Opportunities for precision oncology: Real-world patterns and disparities in FGFR3 testing among US patients with advanced urothelial cancer.

682 Background: Activating fibroblast receptor growth factor-3 (FGFR3) genetic alterations occur in 15-20% of LA/mUC and biomarker directed therapy should be considered for patients with ≥1 prior lines of treatment. Genomic testing for FGFR3 at LA/mUC diagnosis is recommended; yet, real-world adherence to these recommendations and potential disparities in testing are poorly understood. Methods: The Flatiron Health Research Database (FHRD) was analyzed to identify adults diagnosed with LA/mUC between 1/2022 - 3/2025. FGFR3 testing was identified from structured genomic reports or physician documentation. Objectives included: 1) prevalence and timing of FGFR3 testing; 2) how often testing is repeated; and 3) clinical and demographic predictors of testing. Multivariable logistic regression (MLR) evaluated associations between testing and patient characteristics including age, gender, race, ECOG PS, and socioeconomic status; disease specific characteristics including primary tumor location, stage at initial diagnosis and maximum line of therapy (LOT) received; and practice type. Patients with FGFR3 alterations (excluding amplifications and variants of unknown significance) and 22 erdafitinib use were reported. Results: 3,661 patients with LA/mUC were identified. Overall, 41% (n=1495) of patients had ≥1 FGFR3 genomic test, among which 21% (n=319) had ≥2 tests. Of the 2,351 patients who received 1L, 28% (n=648) were tested prior to 1L with 75% (n=483/648) having results prior to initiating therapy. In MLR, significantly lower odds of testing were observed with increasing age (OR, 95% CI; 0.98, 0.98-0.99), Black race (0.66, 0.47-0.92), lower tract primary (0.60, 0.45-0.78), and initial diagnosis of non-metastatic disease (locally advanced 0.65, 0.54-0.78 or muscle invasive 0.67, 0.56-0.79). Patients were less likely to be tested if treated in an academic setting (FHRD-defined). There was a greater likelihood of testing with each subsequent LOT (Table 1). Among tested patients with known results, 19% (n=274) had FGFR3 alterations. Among FGFR3+ patients with 2L+ (n=110), 46% received erdafitinib. Conclusions: FGFR3 testing in LA/mUC is often delayed or omitted with <50% of patients tested. Less than 30% were tested in the guideline-concordant window before 1L, and result availability was timely. Significant differences by age, race, disease stage, and primary site highlight gaps in precision oncology delivery for patients with LA/mUC. Interventions to promote timely and equitable testing are essential to expand access to targeted therapies and improve patient outcomes. Association between maximum LOT and FGFR3 testing. Maximum LOT n Odds Ratio (95% CI) 1L 1558 1.86 (1.57-2.21)* 2L 514 3.10 (2.48-3.89)* 3L 164 5.70 (3.93-8.36)* 4L+ 115 7.49 (4.80-11.99)* Referent group: No LOT (n=1310), *p-value<0.001.