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.
- ~ 1~ Ad Case Studies This chapter describes how the evidence base underpinning the selected nu- trient-disease relationships has changed over the past decade, as expressed in the report, Diet and Health: Implications for Reducing Chronic Disease Risk (D&H) (NRC, 1989) and the Dietary Reference Intake (DRI) reports published by March 2001 (IOM, 1997, 1998, 2000a, 2001~. Following the categorization scheme developed by the committee, the case studies are organized according to whether over time the confidence level of the nutrient-disease relationship was increased, decreased, unchanged, or only evaluated in a DRI report (i.e., not mentioned in the D&H report). For each case study, the types of evidence cited in the relevant report and the reports conclusions are summarized. Each case study concludes with a discussion of the evolution of evidence. Box 2-1 shows how the case studies are ordered. INCREASED CONFIDENCE IN RELATIONSHIP Fluoride and Dental Caries Diet and Health Report The D&H report summarizes the results of epidemiological studies that began in the 1930s and demonstrated an inverse relationship between the preva- lence of dental caries and the fluoride content of water. One paper cited (Dean et al., 1942) reported the incidence of dental caries in children aged 12 to 14 years. 19
20 EVOLUTION OF EVIDENCE The 7,257 children included were lifetime residents of 21 cities with the fluoride content of the public water supply ranging from a not detectable level to 2.6 ppm. The incidence of caries declined markedly as fluoride content increased, up to 0.5 ppm, and declined more slowly above that level. Two papers cited reported the results of additional fluoride on the incidence of decayed, missing, and filled teeth. One report (Driscoll et al., 1981) was a 9.5-year clinical trial. Fluoride tablets were given to first and second grade chil- dren in nine schools once or twice a day for 6 years. A control group received a placebo. The tablets were discontinued when two of the communities fluoridated their water, but children were evaluated after 9.5 years. The incidence of caries decreased during the 6 years of fluoride supplements, and the protective effect continued after the fluoride was discontinued. Another study (Lemke et al., 1970) evaluated the effect of fluoridation of the water supply in one city for over 11 years, and then again 4 years after discontinuation of fluoridation. Children were examined before and after the discontinuation of fluoride. The incidence of dental caries was 50 to 60 percent lower after water was fluoridated than before, but increased to prefluoridation levels when fluoridation was discontinued.
CASE STUDIES 21 Other studies (Anonymous, 1987; Stamm and Banting, 1980) evaluated the effects of differing levels of fluoride in the water supply on root surface caries in adults. Two of the studies compared the incidence of root surface caries in life- time residents of two communities that had different levels of fluoride in the water (Anonymous, 1987~. Both found a lower incidence of caries with the higher level of fluoride. Increased fluoride concentration in the cementum was also observed with increased fluoride in the water. A large trial in which water was fluoridated in three communities (Grand Rapids, Michigan; Newburgh, New York; and Brantford, Ontario) found a 50 to 60 percent reduction in caries prevalence, and no major adverse effects were noted in residents of any age (McClure, 1970~. The results of these studies provide conclusive evidence that fluoridation of the water supply or supplemental fluoride reduces dental caries, and the D&H report stated that "of all dietary components exhibiting a protective effect against caries, the most effective is fluoride" (p. 640~. Dietary Reference Intake Report The DRI report, Dietary Reference Intakes for Calcium, Phosphorus, Mag- nesium, Vitamin D, and Fluoride (IOM, 1997), used the evidence discussed in the D&H report on the cariostatic effect of fluoride as an indicator for an Ade- quate Intake (AI) for fluoride. An AI, one of the DRI reference values, is a rec- ommended intake value based on observed or experimentally determined ap- proximation or estimate of nutrient intake by a group (or groups) of healthy people. An AI is used when a Recommended Dietary Allowance cannot be de- termined (IOM, 2001~. The relationship between dental caries, fluorosis, and fluoride concentration in drinking water, based on earlier work by Dean (Dean, 1942; Dean et al., 1942), was examined. In addition, the DRI report cited more recent evidence from observational and clinical interventions indicating that pre- and posterup- tive exposures to fluoride have cariostatic effects (Hargreaves, 1992), and that the best results are achieved when fluoride is consumed beginning at birth (Groeneveld et al., 1990~. Additional information was also obtained after the D&H report was published on physiological activity by which fluoride produces its cariostatic effect (Marquis, 1995; Whitford, 1996~. The DRI report con- cluded, based on several retrospective clinical studies, that "the earlier children are exposed to fluoridated water or dietary fluoride supplements, the greater the reduction in dental caries in both the primary and permanent teeth" (p. 299), and that "fluoridated drinking water increases resistance to dental caries at all ages" (p. 299~. The conclusions of the DRI report agreed with the conclusions of the D&H report and provided additional supporting evidence, demonstrating that exposure to fluoride at all ages prevents dental caries and that both pre- and pos- teruptive exposure to fluoride has cariostatic effects.
22 EVOLUTION OF EVIDENCE However, studies of the effect of prenatal fluoride supplements produced conflicting results. A prospective study did not support the hypothesis that pre- natal fluoride exposure reduces dental caries (Leverett et al., 19973. These data led to the conclusion that "scientific evidence is insufficient to support a rec- ommendation for prenatal fluoride supplementation" (p. 304~. Evolution Accepted to Accepted + Large retrospective and intervention studies conducted before publication of the D&H report provided conclusive evidence that fluoridated water and dietary fluoride supplements reduce dental caries. The science in this area had already evolved to the point that the level of fluoride required to reduce caries, but not result in fluorosis, was established. Exploration of the mechanisms of the cario- static action of fluoride began before the D&H report was prepared and has ad- vanced since it was published. The research results presented in the D&H report were used to establish an AI for fluoride in the DRI report. Studies conducted after the D&H report was published led to the conclusion that there is no evi- dence to support the need for additional fluoride during pregnancy. Additional studies conducted after the D&H report and considered in the DRI report pro- vided evidence that fluoridated water, dietary supplements, and topical applica- tion of fluoride reduce dental caries at all ages. Calcium and Bone Status This section discusses dietary calcium in relation to bone status and osteo- porosis and in relation to fracture risk. Diet and Health Report Osteoporosis. In the D&H report three major lines of evidence were re- viewed with respect to the association between calcium intake, bone mass accre- tion or maintenance, and osteoporosis: (1) the pathophysiological relationships among dietary calcium, intestinal absorption, and bone mass, (2) epidemiologi- cal studies, and (3) clinical studies of calcium supplementation. In the first case, achievement of optimal peak bone mass and minimizing bone loss in later life were identified as important factors that could modify risk of osteoporosis. However, experimental evidence for the amount of calcium intake and retention needed to support optimal bone gain and minimal bone loss was unknown. It was concluded that the efficiency of calcium absorption declines in the elderly, but the quantitative contribution of this decline to aging-related bone loss and increase in incidence of fractures in the elderly remained undetermined. The D&H report reviewed evidence from epidemiological studies that cen- tered on the relationship between dietary calcium intake (either lifetime by diet
CASE STUDIES 23 history, recent intake by food frequency, or measured intakes) and measures of bone mineral density (BMD) or osteoporosis by one of a variety of techniques. The techniques included single and dual photon absorptiometry, radiology, and computed tomography. The findings did not consistently support a relationship between calcium intake and bone mass or rate of bone loss if measured in women after menopause. In observational studies reporting that people with diagnosed osteoporosis consumed less calcium than age-matched controls, the intakes were usually less than 800 mg of calcium/day. Interpretation of the find- ings was complicated by a lack of information about activity of the subjects and with respect to consumption of dietary compounds known to interfere with cal- cium at the absorptive or excretion levels. Evidence from clinical trials, including both nonrandomized and random- ized designs, was summarized in the report as, "the long-term effects of calcium supplementation on bone mass are not yet established" (p. 352~. Nearly all stud- ies were conducted in postmenopausal women for whom the primary outcome was bone loss as a function of intervention with calcium (in some studies with estrogen replacement therapy compared to estrogen alone, or in addition to vi- tamin D or fluoride). Most of the studies reviewed demonstrated that the higher calcium intakes reduced the amount of bone loss, although some found no ef- fect. The studies were limited in that they were relatively short in duration, the methods of measurement of bone mass and site of measurement were highly variable, and the populations in the different studies varied in age and osteopo- rosis status. None of the studies was designed to demonstrate an association be- tween lifetime intake of calcium and risk of developing osteoporosis. The report identified postmenopausal women and elderly persons as target groups for interventions with calcium (and vitamin D) since it was well estab- lished that production of the active form of vitamin D t1-25-dihydroxyvitamin D] declines with age, and efficiency of calcium absorption is compromised in later life. However, the report concluded that "There is no direct evidence that the impaired intestinal calcium absorption observed during menopause and ag- ing can be overcome by increased calcium intake. Moreover, the evidence that calcium supplementation prevents the trabecular bone loss associated with the menopause is, at best, weak" (p. 360~. And further, "calcium supplementation should therefore not be used as a substitute for sex hormone replacement, which prevents postmenopausal bone loss in most patients" (p. 617~. Bone Fracture. Information about an association between calcium intake and fracture risk was limited at the time of the D&H report. Only one study (Riggs et al., 1982) was cited in the D&H report, and it was not a randomized controlled trial (RCT). All participants were postmenopausal women who had generalized osteopenia and one or more nontraumatic vertebral fractures. The calcium intervention of 1,500 to 2,000 mg/day was combined with high, inter- mittent amounts of vitamin D of 1,250 fig (50,000 IU) once or twice a week.
24 EVOLUTION OF EVIDENCE Fracture risk was reduced by 50 percent with higher calcium and vitamin in- takes. Thus, preliminary evidence was provided for reduction of risk of bone fracture in older women with established osteoporosis on a combination of these therapies, but it was impossible to identify a separate role for calcium alone. Dietary Reference Intake Report Osteoporosis. Calcium was discussed in the report, Dietary Reference In- takes for Calcium, Phosphorous, Magnesium, Vitamin D, and Fluoride (IOM, 1997~. The basis for derivation of the DRI for calcium varied somewhat with age group. It reflected a combination of data from classic metabolic studies of calcium balance (chosen after rigorous review to meet specific criteria) and clinical trials. In the clinical trials, response to an intervention of two or more levels of calcium given for at least 2 years was assessed by bone mineral content (BMC) or BMD at one or more bone sites. Such measures of bone mass are known to be strong predictors of fracture risk (Cummings et al., 1993~. An example follows that uses a combination of balance studies and trials measuring bone mass to derive an AI for calcium for the age group of 51 to 70 years. Results of balance studies showed that calcium intakes between approxi- mately 1,100 and 1,500 mg were associated with higher calcium retention. However, lack of appropriate balance data in women precluded calculation of a plateau in calcium retention. Data were also reviewed from randomized trials of calcium intervention that measured bone mass as an outcome (Table 4-1 of the DRI report, reprinted in Appendix A) and assessed according to bone site meas- ured and time after menopause. This analysis indicated that the effectiveness of calcium intervention varies by bone site measured, menopausal stage, and usual intakes of the participants. Studies in which calcium intake was increased above 750 mg/day (Reid et al., 1995), 800 mg/day (Prince et al., 1995), or 1,000 mg/day (Rids et al., 1987) showed a reduction in loss of bone mineral from corti- cal-rich skeletal sites in postmenopausal women. Taking the two approaches together, an AI of 1,200 mg was chosen. For persons over 70 years of age, there were insufficient data to use balance studies as a predictor of calcium needs. Thus, the report relied on data from four randomized longitudinal studies published during the 1990s (Chapuy et al., 1992; Chevalley et al., 1994; Dawson-Hughes et al., 1997; Recker et al., 1996) in which BMC and BMD were outcome measures. In each of the four studies, supplemental calcium (with or without added vitamin D) resulted in less loss of bone than the lower calcium intake. Bone loss was measured by BMD at one or more skeletal sites including the proximal femur, femoral shaft, spine, forearm, and the total body. The information from randomized trials was more useful than cross- sectional studies because in the latter, "calcium intake is not accurately meas- ured, calcium intake at one point in time may not reflect lifetime calcium intake,
CASE STUDIES 25 and bone mass at a single point in time is the result of the lifelong influence of many confounding variables that are not measured" (p. 86~. Epidemiological data were also reviewed to assess possible associations between life-long dietary intakes of calcium and osteoporosis, but no direct evidence was available. The DRI report reviewed the literature on other dietary components (pro- tein, sodium, and caffeine) or lifestyle habits (physical activity) that affect the utilization of calcium, thus ultimately influencing its availability for deposition in bone. In all cases, any nutrient interaction identified was unlikely to have a significant impact on calcium homeostasis and require a calcium intake above what had been set based on the balance studies and bone mineral measures. Bone Fracture. With respect to bone fracture as the major outcome, several calcium intervention studies conducted during the 1990s in both institutionalized and noninstitutionalized persons revealed a linkage between calcium intake, reduced bone resumption, and a reduction in bone fractures. Of the four studies reviewed, two studies (Chapuy et al., 1992; Dawson-Hughes et al., 1997) were conducted in men and women with participants randomized to placebo or cal- cium supplement groups. Calcium supplements (500 to 1,200 mg/day) provided a total calcium intake of 1,200 mg/day or more for up to 3 years. In both studies, supplementation reduced bone loss moderately (for whole body bone this was significant up to 3 years), and a reduction of nonvertebral fracture rates oc- curred. In two other studies (Chevalley et al., 1994; Recker et al., 1996), calcium supplementation of 800 to 1,200 mg/day resulted in a reduction in vertebral frac- ture rate, although in one study in women with low habitual calcium intakes (Reeker et al., 1996), the reduction in vertebral fractures was noted only in those with a previous fracture but not first vertebral fractures. In conclusion, the DRI report cautioned "additional studies are needed to estimate the magnitude of the impact of calcium intake on fracture rates. Available data do not allow use of fracture outcomes to identify the AI for calcium" (p. 116~. Evolution Uncertain to Accepted The evidence compiled for the D&H report relied primarily on descriptive studies. Outcome measures consisted primarily of circulating bone-regulating hormones, such as vitamin D metabolites and parathyroid hormone, mineral balance, urinary excretion of calcium or phosphorus, or early generation bone densitometry. Studies reviewed in the DRI report consisted mostly of random- ized trials conducted over the past decade. The use of measures of bone mass by dual energy x-ray absorptiometry (DXA), only available since the early 1990s, allowed for a more quantifiable and bone-site specific outcome measure. This technique used a stable photon source of an x-ray tube rather than a radioactive source (such as iodine-125 or gadolinium), providing greater precision and better ability to detect real changes
26 EVOLUTION OF EVIDENCE or differences in BMC. The availability of DXA thus provided new knowledge about a functional measure of calcium intake as indicated by an accretion of bone mass during growth or prevention of bone loss in the elderly population, the age of achievement of peak bone mass (middle to late teen years as opposed to ages 25 to 30 as previously thought), and the peak bone mass velocity from which estimates of deposition of calcium during peak growth could be calcu- lated. Biochemical markers of bone turnover have also been proven to be sensi- tive measures of change in bone status. Many recent studies have assessed change in bone status in response to diet by using plasma osteocalcin to reflect bone formation and urinary pyridinoline, deoxypyridinoline, or plasma C- terminal telopeptide of type 1 collagen to reflect bone resorption. Statements on directions for research from the D&H report such as, "Long- term studies should be conducted to determine the effect of calcium supplemen- tation on the rate at which bone mass is lost in this age group" (in reference to people 65 years of age and older in whom intestinal absorption of calcium is decreased) (p. 622), appear to have set the stage for research in the past decade. Although the DRI report is cautious in its interpretation, data from clinical trials designed to study the effect of varying calcium intakes on bone mass were used along with data on calcium retention to establish the AI of calcium for adult populations. It is unlikely that evidence from epidemiological, ecological, or clinical tri- als will ever define a direct causal link between dietary calcium intake in early life and osteoporosis risk later in life. There are too many confounding variables, both dietary and lifestyle-related, that would obscure a direct link, and con- trolled trials of lifetime duration are not feasible and are extremely costly. Re- search over the past decade has provided fairly convincing evidence that bone mass influences fracture risk and calcium influences bone mass at any age. Par- ticularly revealing in this regard is that for persons over approximately 66 to 70 years of age, calcium intake not only reduces bone loss but also reduces risk of fracture. Application of this knowledge may have an enormous impact on health-care costs in this population. From the perspective of fracture prevention, information has evolved in line with the research agenda set out by the D&H report, specifically with reference to the elderly population. In the four well-designed studies reported over the past decade summarized in the DRI report, the interventions were calcium supple- ments of 500 to 1,200 mg/day (in addition to vitamin D supplements of 8.8 to 20 ~g/day [352 to 800 IU/day]) (p. 1 16), or calcium supplements alone given over 2 to 3 years. For the studies with men and women, the intervention of calcium and vitamin D led to a reduction in fracture rates. In two studies where the inter- vention was calcium alone and the population was only women, fracture rates were reduced in persons with prior or first-time vertebral fractures. Thus, al- though the DRI report indicated that the evidence was not sufficiently robust to use in setting an Estimated Average Requirement (EAR), the studies of the past
CASE STUDIES 27 decade have gone a long way to fulfilling the recommendation of the D&H re- port for additional research to better understand calcium and bone status. (An EAR, one of the DRI reference values, is a nutrient intake value that is estimated to meet the requirements of half the healthy individuals in a life stage and gen- der group [IOM, 2001~. Vitamin D and Bone Status Diet and Health Report Most of the evidence linking vitamin D and osteoporosis and bone fractures was indirect in the D&H report, emerging from observational studies on changes in vitamin D status and calcium absorption with aging. In older persons, circulating plasma 25-hydroxyvitamin D (and 1,25-dihydroxyvitamin D) were lower than in younger persons. Possible reasons suggested for this difference are less exposure to sunlight, a reduced capacity for de novo synthesis of the parent compound in the skin, and also perhaps a decline in the activation of the 1-oc- hydroxylase enzyme that increases synthesis of the active hormone in response to hypocalcemia via parathyroid hormone (PTH). These studies established a relationship between declining vitamin D status in the elderly and lower bone mass or greater incidence of hip fracture. In four studies (Caniggia et al., 1986; Crilly et al., 1981; Gallagher et al., 1982; Riggs et al., 1976) there was a positive relationship between vitamin D status (or intake of vitamin D or the active me- tabolite) and an increase in calcium absorption or bone mass or reduction in bone loss. In summary, the D&H report stated, "Numerous studies during the past two decades suggest that elderly people in the United States, Israel, Great Britain, and Europe are at increased risk for developing vitamin D deficiency" (p. 618~. It was speculated that vitamin D deficiency in this population may be a key factor in reduced calcium absorption leading to reduced bone mass. Dietary Reference Intake Report The observations of the prevalence of vitamin D deficiency in postmeno- pausal women and the elderly and the association with osteoporosis noted in the D&H report were further substantiated by studies conducted in the 1990s (Chapuy et al., 1992; Honkanen et al., 1990; McKenna, 1992; Pietschmann et al., 1990) and reported in Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D, and Fluoride (IOM, 1997~. These led to several ran- domized, double-blind clinical trials (either in women alone or in men and women) to investigate the effectiveness of vitamin D supplementation above that previously recommended to impede bone loss and reduce fracture risk. The response to a vitamin D supplement of 10 to 25 ,ug/day (400 to 1,000 IU/day) was maintenance of normal serum 25-hydroxyvitamin D, reduction in elevated
28 EVOL UTION OF EVIDENCE PTH, and reduced bone resorption as indicated by various urinary markers (Brazier et al., 1995; Chapuy et al., 1992; Fardellone et al., 1995; Kamel et al., 1996; Lips et al., 1988; Sebert et al., 1995; Sorva et al., 1991~. Only in one study (Lips et al., 1996) was the supplement of vitamin D associated with a significant gain in BMD at the femoral neck (but not other bone sites) in women. However, in a classic study by Chapuy and colleagues (1992) that was replicated by Daw- son-Hughes and colleagues (1997) in the United States, a vitamin D supplement of 20 ~g/day (800 IU/day) given with calcium for 18 months resulted in a sig- nificant increase in BMC and a reduction in vertebral and nonvertebral bone fractures. Evolution Uncertain to Accepted Based on the observational studies cited in the D&H report, it was predicted that older persons require higher intakes of vitamin D (or its metabolite, 1,25- dihydroxyvitamin D) to maintain normal vitamin D status, but no clinical trials of the effects of increased vitamin D intake on osteoporosis or fracture were available (or reviewed) for the report. The studies reviewed in the DRI report support the earlier statement about increased risk of vitamin D deficiency with aging, and led to an AI of 15 ,ug/day (600 IU/day) for men and women over 70 years of age. Of important functional significance, recently published clinical trials summarized in the DRI report that used DXA technology to track fracture occurrence and were of sufficient duration demonstrated that vitamin D intakes at or greater than the AI level are associated with a reduction in rate of nonver- tebral fracture. A persistent limiting factor to the studies on fracture outcome in the elderly is that both supplemental calcium and vitamin D were given. Thus, the relative contribution of each nutrient to the increase in bone loss and reduced fracture rate remains to be explored. In this respect, further research is required to delineate the distinct roles for each of the nutrients in fracture prevention. DECREASED CONFIDENCE IN RELATIONSHIP Q-Carotene and Lung Cancer Diet and Health Report At the time of the D&H report, the hypothesis that dietary Q-carotene could prevent lung cancer was promising. The D&H report cited 12 epidemiological studies, including both case-control and cohort studies, that found inverse asso- ciations between either dietary intake of Q-carotene or plasma or serum levels of Q-carotene and risk of lung cancer. While it was recognized that the estimates of intake were derived from measures of fruit and vegetable consumption, there were supportive data from animal studies indicating that p-carotene as a single
CASE STUDIES 29 agent could inhibit carcinogenesis. The animal studies reviewed were not of respiratory tract carcinogenesis, but rather prevention of skin carcinogenesis, where efficacy in skin carcinogenesis was observed in 4 of 5 animal studies cited. While a promising association was noted, the D&H report also noted sev- eral caveats about this relationship. The report noted that "Consumption of caro- tenoid-rich foods does not necessarily serve as a protective factor against lung cancer for persons who smoke. The magnitude of the relative risk . . . has not yet been well characterized" (p. 322~. The report also noted that "Evidence does not yet permit a conclusion that the association is with Q-carotene specifically rather than some other carotenoid" (p. 322~. Given this limitation, the report recom- mended that "consumption of the relevant foods not the putative protective components of those foodsshould be encouraged" (p. 488~. The report also stated that "Clinical trials to determine the effect of dietary p-carotene supple- ments on lung cancer are in progress, but results are not yet available" (p. 313~. Thus, the report was cautious and did not advocate supplemental p-carotene as a general approach for lung cancer prevention. Dietary Reference Intake Report In the years between publication of the D&H report and the DRI report Die- tary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Carotenoids (IOM, 2000a), considerable research evolution occurred. Three major clinical trials investigating supplemental Q-carotene (as a single agent or in combination with another nutrient) and cancer prevention were completed. Two of these trials targeted the primary prevention of lung cancer (ATBC, 1994; Omenn et al., 1996), and a third targeted the primary prevention of total cancers (Hennekens et al., 1996~. These trials failed to substantiate a possible preventive role for p- carotene (high-dose supplements) with regard to lung cancer, and contrary to expectations, lung cancer incidence was significantly increased rather than re- duced in two of the three trials. A possible modifying role for tobacco and/or asbestos exposure was observed. Q-carotene supplementation increased the risk of lung cancer only in persons who were currently smoking or who had signif~- cant prior exposure to asbestos in the two trials where an increased risk was noted. The DRI report stated, "These trials indicate a lack of evidence of overall benefit on total cancer or cardiovascular disease and possible harm in certain subgroups such as current smokers or asbestos-exposed subjects" (p. 371~. Although this indicates a failure to substantiate the hypothesis that Q- carotene might prevent lung cancer, several caveats were noted in the DRI re- port. Plasma concentrations achieved in the clinical trials where adverse effects on lung cancer were noted are outside the range achieved through dietary intake (p. 370~. Also, the trials were not designed to test the hypothesis that dietary p- carotene obtained from foods (fruits and vegetables) would reduce the risk of
30 EVOLUTION OF EVIDENCE lung cancer. This issue remains unresolved. On the basis of the evaluation of observational data, the report suggested that "3 to 6 mg/day of Q-carotene from food sources is prudent to maintain plasma p-carotene concentrations in the range associated with a lower risk of various chronic disease outcomes" (p. 353), a suggested increase in intake from current intake estimates. The report also noted that carotenoid-rich foods are complex mixtures and "it is not clear whether observed health benefits are due to carotenoids per se or to other sub- stances found in carotenoid-rich foods" (p. 351~. Evolution Promising to Uncertain (Diet) and Uncertain to No Relationship (Supplements) A possible role for supplemental p-carotene in lung cancer prevention has not been substantiated. Furthermore, the randomized clinical trial data indicated that high-dose supplements of a nutrient could have the opposite effect of that intended, particularly when the doses tested were widely variant from the doses where benefit is inferred in observational research. This case study also high- lights the notion that lifestyle factors influence whether a given nutrient inter- vention is beneficial or harmful. In this case, it appeared that cigarette smokers and asbestos-exposed individuals were harmed, but former smokers had no in- crease in risk of lung cancer with supplementation. These advances in under- standing were the result of time and a substantial financial investment in ran- domized clinical trials. A clearer understanding of the role of carotenoid-rich foods in the preven- tion of lung and other cancers has yet to emerge, despite some methodological advances. For example, since 1993 there has been a database of the carotenoid content of foods (Mangers et al., 1993~. Prior to this development, researchers were unable to evaluate the association of carotenoids other than p-carotene with risk of lung cancer or other cancers. As noted in the DRI report (p. 343), more recent observational studies of carotenoids and lung cancer risk indicate that p- carotene is not uniquely associated with lung cancer risk; protective associations have been noted for several other carotenoids, such as oc-carotene and lutein (Le Marchand et al., 1993; Ziegler et al., 1996~. The role of these and other carote- noids from foods in the risk of lung and other cancers remains uncertain. This case study also illustrates that the evidence base and evolution of un- derstanding the role of a nutrient in a food matrix and a nutrient provided as a supplement may differ. Because of the differences found in the confidence in the relationship between lung cancer and p-carotene from supplements and from diet, the two sources of intake are considered separately in the discussion of characteristics of evidence.
CASE STUDIES Diet and Health Report 31 Vitamin C and Gastric Cancer At the time of the D&H report, the hypothesis that vitamin C could prevent cancer was most promising for gastric cancer. With regard to this cancer site, the report discussed seven observational epidemiological studies that showed in- verse associations between consumption of vitamin C-rich foods or vitamin C status and gastric cancer risk (pp. 331-3324. The studies included case-control and cohort studies, with vitamin C exposures assessed by dietary intake esti- mates and by blood measures of ascorbate status. The observational epidemio- logical data were supplemented by nearly 20 studies of vitamin C in animal carcinogenesis models (none of gastric cancer), and a limited number of mecha- nistic studies demonstrating inhibition of nitrosation by vitamin C. The report also mentioned four studies on the use of vitamin C supplements in cancer pa- tients (p. 514), but did not draw on these reports because they were therapeutic studies rather than preventive investigations and therefore not in keeping with the preventive focus of the D&H report. The summary statement in the D&H report concluded, "Epidemiologic studies suggest that vitamin C-containing foods and possibly vitamin C itself either may protect against cancer or have no association with the disease. The strongest evidence for a protective effect seems to be for stomach cancer" (p. 341~. Also, recognizing that the epidemiological studies were based on vitamin C-containing foods (fruits and vegetables), the report recommended ". . . in considering appropriate preventive measures, con- sumption of the relevant foods not the putative protective components of those foods [vitamin C for stomach cancer]should be encouraged" (p. 488~. Dietary Reference Intake Report The DRI text on vitamin C and gastric cancer was limited in that the em- phasis of this report, Dietary Reference Intakes for Vitamin C, Vitamin E, Sele- nium, and Carotenoids (IOM, 2000a), was the identification of possible bio- markers and their modulation by vitamin C in order to establish a nutrient requirement for vitamin C. The observational epidemiological literature on vi- tamin C and gastric cancer was not discussed, but studies of vitamin C supple- mentation effects on biomarkers of gastric and bladder cancer were summarized in a table (Table 5-2 of the DRI report, reprinted in Appendix A). Three of four studies showed positive effects of vitamin C on biomarkers of gastric cancer. The summary noted that "Despite the epidemiological associations and the evi- dence that gastric juice vitamin C is protective against nitrosation and oxidant damage, the two vitamin C supplementation studies conducted to date have not shown a subsequent decrease in gastric cancer incidence. Although many of the . . . studies suggest a protective effect of vitamin C against specific cancers
32 EVOLUTION OF EVIDENCE by site, the data are not consistent or specific enough to estimate a vitamin C requirement based on cancer" (p. 125~. Evolution Promising to Uncertain A possible preventive role for vitamin C in gastric cancer has not been demonstrated, although the mechanistic basis for such a role has grown consid- erably. More specifically, an understanding of the role of Helicobacter pylori infection in the etiology of gastric cancer was not available at the time of the D&H report, whereas newer studies of the effects of vitamin C supplementation on biomarkers in patients with H. pylori infection are becoming available (Mannick et al., 1996~. However, 5.25 years of vitamin C supplementation (120 mg/day, in combination with molybdenum) to a large population of poorly nour- ished men and women at high risk of gastric cancer from Linxian County, China, did not reduce the incidence of gastric cancer (Blot et al., 1993~. The relative risk for the study was 1.10, with a 9S percent confidence interval of 0.92 to 1.30. Thus, there appeared to be "no demonstrable short-term benefit" (Blot et al., 1993~. While an understanding of gastric cancer etiology has evolved con- siderably over the past several years, including identification of a possible mechanistic basis for vitamin C preventive effects, large-scale randomized trials were unable to demonstrate this benefit. The negative trial results do not conclu- sively show that vitamin C is unrelated to gastric cancer risk, given the limita- tions of trials of chronic disease prevention (e.g., short duration of intervention, only one dose tested, or results may not be generalizable to other populations with different risk factor profiles). Vitamin E and Cancer Diet and Health Report At the time of the publication of the D&H report, animal studies and obser- vational cohort studies of vitamin E and cancer were available. The epidemiol- ogical studies mentioned in the D&H report used plasma or serum vitamin E levels to assess vitamin E status. While there was no consistent observed effect of only vitamin E on cancer risk, it was noted that "low serum levels of vitamin E and selenium may be related to increased risk of some cancers, such as breast and lung cancers" (p. 322~. With regard to the animal studies, the report noted that "the role of vitamin E in cancer inhibition is inconclusive at this time" (p. 320~.
CASE STUDIES Dietary Reference Intake Report 33 In the time that elapsed between the D&H report and DRI report, Dietary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Carotenoids (IOM, 2000a), several observational and interventional studies of relevance to vitamin E and cancer were published. The largest was a clinical trial of vitamin E for the primary prevention of lung cancer in heavy smoking men from Finland (ATBC, 1994~. No reduction in lung cancer risk was noted in men randomized to receive supplemental vitamin E. The relative risk was 0.98, with a 95 percent confi- dence interval of 0.86 to 1.12. At least five clinical trials with vitamin E to pre- vent the recurrence of colorectal adenomatous polyps were reported, but none found a benefit (pp. 219-220~. Thus, a possible role for vitamin E in the preven- tion of lung and colorectal cancers has not been substantiated, based on data from large, randomized clinical trials. However, a possible preventive role in prostate cancer was suggested in the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study (ATBC) and is discussed later in this chapter. Evolution Uncertain to No Relationship A possible role for vitamin E in lung cancer prevention has not been sub- stantiated. This is based largely upon a large clinical trial, the ATBC trial dis- cussed above. However, the data from existing trials cannot rule out a possible preventive effect of vitamin E for lung or other cancers in persons with low se- lenium status, as none of the trials involving vitamin E targeted populations with this status. Low selenium status is believed to lead to biochemical changes that increase risk of illness associated with other stresses (IOM, 2000a). UNCHANGED CONFIDENCE IN RELATIONSHIP Vitamin E and Coronary Heart Disease Diet and Health Report The D&H report mentions a possible association between large doses of vi- tamin E and a reduced risk of coronary heart disease (CHD), but does not high- light this association. The initial citation is a letter published in the periodical Nature in 1946 that provides a conclusion without actual results; the methods are scant, and it is unclear whether the study is a case series or clinical trial (Vogalsang and Shute, 1946~. Subsequently, the D&H report mentions several clinical trials of persons with angina that are nonconfirmatory (Anderson and Reid, 1974; Donegan et al., 1949; Makinson et al., 1948; Rinzler et al., 19509. The trials were typically small in size and of suboptimal design. A trial of per- sons with intermittent claudication was likewise inconclusive (Farrell, 1980~.
34 EVOLUTION OF EVIDENCE The D&H report then mentions a possibility that vitamin E could raise high- density lipoprotein cholesterol, based on observations in two uncontrolled trials (Barboriak et al., 1982; Hermann et al., 1979~. This finding was not confirmed in a small, randomized controlled trial (Stampfer et al., 1983~. The section con- cludes by mentioning that vitamin E might raise triglycerides and that vitamin E had no effect on atherosclerosis in monkeys (Hayes, 19749. Of the two studies suggesting that vitamin E might raise triglycerides, one was a case-control study (Farrell and Bieri, 1975), and the other a relatively large clinical trial (Tsai et al., 1978~. The relationship of vitamin E and CHD appears to be uncertain in the D&H report. Summary statements in the D&H report do not mention a possible pro- tective effect of vitamin E from either diet or supplements. Also, in contrast to recommendations for additional research on the effects of antioxidants (particu- larly Q-carotene) on cancer, the D&H report makes no recommendations for additional research on the effects of vitamin E on CHD. Overall, it appears that the role of vitamin E in preventing CHD was not considered a highly promising line of investigation. Dietary Reference Intake Report The DRI report, Dietary Reference Intakes for Vitamin C, Vitamin E, Sele- nium, and Carotenoids (IOM 2000a), differs substantially from the D&H report. The DRI report presents a large body of evidence, from a variety of study de- signs, on the potential impact of vitamin E on CHD. More importantly, the DRI report highlights the potential relevance of oxidized low density lipoproteins (LDL) in development of CHD and the role of vitamin E in inhibiting LDL oxi- dation. The DRI report cites Steinberg's seminal paper that states the basic hy- pothesis on oxidative modification of LDL (Steinberg et al., 1989~. Potential mechanisms are then summarized (Azzi et al., 1995; Devaraj et al., 1996; Freedman et al., 1996; Rota et al., 1998), followed by a brief overview of animal studies (Parker et al., 1995~. A large section then summarizes the results of four prospective observational studies (Table 6-3 of the DRI report, reprinted in Ap- pendix A). These studies document significant inverse relationships between vitamin E intake on subsequent CHD events (Knekt et al., 1994; Kushi et al., 1996; Rimm et al., 1993; Stampfer et al., 1993~. Two of the observational stud- ies suggest an inverse association between dietary intake and CHD (Knekt et al., 1994; Kushi et al., 1996), while the other two studies suggest an inverse rela- tionship between total (dietary and supplemental) vitamin E intake (Rimm et al., 1993) or just supplemental vitamin E intake (Stampfer et al., 1993) and subse- quent CHD. On the basis of the mechanistic studies, the animal studies, the pro- spective observational studies, and one clinical trial of high-dose vitamin E
CASE STUDIES 35 supplements (Stephens et al., 1996), the relationship between vitamin E intake and CHD appeared very promising in the 1990s. The results of clinical trials that tested the impact of vitamin E supplements on CHD events were also summarized in the DRI report. In 1994, the ATBC trial, which tested the impact of vitamin E on lung cancer in 29,133 Finnish male smokers, documented a nonsignificant impact of this supplement (50 mg/day of all rac-a-tocopherol acetate) on CHD events over 5 to 8 years of fol- low-up, yet a potential increase in hemorrhagic stroke (ATBC, 1994~. In 1996, the Cambridge Heart Antioxidant Study, which enrolled 2,002 persons with an- giographic evidence of CHD, reported a significant (77 percent) reduction in nonfatal myocardial infarction from vitamin E (400 or 800 IU [268 or 567 mg]/day of RRR-a-tocopherol) over a median follow-up period of 1.4 years, but no significant impact on fatal myocardial infarction or total mortality (Stephens et al., 1996~. In 1999, the GISSI-Prevenzione Trial documented no significant impact of vitamin E (300 mg/day of all rac-a-tocopherol) on CHD clinical events over 3.5 years in 11,324 survivors of a myocardial infarction (GISSI- Prevenzione Investigators, 1999~. In 2000, the Heart Outcomes Prevention Evaluation (HOPE) trial documented no significant effect of vitamin E (400 IU t268 mg]/day of RRR-a-tocopherol) on CHD events over 4.5 years of follow-up in 9,541 persons at high risk for CHD (HOPE Study Investigators, 2000~. The final trial cited (DeMaio et al., 1992) documented a nonsignificant reduction in the extent of coronary stenosis from 1,200 IU (804 mg)/day of all rac-a- tocopherol in 100 persons who had a percutaneous transluminal coronary angio- plasty. The discordance between the nonsignificant large-scale trials and the volu- minous and promising evidence from bench research and observational studies is unclear but might be related to the duration of therapy or the stage of disease. In each of the major trials, participants either had advanced CHD or were at high risk of CHD. Still, in view of the inconsistent data, particularly the results of two major trials (GISSI-Prevenzione Investigators, 1999; HOPE Study Investigators, 2000), the DRI report concludes "as of this date, there are insufficient data on which to base a recommendation of supplemental vitamin E as a heart disease preventative for the general population" (p. 217~. Evolution Uncertain to Promising to Uncertain The section on vitamin E and CHD in the D&H report is very brief, the evi- dence is at best inconclusive, and the topic is not highlighted as promising. One animal study is cited; no observational or mechanistic studies are cited. The summary statement did not mention the potential impact of vitamin E on CHD. This presentation of evidence suggests that at the time of the D&H report, the oxidative modifications hypothesis had not matured and certainly had not caught
36 EVOLUTION OF EVIDENCE the attention of the general public or a broad cadre of researchers. In other words, the relationship was uncertain. In contrast, the DRI report presents a qualitative overview of available evi- dence from a variety of categories. While the mechanistic and animal studies strongly support the oxidative modification hypothesis, it is likely that the pro- spective observational studies, published in the early 1990s, fueled interest in the potential impact of vitamin E supplements on CHD and converted the rela- tionship from uncertain to promising. In the cohort studies, the magnitude of potential risk reduction was substantial, suggesting that increased intake of vitamin E, particularly from supplements, might reduce the risk of CHD by more than 30 percent. Still, the GISSI and HOPE trials effectively ruled out a benefit of this magnitude, at least in the populations studied, namely, persons with pre-existing CHD or those at high risk for CHD. Hence, the overall pattern of evidence evolved from uncertain, as implied in the D&H report, to highly promising in the 1990s, and back to uncertain by the time the DRI report was published. Vitamin C and Colds Diet and Health Report Vitamin C and the common cold was not a major emphasis of the D&H re- port, and the number of citations was limited. The focus was on vitamin C sup- plements and not dietary components. One large and three small, double-blind, placebo-controlled studies of vitamin C from supplements were specifically re- ferred to in the report. One small, randomized clinical trial found a statistical difference between the placebo group and the vitamin C-supplemented group in the number of persons who remained free of illness throughout the study period, and a highly significant 30 percent decrease in the total days of disability (Anderson et al., 19729. A large, randomized clinical trial by this same group did not find any significant difference between the placebo groups and supple- mented groups with respect to the incidence of colds, but the high-dose treat- ment groups (4 or 8 g on the first day of illness) demonstrated a significant re- duction in the severity of the symptoms, similar to the results of their earlier study (Anderson et al., 1974~. A small, double-blind, placebo-controlled study of children in a boarding school reported no difference in the number of respiratory episodes between placebo and vitamin C-supplemented groups, but did report a significant reduc- tion in days of morbidity and a significant increase in the number of children with no sick days reported during the 14-week period in the vitamin C group (Coulehan et al., 1974~. Coulehan and colleagues also found that both boys and girls in the lower grades (6 to 10 years of age) had a significant decrease in sick
CASE STUDIES 37 days, while only the girls in the higher grades (10 to 15 years of age) showed this effect associated with vitamin C supplementation. Karlowski and colleagues (1975) conducted a small, double-blind study with 311 employees of the National Institutes of Health and concluded that vi- tamin C had "at best only a minor influence on the duration and severity of colds," and "the effects demonstrated might be explained equally well by a break in the double blind." The break in the double-blind study may have been due to the curiosity of the scientist participants. The D&H report summarizes the effect of vitamin C on colds by indicating that "Several studies . . . generally indicate that vitamin C taken even in gram quantities does not prevent colds and at best only reduces the frequency and severity of symptoms in cold sufferers" (p. 515~. Dietary Reference Intake Report The relationship of vitamin C to colds was also not a major focus in the DRI report, Dietary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Ca- rotenoids (IOM, 2000a). The report cited several small trials in which supple- mental vitamin C had some effect on colds. Of the ten articles cited, four were published before 1988 (Chalmers, 1975; Coulehan et al., 1976; Ludvigsson et al., 1977; Miller et al., 1977~. Of the other six, five were reviews (Hemila 1996, 1997; Hemila and Herman, 1995; Herbert, 1995; Jariwalla and Harakeh, 1996), and the last was a study that dealt primarily with the antihistaminic effect of vitamin C (Johnston et al., 1992~. The studies included one by Coulehan and colleagues (1976) that followed up their 1974 study (also cited in the D&H report) with a double-blind, placebo- controlled study (868 subjects). They gave 1 g/day of vitamin C, and reported that there was no difference in the number of people becoming ill, number of episodes, or mean illness duration between the supplement and placebo groups. They also reported that children with a high plasma vitamin C level had signifi- cantly longer illness days on average than those with low levels. Coulehan's group was one of the first to suggest that vitamin C may have an antihistaminic effect. Miller and colleagues (1977) carried out a study with 44 school-aged monozygotic twins (6 to 15 years of age) who received 0.5 to 1.0 g of vitamin C/day, depending on body size. They reported no significant overall treatment effect on cold symptoms, but the younger girls receiving treatment had signifi- cantly shorter and less severe illness episodes, as did the younger boys. Ludvigs- son et al. (1977) carried out a double-blind study with 8- and 9-year-old chil- dren, using 1 g/day of vitamin C. They reported a reduction in the duration and severity of colds. The incidence was either unaltered or even increased with high doses of vitamin C. They concluded that "vitamin C in large doses thus had no definitely proved effect against colds."
38 EVOL UTION OF EVIDENCE The reviews presented different evaluations of the role of vitamin C and colds, and were not used for the DRI conclusion. Because of the limitations and variations in the response of colds to vitamin C, the DRI report concluded that "the data are not consistent or specific enough to estimate the vitamin C re- quirement based on the common cold" (p. 127~. Evolution Uncertain to Uncertain The selection of the literature on vitamin C and colds in the D&H report was sparse. It did not include several references available at the time it was pub- lished that were used 1 1 years later in the DRI report. Because of the conflicting nature of the studies, it is doubtful that the D&H report would have been signifi- cantly altered by including these studies. The conclusion from the D&H report was that supplemental vitamin C could not prevent colds; the report was incon- clusive on the question of the frequency and severity of symptoms. Although the DRI report took a systematic approach, the committee concluded that the data were not consistent or specific enough to estimate the vitamin C requirement based on colds. Thus, the report did not discuss whether vitamin C may reduce the duration of colds. Folate and Cervical Dysplasia Diet and Health Report The D&H report cited data from one randomized, controlled intervention study, which suggested a possible benefit of supplemental folate on the progres- sion of cervical dysplasia. In this small-scale study (n = 47), conducted by But- terworth and colleagues (1982), the progression of cervical dysplasia in women treated with large (10 ma) daily doses of folate was compared with that of pla- cebo-treated subjects. Cervical dysplasia, which may or may not progress to cervical cancer, improved only in the subjects consuming folate. The D&H re- port indicated that the data from this study "suggest that oral folate supplements may prevent the progression of cervical dysplasia or promote regression to nor- malcy" (p. 338~. Since cervical dysplasia may spontaneously revert to normal or progress to cervical cancer, the statement in the D&H report about data from this small, randomized trial was cautious. Although there is no concluding statement in the report about cervical dysplasia or cervical cancer, the trial that was suggestive of an effect prompted additional studies related to this possible association. .
CASE STUDIES Dietary Reference Intake Report 39 The section in the DRI report, Dietary Reference Intakes for Thiamin, Ribo- flc~vin, Niacin, Vitamin B6, Folate, Vitamin By, Pantothenic Acid, Biotin, and Choline (IOM, 1998), related to folate and cancer begins with the statement, "Experimental data indicate that changes in folate status may influence the proc- ess of neoplastic changes in certain epithelial tissue: a negative change in folate status may stimulate carcinogenesis. It is unclear if supraphysiological doses obtained from supplements afford any protection" (p. 264~. The report presented a separate review of data relating folate intake and/or status with risk of cervical dysplasia and with risk of advanced cervical neoplasia. The findings of the in- tervention study by Butterworth and colleagues (1982) were also cited in the DRI report, which indicated that "the positive alteration in cytology may have been an attenuation of dysplasia or simply a reduction in megaloblastic cellular changes" (p. 2649. The DRI report also cited a subsequent larger (n = 235) ran- domized, controlled intervention trial conducted by the same research group that failed to reproduce the previous findings (Butterworth et al., 1992b). The hy- pothesis that poor folate status itself might not be carcinogenic but may exacer- bate an underlying predisposition to cancer was investigated by Butterworth and coworkers in a case-control study (n = 464) reported in 1992 and cited in the DRI report (Butterworth et al., 1992a). In this study, risk was five times greater for cervical dysplasia in patients with the human papilloma virus-16 (HPV-16) infection who also had somewhat lower (not clinically deficient) red blood cell folate concentrations. Since cervical infection with HPV-16 significantly in- creases the risk for cervical cancer, this study provides evidence that suboptimal body reserves of folate, as reflected in low red blood cell levels, influence a key risk factor associated with early cervical cancer. The mechanistic animal studies also cited in the DRI report provided sup- porting evidence for the role of impaired folate status on carcinogenesis. The DRI report cited two publications that resulted from a case-control study by Ziegler and coworkers (1990, 1991) of invasive cervical cancer with community controls, conducted in five areas of the United States, in which no association between risk of invasive cervical cancer and folate intake was found. An Austra- lian case-control study (Brock et al., 1988) that did not support an association between invasive cervical cancer and dietary factors including folate was also cited. This study was specifically designed to examine the hypothesis of a reti- nal/carotene protective effect. Folate was included only as an ancillary interest. The DRI report also cited two other case-control studies. One by Verreault and coworkers (1989) failed to find an association between folate intake and cervical cancer, and one conducted in Latin American countries by Potischman and co- workers (1991) failed to find an association between serum folate concentrations and invasive cervical cancer. Potischman and coworkers had previously reported that folate intake was not associated with risk for cervical cancer in the same
40 EVOLUTION OF EVIDENCE study population. The DRI report concluded, "the effect of folate status on car- cinogenesis in the cervix remains uncertain" (p. 265~. Evolution Uncertain to Uncertain The scientific evidence has evolved to include larger, controlled interven- tion trials that do not support an inverse association between supplementation with folate and precancerous cervical dysplasia. Studies cited in the DRI report do support the hypothesis that less than adequate folate status may increase the risk for cervical dysplasia when coexisting risk factors are present. However, data do not support an inverse relationship between cervical cancer and folate supplementation. Butterworth's small intervention trial (Butterworth et al., 1982) suggested that folate supplements might positively affect the progression of cervical dys- plasia. These researchers subsequently conducted a larger and more comprehen- sive study with two major phases. The first phase was a case-control study with 464 participants (Butterworth et al., 1992a). Approximately 50 percent of the subjects in the first phase also volunteered to participate in the second phase, which was a randomized, placebo-controlled, double-blind, clinical intervention trial (Butterworth et al., 1992b). The controlled intervention trial was designed to test the hypothesis that high-dose folate supplements will modify the course of cervical dysplasia and improve its cytological and histological manifestations. The findings from this second intervention trial reject that hypothesis. Possible explanations for the difference between the later trial (Butterworth et al., 1992b) and the earlier study (Butterworth et al., 1982) include a more adequate sample size, exclusion of persons classified as having atypia less than dysplasia, and a longer period of observation. The high rate of apparent regression of dysplasia in both the placebo-treated and folate-supplemented groups of the intervention trial may also have contributed to the lack of significant differences. Case-control study findings indicate that inadequate tissue folate concentra- tion (reflected by red cell folate content) enhances the effects of HPV-16 infec- tion. These data suggest that for less than optimal folate status to enhance cervi- cal carcinogenesis, concurrent factors must be present that predispose to . . carcmogenesls. The early report by Butterworth et al. (1982) that suggested folate supple- mentation could lead to regression of dysplastic cervical lesions also led to a series of case-control studies by other investigators to examine the relationship between folate intake and risk of invasive cervical cancer. Interpretation of these case-control studies, including several cited in the DRI report, was constrained by methodological weaknesses of the studies such as: (a) folate intake was as- sessed with food-frequency instruments that were not validated for folate intake, (b) there was a lack of stratification of subjects for known risk factors such as
CASE STUDIES 41 HPV infection, and (c) the subjects had advanced stages of neoplasia that may be unresponsive to folate. In summary, scientists have been unable to confirm the data cited in the D&H report that suggested folate supplementation may reduce the risk for cer- vical dysplasia. The scientific evidence has evolved to support the hypothesis that less than adequate folate status may increase the risk for cervical dysplasia when coexisting risk factors are present. In addition, research evidence suggests that once the cervical dysplasia has advanced to the neoplastic stage, the inverse association between folate status and disease risk is no longer detectable. Phosphorus and Bone Status Diet and Health Report The review of phosphorus in the D&H report centered on the importance of the ratio of dietary calcium to phosphorus (Ca:P ratio) in determining the effect of excessive dietary phosphorus on calcium absorption, excretion, and bone re- sorption. In rats, high phosphorus intake depressed circulating calcium thereby inducing a mild hyperparathyroidism (Draper et al., 1972~. At the time of the D&H report, the research on calcium and phosphorus interactions focused pri- marily on calcium balance measurements. Also, some biochemical evidence (hyperparathyroidism and/or increased urinary hydroxyproline) indicated that high phosphorus diets induced bone resorption in growing aged animals (Draper and Bell, 1979; Draper et al., 19721. As a result, the prevailing opinion was that the ratio of dietary calcium to phosphorus was critical to establishing the rec- ommended intake for phosphorus. In small, nonrandomized, descriptive studies in humans who served as their own controls (Bell et al., 1977; Spencer et al., 1978), varying high intakes of dietary phosphorus were provided as phosphate additives in foods. These intakes reduced calcium absorption (especially at low calcium intakes) and induced secondary hyperparathyroidism. Compensatory reduction in urinary calcium also occurred. In contrast, in a study with a very small sample size (n = 2 to 4 subjects per diet test group) involving prison in- mates and staff (Maim, 1953), addition of phosphate salts to the diet (for a total intake of 2,000 mg/day) with moderate calcium intake (500 to 600 mg/day) showed no overall detriment to calcium balances. Thus, while there was some evidence in humans of a negative interaction of high phosphorus intakes (up to 3,000 mg/d) on calcium homeostasis (Portale et al., 1984, 1986), there was no evidence of an effect of high dietary phosphorus on bone mass or as a factor in the etiology of osteoporosis. The D&H report concluded, "High-phosphorus diets may decrease calcium bioavailability, but they also reduce urinary calcium excretion and their influence on bone mass and the risk of osteoporosis is un- known" (p. 360~. In the directions for research included in the chapter on "Os- teoporosis," the effect of dietary phosphorus (alone or with protein and fiber) on
42 EVOLUTION OF EVIDENCE calcium economy was stated as a research need, especially for adolescents and the elderly who have greater calcium requirements. Dietary Reference Intake Report The DRI report, Dietary Reference Intakes for Calcium, Phosphorus, Mag- nesium, Vitamin D, and Fluoride (IOM, 1997), summarized the clinical trials of the past decade by stating, "In balance studies in human adults, Ca:P molar ra- tios ranging from 0.08:1 to 2.40:1 (a 30-fold range) had no effect on either cal- cium balance or calcium absorption. Thus, for the reasons cited, there is little or no evidence for relating the two nutrients, one to the other, during most of hu- man life" (p. 1549. Thus, for the first time in the history of setting nutrient-based dietary recommendations in the United States, the recommendation for dietary phosphorus intake was not tied to that of calcium. Rather, the DRI values for phosphorus for children and adolescents were based on tissue accretion applied to a factorial model. For adults, the DRI values were based on the physiological response of maintenance of plasma phosphorus at the bottom end of the normal range (0.87 mmol/L t2.7 mg/dl]) as an indicator to reflect adequacy of phospho- rus intake. No information on a possible relationship between phosphorus intake and bone mass was included. Evolution Uncertain to Uncertain The shift in perspective about the importance of the ratio of dietary calcium to phosphorus may be credited to the many balance studies in humans that did not demonstrate abnormalities in calcium homeostasis imposed by high dietary phosphorus. This led to a search for an alternate approach to establishing a nu- trient-based recommendation for phosphorus as described above. With a better understanding of the endocrine regulation at the intestinal mucosa and renal lev- els that maintain normal plasma phosphorus, there is more confidence that the human body adapts to phosphorus intakes over a relatively wide range. To date, no attempts have been made to relate phosphorus intake to outcomes of bone status or risk of osteoporosis in humans. The importance of the ratio of dietary calcium to phosphorus in humans may need to be reevaluated if concerns about a population-level increase in phosphorus intake through such sources as cola beverages and food phosphate additives are realized, a point noted in the DRI report (IOM 1997~. The studies in animals and humans suggest that in the situation when diets of very low cal- cium and very high phosphorus are consumed over a long period of time or when renal function is not normal, the calcium economy of the body might be compromised. Target populations for such a concern are children and adoles- cents who substitute cola beverages (which have a high phosphoric acid content) for milk (which has a high calcium content). The issue of the influence of high
CASE STUDIES 43 dietary intake of phosphorus (especially in combination with low dietary cal- cium) on bone accretion in children or age-related bone loss later in life remains to be explored. Chromium and Diabetes Diet and Health Report In the D&H report, small intervention trials of 12 to 76 persons were re- ported. Two of the studies found a positive effect of chromium on insulin or glucose (Riales and Albrink, 1981; Simonoff, 1984) and two did not (Anderson, 1986; Rabinowitz et al., 1983~. Another found a relationship in a subset of the participants and no relationship in a larger group (Anderson, 1986~. The D&H report indicates that "no population data have been reported implicating chro- mium deficiency in humans with diabetes, and chromium supplementation does not improve blood glucose or insulin levels in those with the disease" (p. 628~. The report concludes, "the possible role of chromium deficiency in the etiology of diabetes is unresolved" (p. 630~. Dietary Reference Intake Report The DRI report, Dietary Reference Intakes for Vitamin A, Vitamin K, Arse- nic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc (IOM, 2001), cited a study conducted in China in which a positive effect of chromium supplementation was observed on some responses related to diabetes (Anderson et al., 1997), but dietary chromium in- take was not reported and other limitations of the study were noted. The DRI report recommended, "investigation of possible relationships between chromium status and insulin resistance, impaired glucose tolerance, and Type II diabetes" (p. 6-15~. The relationship remained inconclusive. Evolution Uncertain to Uncertain Conflicting results of studies related to chromium and diabetes led the D&H report to conclude that evidence for a relationship was inconclusive. The DRI report cited a study in China suggesting that chromium supplementation in Type II diabetics improved some aspects of glucose tolerance, but evidence from only one study was considered not sufficiently strong to be conclusive.
44 Diet and Health Report EVOL UTION OF EVIDENCE RELATIONSHIP DISCUSSED ONLY IN A DIETARY REFERENCE INTAKE REPORT Folate and Neural Tube Defects The topic of the prevention of birth defects, including neural tube defects (NTDs), was not included in the D&H report. Dietary Reference Intake Report The DRI report, Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B6' Folate, Vitamin By, Pantothenic Acid, Biotin) and Choline (IOM, 1998), made a definitive conclusion about an inverse association between sup- plemental folate intake taken during the periconceptional period and risk of NTDs. The conclusion stated in the DRI report to reduce the risk of neural tube defects was, "women capable of becoming pregnant consume 400 fig of folate daily from supplements, fortified foods, or both in addition to consuming food folate from a varied diet. At this time the evidence for a protective effect from folate supplements is much stronger than that for food folate" (p. 259~. The pri- mary studies on which this conclusion was based are summarized in the DRI report and include six observational studies and six controlled trials (Tables 8-7 and 8-8 of the DRI report, reprinted in Appendix A). The observational studies were published between 1988 and 1995 and with one exception found a signifi- cant reduction in risk for NTDs associated with periconceptional consumption of folate-containing multivitamins. The controlled trials, conducted between 1981 and 1992, provided the most definitive data on which subsequent public health policy was based. EvolutionAccepted in DRI The case study of folate and NTD risk reduction provides an excellent ex- ample of the evolution of scientific evidence from very preliminary suggestive observations to definitive data from randomized, controlled intervention trials. Hibbard (1964) first suggested that folate might be involved in the etiology of NTDs. Following publication of this hypothesis, two controlled trials were conducted. In the first study by Smithells and coworkers (1981), which was not randomized, women were given a mixture of eight vitamins including 360 ,ug/day of folic acid. The risk of recurrence in the supplemented group was about one-seventh that of the unsupplemented group. In the second study, which was a small randomized trial using 4 mg/day (4,000 Friday) of folio acid, a signifi- cantly lower recurrence rate was noted when the "non-compliers" were moved
CASE STUDIES 45 to the unsupplemented group (thus changing the protocol to nonrandomized) (Smithells et al., 1983~. Due to the nonrandomized design of these controlled trials, it was impossible to conclude that the difference in NTD rates between the groups receiving supplements and the unsupplemented groups was due to folate and not some other factor associated with the groups taking the supplements. Another major constraint of the Smithells studies was the fact that folate was provided as one component of a multivitamin supplement; therefore, it was not possible to definitively state that the risk reduction was due to folate alone. To avoid bias and address the lack of acceptance of these early controlled trials, a large Medical Research Council (MRC) study was begun in 1983 that involved an international, multicenter, double-blind randomized trial (MRC Vi- tamin Study Research Group, 1991~. The effect of supplementation was investi- gated by a factorial study design in which folate alone was compared with other vitamins alone, with both folate and other vitamins, and with neither (placebo). By 1991, the results were definitive enough to warrant stopping the trial- the evidence showed a 72 percent protective effect of folate alone. It was also con- cluded that the other vitamins showed no protective effect for NTD. A second significant trial from Hungary was a randomized controlled study that showed that the occurrence of NTDs could be significantly reduced by giving pericon- ceptional multivitamin supplements that included 800 fig of folio acid (Czeizel end Dudes, 1992~. Data from the early controlled trials did provide scientific evidence that periconceptional folate-containing multivitamin supplements reduced the inci- dence of NTDs. However the data were considered inconclusive by the scientific community (Smithells et al., 1983; Vergel et al., 19904. Due to the methodologi- cal constraints including nonrandomization and provision of folate as a compo- nent of a multivitamin supplement, these data were not translated into public health policy in the 1980s. It took scientists an entire decade from these early reports to design, implement, and report conclusive data from appropriately de- signed intervention trials. The results of the controlled intervention trials re- ported in 1991 and 1992 were supported by the collective findings of observa- tional studies published primarily between 1988 and 1995. None of the observational studies could identify folate specifically as the vitamin responsible for reducing NTDs. The definitive data from the MRC and Hungarian controlled trials led scien- tists and public health policymakers to accept the positive association between supplemental periconceptional folate use and NTD risk reduction as conclusive. This evolution of scientific data culminated with the U.S. Public Health Service (USPHS) policy statement issued in 1992 that all women of childbearing age who are capable of becoming pregnant should take 400 ~g/day of folic acid to reduce the risk of NTDs (CDC, 1992~. The USPHS recommendation was fol- lowed several years later by a federal regulation that required all "enriched"
46 EVOLUTION OF EVIDENCE cereal grain products to contain specified quantities of folic acid by January 1, 1998. The DRI report included a review of both the controlled trials and observa- tional studies and provided additional supporting data published since 1992. The DRI report also included a recommendation that all women capable of becoming pregnant consume 400 ,ug/day of folic acid from fortified foods, supplements, or both in addition to consuming food folate from a varied diet to reduce the risk of NTDs. The DRI recommendation thus reinforces and refines the 1992 USPHS public health policy statement. The science related to folate and NTDs has now evolved as discussed in the DRI report to research questions related to mecha- nisms by which supplemental folate reduces the risk of NTDs, and questions related to genetic predisposition and response to folate supplementation. Folate and Colorectal Cancer Diet and Health Report The D&H report does not discuss folate and colorectal cancer. The only re- lated statement was that "an increased incidence of tumors in the liver, colon, and esophagus results from diets deficient in methyl groups" (p. 338~. There was no mention of folate and colorectal cancer in the summary of the D&H report. Dietary Reference Intake Report The DRI report, Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B6, Folate, Vitamin By, Pantothenic Acid, Biotin, and Choline (IOM, 1998), stated, "experimental data indicate that changes in folate status may in- fluence the process of neoplastic changes in certain epithelial tissue: a negative change in folate status may stimulate carcinogenesis. It is unclear if supraphysi- ological doses obtained from supplements afford any protection" (p. 264~. The DRI report presented supporting evidence for modulation of colorectal cancer by folate status. Data from a case-control study by Lashner and coworkers (1989), although nonsignificant, indicated that chronic ulcerative colitis patients who were treated with a drug that inhibits folate absorption (sulfasalazine) were at higher risk for colonic dysplasia. When these patients were treated with folate supplements, the risk for colonic neoplasia was reduced. In a subsequent case- control study by the same research group (Lashner, 1993), red blood cell folate concentration was significantly lower in patients with colorectal neoplasia. The DRI report cited the large-scale observational study of Giovannucci and co- workers (1993), who found in two cohorts that risk of colorectal adenoma was elevated in individuals with low folate intake and high intakes of alcohol. The possible association with colorectal cancer was assessed in a subsequent large- scale observational study (Giovannucci et al., 1995) in which folate intake was
CASE STUDIES 47 assessed and associated with new cases of colorectal cancer in U.S. male health professionals during a 6-year follow-up period. The work by Giovannucci and coworkers supports the hypothesis that increased folate intake is inversely asso- ciated with the incidence of colorectal adenomatous polyps and colorectal can- cers. This study confirmed an inverse relationship between folate intake and colorectal cancer that was striking when combined with high alcohol consump- tion. The large size of the population, the prospective design of the two studies, and the effective control of a number of confounding dietary variables strengthen the conclusion that increased folate intake is inversely associated with the incidence of colorectal adenomatous polyps and of colorectal cancers. The DRI report also cited data from a case-control study in male smokers that was nested within the ATBC study (Glynn et al., 19961. In that study, nei- ther dietary folate intake nor blood folate level alone was significantly associ- ated with colorectal cancer. In contrast, the combination of high alcohol con- sumption and a low folate diet was associated with a significant increase in colorectal cancer risk. The potential influence of folate-related genetic polymor- phisms on colorectal cancer risk, with supporting evidence for a protective role for folate related to colorectal cancer risk, was referred to in the DRI report (Ma et al., 1997~. Studies investigating the hypothesis that a folate deficiency local- ized within the colonic mucosa may increase the risk for colorectal cancer were not supported by clinical observational data cited in the DRI report (Meenan et al., 1997~. In summary, the DRI report stated, "data supporting the modulation of car- cinogenesis by folate status are the strongest for the colorectum" (p. 265~. In addition, the DRI report indicated, "more evidence for or against a causal rela- tionship between folate status and colorectal cancer will be provided by data from prospective controlled intervention trials that are currently under way" (p. 266~. Evolution Promising in DRI Observational studies generally support the hypothesis that diminished folate status is associated with an increased rate of colorectal neoplasia. Data have evolved that strengthen the inverse association between folate status and colorectal cancer. Definitive conclusions related to the impact of folio acid sup- plementation on colorectal neoplasia await the results of ongoing controlled intervention trials. Lashner and coworkers observed that low folate status appears to increase the risk of colorectal adenomas or cancer in a case-control study of patients with chronic ulcerative colitis. Although the data fell slightly short of statistical sig- nificance, the study established the importance of examining this issue. Later work by Lashner's group confirmed these observations by prospectively
48 EVOLUTION OF EVIDENCE comparing red blood cell folate levels in ulcerative colitis patients with neoplas- tic colorectal tissue with matched controls without neoplasia. Observational studies generally support the hypothesis that diminished folate status is associated with an increased rate of colorectal neoplasia, although the nature of the relationship is different in each study. Some of the most con- vincing observational evidence to establish an association between folate status and colorectal neoplasia has come from two observational studies that have carefully examined interactions between folate and other dietary components, especially alcohol. In both studies, moderate to high alcohol intake significantly increased the neoplastic risk of a low-folate diet. In contrast to the early reports in persons with ulcerative colitis, the observational studies in the general popula- tion are particularly important since they show an association in individuals who do not have coexisting conditions that predispose them to colorectal cancer. Even though the data from all of the epidemiological studies related to folate and colorectal cancer are not uniformly supportive of an inverse relationship, a composite of the data provides a supportive argument for the existence of a rela- tionship that is stronger than that observed for cervical cancer. More definitive evidence for a causal relationship between folate status and colorectal cancer may be provided by the ongoing prospective, controlled intervention trials. Vitamin E and Prostate Cancer Diet and Health Report report. Vitamin E and prostate cancer was not specifically discussed in the D&H Dietary Reference Intake Report Vitamin E was discussed in the report, Dietary Reference Intakes for Vita- min C, Vitamin E, Selenium, and Carotenoids (IOM, 2000a). A possible preven- tive role in prostate cancer was suggested in the ATBC study (ATBC, 1994), a large clinical trial of vitamin E for the primary prevention of lung cancer in men from Finland who were heavy smokers. In that trial, a significant 34 percent lower incidence of prostate cancer was seen in the men who received supple- mental vitamin E (Heinonen et al., 19984. Thus, the DRI report concluded that "At present, the data from intervention trials are most suggestive for the ability of vitamin E to prevent prostate cancer, but only a single trial has yet been re- ported, and prostate cancer was not the primary endpoint of that study" (p. 220~.
CASE STUDIES Evolution Promising in DR1 49 The D&H report made no mention of a role for vitamin E in this cancer. As a result of information now available, largely the ATBC trial results, the Na- tional Cancer Institute is funding a large, 12-year, prostate cancer primary pre- vention trial of supplemental vitamin E and selenium, the Selenium and Vitamin E Cancer Prevention Trial. Thus, the relationship between supplemental vitamin E and prostate cancer prevention is perhaps best characterized as promising.
