Erda is an oral pan-FGFR tyrosine kinase inhibitor approved to treat locally advanced or mUC in pts with susceptible FGFR3/2alt who have progressed after platinum-containing chemotherapy. FGFRalt tumors are enriched in luminal 1 subtype and may have limited clinical benefit from anti–PD-(L)1 treatment (tx). This cohort in the randomized, open-label phase 3 THOR study (NCT03390504) assessed erda vs pembro in pts with mUC naive to anti–PD-(L)1.
Pts ≥18 y with unresectable advanced/mUC with select FGFR3/2alt, ECOG PS 0-2, disease progression on 1 prior tx, and naive to anti–PD-(L)1 were randomized 1:1 to receive erda 8 mg QD with pharmacodynamically guided uptitration to 9 mg or pembro 200 mg Q3W. Primary end point: overall survival (OS). Secondary end points: progression-free survival (PFS), objective response rate (ORR), safety.
At 15 Jan 2023 data cutoff, intent-to-treat set (median follow-up 33.2 mo) comprised 175 pts in erda and 176 in pembro arm. The primary end point was not met, with no statistically significant difference in OS between tx arms (Table). Erda had numerically longer PFS and numerically higher ORR but shorter duration of response. Most common tx-related adverse events (TRAEs) of any grade: hyperphosphatemia (73%), stomatitis (45%), diarrhea (45%), and dry mouth (35%) with erda; pruritus (12%), asthenia (10%), hypothyroidism (10%), and fatigue (10%) with pembro. Gr 3-4 TRAEs and serious TRAEs occurred in 43% and 13% in erda and 12% and 10% in pembro arm, respectively. TRAEs leading to death occurred in 0 pts in erda and 3 (2%) in pembro arm. 15% of pts discontinued erda and 5% discontinued pembro due to TRAEs. Table: 2359O
| Erda (n=175) | Pembro (n=176) | |
| OS, median, mo | 10.9 (9.2-12.6) | 11.1 (9.7-13.6) |
| Hazard ratio | 1.18 (0.92-1.51) p=0.18 | |
| PFS, median, mo | 4.4 (4.1-5.5) | 2.7 (1.6-3.0) |
| Hazard ratio | 0.88 (0.70-1.10) p=0.26a | |
| ORR, % | 40.0 (32.7-47.7) | 21.6 (15.8-28.4) |
| Relative risk | 1.85 (1.32-2.59) p<0.001a | |
| DOR, median, mo | 4.3 (3.7-6.9) | 14.4 (7.4-27.8) |
All data are value (95% CI). aNominal p estimate, due to primary end point not being met.
Conclusions
Erda and pembro had similar OS in this anti–PD-(L)1 naive, FGFRalt mUC population, with erda showing a numerically longer PFS and numerically higher ORR rate. Toxicities were manageable with dose modifications.