- Email: [email protected]
Micronutrients and ovarian cancer: A case-control study in Italy
Annals of Oncologv 12 1589-1593. 2001 © 2001 Kluwer Academic Publishers Printed m the Netherlands
Original article Micronutrients and ovarian cancer: A case-control study in Italy E. Bidoli,1 C. LaVecchia,23 R.Talamini,1 E. Negri,2 M. Parpinel,1 E. Conti,4 M. Montella,5 A. Carbone6 & S. Franceschi7 ] Servizio di Epidemiologia, Centro di Rifenmento Oncologwo. Aviano. 2lstiluto di Ricerche Farmacologiche 'Mario Negri', Milan, ^ Istituto di Statistica Medica e Bwmetna, Umversita degli Studi di Milano, Milan, 4Servizio di Epidemiologia e Oncogenesi, Istituto Regina Elena, Rome. ! Serviziodi Epidemiologia, Istituto Tumori 'Fondaztone Pascale', Naples, ^Division of Human Pathology, Centro di Rifenmenlo Oncologico. Aviano, Italy, 1Field and Intervention Studies Unit, International Agency for Research on Cancer. Lyon, France
Summary Background The role of selected micronutnents, vitamins and minerals in the aetiology of epithelial ovarian cancer was investigated using data from a case-control study conducted between 1992 and 1999 in five Italian areas Patients and methods Cases were 1031 patients with histologically confirmed incident epithelial ovarian cancer Controls were 2411 subjects admitted for acute, non-neoplastic diseases to major hospitals in the same catchment areas Dietary habits were elicited using a validated food frequency questionnaire including 78 food groups and recipes Odds ratios (OR) and 95% confidence intervals (95% CI) were computed by quintiles of intake of nutrients Results Inverse associations emerged for vitamin E (OR =
Introduction A role of dietary factors on ovarian carcinogenesis was originally suggested on the basis of descriptive epidemiology and correlational studies, [1-4], but it remains largely undefined. In particular, only a few studies have considered micronutrients and ovarian cancer risk. Some of these included an evaluation of vitamin A [5, 6] or its precursors, mainly beta-carotene [7-9], and showed, in general, inverse relations. No material association between beta-carotene and ovarian cancer was observed, however, in other studies [10, 11]. A population-based case-control study in Massachussetts and Wisconsin found no significant relation with vitamin A or beta-carotene, but a significant inverse trend with lutein/zeaxanthin [12]. Other micronutrients considered were vitamins C, D and E, thiamin, nboflavin, folate and calcium [6, 10, 11], but no consistent association emerged. Most studies, however, were relatively small in size. Some of them were not specifically focused on diet, and hence were not based on validated food frequency questionnaires [5, 7, 9, 10], nor did they include allowance for total energy [5, 7, 9] or other relevant dietrelated factors With reference to serum micronutrients, a prospec-
0 6, 95% CI 0.5-0.8), beta-carotene (OR = 0.8, 95% CI 0.610), lutein/zeaxanthin (OR = 0 6, 95% CI 0 5-0.8 for the highest vs the lowest quintile of intake), and calcium intake (OR = 0 7, 95% CI 0 6-10) When the combined effect of calcium and vitamin E was considered, the OR reached 0 4 (95% CI 0 3-0 7) for subjects in the highest compared to those in the lowest intake tertile of both micronutrients Results were consistent across strata of menopausal status, parity and family history of ovarian or breast cancer. Conclusions The intake of selected micronutrients, which were positively correlated to a diet rich in vegetables and fruits, was inversely associated with ovarian cancer.
Key words cancer, micronutrients, ovary
tive study including 35 cases of ovarian cancer and 67 controls [13] showed a significant inverse relation with selenium and a direct one with cholesterol, but no material association with retinol or carotenoids. This article provides further insight on the relation between intakes of several micronutrients, vitamins, and minerals and ovarian cancer, using data from a large case-control study conducted in Italy and based on a validated food frequency questionnaire
Patients and methods A case-control study on ovarian cancer was conducted between January 1992 and September 1999 in five Italian areas the Provinces of Pordenone and Padua, greater Milan, in northern Italy, Latina. central Italy, and Naples in southern Italy [14] Cases were patients with a first histologically confirmed epithelial ovarian carcinoma diagnosed within one year prior lo interview, identified in major teaching and general hospitals in the areas under surveillance Borderline ovarian tumours were not included A total of 1031 cases (median age 56 years, range 18-79) were included in the present analyses Controls were patients admitted for a wide spectrum of acute conditions to major hospitals in the study areas None of them had been admitted for malignant neoplasms, gynaecological or endocnnological conditions, digestive diseases, or any medical condition usso-
1590 ciated with long-term modification of diet A total of 2411 controls (median age 57 years, range 17-79), comparable to cases with reference to age quinquennia and study center, were interviewed The main diagnostic categories were trauma, mostly sprains and fractures (26%), other orthopaedic disorders, such as lower-back and disc disorders (28%), acute surgical conditions (15%), and other illnesses, such as eye diseases, ear, nose, skin and dental conditions (31%) In each study center, the same structured questionnaire and coding manual were used Interviewers were centrally trained and routinely supervised All interviews were conducted in hospital Interviewing nurses were introduced to patients by attending clinical staff, thus allowing to achieve a participation rate over 95% for both cases and controls The questionnaire included information on socio-demographic characteristics, anthropometnc measures (from which body mass index, kg/m 2 , was derived), history of selected diseases, reproductive and menstrual history, occupational physical activity at age 30-39 [15], use of oral contraceptives and other female hormone preparations An interviewer-administered food frequency questionnaire (FFQ) was used to assess the usual diet during the two years preceding diagnosis or hospital admission for the controls and, therefore, the intake of total energy as well as macro- and micronutnents The FFQ included 78 foods, groups or recipes divided into seven sections (I) bread, cereals and first courses, (n) second courses (e g , meat and other main dishes), (in) side dishes (l e vegetables), (IV) fruits, (v) sweets, desserts and soft drinks, (vi) milk, hot beverages and sweeteners, (vn) alcoholic beverages For vegetables and fruits subject to seasonal variation, consumption in season and the corresponding duration were elicited At the end of each section, one or two open questions were used to report foods not included in the questionnaire but eaten at least once per week For 40 food items, the portion was defined in 'natural' units (e g one teaspoon of sugar, one egg), while for the remainder it was defined as small, average, or large with the help of pictures Dietary supplements were not considered, given the low levels of consumption by this population To compute micronutnent intake, an Italian food composition database was used [16] This source, however, had to be integrated with a North-American source regarding glutathione [17] Losses due to cooking were subtracted from the computation of vitamin contents, when appropriate, except for specific carotenoids, since such information was not available The sum of carotenoids (l e beta- and alfacarotene, beta-cryptoxanthin) weighted by their vitamin activity was different from the value of carotene Reproducibihty and validity of the FFQ were satisfactory, with correlation coefficients between 0 5 and 0 7 for most micronutnents [18, 19]
Statistical analysis Odds ratios (OR), and their corresponding 95% CI, were obtained by means of unconditional, multiple, logistic regression models [20] The models included terms for quinquennia of age, study centre, education ( < 7 , 7-11 or 12+ years), body mass index, year of interview (continuous), parity (0, 1-2. 3-4, 5+ births), oral contraceptive use (never, ever), occupational physical activity, and total energy intake Adjustment for energy was made using the residual model [21] Allowance for a measure of fat intake did not materially change any of the estimates [22] Tests for trend were based on the likelihood-ratio test between the models with and without a linear term for each nutrient's quintile [20] The non linear (e g . quadratic) component of the trend was not calculated
Results Table 1 gives the distribution of ovarian cancer cases and control subjects according to age, education, parity and other selected variables Cases tended to be significantly more educated than controls, to have a lower
Table 1 Distribution of 1031 cases of epithelial ovarian cancer and 2411 controls, according to age and selected variables Italy, 1992-1999 Cases
Age (yrs) <45 45-54 55-64 >65 Education" (yrs)
<7 7-11
£12 Parity" Nulliparous 1-2 3 5=4 Menopausal status b Pre-menopause Post-menopause Family history of ovarian and/or breast cancer" No Yes
Controls
n
(%)
183 287 325 236
(178) (27 8) (315) (22 9)
443 615 724 629
(184) (25 5) (30 0) (26 1)
577 227 227
(56 0) (22 0) (22 0)
1442 620 349
(59 8) (25 7) (14 5)
184 572 229 46
(179) (55 5) (22 1) (4 5)
381 1268 639 123
(158) (52 6) (26 5) (5 1)
346 683
(33 6) (66 4)
803 1603
(33 4) (66 6)
902 129
(87 5) (12 5)
2291 120
(95 0) (5 0)
n
(%)
° Significant difference between cases and controls adjusted for age and residence (P < 0 01) The sum does not add up to the total because of a few missing values
parity, and to report more frequently a family history of ovarian and/or breast cancer. Table 2 gives the ORs of ovarian cancer according to subsequent quintiles of intake of various micronutnents compared to the lowest one, together with the tests for linear trend in risk and the estimates from continuous models With reference to minerals, an inverse association was observed for calcium intake (OR = 0.7 for highest vs lowest quintile of intake; 95% CI: 0.6-1.0), and a direct association, of borderline significance, was observed for zinc (OR = 14; 95% CI: 1.0-1.8). No association emerged for other ions, and phosphorus. Most water-soluble vitamins were unrelated to risk. There was a significant inverse association between ovarian cancer risk and several fat-soluble micronutnents, including vitamin E (OR = 0.6; 95% CI. 0.5-0.8) and several carotenoids, i.e. beta-carotene (OR = 0.8; 95% CI- 0.6-1.0) and lutein/zeaxanthin (OR = 0.6; 95% CI: 0 5-0 8), but not for alpha-carotene, lycopene and beta-cryptoxanthin Most inverse associations were linear across quintiles, with significant trends in risk. The relation between various micronutnents and ovarian cancer risk was further examined in separate strata of menopausal status, parity, and family history of ovarian or breast cancer. No consistent heterogeneity emerged between nulliparous and parous women, or between women without and those with history of ovarian/breast cancer. Most associations were apparently stronger in post- than in pre-menopause Since
1591 Table 2 Odds ratios (OR)Q of ovarian cancer and corresponding 95% confidence intervals (95% Cl) according to intake of selected micronutrients and minerals Italy, 1992-1999 Quintile, OR (95°/ Cl) Micronutnent
Minerals Calcium (g) Potassium (g) Phosphorus(g) Iron (mg) Zinc (mg) Water-soluble vitamins Thiamin (mg) Riboflavin (mg) Vitamin C (mg) Vitamin B6 (mg) Folic acid (meg) Niacin (mg) Fat-soluble vitamins, carotenois and glutathione Retinol (mg) Carotene (mg) oe-carotene (mg) (3-carotene (mg) p-cnptoxanthin (mg) Lutein/zeaxanthin (mg) Lycopene (mg) Vitamin D (meg) Vitamin E (mg) Glutathione (mg) Reduced glutathione (mg)
Meanb
Standard deviation
2
3
4
5
x2
P-value
1 lowest)''
1 0 36 1 4 128 11 9
04 1 1 04 40 3 5
0 9(0 7-1 2) 1 0 (0 7-1 3) 1 1 (0 9-15) 1 0 (0 8-1 3) 1 1 (0 8-14)
0 8 ( 0 6-1 1) 0 8 ( 0 6-1 1) 1 1 (0 8-1 4) 1 0(0 7-1 3) 1 0(0 7-1 3)
0 8(0 6-1 0) 1 1 (0 9-1 5) 1 1 (0 8-1 5) 1 0(0 7-1 3) 1 1 (0 8-1 4)
0 7(0 6-1 0) 0 8 ( 0 6-1 1) 0 8 ( 0 6 - 1 1) 1 0(0 8-1 3) 1 4(1 0-1 8)
6.31 0 82 1 67 001 3 69
001 0 37 0 20 0 92 0 05
08 1 5 1462 1 8 256 0 164
03 05 78 6 05 83 5 49
1 1 (0 8-14) 1 2(0 9-1 6) 1 0(0 8-1 3) 0 9(0 6-1 1) 10(0 7-1 3) 1 1 (0 9-15)
13(10-17) 1 0(0 8-1 4) 1 2(0 9-1 6) 0 9(0 7-1 2) 1 1 (0 8-1 4) 1 2(0 9-1 5)
1 2 (0 9-1 6) 0 9 (0 7-1 2) 0 9(0 7-1 2) 1 1 (0 8-1 4) 1 0(0 7-1 3) 0 9(0 7-1 2)
1 1 (0 8-1 5) 1 0(0 8-1 4) 1 1 (0 8-1 4) 1 1 (0 8-1 4) 1 0(0 7-1 3) 0 9(0 7-1 2)
1 04 0 49 0 10 1 79 0 07 1 89
0 31 0 48 0 75 0 18 0 79 0 17
07 43 08 49 04 48 63 29 140 61 0 46 5
09 22 07 25 04 26 3 7 1 3 60 21 8 167
0 9 (0 6-1 1) 1 0 (0 8-1 3) 0 9 (0 7-1 1) 0 9 (0 7-1 2) 1 1 (0 8-14) 1 1 (0 8-14) 1 0(0 7-1 3) 0 8(0 6-1 1) 0 8(0 6-1 1) 1 0(0 8-1 3) 0 8 (0 6-1 0)
0 9(0 6-1 2) 0 9(0 7-1 1) 0 8(0 6-1 0) 0 8(0 6-1 0) 1 1 (0 8-1 4) 0 9(0 7-1 2) 1 0(0 7-1 3) 0 9(0 7-1 2) 0 6 (0 4-0 7) 0 8(0 6-1 1) 0 7(0 6-1 0)
0 8 (0 6-1 0) 0 8 ( 0 6-1 1) 0 7 (0 5-0 9) 0 6 (0 5-0 8) 1 0 (0 7-1 3) 0 8(0 6-1 0) 1 1 (0 8-1 4) 0 9 (0 7-1 2) 0 7(0 5-0 9) 0 9(0 7-1 2) 0 8(0 6-1 1)
1 1 (0 8-1 4) 0 15 0 8 ( 0 6-1 1) 4 43 0 9(0 7-1 2) 2 62 0 8 (0 6-1 0) 7 63 1 2(0 9-1 6) 1 06 0 6 (0 5-0 8) 1471 1 1 (0 9-1 5) 1.35 0 7 (0 6-1 0) 2 28 0 6 (0 5-0 8) 13 30 1 0(0 8-1 3) 013 0 9(0 7-1 2) 0 29
0 70 004 0 II <00l 0 30 <001 0 25 0 13 <001 0 72 0 59
" Estimates from multiple logistic regression models including terms for age (quinquennia), study center, year of interview, education, body mass index, parity, oral contraceptive use, occupational physical activity, and energy intake, according to the residual model b Among controls, per day c Reference category
about 2/3 of cases and controls were in post-menopause, the estimates are more stable for post- than pre-menopausal women None of the interaction terms, however, were significant. When selected micronutnents significantly related to ovarian cancer were simultaneously included in the same model, the associations become less strong due to the co-linearity among several micronutnents. For instance, beta-carotene and lutein/zeaxanthin displayed a Spearman correlation coefficient of 0.8, while vitamin E displayed values around 0.6 with beta-carotene and lutein/zeaxanthin. The other correlation coefficients were about 0.4. However, the inverse relationship between ovarian cancer risk and calcium, lutein/zeaxanthin and vitamin E remained significant When the combined effect of calcium and vitamin E was considered, the ORs decreased across increasing intake levels of calcium and vitamin E, reaching 0.4 (95% Cl: 0.3-0.7) for subjects in the highest compared to those in the lowest intake tertile of both micronutrients. A similar protection pattern emerged for the combination of calcium and beta-carotene with an OR of 0.5 (95% Cl: 0.3-0.7) for subjects in the highest tertile of both micronutnents.
Discussion Our data supports the hypothesis that dietary carotenoids and vitamin E are inversely associated with epithelial ovarian cancer. The present data are therefore in broad agreement with case-control studies conducted in Italy [7], Greece [6] and the US [8, 9], which showed reduced risk of ovarian cancer associated with high intakes of beta-carotene, other carotenoids, vitamins E and C, or other antioxidants. The results of some previous studies, including those of the Iowa Women's Health Study [11] were, however, largely negative, possibly due to the small number of ovarian cancer cases considered, and of different levels or ranges of micronutnent intake with respect to this Italian population [12]. Vitamin E was mainly derived in our present study from olive oil used in vegetable seasoning [23], and has been shown to protect against breast [24, 25], colorectal [26] and oral [27, 28] cancerogenesis Furthermore, carotenoids and a fraction of vitamin E are derived from fruits and various vegetables, which are commonly used in the Italian diet [29]. The three main sources of carotenoids were carrots, green salad and other greens such as spinaches [23] In a previous analy-
1592 sis of foods, the ORs for the highest tertile relative to the lowest one, were 0.5 for raw vegetables, 0.6 for cooked vegetables and 0 9 for fruits [30]. It is unclear, therefore, whether the protection conveyed by beta-carotene, other carotenoids, and vitamin E is partly or largely related to other constituents of vegetable oils, fruits, vegetables or their correlates Calcium has been inversely related to the hyperproliferative and cancer-promoting effect of high dietary fat intake [31], by limiting the absorption and decreasing the serum levels of specific fatty acids. These mechanisms may be relevant to ovarian cancer risk, but the issue remains open to discussion [7, 10, 32, 33]. The data suggest that the influences of calcium and vitamin E, which has essentially an antioxidant effect, are independent, and apparently multiplicative on ovarian cancer Women in the upper tertile of intake of these micronutrients were at approximately 60% reduced ovarian cancer risk. With reference to potential recall and selection bias in the present study, awareness of any particular dietary hypothesis in ovarian cancer aetiology was very limited in the Italian public, since the issue had not received widespread media attention. It is, however, possible that dietary habits of hospital controls may differ from those of the general population. By study design, great attention was paid to excluding all diagnoses that might have been associated causally with or have determined special dietary habits of controls. Moreover, the comparability of dietary history between cases and controls should have been improved by interviewing all subjects in the same hospital setting. Adjustment for total energy intake should have reduced potential bias due to differential over- or under-reporting of food intakes The greatest strength of this study is due to its uniquely large dataset which allowed us to obtain reasonably precise risk estimates and to identify significant associations for several micronutrients. Other strengths are due to the consistency of findings when major categories of controls were used separately and to reliance on a validated food frequency questionnaire [18, 19]. The assessment of a broad range of micronutrients was based on a comprehensive Italian food composition database [16] Supplementation or food fortification with vitamins or minerals is rare within the Italian population [29]. Thus, our estimates of ovarian cancer risk according to dietary micronutnent intakes should not have been materially biased or diluted by intake of vitamin supplements Finally, our present study took advantage of a nearly complete participation of identified cases and controls, and the substantial heterogeneity of the studied populations. In conclusion, the findings of the present study support the hypothesis that intakes of a few micronutrients are inversely related to epithelial ovarian cancer In particular, in terms of population attributable risk [34, 35], low intakes of calcium accounted for an attributable risk of about 10% of cases while low intakes of carotenoids, lutein/zeaxanthin and vitamin E (l e mainly anti-
oxidants) together accounted for about 30% of cases. Whether this reflects a specific effect of these micronutrients, or the generally favourable influence of a diet rich in olive oil and vegetables [27, 30] remains, however, outside the scope of observational epidemiological studies.
Acknowledgements The authors wish to thank Mrs O. Volpato for study coordination and Mrs L. Mei and Mrs I. Calderan for editorial assistance The contribution from the Italian Association for Research on Cancer, Milan, is gratefully acknowledged.
References 1
2
3
4
5
6
7
8 9 10 11 12
13
14
15
16
17
Haenszel W, Kunhara M Studies of Japanese migrants I Mortality from cancer and other diseases among Japanese in the United States J Natl Cancer Inst 1968,40 43-68 Armstrong B, Doll R Environment factors and cancer incidence and mortality in different countries, with special reference to dietary practices lnt J Cancer 1975, 15 617-31 Decarh A, La Vecchia C Environmental factors and cancer mortality in Italy Correlational exercise. Oncology 1986, 43 116-26 Rose PR, Boyar AP, Wynder EL International comparisons of mortality rates for cancer of the breast, ovary, prostate, and colon, and per capita food consumptron Cancer 1986, 58 2363— 71 Byers T, Marshall J, Graham S et al A case-control study of dietary and non-dietary factors in ovarian cancer J Natl Cancer Inst 1983, 71 681-6 Tzonou A, Hsieh C-C, Polychronopoulou A et al Diet and ovarian cancer A case-control study in Greece lnt J Cancer 1993,55 411—4 La Vecchia C, Decarh A, Negn E et al Dietary factors and the risk of epithelial ovarian cancer J Natl Cancer Inst 1987, 79 663-9 Slattery M, Schuman KL, West DW et al Nutrient intake and ovarian cancer Am J Epidemiol 1989, 130 497-502 Engle A, Muscat JE, Harris RE Nutritional risk factors and ovanan cancer Nutr Cancer 1991, 15 239-47 Shu XO, Gao YT, Yuan JM et al Dietary factors and epithelial ovanan cancer Br J Cancer 1989. 59 92-6 Kushi LH, Mink PJ, Follsom AR et al Prospective study of diet and ovanan cancer Am J Epidemiol 1999, 149 21-31 Bertone ER, Hankinson SE, Newcomb PA et al A populationbased case-control study of carotenoid and vitamin A intake and ovanan cancer (United States) Cancer Causes Control 2001, 12 83-90 Helzlsouer KJ. Alberg AJ, Norlcus EP et al Prospective study of serum micronutrients and ovanan cancer J Natl Cancer Inst 1996,88 32-7 Chiaffanno F, Pelucchi C, Parazzini F et al Reproductive and hormonal factors and ovarian cancer Ann Oncol 2001, 12 33741 Tavani A, Gallus S, La Vecchia C et al Physical activity and risk of ovarian cancer An Italian case-control study lnt J Cancer 2001,91 407-11 Salvini S, Parpinel M, Gnagnarella P et al Banca dati di composizione degli ahmenti per studi epidemiologici in Italia Milano Istituto Europeo di Oncologia 1998 Jones DP, Coates RJ. Flagg E et al Glutathione in foods listed in
1593 the National Cancer Institute's Health Habits and History Food Frequency Questionnaire Nutr Cancer 1992, 17 57-75 18 Franceschi S, Barbone F, Negn E et al Reproducibility of an Italian food frequency questionnaire for cancer studies Results from specific nutrients Ann Epidemiol 1995, 5 69-75 19 Decarh A, Franceschi S, Ferraroni M et al Validation of a food frequency questionnaire to assess dietary intakes in cancer studies in Italy Results for specific nutrients Ann Epidemiol 1996, 6 110-8 20 Breslow NE, Day NE Statistical methods in cancer research Vol I The analysis of case-control studies IARC Scientific Publications No 32 Lyon International Agency for Research on Cancer, 1980 21 Willett WC, Stampfer MJ Total energy intake Implications for epidemiologic analysis A m J Epidemiol 1986, 124 17-27 22 Decarh A, Favero A, La Vecchia C et al Macronutnents, energy intake, and breast cancer risk Implications from different models Epidemiology 1997, 8 425-8 23 Favero A, Salvini S, Russo A et al Sources of macro- and micronutrients in Italian women Results from a food-frequency questionnaire for cancer studies EurJ Cancer Prev 1997, 6 277-87 24 La Vecchia C, Negn E, Franceschi S et al Olive oil, other dietary fats, and the risk of breast cancer Cancer Causes Control 1995, 6 545-50 25 Tnchopoulou A Olive oil and breast cancer Cancer Causes Control 1995, 6 475-6 26 Braga C, La Vecchia C, Franceschi S et al Olive oil, other seasoning fats, and the risk of colorectal carcinoma Cancer 1998,82 448-53 27. Lipworth L, Martinez ME, Angell J et al Olive oil and human cancer An assessment of the evidence Prev Med 1997, 26 18190
28
29 30 31 32
33
34
35
Franceschi S, Favero A, Conti E et al Food groups, oils and butter, and cancer of the oral cavity and pharynx Br J Cancer 1999, 80 614-20 La Vecchia C, Tavani A Fruit and vegetables, and human cancer Eur J Cancer Prev 1998, 7 3-8 Bosetti C, Negn E, Franceschi S et al Diet and ovarian cancer risk A case-control study in Italy Int J Cancer 2001, 93 911-5 Newmark HL, Lipkin M Calcium, vitamin D and colon cancer Cancer Res 1992, 52 (Suppl) 2067s-70s Cramer DW, Welch WR, Hutchinson GB et al Dietury animal fat in relation to ovarian cancer risk Obstet Gynecol 1984, 63 833-8 Risen HA, Jain M, Marrett LD, Howe GR Dietary fat intake and risk of epithelial ovarian cancer J Natl Cancer Inst 1994. 86 1409-15 Bruzzi P, Sylvan BG, Byar DP et al Estimating the population attnbutable risk for multiple risk factors using case-control data A m J Epidemiol 1985, 122 904-14 Mezzetti M, Ferraroni M, D«carli A et al Software for attributable risk and confidence interval estimation in case-control studies Comput Biomed Res 1996, 29 63-75
Received 2 April 2001, accepted 8 June 2001 Correspondence to E Bidoh, MD Servizio di Epidemiologia Centro di Rifenmento Oncologico Via Pedemontana Occ 33081 Aviano Italy E-mail epidemiology@cro it
Original article Micronutrients and ovarian cancer: A case-control study in Italy E. Bidoli,1 C. LaVecchia,23 R.Talamini,1 E. Negri,2 M. Parpinel,1 E. Conti,4 M. Montella,5 A. Carbone6 & S. Franceschi7 ] Servizio di Epidemiologia, Centro di Rifenmento Oncologwo. Aviano. 2lstiluto di Ricerche Farmacologiche 'Mario Negri', Milan, ^ Istituto di Statistica Medica e Bwmetna, Umversita degli Studi di Milano, Milan, 4Servizio di Epidemiologia e Oncogenesi, Istituto Regina Elena, Rome. ! Serviziodi Epidemiologia, Istituto Tumori 'Fondaztone Pascale', Naples, ^Division of Human Pathology, Centro di Rifenmenlo Oncologico. Aviano, Italy, 1Field and Intervention Studies Unit, International Agency for Research on Cancer. Lyon, France
Summary Background The role of selected micronutnents, vitamins and minerals in the aetiology of epithelial ovarian cancer was investigated using data from a case-control study conducted between 1992 and 1999 in five Italian areas Patients and methods Cases were 1031 patients with histologically confirmed incident epithelial ovarian cancer Controls were 2411 subjects admitted for acute, non-neoplastic diseases to major hospitals in the same catchment areas Dietary habits were elicited using a validated food frequency questionnaire including 78 food groups and recipes Odds ratios (OR) and 95% confidence intervals (95% CI) were computed by quintiles of intake of nutrients Results Inverse associations emerged for vitamin E (OR =
Introduction A role of dietary factors on ovarian carcinogenesis was originally suggested on the basis of descriptive epidemiology and correlational studies, [1-4], but it remains largely undefined. In particular, only a few studies have considered micronutrients and ovarian cancer risk. Some of these included an evaluation of vitamin A [5, 6] or its precursors, mainly beta-carotene [7-9], and showed, in general, inverse relations. No material association between beta-carotene and ovarian cancer was observed, however, in other studies [10, 11]. A population-based case-control study in Massachussetts and Wisconsin found no significant relation with vitamin A or beta-carotene, but a significant inverse trend with lutein/zeaxanthin [12]. Other micronutrients considered were vitamins C, D and E, thiamin, nboflavin, folate and calcium [6, 10, 11], but no consistent association emerged. Most studies, however, were relatively small in size. Some of them were not specifically focused on diet, and hence were not based on validated food frequency questionnaires [5, 7, 9, 10], nor did they include allowance for total energy [5, 7, 9] or other relevant dietrelated factors With reference to serum micronutrients, a prospec-
0 6, 95% CI 0.5-0.8), beta-carotene (OR = 0.8, 95% CI 0.610), lutein/zeaxanthin (OR = 0 6, 95% CI 0 5-0.8 for the highest vs the lowest quintile of intake), and calcium intake (OR = 0 7, 95% CI 0 6-10) When the combined effect of calcium and vitamin E was considered, the OR reached 0 4 (95% CI 0 3-0 7) for subjects in the highest compared to those in the lowest intake tertile of both micronutrients Results were consistent across strata of menopausal status, parity and family history of ovarian or breast cancer. Conclusions The intake of selected micronutrients, which were positively correlated to a diet rich in vegetables and fruits, was inversely associated with ovarian cancer.
Key words cancer, micronutrients, ovary
tive study including 35 cases of ovarian cancer and 67 controls [13] showed a significant inverse relation with selenium and a direct one with cholesterol, but no material association with retinol or carotenoids. This article provides further insight on the relation between intakes of several micronutrients, vitamins, and minerals and ovarian cancer, using data from a large case-control study conducted in Italy and based on a validated food frequency questionnaire
Patients and methods A case-control study on ovarian cancer was conducted between January 1992 and September 1999 in five Italian areas the Provinces of Pordenone and Padua, greater Milan, in northern Italy, Latina. central Italy, and Naples in southern Italy [14] Cases were patients with a first histologically confirmed epithelial ovarian carcinoma diagnosed within one year prior lo interview, identified in major teaching and general hospitals in the areas under surveillance Borderline ovarian tumours were not included A total of 1031 cases (median age 56 years, range 18-79) were included in the present analyses Controls were patients admitted for a wide spectrum of acute conditions to major hospitals in the study areas None of them had been admitted for malignant neoplasms, gynaecological or endocnnological conditions, digestive diseases, or any medical condition usso-
1590 ciated with long-term modification of diet A total of 2411 controls (median age 57 years, range 17-79), comparable to cases with reference to age quinquennia and study center, were interviewed The main diagnostic categories were trauma, mostly sprains and fractures (26%), other orthopaedic disorders, such as lower-back and disc disorders (28%), acute surgical conditions (15%), and other illnesses, such as eye diseases, ear, nose, skin and dental conditions (31%) In each study center, the same structured questionnaire and coding manual were used Interviewers were centrally trained and routinely supervised All interviews were conducted in hospital Interviewing nurses were introduced to patients by attending clinical staff, thus allowing to achieve a participation rate over 95% for both cases and controls The questionnaire included information on socio-demographic characteristics, anthropometnc measures (from which body mass index, kg/m 2 , was derived), history of selected diseases, reproductive and menstrual history, occupational physical activity at age 30-39 [15], use of oral contraceptives and other female hormone preparations An interviewer-administered food frequency questionnaire (FFQ) was used to assess the usual diet during the two years preceding diagnosis or hospital admission for the controls and, therefore, the intake of total energy as well as macro- and micronutnents The FFQ included 78 foods, groups or recipes divided into seven sections (I) bread, cereals and first courses, (n) second courses (e g , meat and other main dishes), (in) side dishes (l e vegetables), (IV) fruits, (v) sweets, desserts and soft drinks, (vi) milk, hot beverages and sweeteners, (vn) alcoholic beverages For vegetables and fruits subject to seasonal variation, consumption in season and the corresponding duration were elicited At the end of each section, one or two open questions were used to report foods not included in the questionnaire but eaten at least once per week For 40 food items, the portion was defined in 'natural' units (e g one teaspoon of sugar, one egg), while for the remainder it was defined as small, average, or large with the help of pictures Dietary supplements were not considered, given the low levels of consumption by this population To compute micronutnent intake, an Italian food composition database was used [16] This source, however, had to be integrated with a North-American source regarding glutathione [17] Losses due to cooking were subtracted from the computation of vitamin contents, when appropriate, except for specific carotenoids, since such information was not available The sum of carotenoids (l e beta- and alfacarotene, beta-cryptoxanthin) weighted by their vitamin activity was different from the value of carotene Reproducibihty and validity of the FFQ were satisfactory, with correlation coefficients between 0 5 and 0 7 for most micronutnents [18, 19]
Statistical analysis Odds ratios (OR), and their corresponding 95% CI, were obtained by means of unconditional, multiple, logistic regression models [20] The models included terms for quinquennia of age, study centre, education ( < 7 , 7-11 or 12+ years), body mass index, year of interview (continuous), parity (0, 1-2. 3-4, 5+ births), oral contraceptive use (never, ever), occupational physical activity, and total energy intake Adjustment for energy was made using the residual model [21] Allowance for a measure of fat intake did not materially change any of the estimates [22] Tests for trend were based on the likelihood-ratio test between the models with and without a linear term for each nutrient's quintile [20] The non linear (e g . quadratic) component of the trend was not calculated
Results Table 1 gives the distribution of ovarian cancer cases and control subjects according to age, education, parity and other selected variables Cases tended to be significantly more educated than controls, to have a lower
Table 1 Distribution of 1031 cases of epithelial ovarian cancer and 2411 controls, according to age and selected variables Italy, 1992-1999 Cases
Age (yrs) <45 45-54 55-64 >65 Education" (yrs)
<7 7-11
£12 Parity" Nulliparous 1-2 3 5=4 Menopausal status b Pre-menopause Post-menopause Family history of ovarian and/or breast cancer" No Yes
Controls
n
(%)
183 287 325 236
(178) (27 8) (315) (22 9)
443 615 724 629
(184) (25 5) (30 0) (26 1)
577 227 227
(56 0) (22 0) (22 0)
1442 620 349
(59 8) (25 7) (14 5)
184 572 229 46
(179) (55 5) (22 1) (4 5)
381 1268 639 123
(158) (52 6) (26 5) (5 1)
346 683
(33 6) (66 4)
803 1603
(33 4) (66 6)
902 129
(87 5) (12 5)
2291 120
(95 0) (5 0)
n
(%)
° Significant difference between cases and controls adjusted for age and residence (P < 0 01) The sum does not add up to the total because of a few missing values
parity, and to report more frequently a family history of ovarian and/or breast cancer. Table 2 gives the ORs of ovarian cancer according to subsequent quintiles of intake of various micronutnents compared to the lowest one, together with the tests for linear trend in risk and the estimates from continuous models With reference to minerals, an inverse association was observed for calcium intake (OR = 0.7 for highest vs lowest quintile of intake; 95% CI: 0.6-1.0), and a direct association, of borderline significance, was observed for zinc (OR = 14; 95% CI: 1.0-1.8). No association emerged for other ions, and phosphorus. Most water-soluble vitamins were unrelated to risk. There was a significant inverse association between ovarian cancer risk and several fat-soluble micronutnents, including vitamin E (OR = 0.6; 95% CI. 0.5-0.8) and several carotenoids, i.e. beta-carotene (OR = 0.8; 95% CI- 0.6-1.0) and lutein/zeaxanthin (OR = 0.6; 95% CI: 0 5-0 8), but not for alpha-carotene, lycopene and beta-cryptoxanthin Most inverse associations were linear across quintiles, with significant trends in risk. The relation between various micronutnents and ovarian cancer risk was further examined in separate strata of menopausal status, parity, and family history of ovarian or breast cancer. No consistent heterogeneity emerged between nulliparous and parous women, or between women without and those with history of ovarian/breast cancer. Most associations were apparently stronger in post- than in pre-menopause Since
1591 Table 2 Odds ratios (OR)Q of ovarian cancer and corresponding 95% confidence intervals (95% Cl) according to intake of selected micronutrients and minerals Italy, 1992-1999 Quintile, OR (95°/ Cl) Micronutnent
Minerals Calcium (g) Potassium (g) Phosphorus(g) Iron (mg) Zinc (mg) Water-soluble vitamins Thiamin (mg) Riboflavin (mg) Vitamin C (mg) Vitamin B6 (mg) Folic acid (meg) Niacin (mg) Fat-soluble vitamins, carotenois and glutathione Retinol (mg) Carotene (mg) oe-carotene (mg) (3-carotene (mg) p-cnptoxanthin (mg) Lutein/zeaxanthin (mg) Lycopene (mg) Vitamin D (meg) Vitamin E (mg) Glutathione (mg) Reduced glutathione (mg)
Meanb
Standard deviation
2
3
4
5
x2
P-value
1 lowest)''
1 0 36 1 4 128 11 9
04 1 1 04 40 3 5
0 9(0 7-1 2) 1 0 (0 7-1 3) 1 1 (0 9-15) 1 0 (0 8-1 3) 1 1 (0 8-14)
0 8 ( 0 6-1 1) 0 8 ( 0 6-1 1) 1 1 (0 8-1 4) 1 0(0 7-1 3) 1 0(0 7-1 3)
0 8(0 6-1 0) 1 1 (0 9-1 5) 1 1 (0 8-1 5) 1 0(0 7-1 3) 1 1 (0 8-1 4)
0 7(0 6-1 0) 0 8 ( 0 6-1 1) 0 8 ( 0 6 - 1 1) 1 0(0 8-1 3) 1 4(1 0-1 8)
6.31 0 82 1 67 001 3 69
001 0 37 0 20 0 92 0 05
08 1 5 1462 1 8 256 0 164
03 05 78 6 05 83 5 49
1 1 (0 8-14) 1 2(0 9-1 6) 1 0(0 8-1 3) 0 9(0 6-1 1) 10(0 7-1 3) 1 1 (0 9-15)
13(10-17) 1 0(0 8-1 4) 1 2(0 9-1 6) 0 9(0 7-1 2) 1 1 (0 8-1 4) 1 2(0 9-1 5)
1 2 (0 9-1 6) 0 9 (0 7-1 2) 0 9(0 7-1 2) 1 1 (0 8-1 4) 1 0(0 7-1 3) 0 9(0 7-1 2)
1 1 (0 8-1 5) 1 0(0 8-1 4) 1 1 (0 8-1 4) 1 1 (0 8-1 4) 1 0(0 7-1 3) 0 9(0 7-1 2)
1 04 0 49 0 10 1 79 0 07 1 89
0 31 0 48 0 75 0 18 0 79 0 17
07 43 08 49 04 48 63 29 140 61 0 46 5
09 22 07 25 04 26 3 7 1 3 60 21 8 167
0 9 (0 6-1 1) 1 0 (0 8-1 3) 0 9 (0 7-1 1) 0 9 (0 7-1 2) 1 1 (0 8-14) 1 1 (0 8-14) 1 0(0 7-1 3) 0 8(0 6-1 1) 0 8(0 6-1 1) 1 0(0 8-1 3) 0 8 (0 6-1 0)
0 9(0 6-1 2) 0 9(0 7-1 1) 0 8(0 6-1 0) 0 8(0 6-1 0) 1 1 (0 8-1 4) 0 9(0 7-1 2) 1 0(0 7-1 3) 0 9(0 7-1 2) 0 6 (0 4-0 7) 0 8(0 6-1 1) 0 7(0 6-1 0)
0 8 (0 6-1 0) 0 8 ( 0 6-1 1) 0 7 (0 5-0 9) 0 6 (0 5-0 8) 1 0 (0 7-1 3) 0 8(0 6-1 0) 1 1 (0 8-1 4) 0 9 (0 7-1 2) 0 7(0 5-0 9) 0 9(0 7-1 2) 0 8(0 6-1 1)
1 1 (0 8-1 4) 0 15 0 8 ( 0 6-1 1) 4 43 0 9(0 7-1 2) 2 62 0 8 (0 6-1 0) 7 63 1 2(0 9-1 6) 1 06 0 6 (0 5-0 8) 1471 1 1 (0 9-1 5) 1.35 0 7 (0 6-1 0) 2 28 0 6 (0 5-0 8) 13 30 1 0(0 8-1 3) 013 0 9(0 7-1 2) 0 29
0 70 004 0 II <00l 0 30 <001 0 25 0 13 <001 0 72 0 59
" Estimates from multiple logistic regression models including terms for age (quinquennia), study center, year of interview, education, body mass index, parity, oral contraceptive use, occupational physical activity, and energy intake, according to the residual model b Among controls, per day c Reference category
about 2/3 of cases and controls were in post-menopause, the estimates are more stable for post- than pre-menopausal women None of the interaction terms, however, were significant. When selected micronutnents significantly related to ovarian cancer were simultaneously included in the same model, the associations become less strong due to the co-linearity among several micronutnents. For instance, beta-carotene and lutein/zeaxanthin displayed a Spearman correlation coefficient of 0.8, while vitamin E displayed values around 0.6 with beta-carotene and lutein/zeaxanthin. The other correlation coefficients were about 0.4. However, the inverse relationship between ovarian cancer risk and calcium, lutein/zeaxanthin and vitamin E remained significant When the combined effect of calcium and vitamin E was considered, the ORs decreased across increasing intake levels of calcium and vitamin E, reaching 0.4 (95% Cl: 0.3-0.7) for subjects in the highest compared to those in the lowest intake tertile of both micronutrients. A similar protection pattern emerged for the combination of calcium and beta-carotene with an OR of 0.5 (95% Cl: 0.3-0.7) for subjects in the highest tertile of both micronutnents.
Discussion Our data supports the hypothesis that dietary carotenoids and vitamin E are inversely associated with epithelial ovarian cancer. The present data are therefore in broad agreement with case-control studies conducted in Italy [7], Greece [6] and the US [8, 9], which showed reduced risk of ovarian cancer associated with high intakes of beta-carotene, other carotenoids, vitamins E and C, or other antioxidants. The results of some previous studies, including those of the Iowa Women's Health Study [11] were, however, largely negative, possibly due to the small number of ovarian cancer cases considered, and of different levels or ranges of micronutnent intake with respect to this Italian population [12]. Vitamin E was mainly derived in our present study from olive oil used in vegetable seasoning [23], and has been shown to protect against breast [24, 25], colorectal [26] and oral [27, 28] cancerogenesis Furthermore, carotenoids and a fraction of vitamin E are derived from fruits and various vegetables, which are commonly used in the Italian diet [29]. The three main sources of carotenoids were carrots, green salad and other greens such as spinaches [23] In a previous analy-
1592 sis of foods, the ORs for the highest tertile relative to the lowest one, were 0.5 for raw vegetables, 0.6 for cooked vegetables and 0 9 for fruits [30]. It is unclear, therefore, whether the protection conveyed by beta-carotene, other carotenoids, and vitamin E is partly or largely related to other constituents of vegetable oils, fruits, vegetables or their correlates Calcium has been inversely related to the hyperproliferative and cancer-promoting effect of high dietary fat intake [31], by limiting the absorption and decreasing the serum levels of specific fatty acids. These mechanisms may be relevant to ovarian cancer risk, but the issue remains open to discussion [7, 10, 32, 33]. The data suggest that the influences of calcium and vitamin E, which has essentially an antioxidant effect, are independent, and apparently multiplicative on ovarian cancer Women in the upper tertile of intake of these micronutrients were at approximately 60% reduced ovarian cancer risk. With reference to potential recall and selection bias in the present study, awareness of any particular dietary hypothesis in ovarian cancer aetiology was very limited in the Italian public, since the issue had not received widespread media attention. It is, however, possible that dietary habits of hospital controls may differ from those of the general population. By study design, great attention was paid to excluding all diagnoses that might have been associated causally with or have determined special dietary habits of controls. Moreover, the comparability of dietary history between cases and controls should have been improved by interviewing all subjects in the same hospital setting. Adjustment for total energy intake should have reduced potential bias due to differential over- or under-reporting of food intakes The greatest strength of this study is due to its uniquely large dataset which allowed us to obtain reasonably precise risk estimates and to identify significant associations for several micronutrients. Other strengths are due to the consistency of findings when major categories of controls were used separately and to reliance on a validated food frequency questionnaire [18, 19]. The assessment of a broad range of micronutrients was based on a comprehensive Italian food composition database [16] Supplementation or food fortification with vitamins or minerals is rare within the Italian population [29]. Thus, our estimates of ovarian cancer risk according to dietary micronutnent intakes should not have been materially biased or diluted by intake of vitamin supplements Finally, our present study took advantage of a nearly complete participation of identified cases and controls, and the substantial heterogeneity of the studied populations. In conclusion, the findings of the present study support the hypothesis that intakes of a few micronutrients are inversely related to epithelial ovarian cancer In particular, in terms of population attributable risk [34, 35], low intakes of calcium accounted for an attributable risk of about 10% of cases while low intakes of carotenoids, lutein/zeaxanthin and vitamin E (l e mainly anti-
oxidants) together accounted for about 30% of cases. Whether this reflects a specific effect of these micronutrients, or the generally favourable influence of a diet rich in olive oil and vegetables [27, 30] remains, however, outside the scope of observational epidemiological studies.
Acknowledgements The authors wish to thank Mrs O. Volpato for study coordination and Mrs L. Mei and Mrs I. Calderan for editorial assistance The contribution from the Italian Association for Research on Cancer, Milan, is gratefully acknowledged.
References 1
2
3
4
5
6
7
8 9 10 11 12
13
14
15
16
17
Haenszel W, Kunhara M Studies of Japanese migrants I Mortality from cancer and other diseases among Japanese in the United States J Natl Cancer Inst 1968,40 43-68 Armstrong B, Doll R Environment factors and cancer incidence and mortality in different countries, with special reference to dietary practices lnt J Cancer 1975, 15 617-31 Decarh A, La Vecchia C Environmental factors and cancer mortality in Italy Correlational exercise. Oncology 1986, 43 116-26 Rose PR, Boyar AP, Wynder EL International comparisons of mortality rates for cancer of the breast, ovary, prostate, and colon, and per capita food consumptron Cancer 1986, 58 2363— 71 Byers T, Marshall J, Graham S et al A case-control study of dietary and non-dietary factors in ovarian cancer J Natl Cancer Inst 1983, 71 681-6 Tzonou A, Hsieh C-C, Polychronopoulou A et al Diet and ovarian cancer A case-control study in Greece lnt J Cancer 1993,55 411—4 La Vecchia C, Decarh A, Negn E et al Dietary factors and the risk of epithelial ovarian cancer J Natl Cancer Inst 1987, 79 663-9 Slattery M, Schuman KL, West DW et al Nutrient intake and ovarian cancer Am J Epidemiol 1989, 130 497-502 Engle A, Muscat JE, Harris RE Nutritional risk factors and ovanan cancer Nutr Cancer 1991, 15 239-47 Shu XO, Gao YT, Yuan JM et al Dietary factors and epithelial ovanan cancer Br J Cancer 1989. 59 92-6 Kushi LH, Mink PJ, Follsom AR et al Prospective study of diet and ovanan cancer Am J Epidemiol 1999, 149 21-31 Bertone ER, Hankinson SE, Newcomb PA et al A populationbased case-control study of carotenoid and vitamin A intake and ovanan cancer (United States) Cancer Causes Control 2001, 12 83-90 Helzlsouer KJ. Alberg AJ, Norlcus EP et al Prospective study of serum micronutrients and ovanan cancer J Natl Cancer Inst 1996,88 32-7 Chiaffanno F, Pelucchi C, Parazzini F et al Reproductive and hormonal factors and ovarian cancer Ann Oncol 2001, 12 33741 Tavani A, Gallus S, La Vecchia C et al Physical activity and risk of ovarian cancer An Italian case-control study lnt J Cancer 2001,91 407-11 Salvini S, Parpinel M, Gnagnarella P et al Banca dati di composizione degli ahmenti per studi epidemiologici in Italia Milano Istituto Europeo di Oncologia 1998 Jones DP, Coates RJ. Flagg E et al Glutathione in foods listed in
1593 the National Cancer Institute's Health Habits and History Food Frequency Questionnaire Nutr Cancer 1992, 17 57-75 18 Franceschi S, Barbone F, Negn E et al Reproducibility of an Italian food frequency questionnaire for cancer studies Results from specific nutrients Ann Epidemiol 1995, 5 69-75 19 Decarh A, Franceschi S, Ferraroni M et al Validation of a food frequency questionnaire to assess dietary intakes in cancer studies in Italy Results for specific nutrients Ann Epidemiol 1996, 6 110-8 20 Breslow NE, Day NE Statistical methods in cancer research Vol I The analysis of case-control studies IARC Scientific Publications No 32 Lyon International Agency for Research on Cancer, 1980 21 Willett WC, Stampfer MJ Total energy intake Implications for epidemiologic analysis A m J Epidemiol 1986, 124 17-27 22 Decarh A, Favero A, La Vecchia C et al Macronutnents, energy intake, and breast cancer risk Implications from different models Epidemiology 1997, 8 425-8 23 Favero A, Salvini S, Russo A et al Sources of macro- and micronutrients in Italian women Results from a food-frequency questionnaire for cancer studies EurJ Cancer Prev 1997, 6 277-87 24 La Vecchia C, Negn E, Franceschi S et al Olive oil, other dietary fats, and the risk of breast cancer Cancer Causes Control 1995, 6 545-50 25 Tnchopoulou A Olive oil and breast cancer Cancer Causes Control 1995, 6 475-6 26 Braga C, La Vecchia C, Franceschi S et al Olive oil, other seasoning fats, and the risk of colorectal carcinoma Cancer 1998,82 448-53 27. Lipworth L, Martinez ME, Angell J et al Olive oil and human cancer An assessment of the evidence Prev Med 1997, 26 18190
28
29 30 31 32
33
34
35
Franceschi S, Favero A, Conti E et al Food groups, oils and butter, and cancer of the oral cavity and pharynx Br J Cancer 1999, 80 614-20 La Vecchia C, Tavani A Fruit and vegetables, and human cancer Eur J Cancer Prev 1998, 7 3-8 Bosetti C, Negn E, Franceschi S et al Diet and ovarian cancer risk A case-control study in Italy Int J Cancer 2001, 93 911-5 Newmark HL, Lipkin M Calcium, vitamin D and colon cancer Cancer Res 1992, 52 (Suppl) 2067s-70s Cramer DW, Welch WR, Hutchinson GB et al Dietury animal fat in relation to ovarian cancer risk Obstet Gynecol 1984, 63 833-8 Risen HA, Jain M, Marrett LD, Howe GR Dietary fat intake and risk of epithelial ovarian cancer J Natl Cancer Inst 1994. 86 1409-15 Bruzzi P, Sylvan BG, Byar DP et al Estimating the population attnbutable risk for multiple risk factors using case-control data A m J Epidemiol 1985, 122 904-14 Mezzetti M, Ferraroni M, D«carli A et al Software for attributable risk and confidence interval estimation in case-control studies Comput Biomed Res 1996, 29 63-75
Received 2 April 2001, accepted 8 June 2001 Correspondence to E Bidoh, MD Servizio di Epidemiologia Centro di Rifenmento Oncologico Via Pedemontana Occ 33081 Aviano Italy E-mail epidemiology@cro it