Introduction

Ovarian cancer is one of the most common gynaecological malignancies among women, especially women aged ≥50 years1. It is also associated with the highest mortality and poorest survival because it is usually diagnosed at an advanced stage; the 5-year survival rate is often below 25% to 30%2. Therefore, approaches that could prevent the development of ovarian cancer or decrease the risk of this disease are important and urgent.

The aetiology of ovarian cancer might be multifactorial. In addition to genetic, familial and reproductive factors, population-based studies have showed that patients with ovarian cancer have significantly higher levels of oxidative stress and lower levels of plasma antioxidants, which suggests that oxidative stress might play an important role in ovarian carcinogenesis3,4,5,6. Oxidative stress could increase DNA damage during successive ovulations and result in the malignant transformations of ovarian cells3. Thus, limiting oxidative stress might be an effective defence against ovarian cancer. Furthermore, the differences in the morbidity of ovarian cancer and dietary patterns among various countries suggest that dietary and lifestyle factors might have a close relationship with the risk of ovarian cancer7. Accordingly, high consumption of dietary antioxidants, such as carotenoids, might help to prevent ovarian cancer and reduce the cancer risk8,9.

Lycopene, which is mainly obtained from vegetables and fruits, is the most efficient carotenoid due to its activity as a singlet oxygen quencher and peroxyl radical scavenger. Recent studies have evaluated the relationship between dietary lycopene intake and ovarian cancer risk, but epidemiological studies have obtained inconsistent results. The majority of previous studies have yielded a reverse association between dietary lycopene intake and ovarian cancer risk10,11,12,13,14,15, while others have failed to provide evidence supporting the protective role of lycopene, including the U.S. Nurses Health Study16, an Italian cohort study17 and other studies on dietary fruit and vegetable intake and ovarian cancer risk18,19,20.

Thus, the aim of the present study was to address this suspected association between dietary lycopene intake and ovarian cancer risk among elderly women by conducting a meta-analysis.

Results

Paper selection

We initially obtained 134 papers from the PubMed database, most of which were excluded after reviewing the title and abstracts (i.e., they were reviews, not population-based research, or had research exposure irrelevant to our analysis). After reviewing the full texts of 34 potentially relevant articles, 20 basic experimental studies were excluded. Of the remaining 14 papers, two studies were related to circulating lycopene21,22 and two were duplicate publications23,24. Thus, 10 studies were included in our meta-analysis10,11,12,13,14,15,17,25,26,27.

Basic characteristics of the included studies

The characteristics of the selected studies are shown in Table 1. Of the 10 studies, three were prospective and seven were case-control studies. These seven studies included four hospital-based and three population-based case-control studies. All studies were published from 2001 to 2013. Two were conducted in China10,27, one in the European region17 and the remaining seven in the US11,12,13,14,15,25,26. Eight studies only included postmenopausal women and two included both pre- and postmenopausal women13,14. The majority of all participants were older than 50 years and nonsmokers. The response rate varied from 47.0% to 99.6%. Food consumption was assessed by a food-frequency questionnaire (FFQ) and the lycopene content was calculated using the US Department of Agriculture-National Cancer Institute Carotenoid Database with the exception of one study17, which used the Italian food composition database. Our meta-analysis did not include supplementation of lycopene. The length of dietary recall prior to the date of entry into each study for controls and the diagnosis date for cases varied widely from 3 months to 5 years. Case ascertainment was mainly performed using medical records and pathology reports or by the national cancer registry. In the three prospective studies, the duration of follow-up ranged from 8.3 to 16.0 years (median, 12 years). The number of cases ranged from 71 to 2012 and 2543 cases of ovarian cancer occurred among 668,806 participants. The case-control studies included 3584 cases and 6502 controls. The main confounding factors adjusted for were age, education, race, age at menarche, menopausal status, age at menopause for postmenopausal women, parity, oral contraceptive use, hormone replacement therapy use, body mass index, tubal ligation, total calories, physical activity and smoking status. These data are shown in Table 1.

Table 1 Characteristics of included studies in our meta-analysis
Full size table

Dietary lycopene and ovarian cancer risk

The adjusted ORs of the individual studies with corresponding 95% CIs and the pooled ORs are displayed in Figure 1. Of the 10 selected studies, six showed a negative association between dietary lycopene intake and ovarian cancer risk, with one showing statistical significance. Dietary lycopene consumption decreased the risk of ovarian cancer by 3.7% (OR, 0.963; 95% CI, 0.859–1.080) and no heterogeneity was observed across the studies (p = 0.336, I2 = 11.6%).

Figure 1
figure 1

Forest plot about the association between dietary lycopene and ovarian cancer.

Full size image

The study design had no effect on the relationship between dietary lycopene intake and ovarian cancer risk. The pooled OR for case-control studies and prospective studies was 0.981 (95% CI, 0.846–1.139; n = 7) and 0.871 (95% CI, 0.616–1.233; n = 3), respectively. The final risk estimate for studies conducted in and outside of the US was 0.923 (95% CI, 0.770–1.106; n = 7) and 1.026 (95% CI, 0.849–1.240; n = 3), respectively. Histologically, the OR for serous ovarian cancer and mucinous ovarian cancer was 0.744 (95% CI, 0.545–1.016; n = 4) and 0.878 (95% CI, 0.554–1.393; n = 3), respectively. Limiting the analysis to studies with higher response rates (>90%) resulted in an OR of 0.967 (95% CI, 0.849–1.102; n = 7). Similarly, when stratifying the results by the length of dietary recall before entering the research, omitting the studies with a recall duration of <1 year resulted in an OR of 0.973 (95% CI, 0.823–1.149; n = 3). There was no evidence of heterogeneity in the subgroup analysis.

We conducted sensitivity analysis by omitting one study from each analysis to investigate the effect of a single study on the pooled results. No study had a substantial effect on the final risk estimate; the pooled OR varied from 0.938 to 0.981.

Results of publication bias

Funnel plots (as displayed in Figure 2) with Begg's and Egger's tests did not support the existence of publication bias (Begg, p = 0.251; Egger, p = 0.406).

Figure 2
figure 2

Funnel plot about dietary lycopene and ovarian cancer.

Full size image

Discussion

Our meta-analysis of seven case-control studies and three prospective studies that included 678,892 subjects and 6127 cases demonstrated an insignificant inverse association between dietary lycopene consumption and ovarian cancer risk. Dietary lycopene intake resulted in a 3.7% (OR, 0.963; 95% CI, 0.859–1.08) lower risk of ovarian cancer, as compared to lycopene intake after adjusting for potential confounding factors. This is consistent with previous studies of the beneficial effect of dietary lycopene on ovarian cancer risk10,11,12,13,14,15. Further subgroup analysis showed that the research location, study design, response rate of participants to the study and histological type of ovarian cancer did not affect the relationship between dietary lycopene consumption and the risk of ovarian cancer. We observed no heterogeneity across the studies.

Our insignificant results might be explained as follows. First, the majority of participants in our study were older than 50 years. Cramer et al. reported that the protective role of dietary lycopene against ovarian cancer was much stronger in premenopausal women (<50 years) than in postmenopausal women (>50 years)14. Fairfield's study on the association between fruit and vegetable consumption and ovarian cancer risk deduced similar results16. Accordingly, the effect of lycopene on ovarian cancer risk might be stronger in early life than in later life. Because the postmenopausal period is associated with a higher risk of ovarian cancer and because most studies included postmenopausal women (2 of 10 studies presented their results as stratified by menopausal status13,14), we cannot definitively determine the effect of lycopene on premenopausal patients. Second, the protective role of lycopene against ovarian cancer might be related to the type of ovarian cancer. Chiaffarino et al. found that dietary carotenoids could protect against serous ovarian cancer, but not mucinous, endometrial, or other histological types28. Cramer et al. also found that lycopene could significantly protect against serous, especially borderline, ovarian cancer14. Both of these results suggest a latent relationship between the role of lycopene and the histological type of ovarian cancer. Only four and three studies provided the OR of the effect of lycopene on serous and mucinous ovarian cancer, respectively and the effect of lycopene on serous and mucinous ovarian cancer risk was deemed to be insignificant in all studies. This might be attributed to inadequate statistical power and it is possible that the histological types in our included studies did not include serous ovarian cancer, but other types. Third, underestimation of lycopene intake might have contributed to the observed null association in the present study. Because there was no available database from the Chinese Food Composition tables with which to calculate the lycopene content, two studies10,27 conducted in China used the USDA nutrients database instead. This might have led to underestimation of the lycopene intake because some fruits and vegetables commonly consumed in China are not covered by the USDA database, such as guava10.

The underlying mechanisms by which dietary lycopene reduces the risk of ovarian cancer might be multifactorial, but are likely to involve its activity as a singlet oxygen quencher and peroxyl radical scavenger. These activities may protect ovarian epithelial cells from oxidative stress and prevent malignant transformation of ovarian cells.

Limitations of the present study should be addressed. First, inherent limitations of case-control studies are recall bias and selection bias. Seven of 10 studies in our analysis were case-control studies and dietary assessment was based on a self-reported FFQ, which probably introduced some recall error. Among these seven studies, the use of hospital-based controls might have introduced selection bias because of the possibility that some controls might have had characteristics different from those of the general population. Second, misclassification of dietary lycopene intake may have occurred. The lycopene content was calculated based on the dietary information collected at baseline using only the FFQ, which might have been subject to measurement error. Individual differences in food preparation and the concurrent consumption of other foods could have also affected the lycopene bioavailability29,30,31. Dietary intake just prior to the diagnosis of ovarian cancer does not reflect longer-term dietary lycopene intake levels. Third, no analysis of the association between plasma levels of lycopene and ovarian cancer risk was performed. A biomarker approach would have the advantage of reducing measurement error associated with self-reports of dietary intake32,33. Finally, incomplete control for confounding factors may have existed in some studies, such as the levels of physical activity and alcohol drinking.

In summary, our meta-analysis has demonstrated that dietary lycopene intake could potentially decrease the ovarian cancer risk among postmenopausal women. Although the results did not reach statistical significance, they trended toward a beneficial role of lycopene in decreasing the risk of ovarian cancer. Other factors that should be considered in future research include the relationship between lycopene and different subtypes of ovarian cancer, the effect of menopausal status on the role of lycopene in ovarian cancer risk and the role of lycopene in women of different ages, particularly those aged <50 years. Because individual carotenoids have different antioxidant capacities, total carotenoids might be more effective than any single carotenoid in reducing the cancer risk. Prospective studies with larger numbers of patients from all age groups are required to determine the exact association between dietary lycopene and ovarian cancer risk.

Methods

Literature collection

Studies included in our final meta-analysis were selected after screening the PubMed database and Cochran library through December 2013 using the search terms “lycopene”, “tomato”, “carotenoids” and “ovarian cancer” with limitation to “humans”. Relevant reviews, the PubMed option “Related Articles” and references of included studies were also reviewed to find additional potentially relevant studies. Abstracts and papers without full texts were excluded.

Inclusion criteria

Only studies that met the following criteria were included in our meta-analysis: (1) population-based studies, such as case-control or prospective studies; (2) dietary lycopene or tomato was the exposure factor; (3) the odds ratio (OR) or relative risk (RR) and the corresponding 95% confidence interval (CI) between lycopene and ovarian cancer risk were provided and (4) there was no restriction on the age of the patients.

Data extraction

Data included in the final meta-analysis were extracted in a standardized manner according to the guidelines of the Cochrane Collaboration34 and Stroup et al.35 The last name of the first author, publication year, country, study design, characteristics of the subjects, determination of dietary lycopene intake, diagnosis of ovarian cancer, adjusted risk estimate between lycopene and ovarian risk with the corresponding 95% CI and adjustment for confounding factors were independently extracted by X-L LI and J-H X and disagreement was resolved by discussion.

Statistical methods

ORs were used to evaluate the association between dietary lycopene intake and ovarian cancer risk and RR was directly considered as the OR when the incidence of ovarian cancer was low. For studies that presented graded associations, only the estimate for the highest vs lowest level was selected in the final meta-analysis.

Heterogeneity tests were performed using the Q-test and I2 statistic, with p < 0.10 indicating statistical significance. A random-effects model was used to calculate the final risk estimates when the heterogeneity test was statistically significant; otherwise, a fixed-effects model was used36.

Subgroup analysis was conducted to address the effect of various study characteristics on the pooled risk.

To assess the source of heterogeneity and the robustness of the association between dietary lycopene intake and ovarian cancer risk, a sensitivity analysis was conducted using both fixed- and random-effects models, omitting one study for each analysis.

Funnel plots, Begg and Mazumdar's rank correlation test and Egger's test of the intercept were employed to assess the potential publication bias; and p < 0.10 suggested statistical significance37.

All analyses were conducted using STATA version 11.0 (Stata Corp). A p value of <0.05 denoted statistical significance unless otherwise specified.