Cruciferous Vegetable Intake and Risk of Prostate Cancer: A Systematic Review and Meta-Analysis

Prostate cancer (PCa) is the second most prevalent type of cancer in men and the sixth significant cause of cancer-related fatality [1]. In 2018, it was predicted that there would be 359,000 cancer-related fatalities and 1,266,000 new cases worldwide [1]. Additionally, compared to Asian nations, PCa is substantially lower in non-Asian nations [2]. However, it has risen rapidly in recent decades with a longer average lifespan and a progressively westernized lifestyle [2]. PCa primary prevention is a significant public health concern worldwide.

There is mounting proof that consuming more fruits and vegetables can reduce the risk of some cancer sites [3]. Remarkably, the association involving cruciferous vegetables and the risk of developing cancer has been the subject of much heated discussion. Cruciferous vegetables are a type of vegetable that is also known as cruciferous flowers. People frequently consume cruciferous vegetables: cauliflower, cabbage, and Brussels sprouts [4]. Indole and isothiocyanate, two active components of cruciferous vegetables, are known to have anticancer characteristics [5, 6]. Numerous epidemiological studies have demonstrated reverse relationships between cruciferous vegetables and cancer risks, including endometrial, lung, gastric, colorectal, and other cancers [7]. Meanwhile, Liu et al. [8] found that consumption of cruciferous vegetables was associated with lower risk of PCa. Over the last 30 years, multiple epidemiological studies have looked into the connection between eating cruciferous vegetables and the risk of PCa. Most studies reported that cruciferous vegetables had a statistically insignificant negative correlation with PCa risk, whereas some reports showed that there was a statistically insignificant positive correlation among them. Hence, the ratio of cruciferous vegetable consumption to PCa risk is controversial. A meta-analysis is needed to ascertain their link to one another.

We systematically searched PubMed, Embase, the Cochrane Library, and Web of Science databases from start-up dates to September 20, 2022. The elementary search process for meta-analyses was as follows: (prostatic neoplasms) OR (prostate cancer) OR (prostate neoplasm) OR (prostatic cancer) OR (Cancer of the Prostate) AND (cruciferous vegetables). See online supplementary Material 1 (for all online suppl. material, see https://doi.org/10.1159/000530435) for detailed search strategy. This meta-analysis was performed in light of the PRISMA (registration number: CRD42022361556) statement [9].

The following criteria were used to select the studies: (1) study types included cohort and case-control studies, (2) the study centered on the connection between cruciferous vegetables and PCa risk, (3) the study provided information on the ingestion of cruciferous vegetables for exposure interest, (4) the study involved PCa as a measure of results. The diagnosis of PCa was confirmed through a medical diagnosis, pathology reports, medical records, self-report, or cancer registration form. (5) The study provided estimates of relative risk (RR) or odds ratio (OR) and associated confidence intervals (CIs) to evaluate the relationship between cruciferous vegetables and PCa. Provided that diverse estimates were supplied, multivariable-adjusted risk estimates were preferred. If identified studies were published in more than one publication, older reports or those containing less relevant information were excluded.

Two investigators (J.L. and Z.L.) independently screened all obtained articles for inclusion and exclusion criteria. Any differences in research and literature selection were solved by the third investigator (S.L.).

Each study consisted of the following details: the leading author’s name, study region, type of study, quality score, participant number, patient number, RRs along with 95% CIs, and the latent confounders considered or adjusted.

The potential for bias was assessed using the Newcastle-Ottawa Scale (NOS) [10], which considers the choice of research groups, verification of exposure and result, and comparability of groups. Studies that received 7–9 points were categorically identified as high quality, those that received 5–6 points as intermediate quality, and those that received less than 4 points as poor quality.

The OR value is virtually consistent with the RR value when the low occurrence of PCa is taken into account [11]. Therefore, the effect of all included studies was evaluated using the RR value and its 95% confidence range. We aggregated RR estimates for distinct dosages of cruciferous vegetables relevant to the study. The ability to examine study heterogeneity is a feature of the I² statistic [12]. According to the results of the I², the fixed or random (DerSimonian-Laird) effect model was applied to analyze the article’s heterogeneity [12]. If I² is below 50% (I² < 50%), the heterogeneity between tests is weak or moderate. The RR estimates are therefore calculated using a fixed effect model. If I² is greater than or equal to 50% (I² ≥ 50%), there is moderate to significant heterogeneity between studies. Hence, a random effect model is applied to figure out the RR estimate. When the study is moderately heterogeneous (I² 30–60%), the random effect model is selected as a more conservative approach [13, 14].

Subgroup analyses were stratified in line with study location, study type, year of publication, and adjustment for potential confounders, including a family history of PCa, body mass index (BMI), smoking status, and alcohol intake. We employed funnel plots [15, 16], the Egger linear regression test [17], and the Begg rank correlation asymmetry test [18] to look for evidence of publication bias. First, we visually evaluate whether the scatter points on the funnel plot are symmetrical. To the extent that the funnel chart is asymmetrical, there may be a likelihood of publishing bias. Second, publication bias was formally assessed through the Begg and Egger correlation test. If the p value was below 0.1, it was taken as a statistically significant publication bias. Should a publication bias occur, we will utilize the trim and fill method for analysis [19]. Sensitivity analyses were also conducted to determine if the exclusion from one research each time could have affected the aggregate results.

Furthermore, the probable dose-response link between eating cruciferous vegetables and PCa risk was looked into. In order to use the methodology recommended by Greenland and Longnecker [20], 95% of associated RRs and ICs were collected for dose-response analysis. In order to use this method, the intake of at least three groups was required, and the case number within each group and the participant number within each group/person per year could be provided. We then estimated the overall RR value in light of the trend of each study. At the same time, we tested the linear and nonlinear connections between the consumption of cruciferous vegetables and PCa risk using a restricted quartic spline. Generalized least squares regression was used to estimate the restricted quartic spline, which was 4 knots at each fixed percentile (5%, 35%, 65%, and 95%) of the distribution [21]. We modified the consumption in grams per day, using 80 g as the roughly average portion, given that the included studies utilized various units to measure the intake of cruciferous vegetables. When the original study group provided an intake dose interval for cruciferous vegetables, in the closed interval, the midpoint of the upper and lower bounds of the interval was considered to be the mean intake dose; for the lower open interval, we divided the endpoint of the interval by 1.2; in the upper open interval, we multiply the endpoint of the interval by 1.2.

In all analyses, the p value below 0.05 was found to be statistically meaningful. Stata version 14 (StataCorp, College Station, TX, USA) was utilized for statistical analysis.

703 records were obtained from a total of 4 databases. After removing 212 duplicate records, 491 records were still used in the selection of titles and abstracts, and 473 irrelevant records were ruled out following the selection of headlines and abstracts. Three studies were removed because they lacked relevant risk evaluations or 95% CIs after a thorough review of the remaining 18 trials. Fifteen studies were identified from the full-text screening. One additional study was acquired by examining the reference list of recovered articles. As a result, 16 studies were incorporated into the ultimate analysis (Fig. 1).

Of the 16 papers from the analysis, seven were case-control studies [22, 23, 24, 25, 26, 27, 28], and nine were cohort studies [29, 30, 31, 32, 33, 34, 35, 36, 37]. There were 1,264,437 people enrolled in the study, and 70,201 PCa cases were reported. There were 10 articles [22, 23, 24, 25, 26, 27, 29, 31, 33, 34] on data from North America, three from Europe [28, 30, 32], one from Australia [35], one from Asia [36], and one from Europe and Asia [37]. The characteristics of this investigation are outlined in Table 1. Articles analyzing the consumption of cruciferous vegetables from the dietary habit applied a questionnaire or a proven food frequency questionnaire to gather daily food consumption. In some studies, the food intake of participants was documented weekly or monthly and converted into a daily intake. The entire research reported employing risk estimates such as RR or OR. The quality of each item was evaluated using criteria from the Newcastle-Ottawa Case-Control and Cohort Scales. The mean scores for the case-control and cohort studies were 7.43 (SD = 0.5) and 7.89 (SD = 1.0), respectively.

The association between eating cruciferous vegetables and PCa is depicted in Figure 2. The RR values for the initial study ranged from 1.30 (95% CI: 0.80–2.00) in Ambrosini’s study [35] to 0.54 (95% CI: 0.38–0.78) in Joseph’s study [26]. The highest intake of cruciferous vegetables decreased the incidence of PCa by 13% in comparison to the lowest intake (RR = 0.87; 95% CI: 0.80–0.95). Over the course of the research, moderate heterogeneity was identified (p = 0.001, I² = 59.2%).

The outcome of the dose-response analysis is displayed in Figure 3. The findings suggest that consuming cruciferous vegetables may reduce the likelihood of developing PCa. With a daily increase of 15 g of cruciferous vegetables, the combination RR for PCa was 0.955 (95% CI: 0.928–0.982). In the funnel graph, no substantial publication bias was observed in the included studies. However, Egger’s test (p = 0.001) and Begger’s test (p = 0.034) both showed evidence of publication bias. Then the trim and fill method was applied to identify and rectify the deviation. Its result is illustrated in Figure 4. Finally, there was no missing document, which means there was no publication bias. The pooled RRs using fixed effect and random effect models were 0.96 (95% CI: 0.92–1.00) and 0.90 (95% CI: 0.83–0.97), respectively, which were consistent with the combined effects before using the trim and fill method.

A subgroup analysis was conducted in order to minimize heterogeneity. Table 2 lists the findings of the subgroup analysis of cruciferous vegetable consumption and PCa risk. Consuming cruciferous vegetables reduced PCa risk among participants from North America (RR = 0.85, 95% CI: 0.75–0.94) and Australia (RR = 0.56, 95% CI: 0.20–0.92), but there was no statistically significant difference among participants from Europe (RR = 0.91, 95% CI: 0.79–1.04) and Asia (RR = 0.92, 95% CI: 0.60–1.24). Analysis of subgroups in relation to the type of study found that consuming cruciferous vegetables considerably reduced the PCa risk in the case-control study (RR = 0.79, 95% CI: 0.70–0.87) but not in the cohort study (RR = 0.98, 95% CI: 0.93–1.02). In addition, with the exception of marital status and alcohol consumption, significant negative correlations were found in other confounding correction subgroups.

The combined RR values for the remaining studies are recalculated after excluding a study in the sequence of sensitivity analysis. The total RR from 0.85 (95% CI: 0.77–0.93) to 0.89 (95% CI: 0.82–0.96) has not changed considerably overall.

According to this meta-analysis, consuming cruciferous vegetables is related to the reducing risk of PCa. Liu et al. [8] conducted the first meta-analysis of this topic. We have made a further analysis on this basis.

When analyzing subgroups by geographical region, studies from North America (RR 0.85, 95% CI: 0.75–0.94) and Australia (RR 0.56, 95% CI: 0.20–0.92) are more relevant than those from Europe (RR 0.91, 95% CI: 0.79–1.04) and Asia (RR 0.92, 95% CI: 0.60–1.24), indicating that regional differences may bring about the heterogeneity of observations. Meanwhile, among the 10 studies [22, 23, 24, 25, 26, 27, 29, 31, 33, 34] in North America, there are 6 case-control studies [22, 23, 24, 25, 26, 27], which may be the reason for the internal heterogeneity. Synthetic analysis of case-control studies revealed that eating cruciferous vegetables reduced the PCa risk when subgroups were analyzed by research type. The pooled estimate of the cohort studies was insignificant, indicating that our findings were based primarily on case-control studies. It is worth mentioning that 2 of the included studies were case-control studies from the USA and Australia, where the dietary pattern has grown beyond Europe and focuses on food processing. Furthermore, due to the dietary differences between the USA and Europe, it is necessary to conduct a prospective study to explore this difference. We also studied some essential confounders, including the family-related history of PCa, smoking, drinking as well as obesity.

PCa is a highly hereditary cancer [38]. Epidemiological and family studies have confirmed the apparent family aggregation of PCa [39]. Smoking [40, 41], alcohol, and fat [42‒44] have all been related to an increased risk of PCa in recent epidemiological and follow-up studies. Eating more cruciferous vegetables is often accompanied by a good lifestyle. As a result, healthy habits are linked to lower smoking rates, alcohol use, and BMI. However, after extensive investigation of the confounding factors of smoking, alcohol consumption, and BMI, the inverse relationship between cruciferous vegetables and PCa remains. It also confirms the accuracy and reliability of our research findings, namely that consuming cruciferous vegetables may be a protective factor for PCa.

Cruciferous vegetables have been known for their medicinal value since ancient times [45]. Over the last few decades, the anticancer impact of cruciferous vegetables has attracted extensive attention. Numerous scientists have sought to investigate the relationship between cruciferous veggies and the tumor. However, the preventative effect of cruciferous vegetables on cancer can involve various convoluted mechanisms that have not yet been fully acknowledged. So far, most research has attended to the ability of cruciferous vegetable components to modify the expression and liveness of biotransformation enzymes. Glucosinolates are natural phytochemicals that produce bioactive species in cruciferous vegetables. When broken down by the endogenous plant enzyme myrosinase, they produce two bioactive substances: sulforaphane and indole-3-carbinol. Sulforaphane can induce phase II enzyme [46] and then inhibit Akt signal transduction from inducing apoptosis of PCa cells [47]. The cell cycle inhibitors p21 and p27 are downregulated by indole-3-carbinol, and CDK6 activity is inhibited as a result [48], and then, the growth of PCa is curbed. Additionally, studies have demonstrated that vitamin K is negatively correlated with PCa [49, 50]. As one of the two forms of vitamin K, the intake of menaquinones (vitamin K2) is negatively related to PCa risk, rather than phylloquinone (vitamin K1) [49]. As a product of cruciferous vegetables, phylloquinone is converted to menaquinones in the vitamin K cycle [50].

The strength of our study lies in the dose-response analysis we performed to back up the research hypothesis that a larger consumption of cruciferous vegetables is related with a linear reduction in PCa risk. A thorough subgroup analysis was also performed to identify underlying factors of moderate heterogeneity.

Certain constraints in this meta-analysis cannot be neglected. To begin with, as we all know, Asia is a high-consumption area of cruciferous vegetables, making it an ideal object of research. Nevertheless, only one such study was carried out in Japan. Another study was carried out jointly in Japan and Europe. Because of the scarcity of relevant literature, we also considered controlled trials and cohort studies. There may be an erroneous negative connection between eating cruciferous vegetables and the likelihood of developing PCa due to recall bias in the case-control model. Simultaneously, case-control studies may be flawed due to selection bias, which could skew the results of any association between a cruciferous vegetable diet and the risk of PCa. Therefore, the result of case-control studies should be carefully explained. Moreover, asymmetry in funnels was identified using Begg’s and Egger’s tests. The trim and fill method allowed us to virtually eliminate publication bias, but it also reflected that the publications we included might have disparities in study quality, research heterogeneity, chance, and artefactual effects. Finally, the majority of research is conducted in North America and Europe because of the comparatively high frequency of PCa in those regions. As a result, we should be cautious when extending results to other regions with relatively low impacts.

In summary, meta-analysis results from case-control and published cohort studies reveal that greater consumption of cruciferous vegetables may be linked to a lower PCa risk. However, there are relatively few studies included in the meta-analysis, and the results from the cohort studies are not significant. At present, no clear conclusion can be drawn.

An ethics statement is not applicable because this study is based exclusively on published literature.

The authors have no conflicts of interest to declare.

All authors announced that they had not received any fund support.

J.Y.L. wrote the manuscript. Z.H.L. and S.L. revised the manuscript. J.Y.L. and Z.H.L. sought out and selected studies of relevance. S.L. extracted the data. Z.H.L. contributed to the data analysis. B.X.C. performed manuscript editing and review. A final manuscript was read and approved by all authors.

All the data generated or analyzed during this study are included in this article and its online supplementary material. Further inquiries can be directed to the corresponding author.

Jiaye Long and Zhaohui Liu contributed equally to this work.