Aspirin and celecoxib use and overall survival in ER+/HER2- metastatic breast cancer: A large real-world cohort analysis

Background: Endocrine resistance in ER+/HER2- breast cancer (BC) is frequently linked to PI3K-AKT-mTOR pathway dysregulation. While aspirin benefits PIK3CA -mutant colorectal cancer, randomized trials of adjuvant endocrine therapy ± COX inhibition in BC failed to demonstrate survival benefits, potentially due to lack of biomarker selection. Given that ER+/HER2- metastatic disease is clinically enriched for PI3K pathway activation, we hypothesized that real-world aspirin and celecoxib (COXI) use might be associated with improved survival in this population. This study evaluated the association between COXI use and survival in ER+/HER2- metastatic BC, utilizing this subtype as a clinical proxy for PIK3CA pathway activation. Methods: Using the TriNetX US Collaborative Network, we identified patients with ER+/HER2- metastatic BC via ICD-10 codes and biomarker fields. Patients were grouped into three prevalent-user cohorts based on post-diagnosis medication use: (1) control (no use); (2) single-agent COXI; and (3) combination COXI (aspirin and celecoxib). Propensity score matching (1:1) was utilized to balance cohorts for demographics, comorbidities, cardiovascular risk factors, and concomitant systemic therapies (endocrine agents, chemotherapy, and CDK4/6 inhibitors). We estimated Hazard Ratios (HRs) for overall mortality and Major Adverse Cardiovascular Events (MACE: myocardial infarction, stroke, and cardiovascular death). Results: After matching, 14,955 patient pairs (Single-agent vs. Control) and 1,913 patient pairs (Combination vs. Control) were analyzed. The Combination cohort demonstrated the lowest mortality hazard (HR, 0.429; 95% CI, 0.379-0.486; p < 0.0001). The Single-agent cohort also showed a significantly lower mortality hazard compared to controls (HR, 0.629; 95% CI, 0.604-0.656; p < 0.0001). Conversely, the hazard of MACE was significantly higher in both treated groups compared to controls (Combination: HR, 1.492; Single-agent: HR, 1.641; both p < 0.0001). Conclusion: In this large real-world cohort of ER+/HER2- metastatic breast cancer, aspirin and/or celecoxib use was associated with a graded improvement in overall survival. This association persisted despite the treated groups showing a significantly higher risk of MACE. This difference, even after PSM, reflects residual confounding due to bias by indication. This suggests the survival improvement is driven by a genuine oncologic effect, mitigating concerns about "healthy user" bias. However, findings must be interpreted cautiously given the observational design and potential for biases such as immortal time bias and residual confounding. PIK3CA genomic data was sparse in this dataset, precluding genotype-stratified analyses. These results support further investigation via prospective, biomarker-driven trials specifically targeting PIK3CA -associated ER+/HER2- disease.