Reevaluation of a Twenty-Four-Month Chronic Toxicity/Carcinogenicity Study of Hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) in the B6C3F1 Hybrid Mouse

Abstract

Hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) has been widely used as an explosive in U.S. army munitions formulations since World War II. Two-year carcinogenicity studies revealed RDX to be noncarcinogenic in two strains of rats, but a 2-year carcinogenicity study in B6C3F1 mice revealed an increased incidence of hepatocellular neoplasms in females. Based on results of the study in B6C3F1 mice, RDX has been classified as a possible carcinogen. The authors reevaluated the archived histological sections from the B6C3F1 mouse study, using current histopathologic diagnostic criteria and interpretations. The earlier evaluation showed a statistically significant increase in the incidence of hepatocellular adenoma/carcinoma in female mice from the three highest dose groups (7, 35, and 175/100 mg/kg/day). The revaluation yielded a slightly lower incidence at each of the dose levels in female mice. The reduced number of hepatocellular neoplasms was largely due to reclassification of hepatocellular adenomas as foci of cytoplasmic alteration, in compliance with current diagnostic criteria. The reevaluation was reviewed by a pathology working group (PWG), which arrived at a consensus classification of each lesion. Based on the consensus diagnoses of the PWG, only one female group (35 mg/kg/day) showed a significant increase when compared to controls. The incidence of hepatocellular neoplasms for all groups, including the 35 mg/kg/day group, was within the reported incidence range for spontaneous hepatocellular neoplasms in female B6C3F1 mice. The increased incidence of hepatocellular neoplasms in female mice given RDX at 35 mg/kg/day was interpreted as equivocal evidence of a carcinogenic effect.

Hexahydro-1,3,5-trinitro-1,3.5-triazine (RDX) is a cyclic polynitramine that has been widely used as an explosive in U.S. Army munitions formulations since World War II. The high level of production and widespread dispersal of the compound have resulted in considerable interest in the potential toxicity of RDX (McLellan et al. 1992; ATSDR 1995). Studies of the chronic toxicity/carcinogenicity of RDX in Sprague-Dawely rats (Hart 1977), Fisher 344 rats (Levine et al. 1983), and B6C3F1 mice (Lish et al. 1984) were performed during the 1970s and early 1980s. RDX has been classified as a possible carcinogen (USEPA 2005a) largely due to evidence of a treatment-related increase in hepatocellular neoplasms in female, but not male, B6C3F1 mice (Lish et al. 1984).

Evidence of RDX contamination of ground water near U.S. military installations (Walsh and Jenkins 1992; Major 1999) has rekindled interest in RDX as a potential carcinogen. The U.S. Environmental Protection Agency (EPA) cancer risk assessment process has changed since 1991, and recently EPA proposed new guidelines in the evaluation of carcinogenicity of chemicals (EPA 2005b). Reevaluation of previous carcinogenicity data is not unprecedented. Toxaphene, which was considered a category B2 carcinogen, was reevaluated utilizing current criteria for the classification of hepatocellular adenomas and carcinomas, resulting in a recommendation that the cancer potency factor should be reduced (Goodman et al. 2000). Gentoxicity studies of RDX revealed that it was not genotoxic in vitro or in vivo test systems (Reddy et al. 2005). The present reevaluation of the chronic toxicity/carcinogenicity study that was conducted in B6C3F1 mice in the early 1980s was performed to confirm the original histopathologic findings in the liver of female mice based on current histopathologic diagnostic criteria and interpretations (Harada et al. 1999).

MATERIAL AND METHODS

Original Study Design

Groups of 85 each male and female B6C3F1 mice were given RDX in the feed at dosage levels of 0, 1.5, 7, 35, or 175 mg/kg/day. The dosage level in the high-dose group was reduced to 100 mg/kg/day at study week 11 due to excess mortality in that group. Ten mice per sex/dose group were sacrificed at 6 and 12 months. Surviving mice were sacrificed by carbon dioxide anesthesia and exsanguination at the time of study termination at 24 months. The study was conducted at IIT Research Institute, Chicago, IL.

The original histologic sections from female mice were retrieved from the archives of the original contract laboratory and a second examination was performed by one of the authors (GAP).

Poststudy Review and Pathology Working Group Review

In preparation for the Pathology Working Group (PWG) review, the PWG Chairperson reviewed the pathology data from the initial study, the initial study report that included the pathology narrative, and histologic sections stained with hematoxylin and eosin (H&E) from all tissues of female mice. The PWG Chairperson then selected histologic sections for review by the PWG. The selected histologic sections included all liver lesions in which the study pathologist (Lish et al. 1984) or reviewing pathologist (GAP) observed hepatocellular neoplasms or lesions that had some histologic similarity to hepatocellular neoplasms. The PWG members reviewed the histologic sections without knowledge of treatment group or previous diagnosis by the study pathologist or the reviewing pathologist. Members of the PWG reviewed each individual histologic section in conference with the other PWG members, and provided a diagnosis for each reviewed lesion. A simple majority vote was used to determine the consensus diagnosis of the PWG regarding each lesion.

The following individuals participated in the PWG: Drs. Gary Boorman (National Institute of Environmental Health Sciences, Research Triangle Park, NC [NIEHS]), Ronald Herbert (NIEHS), James Hailey (NIEHS), George Parker (Biotechnics, Inc.), Douglas Wolf (U.S. Environmental Protection Agency, Research Triangle Park, NC), Roxanne Baumgartner (U.S. Army Center for Health Promotion and Preventive Medicine (CHPPM), and Michael Major (CHPPM). Drs. Baumgartner and Major were nonvoting observers of the PWG.

RESULTS

Original Study Results (Lish et al. 1984, available from Defense Technical Information Center [DTIC])

Survival

By study week 10 there was reduced survival of males in the high dose (175 mg/kg/day). Beginning with study week 11 the dosage level for both sexes was reduced from 175 to 100 mg/kg/day, after which survival of high dose males was similar to that of controls. Overall survival of treated females from all dosage groups and males from the lower dosage groups was similar to that of controls.

Clinical Observations

Males in the 175/100 mg/kg/day group exhibited aggressive behavior, including fighting that resulted in numerous skin wounds. Male animals were removed from gang caging to prevent such occurrences.

Body Weights

Body weights of males from the 175/100 mg/kg/day group were significantly less than control values on weeks 95, 101, 103, and 104 (95.6%, 95.0%, 95.0%, and 95.0% of control values, respectively). Body weights of females from the 175/100 mg/kg/day group were significantly less than control values beginning on week 10, and remained lower than control values through the remainder of the study (95.0%, 90.0%, 88.5%, 82.3%, and 80.8% of control values at weeks 10, 51, 69, 101 and 104, respectively). There was no alteration in feed consumption that would explain the altered body weights in males or females.

Necropsy

Unscheduled death mice (30 males and 36 females) from the 175/100 mg/kg/day group that died during the first 6 months of the study commonly had dark red mottled lungs, dark red spleen, and dark red liver, which suggested terminal cardiovascular dysfunction of uncertain specific pathogenesis. The urinary bladder of unscheduled death males from the 175/100 mg/kg/day group was commonly distended with red, yellow, or brown fluid. Terminal sacrifice female mice from all treated groups had an increased incidence of grossly visible hepatic masses and nodules as compared to controls. Tabulations of gross necropsy observations indicated the number of animals with hepatic masses or nodules, but did not indicate the total number of hepatic masses or nodules that were present in individual animals or all animals of a group. Histologic observations during the reevaluation indicated that a number of RDX-treated mice had multiple hepatic masses.

Histopathology

Male mice from all dosage groups that died spontaneously or were sacrificed for humane reasons prior to the scheduled 6-month sacrifice had cytoplasmic vacuolization of renal tubular epithelial cells. The exact pathogenesis and significance of the renal vacuolization was not apparent from histopathologic examination, but the severity of the change suggested it was not the primary cause of death or debilitation. Renal lesions were not observed in male mice from the 12- or 24-month scheduled sacrifices. Lungs of the unscheduled death mice commonly had histologic evidence of vascular congestion, which was attributed to terminal cardiovascular dysfunction of uncertain specific pathogenesis. There was no histologic correlate of the gross necropsy observations in the spleen, liver, and urinary bladder of the unscheduled death mice. At the 24-month sacrifice, male mice from the 35 and 175/100 mg/kg/day groups had testicular changes consisting of necrosis of germinal epithelium, interstitial fibrosis, and aspermia. At the 24-month sacrifice, there was a statistically significant (multivariate statistical methods described by Bock 1975) increase in the incidence of hepatocellular adenoma/carcinoma in female mice from the 7, 35, and 175/100 mg/kg/day groups.

Pathology Working Group Results

The diagnosis of the PWG participants regarding each reviewed lesion was recorded and simple majority opinion was considered to be the final PWG consensus diagnosis. The incidence of hepatocellular proliferative lesions as recorded by the study pathologist and PWG members are presented in Table 1. PWG consensus classifications of non-neoplastic hepatocellular alterations are presented in Table 2.

Proliferative liver lesions were classified by current diagnostic criteria (Harada et al. 1999). Foci of cytoplasmic alteration consisted of clusters of hepatocytes that had altered cytoplasmic tinctorial affinity. Foci of cytoplasmic alteration were subclassified as basophilic, eosinophilic, mixed, or clear cell foci. Hepatocellular adenomas were well circumscribed masses composed of orderly plates of hepatocytes, and typically compressed the surrounding hepatic parenchyma around much of the perimeter of the neoplasm (Figure 1). Hepatocytes in hepatocellular adenomas exhibited little cellular atypism but occasionally had cytoplasmic vacuolization or variation in cytoplasmic tinctorial affinity. Hepatocellular carcinomas tended to be larger than adenomas, often had irregular borders, and exhibited cellular atypism that ranged from mild to severe (Figure 2). Hepatocytes in hepatocellular carcinomas often formed trabeculae, rosettes or organoid patterns, and the neoplasms commonly had areas of necrosis and/or hemorrhage (Figure 3). Pulmonary metastasis was present in two mice from the 175/100 mg/kg/day group that had hepatocellular carcinomas (Figure 4).

The study pathologist’s impression of a treatment-related increase in the incidence of hepatocellular neoplasms in all RDX-treated female groups was supported by PWG consensus diagnoses, but the magnitude of the treatment-related effect was reduced by reclassification of a number of hepatocellular lesions. The review revealed no liver neoplasms that were not recorded by the study pathologist during the initial histopathologic evaluation of the study.

DISCUSSION

Diagnostic criteria for classification of hepatocellular lesions were not included in the initial study report. The initial report included the standard terminology that is currently in use, but changes in diagnostic criteria during the 20-year period since the study was initially evaluated resulted in reclassification of a number of hepatocellular proliferative lesions. Nine hepatocellular lesions that were initially classified as neoplasms by the study pathologist were considered by the PWG members to be non-neoplastic lesions. The reduced number of hepatocellular adenomas as tabulated by the study pathologist resulted primarily from reclassification of hepatocellular adenomas as foci of cytoplasmic alteration. Although there is no formal, documented evidence of this trend in the literature, prolonged experience of one of the authors (GAP) suggests there has been a slight increase in the stringency of the diagnostic criteria required for classification of hepatocellular lesions as adenoma, resulting in an increased number of borderline lesions being relegated to the stature of focus of cytoplasmic alteration.

Necropsy and histology processing records were not available during the reevaluation, thus it was impossible to determine how many grossly visible lesions were present in the liver, nor whether all grossly noted liver lesions were represented in the histologic sections. There were no available laboratory records to indicate that grossly observed liver lesions were tracked through histology and included in the final histologic sections.

The control females in the subject study had a low incidence of hepatocellular neoplasms. Historical incidence data published by the National Toxicology Program (NTP) in 1987, based on data derived from 1781 untreated control female B6C3F1 mice from 59 dietary administration carcinogenicity studies conducted at six subcontract laboratories, indicated that hepatocellular adenomas and carcinomas occurred in B6C3F1 female mice at mean percentage incidence levels of 4% and 5%, respectively, or 8% for adenomas and carcinomas combined (Maronpot 1987). That report from 1987, which would include the time during which the subject RDX study was conducted, indicated a substantial range in the incidence of hepatocellular neoplasms, with hepatocellular adenomas ranging from 0% to 18%, hepatocellular carcinomas ranging from 0% to 15%, and hepatocellular adenoma/carcinoma combined ranging from 0% to 20% of 1781 control female B6C3F1 mice. Another review of neoplasms in control mice from 54 NTP carcinogenicity studies conducted in B6C3F1 mice at five contract laboratories revealed the mean incidence of liver neoplasms in females to be 6.2%, with a range of 0% to 21% (Tarone, Chu, and Ward 1981). Analysis of interlaboratory variation in incidence rates in the latter report revealed significant (p < .05) heterogeneity in the incidence of liver neoplasms in various laboratories (Tarone, Chu, and Ward 1981). Inspection of tabulated data from the latter report revealed that the narrowest range (2% to 10%) in the incidence of liver neoplasms in female B6C3F1 mice occurred at one of the contract laboratories that performed the smallest number (7) of carcinogenicity studies, and that the broadest range (0% to 21%) occurred at the laboratory that performed the largest number (22) of carcinogenicity studies. These observations suggest a wide range in the incidence of spontaneous liver neoplasms in untreated control female B6C3F1 mice. Based on these observations from the NTP experience, the presence of a single hepatocellular adenoma in the 67 control female B6C3F1 mice in the subject study, although not a uniquely low incidence, was a notably low incidence rate (1.49%) for hepatocellular adenomas in B6C3F1 female mice from a 2-year carcinogenicity study. The incidence of hepatocellular neoplasms in RDX-treated female B6C3F1 mice from the subject study were well within the published incidence range of hepatocellular neoplasms in female B6C3F1 mice from carcinogenicity studies during the time frame that the subject study was conducted.

When a control group has an unusually low incidence of a particular lesion, it is common practice to review the fixed tissue remnants of animals from all dosage groups to insure that all grossly visible lesions were included in the histologic sections (Goodman et al. 2000). It was not possible to perform such a tissue specimen review on the present study, as fixed tissue remnants were not available. There were no available laboratory records to indicate that a fixed tissue review took place at the time the study was initially performed. Although there was no indication that an incomplete gross or microscopic examination was performed during the initial conduct of the study, the reassurance of a confirmatory wet tissue review is not available.

Observations regarding technical aspects of the liver sections, though not of major significance on an individual basis, served to arouse some concern with regard to the uniformity histologic processing. The liver sections from many animals were smaller than usual for a carcinogenesis study and only one liver section was present from most animals, as opposed to the minimum of two liver sections that are commonly examined in current carcinogenesis studies. Variation in size and shape suggested the liver sections were not uniformly taken from the same area of the liver of all animals. Most significantly, there was variation in the number of liver sections that were prepared for individual animals.

Variation in the number of liver sections was of particular concern due to the obvious potential impact of a group-related variation in specimen selection and histologic sectioning procedures. In an attempt to address this concern, liver sections of all animals were reviewed a second time with the specific purpose of determining whether there was any group-related bias in the number and/or size of liver sections. This review of the histologic sections revealed that two liver sections were prepared from five females from the control group, nine females from the 7 mg/kg/day group, three females from the 35 mg/kg/day group, and two females from the 175 mg/kg/day group. Four liver sections were prepared from one female from the 35 mg/kg/day group. The remaining females had only one liver section per animal. The increased numbers of liver sections were suspected to be related to gross necropsy observations, as the additional sections typically contained a neoplasm that should have been large enough to be visible at necropsy. As indicated above, absence of necropsy records and histology processing records hindered this evaluation. It was concluded that there was no overt evidence of group-related bias in preparing the liver sections.

Though the purpose of the PWG review was primarily to determine the accuracy of the pathology data, there was discussion as to the weak nature of the evidence supporting RDX-associated hepatic carcinogenicity. Factors that reduced confidence in a positive interpretation of the study included absence of treatment-related precursor lesions such as foci of cellular alteration and, most importantly, the low incidence of hepatocellular neoplasms in the control females.