Data collection and Data elements:
A retrospective cohort study of the National Cancer Database (NCDB) was performed. Jointly sponsored by the American College of Surgeons and the American Society, NCDB is a clinical oncology database sourced from hospital registry data representing more than 70% of newly diagnosed cancer cases nationwide. The database covers more than 1,500 Commission on Cancer (CoC)-accredited facilities. Definition of the database variables are available from the dictionary of NCDB Participant Use Data File (http://ncdbpuf.facs.org). The CoC's NCDB and the hospitals participating in the CoC NCDB are the source of the de-identified data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors.
Patient Cohort and Data analysis:
The NCDB was queried to analyze patients with HER2 + pT1N0M0 BC who received surgery from 2013 to 2017. Pathological staging data for the cohort was based on TNM classification in American Joint Committee on Cancer (AJCC) 7th edition [16]. Patients were excluded if they had metastasis, hormone receptor positive but did not receive endocrine therapy or were missing critical study information (e.g., pathologic stage or HER2 status).
The cohort was categorized by treatment status as immunotherapy alone (HER2 monotherapy) or immunotherapy plus chemotherapy (chemo-HER2 therapy). In the context of stage 1 HER2 + breast cancer, the only immunotherapy agent that patients could have received is trastuzumab, since pertuzumab was not FDA approved for use as adjuvant therapy for HER2 + breast cancer until 2017. The primary outcome was overall survival .
Analysis included univariate comparison of patient factors associated with chemo-HER2 therapy vs. HER2 monotherapy). To compare the two groups, Wilcoxon rank-sum test was utilized for continuous variables and chi-square for categorical data. Multivariable logistic regression determined the association between adjuvant chemotherapy and demographic/tumor factors. Overall survival (OS) differences between the two groups was analyzed using Kaplan-Meier survival estimates and compared by utilizing a log-rank test. To control for confounding effects, cox proportional hazard analysis was also performed. To further account for differences between patient cohorts receiving chemo-HER2 therapy vs. HER2 monotherapy, a 3:1 propensity match was performed, correcting for age, race, Charlson score, insurance, facility, tumor grade, hormone receptor status, mastectomy procedures, lymph nodes removed, lympho-vascular invasion status, and sub pathological T1 stage (T1a, T1b, T1c). The propensity match dropped observations of patients receiving HER2 monotherapy whose propensity score as higher than the maximum or less than the minimum score of the controls (chemo-HER2 therapy). 3:1 nearest neighbor matching replacement was used, with the caliper set to 0.01. Standardized differences before and after matching were reported and graphed during the analysis of propensity matching. Differences in survival among the matched pairs were analyzed using a stratified log-rank test and Cox proportional hazard regression analysis using a clustered “sandwich” robust variance estimator to account for clustering within the matched pairs.
All statistical analysis was performed using STATA/MP, version 16.0 (Stata Corp LLC, College Station, TX). Institutional Review Board (IRB) approval was exempted by the University Hospitals Cleveland Medical Center IRB as all data is de-identified.