL cells as an appropriate tissue to be examined.

The first study was a non-guideline comet assay (Suter, 2016) adopted to explore the sequence of irritation-related events on pulmonary cells harvested by bronchoalveolar lavage (BAL) from rats exposed by inhalation to aerosolized 4,4’-MDI. Groups of rats were nose-only exposure to air and concentration x time (C x t) intensities in the range of 50 and 500 mg/m³xh (10, 20, 100, or 180 mg 4,4’-MDI/m³ for 3 or 6 hours). BAL-fluid/-cells were collected 3 hours (day 0), 1 and 3 days post-exposure to establish a C x t and postexposure duration-related matrix of acute inflammation (lung weights, LDH, total protein, and cytodifferentiation of BAL-cells), restoratively increased annexin V activity, increased caspase 3/7 activity correlated with increased %Tail intensity of BAL-cells. The dose levels, inflammation endpoints, and sampling points were selected based on earlier inhalation toxicity studies in rats (Pauluhn, 2011a).  Aerosolized paraffin at about 50 mg/m³xh 21 served as non-irritant/ genotoxic reference for the phagocytosis of waxy aerosols by alveolar macrophages (AM) and the dose level was selected to match one of the 4,4'-MDI dose levels (but a different time point was examined for that dose).

Results from this study indicated that inhalation exposure to 4,4'-MDI led to a dose dependent mild positive response in the Comet assay observed at day 0 and 1 stating with an exposure of 20 mg/m3 for 3 hours (54.3 mg/m3xh). Increase in tail length (approximately 2.3-fold increase compared to approximately 11 to 22-fold increase in the positive control (MNU administrated orally)) correlated with markers of cytotoxicity, apoptosis and inflammation. A no observed adverse effect concentration (NOAEC) of 10 mg/m3 for 6 hours (61.8 mg/m3xh) was identified for the Comet response but yet still induced a clear acute toxic effect in the BAL. At day three after exposure there was no significant effect on tail length even at a very high C x t of 518.4 mg/m3xh (100 mg/m3 for 6 hours). Effects on markers of alveolar toxicity at 20 mg/m3 4,4'-MDI for 3 hour (54.3 mg/m3xh) and sampled at day 0 correlated with results obtained for the inert organic particle solid paraffin at 25 mg/m3 at day one. For example, increase in alveolar macrophages were not identified, but increase in markers for apoptosis (caspase 3/7 or annexin V) was observed together with a mild comet response. The increased apoptosis observed for at least one of the two apoptosis parameters after inhalation of 4,4'-MDI, but not after exposure to the positive control for direct genotoxicity (MNU), put the positive Comet assays result into perspective of primary versus secondary effects. There is understood to be a link between macrophage activation and reactive oxygen species generation (as seen in the current cytodifferentiation results), leading to an inflammatory response including oxidative burst, finally resulting in oxidative DNA damage. Accordingly, these results do not point to primary, test substance-induced suggest that DNA damage caused the 4,4’-MDI at concentrations ≥ 20 mg/m3 is the result of excessive toxicity (cytotoxicity, apoptosis and/or inflammation) rather than direct genotoxicity.

The second in vivo comet assay was conducted according to OECD TG 489 under GLP conditions (Randazzo, 2017). In addition to the BAL cells, the liver and glandular stomach was included in the analysis. The liver was analyzed since it is the site of primary metabolism (and potential systemic genotoxicity), while the glandular stomach was included due to possible secondary exposure after clearance of 4,4’-MDI via the mucociliary escalator (i.e. local effects at a secondary site of contact). Groups of 12 Wistar rats were exposed to actual concentrations of 2.5, 4.9 and 12 mg/m3 for 6 hours with the maximum dose selected from pilot range-finding study and previous studies as a concentration that will induce significant local cellular damage (Hotchkiss et al., 2017, Pauluhn, 2001, Kilgour, 2002).  Further, results from the previous comet study that showed that ≥20 mg/m3 MDI induced genetic damage via excessive cytotoxicity, apoptosis and/or inflammation (Suter, 2016). As in the exploratory study by Suter (2016), bronchoalveolar lavage (BAL) was performed in all animals at the scheduled necropsies, and the BAL fluid (BALF) was assessed for cytotoxicity and inflammation to determine non-specific or direct toxicity at the sites of contact following acute exposure as recommended by the test guideline. Animals were sacrificed approximately one-hour post-exposure and the other six/group approximately 18 hours post-exposure, or 2 to 4 hours after the second dose for the positive control group (MNU via gavage). An inflammatory response was observed in the lung in the high-concentration group, characterized by a significant increased influx of neutrophils. Also, time and dose-dependent apoptosis/necrosis was induced in the lung in all treatment groups, as shown by Annexin V staining and LDH activity. However, these effects were only persistent to the 18 hour time point at 12 mg/m3.  In the stomach and liver, there were no indications for 4,4’-MDI-induced toxicity as indicated by histopathology. Overall, these results are consistent with local cellular toxicity of the lung as the critical mode of action of MDI toxicity.  Based on the magnitude of the differences noted in the BAL endpoints, 12 mg/m3 (measured concentration) was considered to be the maximum tolerated concentration (MTC) to avoid confounding secondary DNA damage resulting from local cytotoxic effects.  

DNA damage was investigated using the Comet Assay with results clearly negative for all three investigated tissues (BAL cells; liver; stomach) indicating that 4,4’-MDI is not genotoxic at the portal of entry at exposures of up to the maximum tolerated concentration, as indicated by local cytotoxicity.

Systemic mutagenicity

One guideline and several non-guideline studies are available.

In the key study, Pauluhn et al. (2001) conducted an OECD 474 guideline micronucleus study to GLP exposing rats via both whole body and nose only to 4,4’-MDI for 1 hour per week for 3 weeks, with bone marrow examinations at one and two days post-exposure at concentrations up to 118 mg/m3. Although toxic effects at the portal of entry (e.g. respiratory distress, increased lung weights) were observed, there was no evidence of an MDI-induced increase in the frequency of MN-PCE at any of the time points selected. MN-PCEs were significantly increased in rats treated with the positive control when compared to both the negative control and MDI-exposure groups. 

In a supporting study Lindberg et al. (2011) investigated the genotoxicity of inhaled 4,4’-MDI in male C57Bl/6J mice by examining micronucleated polychromatic erythrocytes (PCE) in bone marrow and peripheral blood. Mice were exposed head-only to 4,4’-MDI aerosols (mean concentrations 10.7, 20.9 and 23.3 mg/m³), 1 h/day for 5 consecutive days. Bone marrow and peripheral blood were collected 24 hours after the last exposure. Hemoglobin adducts detected in the exposed mice resulted from direct binding to globin of MDI and adducts originating from the diamine (MDA) were not observed. No significant increase in the frequency of micronucleated PCEs was detected in the bone marrow or peripheral blood of the mice exposed to MDI. The authors concluded that inhalation of MDI (1 h/day for 5 days), at levels that induced toxic effects (decreased respiratory frequency, decreased body weights and an influx of inflammatory cells into the lung were observed) and formation of MDI-specific adducts in hemoglobin, did not have detectable systemic genotoxic effects in mice, as investigated by the micronucleus assay.

An older mouse micronucleus study has been reported in summary form by JETOC (1982) also gave comparable results. In summary, 4,4’-MDI was dissolved in dry DMSO, mixed with corn oil and administered to mice by intra-peritoneal injection at doses of 32, 80 or 200 mg/kg). The mice were killed 24 hours following final treatment and incidence of polychromatic and normochromatic erythrocytes with micronuclei evaluated. There was no difference in incidence of micronuclei between animals treated with MDI and the untreated control group. It was concluded that 4,4’-MDI did not cause micronuclei, and therefore did not induce in vivo genotoxic effects.

In contrast, a bone marrow micronucleus study by Zhong and Siegel (2000a) exposed Brown Norway rats (males, n = 6) were exposed via inhalation (whole body, 2 at a time) to either 7 or 113 mg/m³ 4,4’-MDI for 1 hour, once a week for 3 weeks with sacrifice 1 week later. A dose-dependent increase in the frequency of micronucleated polychromatic erythrocytes (MN-PCEs) was noted: 1.5- and 4.5-fold increases of micronuclei were reported for the two exposure groups over control (a control group of 4 rats). No difference was found in the ratio of PCEs and NCEs between exposure and control groups, suggesting the absence of bone marrow cytotoxicity. However, this study was deemed to be low reliability (Klimisch 3) since the in vivo protocol used was not standard for a micronucleus test (evaluation at 7-days post-exposure but not before, absence of positive controls) and the findings are not consistent with the existing genotoxicity profile of MDI (or MDA).

Based on a weight-of-evidence, inhalation of 4,4’-MDI by rats does not induce MN formation in systemic target organs despite circulating levels of MDI metabolites and detection of radioactivity in the bone following inhalation of radiolabeled 4,4’-MDI in ADME studies (i.e. Gledhill (2003b; 2005).  Taken together it can be assumed that the bone marrow has been exposed to the 4,4’-MDI via metabolites and conjugates. Given the reactivity of 4,4’-MDI and the toxicokinetic profile described above that 4,4’-MDI is available exclusively to MDI metabolites and conjugates (and not free mMDI), the absence of systemic genotoxicity after an inhalation exposure is not unexpected.

Finally, as mentioned above, the in vivo comet assay conducted by Randazzo (2017) also included analysis of the liver to assess the potential for genetic damage at a distal tissue (e.g. site of first pass metabolic clearance) and the stomach for secondary exposure following mucociliary clearance. No DNA-damage was noted in either tissue at either 1 or 18 hours after exposure at the maximum tolerated dose for respiratory toxicity (12 mg/m3) further indicating that 4,4’-MDI is not systemically genotoxic. 


Human information

Some human exposure studies have reported possible alterations in DNA status (Marczynski et al., 1992; Marczynski et al., 1994a; Marczynski et al., 1994b; Pitarque et al., 2002; Marczynski et al., 2003). Data from these studies have been critically reviewed by Greim in the MAK collection for Occupational Health and Safety (DFG, 2008). In this document, it was concluded that these studies suffer from substantial methodological problems or uncertainties. Therefore, the relevance of these results is considered questionable.

In the EU Risk Assessment, abstracts are cited which reported a slight increase in sister chromatid exchange (SCE) in peripheral lymphocytes or micronuclei in buccal mucosa cells in workers exposed to MDI substances. Evaluation was considered not possible as a result of inadequate documentation. In a publication about female workers in two shoe factories with exposure to a complex mixture of various solvents including MDI substance, an increase in micronuclei in the lymphocytes was described, but no increase in SCE (Pitarque et al., 2002). In another publication by the same research group, no DNA damage in monocytes was detected using the comet assay (Pitarque et al., 1999).

Marczynski et al. (1992) described DNA double strand breaks and MDI substance crosslinks in white blood cells observed in a worker two hours after exposure to MDI substance concentrations of 5 to 20 ppb. However, due to several methodological shortcomings (lack of tests for isolation artefacts, the lack of suitable controls or standards and misinterpretation of the results), this study is not considered valid. Two other insufficiently documented studies by this group report DNA fragmentation in the white blood cells of workers after short-term exposure to MDI substance concentrations of 2 ppb for 15 minutes, 5 ppb for 60 minutes and 10 ppb for 5 minutes (Marczynski et al., 1994a; Marczynski et al., 1994b).

In another study by this research group (Marczynski et al., 2003), 10 patients with workplace-related dyspnoea, who had presented themselves for a presumed occupational medical examination, were exposed to MDI substances in exposure chambers. White blood cells were investigated using gel electrophoresis for low-molecular weight (LMW) DNA fragmentation before and after exposure. There was no difference in DNA fragmentation patterns prior to exposure, with one exception. In one female patient who reacted in the challenge to MDI there was already an increase in DNA fragments before exposure. After exposure, changes in the DNA fragmentation pattern such as observed also after exposing cells to H2O2 were found in 4 of 10 exposed persons. This was regarded as evidence of a change in the intracellular redox steady-state rather than direct DNA damage.

For a comet assay conducted in accordance with present-day requirements, 25 patients with workplace-related asthma were exposed to MDI substance concentrations of 0.05, 0.1, 0.2 and 0.3 mg/m3 in chambers for 30 minutes using the same method as described by Marczynski et al. (2003). No increase in DNA strand breaks was observed in the lymphocytes of the exposed patients in a valid comet assay (Marczynski et al., 2005).

MDA formation has also been considered a potential source of mutagenicity. For example, an observation by (Bolognesi et al., 2001) that “in both humans and in laboratory animals, MDA has been found in urine prior to acid hydrolysis” (Schütze et al., 1995; Sepai et al., 1995a; Sepai et al., 1995b; Sepai et al., 1995c) might be taken to indicate concern for MDA formation in-vivo. However, the phrase “prior to acid hydrolysis” recognizes that when biological fluids (e.g. urine, blood) are heated (100°C) under strong acidic (e.g. 6 M HCL) or alkaline (e.g. 5 M NaOH) conditions, diisocyanate-protein / diamine-protein conjugates as well as acetylated diamines are transformed to free diamine (Rosenberg and Savolainen, 1985; Sepai et al., 1995a; Sepai et al., 1995c; Sabbioni et al., 1997; Sennbro et al., 2003). While there have been reports that low levels of free diamine are detected in the “unhydrolyzed” urine of workers (Schütze et al., 1995; Sepai et al., 1995a)  and experimental animals  (Sepai et al., 1995c) a closer inspection of these reports shows that “unhydrolyzed” means urine treated at room temperature with lower concentrations of acid (e.g. 0.75 M HCL; as in Carbonnelle et al. (1996)) or base (e.g. 1.7 M NaOH; as in Schütze et al. (1995) and Sepai et al., (1995a; 1995c) as part of the analytical workup conditions.

A study by Leinweber (2011) duplicating the “unhydrolyzed” conditions used by Schütze et al. (1995; 1996)  demonstrated that the amide bond (a surrogate for a urea bond) in an arylisocyanate conjugate can be hydrolyzed to release a small but significant fraction of amine. Two other studies have shown that the mild (sometimes erroneously referred to as “unhydrolyzing” conditions) conditions (i.e. 0.1 M NaOH at room temperature) are capable of releasing amine from urethane conjugates. Working with arylisocyanate-amino acid conjugates, Sabbioni et al. (1997) reported that weak alkaline conditions can readily hydrolyze the urethane/thiocarbamate bonds of some isocyanate-amino acid conjugates (e.g. of cysteine, tyrosine).  However, to a small but detectable extent, the urea bond of an isocyanate-serine conjugate can also hydrolyze in these conditions, but not the urea bond in other isocyanate-amino acid conjugates (i.e. lysine, valine, aspartate). Likewise, Pauluhn and Lewalter (2002) used these milder conditions to release MDA from hemoglobin-MDI conjugates, presumably by hydrolyzing urethane and thiocarbamate bonds and possibly urea bonds as well. Although Sennbro et al. (2003) detected low level of diamines (close to the level-of-detection) in the “unhydrolyzed” urine of workers, the authors concluded that it could not be stated unequivocally that free diamines were present in the urine since labile conjugates could be hydrolyzed under the mild alkaline conditions they used for the work-up of the non-heated urine samples.

In summary, exposure to MDI substances can result in low levels of DNA adducts in vivo in some tissues at point of contact (e.g. skin, olfactory epithelium) but not others (e.g. respiratory epithelium or in lung tissue where Type II cell adenomas were observed in bioassays). Given the lack of genotoxicity in these point of contact tissues where MDI reactivity is expected to be highest (comet assay; Randazzo (2017)), it is not surprising there is no evidence of genotoxicity at more distant sites (i.e. negative micronucleus studies; Pauluhn et al. (2001)). This lack of genotoxicity is also seen in vitro when MDA formation is not favored (i.e. as a result of proper solvent selection). Incubation with MDI substances can induce positive Ames assay results in vitro if conditions favor its transformation to MDA, a transformation that does not occur in vivo under normal exposure conditions. It should be reiterated that amine detected in the urine and blood of exposed workers and experimental animals is not due to free amine, but instead is a function of the analytical method (see above). Lack of MDA formation is supported by highly unfavorable reaction kinetics and that in vitro GSH reactivity studies do not detect MDA. Further, biomonitoring and metabolisms studies demonstrate that MDI metabolism under physiological conditions is entirely without amine formation. 

Randazoo (2017) showed in reliable in vivo Comet Assay that aersolized, inhlation MDI at the maximum tolerated concentration did not produce DNA strand breaks at the portal of entry (lung), stomach, or liver.

Endpoint conclusion
Endpoint conclusion:
no adverse effect observed (negative)

Additional information

All substances of the MDI category contain the reactive NCO group on the different constituents that is capable of reaction with biological nucleophiles. The chemical reactivity and toxicokinetic behavior of this NCO on the bioaccessible MDI substances is described in Toxicokinetics section  and forms the basis of the hypothesized MoA for site of contact toxicity but is also the basis for the absence of mutagenicity. In the case of mutagenicity, the hypothesized MoA recognizes that under physiological conditions the MDI substance readily polymerizes at the MDI/aqueous interface forming insoluble polyureas and/or reacts with extracellular biological nucleophiles to form MDI-adducts rendering the free NCO completely unavailable to react with DNA thereby negating this concern. Further, the toxicokinetic and metabolic pathways do not indicate the formation of toxicologically relevant metabolites and is described in detail below.


The reaction mechanisms of isocyanates in the lung fluid also explains why the apparently ‘less reactive’ molecules (i.e. the 2,4’- and 2,2’-MDI isomers) are not more likely to be available intracellularly thus not more potent for DNA reactivity and genetic toxicity. As described above, toxicity of MDI substances is a function of adduct formation, and the extent of toxicity is determined by the dissolution kinetics (i.e. rate of dissolution of MDI-glutathione adducts) and not inherent reactivity. The faster the adducts are dissolved, the more rapidly glutathione is depleted and thus the greater likelihood of a toxic effect. However, as this relates to genetic toxicity, even the more ‘stable’ molecules must still react with nucleophiles (like GSH) prior to being systemically or intracellularly available. The rate of dissolution of the adducts into the aqueous lung fluid which dictates severity of site of contact toxicity is not relevant for genotoxicity (either local or systemic). 

Further, the reactivity of the NCO with glutathione (a pre-requisite for dissolution) is functionally comparable between isomers. This concept is demonstrated in experiments by Wisnewski et al. (2018) showing that the 2,2’- and 2,4’-MDI isomers were found to rapidly react with GSH and demonstrate marked similarities with that previously described for 4,4’-MDI. When fully dissolved, at physiological pH and temperature, 2,2’- 2,4’- and 4,4’-MDI-GSH reaction products are primarily bis(GSH)-MDI and form within minutes. The initial rate of bis(GSH)-MDI formation (assuming pseudo first order kinetics) with 2,2’-, 2,4’- and 4,4’-MDI were indistinguishable. Diamine hydrolysis products (MDA) were below the limit of detection (0.03 µM) for all MDI isomer reactions. These experiments show that regardless of the ‘reactivity’ of the molecule, they will equally form GSH adducts and hence the NCO will be unavailable for cellular transport and subsequent mutagenicity. 

The genotoxic potential of MDI substances has been investigated extensively, both in vitro and in vivo and while early in vitro mutagenicity studies were positive, further experiments demonstrated that these results reflect the properties of hydrolysis products (i.e. diamine) formed under specific artificial assay conditions (aprotic solvent) and are not indicative of genotoxic potential of the parent compound under physiological conditions (Herbold et al., 1998; EC, 2005; DFG, 2008).


When assessing the reliability and relevance of the available in vitro genotoxicity data it must be recognized that due to the lack of solubility of unreacted MDI substances in the aqueous media in guideline in vitro genotoxicity studies, the test material must be dissolved in a solvent to enhance dissolution.  In the case of MDI substances and genotoxicity studies, the dielectric properties of particular solvents have been shown to affect the subsequent reactive properties in the experimental solution. This effect stems from the ability of the solvent to shield - or in other words stabilize - electrical charge. Since electronic effects play a significant role in many diisocyanate reactions, including the one with water, solvents with a higher dielectric constant may accelerate chemical reactions involving partial electric charges (e.g. of the reactants or transition states), even without participating in the reaction themselves. There are examples in literature [e.g. Ekberg and Nilsson (1976)] that particularly demonstrate this effect for the hydrolysis of diisocyanates. The work by Herbold et al. (1998) represents the textbook example of this phenomenon, demonstrating that using dimethyl sulfoxide (DMSO) as a solvent modifies the outcome of the Ames test for monomeric MDI substances by accelerating hydrolysis and formation of the respective diamines. More recently, extensive data have been generated on the hydrolysis of MDI substances in a variety of aprotic polar solvents that can provide further insight (Engemann and Pirkl, 2017; Nuthmann et al., 2017).  Thus, it can be concluded that MDA formation is only a function of experimental conditions and when avoided, in vitro mutagenicity tests are negative. Subsequent in vitro studies using solvents that did not favor hydrolysis to amines were consistently negative. Additional in vitro work by Wisnewski et al. (2019a) also did not detect any MDA formation in aqueous GSH solution with any mMDI isomer. 

These results are supported by negative in vivo studies such as a comet and micronucleus (Pauluhn et al., 2001; Lindberg et al., 2011; Randazzo, 2017) and provide convincing evidence for a lack of genotoxicity activity both at the site of contact and systemically. 

Justification for classification or non-classification

Not classified as mutagenic according to CLP.