Introduction:

FGFR3 is the most altered gene in NMIBC, with incidence of alteration increased in lower grade/stage tumors. FGFR3 mutations are known to be directly associated with recurrence and inversely associated with progression. As Erdafitinib, an FGFR1-4 inhibitor, has recently been approved for locally advanced and metastatic bladder cancer, numerous trials that assess the role and efficacy of FGFR inhibitors in the treatment of non-muscle-invasive disease have commenced. We postulate that response to this class of targeted therapies will depend on the presence of co-alterations with FGFR3 in any given tumor. Therefore, this study seeks to characterize the mutational landscape of FGFR3-altered NMIBC and identify co-mutations that may result in an increased risk of recurrence, progression, and treatment resistance.

Methods:

From a prospectively maintained institutional database of 569 NMIBC samples of 503 individual patients that were sequenced with MSK-IMPACT (targeted exome sequencing with a 505-gene panel), we identified 218 samples from 206 patients that were FGFR3-altered. Clinicopathologic characteristics and outcomes, including high-grade recurrence-free survival (HG-RFS) and clinical progression-free survival (PFS), defined as stage ≥T1, progression to radical cystectomy, or development of metastasis, were assessed and compared via the CBio Cancer Genomics Portal, using a Kaplan Meier survival analysis. Comutational frequency was assessed via a chi-squared analysis; the Benjamini Hochberg procedure was used to determine false discovery (q), and significant p-values were only reported when q<0.05.

Results:

Of 218 FGFR3-altered tumors, 77% were high-grade. 68% of these tumors were staged Ta, 31% were T1, and 1% Tis; 90.3% of these tumors had no evidence of concurrent CIS. We found 6 significant co-alterations that were enriched in FGFR3-mutated tumors (CDKN2A[p14/p16], CDKN2B, KDM6A, STAG2, CREBBP, TERT), and 9 that were mutually exclusive (ERBB2, TP53, RB1, KRAS, RAF1, E2F3, HRAS, ATR, SDHA). Among these co-alteration patterns, we found that high-grade RFS was worse for FGFR3-altered tumors when CDKN2A was also altered (Figure 1).  When analyzing other cell cycle genes, we found that co-alteration with CDKN1A was associated worsened clinical-PFS, including when restricting our analysis to patients with high-grade NMIBC treated with BCG (Figure 2). Although TP53 was mutually exclusive with FGFR3, in the 32 (14.67%) tumors where FGFR3/TP53 were co-altered, TP53 was associated with worse RFS and clinical PFS, even when restricting to BCG-treated patients.

Conclusion:

Trials of FGFR-inhibitors are ongoing in NMIBC. We predict that the diversity of response to these drugs in FGFR3-altered NMIBC will be dependent on co-mutational patterns, and identified three co-alterations (CDKN1A, CDKN2A, TP53) that affect the high-grade recurrence-free survival and clinical progression-free survival of patients with FGFR3-altered tumors. Future studies regarding the effect of these and other co-alterations on recurrence, progression, metastasis, and drug resistance in FGFR3-altered tumors are warranted and will provide more insight into the true efficacy of this new class of targeted therapies. 

Funding: This work was supported by the Sidney Kimmel Center for Prostate and Urologic Cancers, NCI Cancer Center Core Grant Number P30-CA008748, NIH/NCI Grant Number R37CA276946, Cycle for Survival, the Bladder Cancer Advocacy Network YIA, and the Marie-Josée and Henry R. Kravis Center for Molecular Oncology.