Clinical and Genomic Landscape of FGFR3-Altered Urothelial Carcinoma and Treatment Outcomes with Erdafitinib: A Real-World Experience

Authors

Brendan J. Guercio

Michal Sarfaty

Min Yuen Teo

Neha Ratna

Cihan Duzgol

Samuel A. Funt

Chung-Han Lee

David H. Aggen

Ashley M. Regazzi

Ziyu Chen

Michael Lattanzi

Hikmat A. Al-Ahmadie

A. Rose Brannon

Ronak Shah

Carissa Chu

Andrew T. Lenis

Eugene Pietzak

Bernard H. Bochner

Michael F. Berger

David B. Solit

Jonathan E. Rosenberg

Dean F. Bajorin

Gopa Iyer

Doi

10.1158/1078-0432.CCR-23-1283

PMID: 37682528 · DOI: 10.1158/1078-0432.CCR-23-1283 · Journal: Clinical Cancer Research (2023)

TL;DR

Guercio et al. integrated MSK-IMPACT tumor sequencing and serial MSK-ACCESS cfDNA profiling on 1,421 patients (1,507 tumors) with urothelial carcinoma to map the landscape of FGFR2/3 alterations and characterize real-world outcomes on erdafitinib in 32 metastatic patients. Oncogenic FGFR3 alterations were common (27.5% overall, enriched in non-muscle-invasive and upper-tract disease), FGFR2 actionable alterations were vanishingly rare (1 fusion), and primary/metastatic FGFR3 status was discordant in 26% (7/27) of paired cases. Erdafitinib produced a 40% response rate but median PFS of only 2.8 months and OS of 6.6 months, with toxicity-driven dose reductions (38%) and interruptions (50%); cfDNA identified putative resistance mutations in TP53 (n=5), AKT1 (n=1), and second-site FGFR3 kinase-domain mutations including V553M and N540S (PMID:37682528).

Cohort & data

  • 1,421 patients / 1,507 urothelial carcinoma tumors sequenced at MSK (largest single-institution UC cohort to date); cancer types: BLCA and UTUC (PMID:37682528).
  • Disease-state breakdown: 504 NMIBC, 526 MIBC, 187 localized upper-tract, 228 distant metastatic specimens (PMID:37682528).
  • Erdafitinib-treated subcohort: 32 patients with metastatic urothelial carcinoma treated between 8/2019 and 7/2022 (PMID:37682528).
  • Tumor assay: MSK-IMPACT FDA-authorized targeted panel; cfDNA assay: MSK-ACCESS (129 genes, >15,000x, 0.1% VAF threshold); copy number via FACETS; oncogenicity via OncoKB (PMID:37682528).
  • Dataset: bladder_msk_2023 on cBioPortal (PMID:37682528).

Key findings

  • Oncogenic FGFR3 alterations frequencies by state: 39% (199/504) NMIBC, 14% (75/526) MIBC, 43% (81/187) localized UTUC, 26% (59/228) metastatic (PMID:37682528).
  • Only 1/1,507 tumors had a potentially actionable FGFR2 alteration (an FGFR2:MARVELD3 fusion of unclear significance) (PMID:37682528).
  • FGFR3 S249C was the most frequent alteration overall; R248C was enriched in upper-tract (22%, 18/81) vs bladder (11%, 37/333; p=0.01) tumors (PMID:37682528).
  • Primary/metastatic FGFR3 discordance in 7/27 (26%) paired cases: 4 metastasis-only, 2 primary-only, 1 with different FGFR3 alterations in each site (PMID:37682528).
  • PI3K pathway co-alterations in FGFR2/3-altered tumors: PIK3CA 28% (115/414), TSC1 13% (52/414), AKT1 2.7% (11/414); total 37% (PMID:37682528).
  • ERBB2 co-alteration in 5% (19/414) of FGFR3-altered tumors, ranging 2.5% NMIBC to 11% MIBC (PMID:37682528).
  • FGFR3-altered tumors showed inverse association with ERBB2, TP53, and RB1 alterations and frequent co-alteration of CDKN2A, CDKN2B, and KDM6A (PMID:37682528).
  • FGFR3-fusion tumors had lower TMB than FGFR3-mutant tumors (median 5 vs 9 mut/Mb, p=0.0006) (PMID:37682528).
  • Erdafitinib ORR 40% (12/30 evaluable); median PFS 2.8 mo (95% CI 2.6–5.4); median OS 6.6 mo (95% CI 4.6–17.3) (PMID:37682528).
  • Dose reductions in 38% (12/32), interruptions in 50% (16/32); TEAEs: hyperphosphatemia 84%, fatigue 59%, mucositis 47% (grade ≥3 16%), PPE 38% (grade ≥3 9%), dysgeusia 31% (PMID:37682528).
  • No PFS/OS difference on immune checkpoint blockade by FGFR3 status (n=26 altered vs 155 wildtype; p=0.47/0.52) (PMID:37682528).
  • Baseline TP53 co-alteration associated with lower ORR (22%, 2/9 vs overall 40%) and enriched among primary progressors (3/5, 60%) (PMID:37682528).
  • Serial cfDNA (n=27) identified on-treatment acquired TP53 (n=5/27), FGFR3 (n=2/27), and AKT1 (n=1/27) oncogenic mutations as putative resistance drivers (PMID:37682528).
  • Acquired FGFR3 kinase-domain mutations at conserved residues (N540S, V553M, K650M, R669G); in silico modeling supports V553M as a gatekeeper allosterically blocking erdafitinib binding (PMID:37682528).

Genes & alterations

  • FGFR3 — oncogenic mutations (S249C, R248C, Y373C) and fusions (FGFR3-TACC3); primary actionable biomarker for erdafitinib. Acquired kinase-domain mutations N540S and V553M implicated as gatekeeper resistance alleles (PMID:37682528).
  • FGFR2 — only one actionable alteration in 1,507 tumors (an FGFR2:MARVELD3 fusion of unclear significance) (PMID:37682528).
  • TP53 — inversely associated with FGFR3 alterations; baseline co-mutation correlated with primary erdafitinib resistance; acquired TP53 mutations seen in 5/27 cfDNA on-treatment series (PMID:37682528).
  • AKT1 — one case acquired an activating AKT1 mutation on erdafitinib as a putative bypass resistance mechanism (PMID:37682528).
  • PIK3CA, TSC1 — PI3K-pathway co-alterations common in FGFR3-altered tumors but did not predict response to erdafitinib (PMID:37682528).
  • ERBB2 — inversely associated with FGFR3 alterations overall but co-altered in 11% of FGFR3-altered MIBC (PMID:37682528).
  • CDKN2A, CDKN2B, CDKN1A, KDM6A — frequently co-altered with FGFR3 (PMID:37682528).
  • RB1 — rare in FGFR3-altered tumors (1.2%, 5/414) and inversely associated with FGFR3 alterations (PMID:37682528).

Clinical implications

  • Archival primary tumor sequencing may misclassify FGFR3 status of metastatic disease in ~26% of patients; authors argue for metastatic-site or cfDNA profiling to guide erdafitinib selection (PMID:37682528).
  • Real-world erdafitinib activity is modest (PFS 2.8 mo) and tolerability-limited, supporting unmet need for FGFR3-isoform-specific inhibitors (e.g., TYRA300, LOXO435) (PMID:37682528).
  • Patients with FGFR3-altered tumors should still be offered immune checkpoint blockade as standard of care, since FGFR3 status did not alter ICB outcomes in this cohort (PMID:37682528).
  • TP53 co-alteration may flag higher risk of primary erdafitinib failure and motivate alternative-therapy selection, pending prospective validation (PMID:37682528).
  • cfDNA detection of FGFR3 V553M/N540S can serve as an early biomarker of evolving erdafitinib resistance (PMID:37682528).

Limitations & open questions

  • Retrospective single-institution cohort subject to sampling bias; erdafitinib-treated subgroup small (n=32) and enriched for adverse prognostic factors relative to BLC2001 (PMID:37682528).
  • cfDNA collection timepoints were inconsistent due to COVID-19-related visit reductions (PMID:37682528).
  • MSK-IMPACT panel does not capture transcriptomic resistance mechanisms (e.g., FGFR3 expression changes) or alterations in off-panel genes (PMID:37682528).
  • Functional validation of V553M and N540S as bona fide gatekeeper mutations, and of TP53-loss as a resistance driver, remains to be established (PMID:37682528).

Citations from this paper used in the wiki

  • “FGFR3 alterations predictive of erdafitinib sensitivity were identified in 39% (199/504) of patients with non-muscle invasive, 14% (75/526) with muscle-invasive, 43% (81/187) with localized upper tract, and 26% (59/228) with metastatic specimens.” (Abstract)
  • “Erdafitinib achieved a response rate of 40% but median progression-free and overall survival of only 2.8 and 6.6 months, respectively (n = 32).” (Abstract)
  • “Putative resistance mutations detected in cfDNA involved TP53 (n=5), AKT1 (n=1), and second-site FGFR3 mutations (n=2).” (Abstract)
  • “Among 27 FGFR3-altered cases with a primary tumor and metachronous metastasis, 7 paired specimens (26%) displayed discordant FGFR3 status.” (Abstract)
  • “in silico analysis predicted that V553M functions as a gatekeeper mutation that allosterically inhibits erdafitinib binding” (Results)

This page was processed by crosslinker on 2026-05-04.