Adverse Events and Perception of Benefit From Duloxetine for Treating Aromatase Inhibitor-Associated Arthralgias

BACKGROUND: Duloxetine effectively treats aromatase inhibitor-associated musculoskeletal symptoms (AIMSS) in women with breast cancer but causes low-grade toxicities. This secondary analysis examines the relationship between adverse events (AE) and patient-perceived benefit, based on patient self-report that the treatment received was beneficial despite side effects. We hypothesized that duloxetine had a favorable effect on patient-perceived benefit, even among duloxetine-treated patients who experienced AEs and who, had they been treated with placebo, would have experienced none. METHODS: Principal stratification was used to estimate the effect of duloxetine vs placebo on patient-perceived benefit and Functional Assessment of Cancer Therapy-Endocrine Scale functional quality of life in the randomized, double-blind trial SWOG S1202 (n = 289). Subgroups of patients were defined by observed and counterfactual (what would have occurred had they been randomly assigned to the opposite study arm) experiences of AEs and the original primary outcome, reduction of average pain after 12 weeks of at least 2 points on the Brief Pain Inventory-Short Form. RESULTS: Duloxetine caused an estimated 23.4% (95% credible interval [CI] = 13.4% to 33.7%) of patients to experience an AE even though they would have experienced none on placebo. Those patients remained more likely to report that their received treatment was beneficial than comparable patients assigned placebo (73.3% vs 41.8%, respectively; 95% CI for difference = 15.4 to 47.2 percentage points), although there was no statistically significant effect of duloxetine on functional quality of life (11.3 vs 9.0, 95% CI for difference = -2.2 to +6.7). CONCLUSION: Duloxetine resulted in higher patient-perceived benefit, even among those who would have an AE on duloxetine but none on placebo. Treatment of AIMSS with duloxetine should be considered for appropriate patients.