Registration Dossier

Administrative data

Key value for chemical safety assessment

Additional information

Genetic toxicity in vitro.

Four valid bacterial mutation studies have been identified, of which two are considered key (Klimisch 1). CTL Report YV6638 (2004) describes a GLP-compliant study of allyl alcohol using 4 Salmonella typhimurium plus 1 E. coli strains (TA 1535, 1537, 98 and 100 plus Wp2 uvrA (pKM101)) in pour-plate and pre-incubation (60 minutes prior to plating) assays with and without rat derived S9 mix: this gave wholly negative results. US NTP Report TR48 (2006) describes GLP-compliant testing of allyl alcohol in two different laboratories, each using 4 Salmonella typhimurium strains (TA97, 98, 100 and 1535) in pre-incubation (20 minutes) assays with and without rat and hamster derived S9 mixes: again all results were negative.


Two other studies (considered Klimisch 2) come from the open literature. Lutz et al (1982) tested both allyl alcohol and acrolein using only Salmonella typhimurium strain TA100 and a pre-incubation (90 minutes) method with and without S9 mix (presumed rat derived): they reported direct-acting mutagenicity of allyl alcohol (750 induced revertants/µmole), reduced 5-fold by S9 inclusion and speculated that mutagenic acrolein formed from the allyl alcohol might be inactivated through reaction with S9 components. Lijinsky and Andrews (1980) used pour-plate and pre-incubation (45 minutes) methods with Salmonella typhimurium strains (TA1535, 1537, 1538, 98 and 100) to examine allyl alcohol and acrolein mutagenicity: allyl alcohol proved active in strain TA1535, only with pre-incubation and hamster (but not rat) S9 while acrolein showed mutagenicity only in a TA98 pour-plate assay without S9 mix.


The microbial mutation assays of allyl alcohol reported by Principe et al (1981) are of uncertain validity for genotoxicity assessment (considered Klimisch 4): Salmonella typhimurium strains TA 1535, 1537, 1538, 98 and 100 were employed only in spot tests, while Streptomyces and Aspergillus spot test and pour-plate mutation assays used test systems not widely used or validated. All experiments gave negative results.


Two valid in vitro cytogenetic studies have been identified (both key). CTL Report SV1223 (2004; considered Klimisch 1) describes a GLP-compliant study of allyl alcohol using cultured human lymphocytes, with and without rat derived S9 mix: significant increases in incidences of cells showing chromosome aberrations (principally breaks and fragments) were seen, mainly at test concentrations above 100 µg/ml in two separate experiments (seen once without S9 mix only, once with and without S9 mix). The second study, described in US NTP Report TR48 (2006) used cultured CHO cells to investigate clastogenicity of the allyl alcohol metabolite acrolein and is hence considered Klimisch 2: no significant increase in frequency of cells with aberrations was seen using acrolein concentrations up to 1 µg/ml.


Two studies of in vitro mammalian cell mutation have been identified. CTL Report VV0306 (2004: key, considered Klimisch 1) describes a GLP-compliant study of allyl alcohol using L5178Y TK+/- cells, with and without rat derived S9 mix: significant increases in mutant colony numbers and mutation frequencies were seen only in the presence of S9 mix. These increases were generally associated with moderate or marked cytotoxicity and were due to increased numbers of small mutant colonies (at highest test concentration not toxicity-limited, ca. 4 - 32 times control small colony numbers, no doubling of control large colony numbers). Smith et al (1990 paper in Carcinogenesis: considered Klimisch 4 due to limited detail, including absence of reported control data) reported that allyl alcohol induced HPRT mutation in excision repair deficient cultured V79 (Chinese hamster) cells, showing activity slightly less than, but comparable to, that of its metabolite acrolein.


Genetic toxicity in vivo.

Six valid studies of allyl alcohol genotoxicity in vivo have been identified. CTL Report SR1290 (2005: considered Klimisch 1) describes a GLP-compliant study of unscheduled DNA synthesis in rats dosed orally with allyl alcohol: hepatocytes from treated rats were cultured in the presence of tritiated thymidine and after autoradiography net nuclear grain counts and percent cells in repair showed no evidence of treatment-induced DNA repair. US NTP Report TR48 describes two micronucleus studies of allyl alcohol considered Klimisch 1, one in rats (bone marrow sampled after 3 daily i.p. injections) and one in mice (peripheral blood lymphocytes sampled after 14 weeks of repeated oral dosing): no evidence of rat bone marrow toxicity and no significant increases in micronucleated rat or mouse cell frequencies were seen.


Jenkinson and(1990 paper in Mutation Research: considered Klimisch 2) described a study of rats dosed (presumed orally) with allyl alcohol daily for up 15 weeks and sequentially mated (weekly, dosing weeks 2-12) with virgin females to evaluate dominant lethal mutations in the sperm of treated males and foetal abnormalities in females terminated on pregnancy Day 21. No significant effect on these parameters resulted from allyl alcohol treatment. Takasawa et al (2010 paper in Mutation Research: considered Klimisch 2) studied the effect of oral allyl alcohol administration on frequency of micronucleated parenchymal hepatocytes in primary isolates from treated, 4-week old rats: this method, validated in an earlier collaborative study, examines chromosomal damage or aneugenic effect in naturally proliferating liver cells. Blood samples were also taken from the treated rats for evaluation of micronucleated peripheral blood erythrocytes: no evidence of genotoxic activity was detected in the examined liver or blood cells.

In addition to the rodent micronucleus assays cited above, US NTP Report TR48 describes sex-linked lethal assays of acrolein in the fruit-fly, Drosophila melanogaster (performed under GLP: considered Klimisch 2): three different treatment regimes (larval feeding, adult feeding, adult injection) were employed, each giving proof of delivery in the form of observed lethality. In all cases no evidence of treatment-related  mutation was recorded. While not a mammalian system and now less commonly used for mutagenicity testing, the fruit fly is known to possess mixed function oxidase, hydroxylase and transferase activity and be sensitive to various mutagens. Of particular note is that Drosophila possess both alcohol and aldehyde dehydrogenase activity (Garcin at al, Alcohol 2(1), 85-89, 1985; Leal and Barbancho, Insect Biochemistry and Molecular Biology 23(5), 543-547, 1993): in mammalian systems these enzymes convert glutathione-conjugated acrolein to the hydroxypropylmercapturic metabolite found in urine of rats and mice dosed with allyl alchol or acrolein (US EPA Toxicity Review Acrolein, EPA/635/R-03/003, 2003; US NTP TR48). It is thus reasonable to suppose that this mammalian detoxification process for acrolein may be mimicked in the fruit fly (in which a general glutathione-mediated detoxification pathway has been identified: e.g. Yepiskoposyan et al, Nucleic Acids Research 34(17), 4866-4877, 2006).  A detailed analysis of the power of the D. melanogaster sex-linked lethal test for detection of induced lethal mutation following chemical exposure has calculated that a 3-fold increase in such mutation over controls has a 95% probability of detection in a (treated X chromosome) sample size somewhat smaller than that examined in each of the three NTP-reported studies (Chapter 9 in Statistical Evaluation of Mutagenicity Test Data, UKEMS sub-committee report, Cambridge Press, 1989). This calculation was based on a spontaneous mutation rate 2-3 fold higher than that seen in the NTP TR48 experiments, reducing test resolving power, so it can be concluded that those experiments provided sensitive assays for induction of in vivo gene mutation by the major allyl alcohol metabolite, acrolein.


Overall conclusions.

It is difficult to draw a conclusion from the apparently contradictory results reported in bacterial mutation assays. It is notable that both key studies used the more sensitive pre-incubation technique but found no mutagenic activity of allyl alcohol: however US NTP TR48 also includes S. typhimurium testing of acrolein, where indications of mutagenic activity were obtained using the pre-incubation technique (a weak positive result in strain TA100 with rat S9, equivocal TA1535 and TA100 results with hamster S9). Considering this together with the findings of Lutz et al and Lijinsky and Andrews, we can conclude that under certain experimental conditions ally alcohol may induce base-substitution mutations in Salmonella typhimurium (strains TA1535 and TA100 typically detect this type of gene mutation and although TA98 typically detects frameshift mutations, presence of the rtf plasmid within this strain broadens its spectrum of sensitivity compared to its rtf-free source strain, TA1538: Ames, McCann and Yamasaki, Mutation Research 31, 1975); where this is seen, it probably reflects conversion of allyl alcohol to acrolein.


It is evident that allyl alcohol treatment of cultured human lymphocytes can lead to chromosome breakage events: whether this clastogenic activity is a consequence of acrolein release has not been determined, but the reported absence of such activity in the US NTP TR48 study using acrolein and CHO cells does not disprove such a hypothesis as the strong cytotoxicity of acrolein may have overridden any clastogenic activity in that study.


Positive results are reported for allyl alcohol in two different types of mammalian cell mutation assay. The CTL study using mouse lymphoma TK+/- cells has been selected as key study (Klimisch 1).  In Experimental Phase 1 with S9 mix, the absolute numbers of mutants were increased at a dose concentration of 20 μg/mL, but not at the next lower dose (10 μg/mL) or the next higher dose (30 μg/mL), which was excessively toxic. Viability (percentage relative total growth) at the effective dose was 40% and the overall mutant frequency at this dose was 2.4 and 3.3/104in two plates, compared with 0.8 and 0.4/104in the two solvent controls. The absolute number of colonies is important in this assay, because reliance only on a proportion of mutants could show an increase only as a result of cell selection and not of mutation induction. These data (absolute numbers of colonies) support a conclusion for an increase in mutagenic events. Small and large colonies were counted separately:small coloniesbeing slow-growth colonies, are likely the result of extensive, but sub-lethal genetic damage (deletions), whilelarge coloniesare faster-growing and likely the result of less extensive damage (point mutations). 


It can be concluded that treatment of L5178Y cells with allyl alcohol in the presence of rat liver-derived S9 mix increases the incidence of small colonies (indicative of large deletions, recognisable in a microscopic examination for structural chromosomal aberrations: Liechty et al, Mutagenesis 13(5), 1998; Hozier et al, Mutation Research 84(1), 1981; etc). However large colonies (indicative of point mutation induction) were increased to a small extent only at 30 μg/ml, where relative total growth was reduced to about 11%, close to the lower limit for data interpretation (10%), making the significance of these large colony counts uncertain. Hence activity of allyl alcohol was S9-dependent and indicative of chromosomal disruption rather than gene mutation.


The supporting mammalian cell mutation study of Smith et al using V79 cells is incompletely reported and hence designated Klimisch 4. However, its description of significant mutation at the HPRT locus (indicated by 6-TG resistance) following exposure to allyl alcohol and acrolein without S9 mix requires some consideration. Equimolar concentrations ( 1 and 2 µM) of allyl alcohol and acrolein produced only slightly lower mutation frequencies with the former, suggesting that allyl alcohol is extensively coverted to acrolein under the test conditions, presumably (as suggested by the authors) due to cellular alcohol dehydrogenase. If such activity is accepted to be real, two explanations are possible. Firstly it has been reported that although the position of the HPRT locus is such that mutational events are likely to reflect localised gene mutation rather than larger scale DNA/chromosomal damage, chemicals more usually characterised as clastogens can also induce HPRT mutation in V79 cells (e.g. coumoestrol, Kulling and Metzler, Food and Chemical Toxicology 35(6), 1997; mitomycin C, Davies et al, Mutation Research 291(2), 1993; potassium bromate, Speit et al. Mutation Research 439(2), 1999). Thus the findings of Smith et al may reflect the same in vitro clastogenic activity of allyl alcohol (probably after conversion to the ultimate clastogen, acrolein) as was seen in the CTL TK +/- study using the different L5178Y rodent cell line. However an alternative mechanism may explain the reported V79 HPRT mutation: Smith et al noted that the V79 cells they used were deficient in DNA (excision) repair capability, that acrolein is known to form DNA adducts and these may be efficiently removed by excision repair in other cell types. Curren et al (Mutation Research 209(1-2), 1988: work also cited by Smith et al) used normal and DNA repair-deficient human fibroblast cells (the latter from xeroderma pigmentosum patients) to show that repair-deficient cells are much more sensitive to the cytotoxicity of acrolein and are strongly mutated by it (HPRT locus mutation, indicated by 6-TG resistance); the normal cells better resisted acrolein cytotoxicity and showed no significant increase in HPRT mutant frequency after exposure. These authors also cited a bacterial mutation study showing acrolein mutagenicity which was highly dependent on absence of DNA excision repair. It may therefore be that the repair deficient V79 cells of Smith et al showed HPRT mutations as a consequence of acrolein-induced DNA adducts which in normal, repair-competent cells would have been excised without leading to mutation. Such a dependency on failure of excision repair for the induction of mutations might also explain the absence of HPRT mutation in CHO cells following acrolein exposure which has been reported by Parent, Caravello and Harbell (Journal of Applied Toxicology 11(2), 1991).


In vivo testing has shown no increase in micronucleated cells (indicative of clastogenic or aneugenic activity) in the bone marrow, peripheral blood and liver of rodents dosed orally or intraperitoneally with allyl alcohol. US NTP TR48 includes a report of negative results in a mouse peripheral blood micronucleus test performed with acrolein (given orally at up to 10 mg/kg/day for 14 weeks), in addition to the report of inactivity of allyl alcohol in the same assay system (orally dosed, up to 50 mg/kg/day for 14 weeks) and in the bone marrows of male rats (given three i.p. injections, up to 30 mg/kg/day where 40 mg/kg/day proved lethal). Takasawa et al (2010) directly investigated the principal target organ for allyl alcohol toxicity (and main site of metabolism to acrolein) in young rats and found no induction of micronuclei after oral dosing at up to 50 mg/kg: exposure of the target hepatocytes was confirmed by a marked reduction in mitotic index in these naturally dividing cells, and again investigation of peripheral blood cells found no treatment-related induction of micronuclei. An absence of dominant lethal mutations (commonly a result of chromosome deletions or similar, large-scale genetic changes) in the sperm of allyl alcohol-dosed rats was confirmed by Jenkinson and Anderson (1990). In addition, an absence of recessive lethal mutations (typically smaller-scale, gene mutations) in fruit flies exposed to the major allyl alcohol metabolite acrolein was demonstrated in the US NTP TR48 studies Together these studies show a complete absence of evidence for transmissible, germ cell mutation.


CTL Report SR1290 (2005) describes an absence of induced DNA repair in hepatocytes from rats dosed orally with allyl alcohol at up to 32 mg/kg (dose selected after a preliminary trial found macroscopic abnormalities in livers of rats given 50 mg/kg and higher dosages: approximately one half of the rat oral LD50 determined by Dunlap et al (1958)).


Thus allyl alcohol can show genotoxic activity in vitro: this may be due wholly or largely to the induction of damage or disruption at the chromosomal level, or such activity may also be associated with localised DNA interactions which under certain conditions can lead to gene mutation. The observed patterns of activity are compatible with dependence on conversion of allyl alcohol to the mutagen acrolein. Such conversion is known to occur with greatest efficiency in the liver of exposed mammals, but being mediated principally by alcohol dehydrogenase it may also occur (with lower efficiency) within exposed bacteria or mammalian cell lines. Uncertainty regarding conditions promoting the conversion in available in vitro test systems and the balance between genotoxicity and cytotoxicity of acrolein itself probably explains the somewhat contradictory findings in bacterial mutation and mammalian cell cytogenetic studies. The available evidence indicates that if gene mutations do occur in mammalian cells exposed in vitro, these may be a consequence of DNA repair deficiency (consequent upon interactions, probably between acrolein and DNA, which in cells with normal DNA repair competence are efficiently repaired).


The in vivo genotoxicity profile of allyl alcohol is clearer: internationally accepted and sensitive test methods for detection of chromosomal disruption (via clastogenic or aneugenic mechanisms) have given consistently negative results, even at dosages where systemic toxicity is expected and (in the work of Takasawa et al, 2010) in the primary target organ for allyl alcohol toxicity with evidence of target cell toxicity provided by reduced cell division. In addition, the CTL SR1290 (2005) UDS study shows no detectable increase in DNA repair in hepatocyte target cells of allyl alcohol dosed rats. This assay system does not monitor changes at individual gene loci but rather detects DNA strand breakages and fragments resulting from a wide range of gene mutation events. It is recognised as a suitably sensitive follow-up test to evaluate the significance of positive results obtained in gene mutation assays: the recent international expert group review “Follow-up actions from positive results of in vitro genetic toxicity testing” (ILSI/HESI Project Committee Review Subgroup: Dearfield et al, Environmental and Molecular Mutagenesis 52, 2011) lists detection of primary DNA damage in the UDS assay as appropriate to evaluate DNA reactivity in vivo in case in vitro gene mutation assays give positive results. It is also notable that DNA interactions which lead to mutation only if not resolved by normal excision repair would, in repair-competent cells, lead to generation of DNA fragments (expected to be detectable in a UDS assay).


Chromosomal changes observed following allyl alcohol exposure in vitro have not been detected in relevant whole animal follow-up tests. In vitro mutational events reported in mammalian cell lines (related to either chromosomal damage or gene mutations apparently dependent on DNA-repair deficiency) were also not detected by the combination of in vivo tests summarised here. It is therefore reasonable to conclude that the genotoxicity profile of allyl alcohol has been adequately defined and induction of mutation following in vivo exposure is not expected. Further testing is not necessary.

Short description of key information:
Allyl alcohol has been tested for induction of chromosome damage and gene mutation using various different bacterial and mammalian cell assays in vitro. Some bacterial assays using non-standard techniques showed induction of gene mutation, and mammalian cell assays demonstrated chromosome-breaking activity. Results obtained with a DNA repair deficient mammalian cell line suggested possible induction of gene mutations. However a range of whole animals assays, capable of detecting chromosomal disruption (by clastogenic or aneugenic mechanisms) and non-specific DNA damage and including assays focussed on the primary allyl alcohol target organ of the liver, found no evidence of genotoxicity.

Endpoint Conclusion: No adverse effect observed (negative)

Justification for classification or non-classification

Genetic toxicity

Based on the in vitro and in vivo genetic toxicity data, allyl alcohol is not classified for mutagenic activity under the classification criteria of:

- Directive 67/548/EEC

- UN GHS criteria

- Regulation 1272/2008 (EU CLP GHS).

(No such classification is proposed in Annex VI of the CLP Regulation).