Citations

e17005 Background: We previously developed a hybrid framework combining LLMs and rule-based algorithms (RBA) for automated risk stratification in localized prostate cancer (locPCa), achieving 89% accuracy. However, clinical T-stage (cTS) inference from narrative MRI reports remained the primary source of errors. Herein, we evaluated whether ensemble approaches using multiple state-of-the-art LLMs could improve cTS inference and overall NCCN risk stratification accuracy. Methods: This study included patients with locPCa presenting with at least one positive prostate biopsy and MRI report available. Building on our previously validated hybrid framework that extracts structured phenotypic variables (PSA, prostate volume, cTS, Gleason patterns, grade group, positive and examined cores) from unstructured reports using LLMs and applies NCCN-concordant RBA for risk stratification, we focused on improving cTS inference. Four HIPAA-compliant LLMs (GPT-4.1, GPT-5.1, Gemini-2.5-Pro, Gemini-2.5-Flash) were evaluated using structured zero-shot prompts iteratively refined through oncologist feedback. Majority voting was employed to generate consensus cTS predictions. Performance was evaluated using weighted accuracy. Results: A total of 358 patients were included. The median age at diagnosis was 65 years (IQR: 60-69); the majority were White (92%) and non-Hispanic (94%). cTS distribution was T2a (51%), T2c (20%), T3a (16%), T3b (7%), T2b (4%), and T4 (2%). The most prevalent risk category was unfavorable intermediate (38%) followed by favorable intermediate (20%), high risk (19%), very high risk (16%), low (5%), and very low (1%). Individual LLM accuracy for cTS inference ranged from 85-90%, with Gemini-2.5-Pro achieving highest accuracy (90%), followed by GPT-4.1 (88%), GPT-5.1 (87%), and Gemini-2.5-Flash (85%). The four-model ensemble reached consensus in 92% of cases, achieving 93% cTS inference accuracy. When integrated into NCCN risk stratification schema, single-model approaches achieved 89-92% accuracy, while ensemble consensus cases achieved 95% accuracy. Conclusions: The ensemble framework using multiple LLMs significantly improves cTS inference and achieves 95% NCCN risk stratification accuracy in consensus cases (92% of patients), representing a 6-percentage-point improvement over our previous single-model framework (89%). This strategy demonstrates that ensemble methods can overcome individual model limitations for complex clinical tasks and holds promise for reliable AI-assisted risk stratification.

e17003 Background: Accurate ascertainment of prostate cancer (PCa) disease states—biochemical recurrence (BCR), castration-sensitive prostate cancer (CSPC), and castration-resistant prostate cancer (CRPC)—is critical for automated clinical trial enrollment, quality improvement initiatives, and real-world evidence generation. Essential information for disease state ascertainment resides in unstructured EHR data, requiring labor-intensive manual abstraction that is inconsistent and not scalable. We developed an automated hybrid AI framework using large language models to identify PCa disease states from real-world EHR data. Methods: This retrospective study included patients with histopathologically confirmed PCa (2017-2022) and ≥3 years follow-up. Sequential EHR data—including PSA and testosterone levels, clinical notes, pathology and radiology reports—were retrieved across temporal points. BCR was identified using rule-based algorithms (post-radical prostatectomy [RP]: PSA ≥0.2 ng/mL ×2; post-radiation therapy [RT]: nadir + 2 ng/mL ×2). CRPC was determined using androgen deprivation therapy (ADT) timing, PSA kinetics, testosterone < 50 ng/dL, and radiographic progression per PCWG/EAU-ASCO criteria. GPT-4o with structured parametrized prompts extracted metastatic status and radiographic progression. Independent clinicians performed manual annotation for validation. The framework was developed on 10% of data with evaluation on 90%. External validation used a held-out cohort of clinical trial patients. Results: Among 200 PCa patients (median age 66 years [IQR 60-71]; 91% White, 93% non-Hispanic), the framework demonstrated high accuracy: PCa diagnosis (92%), RP dates (94%), RT dates (93%), and ADT initiation (92%). For disease state identification, BCR detection achieved 90% accuracy, castration status 93%, and metastatic state (M0/M1) 87%. External validation on trial-enrolled patients confirmed robust performance: 26/31 (84%) CRPC trial patients and 26/27 (96%) CSPC trial patients were correctly classified, validating the framework’s reliability for clinical trial matching. Conclusions: This hybrid LLM-based framework accurately identifies PCa disease states using real-world EHR data with minimal manual intervention. The system offers a scalable, reproducible approach for automated disease state identification, enabling streamlined clinical trial matching and enhanced real-world data curation in oncology.

4589 Background: Recent years have seen a shift in aUC management following approval of enfortumab vedotin–pembrolizumab (EV-P) and gemcitabine/cisplatin plus nivolumab (GC-NIVO) as first-line therapies. Despite improved clinical outcomes, their relative economic value remains uncertain. Herein, we conducted a cost-effectiveness analysis of current 1L regimens in aUC from a U.S. payer perspective using 2026 pricing. Methods: A partitioned survival model with three health states (progression-free [PF], progressed disease [PD], and death) was developed over a 10-year horizon from a U.S. payer perspective. Analyses focused on cisplatin-eligible patients. Survival data were reconstructed from published Kaplan–Meier curves and extrapolated using parametric survival models. Treatment dosing, schedules, and duration were modeled according to trial protocols. Patients receiving gemcitabine/cisplatin without progression were modeled to receive avelumab maintenance (GC→AVEL). Subsequent-line therapy costs were applied at incident progression, and grade ≥3 adverse event costs and disutilities were applied as one-time events. Drug acquisition costs (2026 USD) were based on average sales prices from the Centers for Medicare & Medicaid Services. Costs and quality-adjusted life years (QALYs) were discounted at 3% annually. Incremental cost-effectiveness ratios (ICERs) and net monetary benefit (NMB) were evaluated at willingness-to-pay (WTP) thresholds of $100,000, $150,000, and $200,000 per QALY. Parameter uncertainty was characterized using probabilistic sensitivity analysis. Results: In the base-case analysis, total discounted costs and QALYs were $90,269 and 1.62 for GC→AVEL, $131,836 and 2.09 for GC-NIVO, and $1,451,885 and 2.77 for EV-P. Compared with GC→AVEL, GC-NIVO gained 0.46 QALYs at an incremental cost of $41,567, resulting in an ICER of $89,501 per QALY and positive NMB across all evaluated WTP thresholds. Compared with GC→AVEL, EV-P gained 1.15 QALYs at an incremental cost of $1.36 million (ICER $1.19 million per QALY). Compared with GC-NIVO, EV-P gained 0.68 QALYs at an incremental cost of $1.32 million (ICER $1.93 million per QALY), exceeding all evaluated WTP thresholds. In a scenario analysis limiting EV-P to 10 treatment cycles, total costs decreased to $287,360, with ICERs of $172,000/QALY versus GC→AVEL and $225,000/QALY versus GC-NIVO, remaining above the $150,000/QALY threshold. Conclusions: Using updated 2026 pricing, GC-NIVO is likely to be the most cost-effective first-line regimen for cisplatin-eligible patients with aUC. EV-P offers the greatest clinical benefit but is not cost-effective at current prices. A major limitation of this study is cross-trial variability in study populations and design, limiting the validity of head-to-head economic comparisons.

4514 Background: Maintenance avelumab is a standard of care in patients (pts) with la/mUC who do not progress after first-line PBC. Cabozantinib (CABO) is an oral inhibitor of MET, VEGFR and TAM, active in multiple solid tumors. We hypothesized that an avelumab (Av)-CABO combination would improve outcomes in pts with mUC with an acceptable safety profile compared to avelumab maintenance alone. Methods: MAIN-CAV (NCT05092958) is a phase III randomized, multicenter, international trial for la/mUC pts who did not progress after 4-6 cycles of any PBC (gemcitabine-cisplatin, gemcitabine-carboplatin, MVAC, ddMVAC). Pts were randomized 1:1 to Av 800 mg IV every 2 weeks alone or with CABO 40 mg daily for ≤2 years. Stratification factors were response to first-line PBC (complete response [CR] vs partial response [PR] vs stable disease [SD]) and visceral metastases. The primary endpoint was overall survival (OS); secondary endpoints included progression-free survival (PFS), safety and quality of life. The target accrual was 654 pts; the trial closed early after enrollment of 68 pts following adoption of enfortumab vedotin plus pembrolizumab as the new preferred first-line standard in the US. Results: From 3/2022 to 3/2024, 68 pts were randomized (Av n = 33; CABO-Av n = 35). Median OS was 24.8 months (mo) with Av and 23.0 mo with Avb-CABO (Hazard ratio 1.41; 95% CI 0.66–3.01; p = 0.186). Median PFS was 4.6 mo with Av vs 6.0 mo with Av-CABO (p = 0.343). Objective response rate among pts with PR or SD after first-line PBC was 20.8% with Av vs 11.5% with Av-CABO (p = 0.424). Grade ≥3 treatment-related adverse events occurred in 24.2% pts on the Av arm vs 39.4% of pts on the Av-CABO arm and CABO dose reductions occurred in 63% of pts. Conclusions: Addition of CABO to Av did not improve OS or PFS compared with Av alone in this underpowered, prematurely closed phase III trial; safety was consistent with known profiles. U10CA180821,U10CA180882; U24CA196171, U10CA180863. (CCTG);U10CA180888 (SWOG); https://acknowledgments.alliancefound.org. Clinical trial information: NCT05092958 .

4562 Background: Approximately 50-80% of patients (pts) with la/mUC have tumors with HER2 protein expression. As HER2-directed therapies enter development, a robust understanding of HER2 prevalence and real-world outcomes with standard-of-care (SOC) therapies across HER2 expression subgroups is needed. SGNDV-003 (NCT05902494) is a global retrospective chart review study in which HER2 status was assessed via central lab testing and outcomes were analyzed by HER2 status in pts with la/mUC. Methods: SGNDV-003, conducted 6/2023-7/2025, included pts with la/mUC who received ≥1 prior treatment (tx) (incl. 1 line of platinum-based chemotherapy [chemo]). HER2 status was centrally tested using a UC-specific investigational immunohistochemistry [IHC] and in situ hybridization (ISH) assays. HER2 status was defined as: HER2-positive (IHC3+ or IHC2+/ISH+), HER2-low (IHC2+/ISH- or IHC1+), HER2-expressing (IHC3+, 2+, 1+), and HER2-zero (IHC0). Index line of therapy (LOT) was defined as the first LOT received after platinum-based chemo. Outcomes included real-world objective response rate (rwORR) (per RECIST 1.1 or physician-assessed response) to index LOT, real-world progression-free survival (rwPFS), and overall survival (OS) from start of index LOT. Data was analyzed via descriptive statistics. Results: A total of 353 pts with la/mUC and available HER2 status were analyzed (male, 76%; median age, 70 years). 70% of pts were HER2 expressing (19% HER2-positive, 51% HER2-low); 30% were HER2-zero. Disease characteristics were well balanced by HER2 status. Index LOT tx and real-world outcomes are shown in the Table. Immuno-oncology therapies (IO) and enfortumab vedotin (EV) were the most common index LOTs. For IO, rwORRs were 21% for HER2-expressing and 17% for HER2-zero. For EV mono, rwORRs were 36% (HER2-expressing) and 38% (HER2-zero), respectively. Conclusions: This global study provides a robust estimate of HER2 distribution in la/mUC based on standardized central lab testing, showing high prevalence of HER2 expression. Real-world outcomes with current SOC therapies were broadly consistent across HER2 subgroups, establishing an important benchmark for the clinical development of HER2-directed therapies in this setting. HER2-positive (n=67) HER2-low (n=181) HER2-expressing (n=248) HER2-zero (n=105) Total (n=353) Real-world Outcomes rwORR, n (%)[95% CI] 17 (25.4)[15.5, 37.5] 50 (27.6)[21.3, 34.7] 67 (27.0)[21.6, 33.0] 22 (21.0)[13.6, 30.0] 89 (25.2)[20.8, 30.1] Median rwPFS (95% CI), mo 2.8(2.0, 4.1) 3.7(2.7, 5.1) 3.0(2.7, 4.5) 2.6(2.1, 3.4) 2.9 (2.6, 3.5) Median OS(95% CI), mo 12.2(9.3, 16.5) 11.5(9.5, 13.5) 11.9(10.3, 13.4) 7.7(5.9, 9.9) 10.8(9.1, 12.3) Index LOT, n% IO EV monoChemoOther tx 37 (55)17 (25)10 (15)3 (5) 94 (52)51 (28)24 (13)14 (8) 131 (53)68 (27)34 (14)17 (7) 65 (62)21 (20)12 (11)7 (7) 196 (56)89 (25)46 (13)24 (7)

Abstract Purpose: This multicenter, single-arm, phase II trial evaluated the safety and efficacy of cabazitaxel and carboplatin followed by abiraterone, plus androgen deprivation therapy (ADT), in patients with high-volume metastatic castration-sensitive prostate cancer (mCSPC). Patients and Methods: Eligible patients had high-volume mCSPC and ≤3 months of prior ADT. Patients received six 21-day cycles of cabazitaxel (20 mg/m2) and carboplatin (AUC 4) with continuous ADT, followed by maintenance abiraterone (1,000 mg daily) and prednisone (5 mg daily). The primary endpoint was the proportion of patients free of prostate-specific antigen (PSA) or radiographic progression at 12 months. Secondary endpoints included complete PSA response (≤0.2 ng/mL), objective response, progression-free survival (PFS), and safety. Overall survival (OS) was also reported. Results: Sixty-one participants were enrolled at eight sites. The median age was 64 years, median baseline PSA was 6.6 ng/mL (0.07–745.7 ng/mL), 73.3% had Gleason grade group 5, 69.5% had 10 or more metastases, and 19.6% had a homologous recombination repair (HRR) mutation. PSA PFS at 12 months was 84.6% [95% confidence interval (CI), 72.5%–91.7%], and OS at 12 months was 94.8% (84.7%–98.3%). Complete PSA response was 66.7%, and complete objective response was 30.7%. Contrary to our hypothesis, relative to HRR wild-type patients, HRR-mutated patients (19.6%) had numerically fewer complete PSA responses (33.3% vs. 70.3%) and inferior PFS (HR, 2.43; 95% CI, 0.93–6.39). Common side effects of this quadruplet regimen included fatigue, nausea, and diarrhea. Conclusions: Cabazitaxel and carboplatin followed by abiraterone, together with ADT, were feasible, safe, and efficacious in patients with high-volume mCSPC and warrant further study in larger randomized trials.

703 Background: The association of circulating tumor DNA (ctDNA) reduction with treatment response has been investigated across multiple malignancies. This study aims to assess the impact of ctDNA response on survival outcomes in patients receiving enfortumab-pembrolizumab (EV-P). Methods: We identified 32 patients with advanced urothelial carcinoma who received EV-P and had ≥2 ctDNA assessments with ≥1 detectable level. Deep molecular response was defined as >2-log reduction from highest ctDNA detection. Radiographic progression was determined from radiology reports documenting progression by RECIST criteria. Kaplan Meier analyses were performed to compute Hazard Ratios (HR) with 95% confidence intervals (CI) for radiographic progression free survival (rPFS), defined as time from treatment start to radiographic progression or death and overall survival (OS) defined as time from treatment start to death were computed. Log-rank tests were performed to compare PFS and OS in patients with deep molecular response to those without response. A p-value <0.05 indicated a statistically significant association with response. Results: Thirty-two patients were included (median age 71 years; range 39–90). Median follow-up was 10.6 months. At cutoff (10/1/2025), 26 patients were alive and 6 had died. Seventeen (53%) achieved a ≥2-log ctDNA reduction (responders), while 15 (47%) did not (non-responders). Progression-free survival was significantly longer in responders (log-rank P < 0.001); median PFS was 5.9 months (95% CI 2.4–NR) in non-responders and not reached in responders (HR 0.09, 95% CI 0.02–0.44; P = 0.0026). Median overall survival was 12.7 months (95% CI 10.0–NR) in non-responders versus not reached in responders (log-rank P = 0.011; HR 0.11, 95% CI 0.01–0.88; P = 0.037). At cutoff, 94% of responders and 67% of non-responders were alive. Conclusions: Achieving a >2-log decrease in ctDNA during EV + Pembrolizumab therapy was significantly associated with improved progression free and overall survival. These findings support ctDNA kinetics as a potential early biomarker of response warranting prospective validation. Clinical outcomes by >2-log ctDNA reduction status. Endpoint > 2-log drop (n = 17 ≤ 2-log drop (n = 15) Log-rank p Median (mo) ≤ 2-log > 2-log HR (95% CI) P C-index PFS 2 events (12%) 10 events (67%) 0.0002 5.9 (2.4–NA) NR 0.09 (0.02–0.44) 0.0026 0.77 OS 1 event (6%) 6 events (40%) 0.011 12.7 (10.0–NA) NR 0.11 (0.01–0.88) 0.037 0.73

567 Background: Selecting adjuvant therapy after nephrectomy for renal cell carcinoma (RCC) requires reviewing scattered electronic health records (EHR) and performing accurate eligibility and risk assessments. Large Language Models (LLMs) may assist physicians by extracting key clinical data and generating evidence-based recommendations between active surveillance and adjuvant pembrolizumab. Methods: We selected RCC patients who underwent nephrectomy at Mayo Clinic in Arizona, Florida, and Minnesota between December 2021 and September 2025. Data extraction from EHRs was performed by a HIPAA-compliant LLM (GPT-5 and Gemini 2.0-Flash-001) and compared to human labels. Data on patients’ demographics, medical history, pathology, laboratory results, and imaging were collected. GPT-5 generated treatment recommendations using a structured prompt that incorporated patients’ features and relevant clinical information that is available to clinicians. Concordance between the human labels and LLM was evaluated for extraction. For treatment recommendations, agreement between the LLM’s output and physician’s documented recommendation in clinical notes was assessed, with disagreements reviewed independently by two blinded genitourinary oncologists. Results: The dataset included 85 patients, median (range) age of 65 (34-84) years; 24.7% were females. Patients who declined adjuvant treatment for personal reasons were excluded. LLM achieved an extraction accuracy of 95% for presenting symptoms, 97% for BMI and ECOG, and > 99% for pathology variables. Concordance between LLM and physician treatment recommendations was 84.7% (72/85). Among 13 discrepant cases, one oncologist agreed with the LLM in 9 cases and the other in 8. The main reasons for disagreement between the physicians and the LLM were that the LLM favored active surveillance for patients with declining renal function, whereas physicians still considered adjuvant pembrolizumab appropriate in such cases. Additionally, the LLM demonstrated hesitancy to recommend pembrolizumab for patients with low performance status. Overall, the 13 discrepant cases were nuanced and challenging, with disagreement observed between the two oncologists themselves in three instances. Conclusions: LLMs demonstrated high accuracy in extracting clinically relevant variables and generating objective, evidence-based adjuvant treatment recommendations for RCC after nephrectomy. Importantly, LLM did not generate any inappropriate or potentially harmful recommendations. Our pipeline can serve as a supportive decision tool to enhance consistency in post-nephrectomy management and assist physicians in choosing between active surveillance and adjuvant pembrolizumab. However, final treatment decisions should remain the responsibility of the treating physician after comprehensive patient evaluation.

360 Background: Adding NHT to androgen deprivation therapy (ADT) improves survival in mCSPC, but their comparative impact on patient-reported QoL and time to deterioration (TTD) across FACT-P domains remains underexplored. Methods: MEDLINE and Embase were systematically searched from each database’s inception through August 21, 2025, to identify Phase III trials assessing NHT treatment in mCSPC. Time to FACT-P (total, physical, emotional, functional, and social/family well-being) deterioration across trials with the latest follow up were extracted. Hazards ratios (HRs) and 95% confidence intervals (CIs) were pooled and mixed treatment comparisons were made using network meta-analysis. P-scores were computed; a higher score indicated better QoL. Results: A total of 4 trials (12 references) with 4,037 patients met the inclusion criteria with median follow up ranging from 14.4 months to 30 months. Median time to QoL deterioration ranged from 11.4 to 12.9 months. In terms of FACT-P overall, darolutamide (DARO)+ADT (HR: 0.76; 95% CI: 0.62-0.93), abiraterone acetate (AAP)+ADT (0.85; 0.73-0.98) significantly improved TTD as compared to ADT alone. DARO+ADT was associated with significant improvement in TTD when compared to APA+ADT (0.75; 0.56-0.98). In terms of FACT-P social/family wellbeing, DARO+ADT significantly improved TTD when compared to AAP+ADT (0.75; 0.57-0.96), and ADT (0.79; 0.64-0.98). Likewise, ENZA+ADT significantly improved TTD when compared to AAP+ADT (0.77; 0.63-0.95), APA+ADT (0.88; 0.61-0.99) and ADT (0.82; 0.70-0.95). However, there was no statistically significant difference between DARO+ADT and ENZA+ADT. In terms of FACT-P physical wellbeing, AAP+ADT was associated with statistically significant improvement in TTD when compared to APA+ADT (0.66; 0.52-0.83), ENZA+ADT (0.74; 0.61-0.90), and ADT (0.75; 0.65-0.87). In terms of FACT-P functional wellbeing, DARO+ADT was associated with statistically significant improvement in TTD when compared to ADT (0.78; 0.63-0.96). The ranking analysis is showed in Table. Conclusions: Novel hormonal therapies significantly delay QoL deterioration in mCSPC, particularly across overall, emotional, and functional domains. Domain-specific analyses suggest darolutamide may preserve social/family and functional well-being, while abiraterone favors physical well-being. Rank table. Rank (P-score) FACT-Ptotal FACT-Psocial/family FACT-Pphysical FACT-Pfunctional FACT-Pemotional AAP+ADT 2 (0.73) 5 (0.18) 1 (0.98) 2 (0.62) 3 (0.52) APA+ADT 5 (0.17) 4 (0.22) 5 (0.07) 4 (0.33) 4 (0.52) DARO+ADT 1 (0.90) 1 (0.88) 2 (0.68) 1 (0.92) 2 (0.66) ENZA+ADT 3 (0.50) 2 (0.84) 4 (0.36) 3 (0.52) 1 (0.67) ADT 4 (0.18) 3 (0.38) 3 (0.42) 5 (0.11) 5 (0.13)