Abstract
Purpose
BRCA1/2 mutation testing identifies patients with metastatic castration-resistant prostate cancer (mCRPC) eligible for poly(ADP-ribose) polymerase inhibitors; however, clinical outcomes remain heterogeneous within these genotypes. We evaluated whether plasma-derived genomic instability status (GIS) was associated with outcomes beyond BRCA1/2 and PALB2 mutation status in the phase II TRITON2 study.
Methods
This retrospective biomarker analysis included TRITON2 patients with available baseline plasma profiled using Guardant360 Liquid. GIS is a probabilistic measure of homologous recombination deficiency (HRD) inferred from genome-wide copy number variants and somatic mutational signatures. Analyses were restricted to patients with baseline tumor fraction (TF) ≥ 25%. Radiographic progression-free survival (rPFS) was analyzed using Kaplan-Meier methods and Cox models adjusted for age and baseline TF. Secondary end points included objective response rate (ORR), serum prostate-specific antigen (PSA50) declines, and overall survival (OS).
Results
Among 104 patients with TF ≥ 25%, 43 (41%) had GIS detected. Compared with GIS detected, absence of GIS was associated with shorter rPFS (hazard ratio [HR], 2.81 [95% CI, 1.61 to 4.93]; P <.001). Joint stratification demonstrated patients with both GIS detected and BRCA/PALB2 mutation experienced the longest rPFS, whereas absence of GIS was associated with shorter rPFS regardless of BRCA/PALB2 mutation status. In BRCA2-mutated patients, rPFS trended being longer in those with GIS detected (compared with those without GIS detected; HR, 2.02 [95% CI, 0.95 to 4.29]; P = .069), with numerically higher ORR (25.7% v 14.3%) and PSA50 response rate (65.7% v 42.9%). No statistically significant differences in OS were observed.
Conclusion
Plasma-derived GIS was associated with improved rPFS and higher response rates in rucaparib-treated patients and further stratified outcomes within mutation-defined populations. These findings support genomic instability assessment as a complementary biomarker to mutation testing and warrant prospective validation.
Introduction
Metastatic castration-resistant prostate cancer (mCRPC) remains associated with substantial morbidity and mortality despite therapeutic advances in androgen receptor (AR)–directed therapies and taxane-based chemotherapy.1,2 Following progression on these agents, treatment options are limited, and durable disease control remains challenging. Genomic profiling has demonstrated that approximately 20%-25% of advanced prostate cancers harbor deleterious alterations in DNA damage repair genes, most commonly involving BRCA2, underscoring homologous recombination repair (HRR) deficiency as a clinically actionable vulnerability.3-5
CONTEXT
Key Objective
Does plasma-derived genomic instability status (GIS), a measure of homologous recombination deficiency (HRD), associate with clinical outcomes in rucaparib-treated metastatic castration-resistant prostate cancer (mCRPC) in the phase II TRITON2 study?
Knowledge Generated
Among 104 patients with evaluable tumor fraction, GIS detection was associated with significantly longer radiographic progression-free survival (rPFS) and higher objective and PSA50 response rates compared with absence of GIS detection. Joint stratification by GIS and BRCA/PALB2 mutation status demonstrated the longest rPFS in patients with both GIS detection and BRCA/PALB2 mutation.
Relevance
Plasma-derived genomic instability assessment may complement mutation testing by further stratifying outcomes within mutation-defined populations and supports prospective evaluation of composite HRD biomarkers in poly(ADP-ribose) polymerase inhibitor–treated mCRPC.
Poly(ADP-ribose) polymerase inhibitors (PARPi) exploit synthetic lethality in tumors with impaired HRR. The multicenter, open-label, phase II TRITON2 study evaluated rucaparib in patients with mCRPC who had progressed on androgen receptor–directed therapy and taxane chemotherapy, and harbored deleterious alterations in HRR genes.6,7 In the final analysis, rucaparib demonstrated clinically meaningful activity in patients with BRCA1/2-mutated mCRPC, including an objective response rate (ORR) of 46%, a prostate-specific antigen (PSA) decline of at least 50% (PSA50 response rate) of 53%, and a median radiographic progression-free survival (rPFS) of 10.8 months, leading to regulatory approval in this molecularly selected population.7,8 However, exploratory analyses from TRITON2 and other PARPi clinical trials have demonstrated heterogeneity in clinical outcomes even among mutation-defined cohorts.9-11 These observations suggest that mutation status alone may incompletely capture homologous recombination deficiency (HRD).
Genome-wide measures of HRD, including genomic scar–based metrics and mutational signatures, aim to reflect the cumulative structural and mutational consequences of impaired DNA repair rather than reliance on single-gene alterations.12-14 Such composite approaches provide an additional assessment of HRD. To date, however, most HRD assessment strategies have relied on tissue-based testing, which may be limited by tumor accessibility, sampling bias, and temporal evolution under treatment pressure.
Liquid biopsy platforms offer minimally invasive, real-time characterization of tumor genomic architecture. Genomic instability status (GIS), assessed using Guardant360 Liquid, integrates genome-wide copy number features and mutational signatures from circulating tumor DNA (ctDNA) to probabilistically infer HRD.15 Preliminary analyses in TRITON2 suggested that GIS detection may be associated with improved clinical outcomes in rucaparib-treated patients with mCRPC.15 Building on these findings, we conducted a retrospective biomarker analysis of the TRITON2 cohort to evaluate whether plasma-derived GIS refines clinical outcome stratification beyond BRCA/PALB2 mutation status in patients with mCRPC treated with rucaparib.
Methods
Study Design and Patients
TRITON2 (ClinicalTrials.gov identifier: NCT02952534) was an international, open-label, phase II study evaluating rucaparib 600 mg twice daily in patients with mCRPC who progressed on at least one AR-directed therapy and one taxane-based chemotherapy. Eligible patients harbored deleterious germline or somatic alterations in BRCA1, BRCA2, ATM, or other prespecified HRR genes. The primary end point of the parent study was ORR assessed by independent radiology review (IRR).6,9
This retrospective biomarker analysis included patients with available baseline plasma samples profiled using Guardant360 Liquid. Patients were excluded from the primary analysis if baseline tumor fraction (TF) < 25%, corresponding to the 95% limit of detection for GIS. Analytical validation demonstrated that GIS sensitivity decreases below this threshold, increasing the risk of false-negative classification in low-shedding samples; therefore, analyses were restricted to patients with TF ≥25% to ensure robust assessment of genomic instability.15 Patients were also excluded from the primary analysis if GIS was not evaluable or if the TRITON2 patient enrollment gene did not have a detectable mutation in that same enrollment gene by Guardant360 Liquid. Mutations predicted to be of clonal hematopoiesis (CH) origin were not included among detected enrollment gene mutations. One patient with GIS detected in the absence of a BRCA or PALB2 mutation was excluded from joint stratification analyses. Clinical outcome data were locked at study completion and analyzed independently of biomarker assessment.
Guardant360 Liquid and TF
Plasma samples were analyzed using Guardant360 Liquid, which integrates genome-wide somatic single nucleotide variants (SNVs) and copy number variants (CNVs) profiling with methylation-based TF quantification.16-18 Identification of CH variants was performed using a plasma-only gradient-boosted tree classifier trained on samples with matched plasma and buffy coat sequencing. The model integrates variant-level and sample-level features, including fragmentomic profiles, methylation patterns, and variant allele frequency, along with gene-level annotations informed by public databases. The classifier was optimized for high specificity to distinguish CH from tumor-derived variants and validated on independent data sets.19
Genomic Instability Status
GIS is a probabilistic classifier trained on independent datasets to infer HRD from genome-wide copy number alterations and somatic mutational signatures, including loss of heterozygosity.15 The model generates a probabilistic score (GIS score) based on genome-wide somatic SNV and CNV signatures indicative of deficiency in HRR genes or CNVs alone. Six COSMIC single-base substitution signatures20 with significant differences in samples with BRCA biallelic loss and relevant etiology were selected for inclusion in a projection-based SNV score. The model was trained on breast, prostate, pancreatic, and ovarian cancer clinical samples processed on Guardant360 Liquid to assess the sensitivity and accuracy of detecting biallelic loss of function in BRCA1 and BRCA2. A prostate-specific threshold for binary detection status (0.87) was established using an independent training data set to achieve high sensitivity and specificity.
Mutations included in signature identification were filtered to exclude putative CH variants. Samples deemed not evaluable for GIS were primarily attributable to CNV caller quality control failure.
End Points
The primary end point was IRR-assessed rPFS, defined as time from first dose of rucaparib to radiographic progression or death from any cause, adjudicated by independent central review using PCWG3 and modified RECIST v1.1 criteria.
Secondary outcomes included PSA response rate, ORR, and overall survival (OS). OS analyses were considered exploratory because of limited event maturity and the single-arm design. PSA response rate was measured as ≥50% decrease from baseline.
Statistical Analysis
Time-to-event outcomes were estimated using the Kaplan-Meier method. Comparisons were performed using log-rank tests. Hazard ratios (HRs) with 95% CIs were estimated using Cox proportional hazards models adjusted for age and baseline TF. Joint stratification analyses evaluated rPFS by GIS status and BRCA/PALB2 mutation status determined by ctDNA testing.
The analysis cohort included patients with baseline TF ≥ 25% (n = 116), of whom 104 patients had both detectable HRR gene mutations and evaluable GIS status. For grouped analyses combining GIS and BRCA1/2/PALB2 status, one patient with GIS-positive/BRCA1/2/PALB2-negative status was excluded per study protocol (final n = 103). HRR gene mutations were identified using strict variant-level filters requiring deleterious functional impact for SNVs and rearrangements (excluding somatic putative CH) or homozygous deletions for copy number variants. For patients with multiple HRR gene alterations, gene priority was assigned hierarchically, consistent with TRITON2 analysis7: BRCA2 > BRCA1 > PALB2 > ATM > CDK12 > CHEK2 > other HRR genes (alphabetically). Baseline characteristics of the biomarker-evaluable population were descriptively compared with the overall TRITON2 cohort7 to assess representativeness.
All analyses were performed using R version 4.4.1 with packages survival (version 3.8-3), survminer (version 0.5.0), and gtsummary (version 2.0.4).
IRB/Ethics Statement
TRITON2 (ClinicalTrials.gov identifier: NCT02952534) was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. The TRITON2 protocol was approved by institutional review boards or independent ethics committees at participating institutions. All patients provided written informed consent.
Results
Patient Population and Assay Evaluability
Of 277 patients enrolled in TRITON2, baseline plasma samples were available and evaluable for GIS in 230 (83%) patients, of whom 116 (50%) had TF ≥ 25%. In the TRITON2 study, patients were enrolled if they had a mutation in a specified HRR gene. For our primary analysis, if the patient did not have a detectable mutation in that same enrollment gene by Guardant360 Liquid, we excluded them from our analysis (n = 7). Patients were also excluded if GIS was not evaluable on Guardant360 Liquid because of assay quality control failure (n = 6). One patient met both exclusion criteria. The analytical cohort comprised 104 patients (Fig 1). Among these 104 patients, GIS was detected in 43 patients (41%; Appendix Table A1).
Fig 1.
Patient inclusion and analysis population for GIS assessment in TRITON2. Patients enrolled in the phase II TRITON2 study based on deleterious alterations in HRR genes (BRCA1, BRCA2, ATM, CDK12, CHEK2, PALB2, and other prespecified HRR genes) underwent retrospective baseline plasma profiling using Guardant360 Liquid. Patients with baseline TF < 25% were excluded because of reduced sensitivity of GIS detection below the estimated limit of detection. Patients were excluded if GIS was not evaluable or if the enrollment gene did not have a detectable mutation identified with Guardant360 Liquid. One patient meeting both exclusion criteria was counted once. One patient with GIS detected in the absence of a BRCA or PALB2 mutation was excluded from joint stratification analyses. The final Kaplan-Meier analysis population included 103 patients. GIS, genomic instability status; HRR, homologous recombination repair; TF, tumor fraction.
Baseline demographic and clinical characteristics of the biomarker-evaluable population were comparable with those of the overall TRITON2 cohort (Appendix Table A1). Age, Gleason score, and enrollment gene distribution were similar between cohorts, whereas baseline PSA was modestly higher in the biomarker-evaluable population.
Prevalence of GIS by Individual HRR Gene
GIS detection was observed among 35 of 56 (63%) patients with BRCA2 mutations, four of six (67%) patients with PALB2 mutations, three of seven (43%) patients with BRCA1 mutations, and one of 20 (5%) patients with ATM mutations (Fig 2; Appendix Table A1). Thirty patients had at least two co-occurring HRR gene mutations (30/104, 29%), and of those, 14 (47%) had GIS detected (Appendix Table A2).
Fig 2.
Distribution of GIS score by HRR gene. Box and whisker plots showing GIS score distribution stratified by HRR gene alteration. Points represent individual patients and are color-coded by GIS status (GIS detected v not detected). The dashed horizontal line indicates the predefined GIS detection threshold (≥0.87). Gene assignment followed hierarchical classification (BRCA2, BRCA1, ATM, CDK12, CHEK2, PALB2, followed by other HRR genes) consistent with TRITON2 hierarchy.7 Included variants consisted of deleterious SNV, rearrangements or CNV homozygous deletions. CH, clonal hematopoiesis; CNV, copy number variant; GIS, genomic instability status; HRR, homologous recombination repair; SNV, single nucleotide variant.
Association of GIS With rPFS, ORR, PSA50 Response Rate, and OS
The median rPFS was 9.7 months in the GIS-detected group (n = 43) and 5.3 months in the GIS not detected group (n = 61). Compared with GIS detected, GIS not detected was associated with shorter rPFS (HR, 2.81 [95% CI, 1.61 to 4.93]; P <.001; Fig 3). In a multivariate analysis including age, GIS, and baseline TF, age was not significantly associated with rPFS (HR, 1.02 [95% CI, 0.99 to 1.05]; P = .3), while baseline TF was independently associated with rPFS (HR, 5.62 [95% CI, 1.38 to 22.8]; P = .016). Among patients evaluable for objective response (IRR-evaluable population), ORR was 27.9% (12/43) in the GIS-detected group and 6.6% (4/61) in the GIS-not detected group. A PSA decline of at least 50% (PSA50 response rate) was observed in 67.4% (29/43) and 16.4% (10/61) of patients, respectively.
Fig 3.
rPFS by GIS. Kaplan-Meier estimates of rPFS in patients with metastatic castration-resistant prostate cancer treated with rucaparib, stratified by GIS (GIS detected v not detected). rPFS was defined as time from first dose of rucaparib to radiographic progression or death, adjudicated by independent central review using PCWG3 and modified RECIST v1.1 criteria. Patients with GIS detected (n = 43) had longer rPFS compared with those without GIS detected (n = 61). Compared with GIS detected, absence of GIS was associated with shorter rPFS (HR, 2.81 [95% CI, 1.61 to 4.93]; log-rank P < .001). GH, Guardant Health; GIS, genomic instability status; HR, hazard ratio; ORR, objective response rate; PSA50, prostate-specific antigen decline ≥50%; rPFS, radiographic progression-free survival; TS, tumor fraction.
Patients were stratified by combined GIS and BRCA1/2/PALB2 mutation status as GIS detected/mutation detected (n = 42), GIS not detected/mutation detected (n = 27), and GIS not detected/mutation not detected (n = 34; Fig 4). The median rPFS was 9.7 months in the GIS detected/mutation detected group, 7.4 months in the GIS not detected/mutation detected group, and 3.7 months in the GIS not detected/mutation not detected group. Compared with GIS detected/mutation detected, GIS not detected/mutation detected was associated with an increased risk of progression (HR, 2.09 [95% CI, 1.09 to 3.99]; P = .026), and GIS not detected/mutation not detected was associated with further increased risk (HR, 6.14 [95% CI, 2.91 to 12.9]; P <.001).
Fig 4.
Joint stratification of rPFS by GIS and BRCA/PALB2 mutation status. Kaplan-Meier estimates of rPFS stratified by combined GIS (GIS detected v not detected) and BRCA or PALB2 mutation status: GIS detected/mutation detected (n = 42), GIS not detected/mutation detected (n = 27), and GIS not detected/mutation not detected (n = 34). Patients with both GIS detected and a BRCA or PALB2 mutation had the longest rPFS. Absence of GIS was associated with shorter rPFS regardless of mutation status. Compared with GIS detected/mutation detected, GIS not detected/mutation detected was associated with increased risk of progression (HR 2.09; 95% CI, 1.09-3.99), and GIS not detected/mutation not detected was associated with further increased risk (HR 6.14; 95% CI, 2.91-12.9). GIS, genomic instability status; HR, hazard ratio; ORR, objective response rate; PSA50, prostate-specific antigen decline ≥50%; rPFS, radiographic progression-free survival.
For OS, no statistically significant differences were observed among the GIS detected/mutation detected (n = 42), GIS not detected/mutation detected (n = 27), and GIS not detected/mutation not detected (n = 34) groups (HR, 1.45 [95% CI, 0.79 to 2.68]; P = .2; HR, 1.66 [95% CI, 0.98 to 2.79]; P = .058, respectively; Appendix Fig A1).
GIS and Response in BRCA2-Mutated Patients
Within patients eligible for analysis, the majority had a BRCA2 mutation (56/104, 54%). We compared GIS detected (n = 35) and not detected in this subset of patients (n = 21). rPFS trended toward being longer among patients with GIS detected compared with GIS not detected, although this exploratory analysis did not reach statistical significance (HR for GIS not detected v GIS detected, 2.02 [95% CI, 0.95 to 4.29]; P = .069; Fig 5). ORR was 25.7% (9/35) in the GIS-detected group and 14.3% (3/21) in the GIS-not detected group. PSA50 response was observed in 65.7% (23/35) and 42.9% (9/21) of patients, respectively.
Fig 5.
rPFS in patients with BRCA2-mutated mCRPC stratified by GIS. Kaplan-Meier estimates of rPFS among patients with BRCA2-mutated metastatic castration-resistant prostate cancer treated with rucaparib (n = 56), stratified by GIS. rPFS trended toward being longer in patients with GIS detected compared with those without GIS detected, although this did not reach statistical significance (P = .069). This analysis was exploratory and not powered for definitive inference. HRD, homologous recombination deficiency; GH, Guardant Health; GIS, genomic instability status; HR, hazard ratio; ORR, objective response rate; PSA50, prostate-specific antigen decline ≥50%; rPFS, radiographic progression-free survival; TF, tumor fraction.
Discussion
In this retrospective biomarker analysis of TRITON2, plasma-derived GIS was associated with improved rPFS and higher objective and PSA50 response rates in rucaparib-treated patients with mCRPC. Patients with GIS detected experienced longer rPFS, and this association remained significant after adjustment for age and baseline TF. Baseline TF was independently associated with rPFS, consistent with previous studies demonstrating that ctDNA burden at baseline is a strong prognostic biomarker.21,22 These findings suggest that TF reflects disease burden, whereas GIS captures complementary biological features associated with sensitivity to PARP inhibition.
Although BRCA1/2 and selected HRR gene alterations guide PARP inhibitor treatment in mCRPC,7,10 heterogeneity in clinical outcomes within these mutation-defined populations is well recognized.9,11 In the present analysis, GIS further stratified outcomes among patients harboring BRCA or PALB2 mutations, identifying subgroups with distinct clinical trajectories. Specifically, in patients with BRCA2 mutations where the strongest benefit from PARP inhibition has been documented, GIS detection was associated with numerically longer rPFS; however, given the exploratory nature of this subgroup analysis and limited sample size, these findings should be interpreted cautiously and considered hypothesis generating. GIS detection within this subgroup may provide additional biologic context to understand patient response to rucaparib. Only one patient in this cohort demonstrated GIS detection in the absence of a BRCA1/2 or PALB2 mutation, limiting conclusions regarding the independent role of GIS in mutation-negative populations; future studies including larger numbers of such patients will be important to clarify whether GIS may identify additional individuals who could benefit from PARP inhibition beyond current mutation-defined eligibility criteria. Importantly, these findings do not challenge the established role of mutation testing in treatment selection but rather suggest that mutation status alone may incompletely capture the biologic spectrum of therapeutic sensitivity and that GIS may serve as a complementary biomarker to refine risk stratification within clinically eligible populations.
Genomic scar–based HRD signatures have previously been described in prostate cancer and shown to correlate with underlying BRCA deficiency12,13; however, their prospective evaluation as predictive biomarkers of PARP inhibitor benefit in mCRPC has been limited. Prior phase III studies, including PROfound,10 PROpel,23 and TALAPRO-2,24 have primarily relied on gene-level HRR mutation status rather than composite genomic instability metrics to define treatment-eligible populations. To our knowledge, plasma-derived genomic instability assessment integrating somatic mutational signatures and genome-wide copy number alterations has not previously been systematically evaluated in relation to PARP inhibitor outcomes within a clinical trial cohort.
Specific study limitations warrant consideration. First, TRITON2 was a single-arm study, and this analysis cannot establish GIS as a predictive biomarker in a randomized comparative setting. The observed associations represent enrichment within a treated population rather than discrimination between therapies. Second, subgroup sample sizes were modest, particularly for exploratory analyses within individual gene cohorts, limiting conclusions that can be drawn in the small subgroups and identifying an area for additional research. Third, genomic instability assessment requires sufficient ctDNA levels, and the use of a minimum TF threshold resulted in a study population with higher TF. As a result, the generalizability of these findings to patients with low TF may be limited. Comparison with the overall TRITON2 cohort (published population) demonstrated similar distributions of age, Gleason score, and enrollment gene categories. Baseline PSA was modestly higher in the biomarker-evaluable population, without clear evidence of broader clinical imbalance. Overall, these findings suggest that restriction to patients with evaluable plasma and TF ≥ 25% did not meaningfully alter the clinical composition of the study population.
Taken together, these results suggest that a liquid-based genomic instability assessment may represent a clinically meaningful extension of mutation-based testing in mCRPC. Although limited by the retrospective design and modest sample size of this biomarker analysis, these findings provide hypothesis-generating evidence supporting further evaluation of GIS as a complementary biomarker to mutation testing. Prospective integration of genomic instability assessment into randomized PARP inhibitor trials will be essential to determine whether GIS can inform patient selection and advance precision oncology in advanced prostate cancer.
ACKNOWLEDGMENT
The authors thank the patients and their families who participated in the TRITON2 study, as well as the investigators, research staff, and site personnel involved in trial conduct. We acknowledge Clovis Oncology for providing patient samples and clinical data to Guardant Health for this retrospective, genomic analysis. We would like to also acknowledge Kannan Shah and Jill Lai for their analytical expertise, as well as Scott Newman, Martina Lefterova, and Kimberly Banks for serving as scientific advisors. We also acknowledge ChatGPT for writing assistance.
Appendix
Fig A1.
Joint stratification of OS by GIS and BRCA/PALB2 mutation status. Kaplan-Meier estimates of OS stratified by GIS and BRCA/PALB2 mutation status. No statistically significant differences in OS were observed, consistent with limited event maturity and the single-arm design of TRITON2. GIS, genomic instability status; OS, overall survival.
Table A1.
Baseline Demographic and Clinical Characteristics of the Biomarker-Evaluable Population Compared With the Overall TRITON2 Cohort
| Gene | No. (%) | Age, Median (IQR) | Baseline PSA, Median (IQR) | Gleason ≥8, No. (%) |
|---|---|---|---|---|
| GH enrolled cohort | ||||
| BRCA | 63 (61%) | 71 (64-76) | 114 (37-416) | 43 (68) |
| PALB2 | 6 (6%) | 70 (60-75) | 83 (33-188) | 5 (83) |
| Other | 35 (34%) | 66 (60-74) | 109 (44-342) | 23 (66) |
| Total | 104 | 70 (62-75) | 112 (37-403) | 71 (68) |
| TRITON-2 source data | ||||
| BRCA | 172 (62%) | 72 (66-76) | 69 (19-264) | 115 (67) |
| PALB2 | 11 (4%) | 70 (68-72) | 33 (20-124) | 7 (64) |
| Other | 94 (34%) | 72 (64-76) | 74 (24-254) | 57 (61) |
| Total | 277 | 71 (65-76) | 72 (22-249) | 179 (65) |
NOTE. All statistics computed from individual-level data; No. (%) shows count and percentage of cohort total. GH cohort: BRCA1 and BRCA2 combined into BRCA; other includes ATM and all other HRR genes. TRITON-2 source: BRCA includes BRCA1 and BRCA2; PALB2 classified separately; other includes ATM, CDK12, CHEK2, and remaining HRR genes. Gleason ≥8 derived from TGLEASON field (numeric threshold ≥8); unknown/missing excluded from denominator. No. (%) reflects proportion within each cohort. Baseline demographic and clinical characteristics of patients included in the GIS analysis (biomarker-evaluable population; n = 104) compared with the overall TRITON2 cohort as previously published.7 Continuous variables are reported as median (IQR) and categorical variables as No. (%). The biomarker-evaluable population included patients with baseline TF ≥25% and evaluable GIS. The overall TRITON2 cohort includes all enrolled patients regardless of plasma evaluability. Patients with multiple HRR gene alterations were assigned to a single category using a predefined hierarchy (BRCA2, BRCA1, PALB2, then other HRR genes). Comparisons are descriptive.
Abbreviations: GH, Guardant Health; GIS, genomic instability status; HRR, homologous recombination repair; PSA, prostate-specific antigen; TF, tumor fraction.
Table A2.
Patients With Multiple HRR Gene Mutations
| Patients With Multiple HRR Genes Detected by GH | |||
|---|---|---|---|
| Among Enrolled Patients Only | |||
| Patient ID | Genes Detected | No. of Genes | GIS Status |
| 28059-007 | ATM, BRCA2, CDK12, CHEK2, RAD51B | 5 | Not detected |
| 21025-004 | ATM, BRCA2, NBN | 3 | Detected |
| 28063-004 | ATM, BRCA2, NBN | 3 | Detected |
| 32257-001 | BRCA2, PALB2, RAD54L | 3 | Detected |
| 11015-019 | BRCA2, CHEK2 | 2 | Detected |
| 11016-032 | ATM, NBN | 2 | Not detected |
| 11019-028 | BRCA2, CDK12 | 2 | Detected |
| 12007-002 | BRCA2, PALB2 | 2 | Not detected |
| 12007-009 | BRCA2, RAD51D | 2 | Detected |
| 14025-003 | BRIP1, CHEK2 | 2 | Not detected |
| 16031-020 | BRCA1, BRCA2 | 2 | Not detected |
| 17027-009 | BRCA2, RAD51C | 2 | Not detected |
| 17029-008 | BRCA2, RAD54L | 2 | Detected |
| 19001-005 | FANCA, NBN | 2 | Not detected |
| 19001-014 | ATM, BRCA1 | 2 | Not detected |
| 20017-006 | BRCA2, CHEK2 | 2 | Detected |
| 20020-033 | BRCA1, CDK12 | 2 | Not detected |
| 28031-010 | BRCA2, RAD51D | 2 | Detected |
| 28038-004 | ATM, CDK12 | 2 | Not detected |
| 28038-031 | BRIP1, PALB2 | 2 | Not detected |
| 31029-028 | NBN, RAD51D | 2 | Not detected |
| 32232-002 | BRCA2, NBN | 2 | Detected |
| 32251-002 | BRCA1, CDK12 | 2 | Detected |
| 32251-006 | CDK12, CHEK2 | 2 | Not detected |
| 32254-001 | BRCA2, CDK12 | 2 | Not detected |
| 32261-033 | BRCA2, FANCA | 2 | Detected |
| 32274-014 | ATM, BRCA2 | 2 | Not detected |
| 32288-045 | BRCA2, PALB2 | 2 | Detected |
| 32308-002 | ATM, BRCA2 | 2 | Not detected |
| 32355-004 | ATM, BRCA1 | 2 | Detected |
NOTE. This table summarizes patients in the study cohort with pathogenic alterations in ≥2 HRR genes identified by ctDNA analysis. For each patient, the specific HRR genes detected, total number of HRR genes with pathogenic mutations, and GIS (detected vs not detected) are shown. Among patients with multiple HRR gene alterations (30/104, 29%), 14 had GIS detected and 16 did not.
Abbreviations: ctDNA, circulating tumor DNA; GH, Guardant Health; GIS, genomic instability status; HRR, homologous recombination repair.
Wassim Abida
Honoraria: Roche, Medscape, Aptitude Health, Clinical Education Alliance, OncLive/MJH Life Sciences, touchIME, Pfizer, theMedNet, Prostate Cancer Foundation, Decera, IDEOlogy Health, Research to Practice
Consulting or Advisory Role: Clovis Oncology, Janssen, ORIC Pharmaceuticals, Daiichi Sankyo, AstraZeneca/MedImmune, Pfizer, Laekna Therapeutics, MOMA Therapeutics, Endeavor BioMedicines, K36 Therapeutics, Boundless Bio, Tolmar, DualityBio
Research Funding: AstraZeneca (Inst), Zenith Epigenetics (Inst), Clovis Oncology (Inst), ORIC Pharmaceuticals (Inst), Epizyme (Inst), Nuvation Bio (Inst), Merus (Inst), TransThera Biosciences (Inst), MOMA Therapeutics (Inst), K36 Therapeutics (Inst)
Patents, Royalties, Other Intellectual Property: UptoDate
Uncompensated Relationships: Nuvation Bio, K36 Therapeutics
Alan H. Bryce
Employment: City of Hope
Leadership: City of Hope
Honoraria: Pfizer, Janssen, Novartis, Astellas Pharma, Bayer, Janssen Oncology, Lilly, MOMA Therapeutics, Tolmar, AstraZeneca, Johnson & Johnson/Janssen
Consulting or Advisory Role: Astellas Pharma, Johnson & Johnson/Janssen, Lantheus Medical Imaging
Research Funding: Provectus Biopharmaceuticals
Travel, Accommodations, Expenses: Pfizer (Inst), Bayer, Tolmar, Pharma& (Inst)
Simon Chowdhury
Employment: Medical Watch
Stock and Other Ownership Interests: Telix Pharmaceuticals, Medical Watch
Honoraria: Novartis
Consulting or Advisory Role: Astellas Pharma, Bayer, Pfizer, Janssen-Cilag, Novartis, Telix Pharmaceuticals, Remedy Bio, Huma, AstraZeneca
Speakers' Bureau: Pfizer, Janssen-Cilag
Akash Patnaik
Stock and Other Ownership Interests: Gilead Sciences, Merck, Infinity Pharmaceuticals
Honoraria: Prime Oncology, Medscape, BostonGene
Consulting or Advisory Role: Exelixis, Guidepoint Pharmacy, Gerson Lehrman Group, Curio Science, Novartis, Johnson & Johnson/Janssen, AstraZeneca
Research Funding: Bristol Myers Squibb (Inst), ProGenics (Inst), Clovis Oncology (Inst), Laekna Health Care (Inst), AstraZeneca (Inst), Xencor (Inst), Zenith Pharmaceuticals (Inst), Exelixis (Inst)
Travel, Accommodations, Expenses: Bristol Myers Squibb, Prime Oncology, Exelixis
Brooke Overstreet
Employment: Guardant Health
Stock and Other Ownership Interests: Guardant Health
Research Funding: Guardant Health
Patents, Royalties, Other Intellectual Property: I have patents filed with Guardant Health
Travel, Accommodations, Expenses: Guardant Health
Matthew Margolis
Employment: GRAIL, Guardant Health
Stock and Other Ownership Interests: Roche, Illumina, GRAIL, Guardant Health
Honoraria: Sanofi/Regeneron
Patents, Royalties, Other Intellectual Property: IGF1R Activation Mutations and Uses Thereof
Travel, Accommodations, Expenses: Sanofi/Regeneron
Pegah Safabakhsh
Employment: Guardant Health
Stock and Other Ownership Interests: Guardant Health
Travel, Accommodations, Expenses: Guardant Health
Karim Fizazi
Honoraria: Janssen (Inst), Astellas Pharma (Inst), Bayer (Inst), Novartis (Inst)
Consulting or Advisory Role: Janssen Oncology (Inst), Bayer (Inst), Astellas Pharma (Inst), AstraZeneca (Inst), Amgen (Inst), Bristol Myers Squibb (Inst), Novartis (Inst), Pfizer (Inst), Daiichi Sankyo Europe GmbH (Inst), MSD (Inst)
Travel, Accommodations, Expenses: Janssen, MSD, AstraZeneca, Pfizer, Bayer, Novartis
No other potential conflicts of interest were reported.
Disclaimer
TRITON2 was sponsored by Clovis Oncology. The retrospective plasma genomic and statistical analyses were conducted by Guardant Health. Guardant Health authors had access to the full dataset. Academic authors had access to summarized data outputs and participated in data interpretation and manuscript development. All authors approved the final manuscript and the decision to submit for publication.
Prior Presentation
Presented in part at the 2025 ASCO Annual Meeting, Chicago, IL, May 30-June 3, 2025.
Support
Supported by Guardant Health. The sponsor participated in study design and analysis. W.A. is partially funded by NIH/NCI Cancer Center Support Grant P30-CA008748 and NCI SPORE grant P50-CA092629. A.P. is funded by NIH/NCI Prostate SPORE P50CA180995, NIH/NCI R01 CA273914-01, Prostate Cancer Foundation Challenge Awards, V Foundation Translational Research Award, Department of Defense Prostate Cancer Clinical Consortium Award W81XWH-21-PCRP-CCA, NIH/NCI Cancer Center Support Grant P30CA014599.
W.A., A.H.B., S.C., A.P., and K.F. authors contributed equally to this work.
Data Sharing Statement
A data sharing statement provided by the authors is available with this article at DOI https://doi.org/10.1200/PO-26-00390.
The data that support the findings of this study are available from Guardant Health, but restrictions apply to the availability of these data, which are not publicly available. Data may be made available upon reasonable request and with permission of the sponsor.
Author Contributions
Conception and design: Wassim Abida, Simon Chowdhury, Akash Patnaik, Brooke Overstreet, Pegah Safabakhsh, Karim Fizazi
Provision of study materials or patients: Wassim Abida, Akash Patnaik, Karim Fizazi
Collection and assembly of data: Wassim Abida, Simon Chowdhury, Akash Patnaik, Brooke Overstreet, Pegah Safabakhsh, Karim Fizazi
Data analysis and interpretation: All authors
Manuscript writing: All authors
Final approval of manuscript: All authors
Accountable for all aspects of the work: All authors
Authors' Disclosures of Potential Conflicts of Interest
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.
Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).
Wassim Abida
Honoraria: Roche, Medscape, Aptitude Health, Clinical Education Alliance, OncLive/MJH Life Sciences, touchIME, Pfizer, theMedNet, Prostate Cancer Foundation, Decera, IDEOlogy Health, Research to Practice
Consulting or Advisory Role: Clovis Oncology, Janssen, ORIC Pharmaceuticals, Daiichi Sankyo, AstraZeneca/MedImmune, Pfizer, Laekna Therapeutics, MOMA Therapeutics, Endeavor BioMedicines, K36 Therapeutics, Boundless Bio, Tolmar, DualityBio
Research Funding: AstraZeneca (Inst), Zenith Epigenetics (Inst), Clovis Oncology (Inst), ORIC Pharmaceuticals (Inst), Epizyme (Inst), Nuvation Bio (Inst), Merus (Inst), TransThera Biosciences (Inst), MOMA Therapeutics (Inst), K36 Therapeutics (Inst)
Patents, Royalties, Other Intellectual Property: UptoDate
Uncompensated Relationships: Nuvation Bio, K36 Therapeutics
Alan H. Bryce
Employment: City of Hope
Leadership: City of Hope
Honoraria: Pfizer, Janssen, Novartis, Astellas Pharma, Bayer, Janssen Oncology, Lilly, MOMA Therapeutics, Tolmar, AstraZeneca, Johnson & Johnson/Janssen
Consulting or Advisory Role: Astellas Pharma, Johnson & Johnson/Janssen, Lantheus Medical Imaging
Research Funding: Provectus Biopharmaceuticals
Travel, Accommodations, Expenses: Pfizer (Inst), Bayer, Tolmar, Pharma& (Inst)
Simon Chowdhury
Employment: Medical Watch
Stock and Other Ownership Interests: Telix Pharmaceuticals, Medical Watch
Honoraria: Novartis
Consulting or Advisory Role: Astellas Pharma, Bayer, Pfizer, Janssen-Cilag, Novartis, Telix Pharmaceuticals, Remedy Bio, Huma, AstraZeneca
Speakers' Bureau: Pfizer, Janssen-Cilag
Akash Patnaik
Stock and Other Ownership Interests: Gilead Sciences, Merck, Infinity Pharmaceuticals
Honoraria: Prime Oncology, Medscape, BostonGene
Consulting or Advisory Role: Exelixis, Guidepoint Pharmacy, Gerson Lehrman Group, Curio Science, Novartis, Johnson & Johnson/Janssen, AstraZeneca
Research Funding: Bristol Myers Squibb (Inst), ProGenics (Inst), Clovis Oncology (Inst), Laekna Health Care (Inst), AstraZeneca (Inst), Xencor (Inst), Zenith Pharmaceuticals (Inst), Exelixis (Inst)
Travel, Accommodations, Expenses: Bristol Myers Squibb, Prime Oncology, Exelixis
Brooke Overstreet
Employment: Guardant Health
Stock and Other Ownership Interests: Guardant Health
Research Funding: Guardant Health
Patents, Royalties, Other Intellectual Property: I have patents filed with Guardant Health
Travel, Accommodations, Expenses: Guardant Health
Matthew Margolis
Employment: GRAIL, Guardant Health
Stock and Other Ownership Interests: Roche, Illumina, GRAIL, Guardant Health
Honoraria: Sanofi/Regeneron
Patents, Royalties, Other Intellectual Property: IGF1R Activation Mutations and Uses Thereof
Travel, Accommodations, Expenses: Sanofi/Regeneron
Pegah Safabakhsh
Employment: Guardant Health
Stock and Other Ownership Interests: Guardant Health
Travel, Accommodations, Expenses: Guardant Health
Karim Fizazi
Honoraria: Janssen (Inst), Astellas Pharma (Inst), Bayer (Inst), Novartis (Inst)
Consulting or Advisory Role: Janssen Oncology (Inst), Bayer (Inst), Astellas Pharma (Inst), AstraZeneca (Inst), Amgen (Inst), Bristol Myers Squibb (Inst), Novartis (Inst), Pfizer (Inst), Daiichi Sankyo Europe GmbH (Inst), MSD (Inst)
Travel, Accommodations, Expenses: Janssen, MSD, AstraZeneca, Pfizer, Bayer, Novartis
No other potential conflicts of interest were reported.
REFERENCES
- 1. Fizazi K, Attard G, Azad AA, et al. Advanced and metastatic prostate cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol. 2026;37:590–607. doi: 10.1016/j.annonc.2026.01.015. [DOI] [PubMed] [Google Scholar]
- 2. Gillessen S, Turco F, Davis I, et al. Management of patients with advanced prostate cancer: Report from the 2024 Advanced Prostate Cancer Consensus Conference (APCCC) Eur Urol. 2025;87:157–216. doi: 10.1016/j.eururo.2024.09.017. [DOI] [PubMed] [Google Scholar]
- 3. Pritchard CC, Mateo J, Walsh MF, et al. Inherited DNA-repair gene mutations in men with metastatic prostate cancer. N Engl J Med. 2016;375:443–453. doi: 10.1056/NEJMoa1603144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Nicolosi P, Ledet E, Yang S, et al. Prevalence of germline variants in prostate cancer and implications for current genetic testing guidelines. JAMA Oncol. 2019;5:523–528. doi: 10.1001/jamaoncol.2018.6760. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Schostak M, Bradbury A, Briganti A, et al. Practical guidance on establishing a molecular testing pathway for alterations in homologous recombination repair genes in clinical practice for patients with metastatic prostate cancer. Eur Urol Oncol. 2024;7:344–354. doi: 10.1016/j.euo.2023.08.004. [DOI] [PubMed] [Google Scholar]
- 6. Abida W, Patnaik A, Campbell D, et al. Rucaparib in men with metastatic castration-resistant prostate cancer harboring a BRCA1 or BRCA2 gene alteration. J Clin Oncol. 2020;38:3763–3772. doi: 10.1200/JCO.20.01035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Abida W, Campbell D, Patnaik A, et al. Rucaparib for the treatment of metastatic castration-resistant prostate cancer associated with a DNA damage repair gene alteration: Final results from the phase 2 TRITON2 study. Eur Urol. 2023;84:321–330. doi: 10.1016/j.eururo.2023.05.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Anscher MS, Chang E, Gao X, et al. FDA approval summary: Rucaparib for the treatment of patients with deleterious BRCA-mutated metastatic castration-resistant prostate cancer. Oncologist. 2021;26:139–146. doi: 10.1002/onco.13585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Abida W, Campbell D, Patnaik A, et al. Non-BRCA DNA damage repair gene alterations and response to the PARP inhibitor rucaparib in metastatic castration-resistant prostate cancer: Analysis from the phase 2 TRITON2 study. Clin Cancer Res. 2020;26:2487–2496. doi: 10.1158/1078-0432.CCR-20-0394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. de Bono J, Mateo J, Fizazi K, et al. Olaparib for metastatic castration-resistant prostate cancer. N Engl J Med. 2020;382:2091–2102. doi: 10.1056/NEJMoa1911440. [DOI] [PubMed] [Google Scholar]
- 11. Marshall CH, Sokolova AO, McNatty AL, et al. Differential response to olaparib treatment among men with metastatic castration-resistant prostate cancer harboring BRCA1 or BRCA2 versus ATM mutations. Eur Urol. 2019;76:452–458. doi: 10.1016/j.eururo.2019.02.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sztupinszki Z, Diossy M, Krzystanek M, et al. Detection of molecular signatures of homologous recombination deficiency in prostate cancer with or without BRCA1/2 mutations. Clin Cancer Res. 2020;26:2673–2680. doi: 10.1158/1078-0432.CCR-19-2135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Lotan TL, Kaur HB, Salles DC, et al. Homologous recombination deficiency score in germline BRCA2- versus ATM-altered prostate cancer. Mod Pathol. 2021;34:1185–1193. doi: 10.1038/s41379-020-00731-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Chang JCW, Nandakumar S, Stopsack KH, et al. Genomic features and response to poly (ADP-ribose) polymerase inhibition in metastatic castration-resistant prostate cancer. JCO Precis Oncol. doi: 10.1200/PO-25-00934. 10.1200/PO-25-00934 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Safabakhsh P, Tolkunov D, Lai J, et al. HRD status prediction in patients with advanced breast, prostate, ovarian, and pancreatic cancers using a liquid biopsy assay. Presented at: 2025 ASCO Annual Meeting; Chicago, IL. Abstract 3047.
- 16. Jiang T, Wu I, Kim Y, et al. Analytical validation of a robust integrated genomic and epigenomic liquid biopsy for biomarker discovery, therapy selection, and response monitoring. 2023 Presented at: AACR Annual Meeting. Abstract 6601. [Google Scholar]
- 17. Greenwald WY, Gross A, Jiang T, et al. Accurate epigenomic estimates of circulating tumor fraction in large-scale clinical data. 2022 Presented at: AACR Annual Meeting. Abstract 5189. [Google Scholar]
- 18. Gross A, Zhao J, Quinn K, et al. Landscape of epigenomic tumor fraction in large pan-cancer cfDNA cohort. 2025 Presented at: AACR Annual Meeting. Abstract 5936. [Google Scholar]
- 19. Fiaux PC, Cai M, Safabakhsh P, et al. Analytical validation of a new plasma-only bioinformatics classifier to identify variants from clonal hematopoiesis in cfDNA liquid biopsy assays. 2025 Presented at: AACR Annual Meeting. Abstract 6603. [Google Scholar]
- 20. Alexandrov LB, Kim J, Haradhvala NJ, et al. The repertoire of mutational signatures in human cancer. Nature. 2020;578:94–101. doi: 10.1038/s41586-020-1943-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Liang SI, Quandt Z, Wienke S, et al. Methylation-based ctDNA tumor fraction changes predict long-term clinical benefit from immune checkpoint inhibitors in RADIOHEAD, a Real-World Pan-Cancer Study. Cancer Res Commun. 2025;5:1384–1395. doi: 10.1158/2767-9764.CRC-25-0151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Rolfo CD, Khushman MM, Russo A, et al. Chemotherapy response monitoring with DNA methylation-based ctDNA tumor fraction: Evidence from a real-world cohort of patients with advanced common solid malignancies. J Liquid Biopsy. 2025;10:100442. doi: 10.1016/j.jlb.2025.100442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Saad F, Clarke N, Oya M, et al. Olaparib plus abiraterone versus placebo plus abiraterone in metastatic castration-resistant prostate cancer (PROpel): Final prespecified overall survival results of a randomized, double-blind, phase 3 trial. Lancet Oncol. 2023;24:1094–1108. doi: 10.1016/S1470-2045(23)00382-0. [DOI] [PubMed] [Google Scholar]
- 24. Fizazi K, Azad AA, Matsubara N, et al. Talazoparib plus enzalutamide in men with HRR-deficient metastatic castration-resistant prostate cancer: Final overall survival results from the randomized, placebo-controlled, phase 3 TALAPRO-2 trial. Lancet. 2025;406:461–474. doi: 10.1016/S0140-6736(25)00683-X. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
A data sharing statement provided by the authors is available with this article at DOI https://doi.org/10.1200/PO-26-00390.
The data that support the findings of this study are available from Guardant Health, but restrictions apply to the availability of these data, which are not publicly available. Data may be made available upon reasonable request and with permission of the sponsor.