Abstract
Introduction
Liquid biopsies using cell-free DNA (cfDNA) can identify actionable somatic alterations in prostate cancer (PCa). The presence of alterations related to clonal hematopoiesis (CH) can complicate interpretation of cfDNA results and lead to clinical inaction. We sought to determine alteration profiles identified by cfDNA testing in a large, racially diverse cohort of United States Veterans with metastatic PCa and quantify the extent of CH interference in the identification of actionable alterations.
Methods
Next-generation sequencing of cfDNA biopsy specimens from US Veterans with metastatic PCa was performed using Foundation Medicine through the VA National Precision Oncology Program (NPOP) from August 2020 to January 2025. Actionable and non-actionable alteration rates were compared between tests with and without identification of a CH alteration. Tests with CH interference were then stratified by the presence of a PCa-related alteration (TMPRSS2, ERG, SPOP) and rates of actionable and non-actionable alterations were compared between these groups.
Results
cfDNA alterations were identified in 2501/2657 (94.1%) cfDNA tests. CH interference was common, with 663/2657 (25%) tests identifying CH alterations only and 1615/2657 (60.8%) tests identifying CH and non-CH alterations concurrently. Accounting for tests in which no alteration was identified (156/2657; [5.9%]), only 223/2501 (8.4%) tests exhibited non-CH only alterations. Alteration rates in FDA-approved targetable genes were similar in samples with and without CH interference, except for CDK12 variants, which were more frequently identified in tests without CH interference (8.5% vs 3.8%, P < .01) and BRCA1 variants which were more frequently identified in tests with CH interference (2.0% vs 0%, P = .02). In samples with CH interference, alterations in PTEN, RB1, and TP53 were more likely and homologous recombination repair (HRR) alterations were less likely in samples with a PCa specific alteration versus those without (tumor suppressor gene 72% vs 51%, P < .0001; HRR 36% vs 56%, P < .0001).
Conclusions
CH interference was very common in cfDNA testing in metastatic PCa patients. The absence of a CH alteration or presence of a known PCa-related alteration does not reliably modify identification frequencies of actionable alterations. Robust CH inferential platforms or analysis of a matched lymphocyte sample with cfDNA may improve clinical differentiation of CH versus PCa-derived cfDNA results.
Keywords: prostate cancer, cell-free DNA, circulating tumor DNA, clonal hematopoesis, CHIP
Implications for Practice.
Genomic alterations related to clonal hematopoiesis (CH) are frequently encountered when cell-free DNA (cfDNA) sequencing is pursued for individuals with metastatic prostate cancer. This interference of CH can complicate the interpretation of cfDNA testing results, particularly when the presence of actionable alterations are sought. Importantly, the absence of a CH alteration or the presence of a known PCa-related alteration (AR, TMPRSS2, SPOP) does not reliably modify the identification frequencies of actionable alterations. Robust CH inferential platforms or analysis of a matched lymphocyte sample with cfDNA may improve clinical differentiation of CH versus PCa-derived cfDNA results.
Introduction
The assessment of both somatic and germline genomic alterations is critically important for elucidating candidacy for clinical trials and precision-informed targeted therapies1 in individuals with metastatic prostate cancer (PCa), and this practice has become widely endorsed by numerous highly cited clinical guidelines2 as standard of care. Genomic testing for men with metastatic PCa can be accomplished either through next-generation sequencing of the primary tumor in the prostate, metastatic tissue biopsy, or liquid biopsy using plasma circulating cell-free DNA (cfDNA). Liquid biopsies are an especially attractive option when tissue is unavailable or repeat biopsy is not practical or feasible.3 Plasma cfDNA has also been utilized in landmark precision-medicine based clinical trials4,5 as a noninvasive approach for assessing alteration status and monitoring disease response.6
Nevertheless, the settings in which cfDNA testing are most clinically reliable for identifying relevant and actionable PCa alteration profiles in metastatic disease remain to be fully elucidated. This is particularly true in the context of clonal hematopoiesis (CH), which is a process that results in the clonal expansion of hematopoietic stem cells and increases in frequency with normal aging.7 The frequency of CH is approximately 10% for adults in their 70s, but this number rises to approximately 20% for adults over the age of 90.8 Given that CH can be a major contributor to plasma DNA variants even in cancer patients with metastatic solid tumors,9 the presence of CH alterations may obfuscate efforts to reliably understand the somatic alteration landscape (including actionable alterations) in metastatic PCa—a disease where age at development of metastatic disease (and thus the risk of CH10) skews higher than other malignancies.
The true prevalence of CH as assessed by liquid cfDNA biopsy in metastatic PCa patients is unknown, particularly in diverse cohorts of patients. Therefore, we sought to evaluate and quantify CH interference in cfDNA liquid biopsy testing in men with metastatic PCa in a large, racially, and ethnically diverse real-world cohort of US Veterans. We subsequently evaluated the frequency of actionable alterations depending on the presence of CH in a cfDNA test result and explore how the presence of known PCa alterations influences the identification of actionable alterations in the context of CH.
Methods
Next-generation sequencing of cfDNA liquid biopsy specimens from Veterans with metastatic PCa was performed using the Foundation Medicine platform through the VA National Precision Oncology Program (NPOP) from August 28, 2020 to January 22, 2025. Likely pathogenic/pathogenic alterations with a variant allele fraction (VAF) of at least 2% from each test were extracted from clinical reports and stored on NPOP dashboards; tests classified as “Pass” or “Qualified” were included. For patients who underwent multiple cfDNA NGS tests, the most recent test result was utilized. Clinical and demographic information associated with patients who underwent cfDNA testing were linked to NPOP sequencing data via patient Integration Control Number, as previously described.11,12
CH genes were defined as the 74 genes identified in 4229 individuals with CH by Bick et al.13 Among these 74 genes, 44 are included on the Foundation Medicine Liquid panel (Table S1). All liquid biopsy test results were categorized based on whether an alteration was detected or not, and among tests with a detectable alteration, tests were classified as having non-CH alterations only, CH alterations only, or Non-CH+CH alterations. Demographic information for patients in each test category was extracted and summarized descriptively with comparisons between groups assessed via Kruskal-Wallis (median age) or two-way ANOVA (race and ethnicity).
Frequencies of CH alterations were evaluated and compared between tests with CH alterations only and tests with Non-CH+CH alterations using Fisher’s exact test. To then explore the impact of CH interference on the identification of actionable alterations, frequencies of alterations in individual genes as well as in functional gene groupings (where one or more alterations in at least one gene within the group had to be present) were compared between Non-CH only test results and Non-CH+CH test results using Fisher’s exact testing. After limiting tests to those reporting Non-CH+CH alterations alone, the impact of co-identification of a known PCa alteration (AR, TMPRSS2, SPOP) was evaluated for both actionable and non-actionable gene groupings and rates compared using Fisher’s exact testing. Due to the scale, retrospective design, and deidentified nature of the nongenomic data elements, informed consent was waived. This was a substudy of the IRB approved RESOLVE project (IRB 1729212).
Results
The landscape of liquid biopsy alterations
There were 2657 cfDNA tests identified through NPOP that were related to a diagnosis of metastatic PCa and associated with patient clinicopathologic information. An alteration was detected in 2501/2657 (94.1%) tests. Among these tests, 663 (24.9%) identified CH alterations only, 1615 (60.8%) identified Non-CH+CH alterations, and 223 (8.4%) identified Non-CH alterations only (Figure 1, Table S2), demonstrating that CH alterations commonly co-occur with non-CH alterations in metastatic PCa.
Figure 1.
cfDNA tests analyzed in this study. cfDNA testing of 2657 patients with metastatic PCa was analyzed. Tests were categorized by the presence of absence of any alteration(s), as well as whether the alteration(s) identified were CH-only, CH plus non-CH, or non-CH only. Abbreviations: cfDNA, cell-free DNA; CH, clonal hematopoiesis; PCa, prostate cancer.
Patient and disease characteristics associated within the entire cfDNA testing cohort as well as each alteration result category are presented in Table 1. Median (standard deviation) age at the time of cfDNA testing was 76 (10) years. Tests (8.6% [229/2657]) were carried out in Hispanic or Latino men; 60.1% (1612/2657) and 32.6% (866/2657) of tests were carried out in men self-identified as White and Black, respectively. There were no significant differences in the distribution of self-identified race or ethnicity in tests stratified by CH and non-CH alterations.
Table 1.
Patient and disease characteristics.
| All tests n = 2657 | Negative, n = 156 | CH only, n = 663 | Non-CH only, n = 223 | CH+non-CH, n = 1615 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Median(SD) age at testing | 75.82 | (9.51) | 72.87 | (9.44) | 75.46 | (9.17) | 74.38 | (7.24) | 76.44 | (10.01) |
| Ethnicity | n | % | n | % | n | % | n | % | n | % |
| Non-Hispanic or Latino | 2345 | 88.3% | 140 | 89.7% | 566 | 85.4% | 197 | 88.3% | 1442 | 89.3% |
| Hispanic or Latino | 229 | 8.6% | 11 | 7.1% | 77 | 11.6% | 21 | 9.4% | 120 | 7.4% |
| Unknown | 83 | 3.1% | 5 | 3.2% | 20 | 3.0% | 5 | 2.2% | 53 | 3.3% |
| Self-identified race | ||||||||||
| White | 1612 | 60.7% | 81 | 51.9% | 373 | 56.3% | 122 | 54.7% | 1036 | 64.1% |
| Black or African American | 866 | 32.6% | 61 | 39.1% | 245 | 37.0% | 84 | 37.7% | 476 | 29.5% |
| Native Hawaiian/Pacific Islander | 16 | 0.6% | 1 | 0.6% | 7 | 1.1% | 0 | 0.0% | 8 | 0.5% |
| American Indian/Alaska Native | 16 | 0.6% | 0 | 0.0% | 4 | 0.6% | 2 | 0.9% | 10 | 0.6% |
| Asian | 5 | 0.2% | 0 | 0.0% | 1 | 0.2% | 1 | 0.4% | 3 | 0.2% |
| Mixed | 18 | 0.7% | 4 | 2.6% | 4 | 0.6% | 1 | 0.4% | 9 | 0.6% |
| Unknown | 124 | 4.7% | 9 | 5.8% | 29 | 4.4% | 13 | 5.8% | 73 | 4.5% |
Abbreviations: CH, clonal hematopoiesis; IQR, interquartile range.
Interference of CH in liquid biopsy testing in metastatic PCa patients
Twenty-eight genes were altered at over 2% frequency in cfDNA samples from metastatic PCa patients. The most commonly altered genes in cfDNA testing were TP53 and DNMT3A, both found at significantly higher rates in cfDNA than in a public dataset of PCa tumors (41.0% vs 29.9%, P < .0001 and 37.6% vs 1.0%, P < .0001, respectively) (Table S1). Alteration rates in all CH genes found at >2% in cfDNA samples were found at significantly higher frequencies in cfDNA compared to PCa tumors with the exception of PTEN (Table S1). Among PCa-specific driver genes, TMPRSS2 and SPOP were found less commonly in cfDNA compared to PCa tumors (8.7% vs 26.0%, P < .0001 and 5.1% vs 10.0%, P < .0001, respectively); whereas AR alterations were more common (23.1% vs 14.5%, P < .0001) in cfDNA compared to PCa tumors. The most frequently identified FDA-targetable alterations were in ATM, CHEK2, BRCA2, and CDK12. ATM and CHEK2 alterations were identified at higher rates in cfDNA compared to PCa tumors (13.7% vs 5.7%, P < .0001 and 11.3% vs 1.1%, P < .0001, respectively); whereas BRCA2 and CDK12 alterations were less common in cfDNA compared to PCa tumors (4.7% vs 7.2%, P < .0001 and 3.0%, vs 4.9%, P < .0001) (Table S1). Stratifying tests with CH alterations by presence or absence of non-CH alterations, the CH genes TP53, PTEN, RAD21, and KDM6A were altered significantly more commonly with non-CH alterations than without non-CH alterations (53.0% vs 35.3%, P < .0001, 17.0% vs 3.2%, P < .0001, 8.0% vs 3.2%, P < .0001, and 3.5% vs 1.2%, P = .002, respectively) (Table 2). By contrast, DNMT3A was the only CH alteration found more commonly in tests with CH alterations only (48.7% vs 41.9%, P = .003) (Table 2).
Table 2.
Rates of common CH alterations in tests with or without non-CH alterations.
| CH only, n = 663 |
Non-CH+CH, n = 1615 |
P-value | |||
|---|---|---|---|---|---|
| CH gene | n | % | n | % | |
| TP53 | 234 | 35.3% | 856 | 53.0% | <0.0001 |
| DNMT3A | 323 | 48.7% | 677 | 41.9% | 0.0034 |
| TET2 | 177 | 26.7% | 387 | 24.0% | 0.1816 |
| ASXL1 | 119 | 17.9% | 310 | 19.2% | 0.5166 |
| SF3B1 | 68 | 10.3% | 173 | 10.7% | 0.8220 |
| PTEN | 21 | 3.2% | 275 | 17.0% | <0.0001 |
| RAD21 | 21 | 3.2% | 129 | 8.0% | <0.0001 |
| KDM6A | 8 | 1.2% | 56 | 3.5% | 0.0020 |
| NF1 | 54 | 8.1% | 125 | 7.7% | 0.7324 |
| CBL | 25 | 3.8% | 61 | 3.8% | 1.0000 |
| BRAF | 16 | 2.4% | 66 | 4.1% | 0.0624 |
| U2AF1 | 19 | 2.9% | 55 | 3.4% | 0.6030 |
| GNAS | 20 | 3.0% | 47 | 2.9% | 0.8918 |
| JAK2 | 16 | 2.4% | 49 | 3.0% | 0.4894 |
CH only are tests with only one or more CH alteration(s) identified, non-CH+CH are tests with at ≥1 non-CH and ≥1 CH alteration identified.
Abbreviations CH, clonal hematopoiesis.
Associations with CH and the identification of actionable alterations
We next compared the rates of alterations in non-CH genes in functional groupings and individually in samples with (n = 1615) and without (n = 223) CH. Alterations in PCa-specific driver genes were found more commonly in samples with CH interference than without (45.7% vs 33.6%, P = .0007), driven by higher TMPRSS2 alteration rates (15.4% vs 4.5%, P < .0001) (Figure 2A, Table S3). Alteration rates of FDA approved targetable homologous recombination repair (HRR) genes overall were similar in samples with and without CH interference (51.6% vs 46.4%, P = .15), although alteration rates in BRCA1 and BRCA2 were nominally higher in samples with CH interference (2.0% vs 0.0%, P = .02 and 7.1% vs 4.9%, P = .26) (Table S3). CDK12 alterations were less frequently identified in tests with versus without CH interference (3.8% vs 8.5%, P = .004) (Figure 2A, Table S3). Rates of alterations in putatively targetable genes were higher in samples with versus without CH interference (16.8% vs 8.5%, P = .001) (Figure 2A, Table S3). A significantly higher percentage of samples with versus without CH interference had a high or intermediate tumor mutational burden (17.2% vs 7.0%, P < .0001), although a similar rate of MSI-High status was found with or without CH interference (4.8% vs 2.2%, P = .41) (Table S4).
Figure 2.
Non-CH alteration rates stratified by CH alterations and PCa specific gene alterations. (A) Rates of alterations in functional gene groupings and individual genes for tests with non-CH alterations stratified by the presence of CH interference. (B) Rates of alterations in functional gene groupings and individual genes for tests with non-CH alterations and CH interference stratified by the presence of PCa-specific driver alterations. Abbreviations: CH, clonal hematopoiesis; PCa, prostate cancer.
Associations with CH and the identification of actionable alterations stratified by PCa-specific alterations
The 1615 tests with non-CH alterations and CH interference (non-CH+CH) were subdivided into tests that simultaneously reported alterations specific to PCa (n = 738) compared to those where no concurrent PCa-specific alteration was present (n = 877). When comparing alteration frequencies between tests with and without PCa alterations, the likelihood of identifying alterations in the tumor suppressor genes (TSGs) TP53, PTEN, and RB1 and other targetable genes was higher if a PCa-related alteration was identified (71.8% vs 51.3%, P < .0001 and 20.7% vs 13.5%, P = .0001) (Figure 2B, Table S5). However, the likelihood of identifying an alteration in any targetable HRR gene was significantly lower in tests where a PCa-specific alteration was also identified (35.6% vs 55.5%, P < .0001), driven by lower rates of ATM and CHEK2 alterations in tests with a PCa-specific alteration compared to those without (14.5% vs 23.0%, P < .0001 and 10.8% vs 20.5%, P < .0001, respectively) (Figure 2B, Table S5). Alterations in BRCA1 and BRCA2 were similar in samples with or without a PCa-specific alteration (7.5% vs 10.1%, P = .07) (Figure 2B, Table S5).
Discussion
We analyzed cfDNA liquid biopsy testing in a racially and ethnically diverse cohort of 2657 patients with metastatic PCa. CH interference was very common, with 85.8% of tests identifying at least one CH alteration, with (60.8%) or without (25%) a non-CH alteration, and only 8.4% of tests identifying non-CH alterations only. In tests with non-CH alterations, CH interference did not affect the frequency of identification of US Food and Drug Association (FDA)-targetable Homologous recombination repair (HRR) or mismatch repair (MMR) gene alterations, except for CDK12 which was found more commonly without CH interference. The identification of a PCa-specific alteration in a test with CH interference was associated with a higher likelihood of identifying TSG and putative targetable alterations, although did not increase the likelihood of finding an FDA-approved actionable HRR or MMR alteration.
This work establishes that CH alterations are frequently identified in men with metastatic PCa undergoing liquid biopsy and commonly co-occur with non-CH alterations, extending prior studies in tumor tissue NGS.14 Moreover, many of the most frequently identified CH alterations (TP53, PTEN, RAD21, and KDM6A) were more commonly identified with non-CH alterations than with CH alterations, further illustrating the depth of CH interference in this real-world dataset and suggesting an extensive bystander effect of CH in a metastatic PCa population of median age 76. The notably high detection of CH-related alterations may be related to the unique military exposures and characteristics of the Veteran population.15 It may also suggest a link between CH and PCa, akin to the striking association between CH and cardiovascular disease risk,16 which merits further exploration. This exploration is especially critical in diverse cohorts of patients, since prior work evaluating associations between CH and PCa have been carried out largely in European cohorts.17
Importantly, the presence of CH alterations in a given test result should not invalidate the simultaneous identification of PCa-related actionable alterations. Actionable PCa alterations were frequently identified alongside CH alterations in approximately the same proportions as tests that did not uncover any CH alterations, so disregard of a test due to the identification of CH could lead to under-identification of true actionable alterations. HRR alterations were the most commonly identified targetable alterations overall (32.5%, 865/2657), and were found in a similar fraction of samples with and without CH interference. These rates are higher than those previously reported in a large cohort of racially diverse Veterans with metastatic PCa tissue sequenced through NPOP (11%-13%).12 Sequencing of tissue has previously been reported to result in lower frequencies of DDR alterations compared to plasma cfDNA sequencing,18 likely due to increased rate of ATM and CHEK2 alterations arising from CH versus PCa in cfDNA testing. Our overall rates of actionable alterations using cfDNA testing are on par with a pan-cancer analysis19 with a much smaller sample size and notably less racial and ethnic diversity.
In a retrospective review of 69 patients with advanced metastatic PCa, Jensen et al. also found that CH involving DNA repair genes can interfere with cfDNA testing for Pca.20 In that analysis, CH interference was detected in 19% of patients tested which included CH variants in DNA repair genes that determine PARP inhibitor candidacy, such as ATM, in 7% of patients tested. It is important to note that the ATM alterations in both their and our analysis could also have been due to patients with concurrent chronic lymphocytic leukemia rather than CH. Regardless of etiology, these analyses emphasize the critical importance of matched lymphocyte testing to differentiate hematological malignancy or CH clones from true PCa-derived circulating tumor DNA.
There are limitations of this analysis. In our cohort, the timing cfDNA testing was variable, with some men tested shortly after diagnosis of either de novo or metachronous metastatic disease, while others tested later in their metastatic disease trajectory. As a result, our population of Veterans with metastatic PCa was heterogeneous. Moreover, clinical features defining our metastatic PCa population, including castrate resistant status, burden of metastatic disease (M1a vs M1b vs M1c), and receipt of prior systemic or local therapies (ADT, cytotoxic chemotherapy, or radiation therapy), many of which are thought to influence cfDNA findings,21 were not available for our study cohort and thus could not be accounted or adjusted for in our analysis. We also acknowledge that designation of what is a CH gene is controversial and debated. While we utilized an established and well-cited list of genes,13 different definitions of CH may lead to different degrees of overlap between CH alterations and PCa-related alterations or actionable alterations. Ultimately, regardless of the CH definition used, matched normal testing is the only way to truly identify whether a cfDNA test result is related to CH or not. More sophisticated inferential CH platforms that leverage fragmentomics22 and tumor fraction,23 while still under evaluation, may also represent a valuable tool for interpreting cfDNA testing results and determining the true burden of CH in metastatic PCa patients of diverse ancestry.
While cfDNA testing does identify PCa-related alterations in metastatic PCa patients, the high frequency of CH interference may complicate the identification of true PCa-related actionable alterations. In particular, the actionability of alterations in genes with roles in both PCa and CH may be more easily discernable with inferential algorithms or, a matched lymphocyte control, the latter which enables direct assignment of an alteration to CH or tumor-derived. Finally, further work is needed to determine the optimal clinical scenario to identify actionable alterations by cfDNA testing in metastatic PCa patients.
Supplementary Material
Contributor Information
Luca F Valle, Radiation Oncology Service, Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA, 90025, USA; Department of Radiation Oncology, David Geffen School of Medicine at the University of California Los Angeles, Los Angeles, CA, 90095, USA; Jonsson Comprehensive Cancer Center, University of California Los Angeles, Los Angeles, CA, 90095, USA.
Micaela Scobie, National Oncology Program, Department of Veterans Affairs, Washington DC, 20420, USA.
Kerry Rowe, National Oncology Program, Department of Veterans Affairs, Washington DC, 20420, USA.
Colin C Pritchard, Department of Laboratory Medicine, University of Washington, Seattle, WA, 98195, USA; Brotman Baty Institute for Precision Medicine, University of Washington, Seattle, WA, 98195, USA.
Bruce Montgomery, Medical Oncology Service, VA Puget Sound Health Care System, Seattle, WA, 98108, USA; Division of Oncology, Department of Medicine, University of Washington, Seattle, WA, 98195, USA.
Isla P Garraway, Jonsson Comprehensive Cancer Center, University of California Los Angeles, Los Angeles, CA, 90095, USA; Urology Service, Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA, 90025, USA; Department of Urology, David Geffen School of Medicine at the University of California Los Angeles, Los Angeles, CA, 98195, USA.
Nicholas G Nickols, Radiation Oncology Service, Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA, 90025, USA; Department of Radiation Oncology, David Geffen School of Medicine at the University of California Los Angeles, Los Angeles, CA, 90095, USA; Jonsson Comprehensive Cancer Center, University of California Los Angeles, Los Angeles, CA, 90095, USA.
Martin W Schoen, Medical Oncology Service, Saint Louis VA Medical Center, St. Louis, MO, 63106, USA; Division of Hematology Oncology, Department of Medicine, St. Louis University, St. Louis, MO, 63103, USA.
Amar U Kishan, Department of Radiation Oncology, David Geffen School of Medicine at the University of California Los Angeles, Los Angeles, CA, 90095, USA; Jonsson Comprehensive Cancer Center, University of California Los Angeles, Los Angeles, CA, 90095, USA.
Michael J Kelley, Medical Oncology Service, Durham Veterans Affairs Health Care System, Durham, NC, 27705, USA; Division of Medical Oncology, Department of Medicine, Duke Cancer Institute, Duke University, Durham, NC, 27708, USA.
Matthew B Rettig, Medical Oncology Service, Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA, 90025, USA; Department of Medicine-Hematology/Oncology, David Geffen School of Medicine at the University of California Los Angeles, Los Angeles, CA 90095, USA.
Kara N Maxwell, Medical Oncology Service, Corporal Michael J. Crescenz Veterans Affairs Medical Center, Philadelphia, PA, 19104, USA; Division of Hematology/Oncology, Department of Medicine, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, 19104, USA.
Author contributions
Luca Valle (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review & editing), Micaela Scobie (Formal analysis, Writing—review & editing), Kerry Rowe (Resources, Writing—review & editing), Colin C. Pritchard (Writing—review & editing), Bruce Montgomery (Writing—review & editing), Isla P. Garraway (Resources, Supervision, Writing—review & editing), Nicholas G. Nickols (Investigation, Resources, Supervision, Writing—review & editing), Martin W. Schoen (Investigation, Writing—review & editing), Amar U. Kishan (Supervision, Writing—review & editing), Michael J. Kelley (Resources, Supervision, Writing—review & editing), Matthew B. Rettig (Conceptualization, Data curation, Formal analysis, Investigation, Project administration, Resources, Supervision, Writing—original draft, Writing—review & editing), and Kara N. Maxwell (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Writing—original draft, Writing—review & editing)
Supplementary material
Supplementary material is available at The Oncologist online.
Funding
Luca Valle receives salary and research support through the Robert A. Winn Career Development Award, sponsored by the Bristol Myers Squibb Foundation as well as the UCLA SPORE in Prostate Cancer (P50 CA092131). Kara Maxwell receives research funding through the Prostate Cancer Foundation (22CHAL02), the Basser Center for BRCA, the Burroughs Wellcome Foundation (1017184), and Veterans Affairs Office of Research and Development (1IO1CX002709, 1I01CX002710). Isla Garraway receives research funding through the National Institutes of Health (R01PAR-20-077 and 5P50CA09213), the Prostate Cancer Foundation 22CHAL02), the Department of Defense (W81XWH211075), Veterans Affairs Office of Research and Development (1I01CX002710) and the Jean Perkins Foundation. Matthew Rettig receives research support through the Prostate Cancer Foundation David Geffen Center of Excellence Award.
Conflicts of interest
None of the authors report any potential conflicts of interest.
Data availability
It is not possible for the authors to directly share the individual-level data that were obtained from the patients included in this study due to constraints stipulated in the informed consent. Anyone wishing to gain access to this data should inquire directly to the authors. The data generated from our analyses are included in the manuscript main text, tables, figures, and online supplementary material.
References
- 1. Merseburger AS, Waldron N, Ribal MJ, et al. Genomic testing in patients with metastatic castration-resistant prostate cancer: a pragmatic guide for clinicians. Eur Urol. 2021;79:519-529. 10.1016/j.eururo.2020.12.039 [DOI] [PubMed] [Google Scholar]
- 2.Spratt DE, Srinivas S, Adra N, et al. Prostate Cancer, Version 3.2026, NCCN Clinical Practice Guidelines In Oncology. 2025:23:469-–493.. 10.6004/jnccn.2025.0052 [DOI] [PubMed] [Google Scholar]
- 3. Kwan EM, Wyatt AW, Chi KN. Towards clinical implementation of circulating tumor DNA in metastatic prostate cancer: opportunities for integration and pitfalls to interpretation. Front Oncol. 2022;12:1054497. 10.3389/fonc.2022.1054497 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Fizazi K, Piulats JM, Reaume MN, et al. TRITON3 Investigators. Rucaparib or physician’s choice in metastatic prostate cancer. N Engl J Med. 2023;388:719-732. 10.1056/NEJMoa2214676 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Deek MP, Van der Eecken K, Sutera P, et al. Long-term outcomes and genetic predictors of response to metastasis-directed therapy versus observation in oligometastatic prostate cancer: analysis of STOMP and ORIOLE trials. J Clin Oncol. 2022;40:3377-3382. 10.1200/JCO.22.00644 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ritch E, Wyatt AW. Predicting therapy response and resistance in metastatic prostate cancer with circulating tumor DNA. Urol Oncol. 2018;36:380-384. 10.1016/j.urolonc.2017.11.017 [DOI] [PubMed] [Google Scholar]
- 7. Marnell CS, Bick A, Natarajan P. Clonal hematopoiesis of indeterminate potential (CHIP): linking somatic mutations, hematopoiesis, chronic inflammation and cardiovascular disease. J Mol Cell Cardiol. 2021;161:98-105. 10.1016/j.yjmcc.2021.07.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Jaiswal S, Fontanillas P, Flannick J, et al. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371:2488-2498. 10.1056/NEJMoa1408617 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Razavi P, Li BT, Brown DN, et al. High-intensity sequencing reveals the sources of plasma circulating cell-free DNA variants. Nat Med. 2019;25:1928-1937. 10.1038/s41591-019-0652-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Miller PG, Steensma DP. Implications of clonal hematopoiesis for precision oncology. JCO Precis Oncol. 2020;4:639-646. 10.1200/PO.20.00144 [DOI] [PubMed] [Google Scholar]
- 11. Price MJ, Vashistha V, Winski D, et al. Homologous recombination repair gene variants and outcomes among patients with prostate cancer treated with poly (ADP-ribose) polymerase inhibitors. JCO Precis Oncol. 2022;6:e2100461. 10.1200/PO.21.00461 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Valle LF, Nickols NG, Hausler R, et al. Actionable genomic alterations in prostate cancer among black and white United States veterans. Oncologist. 2023;28:e473-e477. 10.1093/oncolo/oyad042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Bick AG, Weinstock JS, Nandakumar SK, et al. NHLBI Trans-Omics for Precision Medicine Consortium. Inherited causes of clonal haematopoiesis in 97,691 whole genomes. Nature. 2020;586:763-768. 10.1038/s41586-020-2819-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Iranmanesh Y, Omar N, Rasouli A, et al. Inferred clonal hematopoiesis from tumor DNA sequencing among men with prostate cancer: correlation with somatic tumor alterations and outcomes. Oncologist. 2025;30:oyaf049. 10.1093/oncolo/oyaf049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.DeStefano CB, Wilkerson MD, Villar J, et al. Clonal hematopoiesis in Gulf war Veterans. Am J Hematol. 2025;100. 10.1002/ajh.27680 [DOI] [PubMed] [Google Scholar]
- 16.Jaiswal S, Natarajan P, Silver AJ, et al. Clonal hematopoiesis and risk of atherosclerotic cardiovascular disease. N Engl J Med. 2017;377. 10.1056/NEJMoa1701719 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Wang A, Xu Y, Yu Y, et al. Clonal hematopoiesis and risk of prostate cancer in large samples of European ancestry men. Hum Mol Genet. 2023;32:489-495. 10.1093/hmg/ddac214 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Valle LF, Li J, Desai H, et al. Somatic tumor next-generation sequencing in US veterans with metastatic prostate cancer. JAMA Netw Open. 2025;8:e259119. 10.1001/jamanetworkopen.2025.9119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sánchez NS, Kahle MP, Bailey AM, et al. Identification of actionable genomic alterations using circulating cell-free DNA. JCO Precis Oncol. 2019;3:PO.19.00017. 10.1200/PO.19.00017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Jensen K, Konnick EQ, Schweizer MT, et al. Association of clonal hematopoiesis in DNA repair genes with prostate cancer plasma cell-free DNA testing interference. JAMA Oncol. 2021;7:107-110. 10.1001/jamaoncol.2020.5161 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Annala M, Vandekerkhove G, Khalaf D, et al. Circulating tumor DNA genomics correlate with resistance to abiraterone and enzalutamide in prostate cancer. Cancer Discov. 2018;8:444-457. 10.1158/2159-8290.CD-17-0937 [DOI] [PubMed] [Google Scholar]
- 22. Brown DW, Sun D, Fine AD, et al. Fragmentomic-based algorithm to computationally predict tumor-somatic, germline, and clonal hematopoiesis variant origin in liquid biopsy. J Liq Biopsy. 2025;9:100311. 10.1016/j.jlb.2025.100311 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Husain H, Pavlick DC, Fendler BJ, et al. Tumor fraction correlates with detection of actionable variants across > 23,000 circulating tumor DNA samples. JCO Precis Oncol. 2022; 6:e2200261. 10.1200/PO.22.00261 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
It is not possible for the authors to directly share the individual-level data that were obtained from the patients included in this study due to constraints stipulated in the informed consent. Anyone wishing to gain access to this data should inquire directly to the authors. The data generated from our analyses are included in the manuscript main text, tables, figures, and online supplementary material.