Vitamin Supplement Use and Reduced Risk of Oral and Pharyngeal Cancer

American Journal of Epidemiology Vol 135, No 10 Copyright (D 1992 by The Johns Hopkins University School ot Hygiene and Public Health Printed in U S.A All rights reserved ORIGINAL CONTRIBUTIONS Vitamin Supplement Use and Reduced Risk of Oral and Pharyngeal Cancer Gloria Gridley, 1 Joseph K. McLaughlin, 1 Gladys Block, 2 William J. Blot, 1 Maria Gluch, 1 and Joseph F. Fraumeni, Jr. 1 Use of vitamin and mineral supplements was assessed in a population-based casecontrol study of oral and pharyngeal cancer, conducted during 1984-1985 in four areas of the United States. There was no association with intake of multivitamin products, but users of supplements of individual vitamins, including vitamins A, B, C, and E, were at lower risk after controlling for the effects of tobacco, alcohol, and other risk factors for these cancers. After further adjustment for use of other supplements, vitamin E was the only supplement that remained associated with a significantly reduced cancer risk. The adjusted odds ratio of oral and pharyngeal cancer for "ever regularly used" vitamin E was 0.5 (95% confidence interval 0.4-0.6). To the authors' knowledge, this is the first epidemiologic study to show a reduced oral cancer risk with vitamin E use. Although it is not clear that the lower risk among consumers of vitamin E supplements is due to the vitamin per se, the findings are consistent with experimental evidence and should prompt further research on the role of vitamin E and other micronutrients as inhibitors of oral and pharyngeal cancer. Am J Epidemiol 1992;135:1083-92. mouth neoplasms; neoplasms; pharyngeal neoplasms; retrospective studies; risk factors; vitamin E; vitamins It has been hypothesized that different vitamins may prevent cancer by a number of mechanisms, yet few epidemiologic studies have been designed to assess the effects of vitamin supplement use (1, 2). National survey data (3, 4) show that over 20 percent of adult Americans take supplements on a daily basis and that more than 50 percent have taken a supplement some time within the last year. Received for publication July 1, 1991, and in final form November, 1991. Abbreviations 1 Cl, confidence interval; IU, international units; OR, odds ratio. 1 National Cancer Institute, Division of Cancer Etiology, Epidemiology and Biostatistics Program, Bethesda, MD. 2 National Cancer Institute, Division of Cancer Prevention and Control, Bethesda, MD. Reprint requests to Ms. Gloria Gridley, National Cancer Institute, Executive Plaza North, Room 443, Bethesda, MD 20892. 1083 The major risk factors for oral and pharyngeal cancer are tobacco and alcohol use (5). It has been suggested that vitamins A or C and perhaps additional nutrients play a protective role in part by inhibiting the effects of smoking and drinking and other risk factors. Chronic iron and related deficiencies may also predispose the oral and pharyngeal mucosa to malignant transformation (6). These hypotheses have yet to be confirmed, but several case-control studies have found a reduced risk associated with increased fruit and vegetable consumption (7-11) and indices of dietary intake of vitamins A, particularly pro-vitamin A carotenoids, and vitamin C (8, 10, 12-14). We present here the results of an analysis of vitamin and mineral supplement use collected in four areas of the United States in what we believe to be the largest population-based case-control study of oral and pharyngeal cancer to date.

1084 Gridley et al. MATERIALS AND METHODS The study design and statistical methods have been described in detail in earlier reports (5, 8, 10). Briefly, pathologically confirmed incident cases of oral and pharyngeal cancer (hereafter called oral cancer) diagnosed from January I, 1984 to March 31, 1985 were ascertained from the populationbased cancer surveillance systems of the following four areas: New Jersey; Atlanta, Georgia; Los Angeles, California; and Santa Clara and San Mateo counties south of San Francisco/Oakland, California. Cancers of the tongue, pharynx, and other oral cancers were included, but cancers of the lip, salivary gland, or nasopharynx were excluded. Controls matched to cases on age, race, and sex were randomly chosen from the populations of each area using random digit dialing (for persons aged under 65 years) and Health Care Financing Administration (HCFA) rosters (for persons aged 65 years or older). Interviews were administered, usually in the home, to obtain information on demographic variables, tobacco and alcohol use, diet, occupation, and medical history. The vitamin supplement section of the interview included questions about years of use and age started, for both multivitamin preparations (including brand) and use of individual vitamin and mineral supplements such as vitamins A, C, E, B-complex vitamins, vitamin B-12 shots, minerals, iron, and cod liver oil. Information on dose was not ascertained. Ever use was defined as taken "on a regular basis for 6 months or longer." We estimated vitamin and mineral dose based on the type of pills or capsules reported, summing across products used. If brands were given, actual content in recent years was used. Otherwise, assumptions for content of each supplement type (4, ) were as follows: for regular multivitamins, vitamin A = 5,000 international units (IU), B vitamins (B dose was used to represent B- complex vitamins) = 1.5 mg, vitamin C = 60 mg, vitamin E = 30 IU, iron = 18 mg; for "stress"-type multivitamins, B, =, C = 600, E = 30, iron = ; for "therapeutic"- type multivitamins, A = 10,000, B, = 10, C = 200, E =, iron = 12; for prenatal, A = 8,000, B, = 1.5, C = 60, E = 30, iron = 60; for B-complex multivitamin, B, = ; vitamin A capsules, 10,000 IU; cod liver oil, 4,600 IU of vitamin A; B-complex vitamin pills, mg B,; vitamin C pills, 250 mg; vitamin E capsules, 100 IU; and iron pills, 50 mg (one source (16) reported higher doses for vitamin C (500 mg) and vitamin E (400 IU), but using these doses did not change our estimates of risk). For generic multivitamins, the dose-years of vitamin E were adjusted for calendar year of use: no vitamin E in multivitamins before 1974; IU of vitamin E for 1974-1982; and 30 IU of vitamin E for 1983 forward (personal communication from Connie Kent, Miles Inc., 1991). Dose-years were calculated as the sum of dose times years of use for each separate supplement reported. The association between vitamin supplement use and oral cancer was estimated by the odds ratio (OR). Logistic regression was used to calculate adjusted point estimates and 95 percent confidence intervals (CI) (). All odds ratios presented in the tables and results were adjusted for sex, race, and four hierarchical non-overlapping categories of combined tobacco and alcohol use (former or nonsmokers who consumed less than five alcoholic drinks per week; current smokers of 1-19 cigarettes per day or consumers of 5-14 drinks per week; current smokers of 20-39 cigarettes per day or consumers of -29 drinks per week; current smokers of 40+ cigarettes per day or consumers of 30+ drinks per week). Finer stratification of tobacco and alcohol use (the principal risk factors for this cancer) had no further effect on the estimates, nor did adjustment for age, education, study area (region), body mass, or diet (consumption of fruit or vegetables and dietary intake of vitamins C and E (estimated from the food frequency questions (8))). Effect modification was also evaluated for these risk factors by examination of stratum-specific risk estimates, as well as the statistical significance of interaction terms in the logistic regression model. The statistical significance of dose response (trend) was evaluated by treating

Vitamin Supplement Use and Reduced Risk of Oral Cancer 1085 the categorical variable as continuous in a logistic regression model. The significance of differences in the demographic and risk factor profiles of supplement users was tested by computing the chi-square statistic. RESULTS Interviews were completed for 1,114 oral and pharyngeal cancer cases and 1,268 controls, representing 75 percent of all incident cases in the study areas and 76 percent of controls aged 65 years and older. The response rate for controls under age 65 was 79 percent at the household screening phase and 73 percent at the field interview phase. After exclusion of subjects (11 cases and six controls) with missing information for more than two of the vitamin supplement questions, the number of subjects included in this analysis was 2,365, or 1,103 cases and 1,262 controls. Among the cases, 68 percent were men and 32 percent were women, while among the controls, 66 percent were men and 34 percent were women. Most subjects were non-hispanic whites (78 percent); 18 percent of cases and 16 percent of controls were black; and 4 percent of cases and 7 percent of controls were Hispanic. Exclusion of information provided by surrogate respondents (23 percent of cases and 2 percent of controls) did not alter the risk estimates, nor did exclusion of cases and HCFA controls without telephones (6 percent of cases and 1 percent of controls), so these subjects were included in the analyses. Patterns of supplement use were examined first among the controls (table 1). Use of supplements was associated with the following: being female, white, more highly educated, a California resident, having a lower body mass, and consuming more fruits and vegetables. Heavy tobacco and alcohol consumption were only moderately associated with less supplement use. These associations were due almost entirely to differences in use of individual vitamin supplements, and education was the only variable that was significantly associated with use of multivitamin supplements. Age was not consistently associated with supplement use: controls under age 45 years reported the most supplement use, and controls aged 45-54 years reported the least. Fewer cases consumed individual vitamin supplements than controls (31 percent vs. 43 percent). However, the patterns of usage of the cases and controls were similar (not shown) except as follows: there was no association between education and multivitamin use, next-of-kin interviews (254 cases) reported more multivitamin use and less individual supplement use, and more cases with high intake of dietary vitamin C consumed supplements. About two-thirds of the individual supplement users also used multivitamin products, and two-thirds of the multivitamin users took at least one additional individual supplement. Multivitamins were used on a regular basis by 46 percent of the controls, vitamin C by 29 percent, vitamin E by 22 percent, B vitamins by 19 percent, vitamin A by 10 percent, and iron by 9 percent. Use of one supplement was highly correlated with use of another. For example, the correlation between vitamin E and vitamin C supplement use was r = 0.6. Out of 884 subjects who reported use of individual supplements, only 136 reported taking no multivitamin and only one type of individual supplement: 39 took vitamin C only, 23 iron, 23 vitamin B12 shots, 21 vitamin E, 16 B vitamins, 6 cod liver oil, 6 minerals, and 2 vitamin A only. Supplement use of any kind was associated with a significantly reduced risk of oral cancer (odds ratio (OR) = 0.8, 95 percent confidence interval (CI) 0.7-). There was no further reduction in risk for longer duration of supplement use in general (ORs (95 percent CIs) by years of use of the longest used supplement): less than 5 years, 0.8 (0.6-1.1); 5-14 years, 0.8 (0.6-); or more years, 0.8 (0.6-). Use of any individual supplements (i.e., not including multivitamins) was more strongly associated with reduced risk (OR = 0.7, 95 percent CI 0.6-0.8), with a greater reduction in risk for 5 or more years of use (ORs (95 percent CIs) by years of use of the longest used individual supplement): less than 5 years, 0.8 (0.6-); 5-14 years, 0.6 (0.4-0.8); or more years,

TABLE 1. Profile of matched controls reporting supplement use in an oral and pharyngeal cancer casecontrol study in four areas of the United States, 1984-1985 Mrt % any supplement* % only muitivitamin % individual supplement! All controls 1,262 59 43 Sex Men Women Race White Black Hispanic Education (years) <12 12 >12 Region East (New Jersey/ Atlanta) West (California) Tobacco/alcohol Smoking status Nonsmokers Heavy smokers Drinking status Nondrinkers Heavy drinkers Respondent Self Next-of-kin Age (years) <45 45-54 55-64 65+ X 2 Body mass index quartile Fruit/vegetable quartile Dietary vitamin C quartile 831 431 974 202 86 395 361 506 597 665 471 191 387 86 310 127 1,243 19 103 236 401 522 3 320 3 318 314 314 53 71 62 51 45 45 60 69 53 64 62 53 p = 0. 58 52 p = 0.18 57 55 p = 0.57 59 74 p = 0.19 70 55 58 60 p = 0.07 68 50 53 61 p = 0.07 54 61 p = 028 16 18 p = 0.29 20 p = 0.65 13 19 p = 0.03 16 p = 0.64 20 p = 0.22 20 p = 0.31 21 p = 0.12 16 p = 0.93 22 14 16 p = 0.33 20 16 p = 0.12 18 p = 0.75 p = 0.76 37 53 45 36 26 p< 0.001 32 43 50 36 48 45 34 p = 0.06 42 33 P = O.1O 42 34 p = 0.21 42 58 p = 0. 48 40 41 44 p = 0.54 48 34 p = 0.001 35 46 p = 0.02 40 45 p = 0.62 * Ever took a vitamin supplement on a regular basis for 6 months or more. t The questionnaire ascertained use of eight specific supplements: A, C, E, B vitamins, vitamin B-12 shots, iron, minerals, and cod liver oil. Very few controls (6%) took only one kind of specific vitamin and no muitivitamin' vitamin A, 1 subject, vitamin C, 24; vitamin E, 14; B vitamins, 9; vitamin B-12 shots, 6, iron, 11; minerals, 2, cod liver oil, 4. Two-thirds of the sub ects in this column used multivitamins, and many used more than one kind of individual supplement

TABLE 2. Odds ratios (OR) and 95 percent confidence intervals (Cl) for oral and pharyngeal cancer for duration of supplement use, by type of supplement, in four areas of the United States; 1984-1985* Type of supplement and years of use Multivitamins 1-2 (n = 192) 3-4 (n = 113) 5-9 (n = 180) 10-14(n = 180) -24 (n = 7) 25+ (n = 205) Vitamin C 1-2 (n = 106) 3-4 (n = 79) 5-9 (n = 132) 10-14 (n = 97) +(n = 149) Vitamin E 1-2(n = 90) 3-4 (n = 69) 5-9 {n = 90) 10-14 (n = 73) +(n = 71) B vitamins 1-2(n = 84) 3_4 (n = 52) 5-9 (n = 78) 10-14 (n = 68) +(n = 76) Minerals 1-2 (n = 58) 3-4 (n = 35) 5-9 (n = 40) 10+(n = 61) Iron 1-2(n = 57) 3-4 (n = 34) 5-9 (n = 32) 10+ (n = 43) Vitamin A 1-2(n = 43) 3-4 (n = 36) 5-9 (n = 32) 10+(n = 53) B12 shots <5 (n = 58) 5+ (n = 49) Cod liver oil <5 (n = 28) 5+ (n = 31) OR Men 0.6 0.4-0.9 0.5 0.3-1.2 0.7-1.8 0.7 0.5-1.2 1.2 0.7-1.8 0.9 0.6-1.4 p = 0.68 0.9 0.5-1.6 0.6 0.3-1.2 0.6 0.3-0.5 0.2-0.9 0.6 0.4-0.4 0.2-0.8 0.8 0.4-1.5 0.4 0.2-0.7 0.4 0.2-0.9 0.7 0.4-1.4 0.6 0.3-1.4 0.7 0.3-1.8 0.6 0.3-1.3 0.5 0.2-1.1 0.6 0.3-1.3 p = 0.01 0.6 0.3-1.6 0.3 0.8-1.2 0.4-2.5 0.6 0.2-1.3 p = 0.04 0.7 0.3-1.9 0.1 0.0-0.8 0.5 0.1-1.9 0.6 0.2-1.5 p = 0.01 0.3 0.1-0.5 0.2-1.5 0.3 0.1-1.2 0.5 0.2-1.1 p = 0.004 1 0 1.8 0.7-5.1 0.6 0.2-1.4 p = 012 1 0 0 4 0.1-1.4 0.5 0.2-1.7 p = 0.04 OR Women 0.7 0.4-1.2 0.6 0.3-1.2 0.6 0.3-1.2 1.2 0.6-2.1 0.4 0.2-0.8 0.6 0.4-1.1 p = 0.06 0.8 0.4-1.6 0.7 0.3-1.8 0.6 0.3-1.1 0.5-2.2 0.4 0.2-0.8 p = 0.04 0.4 0.2-0.5 0.2-1.4 0.2 0.1-0.6 0.4 0.2-1.1 0.3 0.1-1.1 0.6 0.3-1.3 0.8 0.3-2.1 0.6 0.3-1.2 0.3 0.1-0.8 0.4 0.2- p = 0.002 0.4 0.2-1.1 0.6 0.2-1.8 0.5 0.2-1.4 0.4 0.1-1.1 p = 0.007 0.4-2.2 1.1 0.4-2.8 0.7 0.3-1.9 0.5 0.2-1.6 p = 0.30 0.3 0.1-0.9 0.3-3.0 0.1 0.1-1.4 0.2 0.1-0.7 p = 0.002 0.4-2.2 0.5 0.2-1.3 p = 0.07 1.7 0.4-7.3 0.2 0.0-0.9 p = 0.01 OR Total 3R Total adjusted also for vitamin E$ 0.6 0.4-0.9.0 ( 5.7 0.5-0.6 0.4- ().7 0.5-1.1 0.7-1.4.2 0.8-1.7 0.9 0.6-1.3.1 0.8-1.6 0.8 0.6-1.2.0 0.7-1.5 0.8 0.5-1.1.1 0.7-1.4 p = 0. p = 0.79 0.9 0.5-1.3 1.1 0.7-1.8 0.7 0.4-1.2 0.5-1.7 0.6 0.4-0.9 ( 3.8 0.5-1.3 0.7 0.4-1.1.0 0.6-1.7 0.5 0.4-0.8 ( 3.8 0.5-1.3 p = 0.39 0.5 0.3-0.7 0.7 0.4-1.2 0.3 0.2-0.5 0.4 0.2-0.8 0.6 0.3-1.1 p< 0.001 0.7 0.4-1.1 3.9 0.5-1.6 0.8 0.4-1.6 1.3 0.7-2.5 0.6 0.4-3.9 0.5-1.6 0.4 0.2-0.8 3.6 0.3-1.2 0.5 0.3-0.9 3.8 0.4-1.5 p = 0.27 0.5 0.3-3.7 0.4-1.4 0.5 0.2-1.2 3.7 0.3-1.6 0.7 0.3-1.4 1.2 0.3-2.5 0.5 0.3-0.9 3.8 0.5-1.6 p = 0.001 p = 0.52 0.9 0.5-1.7 1.3 0.7-2.3 0.6 0.3-1.4 3.8 0.4-1.9 0.7 0.3-1.5 0.5-2.3 0.6 0.3-1.2 0.5-2.2 p = 0.01 p = 0 0.4 0.2-0.8 3.6 0.3-1.3 0.7 0.3-1.5 1.1 0.5-2.5 0.4 0.1-0.9 3.6 0.2-1.4 0.4 0.2-0.8 3.6 0.3-1.4 p< 0.001 p = 0. 1.3 0.7-2.4 1.7 0.9-3.2 0.5 0.3-1.1 3.7 0.4-1.4 p = 0.02 p = 0.76 0.7 0.3-1.8 1.1 0.4-2.8 0.3 0.1-0.9 3.5 0.2-1.3 p = 0.003 p = 0.40 All risks relative to those who took no supplements (n - 1,043), and adjusted for race, tobacco, and alcohol intake (and sex in the total columns), but not the other supplements t Subiects in each supplement type include those who used other supplements also. X Ever use of vitamin E supplements

1088 Gridley et al. TABLE 3. Odds ratios (OR) and 95 percent confidence intervals (Cl) for oral and pharyngeal cancer, by estimated daily dose and dose-years* of supplement usef, in four areas of the United States, 1984-1985 Daily dose OR* Adjusted also for dose of vitamin E Dose-years OR* Adjusted also for dose-years of vitamin E Vitamin A (III) 5,000 < 10,000 <20,000 B vitamins (mg) 1.5 < > Vitamin C (mg) 60 <200 <300 310 >325 Vitamin E (IU) 30 100 >100 Iron (mg) < 18 >20 0.8 0.7-1.1 0.9 0.7-1.2 0.4 0.2-0.6 p = 0.003 0.8 0.7-1.1 0.7-1.5 0.7 0.4-0.6 0.4-0.8 p = 0.001 0.9 0.7-1.2 0.9 0.6-1.3 0.6 0.4-0.9 0.6 0.4-0.8 1.1 0.7-1.7 p = 0.003 1.1 0.8-1.5 1.1 0.8-1.4 0.5 0.3-0.7 0.5 0.4-0.7 1.1 0.8-1.5 0.8 0.6-0.6 0.4-0.9 p = 0.007 OR 1.1 0.8-1.4 1.2 0.8-1.6 0.6 0.4-1.1 p = 0.92 0.8-1.4 1.2 0.8-1.8 0.9 0.6-1.4 0.9 0.6-1.4 p = 0.89 1.1 0.8-1.4 10 0.7-1.5 0.8 0.6-1.3 0.8 0.6-1.2 1.6-2.5 p = 0.82 1.3 0.9-1.9 0.8-1.3 0.9 0.6-1.3 p = 0.91 <20,000 <50,000 <100,000 >100,000 <10 <25 <60 <0 >0 <400 >400 >1,000 >2,000 >4,000 <100 <200 <700 >700 <70 <0 <400 >400 0.6 0.4-0.9 0.8 0.6-1.1 0.9 0.6-1.2 0.8 0.6-1.1 p = 0.08 0.6 0.4-0.8 0.9 0.7-1.3 0.9 0.6-1.2 0.8 0.6-1.2 0.7 0.5-0.9 p = 0.02 0.6 0.5-0.8 0.7-1.3 0.7 0.5-0.8 0.6-1.1 0.7 0.5- p = 0.01 0.9 0.6-1.2 1 1 0.8-1.4 0.6 0.4-0.8 0.5 0.3-0.7 0.7 0.5-0.9 0.6-1.2 0.8 0.6-1.1 0.7 0.5-1 0 p = 0.02 OR 0.7 0.5-0.9 0.8 0.6-1.2 0.7-1.5 1.1 0.8-1.5 p = 0.95 0.6 0.5-0.9 0.7-1.4 0.7-1.4 0.7-1.4 0.9 0.6-1.4 p = 0.65 0.7 0.5-0.9 0.7-1.4 0.8 0.6-1.2 0.7-1.4 0.7-1.5 p = 0.54 0.8 0.6-0.9 0.7-1.3 0.7-1.3 0.9 0.6-1.3 p = 0.48 * Estimated daily dose was calculated as vitamin or mineral dose in multiple vitamins (if subject reported using multiple vitamins) plus vitamin or mineral dose in individual supplements (if subject reported taking them). Dose-years were calculated as the sum of dose x years of use for each separate supplement reported. For generic multivitamins, the dose-years of vitamin E were adjusted for calendar year of use. t Subjects in each vitamin type include those who took other types in addition. Odds ratios adjusted for race and sex and tobacco and alcohol use. The reference group is always the 1,043 subjects who took no vitamin supplements. 0.6 (0.4-0.8). Risks also tended to decline with increasing duration of supplement use for each of the vitamin supplements except multivitamins (table 2). However, in a multivariate model with ever use of each of the types of supplement included, only the reduced risk associated with vitamin E supplement use remained statistically significant. In models examining one supplement at a time, after adjustment for ever use of vitamin E supplements, none of the other supplements was still significantly associated with a reduced cancer risk. Vitamin intake from supplements was also examined by estimated average daily intake (dose) and by dose-years of vitamins A, B, C, and E, and iron (table 3). For these calculations, intake was summed across the various types of supplements. The effects were similar to those seen for years of use of individual supplements. Again, there was no significant reduction in risk for intake of any

Vitamin Supplement Use and Reduced Risk ot Oral Cancer 1089 TABLE 4. Odds ratios (OR) for oral and pharyngeal cancer for ever use of individual vitamin E supplements, by reported use of other supplements, intake of fruits and vegetables, and estimated dietary vitamin E intake in four areas of the United States, 1984-1985 No vitamin E supplement Used vitamin E supplement Tobacco/alcohol use, by quartile Second Third Other supplements No other supplement Multivitamins only Vitamin C (± multivitamins) Other supplements Fruit and vegetable consumption, by quartile Second Third Estimated dietary vitamin E intake (IU) <3.5 <4.5 <6 6+ OR* (Cases/controls) t (103/364) 1.6 (92/214) 3.7 (239/248) 14.4 (545/4) t (527/516) (231/209) 1.2 (70/69) 0.8 (1/186) t (372/264) 0.9 (270/240) 0.7 (181/245) 0.6 (6/231) t (258/227) 0.9 (246/258) 0.9 (260/281) (2/214) OR* (Cases/controls) 0.7(20/107) 0.9(/68) 1.5 (28/70) 6.2 (59/37) 0.5 (7/14) 0.4 (8/21) 0.5(18/47) 0.5(91/200) 0.5(34/51) 0.5 (38/72) 0.4 (26/72) 0.2 (26/87) 0.6 (25/45) 0.5 (30/68) 0.4 (34/97) 0.5 (35/72) * Odds ratios adjusted for race and sex (and tobacco and alcohol use except for the tobacco/alcohol section of the table) t Referent category. other vitamin after adjustment for dose or dose-years of use of vitamin E supplements. In both tables 2 and 3, the key indicator of reduced risk was use of vitamin E supplements. The odds ratio associated with any use (0.5 (95 percent CI 0.4-0.6)) was about the same as the odds ratios for short- or long-term use of vitamin E supplements or for estimated high intake of vitamin E from all types of supplements. Twenty-seven percent of vitamin E dose-years were from multivitamins; and subjects in the two higher categories of vitamin E dose-years (200+ IUyears of vitamin E) solely due to long-term multivitamin use were not at reduced risk of oral cancer (OR = ), and their removal from the two higher categories of vitamin E dose-years resulted in both categories having the same risk (both ORs = 0.5). Hence, there was little evidence of a dose-response. After exclusion of subjects who used vitamin E supplements (124 cases and 282 controls), vitamin A supplement use was associated with a significantly reduced risk (OR = 0.3, 95 percent CI 0.1-0.8) (based on nine cases and 26 controls). After exclusion of both vitamin A and E users, no reduction in risk was observed for any other vitamin supplement or combination of vitamin supplements. The combination of vitamin A and E supplement use was not associated with any greater reduction in risk (OR = 0.5, 95 percent CI 0.3-0.7) than vitamin A (OR = 0.3,95 percent CI 0.1-0.8) or vitamin E (OR = 0.5, 95 percent CI 0.4-0.7) use alone. Similarly, the use of both vitamin C and E supplements (OR = 0.5, 95 percent CI 0.4-0.7) was associated with no greater reduction in risk than use of vitamin E alone. There was no difference in the reduction of risk associated with vitamin E supplement use by site of oral cancer: tongue, pharyngeal, or other oral cancers. Stratum-specific estimates were examined for vitamin E supplement use by all the covariates in table 1. There was remarkable consistency of the vitamin E supplement association in every subgroup. For example,

1090 Gridley et al. table 4 shows the interaction tables for tobacco and alcohol use, use of other supplements, and dietary variables. In every stratum, ever use of vitamin E supplements is associated with approximately half the risk in the first column. Even though dietary vitamin E consumption (estimated from the responses to the food frequency section of the questionnaire) was not associated with reduced risk, at every level, vitamin E supplement use was still associated with half the risk. The age at which regular supplement use started was associated with risk in the same pattern as years of use, that is, risk was lower for earlier age started. No effect on risk estimates was seen when years of use was adjusted for age or age started supplement use, nor was there any effect on risk estimates when age started supplement use was adjusted for age or years of use. Recency of supplement use was also examined, although most supplement users were current users (only one-third of the supplement users had stopped more than 2 years prior to interview). Subjects who stopped supplement use, including use of vitamin E, were still found to be at decreased risk of oral cancer. For most types of supplements, and in particular use of vitamin E, there was no effect of stopping usage on risk. None of the findings reported in the tables or text was affected by exclusion of subjects with less than 5 years of vitamin supplement use, subjects who reported a recent change of diet, or surrogate respondents. DISCUSSION This study found a significantly reduced risk of oral and pharyngeal cancer with use of individual supplements, particularly vitamin E supplements. Vitamin E supplements were the only products significantly associated with decreased risk after adjustment for use of the other vitamin products. To our knowledge, only one other epidemiologic study of these cancers (18) has specifically examined vitamin supplement use, and that study found a reduced risk with vitamin A and C supplement use, but not with multivitamins, although vitamin E supplement use was not assessed. It is of interest that our study would have had similar findings had we not assessed supplemental vitamin E. It is not clear whether vitamin E itself is responsible for the reduced risk of oral and pharyngeal cancer among supplement users. In an earlier analysis (8), we did not find an effect of vitamin E from dietary sources. This may reflect the fact that the level of vitamin E from diet is much lower than from supplements. Dietary intakes in the United States average 11 IU for women and 14 IU for men (slightly less than the recommended daily allowance (RDA)) (19), while most multivitamins currently contain 30 IU and individual vitamin E supplements generally contain from 100 to 400 IU. In contrast, dietary versus supplement levels of vitamin A are similar; and dietary levels of vitamin C generally exceed levels of vitamin C in multivitamin tablets and can approach vitamin C supplement levels (20). It also may be that our questionnaire was not well designed to ascertain dietary intake of vitamin E (e.g., we did not distinguish margarine from butter, nor ask about vegetable oils, which are important indicators of vitamin E status (21)). Estimating the dietary intake of vitamin E (alpha tocopherol and other tocopherols and tocotrienols (22)) is complicated by the fact that there are 16 isomers of vitamin E (eight synthetic and eight naturally occurring) (23), making measurement of vitamin E "activity" complex. Dietary vitamin E is also hard to assess because it is unstable during food storage, processing, and cooking (24). In part because of these difficulties, no other studies of oral cancer have evaluated risk in relation to estimated dietary intake of vitamin E, although a recent investigation of gastric cancer in several areas of Italy found markedly lower risks associated with high consumption of alpha tocopherol (25). There is biologic rationale for an inhibitory effect of vitamin E on cancer risk. Vitamin E acts as an antioxidant (26) to neutralize free radicals (27), inhibit nitrosation and protect fatty acids of cell membranes against peroxidation (26, 28). Vitamin E may also enhance immune function (29),

Vitamin Supplement Use and Reduced Risk of Oral Cancer 1091 and it is an inhibitor of tumor initiation in experimental tumor systems (30). Of particular note is that vitamin E has been shown in several studies (31, 32) to prevent buccal pouch cancers in hamsters. This finding may be related in part to the capacity of vitamin E to stimulate immunoreactive Langerhans' cells of the hamster buccal pouch (33). The tumor-inhibitory effects of other vitamins and minerals have also been reported from experimental studies (34), and epidemiologic investigations of oral cancer (7-11) have fairly consistently noted reduced risks associated with intake of fresh fruits and vegetables. We also noted lowered risks among consumers of individual vitamin A, B, and C supplements, but these were not independent of vitamin E supplement use. The Plummer-Vinson syndrome of iron, B vitamins, and other deficiencies has also been linked to pharyngeal cancer (35). We found lower risks among persons who used iron supplements, but, after adjustment for vitamin E supplements, use of supplemental iron was not associated with cancer risk. The profile of vitamin supplement users among our controls was quite similar to that of previous national surveys (3, 4). Supplements were used more often by subjects who were white, female, better educated, from California, thinner (lower body mass or Quetelet's index), and consumed more fruit and vegetables. Because we defined vitamin pill users as those who ever used any vitamin supplement regularly for 6 months or longer, the prevalence of use (61 percent) was not directly comparable to these national surveys, which defined regular use differently. In one survey (4), for example, use of multivitamins ever in the past year was 51 percent, but current daily use was only 23 percent. However, the relative ranking of use of individual vitamins was quite similar to national data, with reported frequency of use in descending order as follows: multivitamins, vitamin C, vitamin E, B vitamins, and vitamin A. The juxtaposition of no association with multivitamin use versus an inverse association with individual vitamin supplement use is noteworthy. The profile of multivitamin users was quite different from that of individual supplement users, suggesting that users of individual supplements may have a particular life-style that contributes to a lower risk of oral and pharyngeal cancer. However, adjustment for smoking, drinking, and dietary variables the major risk factors in this study as well as age, education, race, sex, and region did not account for the reduced risks we observed with individual supplement use. Also, the lack of effect modification and the consistency of our findings for vitamin E supplement use were quite striking. Reduced risks were seen regardless of sex, race, tobacco and alcohol use, education, or several markers of dietary status. One of the limitations of the study is the imprecise information on actual amounts of the various vitamins consumed by the cases and controls. Imputation of a higher dose for vitamin E supplements (400 vs. 100 IU) did not affect our findings; however, the estimation of dose was crude and misclassification of vitamin intake levels could have hindered detection of dose-response trends. If the plateau effect observed for vitamin E is confirmed, it could argue against an etiologic nature of the association, or it could indicate a threshold a minimum level of vitamin E dose necessary to see an effect on cancer risk. The absence of an association with multivitamin use was of interest for this reason. Multivitamins have contained vitamin E only since 1974, and the dose of vitamin E in multivitamins is much lower than separate vitamin E supplements (though higher than from diet). Recall bias is a potential problem in retrospective studies; however, the findings in this study were unchanged by elimination of surrogate respondents, or subjects who reported only recent supplement use, or subjects who reported a recent change in diet. The reduced risk associated with vitamin E supplements was remarkably consistent in all subgroups, and in all time periods (i.e., for both short-term use and long-term use, and for recent use or no recent use). In summary, individual vitamin supplement use, but not use of multivitamins, was associated with a reduced risk of oral and pharyngeal cancer in this case-control study. This finding was most pronounced for users

1092 Gridleyetal. of vitamin E supplements, and provides epidemiologic support for experimental models suggesting that vitamin E can prevent oral carcinogenesis. Further research is warranted to evaluate the possibility that some unknown but correlated life-style variable may account for this effect, and to further clarify the potential of vitamin E supplements as well as other micronutrients to inhibit the development of oral and pharyngeal cancer. REFERENCES 1. Smith PG, Jick H. Cancers among users of preparations containing vitamin A. Cancer l978;42: 808-H. 2. Enstrom JE. Pauling L. Mortality among healthconscious elderly Californians. Proc Natl Acad Sci USA 1982:79:6023-7. 3. Block G, Cox C, Madans J, et al. Vitamin supplement use, by demographic characteristics. Am J Epidemiol 1988:127:297-309. 4. Subar AF, Block G. Use of vitamin and mineral supplements: demographics and amounts of nutrients consumed. The 1987 Health Interview Survey. Am J Epidemiol 1990:132:1091-1101. 5. Blot WJ. McLaughlin JK. Winn DM. et al. Smoking and drinking in relation to oral and pharyngeal cancer. Cancer Res 1988:48:3282-7. 6 Binnie WH, Rankin KV, MacKenzie IC. Etiology of oral squamous cell carcinoma. J Oral Pathol 1983:12:11-29. 7. Winn DM, Ziegler RG, Pickle LW, et al. Diet in the etiology of oral pharyngeal cancer among women from the southern United States. Cancer Res 1984:44:1216-22. 8. McLaughlin JK. Gridley G, Block G. et al. Dietary factors in oral and pharyngeal cancer. J Natl Cancer Inst 1988:80:1237-43. 9. Franco EL, Kowalski LP, Oliveira BV. et al. Risk factors for oral cancer in Brazil: a case-control study. Int J Cancer 1989:43:992-1000. 10. Gridley G, McLaughlin JK, Block G, et al. Diet and oral and pharyngeal cancer among blacks. Nutr Cancer 1990:14:219-25. 11. La Vecchia C, Negri E, D'Avanzo B. et al. Dietary indicators of oral and pharyngeal cancer. Int J Epidemiol 1991:20:39-44. 12. Graham S, Mettlin C, Marshall J, et al. Dietary factors in the epidemiology of cancer of the larynx. Am J Epidemiol 1981;113:675-80. 13. Marshall J, Graham S, Mettlin C, et al. Diet in the epidemiology of oral cancer. Nutr Cancer 1982:3: 145-9. 14. Mackerras D, Buffler PA, Randall DE. et al. Carotene intake and the risk of laryngeal cancer in coastal Texas. Am J Epidemiol 1988:128:980-8.. Physicians' desk reference. 33rd ed. Oradell, NJ: Medical Economics Co, 1979. 16. Park YK, Kim I, Yetley EA. Characteristics of vitamin and mineral supplement products in the United States. Am J Clin Nutr 1991:54:750-9.. Breslow NE, Day NE. Statistical methods in cancer research. Vol 1. The analysis of case-control studies. (IARC publications no. 32). Lyon: International Agency for Research on Cancer, 1980. 18. Rossing MA, Vaughan TL, McKnight B. Diet and pharyngeal cancer. Int J Cancer 1989;44:593-7. 19. Murphy SP, Subar AF, Block G. Vitamin E intakes and sources in the United States. Am J Clin Nutr 1990:52:361-7. 20. Block G, Sorenson A. Vitamin C intake and dietary sources by demographic characteristics. Nutr Cancer 1987:10:53-65. 21. Knekt P, Seppanen R, Aaran RK. Determinants of serum alpha-tocopherol in Finnish adults. Prev Med 1988::725-35. 22. National Research Council. Diet and health. Implications for reducing chronic disease risk. Report of the Committee on Diet and Health, Food and Nutrition Board, Commission on Life Sciences. Washington, DC: National Academy Press, 1989. 23. Horwitt MK. The promotion of vitamin E. J Nutr 1986:116:1371-7. 24. National Research Council. Diet, nutrition, and cancer. Report of the Committee on Diet, Nutrition, and Cancer, Food and Nutrition Board, Commission on Life Sciences. Washington. DC: National Academy Press. 1982. 25. Buiatti E. Palli D, Decarh A, et al. A case-control study of gastric cancer and diet in Italy: II. Association with nutrients. Int J Cancer 1990;45: 896-901. 26. Stahelin HB. Vitamins and cancer. Recent Results Cancer Res 1988:108:227-34. 27. Watson RR. Leonard TK. Selenium and vitamins A, E, and C: nutrients with cancer prevention properties. J Am Diet Assoc 1986;86:5O5-1O. 28. Bieri JG, Corash L, Hubbard VS. Medical uses of vitamin E. N Engl J Med 1983:308:1063-71. 29. Meydani SN, Barklund MP, Liu S, et al. Vitamin E supplementation enhances cell-mediated immunity in healthy elderly subjects. Am J Clin Nutr 1990;52:557-63. 30. Slaga TJ, Bracken WM. The effects of antioxidants on skin tumor initiation and aryl hydrocarbon hydroxylase. Cancer Res 1977:37:1631-5. 31. Trickier D, Shklar G. Prevention by vitamin E of experimental oral carcinogenesis. J Natl Cancer Inst 1987;78:165-9. 32. Shklar G, Schwartz JL, Trickier DP, et al. Prevention of experimental cancer and immunostimulation by vitamin E (immunosurveillance). J Oral Pathol Med 1990; 19:60-4. 33. Schwartz J, Odukoya O, Stoufi E, et al. Alpha tocopherol alters the distribution of Langerhans cells in DMBA-treated hamster cheek pouch epithelium. J Dent Res 1985,64:1-21. 34. Bin DF. Update on the effects of vitamins A, C, and E and selenium on carcinogenesis. Proc Soc Exp Biol Med 1986; 183:311-20. 35. Wynder EL, Hultberg S, Jacobsson F, et al. Environmental factors in cancer of the upper alimentary tract. Cancer 1957;10:470-87.