Patients with breast cancer (BC) and concomitant comorbidity have poorer disease prognosis, which may be related to a reduction in the receipt of curative cancer treatment. This study sought to determine the survival impact of BC treatment concordance in relation to comorbidity burden.
Incident cases of unilateral stage I-III BC diagnosed between 2000 and 2015 were identified from a prospectively collected New Zealand BC register. Comorbidity severity was measured by C3 index score; derived via linkage with national hospitalization data. Treatment concordance (for breast and axillary surgery, adjuvant radiotherapy, adjuvant chemotherapy and endocrine therapy) was assigned in relation to minimum treatment indications as per national tumor standards and St Gallen Consensus Statements from relevant years. Multiple imputation of missing data was performed. Propensity scores for the conditional probability of receiving each treatment modality were modelled and used to create inverse probability of treatment- (IPTW) and standardized mortality ratio-weighted (SMRW) samples which were balanced with respect to baseline confounding variables between treated and untreated groups. For each treatment modality, weighted Cox proportional hazards (for all-cause mortality) and competing risks regression models (for BC-specific mortality) were produced, giving hazard ratio’s (HR’s) and sub-distribution hazard ratio’s (SHR’s) respectively as estimates of the average treatment effect (ATE) in IPTW samples and the average treatment effect on the treated in SMRW samples. Treatment effect heterogeneity by comorbidity severity was evaluated through the use of interaction tests and confirmatory subgroup analysis.
A total of 12,834 patients were included, with 21.5% possessing one or more pre-existing comorbidities. Patients with comorbidity were less likely to receive all four modalities of breast cancer treatment. Comorbidity burden was associated with poorer age and stage-adjusted survival, with greater impact on all-cause than BC-specific mortality. Overall, and for patients without comorbidity, those who received guideline-concordant treatments experienced significant reductions in all-cause and BC mortality risk (Tables 1 and 2; ATE’s from IPTW samples shown only). Heterogeneity in ATE’s by comorbidity severity was noted for surgery (BC mortality, p=.03) and endocrine therapy (all-cause mortality, p=.003), with lesser benefit at higher levels of comorbidity. Amongst patients with the greatest comorbidity severity (C3 score >2.00), surgery to the breast and axilla reduced the risk of all-cause but not BC mortality, with no mortality benefits obtained from the addition of adjuvant therapies.
Patients with BC and concurrent comorbidity do not obtain the same survival benefits from guideline-concordant treatment as their non-comorbid counterparts. The inferior survival of such patients is therefore likely to be mediated through mechanisms other than reduced receipt of curative treatment.