A causal investigation of soy isoflavone intake for primary prevention of post-menopausal breast cancer among Asian women

The incidence of breast cancer is increasing at an alarming rate across Asia, by up to 6% annually, compared to near stable incidence rates in many Western countries. While selective oestrogen receptor modulators and aromatase inhibitors are actively being studied as chemoprevention among high-risk Caucasian women, the risks may outweigh the benefits among Asian women with lower population risk of breast cancer. Modifiable lifestyle targets for primary prevention have long been identified, such as post-menopausal obesity, alcohol intake, and hormone replacement therapy use, but these risk factors are less prevalent among women in Asian countries. There remains an urgent need to find primary prevention strategies that are low risk, acceptable, and effective for Asian women. Epidemiological evidence in Asian women suggests that high soy intake is associated with lower risk of breast cancer, but these findings were not observed in epidemiological studies of Caucasian women nor in clinical trials of soy isoflavone supplements. To date, there are no clinical trials that examine the effect of soy isoflavone intake from diet nor supplement on breast cancer risk among Asian women. In this thesis, I present the research studies undertaken to investigate if soy isoflavone intake is causally and inversely associated with post-menopausal breast cancer risk among Asian women. The objective of the first research study was to identify mammographic density measures that are suitable biomarkers of breast cancer risk in the target population (Chapter 3). In this study, volume-based mammographic density measures and breast cancer risk factors were compared between 1,501 Malaysian women and 4,501 age- and BMI-matched Swedish women with no personal history of cancer. The analysis demonstrated that absolute dense volume, rather than percent density, may be a better biomarker of breast cancer risk among post-menopausal Asian women. Based on the above findings, the second research study sought to determine if mammographic density mediates the association between soy intake and breast cancer risk in the target population (Chapter 4). A cross-sectional analysis of 3,277 healthy Malaysian women showed that mammographic density was lower among women with frequent soy intake compared to non-consumers, by up to 2.5cm3 dense volume or 2.0cm2 dense area, but this was not statistically significant. Intriguingly, regular soy intake was associated with lower mammographic density among overweight or obese women, but for leaner women, regular soy intake was associated with higher mammographic density. This interaction was statistically significant among pre-menopausal women (pinteraction = 0.029). Prior to designing a robust clinical trial to test the causal association between soy intake and mammographic density as a biomarker of breast cancer risk, the feasibility of a dietary soy intervention was assessed in a small sample of the target population (n=10, Chapter 5). Overall, women in the study were able to maintain a diet of 70-90mg/day of soy isoflavones for 2 months, but the target of 100mg/day was not feasible and may have led to some adverse events. Thematic analysis of semi-structured interviews revealed that women participated in the study for altruistic reasons and due to emotional attachments to the cause, and that adherence was largely influenced by the practicability of the new diet or routine. Building from the results of the previous three chapters, the primary objective of the last research study was to test the effect of daily soy isoflavone intake for 1 year on breast cancer risk among Asian women, using mammographic density as a biomarker of risk (Chapter 6). In this clinical trial, 57 healthy post-menopausal Malaysian women were randomized into the Supplement arm (100mg/day isoflavones, with >90% daidzein), the Dietary Soy arm (50mg/day isoflavones), or the Control arm. After 1 year of intervention, women in the Supplement arm experienced 4.1cm2 lower dense area and 2.4% lower area-based percent density compared to women in the Control arm, but these associations were not statistically significant. The associations were weaker for women in the Dietary Soy arm and for volume-based mammographic density measures. Interestingly, stronger associations were observed when the analysis was limited to women with high BMI or low dietary fat intake, but the sample size was too small for robust analyses. In conclusion, the data presented in this thesis suggest a causal association between soy isoflavone intake and lower post-menopausal breast cancer risk among Asian women. However, due to the small sample size, the analysis was underpowered to show statistically significant effects and will require confirmation in a larger trial. Nonetheless, the research undertaken here adds to existing evidence that the soy isoflavone daidzein may be responsible for the protective effect of soy. Furthermore, it proposes new hypotheses in understanding the association between soy intake and breast cancer risk across populations, including possible effect modification by BMI or dietary fat intake