Fibroblast growth factor receptor 3 ( <i>FGFR3</i> ) alteration status and outcomes with immune checkpoint inhibitors (ICPI) in patients with metastatic urothelial carcinoma.

4564 Background: The use of immune checkpoint inhibitors (ICPIs) has expanded in the treatment of metastatic urothelial carcinoma (mUC) but response rates are variable, highlighting the need for preditive biomarkers. Tumor mutational burden (TMB) has been previously shown to predict response to ICPI but FGFR3 alterations are common drivers in mUC and there is preclinical and anectodal evidence that they may predict less favorable outcomes to ICPIs, similar to ALK and ROS1 fusions in lung cancer. We sought to explore the effect of FGFR3 alterations alone and with TMB on response to ICPI in mUC. Methods: 1,416 mUC patients who received hybrid-capture NGS-based genomic profiling were evaluated for their response to ICPI and chemotherapy treatment based on the presence of FGFR3 alterations (activating point mutations, insertions and deletions, rearrangements) and TMB. The nationwide (US-based) de-identified Flatiron Health-Foundation Medicine mUC clinico-genomic database of NGS results linked to deidentified electronic health record-derived clinical data originating from approximately 280 US cancer clincs (~800 sites of care) was used to assess treatment patterns and real-world overall survival (rwOS) and progression-free survival (rwPFS). Propensity analysis was used to match clinical characteristics between patients receiving first-line ICPI and chemotherapy and included age, disease grade, ECOG, and erdafitinib receipt as features. Results: Among 819 patients with mUC who received ICPI, there were no significant differences in rwOS or rwPFS between FGFR3 -altered (alt; n = 161) and wildtype (wt; n = 658) patients. However, among patients with TMB≥10 mut/Mb, FGFR3 -alt patients (n = 39) trended towards longer rwOS (20 vs. 14 months, aHR 0.62, 95% CI 0.37-1.02, p = 0.06) and rwPFS (5.5 vs. 4.9 months, aHR 0.66, 95% CI 0.42-1.03, p = 0.07) than FGFR3 -wt patients (n = 244). Comparing first-line ICPI vs. chemotherapy and adjusting for imbalances, patients with TMB≥10 and FGFR3 -alts who received ICPI (n = 21) also trended towards longer rwPFS than patients who received chemotherapy (n = 18) (14 vs. 7.1 months, HR 0.55, 95% CI 0.25-1.25, p = 0.2), although no significant difference in rwOS was observed. Conclusions: While FGFR3 status alone is not predictive of response to ICPI, FGFR3 combined with TMB emerged as a biomarker that may be predictive for response to ICPI in mUC and may have the potential to reconcile differences in previous observations regarding FGFR3 and ICPI response. Further studies performed with larger patient populations to confirm these findings are warranted.