Lifestyle Risk Factors and Rise of Breast Cancer Overall and by Subtypes Defined by Hormone Receptor Status
Access StatusFull text of the requested work is not available in DASH at this time ("dark deposit"). For more information on dark deposits, see our FAQ.
MetadataShow full item record
CitationWu, You. 2020. Lifestyle Risk Factors and Rise of Breast Cancer Overall and by Subtypes Defined by Hormone Receptor Status. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
AbstractWorldwide, breast cancer is the most commonly diagnosed malignancy in women with over 2 million cases diagnosed each year. Breast cancer is a heterogeneous disease that could be classified into subtypes based on the molecular features, for example, by hormone receptor status. The majority of breast cancer cases are estrogen receptor (ER) positive, and less than 20% are ER- negative. These subtypes have different etiologies, clinical characteristics, and survival rates. The effects of risk factors may also differ by hormone receptor status.
For Chapter 1 through 3, I examined potential dietary factors in three pooled analyses of diet and risk of breast cancer in an international consortium of more than 20 prospective cohorts. Over 1 million women were included in the analyses, among whom around 40,000 incidence breast cancer cases were documented. In each chapter, a 2-stage approach was used for data analysis. In stage 1, Cox proportional hazard regression was used to get study-specific hazard ratios for breast cancer overall and the subtypes defined by ER status. In stage 2, the study-specific multivariable HRs were pooled using random-effects model.
Chapter 1 focused on dietary fiber as an example of nutrient. We found that dietary fiber intake was inversely associated with breast cancer risk. The association could mainly be attributed to fiber from fruits and vegetables, but not grains, and was modified by fat intake, where the association became weaker and nonsignificant among people with higher fat intake. In subtype analysis, the association was stronger for ER- tumors than that for ER+, although difference by ER status reached statistical significance only for fiber from vegetables.
In Chapter 2, dairy products were examined as examples of food items. Individual dairy products showed null or very weak inverse associations with risk of overall breast cancer. Differences by ER status were suggested for yogurt and cottage/ricotta cheese where associations were observed for ER-negative tumors only. Dietary calcium intake was only weakly associated with breast cancer risk, and the effect estimates did not differ by ER status.
In Chapter 3, the focus was on red meat and other major protein sources, to explore the substitution effects of different food groups on risk of breast cancer. Total red meat, processed meat, and unprocessed meat intakes were not significantly associated with risk of breast cancer when holding other protein sources constant. However, inverse associations were observed when substituting red meat with an energy-equivalent amount of mature beans or dairy products. There were no significant substitution effect replacing total or unprocessed red meat with poultry, seafood, eggs, or nuts. The results were similar for ER-positive and ER-negative breast cancer.
To quantify the theoretical impact of interventions, population attributable risk (PAR%) helps set priority and guide personal decision. Wide range of PAR% of cancer incidence by modifiable risk factors has been reported, yet there is no consensus on what contributed to the variation.
Chapter 4 investigated the PAR% of breast cancer by a group of modifiable risk factors and examined its variation by choices of exposures and methods. Fruits and vegetable intake, physical activity, adult weight gain, and alcohol consumption were the exposures of interest in the analysis. Partial PAR% was calculated from three models: baseline only, simple updates of repeated measures, or cumulative averages of repeated measures. For each model, two methods were applied - one based on the four factors individually, and the other based on comparison between an overall high- versus low-risk group. The models based on repeated measures yielded greater estimated PAR%. PAR% estimates by the low-risk method were higher than that based on each factor individually, but in similar pattern. Therefore, PAR% by modifiable risk factors in current literatures likely underestimated the preventable fraction, if relied on studies with baseline data only.
In conclusion, the first three chapters found modest inverse associations with risk of breast cancer for dietary fiber, especially that from fruits and vegetables; for yogurt and cottage cheese consumption and ER- tumors; and when substituting red meat by beans or dairy. The last chapter emphasized the importance of high-quality repeated measure data in PAR% calculation and called for cautious interpretation of PAR% in the current literature.
Citable link to this pagehttps://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37365723
- FAS Theses and Dissertations