Acrylonitrile induction of rodent neoplasia: Potential mechanism of action and relevance to humans

Literature search

A search of the published literature was conducted for combinations of the following terms: acrylonitrile, vinyl cyanide, toxicity, single-dose, repeat dose, acute, chronic, long-term, reproductive, developmental, carcinogenicity, tumorigenicity, cancer, neoplasia, tumor, malignancy, forestomach, brain, Zymbal’s gland, Harderian gland, epidemiologic, biomonitoring, genotoxicity, in vitro, in vivo, binding, macromolecules, protein, DNA, damage, adducts, strand breaks, mutagenicity, clastogenicity, chromosome aberrations, Ames, micronucleus, mouse lymphoma, sister chromatid exchange, study, test, assay, oxidative damage, oxidative stress, mitochondrial damage, inflammation, glutathione, depletion, cytotoxicity, immunosuppression, animal, rat, mouse, human, toxicokinetics, pharmacokinetics, physiological based pharmacokinetic modeling and simulation (PBPK), mechanisms, mode of action, mechanisms of action, metabolism, absorption, uptake, distribution, elimination, excretion, dose, and dose-response, using the following databases:

1. Pub Med (https://www.ncbi.nlm.nih.gov/pubmed) is a free source premier database for biomedical literature indexed in the National Library of Medicine MEDLINE database. It contains over 30 million citations from over 5000 current biomedical journals. Coverage range: 1950s to present.

In addition, safety profiles for acrylonitrile prepared by regulatory organizations including European Chemicals Agency, ³ European Chemicals Bureau, ¹ IARC,^2,26,27 Occupational Safety and Health Administration, ²⁸ and US EPA^29,30 were reviewed. Publications written in languages other than English were excluded.

MoA analyses

To structure the analysis of the reviewed literature, the evidence for key characteristics of carcinogens (Table 1) proposed by IARC ³¹ to support conclusions regarding potential human hazard from exposure to carcinogens was evaluated taking into consideration chemicokinetic and chemicodynamic factors in order to assess potential mechanisms of carcinogenicity and to evaluate whether the findings in rodents are plausible in humans.

OPEN IN VIEWER

Table 1.

Considerations for hypothesized mechanism of action (MoA) analyses.

Key characteristics of carcinogens defined by IARC a
It is electrophilic or can be metabolically activated
It is genotoxic
It alters DNA repair or causes genomic instability
It induces epigenetic alterations
It induces oxidative stress
It induces chronic inflammation
It is immunosuppressive
It modulates receptor-mediated effects
It causes immortalization
It alters cell proliferation, cell death, or nutrient supply
Evolved Bradford Hill weight of evidence considerations b
Biological concordance:
does the hypothesized MoA conflict with broader biological knowledge?
how well established is the MoA in the wider biological database?
Essentiality of key events:
is the sequence of events reversible if dosing is stopped or a key event prevented?
Concordance of empirical observations among key events (dose-response, temporality, incidence)
are the key events observed at doses below or similar to those associated with the adverse effect?
are the key events observed in hypothesized order?
is the occurrence of the end (adverse) effect less than that for the preceding key events?
Consistency:
is the pattern of effects across species/strains/organs/test systems what would be expected?
Analogy:
would the MoA be anticipated based on broader chemical-specific knowledge?

To address some concerns of the robustness of the IARC approach, ³² a weight of evidence (WoE) evaluation for the key events of hypothesized MoA(s) based on evolved Bradford Hill considerations ³³ (Table 1) was also applied.

Dose-response assessment for carcinogenicity findings

In order to characterize carcinogenic potency of acrylonitrile, tumor findings in target organs in rodent chronic bioassays were analyzed using the Benchmark dose (BMD) approach. BMD modeling was performed using the US EPA BMDS software version 3.0, according to the guidance in the software manual. A multistage-cancer model was used to calculate the BMD₁₀ (indicative of a 10% increase in tumor incidence over the control) for each data set. This model was selected for the present analyses since it also calculates a cancer slope factor (CSF). This cancer potency factor is an upper-bound estimate of the increase in cancer risk after a lifetime exposure of 70 years to a chemical by ingestion or inhalation. The fit of the model is given in Supplemental Figures 1–4. The BMDL and BMDU refer to the lower and upper bounds of a two-sided 95% confidence interval for calculated BMD_10.

Summary of acrylonitrile toxicity and carcinogenicity in humans

Humans exhibit acute toxicity to acrylonitrile. Thus, in workers, acute inhalational and/or dermal exposure to acrylonitrile at various concentrations above 5 ppm (10.9 mg/m³) was reported to result in irritation of eyes and nose, cyanosis, and neurological symptoms (headache, dizziness, fatigue, nausea, and vomiting).^2,3 These symptoms were reversible. Cases of occupational contact dermatitis have been reported after dermal exposure to acrylonitrile. ³⁴

A considerable body of epidemiological data has not revealed evidence of human cancer risk from acrylonitrile exposure.^5–12 While a recent retrospective study by the US National Cancer Institute on mortality of workers exposed to acrylonitrile reported a statistically significant (p for trend = .05) association between acrylonitrile exposure and lung cancer, as well as a possible association with increased mortality from bladder cancer and pneumonitis, ³⁵ findings from comprehensive meta-analysis did not support increased risk of lung cancer among acrylonitrile-exposed workers. ⁴

The International Agency for Research on Cancer (IARC) conducted several evaluations of acrylonitrile. In the first evaluation, IARC concluded that acrylonitrile “should be regarded as if it was carcinogenic to humans.” ²⁶ Later the carcinogenic risk of acrylonitrile to humans was assessed as “probable” (Group 2A), based on the limited evidence in epidemiologic studies and sufficient evidence for carcinogenicity to animals. ²⁷ In the latest IARC evaluation, acrylonitrile was downgraded to a “possible” human carcinogen, Group 2B, due to inadequate evidence for the carcinogenicity of this compound in humans based on additional epidemiological studies in occupationally exposed populations. ² The National Toxicology Program (NTP) listed acrylonitrile as a “reasonably anticipated to be a human carcinogen” based on sufficient evidence of carcinogenicity in experimental animals. ³⁶ The Environmental Protection Agency (EPA) classifies acrylonitrile as a “probable human carcinogen” (Group B1), based on the limited evidence of carcinogenicity in humans. ³⁰

Experimental carcinogenicity data and summary of neoplastic findings

Acrylonitrile was reported to produce tumors in multiple sites in rats and mice in a dose-related fashion in several chronic rodent studies, conducted with various protocols, utilizing several modes of acrylonitrile administration. Six of these studies of greater than 18 months in duration have been validated and accepted by regulatory agencies, and hence, were selected as the dataset for the current analyses. Five out of six studies were conducted in rats, with acrylonitrile administered in drinking water,^14,37,38 by oral gavage ³⁷ or inhalation. ³⁹ One study was conducted in mice dosed with acrylonitrile by oral gavage.^15,16 Findings in several other chronic bioassays conducted with acrylonitrile^13,40,41 are also briefly discussed.

Indicators of systemic stress associated with general toxic effects of acrylonitrile were reported in the rodent bioassays. These included increased mortality rates, decrease in food and water consumption, and consequent reduction in body weight gain, as well as mild anemia, evidenced by reduction in hematocrit, hemoglobin, and erythrocyte count levels (Table 2). Hematological changes were attributed to either blood loss from ulcerated tumors or poor nutrition (“nutritional anemia”) of the animals and generalized stress from acrylonitrile exposure, rather than bone marrow toxicity.^38,39

OPEN IN VIEWER

Table 2.

Systemic toxicity-associated findings in rodent carcinogenicity studies of acrylonitrile.

Strain, route of administration and duration	Dosage, mg/kg/d		Change (% incidence)
	M	F	Survival	Water consumption	Food consumption	Body weight gain	Hematologic changesb
1) Quast 2002 38
SD rats, drinking water, 22 (M) or 19 (F) months	3.4	4.4	↓ F	↓ (35–44%)	↓ F	↓ (25–27%)	-
	8.5	10.8	↓ F	↓ (25–27%)	↓	↓ (25–27%)	+
	21.3	25	↓ (19%)	↓ (36–38%)	↓ (50–66%)	↓ (25–27%)	+
2) Johannsen and Levinskas 2000a 37
SD rats, drinking water, 20 months	0.09	0.15	N/C	N/C	N/C	N/C	-
	8	10.7	↓ (10–40%)	↓	↓	↓ (8–10%)	+
3) Johannsen and Levinskas 2002b 14
F344 rats, drinking water, 24 (M) or 23 (F) months	0.1	0.1	N/C	N/C	N/C	N/C	-
	0.3	0.4	N/C	N/C	N/C	N/C	-
	0.8	1.3	↓ M (10%)	N/C	N/C	N/C	-
	2.5	3.7	↓ (20–40%)	N/C	N/C	↓ M (<5%)	-
	8.4	10.9	↓ (40–60%)	↓ (11–13%)	↓	↓ (11–13%)	+ F
4) Johannsen and Levinskas 2000a. 37
SD rats, oral gavage, 7 days per week, 20 months	0.1		N/C	N/C	N/C	N/C	-
	10		↓ (30–40%)	↑ F (6%)	N/C	↓ M (13%)	+
5) Quast et al. 1980 39
SD rats, inhalation, 6 hrs per day, 5 days per week, 22 months	2a	2.6a	↓ F (14%)	↑ M (15%)	N/R	↓ F	-
	9a	13.5a	↓ (14–21%)	↑ (13–40%)	N/R	↓ (7–15%)	+
6) Ghanayem et al. 2002 15 ; NTP 2001 16
B6C3F1 mice, oral gavage, 5 days per week, 18 (M) or 22 (F) months	2.5		N/C	N/R	N/R	N/C	N/R
	10		N/C	N/R	N/R	N/C	N/R
	20		↓ (29–48%)	N/R	N/R	↓ (7%)	N/R

Of the three modes of acrylonitrile administration to rats, intake in drinking water in both SD and F344 strains most severely impacted survival, especially in female rats, which were dosed with higher doses (up to 17% greater) compared to males (Table 2). An increase in mortality rates began at 10 and 16 months at 300 ppm and at 12 and 20 months at 100 ppm in both female and male rats, respectively, and at 18 months in both sexes at 35 ppm in the drinking water. ³⁸ The remainder of acrylonitrile-dosed rats of both sexes were terminated at 22 months. Reduction in survival was similar for the same oral dosages between the SD and F344 strains of rats.^14,37 With gavage administration, increased mortality was evident in rats of both sexes at 8.5 (male)/10.8 (female) mg/kg/day beginning at 14 months, until study termination at 20 months. ³⁷ In B6C3F1 mice, which also were dosed with acrylonitrile by oral gavage,^15,16 the high dose (14.3 mg/kg/day) produced significant reduction in survival in both sexes (males greater than females), after approximately 1 year of dosing. In the rat inhalation study, mortality was increased at 80 ppm, starting at 6 months in males, and 10 months in females, until termination of the study at 22 months. ³⁹

Species-specific differences in target tissues for acrylonitrile-induced carcinogenicity were observed between rats and mice, with exception of the common occurrence of forestomach neoplasia in both species. In all rat bioassays, forestomach, brain, and Zymbal’s gland were common target organs, regardless of the route of administration. In the mouse gavage study, forestomach neoplasia as well as Harderian gland tumors were observed. Hyperplastic changes in target organs were often observed in both rats and mice at carcinogenic dosages of acrylonitrile or one dose level below the carcinogenic doses (Tables 3–6), suggesting that these events were associated with neoplasia. Tumors in the small intestine, mammary gland, and tongue were also observed in several rat bioassays, but these tumors were sex-specific and, thus, were not considered for the current analyses. In female mice, neoplasms of the ovary and lung may have been related to administration of acrylonitrile, however, association with exposure was considered equivocal due to a lack of dose-response; thus, these findings were also excluded.

OPEN IN VIEWER

Table 3.

Percent incidence of pertinent non-neoplastic and neoplastic findings in Sprague-Dawley rats dosed with acrylonitrile in drinking water for 2 years.

Microscopic finding	Sex	Acrylonitrile dosagea,b
		0 ppma,b		1 ppm b	35 ppm a	100 ppma,b		300 ppm a
	M	0 mg/kg/d a	0 mg/kg/d b	0.09 mg/kg/d	3.4 mg/kg/d	8.5 mg/kg/d a	8 mg/kg/d b	21.3 mg/kg/d
	F			0.15 mg/kg/d	4.4 mg/kg/d	10.8 mg/kg/d a	10.7 mg/kg/d b	25.0 mg/kg/d
	HED c	0 mg/kg/d		0.02 mg/kg/d	0.58 mg/kg/d	1.46 mg/kg/d	1.36 mg/kg/d	3.62 mg/kg/d
Forestomach hyperplasia/hyperkeratosis	M	19%	77%	90%	32%	92%*	91%*	94%*
	F	25%	81%	63%	48%*	85%*	83%*	98%*
Forestomach papilloma/carcinoma	M	0	3%	3%	4%	48%*	12%	81%*
	F	1.3%	1%	4%	2%	25%*	7%*	63%*
Glial cell proliferation	M	0			9%*	6%		15%*
	F	0			6%	6%		15%*
Brain astrocytomas d	M	1.3%	2%	3%	17%*	40%*	24%	48%*
	F	1.3%	0	1%	35%*	46%*	33%*	50%*
Zymbal’s gland hyperplasia e	M	3%			9%	2%		33%
	F	1%			8%	6%		21%
Zymbal’s gland adenoma/carcinoma	M	4%	1%	0	9%	6%	20%*	33%*
	F	1.3%	1%	0	10%*	19%*	12%*	38%*

OPEN IN VIEWER

Table 4.

Percent incidence of pertinent non-neoplastic and neoplastic findings in Fischer 344 rats dosed with acrylonitrile in drinking water for 2 years.

Microscopic finding	Sex	Acrylonitrile dosage a
		0 ppm	1 ppm	3 ppm	10 ppm	30 ppm	100 ppm b
	M	0 mg/kg/d	0.1 mg/kg/d	0.3 mg/kg/d	0.8 mg/kg/d	2.5 mg/kg/d	8.4 mg/kg/d
	F		0.1 mg/kg/d	0.4 mg/kg/d	1.3 mg/kg/d	3.7 mg/kg/d	10.9 mg/kg/d
	HED c	0 mg/kg/d	0.02 mg/kg/d	0.04 mg/kg/d	0.14 mg/kg/d	0.43 mg/kg/d	1.39 mg/kg/d
Forestomach hyperplasia/hyperkeratosis	M	6%	3%	19%*	13%*	17%*	9% b
	F	2%	2%	16%*	24%*	13%*	5% b
Forestomach papilloma/carcinoma	M	0	1%	4%*	4%*	4%*	1% b
	F	0–1%	1%	2%	2%	4%*	2% b
Gliosis	M	1%	0	0	0	0	1% b
	F	0	0	0	1%	0	0
Brain astrocytomas d	M	0–2%	2%	1%	2%	10%*	21%*
	F	0–1%	1%	2%	4%	6%*	23%*
Zymbal’s gland papilloma/adenoma/carcinoma	M	0–2%	1%	0	2	7%*	17%*
	F	0	0	2%	4%*	5%*	12%*

*, statistically significant (p < .05) compared to in-study control values for both pairwise and trend parameters; F, female; M, male.

^aJohannsen and Levinskas 2002b. ¹⁴

^bMarked increase in mortality.

^cHED, human equivalent dose. ²⁹

^dNow considered to be malignant microglial tumors. ¹⁸

OPEN IN VIEWER

Table 5.

Percent incidence of pertinent neoplastic findings in Sprague-Dawley rats dosed with acrylonitrile by oral gavage ^a and by inhalation ^b for 2 years.

Microscopic finding	Sex	Acrylonitrile dosage
		0 ppm a	1 ppm a	100 ppm a	0 ppm b	20 ppm b	80 ppm b
	M	0 mg/kg/d	0.1 mg/kg/d	10 mg/kg/d	0 mg/kg/d	2 mg/kg/d c	9 mg/kg/d c
	F				0 mg/kg/d	2.6 mg/kg/d c	13.5 mg/kg/d c
Forestomach hyperplasia/hyperkeratosis	M	79%	69%	81%*	8%	7%	16%
	F	80%	84%	90%*	2%	3%	7%
Forestomach papilloma/carcinoma	M	2%	6%	40%*	1%	2%	5%
	F	2%	4%	17%*	0	0	1%
Glial cell proliferation	M	N/R	N/R	N/R	0	0%	7%*
	F				0	4%	4%
Brain astrocytomas d	M	2%	0	16%*	0	4%	15%*
	F	1%	2%	17%*	0	4%*	16%*
Zymbal’s gland adenoma/carcinoma	M	1%	1%	16%*	2%	4%	11%*
	F	1%	0	15%*	0	1%	11%*

*, statistically significant (p < .05) compared to in-study control values for both pairwise and trend parameters; F, female; M, male; N/R, not reported.

^aJohannsen and Levinskas 2000a. ³⁷

^bQuast et al. 1980. ³⁹

^cEquivalent dose; details for calculations are provided in footnote of Table 2.

^dNow considered to be malignant microglial tumors. ¹⁸

OPEN IN VIEWER

Table 6.

Percent incidence of pertinent non-neoplastic and neoplastic findings in B6C3F1 mice dosed with acrylonitrile and delivered by oral gavage for 2 years.

Microscopic finding	Sex	Acrylonitrile dosage a
		0 mg/kg/d	2.5 mg/kg/d	10 mg/kg/d	20 mg/kg/d
	M	0 mg/kg/d b	1.8 mg/kg/d b	7.1 mg/kg/d b	14.3 mg/kg/d b
	F
Forestomach hyperplasia	M	4	8%	16%*	18%*
	F	4	4%	10%	14%*
Forestomach papilloma/carcinoma	M	6	8%	52%*	64%*
	F	6	14%	50%*	58%*
Harderian gland hyperplasia	M	2	8%	14%*	8%
	F	10	8%	12%	16%
Harderian gland adenoma/carcinoma	M	12	32%*	54% *	60%*
	F	22	20%	52%*	50%*

*, statistically significant (p < .05) compared to in-study control values for both pairwise and trend parameters; F, female; M, male.

^aGhanayem et al. 2002 ¹⁵ ; NTP 2001. ¹⁶

^bAdjusted 7 days per week intake levels of 2.5, 10, and 20 mg/kg body weight administered 5 days per week.

Studies in rats

In the studies with acrylonitrile administration in the drinking water to SD rats, incidences of forestomach and Zymbal’s gland neoplasia were significantly increased in both sexes in groups dosed with 100 ppm and above, while Zymbal’s gland tumors were also observed in females at 35 ppm.^37,38 Brain neoplasms were present in dosed groups in both sexes at 35 ppm and above (Tables 3 and 7).

OPEN IN VIEWER

Table 7.

Thresholds for neoplasms produced by acrylonitrile in rodent carcinogenicity bioassays.

Study and experimental design	Sex	Forestomach neoplasms		Brain neoplasms		Zymbal’s gland neoplasms		Harderian gland neoplasms
		NOAEL, mg/kg bw/d	LOAEL, mg/kg bw/d	NOAEL, mg/kg bw/d	LOAEL, mg/kg bw/d	NOAEL, mg/kg bw/d	LOAEL, mg/kg bw/d	NOAEL, mg/kg bw/d	LOAEL, mg/kg bw/d
1) Quast 2002 38 0, 35, 100, 300 ppm in drinking water to SD rats (48/sex/dose)	M	3.4	8.5	−	3.4	8.5	21.3	N/D
	F	4.4	10.8	−	4.4	−	4.4
2) Johannsen and Levinskas 2002a 37 0, 1, 100 ppm in drinking water to SD rats (100/sex/dose)	M	8	−	8	−	0.09	8	N/D
	F	0.15	10.7	0.15	10.7	0.15	10.7
3) Johannsen and Levinskas 2002b 14 0, 1, 3, 10, 30, 100 ppm in drinking water to F344 rats (100/sex/dose)	M	0.1 a	0.3	0.8	2.5	0.8	2.5	N/D
	F	1.3 a	3.7	1.3	3.7	0.4	1.3
4) Johannsen and Levinskas 2002a 37 0, 0.1, 10 mg/kg/d by gavage, 7 days/wk, to SD rats (100/sex/dose)	M	0.1	10	0.1	10	0.1	10	N/D
	F	0.1	10	0.1	10	0.1	10
5) Quast et al. 1980 39 0, 20, 80 ppm by inhalation, 6 hrs/d, 5 days/wk to SD rats (100/sex/dose)	M	9 b	−	2 b	9 b	2 b	9 b	N/D
	F	13.5 b	−	−	2.6 b	2.6 b	13.5 b
6) Ghanayem et al. 2002 15 ; NTP 2001 16 0, 2.5, 10, 20 mg/kg/d by gavage, 5 days/wk, to B6C3F1 mice (50/sex/dose)	M	1.8 c	7.1 c	N/D		N/D		−	1.8 c
	F	1.8 c	7.1 c					1.8 c	7.1 c

Statistically significant increases in the incidences of squamous carcinomas of Zymbal’s gland and a trend towards an increase of forestomach papillomas were also observed in SD rats that received 500 ppm of acrylonitrile in drinking water. ⁴⁰ The small group size (20 rats per group) limited the power of the study. At 500 ppm, rats exhibited higher mortality rates, a trend towards a decrease in water consumption, and a slower increase in body weight gain. No tumor increases in other organs were reported.

F344 rats were more susceptible to acrylonitrile tumorigenicity compared to SD rats. Thus, acrylonitrile administered in drinking water produced a significant increase in the incidences of forestomach neoplasia in F344 males at 3 ppm and above (with exception of 100 ppm), and in F344 females at 30 ppm. ¹⁴ Brain tumors in both sexes and Zymbal’s gland tumors in males were significantly induced by 30 ppm and above, Zymbal’s gland tumors in females were induced by 10 ppm and above (Tables 4 and 7). The low survival of rats in the highest dose group (100 ppm) after 14 months of the study possibly affected statistical analyses.

In a study to investigate the nature of the brain tumors, F344 rats of both sexes were administered acrylonitrile in drinking water at 100 and 500 ppm for 12–18 months. ¹³ At both dose levels, acrylonitrile produced an increase in incidences of brain tumors, primarily observed in the cortex. While cells adjacent to the lesions were reactive astrocytes, negative staining of the tumors for glial fibrillary acidic protein (GFAP) microscopical, as well as ultrastructural evidence of no glial filaments, indicated that the neoplastic cells were different in origin (i.e., not astrocytes). An increase in incidences of Zymbal’s gland tumors and forestomach and subcutaneous papillomas was also observed in acrylonitrile-dosed rats. In addition to carcinogenicity, rats that received acrylonitrile exhibited signs of systemic and neurotoxicity from 12 to 18 months (i.e., until termination).

Similar to the outcomes described above for F344 rats, acrylonitrile administered by oral gavage to SD rats at 100 ppm (equivalent to male and female doses of 8 and 10.7 mg/kg of acrylonitrile in drinking water per day, respectively) produced a significant increase in tumor incidences in three target organs (forestomach, brain, and Zymbal’s gland) in rats in both sexes (Tables 5 and 7). ³⁷

In an inhalation study of acrylonitrile in SD rats, no increase in forestomach tumor incidences was observed, while brain and Zymbal’s gland neoplasms were present in both sexes at the highest dose of 80 ppm (equivalent dose of 173.6 mg/m³ per day). ³⁹ Brain tumors were also observed at 20 ppm in female rats (43.4 mg/m³ per day) (Tables 5 and 7). The authors also reported inflammatory changes in the respiratory epithelium of the nasal turbinates at both dose levels, with less severe changes observed at 20 ppm. These findings were considered to result from irritation produced by acrylonitrile. ³⁹ Acute suppurative pneumonia observed in the rats administered 80 ppm of acrylonitrile was interpreted to be related to stress from acrylonitrile exposure.

In a study in SD rats, no increase in tumor incidences was observed after oral gavage with acrylonitrile at 5 mg/kg bodyweight for 52 weeks. ⁴¹ In contrast, in inhalation studies, statistically significant, but not dose-related, increases in the total number of animals with tumors were observed in rats of both sexes exposed to acrylonitrile at the dose range from 5 to 40 ppm for 52 weeks. Slight, not significant increases were observed in the incidences of “encephalic glyomas,” although, the numbers were within the historical control range. When administered at 60 ppm for 104 weeks to breeders and offspring rats, acrylonitrile produced an increase in “encephalic glyomas” (oligodendrogliomas) in all dosed groups. In addition, a significant increase in incidences of Zymbal’s gland carcinomas was observed in male offspring.

Study in mice

In a 2-year carcinogenicity study in B6C3F1 mice administered acrylonitrile by gavage, increases in the incidences of neoplasia in two target organs, forestomach and Harderian gland, were observed in both sexes at 10 and 20 mg/kg body weight per day.^15,16 In males, the percent incidences of Harderian gland adenoma and carcinoma were significantly increased at all tested dose levels, from 2.5 to 20 mg/kg/day of acrylonitrile, and in females at 10 and 20 mg/kg/day (Tables 6 and 7). Noteworthy, in both sexes, over 90% of all tumors of the Harderian gland were benign adenomas. Increased incidences of combined alveolar/bronchiolar adenomas and carcinomas, as well as benign/malignant granulosa cell tumors of the ovary, were reported only in female mice from a mid-dose group (10 mg/kg/day). These findings were considered equivocal due to a lack of dose-response.

Summary of carcinogenicity findings with regard to potential MoA

The difference in the target organs of carcinogenicity between rats and mice denotes inter-rodent variations, supporting the concept that acrylonitrile-induced neoplasia is “species-specific,” with exception of the forestomach which was a common tumor site in both species with oral administration. Overall, responses were similar between the sexes, although, in a few studies, female rats were more susceptible to brain or Zymbal’s gland carcinogenicity (Table 7). Target organs of acrylonitrile carcinogenicity are the sites of spontaneous background neoplasms, ²⁵ indicating the susceptibility of these tissues to tumor development.

Rates of “spontaneous” forestomach squamous cell carcinomas or papilloma’s range between 0.08 and 0.25% in F344 and SD rats and between 1 and 2% in B6C3F1 mice.^42–44 Forestomach squamous cell neoplasia is the sixth most commonly induced neoplasm in NTP rodent carcinogenicity assays and is similar in histological appearance in all rodent species and strains.^45,46 Two recognized key events involved in the development of forestomach neoplasms are DNA damage and chronic irritation. ⁴⁶ Moreover, an association between early epithelial cell proliferation and hyperkeratosis and development of forestomach neoplasm was described for several carcinogens, especially following chronic administration by gavage. ⁴⁷

The incidence of “spontaneous” brain glial neoplasms in rodents is low, that is, 1–2%, although, with advancing age, rats develop these neoplasms more commonly than mice.^18,48 In the F344 rat, their incidence is lower than that of the SD strain.^18,48 While previous studies reported that “astrocytomas” have been the most commonly identified brain tumors by conventional histopathology analyses, ⁴⁹ based on immunohistochemical analyses, commonly observed background glial brain tumor in rats were primarily oligodendroglioma. ¹⁸ Staining characteristics of other, less common spontaneous tumors previously diagnosed as astrocytomas suggested origin from monocytic cells, and thus they were diagnosed as malignant microglial tumors. ¹⁸ Genotoxicity is an accepted MoA for several compounds that induce brain tumors in rodents, including ethyl-N-nitrosourea, N-methylnitrosurea, and several electrophilic epoxide or epoxide-forming compounds, including ethylene oxide and acrylamide.^50,51

Spontaneous neoplasms of Zymbal’s auditory gland in rats are not common, even in old age.^45,52 Thus, the historical incidences of Zymbal’s gland tumors in rodents are reported as ranging from 0.4 to 2%.^42–44,53 Many Zymbal’s gland neoplasms, however, are missed because either they are not large enough at necropsy or the tissue is not on the standard protocol organ list. NTP data suggest that chemicals that produce Zymbal’s gland tumors are strongly associated with tumor development in the skin and in preputial glands in male and clitoris and mammary glands in female rats. ⁵⁴ This holocrine auditory integumental accessory tissue is described in rats but is in a vestigial form in mice. To date, no key events in Zymbal’s gland neoplasia have been reported. Non-neoplastic changes observed in Zymbal’s gland, including inflammatory and degenerative changes, are histologically similar to those seen in the mammary gland. ⁵⁵

The incidence of spontaneous neoplasms of the ocular orbit Harderian gland increases with age, especially in the mouse. ⁵² Thus, the rate of “spontaneous” carcinomas or adenomas in B6C3F1 mice is over 12% in males and 7.4% in females, while in F344 and SD rats it does not exceed 0.3%.^42–44 Harderian gland neoplasms are suggested to result from genotoxicity, ⁵² but no direct measure of DNA damage in this gland has been reported for any chemical, and no key events are known.

Dose-response assessment of carcinogenic effects

In all six rodent bioassays, tumors were observed only after at least 12 months of acrylonitrile administration. Delayed occurrence of tumors might suggest that acrylonitrile has at most weak genotoxic activity, as suggested by ethylene oxide.^51,56 Thus, the existence of thresholds for acrylonitrile tumorigenicity is plausible.^57,58

Overall, in rats, the lowest-observed-adverse-effect-level (LOAEL) for neoplasia induction by acrylonitrile in target organs was at least 0.3 mg/kg/day, and the lowest NOAEL was 0.09 mg/kg (Table 7). In mice, the LOAEL was at least 1.8 mg/kg/day, and the NOAEL was below this concentration (Table 7). According to the Carcinogenic Potency Database (CPDB, https://toxnet.nlm.nih.gov/cpdb/chempages/ACRYLONITRILE.html), median tumor doses (TD₅₀) for acrylonitrile are estimated to be 16.9 and 6.32 mg/kg per body weight per day, for rat and mouse, respectively.

BMD modeling showed that overall, the potency of acrylonitrile was consistent for different tumor types across studies with different routes of acrylonitrile administration (Figure 1). Male rats and mice in most cases were more susceptible to carcinogenic effects of acrylonitrile, compared to females. Acrylonitrile was more potent in mice compared to rats. Analyzing each tumor type separately, the lowest BMDL for the brain tumor incidence in rats of both sexes was ∼2 mg/kg body weight (Figure 1a). BMDs for forestomach tumors could not be computed in F344 rats of both sexes from the drinking water study, and in female SD rats from the inhalation study. The failure to estimate BMDs for these groups could be due to the fact that maximum tumor increase in the dosed groups was smaller than specified benchmark response of 10% (Suppl. Figures 2 and 3). Based on the remaining data, the lowest BMDLs for forestomach tumors were 1.7 mg/kg in rats and ∼1 mg/kg in mice (Figure 1b and d). The lowest BMDL for the Zymbal’s gland tumors in rats was 2.7 mg/kg, and 1.2 mg/kg for Harderian gland tumors in mice (Figure 1c and d).OPEN IN VIEWER

OPEN IN VIEWER

Figure 2.

Structure, metabolism and reactivity of acrylonitrile. Solid arrows indicate the major pathway, broken arrow indicates minor pathway. Relevance of rat/mice metabolites/metabolic pathways to humans is noted. References: Friedman et al. 1965, ⁶² Kedderis and Batra 1993, ⁶³ Kedderis et al. 1995. ⁶⁴

OPEN IN VIEWER

Figure 3.

Proposed mechanisms of action (MoA) of acrylonitrile associated neoplasia. Solid arrows indicate direct key event relationships; broken arrows indicate indirect key event relationships. CEO, 2-cyanoethyleneoxide.

CSF for all tumor types was estimated to range from 0.01 to 0.06 per mg/kg per day based on the findings in the rat studies, and from 0.06 to 0.09 per mg/kg per day based on the study conducted in mice. An oral multi-tumor Slope Factor of 0.03 per mg/kg per day was derived from a geometric mean for all rat studies with acrylonitrile administration in drinking water. In order to estimate human risk from inhalation exposure, inhalation cancer unit risk factors (CSF expressed as μg/m³) were also calculated based on the tumor incidences in the inhalation study in rats (Quast et al. 1980) using the concentration in μg/m³ (43.4 × 10⁻³ and 173.6 × 10⁻³ μg/m³). The geometric mean for all tumors in both sexes of rats was estimated to be 1.02 × 10⁻⁶. Individual values for each tumor type and sex are provided in Supplemental Table 1.

US EPA (2006) estimated an inhalation unit risk for acrylonitrile of 6.8 × 10⁻⁵ per μg/m ³ , and that inhalation of 1 μg/m³ acrylonitrile in the air over the lifetime of an individual would result in not greater than one-in-ten thousand increased risk of developing cancer. This estimate is adjusted for smoking and continuous lifetime equivalent of occupational exposure, and thus, is different from the calculations above.

Previous cancer dose-response modeling of acrylonitrile cancer potency, based on the findings in rodent carcinogenicity, PBPK prediction for internal peak dose of CEO in the brain, and mechanistic studies, established that oral doses below 0.009 mg/kg/day and air concentrations below 0.1 mg/m³ are not expected to pose an exposure risk to humans. ⁵⁹

Review of key characteristics of acrylonitrile relevant to its mechanism of action

Review of animal toxicity data

Acute toxicity

Acrylonitrile produced acute toxicity in experimental animals regardless of the route of administration. ³ The toxicity of acrylonitrile could result either from direct effects or from effects of its metabolites, CEO or cyanide (Figure 2). Oral median lethal dose (LD₅₀) values for acrylonitrile range from 25 to 186 mg/kg body weight, with the mouse being the most sensitive species and rat the least (Table 8). Acute inhalation toxicity studies yielded LC₅₀ values of 0.3–1.21 mg/L for 4-hour exposure in various species and another study determined a 4-hour LC₅₀ of 2.05 mg/L in rats.^1,3 In acute dermal toxicity studies, the rat was the most sensitive species with LD₅₀ values ranging from 148 to 282 mg/kg body weight, followed by the rabbit, 226 mg/kg, and the guinea pig, 260–690 mg/kg. Studies with other routes of administration (subcutaneous, intravenous, and intraperitoneal) yielded LD₅₀ values similar to those from oral toxicity studies (Table 8).

OPEN IN VIEWER

Table 8.

Acute toxicity of acrylonitrile in different species.

Route of administration	Reported LD50 and LC50 values				Effects
	Mouse	Rat	Guinea pig	Rabbit
Oral	25–48 mg/kg a	72–186 mg/kg a	50–85 mg/kg a	93 mg/kg b	Seizures, paralyses, dyspnea, agitation, salivation, lacrimation
Inhalation (4 h)	0.3 mg/L a	425 ppm c	576 ppm a	260 ppm d	Dyspnea, tremor, lacrimation
		1.21–2.05 mg/L a	0.99 mg/L a
Skin		148–282 mg/kg a	260–690 mg/kg a	226 mg/kg a	Dermatitis, somnolence
			250 mg/kg b	280 mg/kg b
			202 mg/kg d	63 mg/kg d
Subcutaneous	25–50 mg/kg a	80–96 mg/kg a	130 mg/kg d		Tremor, convulsions, flaccid paralysis with anesthesia
Intravenous			72 mg/kg a	69 mg/kg a	Convulsions, tremor, flaccid paralysis with anesthesia
Intraperitoneal	47–50 mg/kg a	65–100 mg/kg a			Dyspnea, ataxia, corneal damage

LD₅₀, median lethal dose.

^aECHA 2018. ³

^bSnyder 1990. ⁹⁰

^cLewis 2004. ⁹¹

^dChemIDplus database https://chem.nlm.nih.gov/chemidplus/name/acrylonitrile.

Signs of neurotoxicity were commonly observed in the majority of acute toxicity studies. They occurred in stages, starting with agitation and lacrimation in the excitatory phase, salivation, lacrimation, urination, and defecation during a tranquil phase, which was followed by seizures (convulsive phase), paralysis (paralytic phase), and death. ³ Edema of the brain and lungs, as well as bleeding in the gastrointestinal tract and hemorrhagic necrosis of the adrenals, were also observed. A single dose of acrylonitrile at 20, 40, or 80 mg/kg body weight administered to male Sprague-Dawley (SD) rats by gavage or subcutaneous injection also produced neurotoxic effects, cholinomimetic in nature. ⁹² The signs included salivation, lacrimation, miosis, vasodilation, and diarrhea. Pre-dosing with atropine protected rats from the neurotoxicity, suggesting possible involvement of the cholinergic system in acute toxicity produced by acrylonitrile.

A single oral administration of acrylonitrile by gavage at LD₅₀ levels to male SD rats (93 mg/kg) and male Albino-Swiss mice (27 mg/kg) resulted in different symptoms of toxicity in both species. ⁸³ Specifically, while rats developed cholinomimetic signs of toxicity (salivation, lacrimation, diarrhea, wheezing, peripheral vasodilation), mice developed CNS disturbances, such as depression, agitation, convulsion, and asphyxia.

The acute toxicity of acrylonitrile has been proposed to result from its bioactivation by cytochrome P450 enzymes leading to formation of cyanide (Figure 2).^77,93 Thus, in wild-type mice (mixed 129/Sv and C57BL strains) of both sexes dosed at up to 40 mg/kg of acrylonitrile by gavage in a single dose, increase in lethality was associated with dose-dependent increase of cyanide levels in blood (in males up to 120 nmol/ml and in females up to 40 nmol/ml one hour after exposure) and brain (in males up to 60 nmol/ml and in females up to 10 nmol/ml one hour after exposure). In contrast, CYP2E1-deficient mice did not exhibit significant signs of toxicity, and cyanide levels in their blood and brain tissue were comparable to that of the control group.^77,93 Ahmed and Patel ⁸³ suggested a role for cyanide formation in acrylonitrile-produced toxicity in rodents which is species dependent. In mice, which metabolize acrylonitrile faster, cyanide formation is the major mechanism of toxicity, whereas in rats, it plays a minor role.

Others suggest that binding of acrylonitrile to cysteine residues on proteins can lead to inhibition of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). ⁹⁴ Thus, in vitro acrylonitrile at 200 μm for 1 h irreversibly inhibited GAPDH activity by 90% due to alkylation of Cys149 residue of the enzyme. These events might contribute to inhibition of oxidative phosphorylation in mitochondria as well as impaired glycolytic ATP production. While mitochondrial damage was described in vitro after long-term acrylonitrile administration,^95,96 no ATP depletion was observed in the brain tissues of male SD rats exposed to a single subcutaneous injection of a LD₉₀ dose (115 mg/kg) of acrylonitrile, suggesting that other mechanisms also may be involved in the acute toxicity of this compound. ⁹⁷

Depletion of GSH is likely to exacerbate toxicity of acrylonitrile. Thus, a single subcutaneous dose of acrylonitrile administered to male SD rats at 30 or 50 mg/kg body weight produced focal erosions of glandular stomach mucosa, which became more severe with time and were associated with hepatic and gastric glutathione depletion (25% decrease in liver and 50% decrease in glandular stomach). ⁹⁸ In this study, rats which were pretreated with sulfhydryl-containing compounds, as well as atropine, were protected against necrosis and gastric erosions, suggesting that acrylonitrile alters cholinergic receptors by depleting endogenous sulfhydryl groups. A similar effect on GSH depletion was observed in male SD rats administered a single oral dose of 23 mg/kg of acrylonitrile, which decreased gastric glutathione levels by 50%, 4 hrs after administration, and 46 mg/kg of acrylonitrile further decreased the glutathione level to 27% of the control group. ⁹⁹

Szabo and colleagues ¹⁰⁰ reported that intravenous administration of a single dose of acrylonitrile at 150 mg/kg to female SD rats resulted in rapid depletion (80–90%) of glutathione in the liver, lung, and kidney in the absence of cell damage in these organs, while in a target tissue for acrylonitrile carcinogenicity, brain, GSH depletion was more gradual (approximately 30% depletion at the highest dose) but correlated with mortality. GSH levels in the liver at lower dose levels of 10 and 50 mg/kg were decreased by approximately 25% and 60%, respectively. Similar findings were obtained in the rat study with a single subcutaneous dose of acrylonitrile. ¹⁰¹ Thus, 20 mg/kg of acrylonitrile, a dose which did not produce toxicity in acute toxicity studies, produced a 50% decrease in GSH levels in the liver of male SD rats, while the LD₁₀ dose of 50 mg/kg depleted GSH by up to 85%. The two highest doses, 80 mg/kg (equivalent of the LD₅₀ dose) and 115 mg/kg (equivalent of the LD₉₀ dose) almost entirely depleted liver GSH within 30 and 15 min, respectively.

While the brain was more resistant to GSH depletion, a dose-dependent decrease was demonstrated at doses that exhibit toxicity (50, 80, and 115 mg/kg), with 10% depletion at the highest dose. Importantly, GSH levels showed little recovery in the brain, unlike in non-target tissues, liver, and kidney. GSH depletion in the forestomach (another target organ for acrylonitrile carcinogenicity) and glandular stomach was about 40% at the highest dose and occurred at a slower rate compared to other tissues; similar to the brain, no GSH recovery was observed in these tissues. ¹⁰¹ The GSH depletion was accompanied by a dose-dependent increase in cyanide levels, at the highest dose reaching approximately 160 nmol/ml in the blood and over 20 nmol/g in the brain. Toxic levels of cyanide persisted in the blood and brain for 2 hours after exposure to 50 and 80 mg/kg of acrylonitrile and remained at the same level after 4 hrs from exposure to 115 mg/kg of acrylonitrile. Thus, the authors concluded that an increase in cyanide levels result from inability of the liver to detoxify acrylonitrile due to depletion of GSH (Figure 2). ¹⁰¹

Based on the outcomes of dermal and ocular irritation studies, in which acrylonitrile administration to rabbits at 0.5 mL (skin) or 0.1 mL (eye) produced erythema, edema, corneal opacity, and severe conjunctival irritation, acrylonitrile is classified as a skin and eye irritant.^1,3 In addition, the guinea pig maximization test provided clear evidence of skin sensitization following acrylonitrile exposure at 0.2, 0.5 and 1%. ³

In vitro genotoxicity

Acrylonitrile induced mutations, chromosomal damage (chromosomal aberrations and sister chromatic exchange), cell transformation, and elicited gene conversion in the yeast assays for chromosomal segregation effects, and cultured mammalian cell assays, while results in bacterial gene mutation (Ames) assay were inconsistent and did not reveal a clear effect (Table 9).^1,3,116,135 Despite positive findings indicative of chromosomal damage, negative results were obtained in DNA repair assay in rat hepatocytes and human mammary epithelial cells.

OPEN IN VIEWER

Table 9.

Summary of findings for acrylonitrile in short term in vitro genotoxicity assays.

Assay	Results		Dose
	- S9	+ S9	LEF or HID
Bacterial gene mutation assay:
Salmonella typhimurium TA97, reverse mutation	-	-	2500 μg/ml
Salmonella typhimurium TA98, reverse mutation	-	-	3333 μg/ml
Salmonella typhimurium TA100, reverse mutation	-	+	500 μg/ml
Salmonella typhimurium TA100, reverse mutation	-	-	2500 μg/ml
Salmonella typhimurium TA102, reverse mutation	-	-	2500 μg/ml
Salmonella typhimurium TA1530, reverse mutation	-	+	2.5 μg/ml
Salmonella typhimurium TA1530, reverse mutation	N/T	+	372 μg/ml
Salmonella typhimurium TA1535, reverse mutation	-	+	0.05 μg/ml
Salmonella typhimurium TA1535, reverse mutation	-	+	167 μg/ml
Salmonella typhimurium TA1535, reverse mutation	-	-	2500 μg/ml
Salmonella typhimurium TA1537, reverse mutation	-	-	2500 μg/ml
Salmonella typhimurium TA1538, reverse mutation	-	-	5000 μg/ml
Escherichia coli WP2 uvrA, reverse mutation	-	-	5000 μg/ml
Escherichia coli WP2 uvrA, reverse mutation	+	+	53 μg/ml
Escherichia coli WP2 uvrA/pKM101	-	-	5000 μg/ml
Escherichia coli WP2 uvrA polA, reverse mutation	+	+	5.3 μg/ml
Gene conversion in yeast:
Saccharomyces cerevisiae D7, gene conversion	+	-	25 μg/ml
Saccharomyces cerevisiae JD1, gene conversion	-	+	250 μg/ml
Saccharomyces cerevisiae PV-2 and PV-3, gene conversion	-	-	800 μg/ml
Saccharomyces cerevisiae D7-144, gene conversion	+	+	0.8 μg/ml
Saccharomyces cerevisiae PV4a and PV4b, homozygous	-	-	800 μg/ml
Saccharomyces cerevisiae D5, forward mutation	+	N/T	30 μg/ml
Saccharomyces cerevisiae D6 and D61-M, homozygosis	+	+	20 μg/ml
Saccharomyces cerevisiae PV-1, forward mutation	-	-	800 μg/ml
DNA strand break:
Syrian hamster embryo cell	+	N/T	200 μg/ml
F344 primary rat hepatocyte	+	N/T	66 μg/ml
Chinese hamster ovary cells	+	+	3710 μg/ml
Human bronchial epithelial cells	+	N/T	200 μg/ml
Unscheduled DNA synthesis:
F344 primary rat hepatocyte	-	N/T	530 μg/ml
Secondary cultures of human mammary epithelial cells	-	N/T	53 μg/ml
TK6 mutation assay:
Human lymphoblast, tk locus	-	+	40 μg/ml
HPRT assay:
Chinese hamster ovary cells	+	+	200 μg/ml
Mouse lymphoma assay:
L5178Y cells, tk locus	+	+	100 μg/ml
P388 F cells, tk locus	-	+	161 μg/ml
L5178Y cells, hprt locus	±	±	200 μg/ml
Gene mutation:
BALB/c 3T3 cells, Na+/K+ ATPase locus	N/T	+	40 μg/ml
Human lymphoblastoid AHH-1 cells, hprt locus	+	N/T	25 μg/ml
Sister chromatid exchange:
Chinese hamster ovary cells	±	+	30 μg/ml
Chinese hamster ovary cells	-	+	106 μg/ml
Rat liver RL4 cells	-	N/T	5 μg/ml
Human lymphoblast	-	±	26.5 μg/ml
Human bronchial epithelial cells	+	N/T	150 μg/ml
Chromosomal aberrations:
Chinese hamster ovary cells	+	+	53 μg/ml
Chinese hamster lung cells	+	N/T	6.2 μg/ml
Chinese hamster liver cells	+	N/T	12.5 μg/ml
Rat liver RL4 cells	-	N/T	10 μg/ml
Human lymphocyte	-	N/T	5.3 μg/ml
Cell transformation assay:
BALB/c 3T3 Mouse cells	+	+	7 μg/ml
C3H 10T1/2 mouse cells	+	N/T	6.3 μg/ml
NIH/3T3 mouse cells	+	N/T	12.5 μg/ml
Syrian hamster embryo cells, clonal assay	+	N/T	2 μg/ml
Syrian hamster embryo cells, focus assay	+	N/T	50 μg/ml
Micronucleus assay:
Chinese hamster ovary cells	+	+	1600 μg/ml

+, positive; ±, either positive or negative in several assays; -, negative; Fischer 344 rat strain; HID, highest ineffective dose; LEF, lowest effective dose; N/T, not tested; - S9, without metabolic activation; + S9, with metabolic activation; Data from ECHA 2018 ³ ; IARC 1999 ² ; Walker et al. 2020a,b.^22,23

In the majority of the assays, positive responses were observed only at very high dose levels (3 or 4 times higher than the LOAELs) of acrylonitrile which were associated with cytotoxic effects, indicating its weak mutagenic potential. The induction or enhancement of genotoxicity after metabolic activation, suggests that conversion to CEO may be responsible for the in vitro mutagenicity of acrylonitrile.^2,136

In addition, it is possible that oxidative stress could contribute to positive results in some of the assays. Thus, L5178Y cells utilized by the mouse lymphoma assay are known to be sensitive to oxidative damage due to decreased antioxidant defenses. ¹³⁷

In vivo genotoxicity

In contrast to in vitro findings, many in vivo studies did not report mutagenicity or clastogenicity of acrylonitrile (Table 10).^1–3,116 These include investigations of chromosomal aberrations and formation of micronuclei in somatic cells and studies of heritable chromosome level changes, such as dominant lethality test, in rats and mice after acrylonitrile administration by various routes, including inhalation, oral administration, and intraperitoneal injections. Results in Drosophila mutation assay, unscheduled DNA synthesis, and sister chromatid exchange yielded conflicting results; thus, the significance of these endpoints is unclear (Table 10).

OPEN IN VIEWER

Table 10.

Summary of findings for acrylonitrile in short term in vivo genotoxicity assays.

Assay	Results	Dose
		LEF or HID
Mutations in Drosophila melanogaster:
D. melanogaster, genetic crossing over or recombination	-	805 mg/kg
D. melanogaster, sex-linked recessive lethal mutations	-	3500 ppm
D. melanogaster, somatic mutation	+	265 mg/kg
D. melanogaster, aneuploidy	+	2.7 ppm
Comet assay:
Rat stomach	-	62.5 mg/kg
Rat liver	+	62.5 mg/kg
Rat brain	-	100 ppm
Rat Zymbal’s gland	-	100 ppm
Wild type mouse forestomach	+	2.5 mg/kg
CYP2E1-null mouse forestomach	-	60 mg/kg
Wild type/CYP2E1-null mouse stomach	-	20/60 mg/kg
Wild type mouse liver	+	10 mg.kg
CYP2E1-null mouse liver	-	60 mg/kg
Wild type/CYP2E1-null mouse Harderian gland	-	20/60 mg/kg
Wild type/CYP2E1-null mouse lung	-	20/60 mg/kg
Wild type mouse ovary	-	20 mg.kg
CYP2E1-null mouse ovary	+	10 mg/kg
Unscheduled DNA synthesis:
Rat gastric mucosa	+	23 mg/kg
SD rat lung	±	46.5 mg/kg
F344 rat hepatocytes	+	50 mg/kg
F344 rat hepatocytes	-	75 mg/kg
F344 rat spermatocytes	-	75 mg/kg
SD rat spermatocytes	±	46.5 mg/kg
F344 rat brain	-	50 mg/kg
HPRT assay:
F344 rat splenic T-cells	+	21 mg/kg
F344 rat thymus and splenic T-cells	+	224 mg/kg
B6C3F1 mice thymus and splenic T-cells	+	58 mg/kg
lacZ transgenic mice thymus and splenic T-cells	+
Sister chromatid exchange:
C57BL/6 mouse bone marrow	±	45 mg/kg
C57BL/6 mouse bone marrow	-	60 mg/kg
Chromosomal aberrations:
Mouse bone marrow	-	140 mg/kg
Mouse, spermatogonia	-	140 mg/kg
C57BL/6 mouse bone-marrow	-	30 mg/kg
Swiss albino mouse bone marrow	-	40 mg/kg
SD rat bone marrow	-	40 mg/kg
Micronucleus assay:
Mouse bone marrow	-	30 mg/kg
Wild type/CYP2E1-null mouse erythrocytes	-	20/60 mg/kg
Dominant lethal test:
Male mice	-	140 mg/kg
Male F344 rats	-	60 mg/kg

+, positive; ±, either positive or negative in several assays; -, negative; HID, highest ineffective dose; Fischer 344 rat strain; LEF, lowest effective dose; SD, Sprague-Dawley rat strain; Data from ECHA 2018 ³ ; IARC 1999 ² ; Walker et al. 2020a,b.^22,23

The lack of in vivo genotoxicity can be explained by differences in metabolic activation of acrylonitrile and detoxication/elimination capacities of live animals compared to in vitro systems. As a result, in vivo epoxidation may yield lower levels of CEO, which might be insufficient for the production of DNA damage. In support of this hypothesis, one of the studies reported that induction of unscheduled DNA repair synthesis in the gastric mucosa by acrylonitrile administered to male SD rats as a single oral dose up to 46 mg/kg was associated with dose- and time-dependent GSH depletion (27% of control group). ⁹⁹ Furthermore, sulfhydryl compounds, cysteine, and penicillamine, protected gastric mucosa from DNA damage.

Mutagenic effects evident by dose-related elevations in Hprt mutated colonies of splenic T-lymphocytes were reported by Walker et al.^22,23 in female B6C3F1 mice, male lacZ transgenic mice, and F344 rats. Similar effects were also observed in CYP2E1-null mice; however, the dose of acrylonitrile that produced an effect (60 mg/kg/bw) was lethal to wild-type mice. Based on the comparison of the mutagenic effects with analogous compounds, the mutagenic potential was considered as limited, and similar to another epoxide-forming chemical, butadiene. ²³ Weak mutagenic effects could stem from either direct or indirect DNA damage. Based on the mainly negative results in in vivo studies of acrylonitrile DNA binding in target tissues, direct genotoxicity is unlikely. The mutagenicity might be a result of CEO reactivity or oxidative damage.

No direct production of DNA strand breaks was detected in the target tissues, brain, and Zymbal’s gland, of SD and F344 rats and Harderian gland of wild-type B6C3F1 or CYP2E-1-null mice using a conventional comet assay.^23,24 Positive comet assay results were reported in forestomach and liver in wild-type B6C3F1 but not CYP2E-1-null mice, after administration of acrylonitrile at 20 mg/kg by gavage 5 days/week for 6 weeks, indicating the significance of the oxidative pathway in producing DNA damage by acrylonitrile. ²³ The authors noted that the nature of DNA damage detected by comet assay was difficult to determine, since the assay does not differentiate between single-strand breaks and excision-repair and alkali-labile sites (including those formed by repair of N-7-guanine DNA adducts).

Toxicity of acrylonitrile can also affect the outcomes of genotoxicity testing. Thus, while acrylonitrile produced DNA-strand breaks in the livers of male SD rats, no increase in DNA damage was detected in the glandular stomach (non-target tissue), where mild histopathological changes were found compared to remarkable necrosis of squamous cells and ulceration that were observed in forestomach (target tissue) at 62.5 mg/kg administered orally as 3 doses. ¹³⁸

Suggested Mechanisms of Acrylonitrile Carcinogenicity

The reviewed information available in the literature supports the conclusion that acrylonitrile is a highly toxic carcinogen with a complex MoA for tumorigenicity which may involve multiple mechanisms. The biological significance of DNA binding of acrylonitrile for the carcinogenic MoA of the compound is questionable, since, in contrast to in vitro studies, no promutagenic etheno-adducts were observed in vivo, and N7-OEG adducts were observed in tissues that were not targets for carcinogenicity, for example, liver. Moreover, acrylonitrile in vivo exhibited at most weak mutagenic potential even at concentrations that were above tumorigenic doses. The applicability of genotoxicity findings in the in vitro assays is of questionable relevance to hazard assessment, since in vivo results in the target tissues were mainly negative, and in some studies in which genotoxicity was detected, it was observed in non-target tissues for carcinogenicity, for example, liver. Thus, acrylonitrile carcinogenicity is not likely to stem from direct DNA damage. One of the most extensively reported mechanisms of acrylonitrile carcinogenicity was production of oxidative DNA damage, which can induce a combination of both genotoxic and epigenetic effects, cytotoxicity and consequent sustained inflammatory response.

Regarding brain and forestomach tumorigenicity, neurotoxicity and ulcers of forestomach mucosa were commonly observed effects of acute and subchronic toxicity of acrylonitrile, associated with depletion of glutathione. Moreover, oxidation of acrylonitrile can lead to production of a reactive metabolite CEO, and a cytotoxic metabolite cyanide, both of which were associated with adverse effects found in various studies. In addition, acrylonitrile, cyanide, and CEO can penetrate the blood-brain barrier, accumulating in the target tissue.

The four target sites of acrylonitrile carcinogenicity in the six bioassays were in tissues, which are located at the interface of different microenvironments and provide a barrier protection from injury or stress. The absence of response of other barrier tissues (e.g., lung) could reflect chemical kinetic (e.g., different metabolism of acrylonitrile) and dynamic factors and inherent tumor susceptibility. All of the target tissues are composed of parenchymal and non-parenchymal cellular elements. The predominant cellular component of the induced brain neoplasia was a non-parenchymal component (microglia), whereas in the other three tissues the predominant cellular component was parenchymal. When the integrity of a tissue is compromised, for example, injury to the forestomach wall, the parenchymal cellular elements respond predominantly with regenerative compensatory proliferation to correct the tissue perturbation. The integument-associated tissues undergo similar processes, which protect the auditory (Zymbal’s gland) or ocular orbital (Harderian gland) regions. In the blood-brain barrier tissue (region), the microglia, which is a non-parenchymal cellular component derived from macrophage precursor, plays a predominant role in response to stress and injury to neural cells produced by neurotoxins.^143,144 Unlike astrocytes, which previously were considered to be the target cells of acrylonitrile carcinogenicity, ¹⁸ microglia cells can readily undergo proliferation in response to pro-inflammatory signaling or, in case of depletion, recolonization.^144,157 Microglial cells activated in response to neurotoxins by classical M1 activation, release more pro-inflammatory mediators and reactive oxygen species, exacerbating tissue damage.^157,158

Hyperplastic changes in the Harderian gland were observed in mice after administration of various chemicals.^159–161 Since this gland is also considered to be a site of immune response, among other functions, ¹⁶² it is possible that cell proliferation observed in the gland could be a result of immune response activation.

As evident from available extensive chemokinetic data, the rodent target organs of acrylonitrile carcinogenicity were exposed to acrylonitrile and CEO in a dose-proportional manner. However, the mouse eliminated acrylonitrile and CEO faster, resulting in levels of CEO which were higher in rat blood and brain compared to respective mouse levels. This could contribute to the lack of brain tumors in mice.

Tissue- and species-specific induction of oxidative stress and subsequent inflammation is a likely MoA for brain tumor development in rats, while in the forestomach, direct cytotoxicity of acrylonitrile and its metabolite cyanide together with immunosuppression might be more relevant (Figure 3). For the Zymbal’s and Harderian glands, mechanistic studies investigating a potential acrylonitrile MoA are lacking, although absence of direct and oxidative DNA damage in the Zymbal’s gland after acrylonitrile administration was reported. ²⁴ Sustained stress and cytotoxicity produced by acrylonitrile possibly can be involved in carcinogenicity in these two tissues.

Depletion of GSH after prolonged exposure to a high concentration of acrylonitrile is a potential initiating event in acrylonitrile MoA (Figure 3). Decrease of the detoxication route and the presence of an oxidation route of acrylonitrile metabolism permits increased formation of two metabolites, reactive CEO, as well as cytotoxic cyanide. This could lead to accumulation of oxygen-derived free radicals and mitochondrial damage leading to ATP depletion, as shown with long-term exposure of rat astrocytes and glial cells to acrylonitrile.^95,96

Also, oxidative stress is exacerbated by depletion of GSH, which has antioxidant properties. Cytotoxicity and increased oxidative stress may contribute to tissue injury, particularly in the brain, which lacks a lymphatic vascular system. Production of pro-inflammatory mediators induces inflammatory conditions, which can lead to further cell and tissue injury and death or apoptosis of contiguous cells. Subsequently, local tissue immunocytes (e.g., microglia and other histiocytes) are recruited. The extent of immunocyte recruitment differs from site to site. In the case of microglia, immunocyte involvement is significant. In response to the injury, cells undergo compensatory regenerative cell proliferation (hyperplasia), which was observed in all target organs of carcinogenicity with chronic dosing (Tables 3–7). Such prolonged administration of an injurious agent leads to cell proliferation, and in turn, proliferating cells can transform and progress to neoplasia, as reported with numerous other chemical carcinogens. ¹⁶³ To support this hypothesis, time course studies are needed.

It is possible that in addition to direct and indirect cytotoxicity of acrylonitrile and oxidative stress, neoplastic conversion of the target cells can develop through the induction of cell proliferation signals leading to increased DNA synthesis, in combination with inhibition of an apoptosis signal, as reported by Pu et al. ¹⁵² Again, mechanistic studies are required to support this hypothesis.

A direct local injurious effect of acrylonitrile dosing may occur in the forestomach barrier site (the portal of entry of acrylonitrile in five of the six bioassays), where neoplasia developed in both rodent species of both sexes in a similar dose and time-related fashion. Injury was evident by formation of erosions of the forestomach as well as hyperplasia/hyperkeratosis of forestomach and glandular stomach mucosa, which were also observed after a shorter duration of administration of acrylonitrile. In the other systemically accessible sites, the species differences in the permeability of the tissue and the speed of acrylonitrile clearance may have influenced development of the neoplastic response, that is, the absence of brain and Zymbal’s gland tumors in mice and Harderian gland tumors in rats.

The combination of the different sustained stresses, which occurred across species, strains, and sexes, was exacerbated in some studies by a significant chronic reduction in water intake. This was pronounced in all drinking water studies in rats (Table 2), likely resulting from poor palatability. Reduced food consumption and consequent body weight gain reduction in a dose and time-dependent fashion contributed to compromised survival. It is possible that water-electrolyte imbalance induced by hypohydration and malnutrition could have a co-carcinogenic effect, although, specific evidence supporting this hypothesis is lacking. In the oral gavage and inhalation studies, water intake was increased, indicating better tolerability. Nevertheless, the authors reported that acrylonitrile produced kidney toxicity, which could have exacerbated water-electrolyte imbalance.

The above suggested MoAs are based on extensive data, but a definitive MoA for acrylonitrile carcinogenicity in the four primary target tissues has not been established. For Zymbal’s and Harderian glands, evidence to conduct a WoE evaluation is limited or missing. The postulated MoAs should be further investigated by short-term in vivo experiments in both rats and mice, including evaluation of genomic profiles in target tissues. Other key events, such as epigenetic alterations and receptor modulations, also warrant investigation. In addition, biological consequences of covalent binding of acrylonitrile to proteins are not well understood; thus, it would be important to identify the relationship between protein adducts and potential alterations of protein function and their effect on signaling networks.

Weight of evidence evaluation for acrylonitrile MoA

The US EPA/IPCS animal MoA guidance ³³ was applied to two of the target tissues of acrylonitrile carcinogenicity, forestomach and brain, to evaluate weight of evidence for MoAs of acrylonitrile carcinogenicity suggested above (Table 11).

1. Biological concordance: The hypothesized MoA does not conflict with the broader biological knowledge. Although, the contribution of GSH depletion to chemical-mediated carcinogenicity is not established, it is known that GSH plays an important role in the detoxication of xenobiotics and protects tissues from oxidative stress. On the other hand, sustained cytotoxicity that was associated with glutathione depletion by acrylonitrile in gastric tissue and brain and resulting compensatory cell proliferation is a well-established MoA for many chemical carcinogens. ¹⁶³

OPEN IN VIEWER

Table 11.

Weight of evidence analysis for acrylonitrile MoAs.

Evolved Bradford Hill considerations a	Forestomach	Brain
Biological concordance	+	+
Essentiality of key events	+	+
Concordance of empirical observations
Dose–response relationships among key events	+	+
Temporality among key events	+	±
Incidence between key events and end (adverse) effects	+	+
Consistency	+	+
Analogy	+	±

+, supporting data; ±, inconsistent data; -, missing data.

^aAdapted from Meek et al. 2014. ³³

Human biological plausibility

The lack of relevance to human cancer hazard for a variety of rodent tumor responses is discussed by Williams et al. ²⁵ Evaluation of possible MoAs for acrylonitrile in all four target tumor sites in rodents (Figure 3), as well as assessment of acrylonitrile using the key characteristics of carcinogens for use in hazard assessment ³¹ (Table 12), supports the conclusion that the human relevance of the induction of these tumors is doubtful. While direct and indirect (via oxidative damage) DNA damage is known to occur in humans, a lack of relevance of the acrylonitrile-induced tumors in rats and mice to humans and differences in metabolism of acrylonitrile between rodents and humans provides little support for human hazard. Moreover, current human chronic exposures to acrylonitrile in occupational settings are much lower ⁷ compared to the carcinogenic doses in rodents.

OPEN IN VIEWER

Table 12.

Summary of evidence for key characteristics of carcinogen based on mechanistic data for acrylonitrile described in literature.

Key characteristica	Forestomach	Brain	Zymbal’s gland	Harerial gland	Other tissues
Is electrophilic or can be metabolically activated	+	+	N/R	N/R	+ (liver)
Is genotoxic	+ in vitro± in vivo	+ in vitro- in vivo	- in vivo	- in vivo	+ in vitro± in vivo
Alters DNA repair or causes genomic instability	N/R	N/R	N/R	N/R	± in vitro- in vivo
Induces epigenetic alterations	N/R	+ in vitro	N/R	N/R	+ in vitro (leukocytes)
Induces oxidative stress	N/R	+ in vitro+ in vivo	- in vivo	N/R	- (liver)
Induces chronic inflammation	+	+	N/R	N/R	+
Is immunosuppressive	+	N/R			+ (spleen)
Modulates receptor-mediated effects	N/R
Causes immortalization	N/R
Alters cell proliferation, cell death, nutrient supply	±	+	N/R	N/R	N/R

Symptoms of acrylonitrile toxicity in workers were associated with high blood levels of cyanide ranging from 1.9 mg/L to 4.3 mg/L, suggesting that oxidative metabolism of acrylonitrile in humans is more active compared to rodents. ¹⁷⁰ However, under environmentally-relevant low-dose exposures, acrylonitrile metabolism to CEO is limited by liver blood flow and alveolar ventilation rate rather than intrinsic rate of metabolism. ¹⁷¹ Another important species difference in the metabolism of acrylonitrile is the presence of genetic polymorphism in human population. Investigation of hemoglobin adducts associated with acrylonitrile exposure suggested that in humans GSTP1 is more important for detoxication of acrylonitrile compared to GSTM1 subunit which is selective for acrylonitrile binding in rats,^67,69 and potential role of variation in GST genotypes might play a role in susceptibility of humans to acrylonitrile toxicity. ⁶⁸ Specifically, polymorphism of the GSTP1 and GSTT1, which is associated with deficient conjugation, might play a role. No significant influences of CYP2E1 polymorphism were found on acrylonitrile-specific hemoglobin adduct levels in humans, except for a trend pointing to higher levels of adducts with the A-316G mutations, nevertheless, slower metabolism of acrylonitrile via CYP2E1-mediated pathway could be associated with higher adduct levels. ⁸⁰

Three of the target tissues, that is, forestomach and Harderian gland (both present in rats and mice) and Zymbal’s gland (present and described commonly in rats, but only in a vestigial form in mice),are absent in humans. Of these three target tissues, applying data on rodent forestomach tumors (occurring in both sexes of rats and mice) to the prediction of cancer risk in humans has been a particularly controversial aspect of interspecies extrapolation. ⁴⁶ Since the forestomach is lined with squamous epithelium similar to the pharynx and esophagus, some argue that forestomach tumors in rodents may be relevant to human hazard.^164,172 It is generally accepted that genotoxic chemicals that elicit forestomach neoplasia in the absence of forestomach irritation and inflammation are considered to be more likely relevant to human risk.^46,173 Acrylonitrile does not have this property. Additionally, the most relevant exposure route in humans is inhalation and not the alimentary route. No increase in forestomach tumors was observed in the carcinogenicity bioassay with acrylonitrile administration by inhalation, ³⁹ indicating a role for direct contact.

In contrast to observations of acrylonitrile-associated brain tumors in several rat bioassays,^13,14 none of several epidemiological studies reported increases in brain tumors in populations exposed to acrylonitrile in occupational settings.^5,6,10–12 The acrylonitrile-associated brain tumors in rats were initially diagnosed by conventional histopathology as gliomas derived from astrocytes (astrocytomas). Re-examination of these tumors utilizing immunohistochemical markers revealed that they were primarily composed of microglia, ¹⁸ components of glia that function as macrophages and are important for the maintenance of the blood-brain barrier. In humans, the most common malignant primary brain tumor is glioblastoma, which is primarily derived from the astrocytic lineage.^174–176 While microglia represent a major non-neoplastic component of gliomas, ¹⁵⁸ presently, only two cases of true microglioma in humans have been reported.^174,177 This difference in cellular susceptibility may explain the absence of central nervous system carcinogenicity of acrylonitrile in humans. ¹⁷⁸ Thus, in an evaluation of carcinogenic risk to humans, IARC ²⁷ found that the evidence for the carcinogenicity of acrylonitrile was very limited, with brain tumors being highly infrequent in the epidemiological studies. In addition, the human blood-brain barrier offers more effective protection from toxins compared to either rats or mice.^179,180

No macroscopic or histologic changes in the lungs were observed in rats in a 2-year inhalation study of acrylonitrile, although degenerative and inflammatory changes in the nasal passages were observed. ³⁹ As mentioned above, a slight increase in combined incidences of alveolar/bronchiolar adenoma or carcinoma that may have been related to acrylonitrile exposure was observed only in female mice and was not dose dependent.^15,16 In humans, while some studies detected an association between acrylonitrile exposure and increased risk of lung cancer in humans, ³⁵ others have shown that these findings are unsupported based on the absence of dose-response and lack of proper adjustments for confounding factors, such as smoking. ⁴

Currently, the occupational exposure levels to acrylonitrile are estimated to be below 2 ppm.^7,28 Assuming that the highest allowed occupational exposure to acrylonitrile is 2 ppm or 4.3 mg/m³ for 8-hours over a 40-hour work week,^28,30 an average of 261 workdays per year and work duration of 10 years, an average lifetime daily exposure to acrylonitrile is estimated to be 146.4 μg/m³ (lifetime daily exposure = 4.3 mg/m³ x (8/24) hrs x (261/365) days x (10/70) years). Using the inhalation cancer unit risk factor of 1.02 × 10⁻⁶ per μg/m³ calculated above based on the animal data, an estimated excess lifetime cancer risk from acrylonitrile exposure is 1.49 × 10⁻⁴ (risk = 146.4 μg/m³ x (1.02 × 10⁻⁶ per μg/m³)) or approximately 0.015%, which is negligible. Even when calculations are corrected for a 40-year career-long occupational exposure, the lifetime cancer risk does not exceed 0.06%.

In conclusion, the suggested MoA of acrylonitrile carcinogenicity in rodents is consistent with direct and indirect (due to oxidative damage) cytotoxicity, and compensatory cell proliferation, although weak, likely indirect, mutagenicity cannot be ruled out. Overall, the relevance to humans of findings with acrylonitrile in rodent studies is questionable and requires further dose-effect and mechanistic investigation. Combined with the largely negative epidemiology in humans and margin of exposure, human risk is unlikely.