Background Immuno-oncology (I-O) plays a major role in the treatment of advanced melanoma (aMel); however, resistance to therapy remains an important clinical problem. This study examined treatment patterns and overall survival (OS) for aMel patients who progressed on anti-programmed death ligand 1 (anti-PD-1) therapy in a real-world clinical setting.

Methods A retrospective database study of Flatiron electronic medical records (EMR) was conducted with 304 aMel patients who progressed on first or second line anti-PD1 (baseline) therapy with pembrolizumab or nivolumab and received subsequent (index) therapy with ≥3 months of potential follow-up. Patients who discontinued treatment for reasons other than progression (primarily toxicity) were excluded. The primary outcome was OS, defined using EMR data linked to external mortality sources (e.g. Social Security Death Index). OS analysis was stratified by several factors (e.g. age, ECOG, BRAF, LDH, type of index therapy, and best overall response [BOR] to baseline anti-PD-1 therapy). BOR defined as response, stable disease, or disease progression was based on clinician assessment following radiographic imaging. Descriptive and log-rank test statistics for OS were used.

Results Among patients receiving index therapy (n=304), 50% received I-O (n=91/151 combination therapy), 36% received BRAFi/MEKi (n=102/109 combination therapy) and 14% received other therapies (n=34/44 chemotherapy). Median (range) age was 67 (23–85) years, with 65% male, 62% ECOG≤1, 33% elevated LDH, and 51% with BRAF mutations. Most patients received baseline anti-PD1 monotherapy (77%) as first line therapy. Median OS (95%CI) was 7.2 (6.4, 8.8) months, with a significant OS association with ECOG≤1 (p<0.001), normal LDH (p<0.001), and BRAFi/MEKi (p=0.02), with higher median OS of 9 vs 5 months, 11 vs 6 months, and 11 vs 7 and 6 months, respectively, compared to patients with ECOG≥2, elevated LDH, and treated with I-O and other therapies. For a subgroup of index therapy patients with a BOR assessment to baseline anti-PD-1 therapy (n=237), there was a significant association (p<0.01) of OS with BOR to baseline therapy, with higher median OS for those with an initial response (12 months) or stable disease (14 months) compared to a BOR of disease progression (6 months). There was also a significant OS association with BOR to baseline anti-PD-1 therapy for the subgroups receiving I-O therapy (n=119/237, p<0.01) and other therapies (n=37/237, p=0.01).

Conclusions Suboptimal OS in patients who progress on anti-PD-1 therapy in a real-world clinical setting, with predictors of enhanced survival, highlights the need for further research to inform optimal treatment strategies.

Acknowledgements The authors would like to acknowledge the contributions of Bo Zheng, Clemens Krepler, Diana Malandrucollo, and Shelby Marx of Merck & Co, Inc.