Vegetarians, Vegans Have Lower Cancer Risk

You have full access to this open access article Several studies have suggested that vegetarian and vegan vs. non-vegetarian diets are associated with lower cancer risk overall, however, results for specific cancer sites have been less consistent. We conducted a systematic review and meta-analysis of prospective studies on vegetarian and vegan diets and cancer incidence to clarify the associations across cancer sites. PubMed and Embase databases were searched for relevant studies up to 5 July 2025. Summary relative risks (RRs) and 95% confidence intervals (95% CIs) were calculated for the association between vegetarian and vegan diets and cancer incidence. World Cancer Research Fund (WCRF) criteria was used to evaluate the strength of the evidence. Seventeen publications (seven prospective studies) were included. The summary RRs (95% CIs) for vegetarians vs. non-vegetarians was 0.87 (0.84–0.91, I2 = 0%, n = 4 studies) for total cancer incidence, 0.55 (0.36–0.86, I2 = 32%, n = 4) for stomach cancer, 0.86 (0.76–0.97, I2 = 14%, n = 6) for colorectal cancer, 0.79 (0.67–0.93, I2 = 38%, n = 7) for colon cancer, 0.55 (0.31–0.97, I2 = 0%, n = 2) for proximal colon cancer, 0.77 (0.62–0.95, I2 = 0%, n = 5) for pancreatic cancer, 0.79 (0.66–0.94, I2 = 0%, n = 4) for melanoma, 0.92 (0.86–0.99, I2 = 0%, n = 7) for breast cancer, 0.81 (0.69–0.95, I2 = 0%, n = 3) for postmenopausal breast cancer, 0.78 (0.62–0.98, I2 = 0%, n = 5) for bladder cancer, and 0.76 (0.63–0.93, I2 = 0%, n = 4) for non-Hodgkin’s lymphoma. In addition, non-statistically significant inverse associations were observed for some cancers, with summary RRs of 0.85 (0.70–1.04, I2 = 0%, n = 6) for lung cancer, 0.83 (0.68–1.02, I2 = 0%, n = 5) for ovarian cancer, and 0.87 (0.75-1.00, I2 = 43%, n = 6) for prostate cancer. Results for other cancer sites were imprecise or near the null. The summary RRs (95% CIs) for vegans vs. non-vegetarians were 0.77 (0.70–0.85, I2 = 0%, n = 3) for total cancer, 1.02 (0.71–1.48, I2 = 42%, n = 3) for colorectal cancer, 0.80 (0.64–1.00, I2 = 0%, n = 4) for breast cancer, and 0.87 (0.50–1.49, I2 = 49%, n = 3) for prostate cancer. BMI explained a moderate part of the associations. The strength of evidence [judging the likelihood of causality] for vegetarian diets and total, colorectal, colon and breast cancer was judged as probable, and limited suggestive for stomach, pancreatic, and bladder cancers, melanoma and non-Hodgkin’s lymphoma, and for vegan diets and total cancer and breast cancer was considered limited-suggestive. Vegetarian diets compared to non-vegetarian diets are associated with reduced risk of total cancer and seven specific cancer types, while vegan diets are associated with reduced risk of total and breast cancer. Although further studies are needed to assess the long-term adherence to vegetarian and vegan diets and cancer incidence and across less investigated cancers, these results support recommendations to adopt much more plant-based diets for cancer prevention. Cancer is a major cause of morbidity and the second leading cause of death globally with 20 million new cases and 9.7 million deaths occurring in 2022 [1]. The observations of large international variation in cancer rates [1], secular trends in cancer rates over time within countries with Westernization of lifestyles [2], and changing cancer rates in individuals migrating from countries with a low incidence to countries with a high incidence of certain cancers (e.g. colorectal, breast, prostate cancers) [3], provides strong evidence that modifiable risk factors are important in cancer etiology. Smoking [4], alcohol consumption [5], adiposity [5] and low physical activity [5, 6] are established risk factors for a wide range of cancers. Dietary factors also are of major importance for some individual cancers, but less is known with certainty across different cancer types [5]. A growing body of evidence suggests that plant-based diets, high in fruits, vegetables, whole grains, nuts, and legumes and with low or no intake of red and processed meat, is associated with reduced cancer risk [5, 7,8,9]. Consistent with this, some studies have suggested vegetarian diets are associated with reduced incidence of digestive tract, colorectal, breast and prostate cancer, however, not all studies have been consistent. In the Adventist Health Study, vegetarians had nearly half the risk of colon cancer compared to meat eaters [10, 11], and in the Adventist Health Study 2 vegetarians had a 21% lower risk of colorectal cancer compared to meat eaters [12]. Other studies reported non-statistically significant inverse associations between vegetarian diets and colorectal cancer [13, 14]; however, in a combined analysis of the EPIC-Oxford Study and the Oxford Vegetarian Study, no association was observed [15]. Studies on vegetarian diets and breast cancer incidence have reported non-statistically significant inverse associations [14, 16,17,18] or associations close to unity [12, 15]. Studies have also suggested potential inverse associations with stomach [12, 15, 19], lung [10, 12, 15, 18, 19], prostate [12, 14, 15, 18, 20], and bladder [12, 15, 19] cancers, non-Hodgkin’s lymphoma or lymphoma [12, 15, 19] and multiple myeloma [12, 15, 19], however, the associations have not always been consistent across studies or statistically significant. Studies on vegan diets have reported inverse associations with total cancer [12, 15] and breast cancer [12, 15, 16], while results for colorectal [12, 15], and prostate cancer [12, 15, 21] have been more mixed. Although the previous conclusions from the Third Expert Report of the World Cancer Research Fund (WCRF) on vegetarian diets and cancer risk have been inconclusive [5], those conclusions were hampered by a limited number of studies published at the time and are now likely outdated as several additional cohort studies have since been published across cancer sites [12,13,14,15, 18, 19, 22, 23]. We therefore conducted a systematic review and meta-analysis to clarify the associations between vegetarian and vegan diets and cancer incidence to provide an up-to-date summary of the evidence across cancer sites. A protocol was developed for the project as part of a grant application for funding. Although this was not pre-registered before the meta-analysis was conducted, it is available through the Open Science Framework (https://osf.io/c3k54/overview). We searched PubMed and Embase databases up to 5 July 2025 for eligible studies. The search terms used are provided in the Supplementary text 1. We followed the PRISMA criteria for reporting of systematic reviews and meta-analyses [24]. The reference lists of the included publications were screened for further potentially relevant studies. Eligibility, inclusion criteria, and study selection The review question was framed using the PECO(S) elements [25]. Studies of healthy individuals in the general population with no restrictions with regard to age, sex, or pregnancy status (P), who adhered to vegetarian (excluding meat, poultry or fish) and vegan (excluding meat, poultry, fish, eggs, and dairy products) diets (E) vs. non-vegetarian (including meat, poultry, fish, eggs, dairy) diets (C) and were followed up for cancer incidence (O) were eligible for inclusion. Prospective cohort studies (S) that reported adjusted relative risk (RR) estimates (including hazard ratios, risk ratios and odds ratios) and 95% confidence intervals (CIs) for the association between vegetarian and vegan vs. non-vegetarian diets and cancer incidence were eligible for inclusion. Retrospective case-control and cross-sectional studies were excluded. When several publications were published from the same cohort on the same diet-cancer association we included the publication with the largest number of cases. However, when one publication only reported on the main cancer outcome and a second publication reported on a subtype of this cancer, both publications were included in the respective analyses, but each study was only included once in each analysis to avoid double-counting. DA did the full screening and DA and JGS screened the initial selection of studies (n = 242) in duplicate. Any discrepancies were resolved by discussion. A list of the excluded studies and the exclusion reasons can be found in Supplementary Table 1. The following data were extracted from each study: The first author’s last name, publication year, country where the study was conducted, the name of the study, study period and duration of follow-up, sample size, sex, age, number of cases, type of diet and comparison, subgroup, RRs and 95% CIs and variables adjusted for in the analysis. The data extraction was conducted by DA and checked for accuracy by JGS. Study quality was assessed in duplicate by DA and JGS, using a modified version of the Newcastle Ottawa Scale (NOS) which rates studies according to selection, comparability and outcome assessment [26]. The modifications are described in detail in Supplementary text 2. We used WCRF grading criteria to assess the strength of evidence for the associations between vegetarian and vegan diets and cancer incidence [5, 27]. This grading system assesses evidence from different study types, the number of studies available, heterogeneity, study quality, dose-response relationship (when relevant), biological plausibility, and experimental evidence. The evidence is graded as (1) substantial effect on risk unlikely, (2) limited-no conclusion, (3) limited-suggestive, (4) probable, and (5) convincing evidence of a causal relationship using these criteria [5, 27]. The evidence grading was done by discussion between all authors. We used DerSimonian and Laird random effects models to calculate summary RRs (95% CIs) for the association between vegetarian and vegan diets and cancer incidence [28]. The average of the natural logarithm of the RRs was estimated and the RR from each study was weighted using random effects weights [28]. For three articles from the Adventist Health Study that reported on total meat and fish consumption, but not vegetarian status, and risk of colon [11], ovarian [29] and bladder [30] cancer incidence, we pooled categories of meat and fish consumers (vs. non-consumers) using a fixed-effects model and inverted the risk estimates so the comparison became vegetarians vs. non-vegetarians. When estimates were presented without and with BMI adjustment in the same study, estimates without BMI adjustment were prioritized for the main analysis, as there is evidence from randomized trials that vegetarian and vegan diets can help with weight loss [31], and BMI can therefore be considered a mediator. We additionally directly compared analyses without and with BMI adjustment when such results were presented in the same study to quantify how much of any observed associations were due to differences in BMI. Heterogeneity between studies was evaluated using the Q test and I2 statistics [32] and because of a small to moderate number of studies we also present 95% CIs for the I2. I2 is a measure of how much of the heterogeneity is due to between study variation rather than chance, ranging from 0 to 100%. We conducted main meta-analyses (all studies combined) and stratified by study characteristics such as sex, duration of follow-up (≥ 10 vs. <10 years), geographic location, general population vs. Adventist setting, number of cases, study quality and by adjustment for confounding factors (e.g. age, education, socioeconomic status, smoking, alcohol, physical activity, age at menarche, hormone replacement therapy, oral contraceptive use) to investigate potential sources of heterogeneity when there was at least five studies included in the analysis. These subgroups were chosen as they were considered potential sources of heterogeneity or established risk factors for several cancers, in line with our previous meta-analyses [7, 33]. Publication bias was assessed using Egger’s test [34] and by inspection of funnel plots in analyses with at least five risk estimates. Sensitivity analyses excluding one study at a time were conducted when at least five risk estimates were included in the analysis. Because there was a limited number of studies on vegans we repeated the analyses in vegetarians restricted to the same studies that reported on vegans and compared the result to the overall analysis in vegetarians, to assess whether the studies published on vegan diets were representative for the overall findings. We calculated E-values for the association between vegetarian and vegan diets and cancer incidence, to assess the potential impact of unmeasured or uncontrolled confounding [35]. The statistical analyses were conducted using the software package Stata, version 16.0 (StataCorp, Texas, US). A total of 2739 records were screened, and of these we identified 17 publications [10,11,12,13,14,15,16,17,18,19,20, 22, 23, 29, 30, 36, 37] covering seven population-based prospective cohort studies that assessed the association between vegetarian or vegan diets and cancer incidence and were included in the analyses (Fig. 1, Supplementary Tables 2–18). The included studies were Oxford Vegetarian Study, European Prospective Investigation into Cancer and Nutrition (EPIC) - Oxford study, UK Women’s Cohort Study, UK Biobank, Netherlands Cohort Study, Adventist Health Study, and Adventist Health Study II. The publication from Oxford Vegetarian Study and EPIC-Oxford Study pooled the two studies together [15], while the remaining studies reported study-specific results. Five studies were from Europe (four from the UK and one from the Netherlands) and two studies were from the USA. All cohorts included both men and women, except for the UK Women’s Cohort Study which only included women. The number of participants ranged from 10,210 to 472,377 and the duration of follow-up ranged from 6 to 20.3 years. The mean (median) study quality was 7.05 (7.0) out of maximum 8 (Supplementary Table 19). Suboptimal confounder adjustments and lack of reporting on loss to follow-up were the main reasons for suboptimal study quality. Figure 2a and b shows a summary of the results across cancer sites for vegetarians and vegans vs. non-vegetarians, respectively, and the individual forest plots are shown in Figs. 3, 4, 5 and 6. Vegetarian (a) and vegan (b) diets and cancer incidence, summary estimates across cancers Vegetarian diets and total cancer (a), upper aerodigestive cancer (b), stomach cancer (c), colorectal cancer (d), colorectal cancer subsites (e), and pancreatic cancer (f) incidence Vegetarian diets and lung cancer (a), melanoma (b), breast cancer (c), endometrial cancer (d), ovarian cancer (e), and prostate cancer (f) incidence Vegetarian diets and kidney cancer (a), bladder cancer (b), brain cancer (c), non-Hodgkin's lymphoma (d), multiple myeloma (e) and leukemia (f) incidence Vegan diets and total cancer (a), colorectal cancer (b), breast cancer (c), and prostate cancer (d) incidence Four prospective studies (three risk estimates, three publications, 65,523 cases, 613,492 participants) [12, 14, 15] were included in the analysis of vegetarian diets and total cancer incidence. The summary RR (95% CI) was 0.87 (0.84–0.91, I2[95%CI] = 0[0–0]%) for vegetarians vs. non-vegetarians (Figs. 2a and 3a). Three prospective studies (two risk estimates, two publications, 10,562 cases, 141,115 participants) [12, 15] were included in the analysis of vegan diets and total cancer incidence. The summary RR (95% CI) was 0.77 (0.70–0.85, I2 = 0[0–34]%) for vegans vs. non-vegetarians (Figs. 2b and 6c). Upper aerodigestive tract cancer Four prospective studies (three risk estimates, three publications, 1050 cases, 550,225 participants) [12, 15, 19] were included in the analysis of vegetarian diets and upper aerodigestive tract cancer. The summary RR was 0.99 (95% CI: 0.74–1.31, I2 = 0[0–0]%) for vegetarians vs. non-vegetarians (Figs. 2a and 3b). Four prospective studies (three risk estimates, three publications, 935 cases, 550,225 participants) [12, 15, 19] were included in the analysis of vegetarian diets and stomach cancer risk. The summary RR (95% CI) was 0.55 (0.36–0.86, I2 = 32[0–82]%) for vegetarians vs. non-vegetarians (Figs. 2a and 3c). Six prospective studies (five risk estimates, five publications, 7892 cases, 655,849 participants) [ 12,13,14,15, 22] were included in the analysis of vegetarian diets and colorectal cancer incidence. The summary RR for colorectal cancer was 0.86 (0.76–0.97, I2 = 14[0–69]%) for vegetarians vs. non-vegetarians (Figs. 2a and 3d). When analysed by subsite, the summary RR was 0.79 (95% CI: 0.67–0.93, I2 = 38[0–76]%) for colon cancer [11,12,13,14,15, 22] (5271 cases, 624,633 participants), 0.55 (95% CI: 0.31–0.97, I2 = 35[0–87]%) for proximal colon cancer [19, 22] (1905 cases, 441,257 participants), 0.84 (95% CI: 0.57–1.22, I2 = 0[0–18]%) for distal colon cancer [19, 22] (1613 cases, 441,257 participants), and 0.88 (95% CI: 0.70–1.11, I2 = 19[0–71]%) for rectal cancer [12,13,14,15, 22] (2690 cases, 592,582 participants) (Figs. 2a and 3e). There was no indication of publication bias with Egger’s test for colorectal (p = 0.45), colon (p = 0.85) or rectal cancer (p = 0.17) (Supplementary Fig. 1, 2 and 3). Three cohort studies (two risk estimates, two publications, 1111 cases, 141,115 participants) [12, 15] were included in the analysis of vegan diets and colorectal cancer. The summary RR (95% CI) was 1.02 (0.71–1.48, I2 = 42[0–89]%) for vegans vs. non-vegetarians (Figs. 2b and 6b). Five prospective studies (four risk estimates, four publications, 1581 cases and 585,590 participants) [12, 15, 19, 23] were included in the analysis of vegetarian diets and pancreatic cancer. The summary RR (95% CI) was 0.77 (0.62–0.95, I2 = 0[0–11]%) for vegetarians vs. non-vegetarians (Figs. 2a and 3f). Six prospective studies (five risk estimates, five publications, 4240 cases, 594,633 participants) [10, 12, 15, 18, 19] were included in the analysis of vegetarian diets and lung cancer. The summary RR (95% CI) was 0.85 (0.70–1.04, I2 = 20[0–71]%) for vegetarians vs. non-vegetarians (Figs. 2a and 4a). There was no indication of publication bias with Egger’s test (p = 0.75) (Supplementary Fig. 4). Four prospective studies (three risk estimates, three publications, 2575 cases, 550,225 participants) [12, 15, 19] were included in the analysis of vegetarian diets and melanoma incidence. The summary RR was 0.79 (95% CI: 0.66–0.94, I2 = 0[0–30]%) for vegetarians vs. non-vegetarians (Figs. 2a and 4b). Seven prospective studies (six risk estimates, six publications, 10,453 cases, 411,391 participants) [12, 14,15,16,17,18] were included in the analysis of vegetarian diets and breast cancer. The summary RR (95% CI) was 0.92 (0.86–0.99, I2 = 0[0–59]%) for vegetarians vs. non-vegetarians (Figs. 2a and 4c). There was indication of publication bias with Egger’s test (p = 0.02) (Supplementary Fig. 5), but this was explained by one outlying study (18) and when excluded Egger’s test was attenuated (p = 0.22), and the observed association was slightly attenuated (summary RR = 0.93, 95% CI: 0.86–1.00, I2 = 0[0–59]%). Two [17, 19] and three [17,18,19] cohort studies were included in the analysis of vegetarian diets and premenopausal and postmenopausal breast cancer risk, respectively, and the summary RR was 0.97 (0.79–1.19, I2 = 0[0–0]%) for premenopausal breast cancer and 0.81 (0.69–0.95, I2 = 0[0–26]%) for postmenopausal breast cancer (Figs. 2a and 4c). Four prospective studies (three risk estimates, three publications, 1821 cases, 118,048 participants) [15, 16, 38] were included in the analysis of vegan diets and breast cancer incidence. The summary RR (95% CI) was 0.80 (0.64–1.00, I2 = 0[0–0]%) for vegans vs. non-vegetarians (Fig. 6c). When using results at age 65 from the Adventist Health Study II, the summary RR was 0.76 (0.63–0.92, I2 = 0[0–41]%