Background: Soy foods and isoflavones have been hypothesised to be associated with cancer risk through phytoestrogenic, antioxidant, and anti-inflammatory mechanisms. Epidemiological evidence remains inconsistent, and prior reviews have not comprehensively evaluated dose-response relationships, cancer subtypes, or biomarker-based exposures. We conducted a systematic review and meta-analysis within the Word Cancer Research Fund International CUP Global framework to evaluate soy and isoflavone intake in relation to cancer incidence and mortality.
Objectives: To systematically evaluate and quantify associations between soy foods, dietary and circulating isoflavones, and the risk of site-specific cancer incidence and mortality in adults.
Methods: PubMed and Embase were searched up to 31 March 2025 for cohort studies and randomised controlled trials (RCTs) reporting associations between soy, isoflavones, or related biomarkers and risk of cancers. Summary relative risks (RRs) and 95% confidence intervals (CIs) were estimated using random-effects models. Risk of bias was assessed using the RoB-NObs tool for cohort studies and Cochrane RoB 2 tool for RCTs.
Results: A total of 128 cohort publications (49 studies) and 5 RCTs were included, covering 18 major cancer sites. Meta-analyses were conducted for 146 exposure–outcome pairs. Higher dietary isoflavones intake was associated with lower risk of breast cancer (RR per 10 mg/day: 0.93, 95%CI: 0.96-0.99; I2=18%, n=6), particularly postmenopausal (RR per 10 mg/day: 0.95, 0.91-0.99; I2=44%, n=8) and hormone receptor–positive subtypes (RR high vs low: 0.79, 0.62-0.99; I2=0, n=2), with evidence of benefit in the non-linear analysis from intake of >40 mg/day. Biomarker studies supported dietary findings, with higher plasma genistein concentrations linked to lower breast cancer risk (RR high vs low: 0.62, 0.38-1.00; I2=13%, n=2). Inverse associations were also observed for non-fermented soy foods intake and distal stomach cancer and (RR per 25 g/day: 0.82, 0.69-0.98; I2=0, n=2), as well as soy foods and lung cancer (RR high vs low: 0.92, 0.84-1.00; n=12) and isoflavones intake and lung cancer (RR per 10 mg/day: 0.87, 0.77-0.99; I2=48%, n=4), which was confirmed in never smokers (RR per 10 mg/day: 0.84, 0.76-0.93; I2=0, n=6). Positive associations were observed for soy foods or isoflavones intake in relation to advanced prostate, pancreatic, liver, and bladder cancers; however, these findings were based on a small number of studies, were not consistent across studies, and often lacked dose-response evidence. Most other cancer sites showed null associations. RCTs were small, short in duration, and provided limited evidence. Studies were generally at moderate risk of bias, with greatest concerns for confounding, exposure misclassification and reliance on single baseline assessments; fewer studies incorporated repeated exposure measurements, which may have attenuated or obscured true associations.
Conclusions/Implications: This comprehensive synthesis suggests that higher intake of soy isoflavones may be associated with a lower risk of breast and lung cancers, while evidence for other cancer sites remains limited or inconsistent. Findings highlight heterogeneity by cancer type and soy product, underscoring the need for more nuanced dietary guidance. Future research should prioritise large, diverse cohorts with repeated exposure assessment, biomarker validation, and improved confounder control, as well as adequately powered trials to clarify causality.