Pemigatinib in previously treated solid tumors with activating FGFR1-FGFR3 alterations: phase 2 FIGHT-207 basket trial

FGFR genes (FGFR1–FGFR3) harbor pathogenic alterations across diverse cancers and can drive oncogenesis by altering signaling that regulates migration, proliferation and survival. While selective FGFR inhibitors are approved in specific contexts (for example, erdafitinib in urothelial cancer; pemigatinib and futibatinib in cholangiocarcinoma; pemigatinib in myeloid/lymphoid neoplasms with FGFR1 rearrangements), many FGFR-altered tumor types and alteration classes remain clinically unvalidated for FGFR inhibition. Early-phase studies, including the first-in-human FIGHT-101 study of pemigatinib, suggested sensitivity of FGFR1–FGFR3 fusions and point mutations across histologies, and additional rare alterations (for example, in-frame insertions or truncating deletions) have been proposed as oncogenic but not yet established as actionable. Key clinical questions include: which FGFR alteration classes and tumor histologies are sensitive to FGFR inhibition; how co-alterations influence response; and what mechanisms of resistance emerge across histologies. The phase 2 FIGHT-207 basket trial was designed to evaluate the antitumor activity and safety of pemigatinib in previously treated, unresectable/metastatic solid tumors with FGFR1–FGFR3 alterations and to investigate genomic correlates of response and resistance using tissue and circulating tumor DNA.

The paper synthesizes prior evidence that FGFR1–FGFR3 alterations occur in up to ~7% of cancers and have been successfully targeted in select diseases. Approvals include erdafitinib for FGFR2/3-altered urothelial carcinoma, pemigatinib and futibatinib for FGFR2 fusion/rearranged cholangiocarcinoma, and pemigatinib for FGFR1-rearranged MLNs. Early trials (for example, FIGHT-101, futibatinib phase 1, RAGNAR phase 2 with erdafitinib) reported responses across multiple histologies with various FGFR alterations. Preclinical work indicates certain kinase domain mutations reduce sensitivity to ATP-competitive FGFR inhibitors, while rare FGFR2 extracellular domain in-frame deletions and truncating exon 18 variants may be pathogenic and targetable. Prior clinicogenomic analyses (for example, FIGHT-202) linked TP53 co-mutations with lack of response and documented acquired on-target resistance via FGFR gatekeeper and molecular brake mutations, primarily in cholangiocarcinoma and urothelial cancer.

Design: Open-label, single-arm, multicenter phase 2 basket trial (FIGHT-207; NCT03822117; EudraCT 2018-004768-69) evaluating pemigatinib in adults with previously treated, unresectable or metastatic solid tumors harboring FGFR1–FGFR3 alterations. Patients were enrolled between 10/17/2019 and 07/12/2021 at 48 sites in 10 countries; study completion 03/29/2022. Cohorts: A) FGFR1–FGFR3 fusions/rearrangements with intact kinase domains (n=49); B) actionable non-kinase domain FGFR1–FGFR3 SNVs considered activating/actionable based on public databases (OncoKB, ClinVar, OMIM) and prior clinical data (n=32); C) FGFR1–FGFR3 kinase domain mutations or FGFR1–FGFR3 variants of unknown significance with potential pathogenicity (n=26). Genomic eligibility was initially based on local testing, with central confirmation (primarily FoundationOne CDx). A small number of patients initially categorized as VUS were re-assigned for translational analyses following central review. Eligibility: ≥18 years; histologically/cytologically confirmed advanced/metastatic or unresectable solid tumor; measurable disease per RECIST v1.1 or RANO; documented FGFR1–FGFR3 mutation or fusion/rearrangement; progression after ≥1 prior systemic therapy; ECOG PS ≤2; baseline tumor specimen available; no available therapy likely to provide clinical benefit; adequate organ function; contraception requirements. Key exclusions included prior selective FGFR inhibitor therapy; active/untreated or progressing CNS metastases; significant ocular disorders (corneal/retinal), ectopic mineralization; significant cardiac disease or abnormal ECG; active infections (HBV, HCV, HIV); prohibited concomitant CYP3A4 modulators; pregnancy/breastfeeding; and protocol-defined abnormal laboratory values. Treatment: Pemigatinib 13.5 mg orally once daily continuously in 21-day cycles until radiographic progression, unacceptable toxicity, withdrawal, or physician decision. Assessments and Endpoints: Imaging (CT/MRI) at baseline, every 3 cycles, and end of treatment. Primary endpoints: IRC-confirmed objective response rate (ORR) in cohorts A and B per RECIST v1.1 or RANO. Secondary endpoints: duration of response (DOR), IRC-assessed progression-free survival (PFS), overall survival (OS), safety/tolerability (TEAEs graded by NCI CTCAE v5.0) in cohorts A and B. Exploratory endpoints: ORR/DOR/PFS/OS in cohort C; translational analyses of baseline and progression tissue and plasma ctDNA to identify co-alterations associated with response/resistance and acquired resistance mechanisms. Post hoc: clinical benefit rate (CBR: CR/PR or SD ≥6 months) in all cohorts. Genomic platforms: Central tissue NGS (FoundationOne CDx; 324 genes, >500× coverage). Plasma ctDNA NGS (PredicineCARE; 152 genes, ~20,000×) at baseline and progression, when available. Variant interpretation incorporated Foundation Medicine database and COSMIC. Concordance between tissue and ctDNA for FGFR alterations/co-alterations was limited; analyses used combined detection (either tissue or ctDNA) where applicable. Statistics: ORR with 95% CIs (Clopper–Pearson). Time-to-event endpoints (PFS, DOR, OS) via Kaplan–Meier; 95% CIs for medians via Brookmeyer–Crowley with log–log transformation. Associations of co-alterations with ORR/CBR via two-sided Fisher’s exact test; with PFS via log-rank test. Planned sample sizes: cohorts A (n≈60) and B (n≈90) powered to reject ORR ≤15% (one-sided α=0.025) for assumed ORRs of 35% and 30%, respectively; cohort C planned n=20 to have ≥80% chance of ≥4 responders if true ORR=30%. Sponsor terminated the study early for business reasons, reducing enrollment in A and B. Populations: Efficacy population included 107 patients with centrally confirmed FGFR alterations receiving ≥1 dose. Safety population included 111 treated patients (includes 4 without central confirmation). Baseline ctDNA available for 89; paired baseline and progression ctDNA for 73.

- Enrollment and cohorts: 111 treated; 107 efficacy-evaluable divided into cohort A (fusions/rearrangements, n=49), cohort B (actionable non-kinase domain SNVs, n=32), cohort C (kinase domain mutations or potentially pathogenic VUS, n=26). - Primary efficacy (IRC-confirmed ORR): A: 26.5% (13/49; 95% CI 15.0–41.1), including 1 CR; B: 9.4% (3/32; 95% CI 2.0–25.0). Exploratory cohort C ORR: 3.8% (1/26; 95% CI 0.1–19.6). - Secondary efficacy: Median DOR: A 7.8 mo (95% CI 4.2, NE), B 6.9 mo (95% CI 4.0, NE), C 6.2 mo (not estimable CI due to 1 responder). Median PFS: A 4.5 mo (95% CI 3.6–6.3), B 3.7 mo (95% CI 2.1–4.5), C 2.0 mo (95% CI 1.8–3.7). Median OS: A 17.5 mo (95% CI 7.8, NE), B 11.4 mo (95% CI 6.6, NE), C 11.0 mo (95% CI 3.9, NE). CBR: A 28.6% (95% CI 16.6–43.3), B 21.9% (95% CI 9.3–40.0), C 15.4% (95% CI 4.4–34.9). - Tumor types with responses: Objective responses observed in CNS tumors, pancreatic cancers (notably KRAS wild-type cases), cervical cancers, urothelial carcinoma, and cholangiocarcinoma, including contexts without approved FGFR inhibitors or with alteration classes not previously validated (for example, FGFR2 SNVs C382R and extracellular in-frame deletions in cholangiocarcinoma). - Safety (N=111): Grade ≥3 TEAEs in 68%; fatal TEAEs in 6 patients (5.4%), none deemed related. TEAEs leading to interruption 71%, dose reduction 43%, discontinuation 7.2%. Most common any-grade TEAEs: hyperphosphatemia 84% and stomatitis 53%. Nail toxicities 45%, serous retinal detachment 14%. Safety consistent with prior pemigatinib experience. - Genomic correlates of response (baseline co-alterations): Across cohorts A and B, TP53 alterations in tumor tissue associated with lack of response (0/27 objective responses) and shorter PFS; ARID1A and MAPK pathway alterations associated with shorter PFS; BAP1 alterations associated with higher ORR/CBR. Concordance between tissue and ctDNA for FGFR/co-alterations was limited; associations for ARID1A, MAPK, and BAP1 held in ctDNA-only analysis, whereas TP53 association did not. - Acquired resistance (paired ctDNA; n=73): 19% (14/73) acquired secondary FGFR kinase domain mutations at progression, predominantly gatekeeper (FGFR2 V564F/I/L; FGFR3 V555L/M; FGFR1 V559L/M) and molecular brake residues (FGFR1 N546K; FGFR2 N549D/H/K, E565A, K641R). Polyclonal resistance common (71% with multiple concurrent FGFR mutations). Additional emergent co-alterations implicated bypass resistance (PIK3CA, RAS family). On-target resistance documented across histologies, including first report in FGFR1-altered pancreatic cancer (FGFR1-PDE4DIP fusion) with emergent FGFR1 V559L/M and N546K. - Amplifications: Concurrent FGFR gene amplifications observed in a minority; numbers insufficient to assess impact on response.

The trial addressed whether pemigatinib provides antitumor activity across multiple histologies with diverse FGFR1–FGFR3 alterations and identified genomic determinants of benefit and resistance. Pemigatinib demonstrated clinically meaningful activity beyond approved indications (cholangiocarcinoma and urothelial carcinoma), including CNS, pancreatic (KRAS wild type), and cervical cancers, supporting broader tumor-agnostic development for select FGFR alterations. Dedicated evaluation of actionable non-kinase domain FGFR SNVs confirmed responsiveness of specific FGFR2 mutations (for example, C382R and extracellular in-frame deletions) in cholangiocarcinoma, expanding the actionable alteration spectrum. De novo kinase domain mutations showed low response rates, consistent with their role as resistance mutations, though isolated cases of benefit (FGFR1 K656E, N546K) were noted. Co-alteration analyses suggest TP53 and other tumor suppressor alterations and MAPK pathway activation correlate with reduced benefit (shorter PFS, lack of ORR), while BAP1 correlates with higher response, potentially indicating cooperative oncogenic contexts (for example, FGFR2+BAP1 in intrahepatic cholangiocarcinoma). Serial ctDNA analyses generalized known resistance mechanisms (gatekeeper and molecular brake mutations; polyclonality) across FGFR1–FGFR3 and histologies, and revealed emergent MAPK/PI3K pathway variants as potential bypass mechanisms. These findings inform patient selection, support routine comprehensive molecular profiling, and motivate development of next-generation FGFR inhibitors capable of overcoming on-target resistance.

Pemigatinib showed antitumor activity and manageable safety across multiple FGFR1–FGFR3-altered solid tumors in the phase 2 FIGHT-207 basket trial, including tumors and alteration classes previously unvalidated for FGFR inhibition. Objective responses were most frequent with FGFR fusions/rearrangements and occurred in histologies beyond current approvals; selected FGFR2 SNVs (C382R, extracellular in-frame deletions) in cholangiocarcinoma were responsive. Genomic analyses identified TP53 and MAPK pathway co-alterations as correlates of reduced benefit and BAP1 alterations as correlates of higher response, and established gatekeeper/molecular brake mutations as common acquired resistance mechanisms across FGFR1–FGFR3. These data support expanding clinical development of FGFR inhibitors for additional alteration–histology combinations, refining eligibility to include high-sensitivity SNVs, and advancing next-generation inhibitors and rational combinations to prevent or overcome resistance. Future research should prospectively validate co-alteration-based predictive markers, explore ctDNA-guided management, and test strategies addressing polyclonal on-target and pathway-bypass resistance.

- Basket design with heterogeneous histologies and alteration classes yielded small subgroup sizes; early study termination for business reasons underpowered cohorts A and B to definitively assess FGFR dependency by histology/alteration. - Heavily pretreated population may harbor more co-alterations impacting response. - Analyses of co-alterations and outcomes are exploratory and may be prognostic rather than predictive; require prospective validation. - Limited concordance between tissue and ctDNA for FGFR/co-alterations due to technical and biological factors; integrated analysis mitigated but does not eliminate bias. - Numbers with concurrent FGFR amplification were too small to assess impact on efficacy.