CELL INTELLIGENCE · 肿瘤细胞治疗研究

Immune checkpoint inhibitors (ICIs) are standard components of metastatic urothelial carcinoma (mUC) treatment, but the association between FGFR3 alteration status and ICI outcomes remains uncertain.

The protocol was prospectively registered in PROSPERO (CRD420251114235). PubMed/MEDLINE, Scopus, and Embase were searched from inception through 22 November 2025; reference lists of included studies and relevant reviews were also examined. Eligible studies compared adults with FGFR3-altered versus FGFR3-wild-type mUC treated with PD-1, PD-L1, and/or CTLA-4 inhibitors. Random effects models pooled risk ratios (RRs) for objective response rate (ORR) and disease control rate (DCR), and hazard ratios (HRs) for progression-free survival (PFS) and overall survival (OS). Risk of bias was evaluated using RoB 2 and the Newcastle-Ottawa scale as appropriate. Leave-one-out analyses were performed for all endpoints; small-study effects were assessed for OS at the publication level.

13 publications met the eligibility criteria, and 12 contributed to at least one quantitative synthesis. The largest endpoint analysis, OS, included 13 cohort-level comparisons from 11 publications and 2042 patients. FGFR3 alterations were associated with lower DCR (RR 0.74, 95% CI 0.57-0.96; I 2 =12%), but not significantly different ORR (RR 0.82, 95% CI 0.63-1.07; I 2 =0%). OS (HR 1.25, 95% CI 1.08-1.44; I 2 =9%) and PFS (HR 1.63, 95% CI 1.23-2.16; I 2 =35%) were shorter in FGFR3-altered disease. In exploratory second-line-or-later analyses, DCR was lower (RR 0.59, 95% CI 0.41-0.83), whereas ORR remained inconclusive (RR 0.71, 95% CI 0.41-1.20). Leave-one-out analyses preserved the direction and statistical significance of OS and PFS, but the DCR result was sensitive to study omission. Egger regression did not indicate marked funnel-plot asymmetry for OS (p=0.81).

Amongst patients treated with ICIs, FGFR3 alterations were associated with less Disease Control and shorter time-to-event outcomes, whilst ORR was not significantly different. Because the evidence is predominantly retrospective or derived from exploratory biomarker subgroups and lacks a non-ICI comparator, these findings do not establish a predictive treatment interaction or causal immunotherapy resistance. They should be regarded as hypothesis-generating and support prospective biomarker-stratified validation, including trials evaluating FGFR inhibition with immunotherapy.