&EPA United States Environmental Protection Agency Office of Health and Environmental Assessment Washington DC 20460 EPA/600/8-82/004F ]- September 1985 Final Report Research and Development Addendum to the Health Assessment Document for Dichloromethane (Methylene Chloride) Updated Carcinogenicity Assessment of Dichloromethane (Methylene Chloride) ------- EPA/600/8-82/004F September 1985 Final Report ADDENDUM TO THE HEALTH ASSESSMENT DOCUMENT FOR DICHLOROMETHANE (METHYLENE CHLORIDE) Updated Carcinogenicity Assessment of Dichloromethane (Methylene Chloride) ffiiioago, Office of Health and Environmental Assessment Office of Research and Development U.S. Environmental Protection Agency Washington, J.... ------- DISCLAIMER This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. ------- CONTENTS Preface v Authors, Contributors, and Reviewers vi 1. SUMMARY AND CONCLUSIONS 1 1.1. SUMMARY 1 1.1.1. Qualitative Assessment 1 1.1.1.1. Total Data Base 1 1.1.1.2. NTP (1985) Inhalation Bioassay 3 1.1.2. Pharmacokinetics/Metabolism 4 1.1.3. Quantitative Assessment 5 1.2. CONCLUSIONS 9 2. INTRODUCTION 11 3. CARCINOGENIC ITY 12 3.1. NATIONAL TOXICOLOGY PROGRAM INHALATION BIOASSAY (1985, DRAFT) . . 12 3.1.1. Rat Study 13 3.1.2. Mouse Study 24 3.1.3. Summary 33 3.2. PHARMACOKINETICS/METABOLISM 36 3.2.1. In Vitro Metabolism/Pathways .36 3.2.2. In Vivo Metabolism/Effect of Dose 40 3.2.3. Use of Pharmacokinetic/Metabol ism Data for Risk Calculation 50 3.2.4. Human Studies 61 3.2.5. Summary 67 4. QUANTITATIVE ESTIMATION (USING THE NTP INHALATION BIOASSAY) 71 4.1. SUMMARY OF THE NTP FINDINGS USED FOR QUANTITATIVE ANALYSIS ... 71 4.2. DOSE-RESPONSE MODEL SE LECTION 74 4.3. APPLICATION OF THE MULTISTAGE MODEL TO THE NTP BIOASSAY DATA . . 75 4.4. RISK ANALYSIS CONSIDERING TIME-TO-TUMOR INFORMATION 81 4.5. COMPARISON OF RISKS ESTIMATED WITH OTHER DOSE-RESPONSE MODELS 87 4.6. COMPARISON OF THE NTP (1985) RESULTS WITH OTHER BIOASSAYS ... 95 4.7. DERIVATION OF HUMAN UNIT RISK ESTIMATES FOR INHALATION OF DCM 99 ill ------- CONTENTS (continued) 4.8. HUMAN UNIT RISK ESTIMATE FOR INGESTION OF DCM 104 4.9. COMPARISON OF ANIMAL AND HUMAN DATA RELEVANT TO CANCER RISK . . 106 REFERENCES 109 ------- PREFACE The Office of Health and Environmental Assessment has prepared this addendum to serve as a "source document" for EPA use. This addendum updates EPA's Health Assessment Document for Dichloromethane (February 1985); the up- date is being prompted by the release of the draft National Toxicology Program (NTP) inhalation bioassay for mice and rats (1985, draft). Originally the Health Assessment Document (HAD) was developed for the eventual use of the Office of Air Quality Planning and Standards; however, at the request of the Agency, the assessment scope was expanded to address multimedia aspects. The addendum reviews the findings of the NTP draft Technical Report on Carcinogenesis Studies of Dichloromethane (February 1985) with recognition of the changes that were recommended for the report at the August 1985 review meeting of the NTP Board of Scientific Counselors. The addendum refers to the remainder of the toxicology data base as detailed in the HAD. The addendum contains an analysis, not previously contained in the February HAD, that evaluates the reasonableness of using available metabolism and pharmacokinetic data in the development of cancer unit risk estimates. Lastly, the addendum presents an updated derivation of inhalation cancer unit risk values and recommends, for the first time, an estimate of unit risk for ingestion exposure. If a review of the health information indicates that the Agency should consider regulatory action for this substance, a considerable effort will be undertaken to obtain appropriate information regarding the extent of exposure to the population. Such data will provide additional information for the development of regulatory options regarding the extent and significance of the public hazard associated with this substance. ------- AUTHORS, CONTRIBUTORS, AND REVIEWERS The Carcinogen Assessment Group within the Office of Health and Environ- mental Assessment was responsible for preparing this document. PRINCIPAL AUTHORS Dharm V. Singh, D.V.M., Ph.D. Chapters 1 and 3 Hugh L. Spitzer, B.A. Chapters 1, 2, and 3 Paul D. White, B.A.* Chapters 1 and 4 PARTICIPATING MEMBERS Roy E. Albert, M.D., Chairman Steven Bayard, Ph.D. David L. Bayliss, M.S. Chao W. Chen, Ph.D. Arthur Chiu, Ph.D., M.D. Margaret M.L. Chu, Ph.D. Herman J. Gibb, B.S., M.P.H. Bernard H. Haberman, D.V.M., M.S. Charalingayya B. Hiremath, Ph.D. James W. Holder, Ph.D. Robert E. McGaughy, Ph.D., Acting Technical Director William E. Pepelko, Ph.D. Charles H. Ris, P.E., Acting Executive Director Todd W. Thorslund, Ph.D. REVIEWERS The following individuals provided peer review of the Pharmacokinetics/ Metabolism section of this document. Andrew G. Ulsamer, Ph.D. Acting, Associate Executive Director Health Sciences U.S. Consumer Product Safety Commission Washington, DC 20207 Linda S. Birnbaum, Ph.D. National Toxicology Program Department of Health and Human Services Research Triangle Park, NC 27709 *Exposure Assessment Group, Office of Health and Environmental Assessment. vi ------- Russell Prough, Ph.D. Department of Biochemistry University of Texas Health Science Center, Dallas Dallas, TX 75219 Marguerite Coomes, Ph.D. Department of Biochemistry Howard University Washington, DC 20059 Ellen O'Flaherty, Ph.D. Department of Environmental Health University of Cincinnati Cincinnati, OH 45267 EPA Science Advisory Board The substance of this document was independently peer-reviewed in public sessions of the Environmental Health Committee of EPA's Science Advisory Board, vn ------- 1. SUMMARY AND CONCLUSIONS 1.1. SUMMARY 1.1.1. Qualitative Assessment 1.1.1.1. Total Data Base—There have been eight chronic studies in which dichloromethane (methylene chloride, DCM) was administered to animals: five in rats, two in mice, and one in hamsters. The Dow Chemical Company (1980) reported the results of chronic inhalation studies in rats and hamsters. There was a statistically significant increased incidence of ventral cervical sarcomas, probably of the salivary gland, consisting of sarcomas only, and appearing in male rats but not in females. In addition, the study showed a small increase in the number of benign mammary tumors compared to controls in female rats at all doses and in male rats at the highest dose. In hamsters, there was an increased incidence of lymphosarcoma in females which was not statistically significant after correction for survival. In a second inhalation study, the Dow Chemical Company (1982) reported that there was no increase in compound-related tumors in rats; however, the highest dose used in this study was far below that of the previous study. The National Coffee Association (1982a, b) conducted a study in which Fischer 344 rats and B6C3F1 mice were exposed to DCM in drinking water. The results indicated that female Fischer 344 rats had an increased incidence of neoplastic nodules and/or hepatocellular carcinomas, which was significant with respect to matched controls; however, the incidence was within the range of historical control values at that labora- tory. The National Coffee Association (1983) drinking water study in B6C3F1 mice also showed a borderline response of combined neoplastic nodules and hepatocellular carcinomas. The National Toxicology Program (1982, draft) gavage study on rats and mice has not been published due to data discrepancies; 1 ------- however, usable information from the gavage studies has been incorporated by the NTP into the inhalation bioassay (1985, draft). The recently released NTP (1985, draft) inhalation bioassay concluded that "there was some evidence of carcinogenicity of dichloromethane for male F344/N rats as shown by increased incidence of benign neoplasms of the mammary gland. There was clear evidence of carcinogenicity of dichloromethane for female F344/N rats as shown by increased incidence of benign neoplasms of the mammary gland. There was clear evidence of carcinogenicity of dichloromethane for male and female B6C3F1 mice, as shown by increased incidences of alveolar/ bronchiolar neoplasms and of hepatocellular neoplasms." There are several other animal studies in the literature, which are noted but considered to be inadequate. One study (Theiss et al., 1977) reported a marginally positive pulmonary adenoma response in strain A mice injected intraperitoneally with DCM. Two negative animal inhalation studies were judged to be inadequate because they were not carried out for the full lifetime of the animals (Heppel et al., 1944; MacEwen et al., 1972). Positive results in a rat embryo cell transformation study were reported by Price et al. (1978). The significance of these findings with regard to carcinogenicity is uncertain at the present time. The epidemiologic data associated with exposure to dichloromethane con- sists of two studies and two updates: Friedlander et al. (1978), updated by Hearne and Friedlander (1981) and Friedlander et al. (1985), and Ott et al. (1983a, b, c, d, e). Although no study shows excessive risk, each study has sufficient limitations to prevent it from being judged as a negative study. The Friedlander et al. (1978) study lacked the combination of size, exposure levels, and follow-up period necessary to provide sufficient statistical power to detect the excess of total cancers predicted using the animal cancer potency ------- estimates. It should be noted that the data in the Friedlander et al. (1985) study provides some suggestion of an increased incidence of tumors of the pancreas in exposed workers; a rigorous review of this study will have to be done at a later date. The Ott et al. (1983a, b, c, d, e) study, among other deficiencies, lacked a sufficient latency period for site-specific cancer. 1.1.1.2. NTP (1985) Inhalation Bioassay—The NTP (1985, draft) inhalation bioassay of DCM was conducted in male and female F344/N rats and B6C3F1 mice. The animals were exposed at concentrations of 0, 1,000, 2,000, and 4,000 ppm for rats and 0, 2,000, and 4,000 ppm for mice, 6 hours/day, 5 days/week, for 102 weeks. There was an increased incidence of benign mammary gland neoplasms and primarily fibroadenomas in both male and female rats. In female rats there was a significant increase in hepatocellular neoplastic nodules and hepatocellular carcinomas (combined) by the trend test only. There was also a statistically significant increase of mononuclear cell leukemias in female rats by age adjustment. In male rats there was a significant increase in mesotheliomas, primarily in the tunica vaginal is. Lastly, a marginally signi- ficant increase was noted in adrenal pheochromocytomas and interstitial cell tumors in male rats and pituitary gland adenomas and carcinomas combined in male and female rats by the trend test only. In the study using B6C3F1 mice, there was a highly significant increase in alveolar/bronchiolar adenoma and/or carcinoma in both sexes of mice. The incidence of hepatocellular adenoma and hepatocellular carcinoma combined was increased in the high-dose male group and in both dosed groups of female mice. It should be noted that there was also a dose-related increase in the number of mice bearing multiple lung and liver tumors. The control mice had no more than one lung tumor per mouse, whereas 38% of all dosed males and 42% of all dosed females had multiple lung tumors. The incidence of multiple hepatocellular ------- tumors in the exposed groups increased in both sexes in a dose-related manner. Multiple hepatocellular tumors were found in only 4% of the male controls, and none were found in the female controls. In contrast, 28% of the exposed males and 32% of the exposed females exhibited multiple liver tumors. The NTP concluded that, under the conditions of this bioassay, there was some evidence of the carcinogenicity of DCM for male F344/N rats as shown by an increased incidence of benign neoplasms of the mammary gland; there was suffi- cient or clear evidence of the carcinogenicity of DCM for female F344/N rats as shown by an increased incidence of benign neoplasms of the mammary gland; there was clear evidence of carcinogenicity in male and female B6C3F1 mice as shown by increased incidences of lung and liver tumors. 1.1.2. Pharmacoki neti cs/Metabol i sm The available data have been analyzed to determine if there are quali- tative or quantitative metabolic differences or similarities between species that may alter the assumptions used in estimating the carcinogenic risk arising from exposure to DCM. This has in part been accomplished by considering several approaches for using the data that is currently available. The results of both in vitro and in vivo studies indicate that DCM is metabolized via two pathways. One pathway yields carbon monoxide as an end product, and the other pathway yields carbon dioxide as an end product with formaldehyde and formic acid as metabolic intermediates. Each pathway involves formation of a metabolically-active intermediate that is theoretically capable of irreversibly binding to cellular macromolecules. A comparative analysis of the capability of various tissues to metabolize DCM indicates that the liver is the primary site of metabolism, with some metabolism taking place in the lung and kidney. An analysis of the available in vivo data suggests that when rats or mice are exposed to high concentrations of DCM they exhale more carbon ------- dioxide and excrete more formic acid than carbon monoxide. At exposure to low concentrations of DCM, both pathways appear to be utilized about equally. At the present time the implications of these observations in assessing the car- cinogenic potency of DCM are unclear. A comparative analysis of the data from in vivo studies in mice, rats, and humans indicates that all three species metabolize DCM to carbon monoxide. Both mice and rats metabolize DCM to carbon dioxide. No human data are avail- able to verify the metabolism of DCM to carbon dioxide. However, based on uptake data, some investigators have speculated that this pathway is functional in humans. At present, the available data are insufficient for the purpose of esti- mating doses at which metabolism is saturated. The data indicate that, at low doses, little unmetabolized DCM is exhaled. At high doses there is a significant exhalation of DCM immediately postexposure. The available data do suggest that at high doses more DCM is taken up into the body. Currently, the data are insufficient to determine the relationship between exposure con- centration and uptake. Several approaches are presented using the available pharmacokinetic/metabolism and modeling data to evaluate the carcinogenic potential of DCM. These are considered, at this time, to be more illustrative of possible approaches than useful in risk estimation because of data deficien- cies and uncertainty about the validity of assumptions. Based on this analysis it is concluded that the available data do not offer useful parameters for modifying the dose assumptions used in the calculation of the carcinogenic unit risk of DCM. 1.1.3. Quantitative Estimation In the previous carcinogenicity evaluation of DCM (U.S. EPA, 1985), a quantitative estimate for the upper-bound incremental unit risk was developed ------- on the basis of salivary gland tumors seen in an inhalation study with male rats. The upper-bound estimate of incremental unit risk has now been re-eval- uated using the results of the NTP inhalation bioassay (NTP, 1985, draft). The new risk calculations presented in this addendum are based primarily on the NTP findings of carcinogenicity in the liver and lung of male and female mice. In mice, both separate and combined analyses were conducted for benign and malignant tumors. Risk calculations are made for mice developing either the lung or liver tumors in order to indicate the total risk associated with tumors of these two organs. The elevated mammary tumor incidence in female rats and mammary and subcutaneous tumor incidence in male rats were also used in the risk analysis. The available metabolic and pharmacokinetic data are inadequate to support modifications to the experimentally applied doses. Thus, risk calculations are based on the experimentally applied doses (in ppm using the study dose schedule), with subsequent adjustment to estimate human equivalent doses and risks. The multistage dose-response model, as incorporated in the GLOBAL83 computer program (with the number of terms restricted to the number of experimental dose groups minus one), is the primary model utilized in the analysis. Both maximum likelihood estimates (MLE) and 95% upper confidence limit (UCL) values for risk are given. The multistage model was found to provide an adequate fit to the experi- mental data for the tumor sites, tumor pathology types, sexes, and species groups examined. The highest estimate of risk was obtained from the UCL value for combined adenoma and carcinoma response in the lung and/or liver of female mice. To provide comparison with the basic multistage risk estimates, addi- tional calculations were made with other risk estimation approaches and models using the data on mice having lung or liver tumors. ------- An analysis, which excluded animals that died before the first tumors developed, produced similar risk estimates (results were within 10% for female mice with lung and/or liver adenomas and carcinomas combined). A time-to-tumor analysis using the multistage model, as formulated in the WEIBULL82 computer program, was also applied to the data to determine if the inclusion of a time term would influence risk estimates. The time-to-tumor estimates for the UCL of risk at low doses were generally in good agreement with the multistage model. The probit and dichotomous Wei bull models, in both background-independent and background-additive formulations (using the RISK81 computer program), were applied to the mouse data for comparison with the multistage model. For com- bined lung and liver tumors in female mice, the background-additive formula- tions of both the probit and Weibull models are in good agreement with the multistage model. The background-independent formulations of the probit and Weibull models lead to much lower risk estimates. The quantitative risk estimates developed from the NTP inhalation bio- assay data are compared for consistency with findings in earlier long-term bioassays of DCM conducted by the Dow Chemical Company and the National Coffee Association. These studies provide some evidence of DCM-induced tumors con- sistent with the NTP findings. Multistage model UCL calculations using the results from these studies are comparable to, and in some cases exceed, estimates for respective tumor sites in the NTP study. In addition, the Dow inhalation study in rats showed an increase in tumors of the salivary gland region; the multistage UCL risk estimates for mammary tumors in the NTP female rats (the highest risk finding for the NTP rats) exceed the corresponding risk estimate based on the salivary tumors by a factor of three. ------- Equivalent human dose and upper-bound incremental unit risk estimates were developed using the standard assumptions of the Carcinogen Assessment Group (CAG) on the inhalation rates of rodents and humans, and use of a surface area correction (body weight, to the two-thirds power) for interspecies extrapola- tion, there being insufficient data to justify abandoning this assumption in favor of an alternative. Using the multistage UCL estimates for female mice with either adenomas or carcinomas of the lung and/or liver, the upper-limit incremental unit risk for humans from exposure over a lifetime to 1 mg/kg/day DCM is 1.4 x 10~2. Equiva- lently, the unit risk for inhaling 1 ^g/m^ DCM over a lifetime is 4.1 x 10~6; the unit risk for exposure to 1 ppm DCM is 1.4 x 10~2. Estimates of the possible incremental unit risk for humans from exposure to DCM in drinking water are made using two approaches: first, based on the findings of liver (but not lung) tumors in the NTP inhalation bioassay with mice, and second, using the borderline positive finding of liver tumors in the National Coffee Association (1983) drinking water study in mice. Since the risk estimates from these two studies are roughly comparable, the mean of the derived risk values is chosen for the unit risk estimate via ingestion. Using the mean of the UCL risk calculations from these two studies [an average unit risk of 7.5 x 10~3 (mg/kg/day)"l], ingestion of water containing 1 pg/L DCM over a lifetime has an estimated upper-bound incremental unit risk of 2.1 x ICT7. The upper-bound incremental unit risk for inhalation exposure estimated using the NTP bioassay was compared with the findings of the strongest epide- miologic study of workers exposed to DCM. Power calculations showed that the study did not have the ability to detect the estimated increases with any degree of confidence. 8 ------- 1.2. CONCLUSIONS An animal inhalation study showed a statistically positive salivary gland sarcoma response in male rats (Dow Chemical Company, 1980) and a borderline hepatocellular neoplastic nodule response was shown in female rats (National Coffee Association, 1982a, b) from drinking water exposures. There is some evidence of the carcinogenicity of DCM in male rats, as shown by an increased incidence of benign mammary gland neoplasms, and clear evidence in female rats (Dow Chemical Company, 1980; Burek et al., 1984; NTP, 1985). There is clear evidence for the carcinogenicity of DCM in male and female mice, as shown by statistically significant increased incidences of alveolar/bronchiolar neoplasms and hepatocellular neoplasms (NTP, 1985, draft). There is also evidence that DCM is mutagenic. Using EPA's Proposed Guidelines for Carcinogen Risk Assess- ment (U.S. EPA, 1984), the weight-of-evidence ranking for the carcinogenicity of DCM in experimental animals is "sufficient," and for human evidence, the ranking is "inadequate." No excessive risks were shown in the epidemiologic studies, yet the studies had sufficient deficiencies to make them inadequate for properly assessing the human risk potential. Overall, an EPA category of B2 is assigned to DCM, meaning that DCM is to be considered a "probable" human carcinogen. Using the criteria of the International Agency for Research on Cancer (IARC), the weight of evidence for the carcinogenicity of DCM in animals would also be considered as "sufficient," placing it in Group 2B. While the clear evidence for carcinogenicity is in the animal inhalation studies, the induction of distant site hepatocellular neoplasms from inhalation exposure and the borderline significance for hepatocellular neoplastic nodules in a drinking water study is an adequate basis for concluding that DCM should be considered as a "probable" human carcinogen via ingestion as well as i nhalation. ------- An estimate of the carcinogenic potency (unit risk) of DCM has been pre- pared using the incidence data from the 1985 NTP bioassay. The development of these risk estimates is for the purpose of evaluating the magnitude of the public health impact on the assumption that DCM is carcinogenic in humans. The upper-bound incremental unit risk for the inhalation of air contamina- ted with DCM is 4.1 x 10~6 (ug/m3)'1 -[B2]. The upper-bound incremental unit risk for drinking water is 0.21 x 10"6 (pg/L)"1 -[B2]. The CAG potency index for DCM is 1.2 (mmol/kg/day)"!, which places DCM in the lowest quartile of ranked chemicals that the CAG has evaluated as carcinogens. Any use of the risk estimates should include a recognition of the weight of evidence for carcinogenicity in humans and the understanding that the upper-bound nature of these estimates is such that the true risk is not likely to exceed the value, and may be lower. 10 ------- 2. INTRODUCTION In February of 1985 the Office of Health and Environmental Assessment published a Health Assessment Document for Dichloromethane (Methylene Chloride), The document, which contained an analysis of evidence for the carcinogenic potential of dichloromethane (DCM), concluded: "Using the criteria of the International Agency for Research on Cancer (IARC), the weight of evidence for carcinogenicity in animals is judged to be limited. . . . based upon the statistically positive salivary gland sarcoma response in male rats (Dow Chemical Company, 1980) and the borderline hepatocellular neoplastic nodule response in the rat and hepatocellular adenoma and/or carcinoma in male mice (National Coffee Association, 1982-1983)." It was further concluded: "When the absence of epidemiological evidence is considered along with the limited animal evidence, as well as the potential for DCM to cause gene mutations in mammalian systems, DCM is judged to be in IARC Group 3 . . . ." Because the National Toxicology Program (NTP) inhalation bioassay has been completed (NTP, 1985, draft) and the report has been reviewed and approved by the NTP Board of Scientific Counselors, it is appropriate to update the February 1985 Health Assessment Document for Dichloromethane (Methylene Chloride). The purpose of this addendum is to: o Review and integrate the data obtained in the NTP inhalation bioassay, o Analyze the pharmacokinetic/metabolic data presented in Chapter 4 of the Health Assessment Document and determine its usefulness in the quantitative estimation of carcinogenic risk, and e Revise the estimated carcinogenic potency for DCM using the data from the NTP bioassay and pharmacokinetic data if appropriate. 11 ------- 3. CARCINOGENICITY 3.1. NATIONAL TOXICOLOGY PROGRAM INHALATION BIOASSAY (1985, DRAFT) A 2-year carcinogenesis study of DCM (99% pure) was conducted at Battelle Pacific Northwest Laboratories by inhalation exposure to groups of 50 male and female F344/N rats and B6C3F1 mice (6 hours/day, 5 days/week) for 102 weeks. The exposure concentrations used were 0, 1,000, 2,000, or 4,000 ppm for rats and 0, 2,000, or 4,000 ppm for mice. These doses were selected on the basis of results obtained from a 13-week subchronic inhalation study in which animals were exposed to concentrations of 525 to 8,400 ppm, 6 hours/day, 5 days/week. The maximum exposure concentration of 4,000 ppm was selected because mimimal histopathologic changes were found after exposure to 4,000 ppm. The second dose was 2,000 ppm for both species. The third dose, 1,000 ppm, was added for rats because in an earlier inhalation study in male and female Sprague-Dawley rats (Dow Chemical Company, 1980; Burek et al., 1984), reduced survival was observed in the highest exposure group, 3,500 ppm. All animals used in this experiment were produced under strict barrier conditions at Charles River Breeding Laboratories under a contract to the car- cinogenesis program of the National Toxicology Program (NTP). The rats were placed in the study at 7 to 8 weeks of age and mice at 8 to 9 weeks. All animals were housed individually. Food and water were available ad libitum except during exposure periods, when only water was available. All animals were observed twice a day for signs of moribundity or mortality. Clinical signs were recorded every week. Body weight was recorded once a week. A complete quality-controlled environment was maintained during the experiment. The probability of survival was estimated by the product-limit procedure of Kaplan and Meier (1958). Tests of significance included pair-wise comparisons 12 ------- of high-dose and low-dose groups with controls and tests for overall dose- response trends. Life table analysis, incidental tumor analysis, and the Fisher Exact Test were used to evaluate tumor incidence. 3.1.1. Rat Study The mean body weights of experimental and control rats of each sex were similar throughout the studies (Figure 1). Rats exposed to 4000 ppm, the highest dose, were restless and pawed at the eyes and muzzle during the expo- sure period. The survival of male and female rats exposed to DCM is shown in Figure 2. The survival of female rats was significantly lower than that of the controls after week 100, and the survival in all groups of male rats at the termination of the study was low (Table 1). There were many deaths of males in the final 16 weeks of the study. The decreased survival is believed to be related to high incidence of leukemias. A significant positive trend for mammary gland fibroadenoma and adenoma or fibroma (combined) was observed in male and female rats. The incidence in high- dose males (0/50, 0/50, 2/50, and 5/50} and in females (7/50, 13/50, 14/50, and 23/50) was significantly (p < 0.001) higher than in the controls (Tables 2 and 3). Also, subcutaneous fibroma or sarcoma (combined), located in the mammary area in male rats, occurred with a significant positive trend (p = 0.008), and the incidence in the high-dose group was significantly (p < 0.05) greater than in the controls (Table 2). The subcutaneous tumors all occurred in the area of the mammary chain; therefore, the subcutaneous tumors were combined by the NTP for comparative purposes (male rats, 1/50, 1/50, 4/50, and 9/50). The incidence of subcutaneous tumors in the highest dose group was signifi- cantly (p = 0.002) higher than in the controls. The historical incidence of mammary gland tumors at the same laboratory is 0% in males and 16% in females, and the NTP historical control incidence is 3% in males and 28% in females for 13 ------- » •OKS ON STUDY YS~ ! i -*— • won ON STUDY Figure 1. Growth curves for rats exposed to dichloromethane by inhalation for 2 years. SOURCE: NTP, 1985. 14 ------- MALE RATS • •UNTKATED O-UOOO PfW FEUAU RATS •UNTREATED O-1.000 ffM *• 2.000 PfM PfM WOKS ON STUDY Figure 2. Kaplan-Meier survival curves for rats exposed to dichloromethane by inhalation for 2 years. SOURCE: NTP, 1985. 15 ------- TABLE 1. SURVIVAL OF RATS IN THE 2-YEAR INHALATION STUDY OF DICHLOROMETHANE Control 1,000 ppm 2,000 ppm 4,000 ppm Males3 Animals initially in study Nonacci dental deaths before terminationb Killed at termination Survival p values0 Females3 Animals initially in study Nonaccidental deaths before termination^ Killed at termination Survival p values0 50 34 16 0.116 50 20 30 0.006 50 34 16 0.945 50 28 22 0.223 50 33 17 0.935 50 28 22 0.118 50 41 9 0.163 50 35 15 0.006 3Tenninal kill period: week 104. ^Includes animals killed in a moribund condition. °The results of the life table trend test are in the control column, and those of the life table pairwise comparisons with the controls are in the dosed columns. SOURCE: NTP, 1985. ------- TABLE 2. ANALYSIS OF PRIMARY TUMORS IN MALE RATS IN THE 2-YEAR INHALATION STUDY OF DICHLOROMETHANE Control 1,000 ppm 2,000 ppm 4,000 ppm Subcutaneous tissue: Fibroma Overall rates* Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Subcutaneous tissue: Fibroma or sarcoma Overall rates3 Adjusted rates'5 Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd 1/50(2%) 6.3% 1/16(6%) 104 p=0.024 p=0.064 p=0.072 1/50(2%) 6.3% 1/16(6%) 104 p=0.008 p=0.026 p=0.029 1/50(2%) 6.3% 1/16(6%) 104 p=0.764 p=0.764 p=0.753 1/50(2%) 6.3% 1/16(6%) 104 p=0.764 p=0.764 p=0.753 2/50(4%) 9.2% 1/17(6%) 96 p=0.523 p=0.505 p=0.500 2/50(4%) 9.2% 1/17(6%) 96 p=0.523 p=0.505 p=0.500 4/50(8%) 19.5% 0/9(0%) 89 p=0.095 p=0.204 p=0.181 5/50(10%) 22.7% 0/9(0%) 89 p=0.050 p=0.125 p=0.102 Hematcpoietic system: Mononuclear cell leukemia Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Adrenal : Pheochromocytoma Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Adrenal: Pheochromocytoma or pheochromocytoma, Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd 34/50(68%) 80.3% 8/16(50%) 57 p=0.045 p=0.399 p=0.251 5/50(10%) 23.5% 2/16(13%) 75 p=0.035 p=0.131 p=0.192 malignant 5/50(10%) 23.5% 2/16(13%) 75 p=0.034 p=0.134 p=0.186 26/50(52%) 77.0% 9/16(56%) 82 p=0.147N p=0.049N p=0.076N 11/50(22%) 46.4% 5/16(31%) 89 p=0.094 p=0.093 p=0.086 11/50(22%) 46.4% 5/16(31%) 89 p=0.094 p=0.093 p=0.086 32/50(64%) 80.2% 10/17(59%) 71 p=0.400N p=0.434N p=0.417N 10/50(20%) 45.4% 6/17(35%) 89 p=0.149 p=0.131 p=0.131 11/50(20%) 47.0% 6/17(35%) 89 p=0.104 p=0.087 p=0.086 35/50(70%) 89.4% 6/9(67%) 75 p=0.134 p=0.487N p=0.500 10/50(20%) 52.9% 3/9(33%) 80 p=0.039 p=0.108 p=0.131 10/50(20%) 52.9% 3/9(33%) 80 p=0.039 p=0.108 p=0.131 (continued on the following page) 17 ------- TABLE 2. (continued) Mammary gland: Fi broadenoma Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Mammary gland: Adenoma or fi broadenoma Overall rates* Adjusted rates'5 Terminal rates0 Week of first observation Life table tests'1 Incidental tumor tests*1 Cochran-Armltage Trend Testd Fisher Exact Testd Mammary gland or subcutaneous tissue: Adenoma, Overall rates9 Adjusted rates'5 Terminal rates0 Week of first observation Life table testsd Incidental tumor tests'1 Cochran-Armitage Trend Testd Fisher Exact Test*1 Testis: Interstitial cell tumor Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table tests'1 Incidental tumor tests'1 Cochran-Armitage Trend Testd Fisher Exact Testd Tunica vaginalis: Malignant mesothelioma Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armltage Trend Testd Fisher Exact Testd Control 0/50(0%) 0.0% 0/16(0%) p<0.001 p<0.003 p=0.009 0/50(0%) 0.0% 0/16(0%) p<0.001 p<0.001 p=0.003 fi broadenoma, or 1/50(2%) 6.3% 1/16(6%) 104 p<0.001 p=0.003 p<0.001 39/50(78%) 94.9% 14/16(88%) 65 p=0.009 p=0.114 p=0.129 0/50(0%) 0.0% 0/16(0%) p=0.025 p=0.060 p=0.044 1,000 ppm 0/50(0%) 0.0% 0/16(0%) e e e 0/50(0%) 0.0% 0/16(0%) e e e fibroma 1/50(2%) 6.3% 1/16(6%) 104 p=0.764 p=0.764 p=0.753N 37/49(76%) 97.3% 15/16(94%) 69 p=0.420N p=0.385N p=0.478N 1/50(2%) 2.2% 0/16(0%) 69 p=0.496 p=0.473 p=0.500 2,000 ppm 2/50(4%) 11.8% 2/17(12%) 104 p=0.250 p=0.250 p=0.247 2/50(4%) 11.8% 2/17(12%) 104 p=0.250 p=0.250 p=0.247 4/50(8%) 20.6% 3/17(18%) 96 p=0.196 p=0.186 p=0.181 41/50(82%) 95.2% 15/17(88%) 75 p=0.512 p=0.387 p=0.401 0/50(0%) 0.0% 0/17(0%) e e e 4,000 ppm 4/50(8%) 34.0% 2/9(22%) 101 p=0.020 p=0.040 p=0.059 5/50(10%) 36.6% 2/9(22%) 93 p=0.010 p=0.023 p=0.028 9/50(18%) 49.0% 2/9(22%) 89 p=0.002 p=0.008 p=0.008 43/50(86%) 97.7% 8/9(89%) 75 p=0.029 p=0.253 p=0.218 3/50(6%) 19.6% 0/9(0%) 92 p=0.068 p=0.172 p=0.121 (continued on the following page) 18 ------- TABLE 2. (continued) Control 1,000 ppm 2,000 ppm 4,000 ppm Tunica vaginal is: Mesothelioma (all types) Overall ratesa 0/50(0%) 1/50(2%) 4/50(8%) 4/50(8%) Adjusted ratesb 0.0% 2.2% 19.2% 24.4% Terminal ratesc 0/16(0%) 0/16(0%) 2/17(12%) 0/9(0%) Week of first observation 69 96 92 Life table testsd p=0.009 p=0.496 p=0.070 p=0.031 Incidental tumor testsd p=0.030 p=0.473 p=0.062 p=0.097 Cochran-Armitage Trend Testd p=0.029 Fisher Exact Testd p=0.500 p=0.059 p=0.059 All sites: Malignant mesothelioma Overall ratesa 0/50(0%) 2/50(4%) 0/50(0%) 3/50(6%) Adjusted rates'5 0.0% 4.4% 0.0% 19.6% Terminal ratesc 0/16(0%) 0/16(0%) 0/17(0%) 0/9(0%) Week of first observation 69 92 Life table testsd p=0.066 p=0.243 e p=0.068 Incidental tumor testsd p=0.136 p=0.225 e p=0.172 Cochran-Armitage Trend Testd p=0.097 Fisher Exact Testd p=0.247 e p=0.121 All sites: Mesothelioma (all types) Overall ratesa 0/50(0%) 2/50(4%) 5/50(10%) 4/50(8%) Adjusted ratesb 0.0% 4.4% 22.8% 24.4% Terminal ratesc 0/16(0%) 0/16(0%) 2/17(12%) 0/9(0%) Week of first observation 69 96 92 Life table testsd p=0.020 p=0.243 p=0.038 p=0.031 Incidental tumor testsd p=0.063 p=0.225 p=0.030 p=0.097 Cochran-Armitage Trend Testd p=0.052 Fisher Exact Testd p=0.247 p=0.028 p=0.059 aNumber of tumor-bearing animals/number of animals examined at the site. ^Kaplan-Meier estimated tumor incidences at the end of the study after adjusting for intercurrent mortality. C0bserved tumor incidence at terminal kill. dBeneath the control incidence are the p values associated with the trend test. Beneath the dosed group incidence are the p values corresponding to pairwise comparisons between the dosed group and the controls. The life table analysis regards tumors in animals dying prior to terminal kill as being (directly or indirectly) the cause of death. The incidental tumor test regards these lesions as nonfatal. The Cochran-Armitage Trend Test and the Fisher Exact Test directly compare the overall incidence rates. A negative trend or lower incidence in a dose group is indicated by the letter N. eNo p value is presented because no tumors were observed in the dosed and control groups. NOTES: Mammary gland fibroadenoma: Historical incidence at testing laboratory 0/100 (0%); historical incidence in NTP studies 51/1,727 (3%) ± 3%. Mesothelioma—all sites: Historical incidence at testing laboratory 4/100 (4%); historical incidence in NTP studies 44/1,727 (3%) ± 2%. Mononuclear cell leukemia: Historical incidence at testing laboratory 36/100 (36%); historical incidence in NTP studies 458/1,727 (27%) ± 9%. SOURCE: NTP, 1985. 19 ------- TABLE 3. ANALYSIS OF PRIMARY TUMORS IN FEMALE RATS IN THE 2-YEAR INHALATION STUDY OF OICHLOROMETHANE Control 1,000 ppm 2,000 ppm 4,000 ppm Hematopoietlc system: Mononuclear cell leukemia Overall rates3 Adjusted rates'5 Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Liver: Neoplastic nodule Overall rates3 Adjusted rates'3 Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Liver: Neoplastic nodule or hepatocellular Overall rates3 Adjusted rates0 Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Mammary gland: Fibroadenoma Overall rates3 Adjusted rates0 Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Mammary gland: Adenoma or fibroadenoma Overall rates3 Adjusted rates0 Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd 17/50(34%) 41.1% 8/30(27%) 73 p=0.009 p=0.273 p=0.086 2/50(4%) 6.7% 2/30(7%) 104 p=0.030 p=0.097 p=0.078 carcinoma 2/50(4%) 6.7% 2/30(7%) 104 p-0.027 p =0.086 p=0.079 5/50(10%) 15.7% 4/30(13%) 96 p<0.001 p<0.001 p<0.001 5/50(10%) 15.7% 4/30(13%) 96 p<0.001 p<0.001 p<0.001 17/50(34%) 44.4% 4/22(18%) 76 p=0.402 p=0.425N p=0.584N 1/50(2%) 2.4% 0/22(0%) 61 p=0.569N p=0.494N p=0.500N 1/50(2%) 2.0% 0/22(0%) 61 p=0.569N p=0.494N p=0.500N 11/50(22%) 41.2% 8/22(36%) 74 p=0.028 p=0.049 p=0.086 11/50(22%) 41.2% 8/22(36%) 74 p=0.028 p=0.049 p=0.086 23/50(46%) 63.6% 10/22(45%) 73 p=0.049 p=0.189 p=0.154 3/50(6%) 10.2% 1/22(5%) 85 p=0.382 p=0.482 p=0.500 4/50(8%) 14.4% 2/22(9%) 85 p=0.223 p=0.297 p=0.339 13/50(26%) 43.6% 7/22(32%) 65 p=0.009 p=0.025 p=0.033 13/50(26%) 43.6% 7/22(32%) 65 p=0.009 p=0.025 p=0.033 23/50(46%) 58.1% 1/15(7%) 63 p=0.028 p=0.579 p=0.154 5/50(10%) 19.6% 1/15(7%) 73 p=0.080 p=0.229 p=0.218 5/50(10%) 19.6% 1/15(7%) 73 p=0.080 p=0.229 p=0.218 22/50(44%) 79.4% 10/15(67%) 73 p<0.001 p<0.001 p<0.001 23/50(46%) 83.5% 11/15(73%) 73 p<0.001 p<0.001 p<0.001 (continued on the following page) 20 ------- TABLE 3. (continued) Control 1,000 ppm 2,000 ppm 4,000 ppm Mammary gland: Adenoma, fibroadenoma, or adenocarcinoma Overall rates* 6/50(12%) Adjusted rates0 17.8% Terminal ratesc 4/30(13%) Week of first observation 92 Life table testsd p<0.001 Incidental tumor testsd p<0.001 Cochran-Armitage Trend Testd p<0.001 Fisher Exact Testd 13/50(26%) 44.4% 8/22(36%) 74 023 053 p=0 p=0 p=0.062 Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd 7/50(14%) 20.0% 4/30(13%) 92 p<0.001 p<0.001 p<0.001 13/50(26%) 44.4% 8/22(36%) 74 p=0.045 p=0.092 p=0.105 14/50(28%) 44.9% 7/22(32%) 65 p=0.012 p=0.043 p=0.039 Mammary gland: Adenoma, fibroadenoma, adenocarcinoma, or mixed tumor, malignant 614/50(28%) 44.9% 7/22(32%) 65 p=0.022 p=0.083 p=0.070 23/50(46%) 83.5% 11/15(73%) 73 p<0.001 p<0.001 p<0.001 23/50(46%) 83.5% 11/15(73%) 73 p<0.001 p<0.001 p<0.001 aNumber of tumor-bearing animals/number of animals examined at the site. ^Kaplan-Meier estimated tumor incidences at the end of the study after adjusting for intercurrent mortality. C0bserved tumor incidence at terminal kill. dBeneath the control incidence are the p values associated with the trend test. Beneath the dosed group incidence are the p values corresponding to pairwise comparisons between the dosed group and the controls. The life table analysis regards tumors in animals dying prior to terminal kill as being (directly or indirectly) the cause of death. The incidental tumor test regards these lesions as nonfatal. The Cochran-Armitage Trend Test and the Fisher Exact Test directly compare the overall incidence rates. A negative trend or lower incidence in a dose group is indicated by the letter N. eA carcinoma was also present in one of the animals that had a fibroadenoma. NOTES: Mammary gland fibroadenoma: Historical incidence at testing laboratory 16/99 (16%); historical incidence in NTP studies 492/1,772 (28%) ± 10%. Mononuclear cell leukemia: Historical incidence at testing laboratory 27/99 (27%); historical incidence in NTP studies 307/1,772 (17%) ± 6. SOURCE: NTP, 1985. 21 ------- the same strain of rats. The increased incidence of mammary gland tumor is consistent with the results reported by Dow Chemical Company (1980) and Burek et al. (1984) in Sprague-Dawley rats (U.S. EPA, 1985). These studies have been reviewed previously. Sprague-Dawley rats have a spontaneous incidence of mammary gland tumors, about 80% in females and 10% in males. In males (Burek et al., 1984) the mammary tumors increased in the highest dose group to 14/97, as compared to 7/92 in the controls. In females the number of tumors per rat increased with dose. The increased incidence of benign mammary gland tumors in males and females provides some supportive evidence for mammary gland carcinogenesis. Maltoni (1984) also reported at the Food Solvent Workshop (1984) on a study in which DCM was administered by gavage at 500 mg/kg/day, 5 days/week for 64 weeks, followed by an observation period until spontaneous death. Maltoni (1984) observed an increased incidence of mammary gland tumors in Sprague-Dawley rats. Thus, the incidence of mammary gland benign tumors in female rats in the present NTP study is consistent with the reports of Burek et al. (1984), Maltoni (1984), and Nitschke et al. (1982). The incidence of liver neoplastic nodules and hepatocellular carcinomas (combined) in female rats (2/50, 1/50, 4/50, and 5/50) occurred with positive trends (p = 0.030) by life table analysis only (Table 3). The incidence in the high-dose group was not significantly greater than in the controls; this result is consistent with observations made by other investigators (Maltoni, 1984; NTP, 1982, unpublished; National Coffee Association, 1982a, b). Maltoni (1984) reported that gavage administration of 100 or 500 mg/kg/day for 64 weeks induced a dose-related increase in the incidence of nodular hyperplasia of the liver in Sprague-Dawley rats. The earlier gavage study (NTP, 1982, unpublished) indicated that administration of 500 or 1,000 mg/kg/day increased the incidence of hepatocellular nodules in both male and female F344/N rats. 22 ------- Furthermore, there was a significant increase in liver tumors in female rats dosed at 250 mg/kg/day in the drinking water study (National Coffee Associa- tion, 1982a, b), although the number of tumors were within the range of the his- torical control values of the laboratory. The pharmacokinetic data presented by Dr. Kirshman at the Food Solvent Workshop (1984, page 41) indicated that 250 mg/kg/day in the drinking water study was equivalent to a 750 ppm inhalation level, which is 1/5 of the maximum tolerated dose (MTD) used in the Burek et al. (1984) study and in the NTP (1985) study. The NTP reported that the highest exposure concentration (4,000 ppm) in the inhalation study has been estimated to be equivalent to 1,300 mg/kg/day from an oral dose. In male rats the incidence of mesothelioma arising from all sites (0/50, 2/50, 5/50, and 4/50) occurred with a significant positive trend (p = 0.02); the incidence in the mid- and high-dose groups was significantly higher (p = 0.038, p = 0.30) than in the controls (Table 2). This increased incidence may not be due to administration of DCM because the concurrent controls in the same laboratory were low in comparison with earlier inhalation studies (4/100). Mononuclear cell leukemia in male (Table 2) and female (Table 3) rats occurred with a significant positive trend by life table analysis only. The incidence (17/50, 17/50, 23/50, 23/50) in females was significantly greater than in the controls at the mid-dose (p = 0.049) and high-dose (p = 0.028) levels. Some other tumor incidences were increased marginally in experimental groups as compared to the controls. These increases were characterized by a significant trend only. These tumors included adrenal gland pheochromocytoma and interstitial cell tumors in males (Table 2) and pituitary gland adenoma or carcinoma (combined) in males and females (Tables 2 and 3). The squamous cell metaplasia of the nasal cavity in female rats (1/50, 2/50, 3/50, and 9/50) was increased significantly in the high-dose group, but no nasal tumors were 23 ------- found in this group. 3.1.2. Mouse Study The mean initial body weight of males in the 4,000 ppm group was 15% lower than that of controls (Figure 3). The mean body weights were comparable in the high-dose and control groups until week 90, but after week 90 the body weights were 8% to 11% lower than those of controls. During the exposure period, the mice were hyperactive. The probabilities of survival of male and female mice are shown in Figure 4. The survival in both male and female high- dose groups decreased significantly compared with controls (Table 4). The reduced survival may have been due to chemically-induced lung and liver tumors in both male and female mice. The incidences of lung tumors increased significantly (p = 0.0001) in both males (Table 5) and females (Table 6). The latent period for tumor induction was significantly (trend analysis) decreased in the high-dose groups as compared to controls. The tumors observed were alveolar/bronchiolar adenomas (males, 3/50, 19/50, and 24/50; females, 2/50, 23/48, and 28/48) and al veolar/bronchio- lar carcinomas (males, 2/50, 10/50, and 28/50; females, 1/50, 13/48, 29/48, and 29/48). In addition to the dose-related increase in lung tumors in male and female mice, there were dose-related increases in multiple lung tumor- bearing mice (Table 7). The multiplicity of tumors included both alveolar/ bronchiolar adenoma and carcinoma. No lung tumors were found in the controls, whereas 70% of the high-dose males and 71% of high-dose females had multiple tumors (males, 0/50, 10/50, and 28/40; females, 0/50, 11/48, and 29/41). In the experimental groups, 38% of the dosed male mice and 42% of the dosed female mice had multiple lung tumors. These results are consistent with the data obtained in other studies. In the earlier NTP (1982, unpublished) gavage study, DCM produced a significant increase in lung tumors in female mice. 24 ------- «- B *' o 8 o i MlRft ON SIUDT 88 * ,, e o o i a fi M1U ON STUDY Figure 3. Growth curves for mice exposed to dichloromethane by inhalation for 2 years. SOURCE: NTP, 1985. 25 ------- * e 5 •* L MALI kflCE • UNTREATED O«2.000 PPM -4.000 PPM WOKS ON STUDY FEMALE MICE • •UNTREATED O«2.000 PPM 4,000 PPU WOXS ON STUDY Figure 4. Kaplan-Meier survival curves for mice exposed to dichloromethane by inhalation for 2 years. SOURCE: NTP, 1985. 26 ------- TABLE 4. SURVIVAL OF MICE IN THE 2-YEAR INHALATION STUDY OF DICHLOROMETHANE Control 2,000 ppm 4,000 ppm Males9 Animals initially in study 50 50 50 Nonaccidental deaths before termination'5 11 24 38 Accidentally killed 021 Killed at termination 39 24 9 Died during termination period 002 Survival p valuesc <0.001 <0.010 <0.001 Females9 Animals initially in study 50 50 50 Nonaccidental deaths before termination15 24 22 40 Accidentally killed 1 2 1 Killed at termination Oil Died during termination period 25 25 8 Survival p values0 0.002 0.678 0.004 aTerminal kill period: week 104. ^Includes animals killed in a moribund condition. cThe results of the life table trend test are in the control column, and those of the life table pairwise comparisons with the controls are in the dosed columns. SOURCE: NTP, 1985. 27 ------- TABLE 5. ANALYSIS OF PRIMARY TUMORS IN MALE MICE IN THE 2-YEAR INHALATION STUDY OF OICHLOROMETHANE Lung: Alveolar/bronchiolar adenoma Overall rates3 Adjusted ratesb Terminal rates0 Week of first observation Life table tests'* Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Lung: Alveolar/bronchiolar carcinoma Overall rates3 Adjusted ratesb Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Lung: Alveolar/bronchiolar adenoma or Overall rates3 Adjusted ratesb Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Circulatory system: Hemangiosarcoma Overall rates3 Adjusted ratesb Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Control 3/50(6%) 7.7% 3/39(8%) 104 p<0.001 p<0.001 p<0.001 2/50(4%) 4.9% 1/39(3%) 94 p<0.001 p<0.001 p<0.001 carcinoma 5/50(10%) 12.4% 4/39(10%) 94 p<0.001 p<0.001 p<0.001 1/50(2%) 2.6% 1/39(3%) 104 p=0.007 p=0.083 p=0.060 2,000 ppm 19/50(38%) 55.6% 10/24(42%) 71 p<0.001 p<0.001 p<0.001 10/50(20%) 34.0% 6/24(25%) 78 p<0.002 p<0.016 p<0.014 27/50(54%) 74.2% 15/24(63%) 71 p<0.001 p<0.001 p<0.001 2/50(4%) 7.6% 1/24(4%) 101 p=0.352 p=0.495 p=0.500 4,000 ppm 24/50(48%) 78.5% 6/11(55%) 70 p<0.001 p<0.001 p<0.001 28/50(56%) 92.9% 9/11(82%) 72 p<0.001 p<0.001 p<0.001 40/50(80%) 100.0% 11/11(100%) 70 p<0.001 p<0.001 p<0.001 5/50(10%) 21.4% 1/11(9%) 70 p=0.017 p=0.142 p=0.102 Circulatory system: Hemangioma or hemangiosarcoma Overall rates3 Adjusted ratesb Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd 2/50(4%) 4.8% 1/39(3%) 87 p=0.010 p=0.170 p»0.080 2/50(4%) 7.6% 1/24(4%) 101 p=0.558 p=0.643N p=0.691 6/50(12%) 25.8% 1/11(9%) 70 p=0.022 p=0.301 p=0.134 (continued on the following page) 28 ------- TABLE 5. (continued) Control 2,000 ppm 4,000 ppm Liver: Hepatocellular adenoma Overall rates* 10/50(20%) Adjusted ratesb 23.0% Terminal ratesc 7/39(18%) Week of first observation 73 Life table testsd p<0.001 Incidental tumor testsd p<0.075 Cochran-Armitage Trend Testd p=0.194 Fisher Exact Test*1 Liver: Hepatocellular carcinoma Overall rates* 13/50(26%) Adjusted rates0 29.7% Terminal ratesc 9/39(23%) Week of first observation 73 Life table testsd p<0.001 Incidental tumor testsd p=0.016 Cochran-Armitage Trend Testd p=0.004 Fisher Exact Testd Liver: Hepatocellular adenoma or carcinoma Overall rates3 22/50(44%) Adjusted ratesb 48.3% Terminal ratesc 16/39(41%) Week of first observation 73 Life table testsd p<0.001 Incidental tumor testsd p=0.010 Cochran-Armitage Trend Testd p=0.013 Fisher Exact Testd 14/49(29%) 46.9% 9/24(38%) 71 p=0.041 p=0.161 p=0.224 15/49(31%) 43.7% 7/24(29%) 72 p=0.111 p=0.422 p=0.387 22/49(49%) 66.8% 13/24(54%) 71 p=0.048 p=0.305 p=0.384 14/49(29%) 68.3% 6/11(55%) 80 p=0.001 p=0.095 p=0.224 26/49(53%) 76.4% 5/11(45%) 61 p<0.001 p=0.042 p=0.005 33/49(67%) 93.0% 9/11(82%) 61 p<0.001 p=0.020 p=0.016 aNumber of tumor-bearing animals/number of animals examined at the site. bKaplan-Meier estimated tumor incidences at the end of the study after adjusting for intercurrent mortality. C0bserved tumor incidence at terminal kill. dBeneath the control incidence are the p values associated with the trend test. Beneath the dosed group incidence are the p values corresponding to pairwise comparisons between the dosed group and the controls. The life table analysis regards tumors in animals dying prior to terminal kill as being (directly or indirectly) the cause of death. The incidental tumor test regards these lesions as nonfatal. The Cochran-Armitage Trend Test and the Fisher Exact Test directly compare the overall incidence rates. A negative trend or lower incidence in a dose group is indicated by the letter N. NOTES: Alveolar/bronchiolar adenoma or carcinoma: Historical incidence at testing laboratory 31/100 (31%); historical incidence in NTP studies 296/1,780 (17%) ± 8%. Hepatocellular adenoma or carcinoma: Historical incidence at testing laboratory 28/100 (28%); historical inci- dence in NTP studies 540/1,784 (30%) ± 8%. Hemangioma or hemangiosarcoma: Historical incidence at testing laboratory 2/100 (2%); historical incidence in NTP studies 78/1,791 (4%) ± 4%. SOURCE: NTP, 1985. 29 ------- TABLE 6. ANALYSIS OF PRIMARY TUMORS IN FEMALE MICE IN THE 2-YEAR INHALATION STUDY OF DICHLOROMETHANE Lung: AT veolar/bronchiolar adenoma Overall rates3 Adjusted ratesb Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Lung: Alveolar/bronchiolar carcinoma Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Lung: Alveolar/bronchiolar adenoma or Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Liver: Hepatocellular adenoma Overall rates3 Adjusted ratesb Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Liver: Hepatocellular carcinoma Overall rates3 Adjusted rates'5 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd Control 2/50(4%) 6.7% 1/25(4%) 87 p<0.001 p<0.001 p<0.001 1/50(4%) 4.0% 1/25(4%) 104 p<0.001 p<0.001 p<0.001 carcinoma 3/50(6%) 10.6% 2/25(8%) 87 p<0.001 p<0.001 p<0.001 2/50(4%) 6.5% 1/25(4%) 84 p<0.001 p<0.001 p<0.001 1/50(2%) 4.0% 1/25(4%) 104 p<0.001 p<0.001 p<0.001 2,000 ppm 23/48(48%) 66.5% 14/25(56%) 83 p<0.001 p<0.001 p<0.001 13/48(27%) 45.9% 10/25(40%) 89 p<0.001 p<0.001 p<0.001 30/48(63%) 82.9% 19/25(76%) 83 p<0.001 p<0.001 p<0.001 6/48(13%) 21.3% 4/25(16%) 96 p=0.151 p=0.155 p=0.121 11/48(23%) 34.0% 6/25(24%) 83 p=0.005 p=0.004 p=0.001 4,000 ppm 28/48(58%) 91.1% 6/8(75%) 68 p<0.001 p<0.001 p-cO.OOl 29/48(60%) 92.2% 6/8(75%) 68 p<0.001 p<0.001 p<0.001 41/48(85%) 100.0% 8/8(100%) 68 p<0.001 p<0.001 p<0.001 22/48(46%) 83.0% 5/8(63%) 68 p<0.001 p<0.001 p<0.001 32/48(67%) 96.5% 7/8(88%) 68 p<0.001 p<0.001 p<0.001 30 ------- TABLE 6. (continued) Control Thyroid gland: Follicular cell adenoma Overall rates3 Adjusted rates'3 Terminal ratesc Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd 1/48(2%) 4.2% 1/24(4%) 104 p=0.012 p=0.040 p=0.093 2,000 ppm 1/47(2%) 4.0% 1/25(4%) 104 p=0.754N p=0.754N p=0.747 4,000 ppm Liver: Hepatocellular adenoma or carcinoma Overall rates3 Adjusted rates0 Terminal rates0 Week of first observation Life table testsd Incidental tumor testsd Cochran-Armitage Trend Testd Fisher Exact Testd 3/50(6%) 10.4% 2/25(8%) 84 p<0.001 p<0.001 p<0.001 16/48(33%) 48.0% 9/25(36%) 83 p=0.002 p=0.002 p<0.001 40/48(83%) 100.0% 8/8(100%) 68 p<0.001 p<0.001 p<0.001 4/46(9%) 35.0% 2/8(25%) 77 p=0.022 p=0.069 p=0.168 3Number of tumor-bearing animals/number of animals examined at the site. bKaplan-Meier estimated tumor incidences at the end of the study after adjusting for Intercurrent mortality. C0bserved tumor incidence at terminal kill. dBeneath the control incidence are the p values associated with the trend test. Beneath the dosed group incidence are the p values corresponding to pairwise comparisons between the dosed group and the controls. The life table analysis regards tumors in animals dying prior to terminal kill as being (directly or indirectly) the cause of death. The incidental tumor test regards these lesions as nonfatal. The Cochran-Armitage Trend Test and the Fisher Exact Test directly compare the overall incidence rates. A negative trend or lower incidence in a dose group is indicated by the letter N. NOTES: Alveolar/bronchiolar adenoma or carcinoma: Historical incidence at testing laboratory 10/100 (10%); historical incidence in NTP studies 122/1,777 (7%) ± 4%. Hepatocellular adenoma or carcinoma: Historical incidence at testing laboratory 5/100 (5%); historical incidence in NTP studies 147/1,781 (8%) ± 5%. SOURCE: NTP, 1985. 31 ------- TABLE 7. MULTIPLICITY OF PULMONARY TUMORS IN MICE EXPOSED TO DICHLOROMETHANE Exposure groups (ppm) Diagnoses 2,000 4,000 Males One adenoma and one carcinoma Multiple adenomas Multiple carcinomas Multiple adenomas and multiple carcinomas One adenoma and multiple carcinomas Multiple adenomas and one carcinoma Incidence of mice with multiple tumors No. of mice with multiple tumors/no, of mice with pulmonary tumors Females One adenoma and one carcinoma Multiple adenomas Multiple carcinomas Multiple adenomas and multiple carcinomas One adenoma and multiple carcinomas Multiple adenomas and one carcinoma Incidence of mice with multiple tumors No. of mice with multiple tumors/no, of mice with pulmonary tumors 0/50 0/50 0/50 0/50 0/50 0/50 0/50(0%) 0/5(0%) 0/50 0/50 0/50 0/50 0/50 0/50 0/50(0%) 0/3(0%) 1/50 5/50 3/50 0/50 1/50 0/50 10/50(20%) 10/27(37%) 2/48 4/48 1/48 0/48 2/48 2/48 11/48(23%) 11/30(37%) 3/50 4/50 12/50 3/50 3/50 3/50 28/50(56%) 28/40(70%) 4/48 5/48 8/48 2/48 7/48 3/48 29/48(60%) 29/41(71%) SOURCE: NTP, 1985. 32 ------- Maltoni (1984) also reported an increased incidence of lung tumors (one and one-half times) in male mice dosed with 100 to 500 mg/kg/day DCM by gavage. In male mice (Table 5) the hepatocellular adenoma or carcinoma (combined) (22/50, 24/49, and 33/49) and hepatocellular carcinoma (13/50, 15/49, and 26/49) were increased significantly (p < 0.001), especially at 4,000 ppm. In female mice (Table 6), DCM produced dose-related increases in both hepatocellular adenoma (2/50, 6/48, and 22/48) and hepatocellular carcinoma (1/50, 11/48, and 32/48), which were highly significant by any statistical test. The incidence of these tumors in the controls was consistent with the historical control values of this laboratory. The multiplicity of the hepatocellular neoplasms was common in the male and female dosed mice (Table 8). It should be noted that only 4% of the male control mice and none of the female control mice had multiplicity of liver tumors. The multiplicity of hepatocellular tumors in both male and female mice increased significantly in a dose-related manner (males, 2/50, 11/49, and 16/46; females, 0/50, 3/48, and 28/48). There were 27/57 (47%) males and 31/56 (55%) females with multiple tumors. The increased incidence of hepatocellular tumors is consistent with the results of the previous NTP (1982, unpublished) gavage study, the Maltoni (1984) study, and the National Coffee Association (1983) drinking water study. The National Coffee Association study produced a borderline significant increase in liver tumors at a dose significantly less than the maximum tolerated dose. There was also an increase in hemangiosarcoma (1/50, 2/50, and 5/50) or hemangioma and hemangiosarcoma (2/50, 2/50, and 6/50) in male mice (Table 7), which occurred with a positive trend by life table analysis. 3.1.3. Summary The results of the NTP inhalation bioassay (1985, draft) using F344/N rats showed an increased incidence of benign mammary gland neoplasms, primarily 33 ------- TABLE 8. MULTIPLICITY OF LIVER TUMORS IN MICE EXPOSED TO DICHLOROMETHANE Diagnoses Exposure groups (ppm) 2,000 4,000 Males One adenoma and one carcinoma Multiple adenomas Multiple carcinomas Multiple adenomas and multiple carcinomas One adenoma and multiple carcinomas Multiple adenomas and one carcinoma Incidence of mice with multiple tumors No. of mice with multiple tumors/no, of mice with pulmonary tumors Females One adenoma and one carcinoma Multiple adenomas Multiple carcinomas Multiple adenomas and multiple carcinomas One adenoma and multiple carcinomas Multiple adenomas and one carcinoma Incidence of mice with multiple tumors No. of mice with multiple tumors/no, of mice with pulmonary tumors 1/50 0/50 1/50 0/50 0/50 0/50 2/50(4%) 2/22(9%) 0/50 0/50 0/50 0/50 0/50 0/50 0/5(0%) 0/3(0%) 2/49 3/49 3/49 0/49 0/49 3/49 11/49(22%) 11/24(46%) 1/48 0/48 2/48 0/48 0/48 0/48 3/48(6%) 3/16(19%) 3/49 3/49 6/49 1/49 2/49 1/49 16/49(33%) 16/33(48%) 6/48 4/48 10/48 3/48 1/48 4/48 28/48(58%) 28/40(70%) SOURCE: NTP, 1985. 34 ------- fibroadenomas, in both male and female rats. In female rats, there was also a significant increase in hepatocellular neoplastic nodules and hepatocellular carcinomas (combined) by the trend test only and a statistically significant increase of mononuclear cell leukemias by age adjustment. In male rats, there was a significant increase in mesotheliomas, primarily from the tunica vagina- lis. Lastly, there was a marginally significant increase in adrenal pheochro- mocytomas and interstitial cell tumors in males and pituitary gland adenomas and carcinomas (combined) in male and female rats by the trend test only. In the NTP inhalation bioassay using B6C3F1 mice, a highly significant increase in alveolar/bronchiolar adenoma and/or carcinoma was observed in both sexes. The incidence of hepatocellular adenoma and hepatocellular carcinoma (combined) was increased in the high-dose males and in both dosed groups of females. There was also a dose-related increase in the number of mice bearing multiple lung and liver tumors. Only one lung tumor per mouse was found in the controls whereas 38% of all dosed male mice and 42% of all dosed female mice had multiple lung tumors. The incidence of multiple hepatocellular tumors in the exposed groups increased in both sexes in a dose-related manner. Multiple hepatocellular tumors were found in only 4% of the male controls, but none were found in the female controls; in contrast, 28% of the exposed males and 32% of the exposed females exhibited multiple liver tumors. The NTP concluded that, under the conditions of this bioassay, there was some evidence of DCM carcinogenicity for male F344/N rats as shown by an increased incidence of benign neoplasms of the mammary gland, sufficient or clear evidence of DCM carcinogenicity for female F344/N rats as shown by an increased incidence of benign neoplasms of the mammary gland, and sufficient or clear evidence of DCM carcinogenicity in male and female B6C3F1 mice as shown by increased incidences of lung and liver tumors. 35 ------- 3.2. PHARMACOKINETICS/METABOLISM The purpose of this analysis is to evaluate the available pharmacokinetic/ metabolism data that might be useful in the assessment of the carcinogenic risk arising from exposure to methylene chloride (DCM). Most of the data used in this analysis have previously been reviewed (U.S. EPA, 1985) and therefore will not be discussed in detail. Data on routes of exposure, rates of inges- tion, metabolism, and administered versus effective dose in the NTP bioassay have been reviewed in order to determine if there are qualitative or quantita- tive differences or similarities between species that may alter the assumptions used in estimating risks. Relevant new data have been incorporated as appropriate. 3.2.1. In Vitro Metabolism/Pathways The preponderance of data obtained from both in vivo and in vitro experiments indicate that DCM and other dihalomethanes are biotransformed to both carbon monoxide and carbon dioxide. Carbon monoxide is the end product of microsomal oxidation, and carbon dioxide is an end product of cytosolic metabolism. A number of investigators (Kubic and Anders, 1975, 1978; Hogan et a!., 1976; Stevens and Anders, 1978, 1979; Ahmed and Anders, 1978) have studied the metabolism of DCM and other dihalomethanes in experiments using rat liver microsomes. These studies have resulted in the following observations: e NADPH and molecular oxygen are required for maximal activity, e Anaerobic conditions reduce the rate of DCM conversion to carbon monoxide by 80 percent, e There is a high correlation between the in vitro production of carbon monoxide and microsomal P^Q content, 36 ------- o Pretreatment of test animals with P^Q inducers resulted in increased conversion of DCM to carbon monoxide by rat liver mi crosomes, o Pretreatment of test animals with P45Q inhibitors resulted in decreased conversion of DCM to carbon monoxide by rat liver microsomes, and • Cleavage of the carbon-hydrogen bond is the rate-limiting step in dihalomethane metabolism. Based on these observations, Anders et al. (1977) postulated that DCM is biotransformed to carbon monoxide by the microsomal mixed-function oxidases via formation of a formyl halide intermediate. The proposed mechanism for the metabolism of dihalomethanes by liver microsomes is summarized in Figure 5. A number of investigators (Heppel and Porterfield, 1948; Kubic and Anders, 1975; Ahmed and Anders, 1976, 1978) have studied the metabolism of DCM and other dihalomethanes to carbon dioxide, formaldehyde, and formic acid using a rat liver cytosolic fraction. They have made the following observa- tions: o The reaction does not require molecular oxygen, • The reaction is glutathione-dependent, o Chemicals known to complex with glutathione inhibit the reaction, and o Removal of formaldehyde dehydrogenase by ammonium sulfate fractionation results in the formation of formaldehyde only, and not formic acid. Based on these observations, Ahmed et al. (1980) postulated that DCM is bio- transformed to carbon dioxide by the liver cytosolic fraction via formation 37 ------- MicrosomalPathway MFO X2CH2 H NADPH Covalent Binding •*• lipid protein OH .-H Nonenzymatic -Xw rearrangement .H Spontaneous X-C 'O Decomposition -H, -X formyl halide Cytosolic Pathway XX X - CH2 + GSH GS - CH2X transferese NAD GS - C S-halomethylglutathione HOH Nonenzymatic hydrolysis GS - CH2OH S-hydroxymethyl glutathione \ formaldehyde H2C«O + GSH dehydrogenase formaldehyde S-formyl glutathione HOH | S-formyl glutathione hydrolase O »»•: metabolic incorporation Figure 5. Proposed reaction mechanisms for the metabolism of dihalomethanes to carbon monoxide, carbon dioxide, formaldehyde, formic acid, and inorganic halide. SOURCE: Adapted from Ahmed et al., 1980. 38 ------- of a reactive formyl intermediate. The proposed mechanism for the metabolism of dihalomethanes by liver cytosol is summarized in Figure 5. A number of studies have compared DCM metabolism by means of microsomes and/or cytosol prepared from various tissues. Kubic and Anders (1975) compared the biotransformation of DCM to carbon monoxide by liver, kidney, and lung microsomes. They found liver microsomes to be five times more active than lung microsomes and 30 times more active than kidney microsomes. Ahmed and Anders (1976) compared the biotransformation of DCM to carbon dioxide by cytosolic fractions prepared from various tissues of the rat. The highest activities were found in liver, lung, and kidney cytosol. Liver cytosol was found to be 15 times more active than lung cytosol and 12.5 times more active than kidney cytosol. Ahmed and Anders (1978) measured the metabolism of dibromomethane to bromide, formaldehyde, and formic acid using rat liver cytosol, and observed that free bromide was formed at the rate of 27.7 ± 3.8 nmol/mg protein/minute, and that the amount of formaldehyde plus formic acid formed was 12.6 ± 0.8 nmol/mg protein/minute. Kubic and Anders (1975), using rat liver microsomes or cytosol from the same liver preparation, measured dibromomethane metabolism and found that the microsomes yielded 3.9 nmol carbon monoxide/nig protein/ minute and 14.0 nmol free bromide/mg protein/minute. The cytosol fraction from the same liver preparation converted dibromomethane to bromide at a rate of 9.1 nmol/mg protein/minute. Based on these data for dibromomethane, it appears likely that in in vitro experiments similar amounts of DCM/mg protein/unit time might be converted to carbon monoxide or carbon dioxide by microsomes and by cytosol. Since microsomes comprise 2% to 5% of liver protein and cytosol comprises about 10% of liver protein (Estabrook et al., 1971; Hogeboom et al., 1953), it would be anticipated, on a mass basis, that 39 ------- qualitatively the cytosol would metabolize more DCM than would microsomes. It has been postulated that each pathway involves the formation of a reactive intermediate, and each intermediate can be a potential alkylating agent. The available data also indicate that most metabolism of DCM occurs in the liver, although small amounts of activity have been detected in the lung and kidney. 3.2.2. In Vivo Metabolism/Effect of Dose There have been a number of in vivo studies in which investigators have given animals 14C-DCM and then measured the amount of exhaled 14C-carbon monox- ide and 14C-carbon dioxide. Yesair et al. (1977) and Angelo (1985) assessed the metabolism of DCM by mice. Yesair et al. (1977) gave groups of mice 1 mg/ kg or 100 mg/kg 14C-DCM in corn oil by intraperitoneal injection, and then collected metabolic by-products for 96 hours. (The authors stated, without giving data, that most of the observed metabolism took place during the first 12 hours postexposure.) Angelo (1985) gave groups of mice 10 mg/kg or 50 mg/ kg l^C-DCM in 25% polyethylene glycol by tail vein injection, and then collec- ted metabolic by-products for 4 hours. The data collected by these investiga- tors are summarized in Table 9. Direct comparison of the two studies is not possible since the route of administration of the DCM and the postexposure collection period of metabolic products was different. However, both studies show a lack of a linear dose-response relationship between the administered dose and carbon monoxide plus carbon dioxide formed. Thus, these data suggest that the highest doses approach metabolic saturation, but without more dose levels the degree of saturation cannot be estimated. The data from both studies show that the mouse is able to metabolize DCM to carbon monoxide and carbon dioxide, and that over a large dose range, equal amounts of DCM are converted to carbon monoxide and carbon dioxide. 40 ------- TABLE 9. IN VIVO METABOLISM OF DICHLOROMETHANE (DCM) BY MICE Dose intraperitoneal administration'3 1 mg/kg (11.76 ymol/kg) 100 mg/kg (1,176 ymol/kg) intravenous administration0 10 mg/kg (117.6 ymol/kg) 50 mg/kg (588 Umo1/kg) DCM exhaled3 — 470.0 56.9 382.2 Percent of dose exhaled C02a 5.9 40.0 294.0 48.4 23.8 65.0 80.6 C0a 5.3 235.0 17.5 49.4 aValues are pmol/kg. bYesair et al. (1977). cAngelo (1985). One animal study has been reported on the metabolism of inhaled 14C-DCM. McKenna et al. (1982) gave groups of rats a single 6-hour inhalation exposure to 50, 500, or 1,500 ppm 14C-DCM. At the end of the exposure period, exhaled DCM, carbon dioxide, carbon monoxide, and urine were collected for 48 hours. At the end of the 48-hour collection period, the rats were sacrificed and tissue levels of radioactivity were determined. The data obtained from this study are summarized in Tables 10 and 11 and in Figures 6 and 7. The authors interpreted these data to indicate saturability of metabolism because of the following observations: e The percent of administered DCM metabolized to carbon dioxide and carbon monoxide declined with increasing dose, and e There was less than a proportional increase in tissue levels of radioactivity. 41 ------- TABLE 10. BODY BURDENS AND METABOLISM OF 14C-DICHLOROMETHANE (DCM) IN RATS AFTER INHALATION EXPOSURE TO 14C-DCM Exposure concentration 50 ppm 500 ppm 1,500 ppm Total body burden3 mgEq 14C-DCM/kg 5.53 ± 48.41 ± 109.14 ± 0.18 4.33 3.15 Metabolized 14C-DCMa mgEq l4C-DCM/kg 5.23 ± 33.49 ± 49.08 ± 0.32 0.33 1.37 Metabolized 14C-DCM, % 94.6 69.2 45.0 aValues are mean ± standard deviation; number of animals in each group = 3. SOURCE: McKenna et al., 1982. TABLE 11. FATE OF 14C-DICHLOROMETHANE (DCM) IN RATS AFTER A SINGLE 6-HOUR INHALATION EXPOSURE Percent body burden (x ± SD, n = 3) Parameter measured Expired Expired Expired Urine Feces Carcass Skin CH2C12 C02 CO Cage wash 5. 26. 26. 8. 1. 23. 6. 0. 50 ppm 42 ± 20 ± 67 ± 90 ± 94 ± 26 ± 85 ± 75 ± 0. 1. 3. 0. 0. 1. 1. 0. 73 21 00 39 19 62 62 33 500 ppm 30. 22. 18. 8. 1. 11. 6. 0. 40 ± 53 ± 09 ± 41 ± 85 ± 65 ± 72 ± 24 ± 7.10 4.57 0.81 0.90 0.68 1.87 0.13 0.23 1 55 13 10 7 2 7 3 0 ,500 ppm .00 ± .61 ± .23 ± .20 ± .33 ± .24 ± .97 ± .43 ± 1.92 1.20 1.68 0.74 0.05 0.65 0.15 0.15 SOURCE: McKenna et al., 1982. 42 ------- 10.0 u 1 0. A SO ppm • 500 ppm • 1500 ppm I I I I I I 0.01 — 0.001 Figure 6. Plasma levels of dichloromethane in rats during and after dichloromethane exposure for 6 hours. Data points represent mean ± standard deviation for two to four rats. SOURCE: McKenna et al., 1982. 43 ------- 100.0 I ' I ' I ' I ' I ' I i I ' I i I 'E 10.0 0.1 I EXPOSURE' 50 ppm 500 ppm 1500 ppm I I I I I I I I 456/0 TIME, hours Figure 7. Blood COHb concentrations in rats during and after a 6-hour inhalation exposure to dichloromethane. Each data point is the mean ± standard error for two to four rats. SOURCE: McKenna et al., 1982. 44 ------- McKenna et al. (1982) concluded that the disproportionate exhalation of DCM, carbon monoxide, and carbon dioxide indicated saturated metabolism. However, analysis of the data suggest that while the higher doses are approaching metabolic saturation, they are not necessarily saturating. For example, the amount of radioactivity recovered in the urine and feces is proportional to dose (Table 11), which suggests no metabolic saturation. Also, the body burden data (Table 10) accounts for only 40% to 65% of the DCM to which the animals were exposed (assuming 250 g rat, a minute volume of 273 mL/minute and an absorption of 50%). Thus, it is not possible to determine if the amounts of metabolites formed and retained are directly proportional to the administered dose. In these experiments (McKenna et al., 1982) the amount of metabolites measured are formed both during the 6-hour exposure period and in the 48-hour postexposure period. The data suggest that the amount of DCM retained in the tissues postexposure might be significant since the amount of carbon monoxide exhaled exceeds that which could be predicted by the blood carboxyhemoglobin (COHb) level during the exposure period. Consistent with this observation, the data from the experiments by DiVincenzo and Hamilton (1975) also suggest that following intraperitoneal injection of DCM in rats, the residual level of unmetabolized DCM in tissues may make a significant contribution to the amount of carbon monoxide and carbon dioxide exhaled 24 hours postexposure (Table 12). The data reported by McKenna et al. (1982) showed that during the exposure period the COHb levels in rats exposed to 500 and 1,500 ppm DCM were the same (Figure 7). These data indicate that the carbon monoxide pathway was saturated over the range of concentration of DCM studied. Supporting this conclusion are the data from Fodor et al. (1973) and Kurppa et al. (1981) which showed that the COHb level of rate exposed to carbon monoxide plus DCM was additive, 45 ------- TABLE 12. FATE AND DISPOSITION OF 14C-DICHLOROMETHANE (DCM) IN RATS (511 mg/kg) INJECTED INTRAPERITONEALLY Percentage of dose (averages) 2 hours 24 hours Breath Unchanged 14C-DCM 14CO Unidentified 14C Total 84.5 0.14 0.55 0.40 85.59 91.5 2.15 3.04 1.49 98.18 Source: DiVincenzo and Hamilton, 1975. indicating that COHb formation is not the rate-limiting step in these experi- ments. However, since similar data on the carbon dioxide pathway were not obtained, it is difficult to conclude definitively the extent of metabolic saturation over the range of concentrations of DCM studied by McKenna et al. (1982). Gargas et al. (1985) have recently provided data on the relative contri- bution of the microsomal and cytosolic pathways to the metabolism of various dihalomethanes in rats following inhalation exposure. Some animals were given pyrazole in order to inhibit microsomal oxidation, and others were given 2,3- epoxypropan-1-ol to reduce hepatic glutathione and thus inhibit the cytosolic pathway. Dihalomethanes containing at least one bromide group were used to assess the carbon dioxide pathway by determining the level of free bromide in the blood when the microsomal pathway is inhibited. Immediately following 4-hour inhalation exposures, the rats were sacrificed and blood COHb and plasma bromide determined. The data suggest that the carbon monoxide pathway is 46 ------- saturated at exposures to DCM of less than 400 ppm. However, following inhala- tion exposure of 50 to 2,000 ppm dibromo- or bromochloromethane, the carbon dioxide pathway in animals given pyrazole (to inhibit microsomal oxidation) was first order. Gargas et al. (1985) concluded that the carbon monoxide pathway was high affinity, low capacity, and the carbon dioxide pathway could be described as low affinity, high capacity. Thus, the data suggest that at low doses most of the inhaled DCM (or other dihalomethanes) is metabolized via the carbon monoxide pathway with little or no metabolism occuring via the carbon dioxide pathway. Gargas et al. (1985) did not measure carbon dioxide directly. Therefore, it is not possible to directly compare their data with the observations made by Yesair et al. (1977), Angelo (1985), and McKenna et al. (1982), who found that mice and rats excreted equal amounts of carbon monoxide and carbon dioxide. One laboratory reported results from similar experiments using mice and rats. Angelo (1985) gave 10 mg/kg or 50 mg/kg l^C-DCM intravenously to groups of mice or rats and then measured the amount of exhaled DCM, carbon dioxide, and carbon monoxide for 4 hours post-exposure. The data from these experiments are summarized in Table 13. Rats given 10 mg/kg or 50 mg/kg immediately (less than 20 minutes) exhaled about 35% of the administered dose, while mice immediately exhaled 45% of the low dose and 60% of the high dose (Table 13). The data suggest that in both species the amount of DCM metabolized to carbon monoxide and carbon dioxide is not proportional to the administered dose. This suggests that metabolic saturation is being approached. Without more dose levels one cannot determine the extent of the saturation. The data imply that over the dose range studied the rat and mouse appear to be capable of metabolizing the same amount of DCM over a 4-hour period. This cross species comparison must be considered with caution, however, since even at 47 ------- TABLE 13. METABOLISM OF DICHLOROMETHANE (DCM) FOLLOWING INTRAVENOUS ADMINISTRATION OF 10 mg/kg OR 50 mg/kg EFFECT OF DOSE ON EXHALATION OF DCM, CARBON MONOXIDE, AND CARBON DIOXIDE BY MICE AND RATS Dose 10 mg/kg DCM CO C02 50 mg/kg DCM CO C02 10 mg/kg DCM CO C02 50 mg/kg DCM CO C02 0-20 Minutes 32. 2a 0.2 2.4 35.2 0.7 0.8 45.6 1.9 11.9 59.5 0.8 6.2 20 - 40 Minutes 6.6 0.3 3.3 11.0 0.1 1.4 2.1 2.3 4.1 4.1 1.2 3.6 Time 40 - 60 Minutes Rats 2.7 1.0 3.0 5.6 0.4 1.5 Mice 0.4 3.4 1.4 0.9 2.1 1.5 60 - 240 Minutes 2.9 11.1 7.3 5.8 7.4 6.2 0.3 7.3 2.6 0.6 4.3 2.4 Total 44.5 12.7 16.0 57.6 8.7 9.9 48.4 14.9 20.2 65.0 8.4 13.7 aValues are percent of administered dose. SOURCE: Adapted from Angelo, 1985. 48 ------- 50 mg DCM/kg, the maximum rate of conversion to carbon monoxide observed in the rat is about 14 ymol/kg/hour, which is only about 35% of the reported vmax (Rodkey and Collison, 1977a, b; Gargas et al., 1985). The data indicate that mice and rats given small amounts of DCM readily metabolize it to carbon monoxide and carbon dioxide in roughly equal proportions, which is also sup- ported by the findings of Yesair et al. (1977) and McKenna and Zempel (1981), who assessed the metabolism of small amounts of DCM in mice or rats and ob- tained similar results. Rodkey and Collison (1977b) exposed a group of four rats to 1,255 ppm DCM in a chamber having a closed rebreathing system and determined the amount of carbon monoxide exhaled as a function of time. After a short lag period, the rats exhaled carbon monoxide at a rate of 30 ymol/kg/hour. The authors stated that they obtained similar results after giving animals the same dose of DCM intraperitoneally. In a second experiment, a group of four rats exposed to 6,462 ppm DCM exhaled 40 ymol carbon monoxide/kg/hour. These data indicate that in the rat the microsomal pathway is almost saturated at DCM exposures as low as 1,255 ppm and that the Vmax equals about 30 to 40 ymol/ kg/hour. Gargas et al. (1985) recently reported that the Vmax of the micro- somal pathway in rats is 47 ymol/kg/hour. Rodkey and Collison (1977a) also measured DCM biotransformation to carbon monoxide and carbon dioxide. Rats were exposed to 1,630 ppm of DCM in a chamber with a closed rebreathing system, and the amount of carbon monoxide and carbon dioxide produced were determined at 7 to 9 hours postexposure. The amount of carbon monoxide formed was about 90 ymol, whereas the amount of exhaled carbon dioxide formed was only 56 ymol. These data appear to differ significantly from those reported by McKenna et al. (1982) and DiVincenzo and Hamilton (1975), who found that slightly more carbon dioxide than carbon 49 ------- monoxide was exhaled. However, Rodkey and Collison (1977b) accounted for only about 75% of the administered dose. It has been estimated that mice metabolize DCM to carbon monoxide at a rate of about 19 ymol/kg/hour (U.S. EPA, 1985). This value was determined by assuming that the carbon monoxide formation was constant over the 12-hour postexposure period in the experiments performed by Yesair et al. (1977). Given the volatility of DCM, it is unlikely that the substrate concentration was saturating for 12 hours. Thus the maximum rate for carbon monoxide forma- tion in the mouse could well exceed 19 ymol/kg/hour. The in vivo metabolic studies confirm the data obtained in vitro that DCM is biotransformed to carbon monoxide and carbon dioxide. There are some data that indicate that formation of carbon monoxide in the rat reached Vmax at exposures of around 1,200 ppm. The data are limited on the exposure concen- tration required to saturate the carbon dioxide pathway. Comparative data on species differences and similarities are limited. At low doses both rats and mice metabolize similar amounts of DCM to carbon monoxide and carbon dioxide. 3.2.3. Use of Pharmacokinetic/Metabolism Data for Risk Calculation This section will discuss how some of the available pharmacokinetic/ metabolism data might be used in the estimation of carcinogenic risk, and will also discuss one approach based on the use of a physiological-based pharmaco- kinetic model as detailed in a paper submitted to the EPA on July 15, 1985, by the Dow Chemical Company titled "Physiologically-Based Pharmacokinetics and the Risk Assessment Process for Methylene Chloride" (Reitz and Andersen, 1985). The physiologically-based pharmacokinetic inhalation model used by Reitz and Andersen (1985) is a modification of the model developed by Ramsey and Andersen (1984). Some of the data used in the risk assessment for DCM was 50 ------- taken from a paper by Gargas et al. (1985). The inhalation model consists of a series of mass balance differential equations describing parent chemical concentrations within four tissue groups (liver, lung, slowly perfused tissue, and rapidly perfused tissue). In addi- tion, four other equations are included to describe the metabolism of DCM. All metabolism is assumed to occur in the lungs and liver. Physiological constants (volumes of tissue and blood flows) were taken from the published literature. Some partition coefficients were measured in vitro (blood/air partition coefficients for rat, mouse, hamster, and human blood), and others were assumed (lung tissue, and all tissue/air partition coefficients for mouse and human tissue). The respiration rate used in the model is for a person at rest, and the respiration rate used for the mouse is 50% higher than that typically used by the EPA's Carcinogen Assessment Group (CAG). Interspecies conversion of enzyme activities was based on the measured activity for the monooxygenase and glutathione s-transferase activities using surrogate sub- strates in samples of lung and liver tissue from humans, rats, hamsters, and mice. Reitz and Andersen (1985), using data derived from a physiologically- based pharmacokinetic inhalation model, suggest that the CAG has substantially overestimated the carcinogenic risk associated with exposure to DCM. A review of the model parameters shows that the model takes into account many of the necessary physiological parameters, such as blood flow and respira- tion, which might affect the uptake of DCM and its subsequent metabolism. However, as reasoned in the discussion that follows, the model needs validation. The model seems to be reasonably good for predicting blood levels of DCM during exposure via inhalation but appears to overestimate blood concentrations of DCM postexposure. Conversely, the model appears to underestimate both arterial and venous blood levels of DCM following intravenous exposure. Both 51 ------- the overestimation (inhalation) and the underestimation (intravenous) might be due to the assumption made by Reitz and Andersen (1985) that the DCM in the tissues is in equilibrium with the DCM in the blood based on the use of data from in vitro equilibrium studies. For example, possible reasons for the postinhalation overestimation might be a slow diffusion of DCM out of the tissues and the postintravenous administration underestimation might be a lack of blood/tissue equilibrium. The CAG suggests that the assumption regarding equilibrium between blood and tissue made by Reitz and Andersen (1985) may require some adjustment. Indeed, the data from Angelo et al. (1984) suggest that the biological half-life of DCM in tissues is much greater than that predicted by the Reitz and Anderson (1985) model, which would increase the risk associated exposure DCM. Reitz and Andersen (1985) used the monooxygenase (microsomal) and gluta- thione s-transferase (cytosolic) activity in liver and lung tissues to deter- mine the relative amount of metabolism by each pathway in tissue from humans, rats, hamsters, and mice. The enzymatic activities selected by Reitz and Andersen (1985) are from a study by Lorenz et al. (1984). The monooxygenase activities were obtained using 7-ethoxycoumarin as a substrate, and the gluta- thione s-transferase activities were obtained using l-chloro-2,4-dinitrobenzene as a substrate. A limited review of the literature suggests that the values used by Reitz and Andersen (1985) may not reliably reflect the appropriate rates of DCM metabolism in each tissue nor the amount of enzymatic activity in each tissue. Using 7-ethoxycoumarin as a substrate, the monooxygenase activity in rat lung tissue is 0.13 as active as that in rat liver tissue, but lung tissue has only 0.0069 the activity of liver tissue when benzo[a]- pyrene is the substrate (Table 14). The substrate used can also make a significant difference in the interspecies comparison of enzymatic activities. 52 ------- For example, rat liver has twice the monooxygenase activity compared to human liver using 7-ethoxycoumarin as a substrate, but using benzo[a]pyrene, rat liver has more than six times the activity of human liver tissue (Table 14). Similarly, the specific activity of glutathione s-transferase varies signif- icantly depending on the substrate used for assay. Pabst et al. (1974) re- ported a specific activity (vimol product formed/mg protein/minute) of 60 using l-chloro-2,4-dinitrobenzene compared to 4.1 using l,2-dichloro-4-nitro- benzene. Thus, the selection of a surrogate substrate for DCM could markedly influence the apparent metabolism of DCM in various tissues within one species and between species. Lastly, in vitro data suggest that liver microsomes may metabolize DCM at a rate five times faster than lung microsomes (Kubic and Anders, 1975). The enzyme data used by Reitz and Andersen (1985) in their model would predict that the liver microsomes metabolize DCM at a rate eight times faster than lung microsomes. These large differences suggest that the most desirable enzyme data to predict cross-species comparisons should be obtained using DCM as a substrate. Lorenz et al. (1984) obtained enzyme activity data using tissues from Sprague-Dawley rats and NMRI mice, while the NTP carcinogen bioassay used Fischer 344 rats and B6C3F1 mice. Andersen et al. (1980) observed that the hepatic nonprotein sulfhydryl concentration for Holtzman rats was 4.12 mmol/kg and for Fischer 344 rats was 6.24 mmol/kg. Thus, the direct comparison, as performed by Reitz and Andersen (1985), of the carcinogenic response in one strain of rats with the enzymatic activity in another strain should be consid- ered with caution. Reitz and Andersen (1985) calculated a Vmax for the monooxygenase path- way in humans by scaling the Vmax for the rat to the 0.7 power of the body weight ratio for the two species. This gave a value of 85.9 mg/hour. If the 53 ------- TABLE 14. METABOLISM OF VARIOUS SUBSTRATES BY MICROSOMES 7-ethoxycoumari n& human rat mouse hamster benzo[a]pyrenec human rat hamster Liver3 0.42 0.81 1.76 2.57 0.52 3.35 4.38 Lunga 0.0006 0.11 0.73 0.16 0.011 0.023 0.022 Lung/Liver 0.0014 0.13 0.41 0.06 0.02 0.0069 0.0050 aAll values are nmol/min/mg protein. bLorenz et al., 1984. cPrough et al., 1979. f°r humans had been calculated using data from the mouse, the value would have been 272 mg/hour. Similarly, if the Vmax for the rat was calculated by scaling the Vmax for the mouse, the value for the rat would have been 5.0 mg/hour rather than the 1.58 mg/hour obtained experimentally by Reitz and Andersen (1985). These differences suggest that the data selected for inter- species extrapolation, as well as some of the assumptions made by Reitz and Andersen (1985), require refinement. At this time it is not clear how such adjustment would effect the interspecies risk calculation. The human liver enzymatic activity reported by Lorenz et al. (1984) was measured in 15 samples of tissue. For only six samples were both monooxygenase and GSH determined. Some of the tissue samples were taken from patients who 54 ------- were undergoing chemotherapy for Hodgkin's lymphoma. Lorenz et al. (1984) gave no information on individuals; thus, it is not possible to determine variations that might be associated with the chemotherapy treatment. Similarly, two of the patients from whom lung tissue was taken were on drug therapy. No liver and lung tissues were taken from the same patient. The data on human tissue should be considered preliminary and not representative of the whole population. Gargas et al. (1985) performed a series of inhalation experiments using dibromo- and bromochloromethanes as surrogates for DCM. The data suggested to the authors that at low doses most of the inhaled DCM would be metabolized via the carbon monoxide pathway with little or no metabolism occuring via the carbon dioxide pathway (see section 3.2.2. for a discussion of the data). These observations have led Reitz and Andersen (1985) to predict that the reactive intermediate formed by the microsomal pathway is less toxic than the reactive intermediate formed via the cytosolic pathway. To support their hypo- thesis, they cite the lack of a significant increase in the carcinogenic re- sponse to DCM in drinking water at 5, 50, 125 and 250 mg/kg/day (NCA, 1982a, b; 1983). EPA's previous and present assessment agreed that there was no signifi- cant increase in the carcinogenic response in the NCA study, but concluded that the study was borderline for carcinogenicity in Fischer 344 rats and B6C3F1 male mice. The Agency also concluded that the less than significant response was consistent with the calculated risk at higher doses. If Reitz and Andersen (1985) are correct in their assessment that the two reactive intermediates have different toxic properties, additional information would be required to support this hypothesis, such as, data on binding to macromolecules (protein and DNA) of the two intermediates which showed significant differences as an independent way of testing their hypothesis. 55 ------- Preliminary data submitted by Reitz (1985) from three experiments similar to those of Gargas et al. (1985), in which exhaled carbon monoxide and carbon dioxide were measured during exposure to DCM, suggest that the original con- clusion made by Gargas et al. (1985) and Reitz and Andersen (1985) may require some revisions. Reitz (1985) exposed one group of four mice to 1,500 ppm DCM and another group of four mice to 50 ppm DCM. The group exposed to 1,500 ppm DCM exhaled more carbon dioxide relative to carbon monoxide than did the group exposed to 50 ppm DCM. The group exposed to 1,500 ppm DCM exhaled carbon dioxide:carbon monoxide in a range of 1.5 to 4.3 (hours 2 through 6 of the exposure period), while the group exposed to 50 ppm DCM exhaled carbon dioxide: carbon monoxide in a range of 0.62 to 5.3 (hours 2 through 6 of the exposure period). These preliminary data do not support the conclusion of Gargas et al. (1985) that at low doses little or no metabolism occurs via the carbon dioxide pathway. Reitz (1985) also exposed another group of four mice to 1,500 ppm DCM after treatment with pyrazole. The pyrazole reduced the carbon monoxide exhaled by about 81% and the carbon dioxide by about 59%. These preliminary data appear to support the suggestion made by Gargas et al. (1985) that the microsomal pathway may metabolize some DCM to carbon dioxide via a glutathione intermediate. It should be noted that DCM metabolism to carbon dioxide via the microsomal pathway would require the formation of the same reactive intermediate as is formed via cytosolic metabolism. Thus, according to the scheme proposed by Gargas et al. (1985) both pathways become equally toxic. Reitz and Anderson (1985) have stated that the "arbitrary interspecies conversion factors" used by the CAG result in an additional 2.95-fold increase in risk for man when compared to the mouse. A search for the reason for the 2.95 differential shows that the 2.95-fold decrease in the Reitz and Andersen 56 ------- calculations was obtained by reducing human respiration by about 60% to the resting state and increasing the mouse respiration by 50%. Lastly, Reitz and Andersen (1985) have not compared predictions from their quantitative model with measured concentrations of DCM metabolites. Such meta- bolic data is available from several studies (McKenna et al., 1982; Rodkey and Collison, 1977a, b; Yesair et al., 1977; Angelo et al., 1984). An analysis of the available data is a necessary step in the validation of the model. Overall, the Reitz and Andersen (1985) approach, using metabolic data in a physiologi- cally-based pharmacokinetic model, offers an interesting approach to the estima- tion of human risk from exposure to DCM. Given the uncertainty about some of the assumptions used and the variance between some of the modeling components and results with the available data, the model and the use of surrogate sub- strates needs validation. The CAG believes that when models are developed, they may well provide a useful approach for using pharmacokinetic/metabolism data for risk evaluation. Other approaches for utilizing the available pharmacokinetic/metabolism data are worthy of mention. Most of the data currently available on the pharmacokinetics and metabolism of DCM is for the rat. In particular, the study by McKenna et al. (1982) provides information on the metabolism of DCM following a 6-hour inhalation exposure in a dose range (50 to 1,500 ppm) that overlaps with the NTP bioassay dose range (see Table 10). The McKenna et al. data on metabolism are limited in that they were obtained following the cessa- tion of exposure, the strain of rat utilized differed from the strain tested by NTP, and the study included no dose comparable to the 4,000 ppm high dose in the bioassay. Nonetheless, the McKenna et al. data are the most relevant metabolic data available for comparison with NTP bioassay results. Therefore, an exploratory analysis of how metabolic data on DCM might affect the assessed 57 ------- risks was conducted. Table 15 and Figure 8 present estimates on DCM metabolism derived from the McKenna et al. study and show a family of curves that would, by visual inspection, appear to be reasonable extrapolations of the McKenna et al. data (Figure 8) to the higher bioassay doses. These data are expressed in terms of the fraction of the total rat body burden of l^C-DCM equivalents which are metabolized at the end of the experiment (Table 15). Because reactive metabolites may be formed by both pathways of DCM metabolism, it was felt most appropriate to focus on total metabolism in the exercise. It is important to note that the hypothesis inherent in these calculations is that metabolites of DCM alone, and not the parent compound, are responsible for carcinogenicity; the validity of this hypothesis is open to discussion. The estimated effective doses are used in conjunction with the NTP data on the incidence of mammary neoplasms in female rats (the strongest effect in rats) to estimate the 95% upper-bound slope (q^) of the dose-response using the GLOBAL83 multistage dose-response model. The q^ values for rats, obtained using the effective dose estimates, are 0.47 x 10'3, 0.44 x lO'3, and 0.41 x 10~3 ppnr1 under the experimental dose schedule for hypotheses I, II, and III, respectively. Because all three values are close together, the average, 0.44 x 10~3, has been used to calculate human q^ values in Chapter 4 of this document. For projection to humans, the fraction of inhaled DCM metabolized must be estimated. Again, little data are available to estimate these quantities, and the available data show a considerable range. The Health Assessment Document for Dichloromethane (U.S. EPA, 1985) estimated that, on the average, studies showed that 42% of inhaled DCM was absorbed; this value was based on studies in which exposure was for 2 hours or longer in duration, and in all except one case involved exposure to 100 ppm or more of DCM. The one study using a lower concentration, 50 ppm, yielded a retention of 70%; short-term studies also 58 ------- TABLE 15. ESTIMATED FRACTION OF THE ADMINISTERED DOSES USED IN THE NTP BIOASSAY METABOLIZED Administered dose (ppm) Percent of dose metabolized Effective dose3 (ppm) Hypothesis I 1,000 2,000 4,000 Hypothesis II 1,000 2,000 4,000 Hypothesis III 1,000 2,000 4,000 58% 38% 22% 58% 40% 29% 58% 42% 37% 580 760 880 580 800 1,180 580 840 1,480 aThe effective dose is defined as the inhaled dose minus the fraction exhaled unchanged. NOTE: If the quantity of DCM metabolized (mg/kg) is assumed not to decrease with dose, the experimental data point showing 65% of body burden excreted unchanged at 1,500 ppm sets upper limits of 66% and 83% on the fraction exhaled unchanged at 2,000 ppm and 4,000 ppm respectively. 59 ------- 09 PERCENTAGE OF DOSE EXHALED UNMETABOLIZED oo tu CD Oi -1 Q. _i. 3 to Q. O O O o Ol 3 ft) o o o o o O O CO m co co o *-» O 2, T3 O O o rf- O) 1/1 cu rt 3- (D —I "O cr o QJ in Q. O in 0) o o o Ol o o o o I 10 o I CO o I o Ol o I o> o I o I CD o I ------- yielded relatively high values for retention. It is assumed that at low doses the fraction of absorbed DCM that is later exhaled unmetabolized is small and can be omitted from the analysis. Therefore, both values, 42% and 70%, are used as a range to adjust the human inhaled dose. The body surface area inter- species extrapolation is then applied to estimate values of the human unit risk based on the inhaled quantities of DCM. These calculations yield an estimate * Q Q 1 of q^ in humans in the range 2.7 x 10" J to 4.5 x 10"° (mg/kg/day) . For comparison, when extrapolation is done directly on the basis of inhaled doses, * Q 1 qj = 2.4 x 10"° (mg/kg/day)"1 in humans. Thus, in this example, roughly estimated pharmacokinetic/metabolic corrections to the applied doses yielded risk estimates that range somewhat higher than those estimated directly by using the dose administered to animals. In addition to these calculations, an attempt was made to estimate ab- sorbed DCM doses in rodent experiments using information on quasi-steady-state blood levels of DCM at the end of inhalation studies. To generate uptake estimates, we attempted to estimate the effective partition coefficient between blood and alveolar air from experiments in which the exhalation of DCM was measured after the cessation of exposure. Estimates of the partition coeffi- cient showed a wide range and were substantially lower than measured in _in_ vitro experiments. While the wide range of values obtained showed a lack of merit for pursuing this approach using the currently available data, preliminary calculations led to absorbed dose estimates higher than those presented under Hypothesis I. 3.2.4. Human Studies There have been three studies in which volunteers have been exposed to DCM. DiVincenzo and Kaplan (1981a, b) evaluated the conversion of DCM to carbon monoxide in sedentary, nonsmoking individuals and in individuals engaged 61 ------- in physical activity. Exposure, in a chamber, was to 50, 100, 150, or 200 ppm DCM for 7.5 hours or for 7.5 hours/day for 5 consecutive days (Table 16). The metabolism of DCM was also studied in men engaged in physical activity. Di- Vincenzo and Kaplan (1981a) found that in sedentary individuals the pulmonary uptake of DCM was linear over the range studied (50 to 200 ppm), and that excretion of carbon monoxide was proportional to the pulmonary uptake of DCM. The authors noted that only 25% to 34% of the DCM taken up was converted to carbon monoxide, and therefore hypothesized, based on data from animal studies, that up to 70% was metabolized to carbon dioxide. TABLE 16. METABOLISM OF DICHLOROMETHANE (DCM) TO CARBON MONOXIDE IN HUMANS Carbon monoxide Exposure/uptake (pmol/kg) 50 ppm (79.1 ymol/kg) 18.6 100 ppm (152.9 pmol/kg) 28.6 150 ppm (219.7 pmol/kg) 71.4 200 ppm (301.0 ymol/kg) 87.4 NOTES: DCM was given in a single inhalation exposure for 7.5 hours. The amount of postexposure carbon monoxide exhaled was measured for 24 hours. For the purpose of comparison, it was assumed that the weight of each volun- teer was 70 kg. SOURCE: DiVincenzo and Kaplan, 1981a. Sedentary volunteers exposed once for 7.5 hours to 50, 100, 150, or 200 ppm of DCM had peak COHb concentrations of 1.9%, 3.4%, 5.3%, and 6.8%, respec- tively. When sedentary volunteers were exposed to 100, 150, or 200 ppm DCM for 7.5 hours/day for 5 consecutive days, the COHb of those exposed to 150 ppm 62 ------- or 200 ppm increased to levels above that of the single exposure. The volunteer exposed to 200 ppm DCM had a COHb of about 5% on day 1 and a COHb of about 6.5% on day 4. The carbon monoxide concentrations in the breath of the volun- teers also increased each day during exposure to high concentrations (150 to 200 ppm) of DCM. Conversely, the peak level of expired DCM in the breath of the volunteers remained constant (peak level did not change) throughout the exposure period. DiVincenzo and Kaplan (1981b) also measured DCM uptake and metabolism during exercise. The data showed (Table 17) that uptake of DCM is directly related to work intensity, and that the amount of DCM metabolized to carbon monoxide is directly proportional to pulmonary uptake. This is illustrated by the finding that a sedentary volunteer exposed to 200 ppm DCM exhaled 6.1 mmol of carbon monoxide (with a pulmonary uptake of 21.1 mmol DCM), while an exer- cising volunteer exposed to just 100 ppm DCM exhaled 11.8 mmol of carbon monoxide (with a pulmonary uptake of 41.9 mmol DCM). The latter observation suggests that the metabolic capacity of sedentary people exposed to 400 ppm DCM would not be exceeded. McKenna et al. (1980) exposed volunteers to 100 or 350 ppm DCM for 6 hours and measured various parameters, including blood and exhaled air levels of DCM, COHb, and exhaled carbon monoxide (Figures 9 and 10). The data showed that the blood level of DCM for both concentrations reached a steady-state in about 2 hours. At the end of the 6-hour exposure, the COHb concentration of the group exposed to 350 ppm DCM was 1.4-fold higher than that of the group exposed to 100 ppm. Likewise, the concentration of exhaled carbon monoxide in the group exposed to 350 ppm DCM was 2.1-fold higher than that of the group exposed to 100 ppm. McKenna et al. (1980) interpreted their findings to mean that the nonlinearity between administered dose and the COHb and carbon 63 ------- TABLE 17. EFFECT OF EXERCISE ON THE PULMONARY UPTAKE AND METABOLISM OF DICHLOROMETHANE (DCM) DURING EXPERIMENTAL EXPOSURES TO DCM VAPOR Work intensity Volunteer (mL Q^ min kg" ) 1 4 2 14 3 15 19 4 16 28 Pulmonary uptake of DCM (mmol ) 10.7 19.4 30.7 36.4 28.8 41.9 Total pulmonary excretion of CO (mmol ) 2.7 5.1 10.2 14.3 10.4 11.8 SOURCE: DiVincenzo and Kaplan, 1981b. monoxide levels is an indication that metabolic saturation was being approached. However, data from animal studies suggest that both pathways, microsomal and cytosolic, are of about equal capacity. Therefore, it is necessary to establish the capacity of the carbon dioxide pathways in humans. Thus, the authors' conclusion, based on data from only one pathway, that metabolic saturation in humans is achieved at less than 350 ppm DCM requires further investigation. 64 ------- 0 2 4 6 8 10 12 14 16 18 20 22 24 Hours Figure 9. COHb level in volunteers exposed to 100 ppm or 350 ppm dichloromethane. SOURCE: McKenna et al., 1980. 65 ------- o c: o O 7* ID CD rl- IO 00 O CTi O T1 -> -•• to oo c in -» o ro T3 t-' QL -•• m OX O —" "1 O> O Q. fO O r+ O) 3" -J Q) CT 3 O n> 3 o 3 O X Q. n> Exhaled CO (ppm) (t) (T) ft) O to (D Q. O O -o T3 ------- 3.2.5. Summary The results of both In vitro and In vivo studies indicate that DCM is metabolized via two pathways. One pathway yields carbon monoxide as an end product, and the other yields carbon dioxide as an end product with formalde- hyde and formic acid as metabolic intermediates. Each pathway involves formation of a metabolically-active intermediate that is theoretically capable of irreversibly binding to cellular macromolecules (Ahmed et al., 1980). A comparative analysis of the capability of various tissues to metabolize DCM indicates that the liver is the primary site of metabolism, with some metab- olism taking place in the lung and kidney. In vivo data suggest that when rats or mice are exposed to high concentrations of DCM (50 mg/kg or 500 ppm or more), they exhale more carbon dioxide than carbon monoxide (Yesair et al., 1977; McKenna et al., 1982). At exposure to low concentrations of DCM (1 to 10 mg/kg or 50 ppm) both pathways are utilized about equally (McKenna et al., 1982; Yesair et al., 1977). The data from rat studies also suggest route independence since exposure to low doses results in similar metabolic profiles (Rodkey and Collison, 1977a). A comparative analysis of the data from in vivo studies in mice, rats, and humans indicates that all three species metabolize DCM to carbon monoxide. Both mice and rats metabolize DCM to carbon dioxide. There are no human data on the metabolism of DCM to carbon dioxide. However, based on animal data and on uptake data, it is likely that this pathway is functional in humans (Di- Vincenzo and Kaplan, 1981a). Thus, as far as data are available, it appears that mice, rats, and humans metabolize DCM by the same pathways, and at low doses the metabolic profile for all three species appears to be qualitatively simi lar. 67 ------- A few studies are available on the pharmacokinetic/metabolism of animals and humans following inhalation exposure to DCM. In one animal study rats were exposed to 50, 500, or 1,500 ppm DCM (McKenna et al., 1982). The groups of rats exposed to 500 or 1,500 ppm DCM had the same COHb, suggesting that the carbon monoxide (microsomal) pathway has been saturated. There are no similar data on the carbon dioxide (cytosolic) pathway. One group of investi- gators suggested that the carbon monoxide pathway is approaching saturation in humans at less than 350 ppm DCM (McKenna et al., 1980); however, data from another study suggest that the metabolic response to an exposure of 400 ppm DCM might still be linear with dose (DiVincenzo and Kaplan, 1981b). Data from the animal study (McKenna et al., 1982) suggest that nonlinear kinetics were observed at about 500 ppm in the rat. If the conclusions of McKenna et al. (1982) are correct, then nonlinear kinetics may not be a relevant consideration in assessing the exposure concentrations of 1,000, 2,000, and 4,000 ppm DCM used in the NTP rat carcinogenesis study. However, the data reported by Gargas et al. (1985) suggest that the carbon dioxide pathway is not saturated and is a first-order reaction at doses up to 2,000 ppm. If metabolic saturation determines the effective dose, then at doses that exceed metabolic saturation the carcinogenic response should be similar, assuming there is no change in the concentration or biological half-life of the reactive intermediate(s). In the NTP bioassay there is a clear dose-response relationship between the car- cinogenic response and the doses tested. Thus, it would appear that nonlinear kinetics alone does not explain the limits to the carcinogenic response. Currently no data exist with regard to the effect of dose on the biological half-life of the reactive intermediates of DCM nor on what role unmetabolized DCM or any metabolites might play in the carcinogenic response. 68 ------- Approximate estimates of the metabolized doses in the NTP bioassay can be predicted using the data from McKenna et al. (1982) on DCM body burdens at the end of the inhalation experiments in rats. The result of this adjustment in the administered dose of 1,000 ppm represents an effective dose of 580 ppm; the administered dose of 2,000 ppm becomes an effective dose of 760 to 840 ppm and the administered dose of 4,000 ppm becomes an effective dose of 880 to 1,480 ppm. The calculated unit risk associated with exposure to DCM increases about twofold using the effective dose rather than the administered dose. There are, however, uncertainties in this extrapolation which limit the degree of confidence in these calculations. Of most concern is the lack of metabolic data at the dose levels used in the NTP bioassay. Therefore, at this time, it is appropriate to limit the introduction of additional uncertainties by calculating risk estimates based on the administered dose. At present, data from the available pharmacokinetic/metabolism studies suggest that it is appropriate to consider the following information in the risk evaluation for DCM. The available data include: o DCM is metabolized to carbon monoxide and carbon dioxide via formation of a biologically reactive intermediate; e Mice, rats, and humans are capable of metabolizing DCM; e At low doses the profile of metabolic end products excreted by mice, rats, and humans appears to be similar; and e Route of exposure does not strongly affect the metabolic profile in tested species. There are, however, no pharmacokinetic/metabolism data that: « Indicate species similarities or differences at high doses, o Relate the carcinogenic response at the doses used in the NTP bioassay to metabolic saturation, 69 ------- e Allow for the determination of the relationship between the adminis- tered dose and the effective dose, and • Allow for determining the concentration and biological half-life of the reactive intermediate(s) responsible for the carcinogenic response, The available data indicate that qualitatively, cross-species comparison of the carcinogenic response between rodents and humans is appropriate. The lack of data on effective versus administered dose and the lack of detailed metabolic information at high doses do not allow for the use of quantitative considerations of pharmacokinetics/metabolism in the assessment of the carcino- genic response to DCM. Thus, in the absence of reliable information on the effective dose of DCM in the dose range of the NTP carcinogenesis bioassay, the use of administered dose is preferred for the quantitative estimation of risk from exposure to DCM. 70 ------- 4. QUANTITATIVE ESTIMATION (USING THE NTP INHALATION BIOASSAY) In February of 1985 the Office of Health and Environmental Assessment published a comprehensive document on the health effects of dichloromethane (methylene chloride, DCM) (U.S. EPA, 1985). After the completion of this report, the NTP released the findings of an inhalation toxicology and carcino- geoesis study of DCM in F344/N rats and B6C3F1 mice (NTP, 1985, draft). The qualitative findings of this study are discussed in a preceding section of this document; here, the NTP findings are used to develop estimates of unit incremental cancer risks for humans exposed to DCM. Variations in extrapo- lated risks using different cancer end points from the NTP study are discussed, as well as the influence of the dose-response model selected. The quantita- tive findings are also compared with earlier experimental carcinogenesis studies of DCM and with the limited information available from epidemiologic studies. 4.1. SUMMARY OF THE NTP FINDINGS USED FOR QUANTITATIVE ANALYSIS In an earlier section of this document, the end points of mammary and subcutaneous tumors in rats and lung and liver tumors in mice were determined to be the sites where the NTP study produced the strongest findings of car- cinogenicity for DCM. Tables 18 and 19 present the NTP tumor incidence find- ings for these sites. The denominator for each data point is the number of animals that were examined at the specific tumor site. The statistical significance of the NTP findings has been discussed earlier in this report; the findings are summarized in Tables 18 and 19. The data in these two tables will serve as the basis for developing quan- titative risk estimates. The EPA Proposed Guidelines for Carcinogen Risk 71 ------- TABLE 18. SUMMARY OF THE NTP INHALATION STUDY OF DICHLOROMETHANE: FINDINGS FOR MAMMARY AND SUBCUTANEOUS TUMORS IN RATS Dose Site/tumor Control 1,000 ppm 2,000 ppm 4,000 ppm Males Mammary gland: adenoma or fibroadenoma3 Subcutaneous tissue: fibroma0 Mammary gland or subcutaneous tissue: adenoma, fibroadenoma or fibroma3 Females Mammary gland: fibroadenoma3 Mammary gland: adenoma, fibroadenoma, adenocarcinoma, or mixed tumor, malignant3 0/50 1/50 1/50 5/50 7/50 0/50 1/50 1/50 ll/50b 13/50b 2/50 2/50 4/50 13/50b 14/50b 5/50b 4/50 9/50d 22/50d 23/50d 3A11 trend tests for tumor incidence positive at p < 0.01 nominal level. bOne or more positive pairwise comparison(s) with control group p < 0.05 nominal level. C0ne or more positive trend test(s) for tumor incidence p < 0.05 nominal level. ^Al 1 pairwise comparisons with control group positive p < 0.01 nominal level. NOTE: Tumor trend tests reported by NTP: life table, incidental tumor, and Cochran-Armitage tests; pairwise tests: life table, incidental tumor, and Fisher Exact tests. SOURCE: NTP, 1985. ------- TABLE 19. SUMMARY OF THE NTP INHALATION STUDY OF DICHLOROMETHANE FINDINGS FOR LUNG AND LIVER TUMORS IN MICE Dose Tumor Control 2,000 ppm 4,000 ppm Males Alveolar/bronchiolar adenoma3 3/50 19/50b 24/50b Alveolar/bronchiolar carcinoma3 2/50 10/50C 28/50b Alveolar/bronchiolar adenoma or carcinoma3 5/50 27/50b 40/50b Hepatocellular adenomad 10/50 14/49C 14/49^ Hepatocellular carcinomad 13/50 15/49 26/49c Hepatocellular adenoma or carcinoma3 22/50 24/49 33/49c Alveolar/bronchiolar or hepatocellular carcinoma6*^ 15/50 21/49 39/49b Alveolar/bronchiolar or hepatocellular adenoma or carcinomae»f 27/50 34/49^ 45/49b Females •—i CO Alveolar/bronchiolar adenoma3 2/50 23/48b 28/48b Alveolar/bronchiolar carcinoma3 1/50 13/48b 29/48b Alveolar/bronchiolar adenoma or carcinoma3 3/50 30/48b 41/48b Hepatocellular adenoma3 2/50 6/48 22/48b Hepatocellular carcinoma3 1/50 ll/48b 32/48b Hepatocellular adenoma or carcinoma3 3/50 16/48b 40/48b Alveolar/bronchiolar or hepatocellular carcinomae»f 1/50 21/48b 43/47b Alveolar/bronchiolar or hepatocellular adenoma or carcinomae»f 5/50 36/48b 46/47b 3A11 trend tests for tumor incidence positive at p < 0.01 nominal level. bAll pairwise comparisons with control group positive p < 0.01 nominal level. C0ne or more positive pairwise comparison(s) with control group p < 0.05 nominal level. dOne or more positive trend test(s) for tumor incidence at p < 0.05 nominal level. eDenominators are number of animals examined for tumors at both lung and liver sites. ^Tumor grouping not presented in NTP report; significance determined using Fisher Exact Test. NOTE: Tumor trend tests reported by NTP: life table, incidental tumor, and Cochran-Armitage tests; pairwise tests: life table, incidental tumor, and Fisher Exact tests. SOURCE: NTP, 1985. ------- Assessment (U.S. EPA, 1984) call for risk estimation using the combined inci- dence of statistically elevated tumors. The combined incidence of lung and liver tumors in mice is given for quantitative analysis and does not imply that the tumors are biologically related. 4.2. DOSE-RESPONSE MODEL SELECTION EPA's proposed guidelines (U.S. EPA, 1984) express the fact that there is no rigorously established scientific basis for the selection of a dose-response model to predict carcinogen risks at low doses. In the typical situation, where there is limited information on which to base the selection of a model, the guidelines express a preference for the multistage dose-response model. The guidelines place emphasis on the upper confidence limit (UCL) risk estimates derived from this model. The basis for the preference include the following: 1. The multistage model incorporates the current scientific opinion that multiple steps are involved in the process of cancer develop- ment, and that a chemical carcinogen can contribute to one or more of these steps. 2. The UCL of the multistage model produces a risk estimate that is linear at low doses (LLD). An LLD dose-response is expected when a carcinogen accelerates stages of the carcinogenic process that lead to the background occurrence of cancer in unexposed members of the population. 3. The UCL of the multistage model produces a "plausible upper-bound" estimate of risk, i.e., an estimate that is reasonable but is usu- ally as high or higher than estimates derived from other models; models that are not linear at low doses will generally lead to sub- stantially lower risk estimates. 74 ------- 4. The multistage UCL is stable under small changes in the input values for tumor incidence; in contrast, the maximum likelihood estimate (MLE) can be unstable if the results in one or a few animals are changed. Because of the role of genetic and mutational factors in the development of many cancers, a supportive biological argument for LLD can be made strongly for chemicals that are known to cause genetic damage. The Health Assessment Document for Dichloromethane (U.S. EPA, 1985) concluded that the weight of evidence shows that DCM is capable of causing gene mutations and has the potential to cause such effects in exposed human cells. Further testing to determine the strength of mammalian evidence was recommended. These findings provide additional support for the application of an LLD dose-response model in estimating DCM cancer risks. Considering both EPA's policy for carcinogen evaluation and the biological information available specifically for DCM, the multistage model was selected as the primary model to be applied in this risk assessment. Versions of the multistage model that incorporate time-to-tumor information have been devel- oped. While time-to-tumor models generally do not produce risk estimates that differ greatly from similar dichotomous models, the results are presented for comparison. Several other models are also presented for comparison with the multistage model. 4.3. APPLICATION OF THE MULTISTAGE MODEL TO THE NTP BIOASSAY DATA The multistage dose-response model, as incorporated in the GLOBAL83 computer program developed by Howe (1983), has been applied to the NTP bio- assay data given in Section 4.2. The model is applied to the experimentally administered doses, and thus directly produces low-dose risk estimates for rodents exposed to DCM following the same time pattern as the NTP bioassay. 75 ------- Section 3.2. of this document reviews the available data on the phar- macokinetics and metabolism of DCM and concludes that these data do not provide an adequate basis for modifications to the experimental doses for use in quantitative risk assessment. The GLOBAL83 program enables the user to select the highest degree of polynomial that the program will allow in the model. GLOBAL83 runs were made for a polynomial degree equal to the number of doses (counting the control) minus one. Tables 20, 21, 22, and 23 present the results of these computa- tions. Several conclusions can be drawn from these data. 1. The multistage model provides an adequate fit for all tumor groupings analyzed in both rats and mice. 2. In rats, the highest value for the UCL of the linear multistage term was obtained for mammary tumors in female rats. The highest value in males was lower by a factor of three. 3. In mice, the highest value for the UCL of the linear term was obtained in females having either adenomas or carcinomas of the lung and/or liver. The corresponding value for males was lower by a factor of two. 4. In mice, the male and female high-dose groups had a high percen- tage of tumors in both the lung and liver. As discussed in Section 4.4., the high-dose mice also showed elevated mortality in comparison to controls. In these circumstances, competing risks can lead to underestimates of risk attributable to indivi- dual tumor types. 5. The NTP (1985) noted that all male rat groups (including controls) experienced higher than usual mortality before final sacrifice. 76 ------- TABLE 20. GLOBAL83 MODEL PARAMETERS FOR THE NTP (1985) RAT DATA Site Males Mammary gland3 Subcutaneous0 Mammary gland or subcutaneousc Females Mammary gland^ Mammary gland6 q0 qjxlO"3 (ppm-1) 0.0 0.0 0.0191 0.00093 0.0197 0.00075 0.111 0.107 0.161 0.0985 Three-stage q2xlO'6 (ppm-2) 0.00696 0.00389 0.0117 0.0 0.0 model q3xlO-10 (ppm-3) 0.0 0.0 0.0 0.00544 0.00846 (ppnr1) 0.0311 0.0306 0.0540 0.164 0.164 aAdenoma or fibroadenoma. ''Fibroma. cAdenoma, fibroma, or fibroadenoma. ^Fibroadenoma. eAdenoma, fibroadenoma, adenocarcinoma, or mixed tumor, malignant. NOTES: qj = the ith power coefficient in the multistage model, q^ = the 95% upper confidence limit estimate of the linear coefficient. Values for the qi apply for rats exposed under the NTP protocol dose time schedule. Calculations are based on an extra risk analysis. ------- TABLE 21. COMPARISON OF THREE-STAGE GLOBAL83 ESTIMATES WITH OBSERVED TUMOR RESPONSE - RATS —I oo Tumor Males Mammary gland adenoma or f ibroadenoma Subcutaneous Mammary gland or subcutaneous Females Mammary gland f ibroadenoma Mammary gland, Observed 1,000 Control ppm 0.0 0.0 2.0 2.0 2.0 2.0 10.0 22.0 14.0 26.0 response (%) 2,000 ppm 4.0 4.0 8.0 26.0 28.0 4,000 ppm 10.0 8.0 18.0 44.0 46.0 Predi 2,000 Control ppm 0.0 0.7 1.9 2.4 1.8 3.0 10.5 19.7 14.9 22.9 cted response 4,000 ppm 2.7 3.6 6.4 28.1 30.6 (%) 4,000 ppm 10.5 8.2 18.8 43.7 45.6 Chi squared 0.66 0.06 0.42 0.30 0.46 all tumors ------- TABLE 22. GLOBAL83 MODEL PARAMETERS FOR THE NTP (1985) MOUSE DATA Two-stage model Tumor Males Lung3 adenoma Lung carcinoma Lung adenoma or carcinoma Li verb adenoma Liver carcinoma Liver adenoma or carcinoma Lung or liver carcinoma Lung or liver adenoma or carcinoma Females Lung adenoma Lung carcinoma Lung adenoma or carcinoma Liver adenoma Liver carcinoma Liver adenoma or carcinoma Lung or liver carcinoma Lung or liver adenoma or carcinoma ,0 0.0672 0.0398 0.105 0.237 0.285 0.565 0.333 0.771 0.0445 0.0202 0.0619 0.0356 0.0193 0.0576 0.0197 0.105 qjxlO'3 (pprrr1) 0.166 0.0 0.295 0.0306 0.0 0.0 0.0 0.0 0.242 0.0690 0.453 0.0 0.0 0.0 0.0 0.345 q2xlO'6 (ppm-2) 0.0 0.0481 0.0202 0.0 0.0283 0.0338 0.0739 0.107 0.0 0.0394 0.0032 0.0336 0.0655 0.101 0.147 0.148 tiSti 0.223 0.141 0.452 0.0810 0.142 0.195 0.190 0.429 0.310 0.233 0.579 0.0872 0.145 0.160 0.244 0.870 aLung = alveolar/bronchiolar. bLiver = hepatocellular. C0nly the linear parameters are given for the four-stage model. NOTES: qi = the itn power coefficient in the multistage model, qt = the 95% upper confidence limit estimate of the linear coefficient. Values for the q-j apply for mice exposed under the NTP protocol dose time schedule. Calculations are based on an extra risk analysis. 79 ------- TABLE 23. COMPARISON OF TWO-STAGE GLOBAL83 ESTIMATES WITH OBSERVED TUMOR RESPONSE - MICE oo o Observed response (%) Tumor Females Lung adenoma Lung carcinoma Lung carcinoma or adenoma Liver adenoma Liver carcinoma Liver carcinoma or adenoma Liver/lung carci- noma Liver/lung carci- noma or adenoma Males Lung adenoma Lung carcinoma Lung carcinoma or adenoma Liver adenoma Liver carcinoma Liver carcinoma or adenoma Lung/liver carci- noma Lung/liver carci- noma or adenoma Control 4.0 2.0 6.0 4.0 2.0 6.0 2.0 10.0 6.0 4.0 10.0 20.0 26.0 44.0 30.0 54.0 2,000 ppm 47.9 27.1 62.5 12.5 22.9 33.3 43.8 75.0 38.0 20.0 54.0 28.6 30.6 49.0 42.9 69.4 4,000 ppm 58.3 60.4 85.4 45.8 66.7 83.3 91.5 97.9 48.0 56.0 80.0 28.6 53.1 67.3 79.6 91.8 Two-stage prediction (%) Control 4.3 2.0 6.0 3.5 1.9 5.6 2.0 10.0 6.5 3.9 10.0 21.1 24.8 43.1 28.3 53.7 2,000 ppm 41.0 27.1 62.5 15.6 24.5 37.0 45.5 75.0 33.0 20.7 54.0 25.8 32.8 50.3 46.6 69.9 4,000 ppm 63.7 60.4 85.4 43.6 65.6 81.3 90.6 97.9 52.0 55.5 80.0 30.2 52.2 66.9 78.0 91.7 Chi square 1.54 <.01 <.01 0.49 0.93 0.43 0.10 <.01 0.90 0.02 <.01 0.30 0.16 0.06 0.42 0.08 ------- In female rats, there was increased mortality in the high-dose group relative to controls. Under these conditions, competing risks may lead to underestimation of the risk attributable to the tumors observed in the rats. The NTP indicated that the decreased rat survival is likely to be due to the frequent occurrence of leukemia in all groups. EPA's proposed guidelines (U.S. EPA, 1984) indicate that weight should be placed on the analysis of risks using the animals that have tumors in any one of the sites found to be statistically elevated. The guidelines also indicate that weight should be placed on the experimental species and sex group showing the highest risks. On these grounds, combined carcinomas and adenomas of the lung and/or liver in female mice is the end point of most weight. Thus, additional analyses were conducted, with emphasis on the data for mice having lung and/or liver tumors. The following sections describe these analyses. 4.4. RISK ANALYSIS CONSIDERING TIME-TO-TUMOR INFORMATION In the NTP study, there was a small number of deaths in mice before the first lung or liver tumors were found. The first lung or liver tumor was a liver tumor in a high-dose male mouse at week 61. Table 24 shows the numbers of male and female mice alive at 61 weeks that were subsequently examined at the lung and liver sites. While the number of animals lost to early mortality was small, because of the very high incidence of tumor responses seen in the high-dose groups, analyses were conducted to determine if the early deaths might have an effect on estimated risks. Table 25 shows the results of applying the two-stage multistage model to the mouse data using the denominators from Table 24. It should be noted that this simplified analysis has the weakness that animals dying before the first tumor was observed may in reality (as would be seen in 81 ------- TABLE 24. MICE SURVIVING TO 61 WEEKS AND RECEIVING EXAMINATION OF THE LUNG AND LIVER Response Males Females Control 50 46 Dose 2,000 ppm 45 46 4,000 ppm 47 46 NOTE: In all, 13 mice were excluded from counting due to early death. The times of these deaths were: 0-3 months, 5 deaths; 3-6 months, 1 death; 6-9 months, 4 deaths; 9-12 months, 1 death; 12 months-61 weeks, 2 deaths. SOURCE: NTP, 1985. TABLE 25. GLOBAL83 MODEL PARAMETERS FOR MOUSE LUNG AND LIVER TUMORS COMBINED- NTP (1985) DICHLOROMETHANE DATA ON ANIMALS SURVIVING TO 61 WEEKS (WHEN FIRST TUMOR OCCURRED) Tumor Two-stage qn Q-iXlO U /I 1 x (ppm-1) model parameters q2xlO'6 (ppm-1 ) Males Lung or liver carcinoma Lung or liver carci- noma or adenoma Females Lung or liver carci- noma Lung or liver carci- noma or adenoma 0.341 0.777 0.0211 0.114 0.0 0.0371 0.0 0.0 0.0854 0.139 0.159 0.363 0.235 0.582 0.245 0.785 NOTE: Values of the q-j apply for the NTP protocol dose schedule. 82 ------- a large population) have experienced some risk of cancer, and that animals dying after the first tumor was observed are treated as having been at full lifetime risk of cancer independent of their time of death. The data in Table 25, in comparison with the data in Table 22, show that the GLOBAL q^ parameter estimates are consistent with the earlier analysis and are not strongly influenced by the adjustment of the denominators. In contrast, for some of the analyses, the MLE qj estimates are unstable: for male mice the qi estimate for combined adenomas or carcinomas of the lung or liver in Table 22 is zero, whereas in Table 25 it is positive. For female mice, for the same combination of tumors and sites, this situation is reversed. The variability of the MLE linear term is consistent with the CAG's experience that this parameter estimate is often unstable in response to small changes in the input data. To provide a comparison with the dichotomous multistage model risk esti- mates developed in the preceding section, a time-to-tumor formulation of the multistage model, the WEIBULL82 time-to-tumor program, also developed by Crump (1982), was applied to the NTP cancer results in mice. The WEIBULL82 program is based on the following equation: Prob [effect] = 1 - exp C-Q (dose) x (time - T0)K] where Q is a fitted polynomial of the same form utilized in GLOBAL83, and T0 and K are fitted parameters for the time-dependence of the tumor response. Risk calculations using this model can be made for two end points: either tumors are assumed to be "incidental" and unrelated to the cause of an animal's death, or the tumors are "fatal" and are assumed to produce death directly. The NTP studies do not attempt to identify the cause of death for animals in 83 ------- cancer bioassays; therefore, a time-to-tumor analysis must hypothesize as to the causes of deaths in study animals dying before final sacrifice. The NTP (1985) bioassay provides statistical analyses showing that in mice both the male and female high-dose groups experienced elevated mortality in the latter part of the 2-year study. A simple comparison demonstrates that the observed tumors may reasonably have produced this mortality. Table 26 shows the study mortality divided into three time periods: deaths before 61 weeks (when the first lung or liver tumor was observed), deaths between 61 and 103 weeks, and deaths at the final sacrifice at 104 weeks. The table also indi- cates the numbers of animals in each death category that were observed to have carcinomas of either the lung or liver; these are the tumor types that can be most strongly expected to contribute to mortality. Table 27 shows that mortality before the occurrence of the first lung or liver tumor was small and comparable in all groups. In the 61- to 103-week period, where elevated treatment-associated mortality is seen, the number of animals dying without carcinomas was relatively stable among the male dose groups. In high-dose females, few animals died without carcinomas. These data are consistent with DCM-induced tumors leading to the increased mortality observed in the NTP mice. In the following analysis, using the WEIBULL82 program, the effect of the alternate assumptions of incidental or fatal tumors will be compared. For completeness, both "fatal" and "incidental" analyses are presented for adenoma response as well as for the other tumor groupings. However, it is noted that a "fatal" tumor analysis may be less appropriate for adenomas. The data from these analyses are given in Table 27. 84 ------- TABLE 26. MORTALITY IN THE NTP MOUSE DICHLOROMETHANE BIOASSAY Dose Males Control 2,000 ppm 4,000 ppm Females Control 2,000 ppm 4,000 ppm Deaths before 61 weeks 0(0)a 4(0) 3(0) 4(0) 4(0) 4(0) Deaths 61-103 weeks 11(5) 22(10) 36(29) 21(0) 21(8) 38(35) Final sacrifice deaths 39(10) 24(11) 11(10) 25(1) 25(13) 8(8) aNumbers in parentheses indicate the number of animals in each group found to have lung or liver carcinomas. SOURCE: NTP, 1985. 85 ------- TABLE 27. COMPARISON OF WEIBULL82 AND GLOBAL83 PREDICTIONS FOR RISK AT 104 WEEKS- LUNG AND LIVER TUMORS IN MICE3 "Fatal" tumor WEIBULL82 analysis Tumor Males Lung adenoma Lung carcinoma Liver adenoma Liver carcinoma Lung or liver carcinoma Lung or liver carcinoma or adenoma Females Lung adenoma Lung carcinoma Liver adenoma Liver carcinoma Lung or liver carcinoma Lung or liver carcinoma or adenoma (ppm-1) 0.125 -- 0.0344 0.0417 0.0573 0.106 0.0 <.001 0.0 0.002 0.0153 0.0769 q*xlO-3 (ppm-1) 0.173 — 0.0787 0.1502 0.199 0.252 0.148 0.0651 0.0440 0.0975 0.137 0.210 „> 5 - 5 5 4 4 2 4 4 4 4 4 "Incidental" tumor WEIBULL82 analysis (ppm-1) 0.232 0.0561 0.0687 0.0369 0.0526 0.178 0.352 0.109 0.0084 0.0526 0.119 0.0 q*xlO-3 (ppm-1) 0.299 0.199 0.131 0.159 0.304 0.644 0.447 0.248 0.105 0.198 0.392 0.822 n 5 5 5 5 4 4 4 4 4 4 4 4 GLOBAL83 Two-stage (ppm-1) 0.166 0.0 0.0306 0.0 0.0 0.0 0.242 0.690 0.0 0.0918 0.0 0.345 q*xlO-3 (ppm-1) 0.223 0.141 0.0810 0.142 0.190 0.429 0.310 0.233 0.0872 0.145 0.244 0.870 aFor the WEIBULL82 model, q1 is defined gs the first-degree polynomial coefficient multiplied by the time function evaluated at week 104. q^ is derived from model risk predictions at low dose. Dn indicates the degree of the dose polynomial used in the WEIBULL82 analysis. Different degrees were used in this exploratory analysis. NOTE: Values of the qi apply for the NTP protocol dose schedule. 86 ------- The following observations can be drawn from these data: 1. The q1 and q^ values for the "incidental" tumor analysis from the WEIBULL83 program were generally higher than the values from the "fatal" tumor analysis. However, the size of this difference was moderate, constituting approximately a factor of two, with a maximum difference of a factor of four for q^. Such differences are expected because the "fatal" tumor analysis excludes tumors found at final sacrifice, and these tumors provide important quantitative contributions to the NTP findings (see Table 26 for examples). While the q\ estimates generally followed this same pattern, the deviations were greater in some cases. 2. The q^ and q^ estimates from the two-stage GLOBAL83 analysis agreed overall with the range of estimates from the "incidental" and "fatal" tumor WEIBULL82 analyses. In all cases, the GLOBAL83 q^ values fell either within or very close to the range of the two WEIBULL qi values. From this exploratory analysis, it can be seen that use of the WEIBULL82 time-to-tumor analysis does not lead to risk estimates strongly different from those derived from GLOBAL83. The WEIBULL82 program has been less widely utilized than the GLOBAL83 program, and requires assumptions as to the cause of animal death. For these reasons, WEIBULL82 is less well-suited than GLOBAL83 for formal use in the present risk assessment. 4.5. COMPARISON OF RISKS ESTIMATED WITH OTHER DOSE-RESPONSE MODELS In order to provide comparison with the two-stage (restricted) multi- stage estimates presented in Table 22, calculations for combined lung and liver tumors were also made with the one-stage (one-hit) and four-stage (dichotomous) formulations of the multistage model. The one-stage or linear 87 ------- model has been one of the most widely used models in carcinogen risk assess- ment, and provides a simple formulation of the hypothesis that many fundamen- tal carcinogenic processes are linear in nature. The four-stage model shares the multistage rationale of the two-stage model, but allows a sharper upward curvature in risk estimates. Table 28 shows parameter estimates for one-stage and four-stage versions of the GLOBAL83 program for tumor response data from Table 19. The one-stage model fits the experimental data acceptably for the combined carcinomas and adenomas in both the male and female groups. For the females, the one-stage model response estimates are within 4% of the observed response for the three experimental doses; for males, the estimates are within 6% of the observed response. For carcinomas alone, the one-hit model does not provide a good fit to the data in either sex (p < 0.05 males, p < 0.01 females, by the chi square test). As can be seen from Tables 22 and 23, the two-stage model provides an acceptable fit to the data for all four tumor end points. The four-stage model provides an exact fit for all four data sets. These analyses demonstrate that for combined adenomas and carcinomas of the lung and/or liver in both male and female mice, the data are compatible with a linear dose-response, and that the 95% UCL linear term estimates from the three models are consistent within 20% in males and within 10% in females. For the carcinoma response, where the one-stage model did not fit well, the four-stage model yielded positive MLE linear term estimates for both sexes, while the two-stage model did not. The UCL four-stage risk estimates were approximately 50% and 20% higher than the UCL two-stage estimates for com- bined lung and liver carcinomas in the males and females, respectively. To provide comparisons with multistage risk model estimates, two models with different theoretical formulations, the Weibull (dichotomous) and probit models, were used to analyze the combined lung and liver tumor data in both the 88 ------- TABLE 28. ONE-STAGE AND MULTISTAGE MODEL PARAMETERS FOR TUMORS IN MICE3 oo Tumor Male mice Lung or liver carci noma Lung or liver carcinoma or adenoma Female mice Lung or liver carcinoma Lung or liver carci noma or adenoma One-stage Two-stage Four-stage Chi square Chi square 0.238 0.329 4.44 0.348 0.505 1.94 0.418 0.522 6.77 0.736 0.936 1.25 0.0 0.0 0.0 0.190 0.42 0.424 <0.01 0.244 <0.01 0.345 0.870 <0.01 Chi square 0.058 0.223 <0.01 0.125 0.429 <0.01 0.120 0.368 <0.01 0.472 0.870 <0.01 aUnits: ------- male and female mice. These models were fitted to the data using the RISK81 computer program developed by Kovar and Krewski (1981). The RISK81 program provides two formulations of both models, one which is based on the assumption that the observed tumor incidence is independent of the background tumor rates observed in the controls, and a second formulation which assumes that the carcinogen contributes a dose that is additive to the background effects seen in the controls. In the additive case, the probit and Weibull models produce risk estimates that are linear at low doses. Tables 29 through 32 provide the results from these models in comparison to the two-stage GLOBAL83 results. The tables show the DCM doses that are estimated to produce four given levels of risk under the different models, as well as lower confidence limits (LCLs) on the dose that produces the specified effect. In all cases, the doses estimated to produce a given risk by the back- ground-independent probit model are markedly higher than those predicted by the multistage model (four orders of magnitude difference in the female mice com- bined tumor group). While the independent Weibull MLE estimates of dose are substantially below those obtained with the independent probit model, they are substantially higher than those obtained with the two-stage multistage model in all four analyses. The independent Wei bull MLE estimates are broadly com- parable to the multistage MLE estimates for cases in which the MLE multistage linear term is zero. The LCL dose estimates of the independent Weibull model are notably lower than the MLE estimates. The additive background formulation of both the Weibull and probit models converged to provide acceptable parameter estimates only for the female mouse data sets. In these two cases, both models produced estimates of dose for a fixed risk that were linear at low doses and were markedly higher than the 90 ------- TABLE 29. MALE MICE CARCINOMAS OF THE LUNG OR LIVER: DICHLOROMETHANE DOSES ESTIMATED TO PRODUCE GIVEN RISK LEVELS (DOSES IN PPM UNDER NTP PROTOCOL SCHEDULE) GLOBAL83 two-stage Risk level 10-2 10-4 10-6 10-8 MLE 369 36.8 3.68 0.368 LCL 52.4 0.524 5.24 x 10-3 5.24 x 10-5 Independent probit MLE 1,080 543 329 217 LCL 531 185 84.7 44.4 Independent Weibull MLE 723 123 20.9 3.57 LCL 217 11.0 0.556 0.0281 NOTES: Calculations are based on excess risk over background. Probit and Weibull model estimates calculated using RISK81 computer program. LCL estimates are the 95% confidence lower bound on dose (variance based on log dose for RISK81). The additive probit and additive Weibull models failed to converge for this data set. ------- TABLE 30. MALE MICE CARCINOMAS OR ADENOMAS OF THE LUNG OR LIVER: DICHLOROMETHANE DOSES ESTIMATED TO PRODUCE GIVEN RISK LEVELS (DOSES IN PPM UNDER NTP PROTOCOL SCHEDULE) ro 6LOBAL83 two-stage Risk level 10-2 10-4 10-6 10-8 MLE 306 30.6 3.06 0.306 LCL 23.4 0.233 2.33 x 10-3 2.33 x 10-5 Independent probit MLE 885 410 236 150 LCL 353 101 41.0 19.5 Independent Wei bull MLE 493 54.0 5.93 0.652 LCL 100 2.09 0.0439 9.16 x 10-4 NOTES: Calculations are based on excess risk over background. Probit and Weibull model estimates calculated using RISK81 computer program. LCL estimates are the 95% lower confidence limits on dose (variance based on log dose for RISK81). The additive probit and additive Weibull models failed to converge for this data set. ------- TABLE 31. FEMALE MICE CARCINOMAS OF THE LUNG OR LIVER: OICHLOROMETHANE DOSES ESTIMATED TO PRODUCE GIVEN RISK LEVELS (DOSES IN PPM UNDER NTP PROTOCOL SCHEDULE) GLOBAL83 two-stage Risk level 10-2 10-4 S 10-6 10-8 MLE 262 26.1 2.61 0.261 LCL 40.7 0.410 4.10x10-3 4.10x10-5 Independent probit MLE 769 412 260 176 LCL 503 218 117 70.1 Additive probit MLE 163 1.92 0.0192 1.92x10-4 LCL 81.4 0.78 0.0078 7.8x10-5 Independent Wei bull MLE 309 35.8 4.15 0.482 LCL 142 8.03 0.452 0.0255 Additi Weibul MLE 134 1.52 0.0152 1.52x10-4 ve 1 LCL 68.8 0.686 0.006 6.86x10-5 NOTES: Calculations are based on excess risk over background. Probit and Wei bull model estimates calculated using RISK81 computer program. LCL estimates are the 95% lower bound on the dose (variance based on log dose for RISK81). ------- TABLE 32. FEMALE MICE CARCINOMAS OR ADENOMAS OF THE LUNG OR LIVER: DICHLOROMETHANE DOSES ESTIMATED TO PRODUCE GIVEN RISK LEVELS (DOSES IN PPM UNDER NTP PROTOCOL SCHEDULE) Risk level 10-2 io-4 10-6 10-8 GLOBAL83 two-stage MLE 28.8 0.290 2.90x10-3 2.90x10-5 LCL 11.6 0.115 1.15x10-3 1.15x10-5 Independent probit MLE 478 238 141 92.5 LCL 198 67.4 30.2 15.6 Additive probit MLE 45.1 0.464 4.64x10-3 4.64x10-5 LCL 24.6 0.249 2.49x10-3 2.49x10-5 Independent Wei bull MLE 93.4 4.75 0.242 0.0124 LCL 17.2 0.196 2.23x10-3 2.53x10-5 Additive Wei bull MLE 36.3 0.369 3.69x10-3 3.69x10-5 LCL 15.7 0.156 1.56x10-3 1.56x10-5 NOTES: Calculations are based on excess risk over background. Probit and Weibull model estimates calculated using RISK81 computer program. LCL estimates are the 95% lower bound on dose (variance based on log dose for RISK81). ------- corresponding independent background models. The MLE and LCI dose estimates were quite comparable between the two models (maximum differences under 50% at low doses). For female mouse carcinomas, the MLE estimates from these models lead to dose estimates that are markedly lower than the nonlinear MLE multi- stage estimates. For the combined carcinomas and adenomas, the MLE estimates of additive Weibull, additive probit, and multistage models agreed within approximately 50%. For both groupings of tumors in female mice, the LCL dose estimates of the additive probit, additive Weibull, and multistage models were within a factor of two of each other, with the two-stage multistage model leading to the lower estimates of dose for a fixed risk. 4.6. COMPARISON OF THE NTP (1985) RESULTS WITH OTHER BIOASSAYS Several recent long-term animal studies of DCM have been performed in addition to the NTP (1985) bioassay. These studies are reviewed in detail in the Health Assessment Document for Dichloromethane (U.S. EPA, 1985). While there were significant limitations in these studies, which often included maximum doses well below what the animals could tolerate, some positive and suggestively positive carcinogenic responses were obtained. This section develops GLOBAL83 parameter estimates (for a multistage model with the maximum polynomial degree equal to the number of dose groups minus one) for the responses found in rats and mice using data taken from the Health Assessment Document (1985). 1. The Dow Chemical Company (1980) 2-year inhalation study in Sprague- Dawley rats found some statistically positive elevations in mammary tumors in this strain, which has a normally high spontaneous mammary tumor rate, making positive results more difficult to obtain statistically. The strongest find- ing was the elevation of the total number of mammary tumors seen in female rats (165/99 in controls vs. 287/97 in the high-dose groups). However, the GLOBAL83 95 ------- model cannot accommodate data of this form. The percentages of mammary tumors in the different female groups was not statistically elevated, but can be used in GLOBAL83 to calculate an upper-bound risk, q^, on the hypothesis of a real biological effect. The male rat data, which indicated an elevated mammary tumor response at the high-dose point (not significant), are also used in this manner (Table 33). For comparison, the combined mammary tumor data from the NTP (1985) study in F344/N rats (using the same dosing schedule) yielded values of q* = 0.0311 Q * o x 10~J for males and q-^ - 0.164 x 10~° for females. Thus, while the mammary tumor incidences were not statistically elevated in the Dow (1980) study, the upper bounds on q^ that can be derived from the Dow results are in close agreement with the NTP results. At a second site, sarcomas in or around the salivary gland in male rats, the Dow (1980) study found statistically positive results (Table 34). The NTP study did not find an elevated sarcoma incidence in or around the salivary gland, 2. The Dow Chemical Company (1982) inhalation study found evidence of an elevated mammary tumor incidence in female rats. (The tumor percentage was sta- tistically elevated at 200 ppm compared with controls.) The total number of mammary tumors found also showed an increase with dose, as in the Dow (1980) study. GLOBAL83 is applied to the incidence data to determine the maximum linear dose-response component that is compatible with these data (Table 35). The estimate of q^ derived from this study is high compared with that for mammary tumor incidence in female F344/N rats in the NTP (1985) study (qj = 0.164 x 10'3 ppm'1). 3. The National Coffee Association (NCA, 1982a, b) study found evidence of an increased occurrence of liver tumors in female F344 rats given DCM in drinking water. The GLOBAL83 value for q^ is shown in Table 36, calculated 96 ------- TABLE 33. MAMMARY TUMORS IN RATS Dose3 Response Females Males Control 79/96 7/95 500 ppm 81/96 3/95 1,500 ppm 80/95 7/95 3,500 ppm 83/97 14/95 GLOBAL83 qj. ppm"1 0.201 x ID'3 0.040 x ID'3 aDose administered 6 hours/day, 5 days/week, for 2 years. NOTE: Value for q-^ based on administered doses and dose time schedule. SOURCE: Dow Chemical Company, 1980. TABLE 34. SALIVARY GLAND TUMORS IN MALE RATS Dose3 Control 500 ppm 1,500 ppm 3,500 ppm . GLOBAL83 qj (x 10'3 ppm'1) 1/93 0/94 5/91 11/88 0.043 3Dose administered 6 hours/day, 5 days/week, for 2 years. NOTES: Value for q-^ based on administered dose and dose time schedule. These data show a statistically significant (p < 0.001) test for a linear trend, and the high-dose group response is elevated compared to the controls (p = 0.002). SOURCE: Dow Chemical Company, 1980. 97 ------- TABLE 35. MAMMARY TUMORS IN FEMALE RATS Dosea Control 50 ppm 200 ppm 500 ppm 52/70 58/70 61/70 55/70 1.22 x 10-3 aDose administered 6 hours/day, 5 days/week, for 2 years, NOTE: Value for q^ based on administered dose and dose time schedule. SOURCE: Dow Chemical Company, 1982. TABLE 36. NEOPLASTIC NODULES OR HEPATOCELLULAR CARCINOMAS IN FEMALE F344 RATS Control Dose (mg/kg/day)a 50 125 250 experimental dose units * 11 NTP dose units 0/134 1/85 4/83 1/85 6/85 0.470 x ID"3 0.22 x ID'3 aDose delivered in drinking water. NOTE: The tumor responses in the 50 and 250 mg/kg/day groups were elevated in comparison with controls at the p < 0.05 level. SOURCE: NCA, 1982a, b. 98 ------- on the basis of the experimentally applied dose units and also calculated using the same dose units as in the NTP (1985) study. (This second estimate was derived using the ppm to mg/kg/day conversion factors presented in sec- tion 4.7. below.) The NTP (1985) study noted that a positive, but marginal, increase in female rat hepatocellular neoplastic nodules or hepatocellular ie *3 1 carcinomas was observed in that study (q^ approximately 0.03 x 10~° ppm ). 4. The NCA (1983) study in mice found evidence of an increased incidence of liver tumors in males. As with rats, a value of q^ using the dose units of the NTP (1985) study was derived and is shown in Table 37. For comparison, the NTP study, which found elevated rates for the same tumors, yields a q^ = 0.195 x ID'3 estimate. In conclusion, the studies discussed in this section provide some evi- dence for DCM-induced tumors at sites where the NTP (1985) study found tumors; in addition, the Dow (1980) study showed an increase in salivary gland region tumors in male rats. Estimates of q^ have been derived for these sites to indicate the maximum linear component of a tumor dose-response that is consis- tent with study findings. These upper-bound risk estimates are comparable to, or in some cases, larger than, corresponding estimates derived from the NTP (1985) study for the same tumor sites. 4.7. DERIVATION OF HUMAN UNIT RISK ESTIMATES FOR INHALATION OF DCM The Health Assessment Document for Dichloromethane (U.S. EPA, 1985) devel- oped estimates of unit risks to humans on the basis of the Dow (1980) findings of salivary gland tumors in rats. The same standard CAG assumptions that were used in that document to convert between animal and human doses are also applied here. It is noted that in this assessment, the extrapolation between high and low doses in humans and animals is done in a different order than in the earlier CAG assessment; however, this does not affect the results. Table 38 sumarizes 99 ------- TABLE 37. HEPATOCELLULAR ADENOMAS OR CARCINOMAS IN MALE MICE Dose (mg/kg/day)a Control 60 125 185 250 experimental dose units NTP dose units 24/125 51/200 30/100 31/99 35/125 0.995 x 10'3 0.78 x 10'3 aDose administered in drinking water. NOTE: The response at the 125 and 185 mg/kg/day doses were elevated in com- parison with controls (p < 0.05). SOURCE: NCA, 1983. TABLE 38. VALUES OF q, FOR NTP (1985) BIOASSAY USED TO DERIVE HUMAN q-[ ESTIMATES Male rat mammary or subcutaneous tumors Female rat mammary tumors Male mouse lung or liver adenoma or carcinoma Female mouse lung or liver adenoma or carcinoma = 0.0540 x 10 (ppm experimental protocol) = 0.164 x 10~3 (ppm"1 experimental protocol) = 0.429 x 10~3 (ppm"1 experimental protocol) = 0.870 x 10~3 (ppm"1 experimental protocol) 100 ------- the values of qi derived for tumors found in both sexes of rats and mice in the NTP study (data taken from Tables 20 and 21). 1. These values are first converted to equivalent rodent values for con- tinuous exposure to DCM in units of mg/kg/day. To make this conversion, the inhalation rates for the rodents must first be estimated. This is done using the following formulas (U.S. EPA, 1985): For mice: I = 0.0345 (wt/0.025)2/3 m3/day For rats: I = 0.105 (wt/0.113)2/3 m3/day Inhalation rates are calculated using the NTP (1985) average weights for male and female mice and rats at the midpoint of the bioassay (51-week data point; the NTP report does not give the average animal weight over the whole study period). The data are given in Table 39. Dose conversion factors can now be calculated between the NTP schedule and a continuous mg/kg/day exposure. For example, in male rats: 1 ppm NTP schedule = 10~6 x 3,478 g DCM/m3 x 1,000 mg/g x average exposure 4.29 hours/day x 1 day/24 hours x 0.268 m3/day/0.462 kg = 0.361 mg/kg/day These data are given in Table 40. 2. Equivalent human DCM doses and q^ values are now calculated using the CAG methodology for well-absorbed vapors (DCM in air is likely to be ab- sorbed by the lungs to a high degree at low doses in both humans and rodents; the interspecies conversion is being applied for the risks estimated at low doses). The CAG assumes (U.S. EPA, 1985) that humans and animals exposed to 101 ------- TABLE 39. ESTIMATED INHALATION RATES FOR NTP (1985) TEST ANIMALS Weight at Estimated bioassay midpoint inhalation rate (kg) (m3/day) Male rat Female rat Male mouse Female mouse 0.462 0.278 0.037 0.032 0.268 0.191 0.0448 0.0407 TABLE 40. DOSE CONVERSION FACTORS AND EQUIVALENT q? VALUES FOR NTP (1985) STUDY 1 mg/kg/day equivalent (mg/kg/day)-1 of 1 ppm exposure, for continuous NTP protocol exposure3 Male Femal Male Femal rat e rat mouse e mouse 0 0 0 0 .361 .467 .753 .791 0 0 0 1 .149 x .383 x .570 x .10 x 10-3 io-3 ID'3 ID'3 aThese values of q^, which apply to the same^tumor types as are listed in Table 32, are obtained by multiplying the q-^ values in Table 32 by the reciprocals of the values in the first column of this table. 102 ------- equal doses of a carcinogen on a (mg/kg)2/3 basis over equivalent proportions of a lifetime will encounter the same degree of cancer risk. This implies that a rodent with weight WR, exposed to a dose of D mg/kg/day and a human exposed to a dose of D($R) mg/kg/day encounter the same lifetime cancer risks. Table 41 contains human dose equivalents and values for q-. TABLE 41. DOSE CONVERSION FACTORS AND EQUIVALENT qt VALUES FOR RODENTS IN NTP (1985) STUDY AND FOR HUMANS Human mg/kg/day ^ equivalent for rodent q^a 1 mg/kg/day (mg/kg/day)-! exposure human Male rat Female rat Male mouse Female mouse 0.188 0.158 0.0809 0.0770 0.793 2.43 x 7.05 x 14.3 x x ID'3 ID'3 ID'3 10-3 aThese values of qj, which apply to the same^tumor sites as those given in Table 32, are obtained by multiplying the qj values in Table 34 by the reciprocals of the values in the first column of this table. 3. To obtain an estimate of the unit risk for a human inhaling 1 pg/m3 of DCM over a lifetime, the standard CAG assumption of a human inhalation rate of 20 m3/day (U.S. EPA, 1985) is applied. A continuous exposure to 1 yg/m3 of DCM is equal to an exposure of 1 yg/m3 x ID'3 mg/yg x 2o m3/day x 1/70 kg = 2.86 x 1(H mg/kg/day 103 ------- Using the value of q^ in female mice from Table 35, an upper-limit incre- mental lifetime cancer risk estimate of 4.1 x 10~6 is estimated for this exposure. Alternatively expressed, q^ = 4.1 x 10"6 (ug/m3)"1. Using the relation that 1 yg/m3 DCM is equivalent to 2.88 x 10~4 ppm, * O1 q^ = 1.4 x 10 ppm" (continuous exposure). 4. The CAG's potency index is derived by multiplying q^ (mg/kg/day)~^ by the molecular weight of the compound (84.9 g/mol for DCM) to obtain qi (mmol/kg/ day) . For lung and liver tumors, combined, in female mice, qi (mmol/kg/ day)'1 = 1.1. This value is in the fourth quartile of the CA6 histogram for the potency index distribution. 4.8. HUMAN UNIT RISK ESTIMATE FOR INGESTION OF DCM Data from both the NTP inhalation bioassay and from the earlier NCA studies of DCM in drinking water will be considered in connection with a unit risk estimate for DCM ingestion. All of the existing studies have limitations for estimating the risks from ingestion of DCM. DCM is rapidly absorbed and sys- temically distributed following either inhalation or ingestion exposure; thus the use of an inhalation study is one approach for assessing hazards from inges- tion exposure. Nonetheless, exposures via the two routes are likely to lead to differing doses reaching individual organs; in particular, an inhalation study may result in a higher degree of exposure to lung tissue and a lesser exposure to tissues in the digestive system than an ingestion exposure. For this reason the NTP findings for liver tumors, but not lung tumors, in mice are used for quantitative estimation of risk from DCM ingestion. On the other hand, it should be noted that an analysis based on the NTP study may underestimate risks of liver tumors or other digestive system tumors. The NCA (1983) drinking water study in mice yielded suggestive but not conclusive evidence of a treatment-associated increase in hepatocellular carci- 104 ------- nomas and/or adenomas in males (U.S. EPA, 1985). This finding can be directly used to make an upper-bound risk estimate from ingestion exposure to DCM. However, the NCA study utilized doses that were well below the maximum tolera- ted dose (MTD) (U.S. EPA, 1985). Thus, it is possible that elevated tumor incidences would have been found in other tissues if the study had been conduc- ted nearer to the MTD. The unit risk from ingestion exposure is obtained using the q^ esti- mate for liver tumors in female mice (the sex with the higher risk estimate): q^ = 0.160 (ppm~*). The female mouse data are used for this calculation because the stronger qualitative and quantitative data for carcinogenicity in the liver were obtained for this sex in the NTP (1985) study. If the NTP findings for male mice were used instead, the unit risk estimate given below would be 20% higher. Using the appropriate dose-conversion procedures, this ie *3 1 value leads to an equivalent human estimate of q^ = 2.6 x 10~° (mg/kg/day) . If a 70-kg human drinks 2 L/day of water containing 1 yg/L DCM, the average daily exposure is 1 ug/L x ID'3 mg/yg x 2 L/day x 1/70 kg = 2.86 x 10'5 mg/kg/day Using the above value for q^, the lifetime incremental cancer risk is estimated at 7.5 x 10-8 (yg/L)-l. Using the NCA (1983) study, the value for q^ in male mice is 0.995 x 10~3 (mg/kg/day)-1, based on hepatocellular carcinomas and adenomas. The male mouse data from the NCA study are used for this calculation because only the male mice showed evidence for carcinogenicity in the liver in this study, which was conducted well below the MTD. Following the dose-conversion procedure given earlier in this section, the equivalent human value is 1.23 x 10~2 105 ------- (mg/kg/day)-1. The corresponding UCL incremental unit risk estimate for drink- ing water is 3.5 x 10~7 (yg/L)'1. The unit risks calculated on the basis of the two mouse studies are com- parable; therefore, the mean value of 7.5 x 10~3 (mg/kg/day)"1 or equivalently 2.1 x 10'7 (pg/L)-1 is used for the unit risk estimate for drinking water exposure to DCM. 4.9. COMPARISON OF ANIMAL AND HUMAN DATA RELEVANT TO CANCER RISK The risk prediction for human exposure to DCM can be compared with the observed cancer mortality in a cohort of Eastman Kodak employees exposed to DCM (Friedlander et al., 1978; Hearne and Friedlander, 1981; Friedlander et al., 1985).* The study of Kodak employees does not provide evidence for elevated rates of total cancer, other than possibly cancer of the pancreas, in the male DCM-exposed workers. The recent update of the study (Friedlander et al., 1985) noted a possible increase in tumors of the pancreas. The importance of this recent finding has not yet been evaluated by the CAG. An analysis of the power of this study to detect an increase in cancer mortality was included in the Health Assessment Document for Dichloromethane (1985) with reference to the finding of salivary tumors in rats; a parallel calculation will be shown here. The continuous lifetime equivalent DCM exposure of the Kodak employees was estimated to be between 1.88 and 7.52 ppm, based on a 20-year exposure for the 252 long-term workers. Based on the 95% upper-limit slope factor for the preceding section (q = 1.4 x 10~2 ppm" for continuous human exposure), *0tt et al. (1983b) also reported an epidemiologic study of DCM workers; how- ever, the limited follow-up in this study prevents any adequate comparison with estimated lifetime cancer risks. In the Ott study, the overall mortality was less than 10% for all subgroups during the study period. A more detailed discussion of this study can be found in the Health Assessment for Dichloro- methane (U.S. EPA, 1985). 106 ------- the upper bound on the lifetime cancer risk encountered by these workers is estimated to be between 0.026 and 0.105. For the 252 workers, this would translate to a 95% upper limit of 6.6 to 26.5 excess cancer deaths over a lifetime. However, because the study follow-up period was 21 years and most workers were not observed until death, it is probable that only a fraction of the estimated excess cancer deaths were seen. In the absence of a more rigor- ous method, it is estimated that the fraction of cancer cases that would be observed in the follow-up period is approximated by the overall mortality expected in the follow-up: 107.9/252 deaths, or 43%. (The expected values are derived from the Kodak control cohort.) Thus, a 95% upper limit of between 2.8 and 11.3 cancer deaths due to DCM exposure would have been predicted for the cohort. Using the statistical methods presented by Beaumont and Breslow (1981), the power of the Friedlander et al. study, with 27.0 expected cancer deaths, to detect an excess of 2.8 deaths from total cancer (with 95% confidence, two-tailed test) is 0.07; the power to detect 11.3 cancer deaths is 0.51. Tumors of several types were found to be elevated in the NTP mouse and rat bioassays. Additionally, it cannot be generally expected that humans and experimental animals will show a carcinogenic response at the same sites when exposed to a chemical that is carcinogenic to both. These factors prevent rigorous comparison of the DCM cancer risk estimated from the NTP study with findings of an epidemiologic study for particular cancer sites. However, as a tentative example of power comparisons that may be made, the ability of the Friedlander et al. study to detect excess lung cancer deaths is calculated. For this example, an estimate of risk for lung cancer is obtained by ap- plying the observed excess of lung tumors in female mice (the sex with the stronger response) to estimate lung cancer deaths in the Friedlander et al. cohort. Taking the q-^ value for female mouse lung carcinoma or adenoma 107 ------- from Table 22 (0.579 x 10"3) and applying the procedure outlined in Section 4.7., the upper-bound unit risk estimate for humans is 9.5 x 10~3 (ppm~l) for continuous exposure. Using this risk value with the same procedure followed above for total cancers leads to the estimate of an upper bound of from 1.9 to 7.7 lung cancers due to DCM exposure; the power of the Friedlander et al. study, in which 8.1 lung cancer deaths were expected, to detect such risk is 0.09 and 0.62, respectively. The preceding calculations show that the Friedlander et al. study does not have the power to rule out an overall cancer risk, or in the example presented, a lung cancer risk, that is predicted using the upper-bound slope derived from the NTP study. 108 ------- REFERENCES Ahmed, A.E.; Anders, M.W. (1976) Metabolism of dihalomethanes to formalde- hyde and inorganic chloride. Drug. Metab. Dispos. 4:356-361. Ahmed, A.E.; Anders, M.W. (1978) Metabolism of dihalomethanes to formalde- hyde and inorganic halide. II. Studies on the mechanism of the reaction. Biochem. Pharmacol. 27:2021-2025. Ahmed, A.E.; Kubic, V.L.; Stevens, J.L.; Anders, M.W. (1980) Halogenated methanes: metabolism and toxicity. Fed. Proc. 39(13):3150-3155. Anders, M.W.; Kubic, V.L.; Ahmed, A.E. (1977) Metabolism of halogenated methanes and macromolecular binding. J. Environ. Pathol. Toxicol. 1:117-121. Andersen, M.E.; Thomas, O.E.; Gargas, M.L.; Jones, R.A.; Jenkins, L.J. (1980) The significance of multiple detoxification pathways for reactive metabolites in the toxicity of 1,1-dichloroethylene. Toxicol. Appl. Pharmacol. 52:422-432. Angelo, M.J.; Bischoff, K.B.; Pritchard, A.B.; Presser, M.A. (1984) A physio- logical model for the pharmacokinetics of methylene chloride in B6C3F1 mice following i.v. administrations. J. Pharmacokinet. Biopharm. 12:413-436. Angelo, M.J. (1985) Personal communication to H.L. Spitzer, U.S. Environ- mental Protection Agency. Beaumont, J.J.; Breslow, N.E. (1981) Power considerations in epidemiologic studies of vinyl chloride workers. Am. J. Epidemiol. 114:725-734. Burek, J.D.; Nitschke, K.D.; Bell, T.J.; Wackerly, D.L.; Childs, R.C.; Beyer, J.E.; Dittenber, D.A.; Rampy, L.W.; McKenna, M.J. (1984) Methylene chloride: a two-year inhalation toxicity and oncogenicity study in rats and hamsters. Fund. Appl. Toxicol. 4:30-47. Crump, K.S. (1982) WEIBULL82. K.S. Crump and Co., Inc. Ruston, LA. Unpublished. DiVincenzo, G.D.; Hamilton, M.L. (1975) Fate and disposition of ^C-methy- lene chloride in the rat. Toxicol. Appl. Pharmacol. 32:385-393. DiVincenzo, G.D.; Kaplan, C.J. (1981a) Uptake, metabolism, and elimination of methylene chloride vapor by humans. Toxicol. Appl. Pharmacol. 59:130-140. DiVincenzo, G.D.; Kaplan, C.J. (1981b) Effect of exercise or smoking on the uptake, metabolism, and excretion of methylene chloride vapor. Toxicol. Appl. Pharmacol. 59:141-148. 109 ------- Dow Chemical Company. (1980) Methlene chloride: a two-year inhalation toxicity and oncogenicity study in rats and hamsters. FYI-OTS-0281-0097. Follow-up response A. U.S. Environmental Protection Agency, Office of Toxic Substances. Dow Chemical Company. (1982) Methylene chloride: a two-year inhalation toxicity and oncogenicity study in rats. Toxicology Research Labora- tory, Health and Environmental Sciences, Dow Chemical Company, Midland, MI. Estabrook, R.W.; Franklin, M.; Baron, A.; Shigematsu, A.; Hildebrandt, A. (1971) Drugs and cell regulation (E. Minich, ed.). New York: Academic Press, pp. 227-254. Fodor, G.; Prajsnar, D.; Schlipkoter, H.W. (1973) Endogenous CO-Bildung durch inkorporierte Halogankohlenwasser-Staffe der Methanreihe: Staub-Reinhalt. Luft 33:258-259. Friedlander, B.R.; Hearne, F.T.; Hall, S. (1978) Epidemiologic investigation of employees chronically exposed to methylene chloride. J. Occup. Med. 20:657-666. Friedlander, B.R.; Pifer, J.W.; Hearne, F.T. (1985) 1964 methylene chloride cohort mortality study, update through 1984. Submitted to the U.S. Environmental Protection Agency (Docket No. OPTS-48503) by Eastman Kodak Company. July 1985. Gargas, M.L.; Clewell, H.J.; Andersen, M.E. (1985) Metabolism of inhaled dihalomethanes in vivo: differentation of kinetic constants for two independent pathways. Toxicol. Appl. Pharmacol. (in press) Hearne, F.T.; Friedlander, B.R. (1981) Follow-up of methylene chloride study. J. Occup. Med. 23:660. Heppel, L.A.; Neal, P.A.; Perrin, T.L.; Orr, M.L.; Porterfield, V.T. (1944) Toxicology of dichloromethane (methylene chloride). I. Studies on effects of daily inhalation. J. Ind. Hyg. Toxicol. 26:8-16. Heppel, L.A.; Porterfield, V.T. (1948) Enzymatic dehalogenation of certain brominated and chlorinated compounds. J. Biol. Chem. 176:763-769. Hogan, G.K.; Smith, R.G.; Cornish, H.H. (1976) Studies on the microsomal conversion of CH2C12 to CO. Toxicol. Appl. Pharmacol. 37(1)-.112. Hogeboom, G.H.; Schneider, W.C.; Striebich, M.J. (1953) Localization and integration of cellular function. Cancer Res. 13:617-632. Howe, R.B. (1983) GLOBAL83: an experimental program developed for the U.S. Environmental Protection Agency as an update to GLOBAL82: a computer program to extrapolate quantal animal toxicity data to low doses (May, 1982). K.S. Crump and Co., Inc., Ruston, LA. Unpublished. 110 ------- Kirshman, J. (1984) Food solvents workshop I: methylene chloride. Proceedings of the workshop sponsored by the Nutrition Foundation, Inc., Washington, D.C. March 8-9, 1984, Bethesda, MD, p. 41. Kovar, J.; Krewski , D. (1981) RISK81: a computer program for low-dose extrapolation of quantal response toxicity data. Health and Welfare, Canada. Kubic, V.L.; Anders, M.W. (1975) Metabolism of dihalomethanes to carbon monoxide. II. In vitro studies. Drug Metab. Dispos. 3:104-112. Kubic, V.L.; Anders, M.W. (1978) Metabolism in dihalomethanes to carbon monoxide. III. Studies on the mechanism of the reaction. Biochem. Phannacol. 27:2349-2355. Kurppa, K.; Kivisto, H.; Vainio, H. (1981) Dichloromethane and carbon monoxide inhalation: carboxyhemoglobin addition, and drug metabolizing enzymes in rat. Int. Arch. Occup. Environ. Health 49:83-87. Lorenz, J.; Glatt, H.R.; Fleischman, R.; Ferling, R.; Oesch, F. (1984) Drug metabolism in man and its relationship to that in three rodent spe- cies: monooxygenase, epoxide hydrolase, and glutathione s-transferase activities in subcellular fractions of lung and liver. Biochem. Med. 32:43-56. MacEwen, J.D.; Vernot, E.H.; Haun, C.C. (1972) Continuous animal exposure to dichloromethane. AMRL-TR-72-28, Systems Corporation Report No. W-71005. Wright-Patterson Air Force Base, Ohio, Aerospace Medical Research. Maltoni, C. (1984) Food solvents workshop I: methylene chloride. Proceedings of the workshop sponsored by the Nutrition Foundation, Inc., Washington, D.C. March 8-9, 1984, Bethesda, MD. McKenna, M.J.; Zempel, J.A. (1981) The dose-dependent metabolism of methylene chloride following oral administration to rats. Food Cosmet. Toxicol. 19:73-78. McKenna, M.J.; Zempel, J.A.; Braun, W.H. (1982) The pharmacokinetics of inhaled methylene chloride in rats. Toxicol. Appl. Pharmacol . 65:1-10. McKenna, J.J.; Saunders, J.H.; Boeckler, W.R.; Karbowski, R.J.; Nitschke, K.D.; Chenoweth, M.B. (1980) The pharmacokinetics of inhaled methylene chloride in human volunteers. Paper #176, 19th Annual Meeting, Society of Toxicology, Washington, D.C., March 3-13. National Coffee Association. (1982a, Aug. 11) Twenty-four month chronic toxicity and oncogenicity study of methylene chloride in rats. Final report. Prepared by Hazleton Laboratories America, Inc., Vienna, VA. Unpubl ished. Ill ------- National Coffee Association. (1982b, Nov. 5) Twenty-four month chronic toxicity and oncogenicity study of methylene chloride in rats. Addition to the final report. Prepared by Hazleton Laboratories America, Inc., Vienna, VA. Unpublished. National Coffee Association. (1983, Nov. 30) Twenty-four month oncogenicity study of methylene chloride in mice. Prepared by Hazleton Laboratories America, Inc., Vienna, VA. Unpublished. National Toxicology Program (NTP). (1982) Draft technical report on the carcinogenesis bioassay of dichloromethane (methylene chloride), gavage study. Research Triangle Park, NC and Bethesda, MD. Unpublished. National Toxicology Program (NTP). (1985, Feb.) NTP technical report on the toxicology and carcinogenesis studies of dichloromethane in F344/N rats and B6C3F1 mice (inhalation studies). NTP TR 306. Board draft. Nitschke, K.; Burek, J.; Bell, T., Rampy, L., McKenna, M. (1982) Methylene chloride: a two-year inhalation toxicity and oncogenicity study. Final report of studies conducted at the Toxicology Research Laboratory, Health and Environmental Sciences, Dow Chemical, Midland, MI, USA. Cosponsored by Celanese Corporation, Dow Chemical, USA, Imperical Chemical Industry Ltd., Stauffer Chemical Company, and Vulcan Material Company. Ott, M.G.; Skory, L.K.; Holder, B.B.; Bronson, J.M.; Williams, P.R. (1983a) Health evaluation of employees occupationally exposed to methylene chloride. General study design and environmental considerations. Scand. J. Health 9(Suppl. l):l-7. Ott, M.G.; Skory, L.K.; Holder, B.B.; Bronson, J.M.; Williams, P.R. (1983b) Health evaluation of employees occupationally exposed to methylene chloride. Mortality. Scand. J. Work Environ. Health 9:8-16. Ott, M.G.; Skory, L.K.; Holder, B.B.; Bronson, J.M.; Williams, P.R. (1983c) Health evaluation of employees occupationally exposed to methylene chloride. Clinical laboratory evaluation. Scand. J. Work Environ. Health 9:17-25. Ott, M.G.; Skory, L.K.; Holder, B.B.; Bronson, J.M.; Williams, P.R. (1983d) Health evaluation of employees occupationally exposed to methylene chloride. Twenty-four hour electrocardiographic monitoring. Scand. J. Work Environ. Health 9:26-30. Ott, M.G.; Skory, L.K.; Holder, B.B.; Bronson, J.M.; Williams, P.R. (1983e) Health evaluation of employees occupationally exposed to methylene chloride. Metabolism data and oxygen half-saturation pressures. Scand. J. Work Environ. Health 9:31-38. Pabst, M.J.; Habig, W.H.; Jakoby, W.B. (1974) Glutathione s-transferase A. J. Biol. Chem. 249:7140-7148. Price, P.J.; Hassett, C.M.; Mansfield, J.I. (1978) Transforming activities of trichloroethylene and proposed industrial alternatives. In Vitro 14:290-293. 112 ------- Prough, R.A.; Patrizi, V.W.; Okita, R.T.; Masters, B.S.S.; Jakobsson, S.W. (1979) Characteristics of benzo(a)pyrene metabolism by kidney, liver and lung microsomal fractions from rodents and humans. Cancer Res. 39:1199-1206. Ramsey, J.C.; Andersen, M.E. (1984) A physiologically-based description of the inhalation pharmacokinetics of styrene in rats and humans. Toxicol. Appl. Pharmacol. 73:159-175. Reitz, R.H. (1985) Personal communication to H.L. Spitzer, Carcinogen Assess- ment Group, U.S. Environmental Protection Agency. Reitz, R.H.; Andersen, M.E. (1985) Physiologically-based pharmacokinetics and the risk assessment process for methylene chloride. Submitted to the U.S. Environmental Protection Agency by Dow Chemical Company, Midland, Michigan. July 15, 1985. Rodkey, F.L.; Collison, H.A. (1977a) Biological oxidation of 14C-methylene chloride to carbon monoxide and carbon dioxide by the rat. Toxicol. Appl. Pharmacol. 40:33-38. Rodkey, F.L.; Collison, H.A. (1977b) Effect of dihalogenated methanes on the in vivo production of carbon monoxide and methane by rats. Toxicol. Appl. Pharmacol. 40:39-47. Stevens, J.L.; Anders, M.W. (1978) Studies on the mechanisms of metabolism of haloforms to carbon monoxide. Toxicol. Appl. Pharmacol. 45:297-298. Stevens, J.L.; Anders, M.W. (1979) Metabolism of haloforms to carbon monoxide. III. Studies on the mechanism of the reaction. Biochem. Pharmacol. 28:3189-3194. Theiss, J.C.; Stoner, G.D.; Shimkin, M.B.; Weisburger, E.K. (1977) Test for carcinogenicity of organic contaminants of United States drinking waters by pulmonary tumor response in strain A mice. Cancer Res. 37:2717-2720. U.S. Environmental Protection Agency. (1984, Nov. 23) Proposed guidelines for carcinogen risk assessment. Federal Register 49:46294-46301. U.S. Environmental Protection Agency. (1985, Feb.) Health assessment docu- ment for dichloromethane (methylene chloride). Final report. EPA-700/ 8-82-004F. Prepared by the Office of Health and Environmental Assess- ment, Washington, D.C. Yesair, D.W.; Jaques. P.; Shepis, P.; Liss, R.H. (1977) Dose-related pharma- cokinetics of l^C-methylene chloride in mice. Fed. Proc. Am. Soc. Exp. Biol. 36:998 (abstract). 113 ------- |