5017 Background: A subset of mCRPC patients (pts) will benefit from enzalutamide (ENZA) following disease progression on AAP. Preliminary data suggests that higher plasma ctDNA/total cfDNA fraction (ctDNA%) post-AAP predicts poor response to ENZA but preserved sensitivity to docetaxel (DOC). We hypothesize that ctDNA%-guided selection of enzalutamide or docetaxel in chemotherapy-naïve mCRPC post-abiraterone will improve clinical outcomes versus patient/clinician’s choice of treatment. Methods: A multi-centre, open-label, phase II trial randomized mCRPC pts progressing after AAP 1:1 to biomarker-directed therapy (Arm A: ctDNA% ≥2% receives DOC; ctDNA% <2% receives ENZA) or clinician’s choice of ENZA or DOC (Arm B, ctDNA% blinded). Baseline ctDNA% was assessed at screening using somatic mutations and genome-wide ploidy models. At progression, eligible pts could cross over to the alternative therapy. The primary endpoint was progression-free survival (PFS) defined as the time from treatment initiation to first documented disease progression (clinical, radiological, or PSA) or death from any cause. PFS was estimated using the Kaplan-Meier method and compared between arms using the log-rank tests. Secondary endpoints included PSA50 response (PSA decline ≥50% from baseline) and overall survival (OS). Results: Between October 2020 and June 2025, 56 pts were screened and 42 randomized 1:1 to Arm A (n=17) or Arm B (n=25). The trial was terminated due to slow accrual prior to the planned enrollment of 100 patients. At data cut-off (January 7, 2026), median follow-up was 38.0 months, and 94% (Arm A) and 92% (Arm B) had experienced disease progression or death. Baseline characteristics were balanced. ctDNA% <2% was observed in 35.3% and 32% of pts in Arm A and B, respectively. In Arm A, 11 pts were assigned to and received DOC (ctDNA% ≥2%) and 6 pts to ENZA (ctDNA <2%). In Arm B patient/clinician choice was DOC in 4 patients (3 pts with ctDNA ≥2%) and ENZA in 21 patients (7 pts with ctDNA <2%). PFS, OS and PSA50 all favoured biomarker directed therapy (Arm A) (Table). Conclusions: In this clinical utility study, ctDNA% to guide ENZA or DOC selection in patients previously treated with AAP led to superior PFS and OS compared to patient/clinician-selected therapy. These hypothesis-generating results illustrate the potential for ctDNA% as a predictive biomarker to assist in clinical decision making and support further studies. Clinical trial information: NCT04015622 . Primary & secondary outcomes. Biomarker-Directed (Arm A, n=17) Clinician’s Choice (Arm B, n=25) HR (95% CI) P-Value Progression-Free Survival (PFS), Months, median (95% CI) 5.6 (3.2-8.2) 2.5 (1.5-3.7) 0.4 (0.2-0.8) p = 0.01 PSA50 Response (%) 52.9% 28.0% p = 0.12 Overall Survival (OS), months, median (95% CI) 46.3 (12.2-NR) 15.3 (13.5-20.7) 0.4 (0.2-0.9) p = 0.04

e16544 Background: Current first-line treatments for mRCC include immune check point inhibitors (IO) and tyrosine kinase inhibitors (TKIs). There is limited data on attrition rate between lines of therapy and clinical outcomes in relation to the institutional volume and/or hospital setting. Objective: To analyze treatment patterns, attrition rate and clinical outcomes in mRCC pts following first-line therapy in high vs. lower volume or community setting within BC. BC Cancer currently operates six regional cancer centres providing care for pts across BC. Methods: BC Cancer’s centralized pharmacy data were screened to identify 547 mRCC pts treated with first-line IO combinations or TKIs from May 2019 to December 2024. Patient charts were reviewed for baseline characteristics, relevant treatment data, reasons for treatment choices, and outcomes. Treatments received in each line of therapy and attrition rate were tabulated using frequencies and percentages. Results: In the first 327 pts, median age was 65, with 79.2% male. 86 percent of pts received first-line IO combinations (193 IO-IO, 103 IO+TKI). 60 pts have not yet progressed on first line therapy. 76% of pts treated at the highest volume centre received second-line therapy, compared with 55% of pts treated at other regional centres and 48% of pts treated in the community. After first line IO-IO, 28% of pts treated in the highest volume centre had third-line therapy versus 25% of pts treated at other regional centres and only 6% in the community. Following first-line IO-TKI, 23% of pts from the highest volume centre had third-line therapy versus only 14% at the other regional centres and in the community. Due to the lack of funding for third-line therapy after IO/TKI start in BC, assessment of attrition rate to third-line therapy has to be interpreted with caution. The most common therapy in subsequent lines included TKI monotherapy. Conclusions: In mRCC pts treated with first-line IO combinations and TKIs, the most prevalent subsequent regimens are TKI-based therapies. The attrition rate with each additional line of therapy remains substantial. Lower attrition rates were observed among pts treated in high-volume specialized settings. These findings reflect the increasing complexity of RCC management and emphasizes the need to manage these pts in or in cooperation with specialized, high-volume centers within a multidisciplinary setting.

Supplementary Table 1. Representativeness of Study Participants.

Purpose: PD-L1 is overexpressed by dendritic cells in patients with metastatic castration-resistant prostate cancer (mCRPC) progressing on androgen receptor pathway inhibitors. We tested whether checkpoint blockade could enhance antitumor activity in mCRPC.Patients and Methods: In a multicenter open-label noncomparative randomized phase II study, patients with mCRPC treated with <= 1 prior cytotoxic chemotherapy, with measurable disease and progression on abiraterone and/or enzalutamide, were randomized to durvalumab 1,500 mg intravenously every 4 weeks +/- 4 doses of tremelimumab 75 mg intravenously. The primary endpoint was objective response (OR) by iRECIST using a Simon two-stage design. Correlative testing included PD-L1/cluster designation 8 IHC on baseline tumor biopsies and deep targeted sequencing of plasma cell-free DNA.Results: Fifty-two patients were enrolled. Median age was 70 years (range, 50-83 years), and 52% had prior taxane therapy for mCRPC. In stage I, 13 patients were randomized to durvalumab with no OR observed. Durvalumab + tremelimumab advanced to stage II with 39 patients enrolled (receiving a median three cycles, range 1-53). Durvalumab + tremelimumab-related adverse events were mainly <= grade 2 but led to discontinuation in seven patients. There were seven ORs [19.4% (95% confidence interval: 8.2%-36.0%); intention to treat 17.9% (95% confidence interval: 7.5%-33.5%)]. Five responding tumors were PD-L1-positive and two exhibited DNA damage repair defects. Responses were observed without high tumor mutational burden or other genomic indices of immunotherapy sensitivity.Conclusions: Durvalumab + tremelimumab is active in mCRPC, but patient selection remains a challenge. Further studies to develop predictive biomarkers are warranted.

189 Background: Patients (pts) with metastatic prostate cancer and liver metastases have poor prognosis, but clinicogenomic analyses are limited. We examined the clinical and genomic features of a cohort with prostate cancer liver metastases (PCLM). Methods: Pts were identified from a prospective population-based biobank in British Columbia, Canada. Eligible pts had PCLM identified on imaging prior to any line of treatment. We collated clinical outcomes with cell-free DNA (cfDNA) sequencing results including circulating tumor DNA fraction (ctDNA%) and genomic alterations. Results: 2048 metastatic prostate cancer pts were enrolled from October 2016 to July 2024, of whom 1258 had sufficient clinical annotation to evaluate PCLM status. Of these, 167 (13%) were diagnosed with PCLM: 36 (22%) with castration-sensitive prostate cancer (mCSPC), 48 (29%) prior to first-line castration-resistant prostate cancer treatment (1L mCRPC), 30 (18%) prior to 2L mCRPC, and 53 (32%) prior to ≥3L mCRPC. Median follow-up was 60.3 months. Median age at metastatic diagnosis was 68.7 (IQR 61.8-74.7) years, 93 (56%) had de novo metastatic disease, 20 (12%) had histologically proven small cell carcinoma, 27 (16%) had low PSA (<5ng/mL) at baseline, and 111 (66%) had ≥3 liver metastases. For mCSPC pts, 27 (75%) received treatment intensification beyond ADT alone, including 5 (14%) with platinum chemotherapy. Median overall survival (mOS) for mCSPC pts was 15.5 months (95% CI 11.7-28.1), and median time to castration resistance was 8.1 months (95% CI 5.7-10.7). For mCRPC pts, 94 (72%) had prior exposure to ARPI, 56 (43%) to taxane, and 8 (6%) to platinum. mOS for 1L mCRPC pts was 9.4 months (95% CI 6.7-12.6), for 2L mCRPC 4.9 months (95% CI 3.5-9.7) and for ≥3L mCRPC 5.6 months (95% CI 4.1-8.4). For 134 pts with cfDNA results the median ctDNA% was 24.7 (IQR 4.6-53.9). Alterations were found in TP53 (47%), PTEN (26%), RB1 (19%), with 82 pts (61%) having alterations in at least one tumor suppressor gene (TSG) and 34 (25%) in more than one TSG. 14 pts (10%) had a BRCA2 alteration. ctDNA% and detectable TSG alterations were associated with survival of pts with PCLM (Table). Of 17 long survivors (pts who lived >24 months from start of next line treatment), 12 had cfDNA results available, median ctDNA% was 13.6%, and only 1 had TSG loss detected ( PTEN ). Conclusions: Pts with PCLM exhibit poor clinical outcomes; both elevated ctDNA% and TSG loss correlate with reduced overall survival. mCSPCmOS (months)N = 24 pts with cfDNA 1L mCRPCmOS (months) N = 35 pts with cfDNA 2L mCRPCmOS (months)N = 25 pts with cfDNA TSG status No alteration detected 57 23.6 8 Alteration detected 11.7 6.8 4.3 Univariable HR (95% CI) 9.3 (2.0-42.7)p<0.01 5.6 (2.1-15.5)p<0.01 1.4 (0.6-3.2)p=0.4 ctDNA% ≤ median ctDNA% (24.7%) 57 18.1 9 > median ctDNA% (24.7%) 14.7 5.8 12 Univariable HR (95% CI) 2.4 (1.0-5.4)p<0.05 2.7 (1.5-5.0)p<0.01 1.1 (0.6-2.0)p=0.8

Abstract Fibroblast growth factor receptor (FGFR) gene alterations are common in urothelial cancer (UC) and targetable with erdafitinib. The co-approved companion diagnostic is an amplicon-based tumor tissue test that detects a limited number of recurrent FGFR alterations. However, FGFR status may change during disease progression, and rare FGFR alterations not typically covered by existing tests may be of clinical importance. Therefore, in a prospective pan-Canadian study (NCT06129084), we compared standard archival tissue testing to broad targeted sequencing of cell-free circulating tumor DNA (ctDNA) to detect FGFR alterations in UC. Eligible patients had progressing metastatic UC and were undergoing tissue FGFR testing. Blood was drawn at study enrollment and plasma cell-free DNA and matched white blood cell DNA underwent deep-targeted sequencing with a hybridization capture panel including the introns and exons of FGFR1-4 and the common breakpoint region of TACC3. Archival tissue selection and testing was initiated by the treating physician and typically used the Oncomine Focus assay. 210 patients were enrolled across 12 sites, and 140 patients have tissue and ctDNA results available for concordance assessment. 43/140 (31%) of cell-free DNA samples had low tumor fraction (<0.5%) and were considered inconclusive. For the remaining 97 same-patient tissue and ctDNA pairs, FGFR status was identical in 88 (91%). In 7 of 9 discordant cases, the tissue was negative, but ctDNA was positive for FGFR alterations: these were enriched for FGFR3-TACC3 fusions (4/7). Two discordant fusions had canonical breakpoints that can be identified by current amplicon-based companion diagnostics, indicating potential spatiotemporal heterogeneity in somatic FGFR status. The other two discordant (ctDNA-only) fusions had non-canonical breakpoints that are not amenable to detection by current amplicon-based companion diagnostics. In two mutation discordant cases (S249C positive in ctDNA only), lack of signal in tumor tissue could be explained by the presence of additional FGFR3 variants within 25bp of S249 and in cis - disrupting primer binding sites. There were only two cases of tissue-only FGFR alterations among the 97 evaluable pairs (FGFR3-TACC3 fusion, S249C). ctDNA testing of metastatic UC with a broad capture-based approach covering all exons and introns of FGFR genes can identify additional alterations not detected with current amplicon-based companion diagnostics applied to archival tissue. Our results support ctDNA testing as a valuable adjunct to tissue testing, but not as a replacement due to the high frequency of inconclusive ctDNA results, secondary to tumor fraction below the threshold. Assay design, including target enrichment technique, is a significant source of variability in tissue and ctDNA FGFR status and should be carefully considered in future clinical biomarker testing strategies. Citation Format: David C. Müller, Gillian Vandekerkhove, Andrew J. Murtha, Jack V.W. Bacon, Carlos Vasquez-Rios, Maria Stephenson, Kimia Rostin, Elena Schönlau, Connor Wells, Sunil Parimi, Krista Noonan, Naveen S. Basappa, Jenny J. Ko, Daygen Finch, Nimira Alimohamed, Tarek A. Bismar, Gang Wang, Andreas Papadakis, Lucia Nappi, Matti Annala, Cecily Q. Bernales, Alan Spatz, Kim N. Chi, Alexander W. Wyatt, Bernhard J. Eigl. Clinical test design affects tumor tissue and ctDNA FGFR gene status in metastatic urothelial cancer: a prospective study [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2024 May 17-20; Charlotte, NC. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(10_Suppl):Abstract nr A004.

No consensus strategies exist for prognosticating metastatic castration-resistant prostate cancer (mCRPC). Circulating tumor DNA fraction (ctDNA%) is increasingly reported by commercial and laboratory tests but its utility for risk stratification is unclear. Here, we intersect ctDNA%, treatment outcomes, and clinical characteristics across 738 plasma samples from 491 male mCRPC patients from two randomized multicentre phase II trials and a prospective province-wide blood biobanking program. ctDNA% correlates with serum and radiographic metrics of disease burden and is highest in patients with liver metastases. ctDNA% strongly predicts overall survival, progression-free survival, and treatment response independent of therapeutic context and outperformed established prognostic clinical factors. Recognizing that ctDNA-based biomarker genotyping is limited by low ctDNA% in some patients, we leverage the relationship between clinical prognostic factors and ctDNA% to develop a clinically-interpretable machine-learning tool that predicts whether a patient has sufficient ctDNA% for informative ctDNA genotyping (available online: https://www.ctDNA.org ). Our results affirm ctDNA% as an actionable tool for patient risk stratification and provide a practical framework for optimized biomarker testing.

Supplementary Figure S1 shows independent estimation of circulating tumor DNA fraction using whole exome sequencing.

Supplementary Figures S1 - S18

INTRODUCTION:In patients with prostate cancer (PCa), the identification of an alteration in genes associated with homologous recombination repair (HRR) has implications for prognostication, optimization of therapy, and familial risk mitigation. The aim of this study was to assess the genomic testing landscape of PCa in Canada and to recommend an approach to offering germline and tumor testing for HRR-associated genes.METHODS:The Canadian Genitourinary Research Consortium (GURC) administered a cross-sectional survey to a largely academic, multidisciplinary group of investigators across 22 GURC sites between January and June 2022.RESULTS:Thirty-eight investigators from all 22 sites responded to the survey. Germline genetic testing was initiated by 34%, while 45% required a referral to a genetic specialist. Most investigators (82%) reported that both germline and tumor testing were needed, with 92% currently offering germline and 72% offering tissue testing to patients with advanced PCa. The most cited reasons for not offering testing were an access gap (50%), uncertainties around who to test and which genes to test, (33%) and interpreting results (17%). A majority reported that patients with advanced PCa (74-80%) should be tested, with few investigators testing patients with localized disease except when there is a family history of PCa (45-55%).CONCLUSIONS:Canadian physicians with academic subspecialist backgrounds in genitourinary malignancies recognize the benefits of both germline and somatic testing in PCa; however, there are challenges in accessing testing across practices and specialties. An algorithm to reduce uncertainty for providers when ordering genetic testing for patients with PCa is proposed.

Supplementary Figure S7 shows the concordance of gene coverage logratios between 72-gene targeted sequencing and whole exome sequencing

<p>Supplementary Figure S17 provides whole exome somatic mutation profiles in patients that did not show temporal changes in 72-gene panel mutation profiles</p>

Supplementary Figure S8 shows that exome-wide sequencing supports ctDNA fraction estimates inferred with the 72-gene panel

Supplementary Figure S13 shows temporal changes in copy number profiles detected via 72-gene targeted sequencing

4577 Background: Fibroblast growth factor receptor (FGFR) inhibitors (e.g., erdafitinib) are increasingly important in the management of FGFR-mutated urothelial carcinoma. FDA-approved archival tissue testing for specific FGFR alterations was implemented as a companion diagnostic for erdafitinib. However, longitudinal sequencing studies indicate variable tumor FGFR status over time, and erdafitinib resistance mechanisms in metastatic urothelial carcinoma (mUC) are underreported. This ongoing study aims to evaluate the accuracy of cell-free DNA (cfDNA) compared to archival tissue testing in mUC for FGFR alterations detection, and to evaluate genomic mechanisms of erdafitinib resistance in cfDNA at progression. Methods: Patients with progressing mUC who were undergoing archival tissue testing for FGFR1-3 mutations and/or fusions and who had blood samples drawn during the management of their metastatic disease were eligible. Plasma cfDNA and matched leukocyte DNA were subjected to deep targeted sequencing with a custom panel including UC-specific gene loci and all clinically approved hotspots in FGFR1+2 and all exons and introns of FGFR3. Results: As of January 2023, 109 patients from 6 sites were enrolled. Median age at diagnosis was 71, 33% had upper urinary tract primaries, and 76% were male. Tissue and cfDNA results for comparison were available for 69 patients to date. Actionable somatic FGFR alterations were found in the tissue of 15 patients (31%); the most common alteration was the FGFR3 p.S249C mutation (67%). 50 of the analyzed cfDNA samples had detectable somatic circulating tumor DNA (ctDNA) variant allele fraction of 0.5% (72%). Of those, 49 had an evaluable tissue test result. Analysis of this subset revealed high concordance (92%) between the two test methods. With the assumption that archival tissue testing is considered the ‘gold standard’, sensitivity is 93% and specificity is 91%. The four discordant results comprised one cfDNA test with undetectable FGFR3-TACC3 fusion, which was detected in tissue and three positive ctDNA test results in patients with FGFR wild-type tissue tests. In one case at erdafitinib progression, ctDNA revealed multiple subclonal populations with distinct FGFR3 gatekeeper mutations suggesting polyclonal resistance. Conclusions: This ongoing study suggests cfDNA is a valuable minimally invasive adjunct to tissue-based assays for the detection of FGFR alterations to identify patients for FGFR inhibitor therapy and to monitor for mechanisms of resistance.

Supplementary Figure S10 provides Kaplan-Meier plots showing duration of treatment response in patients with different baseline circulating tumor DNA fractions

Supplementary Figure S25 shows concordant identification of intragenic copy number changes inside the androgen receptor (AR) gene based on a coverage-based approach and breakpoint-based approach

Supplementary Figure S24 shows that amplification of the enhancer region 640 kb upstream of the androgen receptor (AR) is common.

Supplementary Figure S12 shows the re-detection rate for somatic mutations identified in an earlier cfDNA sample

Supplementary Figure S14 shows temporal changes in whole exome somatic mutation profiles in patients that had evidence for somatic mutation profile change based on 72-gene targeted sequencing.