Registration Dossier

Diss Factsheets

Toxicological information


Currently viewing:

Administrative data

Description of key information

Several long-term carcinogenicity studies with rats, mice and hamsters with furfural are available. Based on the results of these studies it is concluded that furfural is carcinogenic in mice after oral administration. From a long-term dermal study in mice, it is concluded that furfural might have tumour initiating properties. After intratracheal administration or exposure to vapour in hamsters, there was no evidence that furfural possesses carcinogenic activity of its own.

Key value for chemical safety assessment

Carcinogenicity: via oral route

Endpoint conclusion
Dose descriptor:
53 mg/kg bw/day

Justification for classification or non-classification

Based on the available information, IARC (1995) has concluded that there is inadequate evidence in humans for carcinogenicity of furfural, and limited evidence in experimental animals. In November 2003 the CMR Working Group decided that furfural should be classified as Carc Cat 3. This decision was included in the EU-RAR 2008 on furfural. Furfural is classified in the 21 ATP of 67/548/EC as a Carc. Cat 3.

In accordance to these decisions, furfural needs classification with Carc Cat 2 H351 according to CLP.

Additional information

In a 103 week oral carcinogenicity study with Fischer 344 rats, there was no evidence of carcinogenic activity for females at 30 and 60 mg/kg bw/day. The study showed some evidence for carcinogenic activity for male F334/N rats at 60 mg/kg bw/day, based on the occurrence of uncommon cholangiocarcinomas in two high-dosed males and bile duct dysplasia with fibrosis in two other high-dosed males

It should be noted that the Pathology Working Group re-evaluated the slides of the 2-year NTP oral gavage study in rats, and unanimously concluded that the two lesions that were previously diagnosed as cholangiocarcinomas as well as the two lesions that were previously diagnosed as non-neoplastic proliferative lesions, are in fact to be correctly diagnosed as cholangiofibrosis.

In a 103 week oral carcinogenicity study with B6C3F1 mice, there was evidence of carcinogenic activity based on the significantly increased incidences of hepatocellular adenomas and hepatocellular carcinomas in males at a dose of 175 mg/kg bw/day. In female mice, the incidences of hepatocellular adenomas increased with dose and were significantly increased at 175 mg/kg bw/day. Although this response was not as strong as that observed in males, because of a lack of an increase in carcinomas, the dose-response and the significant increase in the high dose group were considered as some evidence of carcinogenic activity. Renal cortical adenomas or carcinomas in male mice and squamous cell papillomas of the forestomach in female mice may have been related to exposure to furfural. However, given the absence of substantial renal tubular cell hyperplasia and the fact that there was no dose relationship in the incidence of the neoplasms, the renal adenomas/carcinomas are considered not to be treatment-related. Furthermore, the squamous cell papillomas of the forestomach are difficult to associate with furfural exposure because of the low incidence, the uncertain biological potential (none progressed to malignant neoplasms) and their possible relationship to gavage administration. Based on the results of these studies, it is concluded that furfural is carcinogenic in mice after oral administration.

Long-term exposure (12 months) of hamsters to furfural vapour showed, that there was no evidence of furfural possessing carcinogenic activity. Simultaneously exposure with benzo[a]pyrene or with diethylnitrosamine showed that furfural is not a co-factor in respiratory tract carcinogenesis.

In a 36-week intratracheal instillation study in hamster with furfural , there was no evidence that furfural possesses carcinogenic activity of its own.

In 47 -week dermal toxicity studies in mice, Furfural in combination with TPA treatment induced eight skin tumours (7 papillomas, 1 squamous cell carcinoma) in 25% of the mice (average 0.40/mouse) whereas DMBA in combination with TPA induced tumours in all animals (average 6.7/mouse). No tumours appeared in mice treated with furfural alone. DMBA alone induced skin tumours in 35% of the mice (average 0.35/mouse) whereas TPA alone resulted in skin tumours in one animal only (5%; average 0.05/mouse) (Miyakawa et al., 1991). Given the increase in number of tumours in the 'furfural + TPA' group, compared with the group treated with TPA or furfural only, it is concluded that furfural may possess tumour initiating activity. Only data on the incidence and number of skin tumours and histological types were given. The occurrence of other effects, e.g. skin irritation, was not reported.

In the Risk Assessment final report of Furfural (2008), the following was concluded: "The mode of action underlying the hepatocarcinogenic activity of furfural after oral exposure has not fully been elucidated. However, a genotoxic component clearly is not involved, as evidenced by the in vivo test using transgenic animals. The data do, however, point to a possible role for chronic cytotoxicity that is found in conjunction with the induction of tumours; a pathway that has also been accepted for other non-genotoxic hepatocarcinogens. It may be argued that the observed cytotoxic effects were not extended and severe enough to explain this. However, this may be regarded as being in line with the observed tumour response. The tumour incidence in the rat is also very low, and the very sensitive mouse strain B6C3F1 already has a very high background incidence. Thus, the weak, though chronically sustained hepatotoxicity may have been sufficient to induce the low level of tumours. Secondly, it is unclear what exactly the (quantitative) nature of the relationship between toxicity and tumour-induction for different non-genotoxic hepatocarcinogens is. The true mechanisms underlying toxicity most probably differ both in qualitative and in quantitative for any chemical. For furfural, as compared with other non-genotoxic hepatocarcinogens, there may be a more prominent role for induction of mitosis by furfural, i.e. instead of clear hepatotoxicity and necrosis, as suggested by the results of the acute Wistar rat study by Shimizu and Kanisawa (1986). It is known, that a regenerative response in rat liver to toxicity or necrosis (even if this is observed only in the centrilobular region) is often located near the bile ducts, noticeable marked by the generation of so-called ‘oval cells’ (Laurson et al., 2005), and the bile duct area is exactly the location of the cholangiocarcinomas, which have been found after treatment with furfural. Therefore, it is assumed that the observed liver tumours were induced via some mechanism involving liver toxicity, and that at levels at which no liver toxicity is induced, tumours will not arise. Hence, as starting point for the risk characterisation for carcinogenicity the oral NOAEL for liver toxicity by the relevant route of administration (i.e. 53 mg/kg bw/d, from the dietary study as established under ‘repeated dose toxicity’) is taken". For the present evaluation, no new data are available, and therefore, the conclusion on oral carcinogenicity as described in the Risk Assessment Report is considered reliable for the present evaluation of furfural. 

Carcinogenicity: via oral route (target organ): digestive: liver