Abstract 89: Genomic and clinical characteristics associated with metastatic patterns in urothelial carcinoma patients

Abstract Introduction: To identify the patients with urothelial cancer (UC) at greatest risk of metastasis, we performed an integrated analysis of clinicopathologic and somatic mutation data from a prospectively generated tumor genomic sequencing cohort. Methods: We collected detailed clinicopathologic data from 1, 617 patients with bladder and upper tract UC treated at Memorial Sloan Kettering. Tumors were sequenced with MSK-IMPACT, a targeted DNA sequencing assay that identifies genomic alterations in 341-505 genes. Metastatic site involvement was inferred using a natural language processing model, as reported in the MSK-CHORD dataset. Patients were considered never metastatic if there was no evidence of metastases after 2.5 years of follow-up from diagnosis. Comparisons were made between ever-metastatic (EM) and never-metastatic (NM) patients. In a time to metastasis analysis, metastatic patients at diagnosis and patients that had metastasis within 6 months of diagnosis were excluded. Continuous variables were analyzed using Wilcoxon rank sum test and two-sided Fisher’s exact test for categorical variables. Univariable Cox proportional hazard models were used to assess time-to-metastasis. Benjamini-Hochberg was used to correct for multiple hypothesis testing and q-values <0.1 were considered significant. Results: EM patients were more likely to be former smokers (64% vs 50%, q<0.001) and be diagnosed after age 80 (17% vs 10%, q=0.033). NM patients were more likely to be never smokers (48% vs 34%, q<0.001) and diagnosed before age 65 (36% vs 28%, q=0.016). EM tumors had a higher fraction genome altered (0.20 vs 0.10, q<0.001) and higher rates of TP53 (55% vs 44%, q=0.003), ELF3 (15% vs 9%, q=0.018), and CDKN2A (29% vs 21%, q=0.069) alterations. NM tumors were more likely to have FGFR3 (18% vs 26%, q=0.002) and ERCC2 (6% vs 13%, q<0.001) alterations. In a time to metastasis analysis: former smoking (HR: 1.44, q=0.041), CDKN2A (HR: 1.59, q=0.005), and ELF3 (HR: 1.64, q=0.030) alterations were associated with a higher risk of bone metastasis. CDKN2A (HR: 1.98, q<0.001) and NRF2 pathway alterations (HR: 2.13 q=0.026) were associated with higher risk of lung metastasis. Former smoking (HR: 1.41, q=0.029), CDKN2A alterations (HR: 1.33, q=0.086), ERBB2 amplifications (HR: 1.81, q=0.009), NRF2 (HR: 1.92 q=0.026) and HIPPO (HR: 1.62 q=0.031) pathway alterations were associated with higher risk of lymph node metastasis. ERCC2 alterations were associated with lower risk of bone, (HR: 0.33, q=0.017), lung (HR: 0.48, q=0.098), and lymph node metastases (HR: 0.43, q=0.026). Conclusion: Through integration of tumor genomic profiling data with NLP-derived clinical annotation, we identified clinical and genomic features associated with a higher risk of developing metastatic disease in patients with UC, as well as alterations associated with organ-specific metastatic tropism. Citation Format: Jordan E. Eichholz, Henry Walch, Jacob Tallman, Syed Alam, James Rodrigues, Kwanghee Kim, Christopher Fong, Ecenur Turkay, Alejandra Lopez Rojas, Gopa Iyer, Eugene Pietzak, Jonathan E. Rosenberg, Hikmat Al-Ahmadie, Nikolaus Schultz, Walid K. Chatila, David B. Solit. Genomic and clinical characteristics associated with metastatic patterns in urothelial carcinoma patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 89.