Registration Dossier

Data platform availability banner - registered substances factsheets

Please be aware that this old REACH registration data factsheet is no longer maintained; it remains frozen as of 19th May 2023.

The new ECHA CHEM database has been released by ECHA, and it now contains all REACH registration data. There are more details on the transition of ECHA's published data to ECHA CHEM here.

Diss Factsheets

Administrative data

Link to relevant study record(s)

Description of key information

Key value for chemical safety assessment

Additional 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 of key parameters (i.e. mMDI content and NCO value) within the MDI category that determine the hypothesized the extremes 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.


With respect to toxicokinetics, all MDI substances are highly reactive and exhibit strong reactivity with biological nucleophiles which are directly related to the toxicity mode of action. Accordingly, the toxicokinetic behavior of the substances of the MDI category at the extracellular site of contact is directly responsible its substance toxicity. Thus, the chemical and toxicokinetic profile of substances of the MDI category is predictive of the toxicity response for all endpoints. This data provides high confidence that all substance of the MDI category will demonstrate a consistent toxicological trend and no further testing is necessary.   


 ECHA Guidance on Information Requirements and Chemical Safety Assessment Chapter R.7a: Endpoint Specific Guidance (Version 3.0, August 2014; page 20 “Selection of the appropriate route of exposure for toxicity testing“) is indicating that “the overall objective of toxicity testing is to determine the potential hazard of the test substance to human beings”. It is stressed that “toxicity data obtained using the appropriate route of exposure are preferred” and that “route-to-route extrapolation should be considered on a case-by-case basis and may introduce additional uncertainties”. The evaluation of toxicokinetic information should be taken into account to address such uncertainties. For reactive substances like diisocyanates significant toxicokinetic differences can be demonstrated, e.g. with respect to primary reaction products at the portal of entry (oral, dermal, inhalation). Since the most relevant route of exposure is inhalation hazard characterization shall preferably be performed via inhalation. 


Oral exposure: 


While no oral toxicokinetic studies of oral MDI exposure are available, the behavior of MDI substances in the stomach following gavage is expected to be similar to another aromatic diisocyanate. The US NIEHS contracted studies on the aromatic diisocyanate toluene diisocyanate (TDI), demonstrating that TDI polymerized in the acid environment of the stomach to solid polyurea before absorption into the body (Jeffcoat et al., 1985; Dieter et al., 1990; Timchalk et al., 1994). The formation of ureas is the typical reaction of MDI in aqueous environments, and reaction rates for both MDI and TDI have been shown to be similar (Allport et al., 2003). Due to the common reactivity of the isocyanate group and the relatively similar rate constants for the two substances, it is expected that substances of the MDI category will react similarly. This is corroborated by the results from an inhalation study with 4,4’-MDI reported by Gledhill (Gledhill, 2003a; Gledhill, 2003b). Following an inhalation exposure of rats to radiolabeled MDI, 79 % of the dose was excreted in feces. The fecal excretion in these animals was considered primarily due to ingestion of radioactivity from grooming and ingestion of deposited material from the nasopharyngeal region via the mucociliary escalator, i.e. not following systemic absorption. The fecal radioactivity was tentatively identified as mixed molecular weight polyureas derived from MDI polymerization. Diamine was not present as a metabolite in urine or fecal extracts. This shows that intake of MDI substances to the digestive tract by grooming or by the mucociliary escalator following inhalation results in formation of polyureas without detectable diamine formation. For MDI substances and diisocyanates in general the oral bolus gavage dosing is inappropriate for toxicological studies and risk assessment, due to formation of hydrolysis and polyurea products in the acidic stomach environment that would be less pronounced via more relevant inhalation or dermal routes.  


 Dermal exposure: 


Two studies on the dermal absorption and distribution of radioactive labelled 4,4’-MDI in rats are available (Vock and Lutz, 1997; Leibold et al., 1999). Vock and Lutz (1997) was designed primarily to investigate possible binding of dermally applied MDI to cellular DNA. Female Wistar rats were treated topically with [14C]4,4’-MDI in dried acetone on the clipped back, onto an area of about 3x3 cm, 1 or 3 days after clipping, without occlusion. Although the EU Risk Assessment (EC, 2005) suggested that the study appeared reliable, the very low recoveries reported by Vock and Lutz (1997) and the reported opportunity for grooming of the unoccluded application site by animals housed together (evidenced by high radioactivity on the tongue) calls into question the validity of the excretion data, particularly for feces. In contrast, in the specifically designed study reported by Leibold et al. (1999), published by Hoffmann et al. (2010), animals were housed singly, the site of administration was protected with a silicone ring, a nylon mesh gauze and a porous bandage, to prevent contact by grooming and to minimize contact of the applied dose by the dressing. Nevertheless, there was high recovery from the dressing which would also suggest removal and ingestion of radioactivity from the application site was easily possible in the study of Vock and Lutz (1997).


Leibold et al. (1999), published by Hoffmann et al. (2010) investigated the absorption, distribution and excretion of radioactivity in groups of 4 male Wistar rats following a single dermal or intradermal administration of [14C]4,4’-MDI (4,4’-Methylenebis[ring-U-14C]-phenyl-isocyanate). These results indicate that dermal application of 4,4’-MDI, in acetone solvent results in low (less than 1 %) penetration of MDI or metabolites to the systemic circulation, with appreciable levels remaining in skin (1-7 %), most probably as covalent adducts with proteins such as keratin (Wisnewski et al., 2000; Hulst et al., 2015). Consequently, a dermal uptake of 1 % (Leibold et al., 1999) is used to calculate the body burden in the dermal exposure assessments. Note that, based on the known high reactivity of MDI with nucleophiles such as glutathione (GSH) (Wisnewski et al., 2019a) and the subsequent formation of protein adducts in skin and plasma (Kennedy and Brown, 1998; Wisnewski et al., 2011a; Hulst et al., 2015), it is expected that absorbed radioactivity from MDI dermal exposures consists of GSH and/or protein adducts, with little to no free diisocyanate.  However, even <1 % penetration into the skin may be sufficient to initiate immunological processes (discussed in more detail below in

To compare the relative behavior of all substances in the MDI category, in silico predictions of dermal absorption and bioavailability in humans were made with GastroPlus®, a physiologically relevant model of dermal penetration which incorporates well-validated QSARs for lipophilicity and water solubility and includes optional metabolic clearance to non-absorbable adducts (i.e. GSH) at the site of application. Here, absorption is defined as the percent of the applied dose that is taken up into the skin.  Bioavailability is the percent of the dose that, once absorbed, passes through the skin into the systemic circulation and then other tissues.  A comparison of GastroPlus® predictions of dermal uptake and/or systemic exposure in rats for 4,4’-MDI isomer in rat to previously reported in vivo data demonstrated that the model was predictive (modeled uptake and bioavailability of 4,4’-MDI was 55 % and 0.3 %, respectively, versus measured values of 26-56 % (Bartels, 2021) and <1 %, respectively (Leibold et al., 1999)).  The results of the GastroPlus® dermal absorption modeling show that non-monomeric MDI category constituents would have lower uptake into the skin and bioavailability, than the three MDI isomers (mMDI > 3-ring oMDI > modified MDI constituents). These results are consistent regardless of vehicle solvent, test substance concentration, exposure time, or metabolic clearance rates within the skin compartments (Bartels, 2021).  Relative differences in predicted absorption of the mMDI isomers and the rest of the MDI substance category further supports the hypothesis that the lowest molecular weight species are the most bioavailable and therefore can be considered the worst-case. Since systemic toxicity has never been observed in any study or study type, it is logical that the less bioavailable non-monomeric constituents would be even less likely to elicit systemic effects in the unlikely event they even become bioaccessible (i.e. reactively dissolved by GSH adduct formation). 


 Inhalation exposure: 


Distribution and regional deposition

With respect to inhalation exposure, there are good and reliable data regarding absorption/distribution/excretion of substances of the MDI category in experimental animals. In a two-part inhalation metabolism/toxicokinetics/distribution study performed by (Gledhill, 2003a; Gledhill, 2003b), tissue distribution of radioactivity after nose-only inhalation exposure to radiolabeled 4,4’-MDI, when considered with excretion data, imply that tissue residues resulted from absorption of radioactive material primarily from the gastrointestinal (GI) tract after ingestion of an inhaled dose.  Further, pulmonary absorption of radioactivity deposited in the lungs accounts for only a minor portion of the administered dose. Similar results were obtained in an older study by Centre d`Etudes (1977) where the fecal elimination of 4,4-MDI was greater than the urinary elimination. In both studies, radioactivity was widely distributed with the respiratory and excretory organs containing the highest concentrations. The highest concentration of radioactivity was present in the respiratory nasal tissue (Gledhill, 2003b). Due to grooming and respiratory clearance a combined oral/inhaled exposure existed in the study of Gledhill (2003b) with 25 to 32 % of the applied dose entered the systemic circulation as adducts with nucleophiles such as GSH, based on the known high reactivity of mMDI with this endogenous tripeptide (Wisnewski et al., 2019a).  There is no unreacted mMDI systemically available (Gledhill et al., 2002). 


Taken together, these data support that following exposure, most inhaled test material or reaction products (polymerized insoluble particles and other insoluble adducts) are removed via mucociliary transport and swallowed with a small portion absorbed via solubilized macromolecular adducts via lung into blood stream. This is consistent with the Hydrolysis concept presented in Section 2.2 describing what happens to substances of the MDI category in an aqueous environment. While the addition of electrophilic nucleophiles to the aqueous matrix somewhat changes this dynamic (and will be discussed in more detail below), the basic pathway described there dominates and is supported by the Gledhill experiments. In other words, in the aqueous matrix of the respiratory tract or GI tract, isocyanate functionalities mostly form toxicologically inert and insoluble complex oligoureas and polyureas, which are excreted via the feces.


Alveolar Dissolution and Conjugation

The remaining respirable MDI substances that reach the lower respiratory tract must be transported through the aqueous matrix of the airway lining fluid before it can interact with and be absorbed by the alveolus. While many of the principles described previously regarding the behavior of MDI substances in an aqueous matrix may be relevant here (see hydrolysis), protective components of the lung fluid result in some critical differences in reaction profile in airway lining fluid versus in water. 


The airway and alveolar lining fluid is an aqueous buffer containing surfactant (phospholipids), protein, and various antioxidants (e.g. glutathione) which play an important role not only in maintaining proper surface tension and airway stabilization, but also pulmonary defense mechanisms each having an important role in the response pathway of MDI substances. Binding of the isocyanate functional group with these biomolecules in the lung fluid is the critical step in the detoxification of MDI, the kinetics of which are described here. The resultant toxicological mode of action that is consequence of these reactions will be described in the discussion of the relevant endpoints.



Glutathione (GSH) is a very potent nucleophile and constitutes the main anti-oxidant in the lung with the extra-cellular epithelial lining fluid containing 140-fold higher concentrations then in the plasma (Cantin et al., 1987). In lung exposed to MDI substances, formation of GSH-adducts (Hu et al., 1989; Day et al., 1997; Karol et al., 1997; Fleischel et al., 2009; Spencer et al., 2013; Ferreira et al., 2018; Wisnewski, 2018) is the preferred reaction path in the airway fluid due to the highly reactive nature of the thiol group on the GSH with the isocyanate group in MDI. As discussed in Chapter 2.1.2, solubility of MDI substances in aqueous media is extremely low. In consequence, the reaction is usually heterogeneous, meaning the reaction is biphasic at the interface between the aqueous and MDI particle phase. At this interface one of the isocyanate groups will react with the thiol group on the glutathione to form S-adducts increasing the solubility of the MDI molecule, allowing the remaining isocyanate groups to be more available to react (for more detail, see Plehiers et al. (2019). This ‘reactive dissolution’ of the molecule drives ‘bioaccessibility’ which is the ability of a substance to be released from the MDI particle matrix and into the lung fluid which ultimately determines the toxic effects.  

Following the partial reactive dissolution to form the mono-GSH-MDI adduct, the remainder of the particle will typically follow the reaction path to form oligo- and poly-ureas (Yakabe, 1994; Yakabe et al., 1994; Allport et al., 2003).  As solids, they are typically phagocytized (Pauluhn et al., 1999b; Pauluhn, 2002a) transferred via the muco-ciliary elevator to the gastro-intestinal tract, and ultimately excreted via feces with a smaller portion eliminated via the lymph. Ureas are stable in the stomach environment and do not decompose to form amines (Mráz and Bousková, 1999; Hettick et al., 2009). This is consistent with the high level of fecal excretion observed by (Gledhill, 2003a) with inhalation-exposed rats. When higher molecular weight constituents of the MDI category are present (e.g. as glycol adducts in ‘MDI and its reaction products with glycols’ subgroup), the ability of GSH to bring these constituents into solution will be reduced, since they will be inherently more hydrophobic and less soluble in water (Muuronen et al., 2018).

If sufficient GSH is present, the second isocyanate group in the above-mentioned MDI mono-adduct can also react with another GSH thiol to generate the MDI-GSH bis-adduct  (Reisser et al., 2002; Wisnewski et al., 2010; Wisnewski and Liu, 2013; Wisnewski, 2018; Wisnewski et al., 2019a). GSH S-adducts, which are thiocarbamates, are stable in acidic environment (e.g. in the stomach) (Stark, 1964; Sabbioni et al., 1997; Reisser et al., 2002; Gledhill, 2003a), a property which is commonly used to stabilize adducts (Wisnewski, 2018; Wisnewski et al., 2019a). However, in a neutral or alkaline environment with limited GSH availability (e.g. in the blood stream), S-adduct formation is reversible (Stark, 1964; Reisser et al., 2002; Mormann et al., 2008)). Under these conditions and in the presence of proteins, the transcarbamoylation reaction (i.e. MDI/GSH à MDI/protein) is dominant over potential hydrolysis reactions of the S-adducts (Reisser et al., 2002; Chipinda et al., 2006; Mormann et al., 2008). It is in the form of these solubilized adducts (MDI/GSH and MDI/protein adducts) that the MDI substance is translocated from the extracellular matrix into the blood stream. Further, these protein (primarily albumin and hemoglobin which are not toxicologically active) adducts are the primary form of MDI substances found in the bloodstream and subject for further metabolism and excretion (further discussed below). The transcarbamylation reaction creates a urea bond, therefore the MDI/protein adducts are not hydrolytically active.

The transcarbamoylation reaction does not require the “release” of the isocyanate moiety from the S-adduct (Mormann et al., 2008; Fleischel et al., 2009). This view is supported by modern quantum-chemical calculations on the similar reaction between an amine and a carbamate group (Ilieva et al., 2013). The protein adducts formed by the transcarbamoylation reaction are substituted N-aryl-ureas, which are stable to hydrolysis over a very broad pH range (Audu and Heyn, 1988; Salvestrini et al., 2002; Sendijarevic and Sendijarevic, 2003; Sendijarevic et al., 2004; Luna et al., 2014). All studies that have investigated the potential formation of the aromatic diamines by hydrolysis of S-adducts of GSH or cysteine, either in vitro or in vivo, have failed to detect the diamine (Day et al., 1997; Reisser et al., 2002; Mormann et al., 2008; Wisnewski, 2018; Wisnewski et al., 2019a). Hence, formation of “free” diamine is not expected in the airway lining fluid or in the blood stream. This is supported by chemical kinetics (described above), in vitro studies (Wisnewski et al., 2019a), by inhalation toxicokinetic or cell studies (Timchalk et al., 1992; Timchalk et al., 1994; Zhong and Siegel, 2000b; Gledhill, 2003b; Gledhill, 2003a; Lindberg et al., 2011), and by human or animal biomonitoring studies (Sabbioni et al., 2000; Pauluhn and Lewalter, 2002; Pauluhn et al., 2006). Results from a study on MDI-exposed worker in a plant that manufactured MDI based rigid polyurethane products showed that there are essentially no low molecular weight MDI derivatives that can be hydrolysed to MDA in plasma and serum albumin is the major protein in plasma that form adducts in vivo with unspecified MDI (Johannesson et al., 2004).

Pulmonary Surfactant

While the thiol group on glutathione is the primary reactant for isocyanate, there is also evidence that MDI will interact with pulmonary surfactant present in the lung fluid. While the primary role of the surfactant is to reduce surface tension in the alveolus, it can also support non-specific defense mechanisms in the lung via antioxidant activities (Hamm et al., 1996).  Pulmonary surfactant mainly consists of phospholipids (approximately 90 %) and four specific surfactant proteins (SP-A, SP-B, SP-C, SP-D). The most prevalent phospholipid is phosphatidylcholine (>80 %; within that group 30-60 % dipalmitoylaphosphatidylcholine) among other phospholipids. The hydrophobic side of the phospholipids is the most likely to interact with the hydrophobic MDI substances. It would be expected that the phospholipids mainly act as a dispersant for MDI particles (much like any surfactant), since the hydrophobic side in essence is devoid of nucleophilic reactive centers. The MDI-associated phospholipids are then phagocytized by macrophages in the extracellular matrix as evidenced by an increase in MDI-associated polar phospholipids in ‘foamy’ macrophages[1] in rats exposed to pMDI (Kilgour et al., 2002) and 4,4’-MDI/DPG/HMWP (Ma-Hock, 2021) via inhalation.

Effects of Solubility on GSH reactions

The toxicological effects of MDI substances in the lower respiratory tract is driven primarily by the reaction of the isocyanate group with biological molecules to form MDI-adducts (especially with GSH).  Depletion of the GSH in the lung fluid results in a cascade of events that includes impaired macrophage clearance, oxidative stress, and inflammation ultimately leading to dysfunction of the surfactant and membranes (Pauluhn, 2011a). It is the extent and rate of this GSH depletion that determines the severity of effect. However, as discussed above, the relative reactive dissolution of the NCO presenting molecule in the mixture (i.e. the rate of the nucleophile pulling the isocyanate into solution) is the limiting step in the adduct formation and not the relative reactivity differences of the NCO moiety on the molecules. A study by Wisnewski et al. (2018) showed that the 2,2’- and 2,4’-MDI isomers were found to react rapidly with GSH and demonstrate marked similarities with that previously described for 4,4’-MDI. When the MDI dissolved in 1 % acetone, 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. However, when the same experiments are conducted at 10 % acetone (to increase dissolution) the rate of bis(GSH)-MDI markedly increased (approximately 10 fold) and reached maximal levels within 1 minute. The initial rate of bis(GSH)-MDI formation (assuming pseudo first order kinetics) with 2,2’-, 2,4’- and 4,4’-MDI in both experiments were indistinguishable. Hydrolysis products (MDA) were below the limit of detection (0.03 µM) for 2,2’-, 2,4’-, and 4,4’-MDI.

This observation is extended by a series of similar in vitro experiments with other modified MDI substances representing each of the sub-groups of the MDI category (Zhang et al., 2021) and including the boundary substances. In these experiments, 4,4’-MDI, pMDI, 4,4’-MDI/4,4’-MDI homopolymer, and 4,4’-MDI/DPG/HMWP were solubilized in solvent and added to the aqueous media containing radiolabeled GSH to qualitatively and quantitatively analyze in vitro adduct formation. A high amount of solvent was used with the intent to dissolve as much of the test substances as possible. Overall, results demonstrated that mMDI constituents (i.e. the most reactive soluble) in the MDI-based test materials react very rapidly (essentially instantaneously) to form adducts with glutathione (GSH). The formation of these GSH adducts with mMDI and their subsequent transcarbamoylation with amine functionalities proceed at similar rates for the four tested materials. The relative formation of various types of adducts (bis-, mono-, and cyclic) is consistent with prior experiments (Wisnewski, 2018), and depends on isomer of mMDI, but not on the presence of the higher molecular weight constituents such as oligomers, condensation adducts or glycol adducts in the test materials. 

Conversely, no GSH adducts other than those with mMDI could be identified from radio-chromatographic separations of reaction mixtures containing these higher molecular weight constituents and radio-labeled GSH. The reason for the absence of GSH adducts with less soluble non-monomeric MDI constituents was due to their precipitation prior to addition of GSH to the test medium. Additional testing demonstrated that a GSH adduct of a three-ring MDI oligomer could be generated and detected under homogeneous (i.e. pre-solubilized in solvent) reaction conditions. Under heterogeneous reaction conditions (i.e. no solvent), the availability of such adducts may be further limited by solubility.

As will be discussed in more detail in the subsequent chapters for the individual endpoints, the key event in toxicity of MDI substances is the extracellular reaction of NCO with biological molecules (primarily GSH).  As reactive dissolution is determined by hydrophobicity (i.e. octanol-water partition coefficient), the least hydrophobic constituents will be most readily solubilized as a GSH mono-adduct, and subsequently further depleting protective nucleophiles as the second NCO reacts with another GSH to form the bis-MDI/GSH adduct. In modified MDI substances (e.g. belonging to ‘Oligomeric MDI’, ‘MDI and its reaction products with glycols’, ‘MDI and its condensation products’, etc. subgroups), the least hydrophobic constituents (e.g. mMDI isomers and three-ring MDI oligomer) with the lowest molecular weight are most easily solubilized as adducts, and as a result, they drive the reaction with endogenous nucleophiles. The higher molecular weight constituents of the MDI substances (e.g. more hydrophobic glycol adducts, condensation products, etc.) are unlikely to become biologically accessible (i.e. solubilized) even if the NCO’s on these molecules have an opportunity to react with GSH to form adducts. Therefore, they can be considered not toxicologically relevant. 

This is confirmed in vivo by Pauluhn and Lewalter (2002) demonstrating that inhalation exposure with ‘pMDI’ (50 % of mMDI, 34 % of three-ring constituent, ratio about 3/2 mMDI / pMDI) in rats subsequent adducts detected (measured as two- or three-core MDA after hydrolysis) were present in a ratio of approximately 10/1 (mMDI / pMDI). In other words, the presence of three-core MDA was seven times lower than expected from the composition of the test substance. Taken together, increasing the molecular weight of modified MDI substances reduces their availability to react with nucleophiles and elicit a toxicological effect. As the 3-ring oligomer is already significantly reduced and it is still the most available of the non-monomeric constituents, even larger constituents are less available and therefore do not need to be considered for the prediction of effects.  Predictions for all endpoints can be based on the amount of bioaccessible NCO as described in the hypothesis. 

Other functional groups

It should also be noted that other functional groups may be present in the substances of the MDI category, besides the isocyanate group including dimer (uretdione) and condensation products (isocyanurate, uretonimine). However, the reactions of the NCO groups at the respective ring structures proceed much slower than the reaction of the NCO group on mMDI. Further, the ring structures are either resistant to hydrolysis or yield stable ureas. 

In the ‘MDI and its reaction products with glycols’ subgroup (i.e. prepolymers subgroup), the aromatic diisocyanates are modified with aliphatic polyols, which results in the formation of alkyl-N-aryl-urethane bonds. The kinetic data on urethane hydrolysis published by Christenson (1964) show that the reaction takes place with the hydroxyl ion (OH-). At pH 8, only 0.001 % conversion (hydrolysis) of the urethane bond is expected to occur in one day. At lower pH, the reaction is even slower. Similar conclusions are arrived at from the results of Williams (1973). Under neutral conditions, an occasional ether bond between polyol entities is even more resistant to hydrolysis.

In water extractability experiments with MDI-based prepolymer substances at a loading of 10,000 mg/L, Neuland (2017) detected no dissolved species such as diols or other hydroxyl-terminated molecules that would indicate hydrolysis of the urethane bond.

Potent nucleophiles such as amines (proteins) and thiolates (GSH) could react with the urethane groups, in which case the isocyanate moiety would be converted into a protein- or a GSH-adduct respectively. Therefore, any effects of the modified aromatic diisocyanate moiety with respect to aromatic diamine formation cannot exceed those of the mMDI.

Muuronen et al. (2018) have shown by quantum-chemical calculations that the reactivity of the free NCO-groups in the prepolymer modified substances is not higher than the reactivity of the NCO-groups in 4,4’-MDI (see Section This is supported by theory, since a urethane group has a slight e-donating effect which reduces NCO reactivity.


An MDI substance is carried from the lung fluid into the lungs and systemic circulation in the form of glutathione and proteins adducts. There is no evidence of unreacted mMDI or diamine being systemically available (Day et al., 1997; Reisser et al., 2002; Mormann et al., 2008) and in vitro experiments have not been able to detect free amine when the mMDI is dissolved in a GSH solution (Wisnewski et al., 2019a; Zhang et al., 2021).

Systemic toxicity has not been observed in any in vivo study with either 4,4’-MDI, pMDI or 4,4’-MDI/DPG/HMWP which is attributed to the lack of systemic bioavailability of the reactive isocyanate functional group. Further, as demonstrated above by Bartels (2021), higher molecular weight constituents of modified MDI substances are likely unavailable for absorption (due to high octanol-water partition coefficient) indicating an even further reduced bioavailability. Similar to the studies described above for dermal absorption, in silico predictions using mechanistically-based pharmacokinetic software, GastroPlus® were conducted to model inhalation absorption of mMDI, the major MDI-GSH conjugates, as well as the modified MDI prepolymers and demonstrated the relative systemic uptake of these substances. Metabolic rate data generated in the in vitro studies above were used to refine the models to more accurately predict behavior. 

Like for dermal absorption, the behavior of higher molecular weight constituents of the category substances relative to that of mMDI has been investigated in silico using the GastroPlus® model (Bartels, 2021). It should be noted that GastroPlus® can only model chemical species, not entire substances (see report for the chemical constituents modeled).

With respect to modeling inhalation exposure, the GastroPlus® model includes barriers related to molecular size and hydrophobicity, as well as the clearance mechanism via the mucociliary escalator and gastro-intestinal tract. GastroPlus®, however, does not by default model reactive detoxification mechanisms like adduct formation and the conversion of MDI-based particles into inert and stable poly-ureas.  These two processes were therefore incorporated into the pulmonary tissues and GI tract compartments of the model, respectively, to account for presystemic loss of parent compounds prior to absorption, with rates for both processes derived from the prior in vivo inhalation study (Gledhill et al., 2005).  The model was found to provide a conservative prediction of systemic uptake to 4,4’-MDI (48% via pulmonary + GI tract), vs. 25-32% measured by Gledhill et al. (2005).  It is important to note, however, that GastroPlus® predictions of systemic exposure are based estimates of systemic uptake of parent 4,4’-MDI vs. measurements of 14C-labeled parent compound and/or metabolites in the in vivo study.  The modeling undertaking therefore represents an overestimate of absorption in absolute terms, both via the lung and the gastrointestinal tract. The results should only be used to compare constituents in relative terms. Additionally, modeled absorption of monomeric MDI and the mono- and bis-GSH adducts is comparable and suggests that the relative results described above for unreacted substances would also be applicable to any potential non-monomeric adducts. It should be reiterated that while adduct formation doesn’t affect lung or GI absorption, it does affect solubility for which molecular weight (and octanol-water partition coefficient) plays a significant role.

Within these limitations, the following trends can be derived from the modeling results:

  • Uptake in the lung is highest for monomeric MDI isomers;

  • For constituents up to a molecular weight of 750 g/mol, uptake in the lung is reduced and more material is excreted via the gastrointestinal tract;

  • For constituents with a molecular weight above 750 g/mol, absorption by either route (lung vs. GI) is greatly reduced.


Relative differences in predicted systemic uptake of the mMDI isomers and the rest of the MDI substance category further supports the hypothesis that the lowest molecular weight species are the most bioavailable and therefore can be considered the worst-case. Since systemic toxicity has never been observed in any study or study type, it is logical that the less bioavailable non-monomeric constituents would be even less likely to elicit systemic effects in the unlikely event they even become bioaccessible (i.e. reactively dissolved by GSH adduct formation).  

Metabolism and Excretion

As described above, the fraction of mMDI absorbed by the lung is exclusively via MDI adducts, consisting primarily of soluble low molecular weight MDI-GSH adducts. In the slightly basic and low GSH environment of the blood, a significant amount of the MDI-GSH adducts are transcarbamoylated with systemic proteins to form stable albumin and hemoglobin adducts.  Transcarbamoylation reactions do not proceed through the formation of MDA. These protein adducts are eventually eliminated via urine without further modification of the isocyanate moiety. Polymerized ureas (e.g. higher molecular weight, low solubility) are removed via mucociliary clearance (Gledhill, 2003b; Gledhill, 2003a; Muuronen et al., 2018)  and excreted via the GI tract. 

Alternatively, a minor portion of absorbed MDI-GSH adducts are metabolized via deconstruction into Cys-based adducts (Wisnewski et al., 2016; Wisnewski et al., 2019b), followed by acetylation of the N-atom of the Cys-residue to form mercapturic acids (Bruggeman et al., 1986; Hinchman et al., 1991; Bartels et al., 2009). Occasional amino-groups (e.g. in or from diisocyanate mono-GSH-adducts) can be acetylated and there with detoxified in the liver (Hinchman et al., 1991; Sabbioni et al., 2000)).  No free MDA was detected in the available studies, and the metabolic pathway proposed does not require MDA as an intermediate. Formation of the diamine would require a bis-glutathione-MDI conjugate to release both glutathione moieties simultaneously, with neither unmasked isocyanate group having a readily available reaction partner (e.g. albumin) and so both being simultaneously hydrolyzed. While this may theoretically happen, in reality the main process is for unmasking of one isocyanate group at a time and its reaction with protein.

Biomonitoring and metabolism studies have consistently failed to identify free diamine from diisocyanate metabolism (Timchalk et al., 1994; Gledhill, 2003b; Gledhill, 2003a Lindberg, 2011 #27802). Although Sepai et al. (1995a), amongst others, reported free diamine from diisocyanate dosed animals, it is known that the mild room temperature weak alkaline conditions used for sample workup can actually release a small portion of amine from urea bonds (Leinweber, 2011) and certainly amine from urethane conjugates (Sabbioni et al., 1997). In urine and bile four main metabolites were identified (N-acetyl-4,4´-diaminophenylmethane, diacetyl-4,4´-diaminophenylmethane, N,N´-diacetyl-4,4´-diaminobenzhydrol and N,N´-diacetyl-4,4´-diaminobenzophenone) which arise from progressive acetylation and oxidation of the test material. None of these specified low molecular weight metabolites were found in feces. The radiolabeled components of feces and bile were tentatively identified as oligo-ureas of MDI, derived from ingestion of MDI by grooming or via the mucociliary escalator of inhaled MDI from the respiratory tract. 

As part of the in vivo testing program, a guideline inhalation OECD 412 (28-day repeated dose study in rodents) was conducted on the boundary substance 4,4’-MDI/DPG/HMWP that has been identified by mMDI concentration and NCO content (which is a function of molecular weight) as the boundary substance in the MDI category. As part of this study, urine was collected from the test animals at sacrifice and analyzed for total MDA as an indicator of exposure.  Results are unavailable at the time of submission but will be included in future updates.   

No data is available on the metabolic pathways of potentially absorbed fraction of non-monomeric MDI constituents. However, due to the reactive nature of the isocyanate group with the thiol groups on the isocyanate, it is expected that these constituents will conjugate with glutathione to the extent that it is available (limited by octanol-water partition coefficient), and therefore the subsequent metabolic pathways are likely comparable. However, as stated previously, even glycol adducts of the MDI substances that have reacted with GHS have very high predicted octanol-water partition coefficient and are not expected to be solubilized and absorbed.

 Application of biomonitoring to humans: 

The methods to identify and quantify MDI-adducts to plasma proteins particularly albumin (Alb) and to haemoglobin (Hb) have now been applied to biological monitoring, particularly useful since the amount of adducts would be indicative of an integrated exposure. In addition, these adducts are specific for MDI exposure. However the total amount of the analyte MDA, retrievable from Hb-adducts and urinary precursors, accounts for less than 0.5 % of the applied dose of MDI (Pauluhn et al, 2006), and the lack of linearity of biomarker to exposure dose makes uncertain the extrapolation from the yield of biomarkers in urine or blood towards inhalative MDI exposure. The use of protein adducts for biomonitoring appears to overcome some of these difficulties and benefits from specificity of the analyte. To date no diisocyanate specific urinary biomarker has been identified. For blood, the MDI-specific methods developed are: Hb-conjugate derived hydantoin, Alb-lysine conjugates and peptide conjugates. On the basis of limits of detection, the Hb-hydantoin method is most sensitive compared to the Alb-lysine method which in turn is more sensitive than the signature peptide method. The Hb-hydantoin method covers a longer period of exposure than the Alb-lysine, due to the longer half-life of the erythrocyte compared to serum albumin.

Biomonitoring for exposure to diisocyanates typically looks to assay derivatives (diamines) following hydrolysis of biological fluids and is routinely employed to measure occupational exposure to MDI and other diisocyanates. For diamine analysis, samples of urine are typically used, although blood samples can also be used. However, these markers are not specific for the diisocyanate exposure. Schütze et al (1995) was the first report of biomonitoring of MDI exposed workers for biomarkers in urine and in blood as Hb adducts. The report indicated that the urine biomarkers (after acid or base hydrolysis) reflected recent exposures whereas the Hb biomarkers did not necessarily correlate with the urine biomarkers, and were considered to reflect overall exposures over a longer term. The hydrolysis methods and conditions used release differing amounts of the diamine analyte (Sennbro et al 2003; Skarping et al 1994). Hydrolysis to release MDA as the analyte is not specific to MDI, for example MDA exposure will also give MDA as analyte after hydrolysis.  Hydrolysis analytes are at low concentrations and proportionally little of the dose is in urine or blood, and that there is no standardised method for measuring biomarkers in hydrolyzed urine. Investigation of diisocyanate specific biomarkers has focused on the conjugated molecules in blood. Typically, conjugates with Hb or albumin (Alb) have been assessed, and recently there has been progress in application of experiments in animals to biomonitoring of human exposures to MDI.

Gries and Leng (2013) have successfully developed a method to detect the haemoglobin adduct 5-isopropyl-3-[4-(4-aminobenzyl)phenyl]hydantoin (ABP-Val-Hyd) in human blood. They reported values up to 5.2 ng ABP-Val-Hyd/g globin (16 pmol/g) in blood samples of workers exposed to MDI. The analytical method focused on optimal sensitivity and selectivity, using gas chromatography–high resolution mass spectrometry. A detection limit of 0.02 ng ABP-Val-Hyd/g globin (0.062 pmol/g) was achieved, with a linear detection range up to 10 ngABP-Val-Hyd/g globin (31 pmol/g). The authors conclude the method described is optimized for screening studies of the human population.

Others have used the MDI-Alb conjugate to derive specific biomarkers. Sabbioni et al (2010) used post-shift blood and urine samples from potentially MDI-exposed workers. Alb was isolated and subject to a pronase digestion and the digest treated to isolate MDI-lysine and acetyl-MDI-lysine conjugates. Quantification was by LC-MS/MS. The MDI-Lysine conjugate was identified in about 63% of workers, the range being 0 – 900 fmol/mg Alb. The authors found the method to detect exposure in more individuals than did the hydrolysis of Hb method, but the urinary hydrolysis method detected exposure in all individuals including unexposed controls, which illustrates the need for a specific biomarker. Luna et al (2014) similarly used a trypsin digestion stage to derive signature peptide conjugates of MDI but concluded that the peptide method was about 18 times less sensitive than the MDI-lysine method.     

The binding of MDI to Alb appears to be an important area of research for both the use of an MDI-specific biomarker of exposure, and for investigation of the ultimate antigen involved in the sensitisation process. Wisnewski et al (2010) identified 14 MDI conjugation sites (12 lysines and 2 asparagines) on human Alb and highlighted reaction specificity for the second lysine in dilysine motifs which may be a common characteristic of ‘‘immune-sensitizing” chemicals. Several of the MDI conjugation sites are not conserved in albumin from other species, and this may suggest species differences in epitope specificity for albumin–isocyanate conjugates.

All methods require sophisticated laboratory facilities and capability but can be used for practical biomonitoring for exposures specific to MDI. It may be that biomonitoring can be a two-tier process, with urine analysis as a simple first step and protein adduct analysis for specific detailed investigations.