Aromatase inhibitor use, co-morbidities, and cardiovascular disease risk in breast cancer survivors

Breast cancer survivors are a growing population with significant comorbidity burden. Approximately 80 percent of breast cancer cases are hormone receptor-positive and can be treated by endocrine therapy. Aromatase inhibitors (AIs) and tamoxifen are the two major types of endocrine therapies. Although AIs have more favorable breast cancer clinical outcomes than tamoxifen, AI-related cardiovascular disease (CVD) risk is a growing concern. Previous observational studies investigating the association between AI use and CVD risk have inconsistent results and may be biased for a few reasons, including not accounting for important confounders, not adequately accounting for immortal time bias, and misclassification of endocrine therapy use. Despite breast cancer guidelines and results from clinical trials emphasizing the importance of using AIs for more than 5 years, few studies have examined the duration of AI exposure when investigating AI-related CVD risk. In addition, no prior studies have examined the association between AI use and CVD risk in subgroups defined by breast cancer comorbidities, such as obesity, depression, and history of CVD prior to breast cancer diagnosis. Therefore, we leveraged data from the Pathways Heart Study and conducted three studies to address the gaps in the literature: In Chapter 1, we examined the associations of endocrine therapy use with CVD and mortality risk in postmenopausal women with early-stage hormone receptor-positive breast cancer and whether the associations varied in strata defined by obesity, depression, and history of CVD prior to breast cancer diagnosis. We found that, compared with non-user of endocrine therapy, women who used AIs within 12 months after breast cancer diagnosis were less likely to develop major adverse cardiovascular events (hazard ratio [HR]= 0.85, 95% confidence interval [CI] =0.73 to 0.98) and heart failure (HR = 0.79, 95% CI = 0.65 to 0.97). No associations between tamoxifen use within 12 months and CVD outcomes were observed. Both AI and tamoxifen use within 12 months after breast cancer diagnosis were associated with lower risk of all-cause mortality and non-CVD-related mortality, and AI users were at approximately 30% lower risk of CVD-related mortality, compared with non-users of endocrine therapy. Tamoxifen users with normal body mass index (BMI) were at decreased risk of all-cause mortality (HR = 0.66, 95% CI = 0.51 to 0.87) and non-CVD-related mortality (HR = 0.57, 95% CI = 0.42 to 0.79) compared with non-users of endocrine therapy with normal BMI, whereas no associations of tamoxifen use with all-cause mortality and non-CVD-related mortality were observed in the overweight/obese subgroup (P for interaction <0.05). Our findings suggest that endocrine therapy reduces all-cause mortality risk and may not increase CVD risk in postmenopausal women with early-stage hormone receptor-positive breast cancer.In Chapter 2, we examined the associations of duration of AI use with CVD and mortality risk in postmenopausal women with early-stage hormone receptor-positive breast cancer and whether the associations varied in strata defined by obesity, depression, and history of CVD prior to breast cancer diagnosis. We observed a lower risk of stroke in intermediate-term AI users (HR = 0.60, 95% CI = 0.37 to 0.96) and long-term AI users (HR = 0.51, 95% CI = 0.30 to 0.85), than in short-term AI users. The longer duration of AI use was also associated with lower risk of all-cause mortality and non-CVD-related mortality. In addition, long-term AI users were at 37% lower risk of CVD-related mortality (HR = 0.63, 95% CI = 0.42 to 0.95) than short-term AI users. No significant differences were observed in risks of major adverse cardiovascular events, ischemic heart disease, and heart failure across the three groups. The association between long-term AI use and non-CVD-related mortality differed in women with versus without a history of CVD prior to breast cancer diagnosis (P for interaction = 0.01); long-term AI users without CVD history were at lower risk of non-CVD-related mortality (HR = 0.47, 95% CI = 0.37 to 0.60) than short-term AI users without CVD history, whereas no such association was observed in the subgroup with CVD history. Our findings suggest that long-term AI use reduces all-cause mortality risk and may not increase CVD risk in postmenopausal women with early-stage hormone receptor-positive breast cancer. In Chapter 1 and 2, we examined the association between endocrine therapy use and CVD risk stratified by depression defined by the CES-D scale. However, only a small portion of participants completed the CES-D questionnaire, and thus our stratified analyses were underpowered. If the CES-D scale shows good agreement with ICD codes for depression, which are available for all study participants, we can use ICD codes to define depression and the stratified analysis by depression would have better power in Chapter 1 and 2. Hence, In Chapter 3, we leveraged data from the Pathways Study and examined the agreement between the self-reported Center for Epidemiologic Studies Depression (CES-D) scale for depressive symptoms and International Classification of Diseases (ICD)-9 diagnosis codes for depression in breast cancer survivors. We found that a total of 1,015 (25.7%) women had a CES-D score of 16 or more and were classified as depressed, while a total of 660 (16.7%) women were classified as depressed using ICD-9 codes. Depressive symptoms defined by the CES-D scale and depression defined by ICD-9 codes had slight agreement (kappa = 0.16). Their classification was significantly different (P <0.0001). Using the CES-D scale as the reference standard, the sensitivity and specificity of ICD-9 codes for depression were 42.3% and 77.6%, respectively. Our findings suggest that ICD-9 codes are not a valid measure of depression in epidemiology studies of breast cancer.