Examining Risk of Heart Failures With Trastuzumab (Herceptin) Use Among Patients With HER2+ Breast Cancer Using Surveillance, Epidemiology, and End Results (Seer) Medicare: A Retrospective Cohort Study

Abstract

Background: Advancements in human epidermal growth factor receptor 2 (HER2) targeted therapies such as trastuzumab have improved survival rates in HER2-positive (+) breast cancer. Still, cardiovascular complications pose significant concerns. Importantly, understanding risk factors for trastuzumab-induced cardiotoxicity and appropriate patient monitoring during trastuzumab treatment allows for safe and effective use. However, existing studies have yielded inconsistent findings regarding risk factors for trastuzumab-induced heart failure (HF)/congestive heart failure (CHF) or cardiotoxicity. Aims and objectives: This dissertation is structured to examine risk factors of CHF in patients with HER2+ breast cancer treated with trastuzumab. The work reported in this dissertation offers a general overview of breast cancer and the treatment landscape for HER2+ breast cancer. Three main objectives are further explored and reported in a two-article format. The first article addresses objectives (1) and (2); a systematic literature review to identify peer-reviewed studies on risk factors of trastuzumab induced cardiotoxicity in patients with HER2+ breast cancer, supported by a meta-analysis to synthesize the evidence collated from these studies regarding assessed risk estimates. The second article, objective (3) is a retrospective cohort study utilizing an earlier version of the 2010-2019 linked Surveillance, Epidemiology, and End Results (SEER)-Medicare dataset to identify risk factors attributed to the development of CHF in women aged 65 years and older with HER2+ breast cancer treated with trastuzumab. And, to explore the feasibility of machine learning methods and application with real world data which is gradually gaining grounds in health outcomes and oncology research. Methods: We conducted comprehensive searches of the literature in PubMed, Medline, Embase, Scopus, and Cochrane Library databases from inception to February 2023. We included grey literature and American Society of Clinical Oncology conference proceedings. Inclusion criteria encompassed English studies on trastuzumab and HER2+ breast cancer. Two reviewers independently screened titles/abstracts, assessed full texts, and extracted data. We evaluated the risk of bias using appropriate tools. We performed a meta-analysis of odds ratios (OR) of risk factors using the inverse variance method and reported results following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We adopted a proof-of-concept approach using an earlier version of the linked SEER-Medicare databases for the retrospective cohort study. We included female patients aged 65 years and older with a histologically confirmed new primary diagnosis of breast cancer reported between January 1, 2010, and 2019. We included patients enrolled in Medicare Parts A, B and D. We did not include patients enrolled in Part C (Medicare Advantage/ managed care/ a health maintenance organization; HMO). We included only trastuzumab based-treated patients with HER2+ breast cancer who were CHF-free at first dose of trastuzumab. We defined the index as the date a patient received trastuzumab for the first time after breast cancer diagnosis. We used multivariate logistic regression model and explored two machine learning (ML) classification algorithms; random forest (RF) and adaptive boost (adaboost) to identify risk factors associated with CHF among HER2+ breast cancer patients treated with trastuzumab. Results: We included 18,099 patients with HER2+ breast cancer from 28 articles from the literature review in the meta-analysis. The estimated crude incidence of cardiotoxicity was 10.1%. Hypertension (OR=1.69; 95% CI=1.39-2.04) and diabetes mellitus (OR=2.02; 95% CI=1.33-3.07) were significant risk factors. Anthracycline use had mixed results. Publication bias and risk of bias were observed in some studies. We included a total of 4,922 patients from the SEER-Medicare databases in the retrospective study. The median age at trastuzumab treatment initiation was 72 years (range; 65 to 100 years). The median follow-up for the entire cohort was 35.0 months. CHF was identified in 9.7% of the patients. An additional treatment cycle of trastuzumab (odds ratio [OR]= 0.99, 95% confidence interval [CI]; 0.97 to 1.01) was not an independent significant risk factor of CHF per the logistic regression model. Advanced age (greater than 83 years ; OR 1.73, 95% CI; 1.18 to 2.54), atrial fibrillation and flutter (OR= 3.32, 95% CI; 2.14 to 5.15), SEER’s combined summary cancer stage regional by direct extension HER2+ breast cancer (OR 1.34, 95% CI; 1.08 to 1.67), diabetes with complications (OR =1.63, 95% CI; 1.10 to 2.41), ischemic heart disease (OR= 1.65, 95% CI; 1.23 to 2.21) and obesity (OR= 1.51, 95% CI; 1.14 to 1.99) were significant risk factors of CHF per the logistic regression model. The RF model identified atrial fibrillation and flutter, diabetes with complications, and ischemic heart disease as the top three important variables. Adaboost identified total treatment cycles of trastuzumab, region of the United States, and age at trastuzumab initiation as the top three important variables of CHF. The RF model had the best performance with an area under the curve of 0.68. Conclusion: CHF is not uncommon in breast cancer therapies. Identifying risk factors associated with CHF in patients with HER2+ breast cancer undergoing trastuzumab therapy highlights the need for personalized therapeutic and monitoring approaches. Continued research, particularly through “real world” observations, is necessary to mitigate existing limitations and enhance the quality of patient management. Finally, the work and results reported here reflect efforts from a proof-of-concept approach due to using the earlier version of the 2010-2019 SEER-Medicare dataset, hence data should be retested to make clinical decisions in practice.