Click for next page ( 52


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 51
FLUORIDE ExPosuRE AND 3 RISK OF BONE FRACTURE BONE FRACTURE IN HUMANS Information from two types of studies on bone fracture and case reports of skeletal fluorosis address a possible link between fluoride ant! bone fracture. One group of studies involves] clinical trials clesignec} to test the efficacy of fluoride supplements in strengthening bone and preventing further fractures in those with osteoporosis. Physicians have used fluo- ride for those purposes for almost 30 years, but until recently, there were no data from systematic, well-controlled clinical trials that addressed the question. Typically, exposure to sodium fluoride in these studies ranged from 50 to 80 mg per day, more than an order of magnitude above the typical exposure to fluoride from fluoridated drinking water. A second group of studies analyzed bone-fracture rates among persons exposed to fluoridated and to nonfluoridated drinking water. With two exceptions, these studies used population data and an ecological design. (Studies of geographic or temporal association, in which population rather than individual data are used, are called ecological studies.) The limitations of this type of study are well known and are described later. Some of the information reviewed here was presented at a workshop on fluoride, hip fractures, and bone health held at the National Institutes of Health (Gordon and Corbin, 1992~. 51

OCR for page 51
52 Health Effects of Ingested Fluoride Clinical Trials of Fluoride Four recent clinical trials with random, controlled designs provide information on exposure to relatively high concentrations of fluoride among high-risk groups for fractures related to osteoporosis. The daily intake of the human subjects in these trials ranged from 50 to 80 mg of sodium fluoride (about 40% fluoride, or 20-32 ma) per day. These fluoride intakes are approximately lO-IS times those of people in the highest decile of drinking-water consumption who live in areas where water is fluoridated at ~ mg/~. Dambacher et al. (1986) conducted a clinical trial of 15 osteoporotic patients treated with sodium fluoride at 80 me per day and 14 placebo- treated controls. In the first year, the fracture rate of treated patients was significantly higher than that of the untreated group, but no differences were observed in the second and third years of the trial. After 3 years, the cancellous bone density of treated patients was 8% higher than untreated patients, but total bone density was not increased. Forty-seven percent of the treated group experienced osteoarticular pain and swelling in the lower extremities, attributed to stress fractures, whereas no un- treated osteoporotic patients experienced these symptoms. Gastric dis- tress was not observed. Based on results of previous studies by the authors, neither calcium nor vitamin D was added cluring the trial. Mamelle et al. (1988) conducted a randomized trial of osteoporotic patients in France with at least one nontraumatic vertebral crush frac- ture. Treatment with 50 mg of sodium fluoride per day, supplemented with ~ g calcium and BOO international units of vitamin D, was adminis- tared to 257 patients, and 209 patients received other stand are! treatment regimens that did not include sodium fluoride. The number of vertebral crush fractures in the first year was similar in the two groups, and the mean number was significantly lower in the soctium-fluoricle-treatec} patients in the second year. Ankle and foot pain was elevates! in the treated group and digestive disorders occurred with equal frequency in the sodium fluoride and other treatment groups. Two double-blind, placebo-controlled trials were supported by the National Institute of Arthritis ant! Musculoskeletal and Skin Diseases and conducted at the Mayo Clinic (Riggs et al., 1990) ant] the Henry Forc! Hospital (Kleerekoper et al., 1989~. Sodium fluoride was administered

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture 53 at 75 mg per day in both trials, and patients received 500 mg of calcium daily in addition to sodium fluoride or placebos. Riggs et al. (1990) reported data from a 4-year trial of 202 randomized patients with osteoporosis and vertebral fractures. Sixty-six women in the fluoride group and 69 women in the placebo control group completed the trial. Among those receiving fluoride, bone-mineral density, as measured by dual-photon absorptiometry, increased by 35% (o < 0.0001) in the lumbar spine (mostly trabecular bone) and 12% in the femoral neck (n < 0.0001) but decreased by 4% ~ < 0.02) in the radius (predominantly cortical bone). The number of new vertebral fractures was slightly lower in the sodium-fluoride treatment group, but the number of nonvertebral fractures was significantly higher in the treatment group (n ~ 0.01) than in the control group. The relative risk for nonvertebral fractures, either incomplete or completefractures, was 3.2 (95% confidence interval (CI) = .X-5.6; 61 patients treated with sodium fluoride had 72 nonvertebral fractures, and 24 controls had 24 fractures, either complete or incomplete. There were six hip fractures in the sodium-fluoride treatment group (three incomplete and three complete), and one (complete) hip fracture in the control group. Fifty- four women treated with sodium fluoride and 24 controls experienced side effects, mostly gastrointestinal symptoms and lower-extremity pain, warranting dose reduction. In the Henry Ford Hospital study, Kleerekoper et al (1989) found no significant difference in vertebral fractures between the sodium fluoride and the control groups. As in the Mayo Clinic study, gastrointestinal side effects and episodes of lower-extremity pain were much more common in the sodium-fluoride group than in the control group. On bone biopsy, 17% of those treated with sodium fluoride had mineraliza- tion defects, whereas none was present in the controls. Population Studies The risk of bone fracture in the elderly has been studied in populations exposed to naturally occurring or adcted fluoride in drinking water and compared with groups exposed to low concentrations of fluoride in drinking water. Of the 10 studies considered by the subcommittee, two

OCR for page 51
S4 Health Effects of Ingested Fluoride of the investigations used information from individuals and eight relied on population-based statistics collected for other purposes. studies Using Information from Incubi - duals Cauley et al. (1991) recently reported on data from participants of an ongoing study of osteoporotic fractures in women in the Pittsburgh area. Residential histories and drinking-water sources were gathered from 1950 to 1990 for 1,878 white women ranging in age from 65 to 93 years (mean = 70.9 years). About 10% of the women had exposures to fluoride in drinking water of ~ mg/L for more than 20 years, and 58% had negligible exposures. Public water constituted 73% of exposure- years, and the mean duration of exposure was 6.0 + 9.24 years (range 0-3X years). No association was found between years of fluoride ex- posure and bone mineral density with or without adjustment for body- mass index and age. There was also no association with history of fracture. The authors concluded that "these data do not support a protec- tive relation between exposure to fluoridated water ant! bone mineral density in this population of elderly women." Women in three small communities in northwest Iowa, who hac! resided in the same town for at least 5 years, were the subjects in a study by Sowers et al. (1991~. The communities were ethnically similar ant! all eligible women were of Northern European background. One com- munity had water with natural fluoride at high concentrations (4 mg/~), a second] had unusually high calcium concentrations, ant! the third, used as a comparison community, had fluoridated water (! mg/~) and normal calcium concentrations. Here, we restrict our attention to the comparison between the two communities with high and normal fluoride concentra- tions. There were 417 subjects in the high-fluoride-concentration com- munity and 194 in the comparison area; they were first enrolled in 1983- 1984 and contacted for followup 5 years later; overall participation rates were 74% and 63%, respectively, at followup. Significantly lower radial bone mass was observed in both younger and older women in the high-fluoride-concentration community. Bone density of the proximal femur was clinically similar in women in the two com- munities. Postmenopausal, but not premenopausal, women in the high

OCR for page 51
Fluoride Exposure arid Risk of Bone Fracture 55 fluoride community reported significantly more fractures. Relative risk estimates of fracture were adjusted for body-mass index, age, and thia- zide use. The S-year relative risk for any fracture among menopausal women was 2. ~ (C! = i.02-4.4) and for wrist, spine, or hip fractures, 2.2 (CT = .-4.7. The S-year fracture rate among premenopausal women was also elevated in the high-fluoride community, but chance could not be excluded (relative risk = ~ . 8, C! = 0.45-~.2~. This study from Iowa is discussed here because extensive information on water consumption, use of replacement estrogens, and other factors was gath- erec! from individual participants. However, little of the data from individuals has been reported, and the analyses were essentially ecologi- cal in character. Studies Using Population Characteristics Eight studies have examined the geographic correlation between hip fracture rates ant! consumption of fluoridated water or changes in hip fracture rates in the same geographic area before anti after fluoridation was begun. The ability of ecological studies to detect excess risk is limited in several respects: (~) lack of indiviclual information on specific exposures, including fluid ingestion patterns ant! exposure to fluoride from sources other than drinking water, of affected and unaffected members of the population; (2) inability to measure fluoride exposures of study subjects who migrated into study areas before diagnosis of disease or death; and (3) limited or nonexistent possibility of adjusting for differences in risk (confounding) factors, such as smoking, occupational exposures, exogenous estrogens, and dietary patterns, that might in- fluence differences in disease rates in fluoriciatec! anti nonfluoriciatect areas. Confounding can give rise to spurious results, such as positive associations that are not truly present, or alternatively, positive associa- tions that are masked. To the extent that migration is a factor, it is likely to diminish the sensitivity of a geographic correlation study to detect excess risk. An additional problem can arise in correlational studies that use mortality statistics, as occurs in many of the studies of fluoride and cancer, because geographic variation in mortality might reflect differ- ences in access to, or quality of, meclical care, and not differences in

OCR for page 51
S6 Health Effects of Ingested Fluoride underlying disease incidence. In this regard, it is helpful to distinguish between ecological studies that make geographically based comparisons and those with a time-trend design. The latter are less subject to con- founding from many factors, because they use rates from the same geographic area. Consequently, time-trend studies are considered to have the stronger design. In some cases, ecological studies that show positive associations might be more likely to be published or more rapidly pub- lished than those that show no associations. When such publication bias exists, literature is likely to be biased toward showing positive effects. The phenomenon of publication bias is less likely to occur with results from case-control or retrospective-cohort studies, because the large effort involved in conducting these studies is usually attended by an early commitment to publish, regardless of outcome. Given the limitations of ecological studies, most epidemiologists consider them valuable in in- dicating the likelihood of positive links or in demonstrating the feasibility of hypotheses. When results from a number of such studies (conducted in different times and places) converge to indicate either an exposure- disease relation or no such relation, confidence in the collective findings is bolstered. When many studies fail to observe exposure-disease rela- tions, the possibility of small risks or protective effects, undetected at a population level, cannot be excluded. In a report presented at a workshop on drinking-water-fluoride in- fluence on hip fracture and bone health held at the National Institutes of Health (see Gordon and Corbin, 1992), Keller (1991) compared hip fracture rates in 216 U.S. counties with natural fluoride concentrations greater than 0.7 mg/L and with rates in 95 counties with naturally low fluoride concentrations (less than 0.4 mg/~) in drinking water. The counties with high concentrations were placed in four groups (0.7-~.2, I.3-2.0, 2.~-3.9, and 4.0 mg/L and above). Hip fracture ratios were calculated as the reported fracture rate in these county groups divided by the rate in counties with low fluoride concentrations. At optimal fluoride concentrations (0.7-~.2 mg/~), no significant increase in hip fracture ratio was found (risk ratio = i.016~. However, Keller found significant increases in hip fracture ratio at higher concentrations of fluoride in drinking water. The adjusted risk ratio was 1.224 for counties in the highest exposure group. In addition to the usual limitations of ecological studies, this investigation was limited by lack of control for other demo- graphic factors in the computations. In another study presented at the workshop, May and Wilson (1991)

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture S7 used data from 438 U.S. counties with populations greater than 100,000, representing 70% of the U.S. population. The percentage of the popula- tion that received natural or added fluoride in drinking water (approxi- mately ~ mg/~) was estimated for each county. As the percentage of persons exposed to fluoride in water increased, the hip fracture rate generally increased. The regression coefficients (R) were calculated as change in hip fractures per 1,000 persons at risk for each ~ % of the population receiving fluoridated water at a concentration of approximately ~ mg/~. The value was significant for white men (R = 0.0037, cr = 0.002-0.005) but not significant for white women (R = 0.0016, C} = 0.001~.005~. In an additional analysis using data from 51 counties (more than 80% of the population was exposed to fluoride), including data on duration of exposure to fluoridated water, the hip fracture rate was higher in counties with up to 10 years of exposure, about 20% lower in counties with Il-~8 years of exposure, and intermediate in counties with more than IS years of exposure. As in the Keller study, this study was limited by the lack in regression models of county demographic factors other than fluoridation status. Jacobsen et al. (1990) examined hip fracture risk among white women in more than 3,000 counties throughout the United States. Weighted least-squares regression methods were used to examine the association of age-adjusted county hip fracture incidence rates in each county with the percentage of the county population exposed to natural or added fluoride in drinking water. The data were weighted by the number of white women over the age of 65 years. After adjustment for other study variables (i.e., poverty rate, percentage of land in farms, water hardness, sunlight, and latitude), the regression coefficient was statistically sig- nif~cant (R = 0.003, p = 0.0009~. It should be noted that weighting of regression equations directly by population size might place too much emphasis on counties with large populations. It is often more appropriate to weight least-squares regression equations by the square root of the population, proportional to the standard error of the estimate. In another geographic comparison study, Jacobsen et al. (1992) ex- amined hip fracture rates in 129 fluoridated and 194 nonfluoridated counties that had been the subject of a National Cancer Institute study of cancer mortality and fluoridation. More than half of the eligible counties were urban and had natural fluoride concentrations of less than 0.3 mg/~. The proportion of the population receiving fluoridated water in fluori- dated counties increased from less than 10% to more than 67% within a

OCR for page 51
58 Health Effects of Ingested Fluoride 3-year period. Nonfluoridated counties had less than 10% of the popula- tion receiving fluoridated water. A small but significant positive associa- tion was found between fluoridated water and hip fracture among white men (relative risk = 1.17, C! = i. I-~.2) and white women (relative risk = I. I, C! = I.06-~.10) over 65 years of age. The counties most recent- ly (0-5 years) fluoridated had the highest rates of hip fracture; rates were lower in counties with longer fluoridation exposure. Jacobsen et al. (1993) also examined hip fracture rates in Rochester, Minnesota, for the lO-year period before and the lO-year period after the drinking-water supply was fluoridated in 1960. The overall annual incidence among persons 50 years of age and older was 483 (C! = 370- 597) per 100,000 in 1950-1959, and 450 (CI = 362-537) per 100,000 in 1960-1969. Using Poisson regression models to control for calendar time and age, they found that the relative risk associated with fluoridation was 0.60 (CT = 0.46-0.86) for women ant! 0.78 (C} = 0.37-~.66) for men. In another time-trend study, Goggin et al. (1965) compared femoral fracture rates in Elmira, New York (1960 population 46,517) during the S years before and the S years after fluoridation of the city's water supply in 1959. Femoral fracture rates for Elmira women 60 years of age or older did not differ significantly before or after fluoridation occurred. In a geographic comparison study from Utah, Danielson et al. (1992) compared the incidence of femoral neck fractures in patients over 65 years of age living in a fluoridates] community with the incidence in two nonfluoridated areas. The outcome measure was the rate of hospital discharge for hip fracture. Among women, the age-acljusted risk ratio was 1.27 (C! = I.08-~.46) and among men, 1.41 (C! = 1.00-~.~. These ratios were based on hip fractures in 65 women and 19 men in fluoridated communities and in 130 women and 32 men in nonfluoridated communities. The criteria for choosing the two nonfluoridated control areas were not presented, nor were hip fracture rates given separately for the two areas. That there might have been important differences between the exposed and nonexposed populations other than fluoridated drinking water is suggested by relative differences in their 1980 and 1987 popula- tions. In the fluoriciatec! community, the over-65 population decreased by 8% from 1980 to 1987, and in the two nonfluoridatec! areas, the over- 65 population increased by 64% and 30%, respectively. A geographic correlational study by Cooper et al. (1990, 1991) con- sidered hip fracture hospital discharge rates in 39 urban areas of England,

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture S9 9 of which had fluoridated water supplies and 30 of which did not. Initial analyses, which did not include adjustment for the precision of fracture rate estimates, did not find an association with fluoride con- centrations in drinking water (R = 0.016, p < 0.34) (Cooper et al., 1990~. However, in subsequent analyses, which used a weighted least- squares regression technique (weighting each district's rate estimate by the size of the population 45 years of age and older), a significant as- sociation between hip fracture discharge rate and fluoride concentration was found (R = 0.41, p < 0.001~. As with Jacobsen et al. (1990), the use of this type of weighting in regression models might have placed too much emphasis on areas with large populations. Skeletal Fluorosis Finally, the subject of human skeletal fluorosis will be discusser] briefly. Smith and Hodge (1979) have described the preclinical and clinical stages of skeletal fluorosis. The asymptomatic preclinical stage is characterized by slight increases in bone mass that are cletectable radiographically and bone-ash fluoride concentrations between 3,500 and 5,500 ppm. The typical fluoride concentrations in bone ash from persons who have chronically consumed optimally fluoridated water are less than 1,500 ppm. In stage ~ of skeletal fluorosis, there might be occasional stiffness or pain in the joints and some osteosclerosis of the pelvis and vertebral column. Bone-ash fluoride concentrations in stage ~ usually range from 6,000 to 7,000 ppm. When bone-ash fluoride concentrations are 7,500-8,000 ppm or more, stages 2 and 3 of skeletal fluorosis are likely to occur. The clinical signs of these stages are chronic joint pain, dose-related calcification of ligaments, osteosclerosis, possibly osteoporo- sis of long bones, and in severe cases, muscle wasting and neurological defects. Crippling skeletal fluorosis might occur in people who have ingested 10-20 mg of fluoride per day for 10-20 years. During the last 30 years, only five cases have been reported in the United States. The history of fluoride intake for two of the cases was determined with reasonable accuracy (Sauerbrunn et al., 1965; Goldman et al., 1971~. The individu- als consumed up to 6 ~ of water per day containing fluoride at 2.4-3.5 mg/L in one case and 4.0-7.8 mg/L in the other. The cIaily fluoride

OCR for page 51
CO Health Effects of Ingested Fluoride intake was estimated at 15-20 mg for 20 years. In general, this intake would be associated with a drinking-water supply containing fluoride at about 10 mom. Thus, crippling skeletal fluorosis in the United States has been rare and not a public-health problem Peons et al., 1954; Stevenson and Watson, 1957), even though for many generations there have been communities with drinking-water fluoride concentrations in excess of those that have resulted in the condition in other countries (Singh and folly, 1970~. The puzzling geographic distribution of the disorder usually Is ascribed to unidentified dietary factors that render the skeleton more or less susceptible. The small number of cases of skeletal fluorosis in the United States has rulM out the possibility of systematic epidemiological evaluation. Based on limited data in the literature on skeletal fluorosis, the subcom- mittee concludes that skeletal fluorosis is not a public health issue in the United States. Discussion Of the six epidemiological studies that used geographic comparisons (where no actual intake data were available), four found a weak associa- tion between fluoride in drinking water and a small increase in the risk of hip (or other bone) fracture. In two of the four studies that observed associations, the weighting scheme for regression models might have been inappropriate. The risk increase in the positive studies was small. Of two additional studies that examined time trends in fracture rates before and after water fluoridation, one showed no association (Goggin et al., 1965) and the other found a negative association Jacobsen et al., 1993~. The time-trends ecological method is consiclered the stronger approach because there is less opportunity for confounding than in geographic comparison studies. Given the multiple limitations of ecologi- cal analyses, the possibility for publication bias in favor of positive findings and Me potential in all the studies for confounding from factors common to all, these studies offer only limited support for a hypothesis of a weak association between fluoridated water and hip fracture, which requires confirmation in studies of individuals. Of the two studies with information on individuals, the analytical approach in one was essentially

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture 61 ecological (risk of hip fracture increased with fluoride at 4 mg/L of drinking water), and the over showed no difference in fracture risk in women who drank fluoridated or nonfluoridated water. In view of the conflicting results and limitations of the current data base on fluoride in drinking water and risk of hip or other fractures, there is no basis at this time to recommend that EPA lower the current maximum contaminant level (MCL) of fluoride of 4 mg/~. The subcommittee recommends that additional studies of hip and over fractures be conducted in geographic areas with high and low concentra- tions of fluoride in drinking water, and that studies should use informa- tion from individuals rather than population groups. In these studies, it is important that individual information be collected about fluoride intake from drinking water and from all other sources, reproductive history, past and current hormonal status, intake of dietary and supplemental calcium and other cations, bone density, and other factors that might influence risk of fracture. BONE FRACTURE IN ANIMALS Studies with laboratory animals designed to determine the effects of fluoride on bone strength or fracture resistance have been done wig several species, a variety of doses and periods of exposure, different stimuli to influence bone growth or resorption, and different measurement techniques. The associated bone fluoride concentrations have been documented in only some of the reports. The results of these studies on bone strength have yielded all possible outcomes (i.e., no effect, in- creased strength, decreased strength, and Aphasic response). The most frequent finding, however, has been the absence of an effect. A repre- sentative sample of these reports will be discussed. The main features and findings of the studies are shown in Table 3-~. Outcomes of studies showing a decrease in bone strength associated with high fluoride intake include those by Gedalia et al. (1964), Faccini (1969), and Wolinsky et al. (19721. Beary (1969) and Riggins et al. (1974) reported fluoride-associated decreases in bone strength in animals fed a calcium-deficient diet but not in animals fed a diet adequate in calcium. In the study by Gedalia et al. (1964), weanling female "Sabra"-strain

OCR for page 51
62 .= ,_ Lo 4 - Cat En U. .g .s at: o o o o m ~ e o ~ O - i ~ 3~ so ~ d ~ ox ~a, ~ cat ~2,, ~ ~ At Vat i CCI ~ ~m ~ ~ rat& ei ~ ~i ;, .5 ~ ~5 = ~5 8 C ~ o in, O at ~ ~ g S ~ O Z ~ m~ E i, ~^ ~ ~ A rat A A. 0 I,, as _ z z ~ 3 ~ z 3 ~g =^ o 8 Ha, ~o^ ~^ ~ . ~As- ~ ~ - ~,D - ~ - ~_ _ ~

OCR for page 51
63 t To in: ~ ~>4 3 ~ ~ 3 ~ ~ ~ r ~ ~ at, ~ Ida i u f ~ O O ,3 ~ ~ r C ~,~ ~Z ~ _ O C ~ E: ~E: E: ~ ~c: ~ m ~m All: ~-= O O ~ 8 o O ,,8 O O ~e,) O Ir) ~ ~ ~,, ~- ~t ~C X ~ ~ ~ O O O t ~ O ~ C V, :3 o ~ ~ ~ ~ ~Z ~o 3 s ~ o ~ ~o ~_ V) ~ _ _ C ~ (f) _ ~ ~. o ._ o ~ o _ _ ~o ~o ~,> =~- c~- ~ - 3 ~- 3 _ o _o . ~ ~_ 8 o ~o ~o o ~o ~o ~ ~o _ _ _ ~_ _ ~b~- ~

OCR for page 51
64 a a: V) : .= .5 i_ = - - ~ ~s ~s ~ Z : ~ r 0 0 Z ~ . ~ ~ fi e sit a ~c 3 ~ E ~ | ~-~ o ~ o ED ~ ~O o - - so Cal o ,Y V) - - o 8 V) Ct 3 .2~4 o ;: .O o V] ~ go A- ~ .= 0 ~ ~ ~ lo: via au - i. ID cat be ~ _ c O O ~ c: O ~= _e ~ tD ao ~ 3 ~D Ct Ct bO .e Ct .5: Ct ~: o o

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture 65 rats were used. The rats were assigned to four groups: (~) untreated, (2) injected with estradio' twice weekly and given fluoridated water (50 mg/~), (3) injected with estradio! twice weekly and given nonfluoridated water, and (4) given fluoridated water without estradio] injections. The animals were killed after 4 weeks. The fluoride concentrations of femur ash in the untreated and estradio} groups were similar at about 300 ppm, and those of the fluoride-treated groups were 640 ppm. Compared with the untreated group, the average breaking strength (the force required to cause fracture), measured by bending, of the femurs was 22% higher in the estradiol-only group and about 25% lower in the fluoride-only and estradiol-fluoride groups. In his review article of the effects of fluoride in bone, Faccini (1969) commented that a significant reduction in breaking strength of femurs occurred in rabbits that had received fluoride at 200 mg/L of drinking water for ~ weeks. No other details were provided. Wolinsky et al. (1972) provided weanling male rats with food (fluoride at 5 mg/kg per day) and water containing no fluoride or fluoride at 200 mg/L per day for 2 weeks. The femur-ash fluoride concentrations were 133 ppm and 7,398 ppm, respectively. Femur strength was measured by bending and was found to be 38% lower in the fluoricte-supplemented group. Beary (1969) studied 25-day-old Sprague-Dawley rats that were given diets containing adequate (0.6%) or deficient (0.! %) amounts of calcium and drinking water that contained fluoride at 0, 3.4, 10, or 45 mg/~. After 15-16 weeks, femur strength was measured with the 3-point bend- ing test. In the groups receiving adequate calcium, bone strength was slightly but not significantly lower in the fluoride-treated groups than in the untreated group, and a dose-response relation was not seen. Femur- shaft fluoride concentrations (calculated for ash content) ranged from 170 ppm in the control group to 5,000 ppm in the 45-mg/L~ group. In the calcium-deficient groups, fluoride-treated bone strength was reduced in a dose-dependent manner. Their bone fluoride concentrations were about twice as high as those in the groups receiving adequate calcium. Riggins et al. (1974) used an experimental design similar to that of Beary's (1969), except that (~) the diet adequate in calcium contained I. ~ % calcium, (2) drinking-water fluoride concentrations were 0, 50, and 100 mg/L, (3) fluoride exposure was 3 months, and (4) femur strength was measured by using fresh bones rapidly loaded in torsion, which was

OCR for page 51
66 Health Effects of Ingested Fluoride said to be "more analogous to the situation in which most long-bone fractures occur." Among the groups fed the 0.1%-calcium diet, bone strength was not affected in those receiving fluoride at 50 mg/L, but it was reduced by 2X% in those at 100 mg/~. Tibia-ash fluoride concentra- tions in the 50- and lOO-mg/L groups were 2,982 ppm and 7,393 ppm, respectively. In the groups fed the ~ . ~ %-calcium diet, bone strength was not affected significantly by administration of fluoride. Tibia-ash fluoride concentrations in the control, 50-mg/L, and lOO-mg/L groups were about 100, 1,580, and 5,200 ppm, respectively. Thus, as noted by Beary (1969), bone fluoride concentrations were significantly higher and break- ing strengths were significantly lower in the calcium-def~cient groups. Turner et al. (1992) measured femur strength with the 3-point bending test and reported a biphasic response to the concentration of fluoride intake. Nine groups of weanling rats received a low-fluoricle diet (less than 2 mg/kg per day) and water containing fluoride at 0-128 mg/L for 16 weeks. For reasons that are not clear, the range of bone fluoride concentrations (vertebral ash) in all nine groups was unusually high. As a result, average values were difficult to `determine accurately. Neverthe- less, average values were less than 1,000 ppm for the groups receiving fluoride at 0-8 mg/L (bone fluoride at about 100 ppm), and there was no difference in bone strength among them. The 16-mg/L group had an average bone fluoride concentration of about 1,500 ppm and an increase in strength that was significantly higher than that in the 0-, I-, 8-, 64-, and 128-mg/L groups. The average femur fluoride concentrations of the 64- and 128-mg/L groups were about 5,000 and ~ 1,000 ppm, respective- ly. The bone strengths of these groups were lower than those of all other groups, but the differences were statistically significant only when com- pared with the 16-mg/L group. Because the difference between bone strength in the 64- and 128-mg/L groups and the I- and 8-mg/L groups was not statistically significant, however, the authors' interpretation of a "biphasic response" appears tenuous. Rich and Feist (1970) provided pregnant rats with a low-fluoride diet during the last 5 days before delivery and during the nursing period in an effort to reduce the bone fluoride concentrations of the offspring. Sixty female weanling pups were then assigned to five groups that receiver! drinking water containing fluoride at 0, 5, IS, 30, or 70 mg/L for 16 weeks. The breaking strengths of femurs were determined with a 3-point bending test, and the fluoride concentrations of the second lumbar verte

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture 67 brae were determined with the ion-specific electrode. The breaking strength of the femurs in the 30-mg/L group was 26% higher (p < 0.01) than the breaking strength of the control group. None of the other groups differed from the control group in breaking strength, but as determined by linear regression analysis, there was a positive relation ~ ~ 0.01) between bone fluoride concentration and breaking strength in the 0-, 5-, 15- and 30-mg/L groups. The strength of the regression was not affected significantly by differences in body weight, bone weight, bone length, bone diameter, or cortical thickness. The bone fluoride con- centrations ranged from 75 to 5,400 ppm in those groups. Although the bone fluoride concentration (calculated in terms of ash) of the 70-mg/L group was 10,580 ppm, the breaking strength was not significantly different from that of the O-mg/L group (14.2 vs. 13. ~ kg, respectively). Rosen et al. (1975) notes! an increase in vertebral bone strength as- sociated with increased fluoride intake in rats with disuse osteoporosis. Their study involved adult male Sabra-strain rats. The tails of one-half of the rats were surgically immobilized to induce disuse osteoporosis. The four subgroups received distilled drinking water, fluoride at 50 mg/L of drinking water, distilled drinking water and supplemental calcium and vitamin D, and fluoride at 50 mg/L of drinking water for 30 days fol- lowec! by distilled drinking water and supplemental calcium and vitamin D for 30 clays. The study lasted 60 days. The compressive strength of the tad] vertebrae did not differ among the groups whose tails were not immobilized. In the groups with disuse osteoporosis, bone strength was increaseci by about 30% with fluoride administration ant] also by 15% with supplemental calcium and vitamin D administration. The average bone-ash fluoride concentration of the groups that did not receive sup- plemental fluoride was 440 ppm, and the concentrations of the 50-mg/L groups were 2,090 ppm (60 days on fluoridated water) and 1,324 ppm (30 days on fluoridated water). Studies showing no effect of fluoride administration on bone strength have involved rats (Seville, 1967; Naylor ant] Wilson, 1967; Kuo and Wuthier, 1975; Einhorn et al., 1992), dogs (Romanus, 1974), guinea pigs (Sharma et al., 1977), and cows (Rahn et al., 1991~. Kuo and Wuthier (1975) used weanling male Sprague-Dawley rats in their study to deter- mine the possible preventive effects of fluoride in cliet-induceci osteoporo- sis. The rats were fed a nutritionally adequate diet with or without fluoride at 50 mg/L of drinking water for 15 weeks, at which time 6-7

OCR for page 51
68 Health Effects of Ingested Fluoride rats were killed to provide baseline data. One-half of the remaining rats in each group continued to receive the nutritionally adequate diet and the others were given a low-calcium (0.014%) and phosphorus (0.018%) diet. All rats were then given distilled water without fluoride to drink until Hey were killed 4, 8, or 12 weeks later. The baseline (15-week) fluoride concentrations of femur ash were 4,000-5,000 ppm in the 50- mg/L group and 50~0 ppm in the control group. There was no dif- ference between the 0- and 50-mg/L groups in bone strength at that time. The concentrations were said to "gradually increase" thereafter but at a "significantly greater rate" in rats receiving the low-calcium, phosphorus diet. This diet caused a progressive loss of bone mass and an important decrease in mid-shaft femur strength. Fluoride administration, however, had no effect on these changes. Saville (1967) assigned 25~ay-old Charles River CD-strain male rats to four groups that received fluoride at 0, 2, 5, or 20 mg/L of drinking water. The rats were killed at selected times up to 9X days. The com- pressive strength of the femurs and humeri was determined and found to be a linear function of body weight but did not differ among the groups. Naylor and Wilson (1967) performed a similar study with weanling albino rats but with higher drinking-water fluoride concentrations (0, 10, 25, 100, or 250 mg/~) for up to 52 weeks. No differences were found among the groups in radiographic appearance, breaking strength, defle- xion pattern on bending, or ash content of femurs. These two reports did not contain bone fluoride concentrations. Romanus (1974) examined physical properties and chemical composi- tion of canine (Beagles) femoral cortical bone strips after nutritional osteopenia was induced by feeding a low-calcium, high-phosphorus diet for 41 weeks. Fluoride was added to the diet at 0, I, 3, 9, or 27 mg/kg. The same analyses done on bone samples obtained from dogs were continued in the study to test for "reversibility." This involved feeding a diet enriched in calcium and phosphorus for up to 28 additional weeks. The physical properties of bone were determined with 3- and 4-point bending tests and application of tension. Dietary fluoride supplementa- tion had no effect on the composition of bone (except for a direct correla- tion with bone fluoride concentrations). Similarly, bone strength was not affected significantly by administration of fluoride. The author offered a concluding comment: "A better source for studies of bone is bone from individuals with skeletal disease, as nutritional and disuse osteopenia does

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture 69 not seem to change the physical properties of the bone material to any major extent." Sharma et al. (1977) assigned adult male Hartley guinea pigs to four groups that were given distilled drinking water containing fluoride at 0, 2, 10, or 20 mg/~. The Purina diet was stated to contain fluoride at less than 17 mg/kg. This diet was a "maintenance" diet, so that the bioavail- ability of fluoride would have been approximately 50% (Whitford, 1991~. Thus, there was no true low-fluoride group. The animals were killed at 15, 27, or 42 weeks. The physical properties of selected long bones were determined with tension, bending, and torsion tests. The fluoride concentrations of rib, radius, and ulna were directly related to time of fluoride exposure and to concentration of fluoride exposure. After 42 weeks, the fluoride concentrations of the bones (dry) in the 20-mg/L group were 2,565, 1,990 and I,X55 ppm, which would correspond to an average concentration of 3,560 ppm in bone ash. There were no statisti- cally significant differences among the groups with respect to tension strength (fracture upon pulling), torsion modulus (fracture upon twisting), or modulus of elasticity (fracture upon bending). Einhorn et al. (1992) provided weanling female rats with a main- tenance diet containing fluoride at 25 mg/kg and drinking water with fluoride at 0, ~ I, 23, or 34 mg/L for 86 days. As in several of the other studies, this study did not have a low-fluoride group. The average femur-ash fluoride concentrations for these groups were 2,017, 6,102, 9,097 and Il,688 ppm, respectively. The torsional strength of the femurs did not differ among the groups. In agreement with Riggins et al. (1974), it was stated that testing animal bones in torsion is more akin to the stress pattern most often encountered by human patients. Einhorn et al. also examined bone with histomorphometric techniques and found no significant intergroup differences. They concluded that, in the ab- sence of effects on bone mass, remodeling, or formation rate, incorpora- tion of even very high concentrations of fluoride does not significantly alter the capacity of bone to withstand mechanical loads. Rahn et al. (1991) assigned 17 Holstein heifer calves 5~ months old to three groups fed a diet containing fluoride at 0, 30, or 50 mg/kg. The fluoride concentration of the control diet was not stated but must have been fairly low because of the bone fluoride concentrations. The ex- perimental groups received fluoride at 0.5-~.2 mg/kg per day or 0.~-~.8 mg/kg per day. After 6 years, the metacarpal bones were analyzed for

OCR for page 51
70 Health Effects of Ingested Fluoride fluoride, density, and compressive strength. The bone-ash fluoride concentrations were 595, 2,664, and 4,500 ppm in the 0-, 30-, and 50- mg/l~ groups, respectively. The relation between bone density and fluoride concentration appeared to be inverse, but the trend was not statistically significant. Compressive strength clid not differ among the groups. Discussion The results of the laboratory animal studies designed to determine the effects of fluoride administration on bone strength have yielded all possible outcomes. Although most of the reports indicate little or no effect even with extremely high fluoride intake and bone concentrations, some have shown positive, negative, or biphasic effects. The explana tions for these discrepant results are not apparent. Several potential or real problems in experimental design, however, have been identified. Insofar as can be determined, only three of the studies used a diet with a reasonably low fluoride concentration (Rich and Feist, 1970; Rahn et al., 1991; Turner et al., 1992) and most used diets with frankly high concentrations. Consequently, the control groups in the latter studies often had high concentrations of bone fluoride, which could have ob- scured positive or negative effects on bone strength in the treatment groups. That is to say, when diets with high fluoride concentrations are used and bone fluoride concentrations in the control groups are clearly high, no true low-fluoride group exists that can be used to judge the effects of fluoride in the treatment groups. In the three studies that used low-fluoride diets, the investigators found that bone strength was not affected or was actually increased in one or more of the treatment groups. Another problem in some of the reports is the method used for deter- mining bone fluoride. For example, Beary (1969) said that the solution of bone mineral to be analyzed was adjusted to a pH of "2.07 + 0.03 since this is considered a desirable range for reading fluoride ion ac- tivity" with the fluoride electrode. The pH range actually recommended is 5-6. This range is well above the pK of hydrogen fluoride (3.4), which is not "seen" by the electrode but is sufficiently low so that hy- droxy} ions do not interfere (the electrode recognizes both fluoride ant]

OCR for page 51
Fluoride Exposure and Risk of Bone Fracture 71 hydroxy! ions). At a pH of 2.07, only 5% of the fluoride in solution is detectable by the electrode. Other investigators dissolved the bone samples in strong acid while the vessels were open to the air. The boiling point of hydrofluoric acid is 19C, so fluoride loss from the sample might have occurred. The hexamethyIdisiloxane-facilitated diffusion method of Taves (1968), as modified by Whitford (1989), is a preparative technique that has been shown to be reliable and accurate. The subcommittee concludes that the weight of evidence currently available indicates that bone strength in animals fed a nutritionally ade- quate diet is not adversely affected unless chronic exposure to fluoride is at least 50 mg/kg in diet or 50 mg/L in water. These data indicate that the current EPA guidelines of fluoride at 4 mg/L of drinking water for humans are appropriate. Recent reports from epidemiological studies of human populations have provided conflicting evidence on this subject, however, ant] indicate the need for additional research. One uncertainty in all the studies is the appropriateness of the methods used to cause bone fractures. Most investigators have used a bending test, although torsion, tension, or compression tests have also been used. Compression tests would be appropriate for vertebrae but rarely for long bones. Pure tension stress would almost never be involved in the frac- ture of any human bone. Bending has a torsional component, but forces are applied to specific points that might not reflect the overall strength of the bone. Torsion gives a uniform strain field that might yield a more realistic estimate of the load required to cause fracture. According to Riggins et al. (1974) and Einhorn et al. (1992), torsion or twisting is most analogous to the stress experienced by humans who have fractures of long bones. To resolve the uncertainties that surround this important area of investigation, the subcommittee recommends that a workshop be con- ducted to evaluate the advantages and disadvantages of the various doses, treatments, laboratory animal models, weight-bearing versus non-weight- bearing bones, and testing methods for bone strength that can be used to determine the effects of fluoride on bone.

OCR for page 51