Fax: (011) 31-(0) -43-3884128
Consumption of vegetables and fruits and risk of ovarian carcinoma
Results from the Netherlands Cohort Study on Diet and Cancer
Article first published online: 15 AUG 2005
Copyright © 2005 American Cancer Society
Volume 104, Issue 7, pages 1512–1519, 1 October 2005
How to Cite
Mommers, M., Schouten, L. J., Goldbohm, R. A. and van den Brandt, P. A. (2005), Consumption of vegetables and fruits and risk of ovarian carcinoma. Cancer, 104: 1512–1519. doi: 10.1002/cncr.21332
Fax: (011) 31-(0) -43-3884128
- Issue published online: 17 SEP 2005
- Article first published online: 15 AUG 2005
- Manuscript Accepted: 4 MAY 2005
- Manuscript Revised: 21 APR 2005
- Manuscript Received: 22 FEB 2005
- World Cancer Research Fund-UK. Grant Number: 2003/30
- ovarian neoplasm;
- cohort study
To the authors' knowledge, only a few prospective studies to date have investigated the correlation between vegetable and fruit consumption and the risk of ovarian carcinoma and their results have been inconclusive.
Vegetable and fruit intake was assessed in relation to ovarian carcinoma, among 62,573 postmenopausal women participating in The Netherlands Cohort Study on Diet and Cancer. Women reported on dietary habits and on other risk factors for cancer in a self-administered questionnaire in 1986. Follow-up of cancer was implemented by annual record linkage with The Netherlands Cancer Registry and a pathology register. After 11.3 years of follow-up, data regarding 252 incident invasive epithelial ovarian carcinoma cases and of 2216 subcohort members were available for case–cohort analyses.
Multivariable-adjusted rate ratios (RR) of ovarian carcinoma for women in the highest compared with the lowest quintile of intake (RRQ5 vs. Q1) were 0.98 (95% confidence interval [95% CI], 0.61–1.58) for total vegetables and 1.11 (95% CI, 0.70–1.78) for total fruit. The RRQ5 vs. Q1 values of ovarian carcinoma with intake of cooked vegetables, raw vegetables, brassicas, legumes, cooked leafy vegetables, and raw leafy vegetables were 1.35 (95% CI, 0.83–2.21), 0.75 (95% CI, 0.48–1.18), 1.42 (95% CI, 0.88–2.29), 0.93 (95% CI, 0.60–1.44), 1.05 (95% CI, 0.66–1.67), and 1.23 (95% CI, 0.75–2.02), respectively. With the exception of raw endive (multivariable-adjusted RR, 0.24; 95% CI, 0.07–0.78), none of the individual vegetable or fruit items showed a statistically significant association with ovarian carcinoma.
The results of the current study did not support a significant association between vegetable or fruit consumption and ovarian carcinoma risk in a cohort of postmenopausal women. Cancer 2005. © 2005 American Cancer Society.
Ovarian carcinoma ranks as the fifth most common malignancy among women living in Europe, with approximately 34,500 newly diagnosed ovarian carcinoma cases in 1998.1 Approximately 90% of all malignant ovarian tumors in industrialized countries are epithelial tumors.2 To our knowledge, little is still known regarding the etiology of ovarian carcinoma, but international variation in incidence and mortality rates suggests an influence of lifestyle as well as of dietary patterns. Putative models of ovarian carcinoma etiology include Fathalla's incessant ovulation theory3 and the gonadotropin hypothesis proposed by Cramer and Welch,4 and, in accordance with these theories, evidence for a role of reproductive and hormonal factors in the etiology of ovarian carcinoma is consistent. In addition, diet also has been believed to affect ovarian carcinoma risk. In particular, greater intake of vegetables and fruit has been suggested to reduce the risk of ovarian carcinoma. Vegetables and fruit contain a variety of nutrients and may be involved in many processes including modulation of detoxification enzymes, antioxidant activities, stimulation of the immune system, and modulation of cholesterol synthesis and hormone metabolism.5
Evidence for a protective influence of vegetables or fruit consumption on ovarian carcinoma risk has been derived for the most part from case–control studies,6–11 which are typically susceptible to both recall and selection bias. Few prospective studies have been conducted on the subject and the results from these prospective studies do not consistently show protective effects of a high intake of vegetables and fruit. High intake of vegetables was inversely associated with ovarian carcinoma risk in the Swedish cohort study,12 but in both U.S.-based cohort studies, no significant association with high vegetable intake was observed, although the risks tended toward a protective effect.13, 14 No association was found between ovarian carcinoma and fruit consumption in any of the three prospective studies.
It is possible that only specific vegetables or fruits constitute an effect, which may have remained unnoticed when investigating overall groups of vegetables or fruit. In The Netherlands Cohort Study on Diet and Cancer, detailed information was collected concerning a number of specific vegetables and fruits, thereby providing the opportunity to study consumption of specific vegetables and fruits individually in relation to ovarian carcinoma, in addition to overall categories of vegetables and fruit.
MATERIALS AND METHODS
The Netherlands Cohort Study on Diet and Cancer is a prospective cohort study that started in September 1986 with the enrollment of 120,852 subjects with an age range of 55–69 years. A total of 62,573 of the participants were female. The study population originated from 204 municipalities with computerized population registries, located throughout the country.
A detailed description of the study design has been published elsewhere.15 In brief, a case–cohort approach was used for data processing and analysis. Cases were enumerated for the entire cohort, whereas accumulated person-years for the entire cohort were estimated using a subcohort of 5000 participants, 2589 of whom were female, and randomly sampled from the cohort after baseline exposure measurement. The subcohort has been followed up biennially for vital status information and, after 11.3 years of follow-up, information regarding vital status was available for all female subcohort members. After the exclusion of women with prevalent malignancy (other than nonmelanoma skin cancer) at baseline, and women who, at baseline, had reported to have undergone an oophorectomy, 2406 female subcohort members remained available for analyses.
Identification of Cases
Incident cancer cases were identified by computerized record linkage of the entire cohort to the regional cancer registries and the Netherlands Pathology Registry.15, 16 The completeness of cancer follow-up has previously been estimated to be > 95%.17 During 11.3 years of follow-up, 300 incident, microscopically confirmed, primary ovarian tumor cases (International Classification of Diseases for Oncology [ICD-O]-3:C56.9) were identified. After excluding nonepithelial tumors (n = 9) and borderline invasive tumors (n = 9), 282 invasive epithelial tumor cases remained eligible for analysis.
Information regarding dietary habits and potential confounders, such as smoking behavior, reproductive history, and personal and family history of cancer, was collected at baseline from all cohort members with the aid of a self-administered questionnaire. The dietary section consisted of a 150-item semiquantitative food frequency questionnaire used to assess regular food and beverage consumption in the year preceding the start of the study, which had been validated against a 9-day diet record, attaining Spearman correlation coefficients of 0.4 for total vegetable and of 0.6 for total fruit consumption.18 Data were obtained concerning consumption frequency of vegetables (i.e., Brussels sprouts, cauliflower, cabbage [white/green], kale, string beans, broad beans, spinach, endive [raw and cooked], lettuce, carrots [raw and cooked], sweet peppers, sauerkraut, tomatoes, red beets, mushrooms, gherkins, rhubarb, leek, and onions), for summer and winter separately, and fruits (i.e., mandarins, oranges, grapefruits, orange/grapefruit juice, grapes, bananas, apples/pears, and strawberries). The questionnaire covered most vegetables and fruits eaten regularly in 1986, with the exception of chicory, red cabbage, and cucumber. In an open-ended question, participants could enter which other foods they consumed on a regular basis as well as the frequency (number of times per wk) and amount of consumption on each occasion.
Consumption frequency was specified using categories ranging from “never or less than once per month” to “three to seven times per week” for vegetable consumption (summer and winter combined) and to “six or seven times per week” for fruit consumption. In addition, for individual fruit items, the amount consumed on each consumption day was asked. For onions and tomatoes, participants were asked to report the number they usually ate per week and for sweet peppers and mushrooms the number they usually ate per month.
Mean daily vegetable consumption (grams per day) was calculated by multiplying the frequency of consumption and serving size. Frequency of intake and standard weight were used to calculate consumption of individual fruit items in grams per day.
After excluding subjects with incomplete or inconsistent dietary data,18 252 ovarian carcinoma cases and 2216 subcohort members remained for analyses. To check the quality and consistency of response to vegetable items, a vegetable error index was computed. Subjects with more than three errors were excluded from analyses regarding vegetable consumption. As a consequence, analyses regarding vegetable consumption included 240 cases and 2088 subcohort members.
For analyzing overall vegetable or fruit consumption as well as for analyzing consumption of subgroups of vegetables (i.e., cooked vegetables, raw vegetables, brassicas, legumes, cooked leafy vegetables, raw leafy vegetables, and other vegetables) or fruits (citrus fruits), subjects were categorized into quintiles of consumption (grams per day), according to the distribution of intake in the subcohort. Individual vegetable and fruit items were modeled as continuous variables (grams per day [g/day]) and rate ratios (RR) were presented per increment of 25 g/day, corresponding to an approximate consumption frequency of once a week for cooked vegetables.
As potential confounders, we included age (yrs), duration of oral contraceptive use (yrs), and parity (continuous) because they have consistently been found to be associated with ovarian carcinoma, as well as cigarette smoking (current cigarette smoker [yes/no], duration of cigarette smoking [yrs], and number of cigarettes smoked daily [continuous]) because of its relation to vegetable and fruit intake. In addition, we examined if other factors ([post]menopausal hormone therapy, height, weight, body mass index, family history of ovarian or breast carcinoma, hysterectomy, age at menarche, age at menopause, tubal ligation, and socioeconomic status [SES]) were confounders for the association between vegetable and fruit consumption and ovarian carcinoma in our data. Risk factors were considered to be confounders for our data if they were associated with ovarian carcinoma risk, were associated with vegetable and fruit consumption and, after inclusion in the model, changed the association of vegetable and fruit consumption with the risk of ovarian carcinoma by > 10% compared with the model not including this risk factor. Only height (continuous) met these above criteria and was thus considered to be an additional confounder for our data.
Person-years of follow-up were calculated for the female subcohort members from the date they returned the questionnaire (September 1986), until the date of ovarian carcinoma diagnosis, death, or end of follow-up (December 1997).
Incidence RR and corresponding 95% confidence intervals (95% CI) for ovarian carcinoma risk were estimated in age-adjusted and multivariable case–cohort analyses using the Cox proportional hazards model19 processed with the Stata statistical software package (StataCorp, College Station, TX).20 Standard errors were estimated using the robust Huber–White sandwich estimator to account for additional variance introduced by sampling from the cohort.21 The proportional hazards assumption was tested using the scaled Schoenfeld residuals.22 By fitting ordinal exposure variables as continuous terms, tests for dose-response trends in risk of ovarian carcinoma were assessed. Two-sided P values are reported throughout the article and were considered statistically significant at a P value < 0.05.
During 11.3 years of follow-up of 62,573 women, 282 epithelial ovarian carcinoma cases occurred. A total of 252 cases occurred in women with complete and consistent dietary data. Of these 252 epithelial ovarian carcinoma cases, 126 were serous carcinoma, 24 were endometrioid carcinoma, 26 were mucinous carcinoma, 12 were clear cell carcinoma, and 51 were histologically diagnosed as adenocarcinoma not otherwise specified; the histologic subtype of 13 was not specified.
In general, the mean intake of vegetables was comparatively similar between cases and subcohort members (Table 1), although total vegetable consumption was slightly higher among subcohort members. Mean total fruit consumption was somewhat higher in cases compared with the subcohort, most likely owing to a greater intake of bananas among cases. Cases were of more advanced age. For example, at baseline, 36.2% of the cases were ages 65–69 years compared with 28.6% in the subcohort. More of the cases were nulliparous compared with subcohort members (25.0% vs. 18.2%). Conversely, proportionally fewer cases than subcohort members had ever used oral contraceptives (15.6% vs. 24.1%) or had undergone a hysterectomy (8.2% vs. 17.1%). Notably, a family history of ovarian or breast carcinoma was not more prevalent among ovarian carcinoma cases (8.2%) compared with among subcohort members (8.6%). No large differences were observed for SES or current smoking status between ovarian carcinoma cases and subcohort members. Among subcohort members, mean daily vegetable consumption was positively associated with ever use of oral contraceptives and with SES and was negatively associated with age. Fruit consumption was found to be associated positively with a family history of ovarian or breast carcinoma and SES, but was negatively associated with current cigarette smoking. Vegetable consumption and fruit consumption were positively associated.
|Mean grams per day (±SD) (n = 240)||Mean grams per day (±SD) (n = 2088)|
|Total vegetables and fruits||385.7 (141.3)||384.3 (152.6)|
|Total vegetables||186.1 (68.2)||190.3 (75.2)|
|Cooked vegetables||146.8 (52.9)||148.8 (59.4)|
|Raw vegetables||39.4 (31.5)||41.5 (30.3)|
|Brassicas||31.6 (17.3)||31.1 (19.6)|
|Brussels sprouts||7.4 (6.5)||7.6 (7.3)|
|Cauliflower||14.0 (9.7)||13.7 (10.2)|
|Cabbage (white/green)||6.9 (7.7)||6.8 (8.0)|
|Kale||3.2 (3.2)||3.1 (3.3)|
|Legumes||28.9 (19.5)||30.6 (20.8)|
|String beans||18.9 (14.7)||19.3 (14.5)|
|Broad beans||4.3 (6.9)||4.3 (6.6)|
|Leafy vegetables, cooked||20.9 (14.4)||21.2 (15.7)|
|Spinach||9.7 (8.1)||9.4 (8.6)|
|Endive||11.2 (10.7)||11.8 (10.3)|
|Leafy vegetables, raw||9.4 (7.3)||10.1 (8.8)|
|Endive||1.9 (3.2)||2.5 (4.4)|
|Lettuce||7.5 (6.5)||7.6 (6.9)|
|Other vegetables||17.5 (14.3)||18.5 (15.5)|
|Carrots, cooked||9.4 (8.0)||8.7 (8.2)|
|Carrots, raw||4.4 (16.9)||3.5 (9.3)|
|Sweet peppers||2.4 (3.7)||3.1 (4.7)|
|Sauerkraut||6.1 (4.7)||5.7 (5.1)|
|Tomatoes||21.7 (17.7)||23.8 (20.9)|
|Red beets||7.7 (7.2)||7.9 (7.7)|
|Mushrooms||4.0 (4.7)||3.9 (4.4)|
|Gherkins||1.5 (5.4)||1.7 (5.9)|
|Rhubarb||3.0 (6.2)||2.5 (5.7)|
|Leek||9.3 (10.7)||9.3 (11.5)|
|Onions||18.5 (14.5)||20.1 (18.2)|
|n = 252||n = 2216|
|Total fruits||202.6 (108.8)||196.7 (121.6)|
|Citrus fruits||90.4 (71.2)||89.4 (76.6)|
|Mandarins||5.8 (9.3)||5.6 (9.1)|
|Oranges and fresh orange juice||56.1 (54.9)||56.2 (57.7)|
|Grapefruits and fresh grapefruit juice||13.9 (31.0)||12.1 (27.8)|
|Grapes||4.9 (8.4)||5.1 (10.2)|
|Bananas||16.2 (28.7)||12.8 (26.7)|
|Apples, pears||85.0 (72.2)||84.0 (82.2)|
|Strawberries||7.2 (7.3)||8.0 (8.6)|
|Other fruit juices||15.6 (35.1)||13.2 (34.8)|
Table 2 shows age-adjusted and multivariable-adjusted RRs for ovarian carcinoma according to quintiles of vegetable or fruit consumption. Women in the highest quintile of total vegetable and fruit intake had a multivariable-adjusted RRQ5 vs. Q1 of 1.13 (95% CI, 0.70–1.82) compared with women in the lowest quintile of intake. For the various vegetable subgroups, no statistically significant associations with ovarian carcinoma risk were found. Total fruit intake appeared not to be associated with ovarian carcinoma (multivariable-adjusted RRQ5 vs. Q1, 1.11; 95% CI, 0.70–1.78), although age-adjusted RRs showed a higher risk of ovarian carcinoma with higher fruit intake.
|Vegetable/fruit group||Quintile (Q) of intake||P value for trend|
|Total vegetables and fruits|
|Median intake in grams per day||207||297||367||445||583|
|RR (95% CI)a||1.00||1.41 (0.92–2.16)||1.33 (0.86–2.05)||1.07 (0.68–1.68)||1.21 (0.78–1.88)|
|RR (95% CI)b||1.00||1.40 (0.89–2.21)||1.29 (0.81–2.05)||1.03 (0.63–1.67)||1.13 (0.70–1.82)||0.53|
|Median intake in grams per day||105||147||180||220||291|
|RR (95% CI)a||1.00||0.98 (0.64–1.49)||1.05 (0.69–1.60)||1.17 (0.77–1.78)||0.87 (0.56–1.35)|
|RR (95% CI)b||1.00||.10 (0.69–1.75)||1.15 (0.72–1.83)||1.25 (0.79–1.98)||0.98 (0.61–1.58)||0.83|
|Median intake in grams per day||79||114||141||173||228|
|RR (95% CI)a||1.00||1.27 (0.82–1.96)||1.16 (0.74–1.80)||1.37 (0.89–2.11)||1.18 (0.76–1.84)|
|RR (95% CI)b||1.00||1.42 (0.88–2.28)||1.24 (0.76–2.01)||1.51 (0.94–2.43)||1.35 (0.83–2.21)||0.50|
|Median intake in grams per day(g/day)||9||24||36||51||77|
|RR (95% CI)a||1.00||0.76 (0.51–1.14)||0.75 (0.50–1.13)||0.75 (0.50–1.13)||0.73 (0.48–1.11)|
|RR (95% CI)b||1.00||0.74 (0.48–1.16)||0.78 (0.50–1.23)||0.80 (0.52–1.24)||0.75 (0.48–1.18)||0.66|
|Median intake in grams per day||11||20||27||37||57|
|RR (95% CI)a||1.00||1.17 (0.75–1.83)||1.45 (0.95–2.23)||1.22 (0.78–1.90)||1.29 (0.83–2.01)|
|RR (95% CI)b||1.00||1.33 (0.82–2.16)||1.49 (0.93–2.38)||1.38 (0.85–2.25)||1.42 (0.88–2.29)||0.54|
|Median intake in grams per day||10||18||26||37||59|
|RR (95% CI)a||1.00||0.87 (0.57–1.31)||0.91 (0.60–1.37)||0.75 (0.49–1.15)||0.88 (0.58–1.33)|
|RR (95% CI)b||1.00||0.80 (0.51–1.25)||0.88 (0.57–1.37)||0.71 (0.45–1.14)||0.93 (0.60–1.44)||0.65|
|Leafy vegetables, cooked|
|Median intake in grams per day||4||12||19||27||41|
|RR (95% CI)a||1.00||1.04 (0.67–1.60)||1.16 (0.76–1.77)||1.15 (0.75–1.75)||1.02 (0.66–1.58)|
|RR (95% CI)b||1.00||1.00 (0.62–1.60)||1.22 (0.78–1.92)||1.23 (0.78–1.94)||1.05 (0.66–1.67)||0.79|
|Leafy vegetables, raw|
|Median intake in grams per day||0.9||4||7||12||22|
|RR (95% CI)a||1.00||1.09 (0.70–1.68)||1.47 (0.97–2.23)||0.95 (0.60–1.50)||1.12 (0.72–1.74)|
|RR (95% CI)b||1.00||1.18 (0.74–1.88)||1.44 (0.91–2.28)||0.99 (0.60–1.65)||1.23 (0.75–2.02)||0.44|
|Median intake in grams per day||2||9||16||23||39|
|RR (95% CI)a||1.00||1.21 (0.80–1.83)||0.94 (0.61–1.46)||1.15 (0.76–1.75)||0.90 (0.58–1.40)|
|RR (95% CI)b||1.00||1.19 (0.77–1.83)||0.81 (0.51–1.29)||0.96 (0.61–1.51)||0.84 (0.53–1.33)||0.45|
|Median intake in grams per day||62||124||177||238||343|
|RR (95% CI)a||1.00||1.48 (0.95–2.30)||1.30 (0.83–2.03)||1.68 (1.10–2.57)||1.33 (0.85–2.07)|
|RR (95% CI)b||1.00||1.32 (0.83–2.09)||1.06 (0.66–1.71)||1.42 (0.91–2.22)||1.11 (0.70–1.78)||0.46|
For specific vegetables and fruits, age-adjusted and multivariable-adjusted RRs are presented as 25-g per day increments in mean daily vegetable or fruit consumption (Table 3). A negative association was observed between ovarian carcinoma and the consumption of raw endive (multivariable-adjusted RR, 0.24; 95% CI, 0.07–0.78). Cooked endive demonstrated a nonsignificant, inverse association (multivariable-adjusted RR, 0.78; 95% CI, 0.48–1.26). No statistically significant positive associations with ovarian carcinoma were observed. For raw carrots (multivariable-adjusted RR, 1.60; 95% CI, 0.95–2.69), there was a positive association with ovarian carcinoma, but the 95% CI included unity. In addition, no significant association was found between ovarian carcinoma and any of the individual fruit items studied. For bananas, the age-adjusted RR of 1.10 (95% CI, 1.00–1.20) suggested a positive association with ovarian carcinoma. However, after adjustment for confounding, the RR became smaller and was statistically nonsignificant (multivariable-adjusted RR, 1.07; 95% CI, 0.95–1.20).
|Vegetable/fruit items||RR (95% CI)a||Adjusted RR (95% CI)b|
|Total vegetables and fruits||1.00 (0.98–1.02)||1.00 (0.98–1.02)|
|Total vegetables||0.98 (0.94–1.03)||0.99 (0.95–1.04)|
|Individual vegetable items|
|Brussels sprouts||0.95 (0.62–1.47)||0.78 (0.42–1.44)|
|Cauliflower||1.13 (0.83–1.55)||1.06 (0.68–1.65)|
|Cabbage||1.05 (0.71–1.55)||1.19 (0.73–1.96)|
|Kale||1.15 (0.44–3.01)||1.02 (0.31–3.35)|
|String beans||0.95 (0.75–1.22)||0.86 (0.64–1.17)|
|Broad beans||1.01 (0.60–1.69)||1.23 (0.66–2.27)|
|Spinach||1.12 (0.78–1.59)||1.43 (0.86–2.37)|
|Endive, cooked||0.87 (0.60–1.25)||0.78 (0.48–1.26)|
|Endive, raw||0.38 (0.15–0.96)||0.24 (0.07–0.78)|
|Lettuce||1.00 (0.62–1.60)||1.26 (0.71–2.22)|
|Carrots, cooked||1.25 (0.91–1.73)||1.28 (0.89–1.85)|
|Carrots, raw||1.33 (0.83–2.12)||1.60 (0.95–2.69)|
|Sweet peppers||0.38 (0.15–0.95)||0.34 (0.10–1.12)|
|Sauerkraut||1.53 (0.90–2.60)||1.67 (0.86–3.25)|
|Tomatoes||0.88 (0.75–1.04)||0.90 (0.74–1.10)|
|Red beets||0.93 (0.61–1.41)||0.89 (0.55–1.44)|
|Mushrooms||1.32 (0.61–2.86)||1.88 (0.76–4.62)|
|Gherkins||0.81 (0.37–1.78)||0.93 (0.52–1.68)|
|Rhubarb||1.30 (0.84–2.02)||1.11 (0.61–2.04)|
|Leek||1.02 (0.78–1.34)||1.14 (0.79–1.65)|
|Onions||0.89 (0.75–1.05)||0.93 (0.75–1.16)|
|Total fruits||1.01 (0.99–1.03)||1.00 (0.97–1.02)|
|Citrus fruits||1.00 (0.97–1.04)||1.01 (0.97–1.05)|
|Individual fruit items|
|Mandarins||1.08 (0.76–1.53)||1.18 (0.80–1.75)|
|Grapes||0.92 (0.68–1.24)||0.83 (0.53–1.30)|
|Bananas||1.10 (1.00–1.20)||1.07 (0.95–1.20)|
|Apples, pears||1.00 (0.97–1.04)||0.99 (0.94–1.03)|
|Strawberries||0.73 (0.49–1.09)||0.66 (0.41–1.06)|
|Oranges and fresh orange juice||1.00 (0.94–1.05)||1.00 (0.94–1.07)|
|Grapefruits and fresh grapefruit juice||1.06 (0.95–1.18)||1.05 (0.92–1.20)|
|Other fruit juices||1.05 (0.97–1.14)||1.08 (0.98–1.18)|
Because the presence of subclinical disease at baseline could possibly have biased the results, we additionally analyzed the data after excluding cases and subcohort members with < 2 years of follow-up. The results were not substantially different (data not shown). Excluding women who had undergone a hysterectomy or women with a family history of ovarian or breast carcinoma did not appear to change the results significantly, nor did including total energy intake as a confounder in the analyses.
No significant association between vegetable or fruit intake and ovarian carcinoma risk was found among postmenopausal women in this large prospective cohort study. Although raw endive was inversely associated with ovarian carcinoma and the data suggested an association of ovarian carcinoma risk with higher consumption of raw carrots, these results should be interpreted with caution considering the number of statistical tests performed in the current study.
There exist a range of mechanisms through which vegetables and fruits might influence cancer risk in general (reviewed by Vaino and Bianchini23), and these activities might be enhanced by interactions between dietary components. Among others, vegetables and fruits may modulate the bioavailability of carcinogens, they contain many natural antioxidants that might prevent damage to cellular structures, and they may modulate enzyme systems necessary for metabolic activation of carcinogens. Because these enzyme systems are also participants in steroid hormone metabolism, their modulation may particularly affect the risk of hormone-dependent cancers,23 such as ovarian carcinoma. However, the evidence for a cancer-preventive effect of vegetable consumption for ovarian carcinoma is limited.23
The majority of evidence for a possible protective effect of high vegetable consumption on ovarian carcinoma risk has been provided by case–control studies, although these may have suffered from recall or selection bias.The protective effects of total vegetable consumption,6–8 and of subgroups of vegetables such as raw,9 cooked,9 green,10 cruciferous,11 and allium vegetables,7 have emerged from case–control studies. No statistically significant positive associations between vegetable consumption and ovarian carcinoma have been reported from case–control studies. Most case–control studies regarding fruit intake found no association,6, 8, 10, 11 although a Chinese case–control study found an inverse association between high fruit intake and ovarian carcinoma risk7 and an Italian case–control study reported a positive association between a high intake of citrus fruits and ovarian carcinoma.9
To our knowledge, the association between vegetables and fruit intake and the risk of ovarian carcinoma has been the subject of study in only three prospective cohort studies to date—two conducted in the U.S.13, 14 and a third one conducted in Sweden.12 These prospective studies reported a tendency toward a protective influence of higher vegetable intake, with RRs (95% CI) of 0.77 (95% CI, 0.48–1.24), 0.61 (95% CI, 0.38–0.97), and 0.76 (95% CI, 0.42–1.37) for the Nurses Health Study,14 the Swedish Mammography Cohort Study,12 and the Iowa Women's Health Study,13 respectively. In contrast, in our data, total vegetable consumption was not found to be inversely associated with ovarian carcinoma risk (multivariable-adjusted RRQ5 vs.v Q1 1.13; 95% CI, 0.70–1.82). In particular, the protective effect of a higher intake of green leafy vegetables reported for these previously conducted prospective studies, reaching statistical significance in the Iowa Women's Health Study (RR, 0.44; 95% CI, 0.25–0.79), was not supported by our data. Neither cooked (multivariable-adjusted RRQ5 vs. Q1 1.35; 95% CI, 0.83–2.21) nor raw (multivariable-adjusted RRQ5 vs. Q1 0.75; 95% CI, 0.48–1.18) leafy vegetable consumption was found to demonstrate a statistically significant association with ovarian carcinoma in our study population. Variation in the number of items included in the questionnaires, as well as in age and food intake distributions between cohorts, may have contributed to the observed differences between the studies.
Adjusted risk estimates for overall fruit consumption in these prospective studies ranged from 1.13 (95% CI, 0.66–1.93) for the Iowa Women's Health Study13 to 1.37 (95% CI, 0.90–2.06) for the Swedish Mammography Cohort Study,12 and are in line with the RRQ5 vs. Q1 of 1.11 (95% CI, 0.70–1.78) found in the current study.
Important strengths of the current study are its prospective design, reducing the potential for recall bias, and the nearly complete follow-up of cases as well as of subcohort members, making selection bias unlikely. The number of epithelial ovarian carcinoma cases was comparable to those encountered in prospective cohort studies on the subject published to date. For risk factors analyzed as quintiles, the power was sufficient to detect an RR ≤ 0.70 and ≥ 1.30. After excluding cases and subcohort members with < 2 years of follow-up, the risk estimates did not substantially change, indicating that the presence of subclinical disease at baseline had not contributed importantly to our results. Moreover, we were able to control confounding by most known ovarian carcinoma risk factors. Another possible limitation of the current study concerns measurement of vegetable and fruit intake, which has generally been considered difficult to assess. We relied on self-reported dietary information, determined by means of a food frequency questionnaire, which may have been subject to measurement error and, consequently, may have biased our results toward null, as a result of nondifferential misclassification of exposure. In addition, multivariable modeling itself may have added considerable uncertainty, because measurement error possibly has affected confounders as well.24 In addition, with respect to overall vegetable consumption, our study population may have been too homogeneous and therefore may have yielded too little contrast between highest and lowest quintiles of intake to detect differences in ovarian carcinoma risk. However, for specific vegetables as well as for fruit consumption, contrasts may have been larger due to individual preferences and, as a consequence, we do not expect the lack of contrast to have influenced the risk estimates for these items.
Similarly, a lack of association may also be explained by a relatively high range of vegetable and fruit intake among participants of The Netherlands Cohort Study on Diet and Cancer, assuming that cancer risk may only be elevated for individuals with low intake of vegetables and fruit. For the same cohort, statistically significant inverse associations have been described between most categories of vegetable or fruit consumption and lung carcinoma,25 as well as between brassica vegetables and cooked leafy vegetables and colon carcinoma,26 arguing against large measurement error or too little contrast obscuring an association between vegetable and fruit consumption and ovarian carcinoma risk, provided there is one. An alternative explanation for the lack of association might be that the time window in which exposure exerts an influence on ovarian carcinoma risk lies well before middle age. Previous studies have suggested that the influence of certain risk factors on ovarian carcinoma risk may differ between women with and without a family history of breast or ovarian carcinoma.27, 28 The majority of women in the current study cohort did not have a positive family history of ovarian or breast carcinoma and excluding women with a positive family history did not change the results significantly. In addition, the effects of certain risk factors may differ between histologic types of ovarian carcinoma. Cramer et al.29 observed that the association with α-carotene was more apparent in women with invasive serous and mucinous tumors. Therefore, studying all ovarian carcinoma subtypes together may have diluted possible associations. In addition, an influence from unmeasured confounders on associations presented cannot be excluded.
In conclusion, with the exception of raw endive, the results from The Netherlands Cohort Study on Diet and Cancer do not support previous observations of an inverse association between higher vegetable intake and ovarian carcinoma in postmenopausal women. In accordance with previous cohort studies, no association between fruit consumption and ovarian carcinoma risk was found.
The authors are indebted to the participants in the current study. They also thank the cancer registries (IKA, IKL, IKMN, IKN, IKO, IKR, IKST, IKW, IKZ, and VIKC) and The Netherlands nationwide registry of pathology (PALGA); Dr. A. Volovics and Dr. A. Kester for statistical advice; S. van de Crommert, H. Brants, J. Nelissen, C. de Zwart, M. Moll, W. van Dijk, M. Janssen, and A. Pisters for assistance; and H. van Montfort, T. van Moergastel, L. van den Bosch, and R. Schmeitz for programming assistance.
- 1EUCAN: cancer incidence, mortality and prevalence in the European Union 1998, version 5.0. IARC CancerBase No. 4. Lyon: IARC Press, 1999., , , .
- 17Estimation of the coverage of Dutch municipalities by cancer registries and PALGA based on hospital discharge data. Tijdschr Soc Gezondheidsz. 1994; 72: 80–84., , .
- 19Regression models and life-tables (with discussion). J R Stat Soc B. 1972; 34: 187–220..
- 20An introduction to survival analysis using Stata. College Station, TX: Stata Press, 2002., , .
- 23VainioH, BianchiniF, editors. Fruit and vegetables. IARC handbooks of cancer prevention. Volume 8. Lyon, France: IARC Press, 2003.