Sequencing of erdafitinib (erda) and enfortumab vedotin (EV) in patients (pts) with fibroblast growth factor receptor (FGFR2/3) altered (alt) advanced urothelial cancer (aUC): Analysis of UNITE database.

Abstract

616 Background: The optimal therapy sequencing in aUC with FGFR2/3 alt is unclear. FGFR inhibition may increase Nectin-4 expression. We hypothesized that outcomes with EV would be more favorable when given after E compared to before E in pts with aUC harboring FGFR2/3 genomic alt. Methods: Pts with FGFR2/3 alts who received EV only (EVo), E then EV (E->EV), or EV then E (EV->E) were included. Overall survival (OS) measured from E or EV start (based on agent given first) and progression-free survival (PFS) measured from EV start were assessed with KM method and group comparison by log rank test with Cox proportional hazards regression model. Variables with p ≤0.2 were included in multivariate analysis (MVA). Wilcoxon rank sum and Fisher’s exact test were used to compare continuous and categorical variables, respectively. Observed response rate (ORR) was compared between groups using χ2 test. Results: Among 633 pts from 16 US sites, 487 had NGS data, and 94 (19%) had FGFR2/3 alterations with 24 receiving E->EV, 15 EV->E, and 55 EVo. Median age: 72 yrs; 87% Caucasian, 65% primary bladder tumor, 73% pure urothelial histology, 76% BMI <30, 83% ECOG PS 0-1, 60% visceral metastases (mets) with 23% liver mets, 93% prior IO, and 63% prior Plt. Pts treated with E->EV vs EV->E had similar baseline characteristics, but at EV start ECOG PS 0-1 was 93% in EV->E vs 67% in E->EV (p=0.04). ORR to EV for E->EV pts was 32% (CI- 12%-51%) vs 67% (CI 54%-91%) for EV->E pts (p=0.04). ORR for EVo was 49%. Median OS was 21 months (mo) E->EV, 19 mo EV->E, and 12 mo EVo; median PFS was 5, 5, 6 mo, respectively. Univariate analysis (UVA) and MVA results are shown in Table. In MVA, longer OS was seen in pts who received both EV and E vs EVo. Conclusions: We showed longersurvival in pts with FGFR2/3 alt aUC who received both EV and E regardless of sequence, which may reflect guarantee-time bias. Additional limitations include retrospective nature, modest sample size, no central scan review, lack of data for pts who received E only, selection and confounding biases. Larger prospective studies are needed to determine optimal sequencing and putative biomarkers. [Table: see text]