Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
CARCINOGENICITY 7 OF FLUORIDE FLUORIDE CARCINOGENICITY IN HUMANS More than 50 epidemiological studies have evaluated the possibility of an association between fluoride concentrations in drinking water and human cancer. With one exception, the available studies are geographic correlation studies or geographic time-trend correlation studies in which measures of exposure and disease are made at the community level. The strengths and limitations of such ecological studies are addressed in Chapter 3. Because of the continuing importance of the question of fluoride in drinking water and human cancer, the relevant scientific literature has been exhaustively reviewed by several independent expert panels of epidemiologists. The two most comprehensive evaluations were con- ducted by the British Working Party on the Fluoridation of Water and Cancer under the chairmanship of E.G. Knox (Knox, 1985) and by an international pane! of epidemiologists convened by the Monographs Programme of the International Agency for Research on Cancer in Lyon, France QARC, 1982~. In addition, the epidemiological literature was reviewed by a subcommittee of the Drinking Water Committee of the National Research Council (NRC, 1977~. The latter review was less 109
Il0 Health Effects of Ingested Fluoride detailed than the Knox and lARC reports but reached similar conclusions; it will not be considered further here. The Knox and lARC panels considered all the available evidence relating to fluoride in drinking water and cancer in human populations. Given the high quality of the critical literature reviews already completed by Knox and co-workers, lARC, and other groups, the subcommittee elected to briefly summarize their findings and conclusions rather than conduct another independent review of the available literature. There have been few contributions to the literature since the publication of the Knox report (Knox, 19SS). Eight studies deserve consideration (Lynch, 1984; Hrudey et al., 1990; Hoover et al., 199la,b; McGuire et al., 1991; Mahoney etal., 1991; Cohn, 1992; Freni and Gaylor, 1992~. They will be described and evaluated separately. As noted earlier, correlations of exposure and disease or mortality rates among population aggregations made in ecological studies are subject to certain limitations. However, the exposure measure used in correlational studies of fluoridated water and cancer is unusual in that it applies to most or all individuals within a study area. Therefore, such studies might be better indicators of risk than other correlational studies, such as occupational or ethnicity studies, in which only a small fraction of a country's population usually is truly exposed. All the expert panels noted the relative strengths and weaknesses of the correlation studies. The expert pane} reviews generally agree that available data provide no credible evidence for an association between either naturally occurring fluoride or added fluoride in drinking water and risk of human cancer. A series of studies begun in 1975 by Yiamouyiannis and colleagues showed both geographic and temporal associations between fluoride in drinking water and risk of cancer mortality (e.g., Yiamouyiannis and Burk, 1977~. However, as revealed in great detail in Chapters 3 ant! 4 of the Knox report (Knox, 198S) and in many other critiques, those studies did not adjust adequately for differences in age, race, and sex of the compared populations. That resulted in inappropriate comparisons of groups that differed in one or more of those demographic factors. The Knox report concluded that there is "no reliable evidence of any hazard to man in respect to cancer." The lARC group (1982) came to a similar conclusion, namely, that "Variations geographically ant! in time in the fluoride content of water supplies provide no evidence of an association between fluoride ingestion and mortality from cancer in humans." The panels that evaluated the available epidemiological data on fluoride in
Carcinogenicity of Fluoride 111 drinking water recognized the limitations of demographic correlational studies in providing fully adequate data to make such evaluations. Lynch (1984) conducted a study using cancer incidence in Iowa muni- cipalities for the years 1969-1981. The relation between cancer incidence and added or natural fluoride in drinking water was analyzer] in 158 municipalities with a total 1970 population of 1,414,X78. A total of 66,572 cancer cases (comprising cancer of all sites, the bladcler, female breast, colon, lung, prostate, rectum, and other sites combined) were evaluated in four study groups, two each for aciclec] fluoride and for natural fluoride. In addition, the duration of exposure to fluoridated water was evaluated. Univariate and multivariate cancer-site, sex-specific statistical analyses were performed. Eight socioclemographic variables were evaluated in multivariate models to account for any effect they might have had on the fluoride-cancer relation. The results showed inconsistent relations between the fluoride variable ant! cancer incipience and failed to support a fluoride-cancer association. Hoover et al. (1991a) updated an earlier analysis of cancer mortality by county in the United States (Hoover et al., 1976) as related to county drinking-water fluoridation. The analysis used records of over 2,20S,000 cancer deaths by county for 1950-1985 to examine possible changes in county cancer mortality over that period. The study was restricted! to the white population to avoid confounding by racial variations in cancer mortality rates. Comparisons of age-acljusted cancer mortality by ana- tomic site and sex were macle between fluoridated and nonfluoridated counties for time periods preceding and following fluoridation. Cancers of the bones and joints (osteosarcoma is not reporter! separately in mor- tality statistics) were singled out for detailed analysis because of results of animal studies. Among both males and females residing in counties with rapid fluoridation (at least two-thirds of the population receiving fluoridated water within 3 years), the risk of death from cancers of bones and joints 20-35 years after fluoridation was the same as it was in the years immediately preceding fluoridation. Cancer-incidence data, as relater] to fluoriciation status of drinking water, were also analyzed in counties of two geographic regions covered by the Surveillance, Epictemi- ology and End Results Program (SEER), a large tumor registry supported by the National Cancer Institute. Both regions comprised fluoridated and nonfluoriciated counties. Over 125,000 newly diagnosed cases were included. For all types of malignancy, there was no eviclence of a consistent relation between cancer incidence or mortality ant! patterns of
Il2 Health Effects of Ingested Fluoride fluoridation. An additional study of cancer incidence in SEER areas analyzed incidence rates of osteosarcoma as well as all bone and joint cancers with respect to time of fluoridation (Hoover et al., 1992b). Although the rates of osteosarcoma were generally higher in the fluori- dated areas than in the nonfluoridated areas, they bore no relation to time of fluoridation. McGuire et al. (1991) conducted a small case-control study of osteo- sarcoma that included telephone interviews with 22 patients and matched controls. No association was found between osteosarcoma and average lifetime or childhood exposure to fluoride in drinking water. The small size of the population restricted the statistical power of this study to detect associations. A study of cancer-incidence rates in fluoridated and nonfluoridated communities in New York conducted by Mahoney et al. (1991) evaluated time trends of all bone-cancer rates between 1950 and 1987 and time trends of osteosarcoma rates between 1970 and 1987; a comparison of bone-cancer and osteosarcoma rates among males and females between 1976 and 1987 was also performed. Total bone cancer increased sig- nificantly in males less than 30 years of age, amounting to a 54% in- crease between 1955 and 1987, but decreased significantly in males and females (separately) 30 years of age and older. Osteosarcoma did not increase significantly in males of either age group or in females. The average annual sex- and age-specific rates of bone cancer and osteosar- coma for 1970-19X7 did not differ in areas with or without fluoridated drinking water. An ecological study in seven central New Jersey counties observed a higher rate of osteosarcoma in fluoridated communities than in nonfluori- dated communities in 1979-1987 (a risk ratio of 3.4 among males under 20 years of age) (Cohn, 1992~. Osteosarcoma was diagnosed in 12 males under 20 years of age in fluoridated communities versus eight males in nonfluoridated communities. Rate ratios comparing fluoridated with nonfluoridated communities were not elevated among females or among males in older age groups. The question of osteosarcoma rates in young males with respect to time of fluoridation was not examined in this study. A study in the Province of Alberta, Canada, compared the annual incidence rates of osteosarcoma for 1970-1988 in Edmonton, where water was first fluoridated in 1967, with rates in Calgary, where water was fluoridated in the fall of 1989 (Hrudoy et al., 1990~. The data showed no difference in rates between the municipalities and showed no time
Carcinogenicity of Fluoride 113 trend. The average annual incidence rates for 1970-1988 were 0.27 per 100,000 in Edmonton and 0.29 per 100,000 in Calgary, based on 26 and 29 cases, respectively. Freni and Gaylor (1992) conducted a time-trend analysis of bone- cancer incidence in the United States, Canada, and Europe. Bone-cancer incidence rates were obtained from 40 cancer registries covering a population of 150 million people for periods of 3~ years spanning the period 1958-1987. A registry area was considered fluoridated if fluoride at approximately ~ mg/L was added to the drinking water of a least 50% of the registry population in the 1960s or before. Because bone cancer is rare, the cumulative risk among people 10-29 years of age or 0-74 years of age was used to examine time trends. Significant increases in the cumulative risk of bone cancer were noted primarily among young males in some registry areas in the United States; significant decreases in lifetime risk were noted among both sexes in Europe. Neither the American increase nor the European decrease in risk was relate to fluoridation. Findings from the additional studies support the conclusions of the Knox report and the lARC pane! in that they provide no credible evi- dence for an association between fluoride in drinking water and risk of cancer. FLUORIDE CARCINOGENICITY IN ANIMALS The subcommittee reviewed six carcinogenicity studies in animals that have been reported in the literature to date (Table 7-~. However, only two carcinogenicity studies conducted by NTP and Procter & Gamble are considered adequate for determining the carcinogenic activity in animals. The first four studies are inadequate because of deficiencies in design or documentation of results, although they were consistently negative for an association of fluoride with carcinogenicity. Early Toxicological Studies Of Fluoride and Cancer Tannenbaum and Silverstone (1949) showed that the addition of 0.09% (10 mg/kg of body weight) sodium fluoride to the diets of female mice
I l4 Health Elects of Ingested Fluoride TABLE 7-1 Studies of Fluoride Carcinogenicity in Animals Study Sex/Species Route Dose Results Tannenbaum and Female mice Diet 10 mg/kg Negative Silverstone, 1949 Taylor, 1954 Female mice Dnnking 0.4-10 Negative water mg/L Taylor and Micea Dnnking 1-5 mg/L Accelerated Taylor, 1965 water growth of mammary gland Kanisawa and Male and Dnnking 10 mg/L Negative Schroeder, 1969 female mice water NIP study Male rats Drinking 25, 100, Equivocal (Bucher et al., water 175 mg/L increase in 1991) osteosarcomas NTP study Female rats, Dnnking 25, 100, Negative (Bucher et al., and male and water 175 mg/L 1991) female mice P&G study Male and Diet 4, 10, 25 Negative (Maurer et al., female rats mg/kg 1990) P&G study Male and Diet 4, 10, 25 Observation of (Maurer et al., female mice mg/kg osteomas in press) aSex not specified. resulted in a 10-40% reduction in body weight without a reduction in caloric intake. Mice on the fluoride diets also drank notably more water than did controls. There was a marked reduction in the incidence of spontaneous mammary tumors: after 100 weeks on the diet, 20 of 50 exposed mice developed mammary tumors as compared with 37 of 50 controls. (The subcommittee felt that the reduction in mammary neo- plasms in the exposer] mice was probably related to reclucec! body weight rather than to a true protective effect of fluoricle.) The authors also evaluated the tumor-promoting effects of fluoride by administering sodium fluoride in the diets of mice after subcutaneous injection of 0.15 mg of benzopyrene. After 52 weeks, 13 of 40 fluoride-exposed mice developed sarcomas, as compared with 7 of 40 controls. In contrast,
Carcinogenicity of Fluoride 115 there was a marked reduction in the incidence of primary lung tumors in fluoride-exposed Swiss and ABC mice after 60~2 weeks of treatment. The authors were unable to provide a simple explanation for the diver- gent effects of sodium fluoride on subcutaneous sarcomas and lung tumors induced by benzopyrene. Taylor (1954) exposed female DBA and C3H mice to low concentra- tions of sodium fluoride in drinking water (fluoride at 0.4-10 mg/~) for 7-17 months. Each treatment group comprised 12~2 mice, most groups comprising 20 animals. Many of the groups were maintained on a chow diet that contained fluoride up to 20 mg/kg; the remaining groups were fed a diet containing a negligible amount of fluoride. Mortality was greater in the fluoride-exposed groups, although the cause of the early deaths of animals given fluoridated water was not generally given. Mammary adenocarcinomas were observed in both exposed and control groups; in total, 59% of the deaths of exposed animals during the course of these studies were considered to be due to those tumors, as compared with 54% of the controls. The four groups of mice exposed to fluoride at 10 mg/L of drinking water averaged 63% mortality from cancer (all types of neoplasms combined) in comparison with S! % of the controls. The authors concluded that those ciata provided no indication that the incidence of cancer was increased as a result of exposure to fluoride. Despite the negative findings, the subcommittee felt that the stucly was inadequate because of the comparatively small number of animals as- signed to each treatment group and the relatively short period of observa- tion. The fluoride doses used in this study were also appreciably lower than current estimates of the maximum tolerated dose for fluoride (Bucher et al., 1991; Maurer et al., 1990, in press). Taylor and Taylor (1965) showed that low concentrations of sodium fluoride can increase the rate of growth of transplanted mammary adeno- carcinomas in DBA mice. In the study, sodium fluoride was added to the tumor tissue suspension before transplantation, to the drinking water of the host animals after transplantation, or by subJermal injection after transplantation. Acceleration of tumor growth was also demonstrated with tissue cultured in the yolk sacs of embryonated eggs. Tumor growth was enhanced by more than 100% in both cases. However, with ex- posure to higher concentrations of sodium fluoride, tumor growth was inhibited. The subcommittee observed that this mode! is unusual for
Il6 Health Effects of Ingested Fluoride studying tumor growth and noted the incongruous results at high and low levels of exposure. Kanisawa and Schroeder (1969) exposed white Swiss mice of the Charles River strain (Card) to fluoride at 10 mg/L (administered as sodium fluoride) in drinking water for life. Male mice exposed to fluoride survived I-2 months longer than unexposed controls. Ten percent of the males were still alive at 752 days of age; female mortality did not reach 90% until 789 days of age. Pathology examination was limited to a small number of tissues and macroscopically visible tumors. Of the 72 animals exposed to fluoride, 22 developed neoplasms (five of these were malignant) primarily in the lung. The lifetime incidence of specific tumors observed in those animals was comparable to that in the 71 controls. The authors concluded that oral ingestion of fluoride cannot be considered carcinogenic in mice at the dose given. The subcommittee felt that exposure was not adequately documented, particularly with respect to fluoride concentrations in bone. The pathological evaluation was also inadequate because bone lesions were not adequately assessed. Recent Carc~nogenicity Bioassays Two recent studies have raised some concern about the ability of fluoride to induce cancer in animals. The study conducted by NTP (NTP, 1990; Bucher et al., 1991) indicated a possible increase in the incidence of osteosarcomas in F344/N male rats, although no such lesions were observed in female rats. B6C3F~ mice also failed to demonstrate a carcinogenic response to fluoride. The incidence of osteosarcomas in male and female rats in the study is shown in Table 7-2. Sodium fluoride was administered in the drinking water at concentrations of 25, 100, and 175 mg/L for 103 weeks. Fifty animals of each sex were assigned to the two lowest-dose groups, and 80 were assigned to Me high-dose and control groups. Mild fluorosis was observed in teeth of all dose groups of both mice and rats, although more so in rats. Four osteosarcomas were observed in male rats: one in the group exposed to 100 mg/L and three in the group exposed to 175 mg/~. Two of the three osteosarcomas in the high-dose group were detected wig radiographs. Although the incidence of osteosarcomas in the high- dose group was not significant ~ ~ 0.05) in relation to the control
Carcir~ogenicity of Fluoride 117 TABLE 7-2 Incidence of Oteosarcomas in Rats Exposed to Sodium Fluoride in Dnnlcing Water Females Concentration, Rats mg/L Males 0 25 100 175 o No. of Animals 80 51 No. of Osteosarcomas o o 50 80 80 25 100 50 50 o o o 175 81 0 Source: NTP, 1990. response rate, the trend in response to increased dose was significant (p = 0.027~. NTP considered those data "equivocal evidence of carcino- genic activity" in male rats, and "no evidence of carcinogenicity" in female rats or in male or female mice. The subcommittee felt that the study was generally well conducted in accordance with current bioassay standards ant] agreed with the interpretation of the results in isolation. However, the subcommittee also felt that the results must be examined in conjunction with results of other studies, particularly those of Maurer et al. (1990, in press). The Maurer et al. (1990, in press) studies for Procter & Gamble (P&G) clearly indicated that sodium fluoride increased the incidence of osteomas (noncancerous bone tumors) in male and female CD-! mice (Table 7-3) but not in male and female Sprague-Dawley rats. In the experiments, 60 female and 60 male mice and 70 female and 70 male rats were exposed to dietary sodium fluoride at concentrations of 4, 10, or 25 mg/kg of body weight per day. The length of the rat study, 95 weeks for males and 99 weeks for females, was based on a termination criterion of 20% survival. In the mice study, the 20% survival rate was reached at 95 weeks for males and 97 weeks for females. Fluoride exposure induced severe fluorosis of bones and teeth in mice and rats ant! osteomas in both sexes of mice. The investigators cautioned that the increase in osteomas in mice might have resulted from a con
Il8 Health Effects of Ingested Fluoride TABLE 7-3 Incidence of Osteomas in Male and Female CD-1 Mice Exposed to Sodium Fluoride in Their Diet . . . . No. of Concentration, No. of Animals with No. of Mice mg/kg Animals Osteomasa Osteosarcomas Male 0 50 1 0 0 45 0 0 4 42 0 0 10 44 2 0 25 50 13 0 Females 0 50 2 0 0 45 3 0 4 42 4 0 10 44 2 0 25 50 13 0 Historical control incidence in CD-1 mice is ~ 1%. Source: Maurer et al., in press. laminating retrovirus. The subcommittee noted, however, that the virus would have to preferentially affect the exposed groups in relation to the control groups to be solely responsible for the observed dose-response relation. This point is discussed further below. In attempting to resolve the apparent differences between the NTP and P&G stuclies, several issues need to be considerecI. · Different routes of exposure were used in the NTP and P&G stud- ies. To evaluate the different results for rats in the two studies, a com- mon measure of close is neecled. Doses in the two studies can be com- pared by using bone fluoride concentrations. Under the exposure condi- tions imposed in both studies (continuous exposure to fluoride in water or feed), fluoride concentration in bone represents the most reliable indicator of the total body burden of fluoride. Using that indicator, the highest dose used in the P&G study was 2 to 3 times greater than the highest dose used in the NTP stucly (Table 7-4) and yet it did not induce bone neopiasms. Doses of fluoride at 4 mg/kg in feed in the P&G stucly resulted in bone-ash fluoride concentrations comparable to those in the
Carcinogenicity of Fluoride 119 TABLE 7 - Comparison of NTP and P&G Studies Fluorde Concentrations in Bone Ash (pa of fluor~de/mg of bone ash) Fluoride Rats Mice Dose Males Females Males Females NTP Study Control 445 (O)a 554 (0) 719 (o) 917 (o) 25 mg/L 978 (o) 1~348 (0) 19606 (0) 1~523 (0) 100 mg/L 3~648 (1) 3~726 (0) 3~585 (0) 4~370 (0) 175 mg/L 5~263 (3)b 5~554 (0) 5~690 (0) 69241 (0) P&G Study Control 1 467 (0) 505 (0) 1~582 (1) 971 (2) Control 2 691 (o) 785 (o) 1~676 (0) 1~295 (3) 4 mg/kg 5~014 (0) 4~541 (0) 4~405 (0) 3~380 (4) 10 mg/kg 8~849 (0) 8~254 (0) 7~241 (2) 69189 (2) 25 mg/kg 16~761 (0) 14~428 (0) 137177 (13)C 10~572 (13)C : - aNumber in parenthesis is the number of osteosarcomas (in NTP study) or osteomas (in P&G study). bEquivocal evidence of increased incidence of osteosarcomas (Bucher et al., 1991). CIncreased incidence of osteomas (Maurer et al., in press). highest-dose group (NaF at 175 mg/L of drinking water) in the NTP study. · It is important to evaluate the NTP tingling of equivocal evidence of carcinogenic activity in male rats in the context of other studies. Such a weight-of-the-evidence approach is necessary to make an overall ju~ig- ment concerning fluoride carcinogenicity. Even though the four earlier animal studies each suffered from deficiencies in design or interpretation, none gave any indication of an association between fluoride and osteosar- coma. Collectively, they provide limited support for the hypothesis of no association between exposure to fluoride and increased risk of osteo- sarcoma. Even though the highest dose of fluoride in the P&G rat study was more than twice the highest dose in the NTP study in terms of bone fluoride concentrations (Table 7-4), no osteosarcomas were observed in
120 Health Effects of Ingested Fluoride either sex of rats in the P&G study. Therefore, He P&G study does not support the equivocal findings in the NTP study. However, different rat strains were used in these two studies; therefore, possible differences in susceptibility in these two strains cannot be ruled out. · The role of the retrovirus in the P&G mouse studies is not clear. Because the virus is transmitted vertically, mice in all dose groups were infected. Also, there is no evidence that viral infection was linked to the occurrence of osteomas. An association between this retrovirus and osteoma has not been previously established: therefore, the subcommittee sees no convincing evidence that the retrovirus alone caused the osteo- mas. The virus might conceivably have had a potentiating effect on the induction of osteomas, although evidence for that is lacking. The sub- committee believes that the most obvious explanation for the increased incidence of osteomas in mice is that the dose in the P&G study (based on bone fluoride concentrations) was more than twice as high as the dose in the NTP study (Table 7-4) and that the occurrence of osteoma in the P&G study was related to exposure to fluoride. · Of equal importance is the relevance of the mouse osteomas in terms of their relation to the potential carcinogenic activity of fluoride in animals and, by extension, in humans. The subcommittee considered the biological importance of the osteomas observed in mice in the P&G study from several perspectives. Are they part of a neoplastic continuum? Do they have the potential for malignant change? Should they be considered evidence of carcinogenic activity? Should they be considered a hyper- plastic response? Are they a unique lesion induced by exposure to fluoride? A critical issue in the P&G mouse study is whether the osteomas observed at the highest dose in male and female mice are related to osteosarcomas or to other neoplastic diseases. In an attempt to answer that and other questions, the subcommittee asked the Armed Forces Institute of Pathology (AF1P) to review the osteomas in the P&G study (Appendix 2~. The results of their review are in Appendix 3. According to the AF1P, arguments that support the view that the osteomas are nottrue neoplasms include the following: (~) none of the lesions progressed beyond the benign condition; (2) none of the tumors showed characteristics of precancerous change; (3) in contrast to primary bone neoplasms, which are usually unicentric in origin (in animals as
Carcinogenicity of Fluoride 121 well as humans), many of the osteomas in mice were multicentric; and (4) there is no human counterpart to this lesion. · In the scientific community, there are clear differences in opinion concerning the relevance of high-dose toxicological studies in rodents to humans (Rail, 1991; Ames, 1991~. Factors to be considered include the relative bone and plasma fluoride concentrations in animals exhibiting lesions in relation to the concentrations observed in human populations. Another factor to be considered is the relative plasma fluoride concentra- tions. Unfortunately, reliable plasma fluoride values are not available from the NTP study. Suppose that, on the basis of the results of the NTP study, fluoride induces osteosarcoma in male rats at concentrations of approximately 5,300 ppm in bone ash. The bone-ash concentrations in mice where osteomas appeared to be related to fluoride exposure were approximately 13,200 ppm in males and 10,500 ppm in females Crable 7-4~. Studies of fluoride in human bone ash indicate that concentrations in people 70 years of age and older might exceed 4,000 ppm in exceptional cases where exposure to fluoride has been high. Osteosarcoma occurs in humans almost exclusively in the first three decades of life. Bone-ash concentrations in people ofthat age are substantially lower, generally less Wan 3,000 ppm. Osteosarcoma is associated with long-bone growth in humans. The long bones cease to grow in teenaged humans but continue to grow throughout the lifetime of rats. Thus, rats are susceptible for a longer portion of their lifespan than humans. In fact, the osteosarcomas ob- served in the NTP study were found exclusively in older rats. CONCLUSIONS More than 50 epidemiological studies have been conducted to evaluate the relation between fluoride concentrations in drinking water and human cancer. With minor exceptions, these studies used the method of geo- graphic or temporal comparisons of fluoridation status and regional cancer rates. There is no consistent observation of increased cancer risk wig drinking-water fluoridation; most of the studies show no association. The large number of epidemiological studies combined with their lack of positive finclings implies that if any link exists, it must be very weak.
122 Health Elects of Ingested Fluoride Based on the weight of the available evidence, the subcommittee believes that the collective data from the rodent fluoride toxicological studies do not present convincing evidence of an association between fluoride and increased occurrence of bone cancer in animals. The equivo- cal result of osteosarcoma in male rats in the NIP study was not sup- ported by results in females in the same study or by the P&G rat study, even though the latter had much higher exposure levels. That suggests that the male rat osteosarcomas observed in the NTP study were not re- lated to fluoride exposure. The subcommittee concluded that the increased incidence of osteomas in male and female mice in the P&G study is most likely related to fluoride, although the presence of a contaminating retrovirus was considered a confounding factor. Consideration was given to the biological significance of the treatment- related increased incidence of benign osteomas in male and female mice at the highest dose in the P&G study. The key question is whether the osteomas are true neoplasms or not. The AF1P concluded that these lesions are not true bone neoplasms, and, therefore, it would be inappropriate to use them to establish carcinogenic activity. in fact, they are more reminiscent of hyperplastic (hyperostoses) than neoplastic lesions. The subcommittee concluded that the osteomas observed in mice at the highest dose in the P&G study are of questionable biological significance in terms of their relevance to humans, especially in light of their occurrence at fluoride exposures far greater than those that are likely to occur in humans. The subcommittee concludes that the available laboratory data are insufficient to demonstrate a carcinogenic effect of fluoride in animals. The subcommittee also concludes that the weight of the evidence from more than 50 epidemiological studies does not support the hypothesis of an association between fluoride exposure and increased cancer risk in humans. RECOMMENDATIONS The subcommittee observed that weak associations between cancer risk and exposure to fluoride, if they exist, might be more readily identified
Carcinogenicity of Fluoride 123 in analytical studies based on individual outcome and exposure informa- tion than in ecological studies based on aggregate outcomes and expo- sures. Thus, conducting well-designed studies with information from individuals is important in the ongoing evaluation of fluoride carcinogeni- city. The subcommittee therefore recommends conducting one or more highly focused, carefully clesigned analytical studies (case control or cohort) of the cancer sites that are most highly suspect, based on data from animal studies and the few suggestions of a carcinogenic effect reported in the epidemiological literature. Such studies should be cle- signec! to gather information on individual study subjects so that adjust- ments can be made for the potential confounding effects of other risk factors in analyses of indivicluals. Information on fluoricte exposure from sources other than water must be obtained, and estimates of exposure from drinking water should be as accurate as possible. In addition, analysis of fluoride in bone samples from patients and controls would be valuable in inferring total lifetime exposure to fluoride. Among the disease outcomes that warrant separate study are osteosarcomas ant! cancers of the buccal cavity, kidney, and bones and joints.