Predictive molecular alterations of prostate cancer brain metastases based on a companion diagnostic assay

Abstract

The current first-line treatment standard for patients with metastatic prostate cancer (mPCa), is a combination of androgen deprivation therapy (ADT) and androgen receptor pathway inhibitors (ARPI), plus the addition of docetaxel for fit patients with high-volume or high-risk disease. This upfront treatment is highly effective, however, the emergence of acquired resistance is nearly universal, upon which, molecular predictive biomarkers are needed for further targeted therapies. Genomic studies have identified a population of 20–30% prostate cancer (PCa) patients harboring tumor DNA damage repair (DDR) alterations. Recent clinical trials have shown that treatment with PARP inhibitors (PARPi) achieve high response rates in tumors with BRCA1- and BRCA2-pathogenic alterations. Patients with metastatic castration-resistant PCa (mCRPC) harboring these alterations showed the highest overall survival benefit. Prostate cancer brain metastases (PCBM) is a usual exclusion criterion for clinical trials in part due to expected poor outcome. Recent work from our group in Switzerland has shown that 19.6% of patients with prostate cancer brain metastases (PCBM) had genomic alterations in homologous recombination repair (HRR) genes. However, the study was done using a research genomic assay. To maximize clinical relevance, in this work we analyzed 46 men suffering from PCBM from the same cohort using the FDA approved companion diagnostic FoundationOne®CDx assay. We found that 12/44 patients (27.3%), from which a metastatic sample was available, harbored qualifying alterations for PARPi therapy. Additionally, we found 7/44 patients (15.9%) qualifying for immune check-point inhibitors therapy. We anticipate that these findings will improve the rate of molecular testing in mCRPC patients with brain metastases and advance personalized management of these patients.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-04150-2.

Introduction

Men with locally advanced prostate cancer (PCa) are almost all initially treated with androgen deprivation therapy (ADT). Second-generation potent androgen receptor signalling inhibitors (ARSi), such as abiraterone and enzalutamide, are used when their disease progresses. An important emerging clinical opportunity in metastatic PCa therapy is to target the frequently observed homologous recombination repair (HRR) defects with PARP inhibitors (PARPi). Despite the different cohort composition and which genes were included, HRR defects were observed in 10–40% of the cases in prior studies [1–6]. In metastatic castration-resistant prostate cancer (mCRPC), the most commonly altered HRR genes are BRCA2 and ATM [5, 6]. As suggested by the PROfound and TOPARP-B trials’ results [4, 7], PARPi for patients harbouring BRCA1/2 and ATM alterations, and potentially alterations in further HRR genes, is an effective therapy. The recently published PROfound phase 3 trial demonstrated prolonged overall survival (18.5 vs. 15.1 months, hazard ratio for death 0.64; P = 0.02) in the cohort of patients with mCRPC harbouring BRCA1, BRCA2 or ATM alterations when treated with the PARPi olaparib after progression on enzalutamide or abiraterone [7]. This was the first successful PCa trial using biomarker selection. The assay used for this trial was FoundationOne®CDx. Based on these and other studies, the FDA approved PARPi therapy for mCRPC on May 22, 2020.

Due to the development of novel and more efficient therapeutic options for treating PCa, a remarkable improvement of the average patient survival has been noted with an extension to 40 months compared to 15 months with earlier therapies [8]. However, the extended survival has favored the metastatic spread to otherwise atypical locations such as the brain. Nevertheless, patients with prostate cancer brain metastases (PCBM) are usually excluded from clinical trials, including the previously mentioned PROfound study. Therefore, patients presenting PCBM, even when their tumors harbor qualifying alterations for trial inclusion, do not get access to targeted therapies from which they might benefit.

In our work recently published in Nature Communications [9], we collected samples from 51 patients suffering from PCBM with non-synchronous matched primary tumors for 20 cases to address the frequency of homologous repair deficiency (HRD) in PCBM. We found a significantly greater representation of the HRD signature in PCBM compared to the Stand Up to Cancer non-brain metastatic castration-resistant prostate cancer cohort [1] (CRPC500). Compared to CRPC500, the PCBM cohort was enriched in mutations affecting the subset of 15 HRR genes investigated for patient inclusion in the PROfound clinical trial [9]. We demonstrated that PCBM has genomic dependencies that may be exploitable through clinical interventions, including PARP inhibition. However, for the molecular definition of the PCBM cohort, we did not use a companion diagnostic test which is FDA approved and employed in routine diagnostics in a clinical setting.

To close this knowledge gap, in the current study we analyzed a subset (N = 44 patients with PCBM) of the PCBM cohort [9] using the FoundationOne®CDx assay. The same assay was also used in the PROfound study. FoundationOne®CDx assay is the first FDA-approved tissue-based broad companion diagnostic (CDx) that is clinically and analytically validated for all solid tumors. Thus, the obtained data in the current study represent a real-case scenario of the molecular routine diagnostic conducted in world-wide health care systems for patients presenting with oncological diseases.

Patients and methods

Patient selection and tumor procurement

Tumor samples were collected from Pathology Departments in university and Cantonal Hospitals across Switzerland (Institute of Pathology, Bern/ Institute of Neuropathology, Zurich/ Institute of Medical Genetics and Pathology, Basel/ Institute of Pathology, Aarau/ Institute of pathology, Münsterlingen/ Institute of Pathology, Liestal/ Institute of Pathology, St. Gallen) and from the Department of Pathology and Laboratory Medicine, and Urology, Emory University School of Medicine, Atlanta, USA [9]. Inclusion criteria were defined as patients having available formalin-fixed paraffin-embedded (FFPE) blocks from confirmed central nervous system (CNS) or meningeal metastases of prostate carcinoma (from now CNS/meningeal metastases referred as PCBM). If available, tissue from the matched primary tumour and normal tissue was also collected. Additionally, matched non-PCBM metastases (i.e. liver, lung, spleen, lymph-node metastases) from 4 patients (P43, P45, P46, P49) were included for comparison reasons. Non-PCBM metastases of P45 could not be sequenced due DNA degradation. All analyses followed protocols approved by the Ethical Committee Bern (Project ID: 2019–00328). No participant compensation was applied for the current study.

Study population

Our cohort includes samples from 48 patients. Patients qualified for inclusion in this study if written consent or no documented refusal was available (Human Research Act, HRA, Swiss Confederation; Art. 34). More detailed information can be found under the “Declarations” section below. We collected archived FFPE tissue from CNS (brain/spinal cord) and meningeal metastases with matched primary tumors in 16 cases and matched non-PCBM metastases (i.e. liver, lung, spleen, lymph-node metastases) from 4 patients (P43, P45, P46, P49). Most tumor samples corresponded to diagnostic biopsies (from prostate or PCBM), transurethral resections (TURP), or prostatectomy specimens. Primary tumors and metastases from patients P1, P32, P43-46, P48, and P49 were taken from autopsy tissue.

DNA extraction and FoundationOne®CDx assay

During the histological review up to three regions of interest (ROIs) from primary tumors and metastases were selected. FFPE core biopsies (1 mm diameter) were taken from the ROIs using the TMA Grand Master (3DHISTECH). After deparaffinization, DNA was extracted using the QIAamp DNA micro kit (Qiagen). Quality and quantity were determined by Qubit dsDNA BR kit and Nanodrop. Up to 300 ng of DNA were sent to the Foundation Medicine laboratory at the Department of Pathology and Molecular Pathology, University Hospital Zurich.

Samples were sequenced using FoundationOne®CDx genomic assay, as previously described [10]. The FoundationOne®CDx assay detects substitutions, insertion and deletion alterations (indels), and copy number alterations (CNAs) in 324 genes and select gene rearrangements, as well as genomic signatures including microsatellite instability (MSI), tumor mutational burden (TMB), and homologous recombination signature (HRDsig) (https://www.foPMIDundationmedicine.com/test/foundationone-cdx).

The analysis of samples used in this study was authorized by the Ethical Committee Bern (2021-01935).

Data analysis

FoundationOne®CDx results were analyzed in R v.4 using basic functions. Oncoplots were generated using the maftools R package (v.2.22.0). Heatmaps were generated using the ComplexHeatmap R package (v.2.22.0).

For defining the HRR status of the tumors, we focused on 15 genes related to the HRR-pathway, as defined in the PROfound study (i.e. BRCA1, BRCA2, ATM, BRIP1, BARD1, CDK12, CHEK1, CHEK2, FANCL, PALB2, PPP2R2A, RAD51B, RAD51C, RAD51D, and RAD54L). Further, for assessing the oncogenicity of the alterations found within the previously mentioned genes, we also applied the inclusion criteria used in the PROfound clinical trial. An alteration was defined as deleterious when resulting in protein truncation (including nonsense, frameshift, or consensus splice site alterations), large-scale alterations, such as genomic truncating rearrangements or homozygous deletions, or missense alterations with the “Functional Impact” assigned as “known” or “likely” in the FoundationOne®CDx report [7].

Results

Demographic and clinical data

The FoundationOne®CDx assay was used on 208 samples and was successful for 170 (81%), comprising 57 samples from primary prostate cancer and 113 samples from metastases (Table S1.A). Metastatic tissue was collected predominantly from PCBM (N = 97) but also from liver (N = 5), lung (N = 5), spleen (N = 3) and lymph node (N = 3). Successfully sequenced samples were derived from 46 patients, of which 16 had paired primary tumor and metastatic tissues, 28 had only metastatic tissue, and two had only primary tumor (Table S1.B). Brain metastasis tissue was available from 44 patients. Poor DNA quality and/or quantity was likely the reason for the failed samples [11] (Table S1).

The average age at diagnosis of PCBM was 70.4 years old. At this time point, 25/46 patients (54%) presented with multiple PCBM and 41/46 (89%) patients suffered from additional metastases in other organs. Therapeutically, 36/46 patients (78%) were treated with androgen deprivation therapy (ADT) or orchiectomy. Additionally, 9/46 (20%) and 2/46 (4%) had received Abiraterone or Enzalutamide, respectively. Some clinical data from some patients were not appropriately documented in the health records (Table S2).

Histomorphology

Histologically most of the metastases presented as acinar adenocarcinoma (N = 42) and two tumors showed overlapping features of small cell and acinar adenocarcinoma. One metastasis harbored a morphological and immunohistochemical spectrum. from acinar adenocarcinoma, through areas with overlapping features between adenocarcinoma/small cell up to bona fide small cell neuroendocrine carcinoma. The morphology of all metastases was reminiscent of high-grade Gleason patterns (i.e. 4 and 5). All patients had at least one primary tumor classified as high-grade acinar adenocarcinoma (ISUP 4/5). The morphology of two primary tumors (P48 and P49) was not assessable due to poor tissue preservation in the autopsy setting (Table S2).

Molecular landscape

The most frequently mutated genes across the brain metastasis samples (44 patients) were TP53 (15/44 patients, 34.1%), APC (8/44 patients, 18.2%), AR (8/44 patients, 18.2%), and MLL2 (8/44 patients, 18.2%) (Table S4). Additionally, brain metastases showed frequent copy number alterations (Table S5), such as amplifications in AR (22/44 patients, 50.0%) and MYC (12/44 patients, 27.3%). PTEN homozygous deletions and coding mutations were found in PCBMs of 14/44 patients (31.8%) and 7/44 patients (15.9%), respectively. Further, RB1 presented homozygous deletions in PCBMs of 3/44 patients (6.8%) and mutations in 6/44 patients (13.6%) (Figure S1, Table S4, Table S5).

A high microsatellite instability (MSI-H) was detected in PCBMs of 3 patients (3/44, 6.8%) (Figure S1, Table S3). PCBM presenting a tumor mutational burden (TMB) > 10 Mutations/Mb could be found in 7/44 patients (16%) and the highest TMB (> 30 Mutations/Mb) was found in the PCBM of P39, P49, and P58. P39 and P58 also harbored MSI-H tumors. In P49, several missense alterations of PMS2 were found with an unknown pathogenic potential and a low allele frequency. No POLE mutation was present in this patient (Table S3).

Further, early driver molecular alterations were investigated in the primary tumor and metastatic component of the entire cohort (N = 46). TMPRSS2-ERG fusions were found in 13/46 patients (28.3%), and TMPRSS2-ETV4 was detected in 1/46 additional patient (2.2%). The mutual exclusivity between TMPRSS2 fusions and SPOP or IDH1 alterations was also observed in this cohort, as previously described [12] (Figure S2; Table S6).

Finally, to investigate tumor heterogeneity, we compared the molecular landscape of different ROIs arising from primary tumors and metastases of the same patient. Multiple samples from the same patients showed a high degree of variant concordance, while the small fraction of unique variants usually consisted of low variant allele frequencies (Figure S3). Similar to short variants, within-patient samples had mostly concordant CNA profiles (Figure S1, Figure S4).

Homologous recombination repair alterations

The main focus of this study was to determine the landscape of HRR alterations in PCBM using the FoundationOne®CDx assay. For this analysis, we considered the 15 HRR genes from the PROfound study (BRCA1, BRCA2, ATM, BRIP1, BARD1, CDK12, CHEK1, CHEK2, FANCL, PALB2, PPP2R2A, RAD51B, RAD51C, RAD51D, and RAD54L) and the inclusion criteria used for qualifying alterations for the PROfound clinical trial [7] (see material and methods). Qualifying alterations were found in 24/97 brain metastases (24.7%), corresponding to 12/44 patients with brain metastasis samples (27.3%). Short variant alterations, homozygous deletions and rearrangements were present in 11/44 (25%), 2/44 (4.5%) and 1/44 (2.3%) of patients with brain metastases samples, respectively. Alterations in BRCA2, ATM and BRCA1 were identified in the PCBM of 4/44 (9.1%), 3/44 (6.8%) and 2/44 (4.5%) patients, respectively (Fig. 1; Table 1, Table S3).

Fig. 1.

Summary of the qualifying alterations in the PROfound HRR genes in the PCBM cohort, detected by the FoundationOne®CDx clinical assay. Alteration types are colored according to the legend. Patient numbers are colored and annotated with numbers to improve legibility.

Source data are provided in Supplementary Tables S2 and S3

Table 1.

Alterations in the PROfound HRR genes (per patient) in the PCBM cohort

GENE	Patients with qualifying alteration	Patients with qualifying alteration (PCBM only)
BRCA2	5 (11.4%)	4 (9.1%)
ATM	4 (9.1%)	3 (6.8%)
BRCA1	2 (4.4%)	2 (4.4%)
BRIP1	2 (4.4%)	1 (2.2%)
CDK12	1 (2.2%)	1 (2.2%)
CHEK2	1 (2.2%)	0 (0%)
RAD51C	1 (2.2%)	1 (2.2%)
RAD51D	1 (2.2%)	0 (0%)
BARD1	0 (0%)	0 (0%)
CHEK1	0 (0%)	0 (0%)
FANCL	0 (0%)	0 (0%)
PALB2	0 (0%)	0 (0%)
PPP2R2A	0 (0%)	0 (0%)
RAD51B	0 (0%)	0 (0%)
RAD51L	0 (0%)	0 (0%)

Primary tumors presented qualifying alterations in HRR genes in 7/18 patients (38.9%). Of them, 5/7 patients (71.4%) harbored the same mutations as in their matched PCBM. For patient P37, only the primary tumor was sequenced, and showed HRR alterations. For patient P43, both liver metastases had the same ATM mutation as in the primary and PCBM samples. On the other hand, P49 showed a HRR alteration in one PCBM sample, but not in the other primary or metastatic samples (Fig. 1, Table S3).

Discussion

An increase in the number of patients with mCRPC presenting metastases in more atypical locations, such as the brain, has been noted. This may be due to improvements in the therapeutic alternatives during the last decades [8]. Nevertheless, inclusion in clinical trials testing new targeted therapies is usually refused for this subset of patients. Indeed, in the current era of precision medicine, we might want to reformulate classical conventions and also allow the access to personalized treatments to patients suffering from advanced oncological diseases. Such a clinical situation is found in patients with mCRPC presenting PCBM. To address whether patients with PCBM might have been considered in a study such as the PROfound, we molecularly investigated a large PCBM cohort (N = 44) by using the FDA-approved companion diagnostic FoundationOne®CDx assay. FoundationOne®CDx assay proved to be valuable in cancer clinical trials, with increasing applications as companion diagnostics for precision medicine [7, 12–17].

We observed frequent molecular alterations in tumor suppressor genes such as P53, PTEN or RB1 as well as in oncogenes as MYC. These molecular events correlate with the clinical advanced status of the present cohort [1]. At the time of PCBM diagnosis, over half of the patients (55%) showed several foci of PCBM and most of them (95%) presented additional metastases in other organs. Thus, a large number of the patients (84%) had received systemic ADT therapy or underwent orchiectomy, with a subset of them having been additionally treated with ARSis (Abiraterone or Enzalutamide). Effects of the therapeutic pressure the patients were exposed, can be also observed in our cohort with 66% of the patients harboring alterations in AR gene.

We investigated the subset of HRR-genes in our PCBM cohort and applied the qualifying criteria for oncogenicity within their alterations, as defined in the PROfound clinical trial, which led to the first FDA approval for PARPi in mCRPC. In particular, the use of FoundationOne®CDx assay, allowed to directly compare the results in the current cohort with the ones published in the PROfound clinical trial. We found qualifying alterations in the PCBM of 12 patients (27.3%), with 6 of them (13.6%) harboring alterations in BRCA2 and BRCA1 genes. The most frequent alterations were mutations present in BRCA2 and ATM genes, as observed in other studies [18–20].

The PROfound study was designed with two cohorts, the first one (cohort A) including patients with alterations in BRCA1, BRCA2 and ATM; the second one (cohort B), including patients with qualifying alterations in one of the other HRR genes, as defined in the inclusion criteria (see methods). Overall survival (OS) benefit for olaparib vs physician's choice was observed for patients included in the cohort A, with a median OS of 18.5 vs 15.1 months (hazard ratio (HR): 0.64, p = 0.02) [7]. In our cohort, considering only qualifying alterations for inclusion in the cohort A of PROfound trial (i.e. alterations in BRCA1, BRCA2 and ATM genes), PCBM of nine patients (20.5%) demonstrated alterations.

However, exploratory subgroup analyses showed an OS benefit uniquely for patients with tumors harboring alterations in BRCA1 and BRCA2, and no differences in OS for patients with ATM alterations. This finding was later confirmed by further trials evaluating efficacy of PARP inhibitors in mCRPC [21–24]. One possible explanation for the unsuccessful response to PARPi of ATM-mutated prostate cancer might be the nature of those mutations, representing alterations as part of the clonal hematopoiesis (CH) instead of driver alterations of the underlying malignancy. CH has been demonstrated to be an aging process and ATM one of its frequently affected genes [25].

Moreover, recent first-line trials in mCRPC showed OS benefit for ARSis plus PARPi combinations for patients with HRD-altered tumors, and, specifically, for patients with BRCA1 and BRCA2 alterations [22–24]. Lastly, according to the Swiss healthcare system, PARP inhibitors are only regularly approved in monotherapy for Swiss patients with mCRPC and deleterious somatic or germline alterations in BRCA1 or BRCA2. In the present study, PCBM presented qualifying alterations in BRCA1 or BRCA2 in six patients (13.6%).

Further, the FoundationOne®CDx assay allowed us to identify additional predictive biomarkers in patients suffering from PCBM. A high microsatellite instability (MSI-H) was detected in three patients and four patients presented a tumour mutational burden (TMB) higher than 10 mutations/Mb. All together, seven patients (16%) might have benefited from a targeted therapy with immune checkpoint inhibitors (ICIs) according to FDA-approved criteria (https://www.cancer.gov/about-cancer/treatment/types/agnostic-cancer-therapies-hp-pdq).

One limitation of this study is the lack of direct comparison of the current FoundationOne®CDx-based results and the published WES-based results from the same cohort [9]. However, a head-to-head comparison between the two methods is quantitatively and qualitatively not appropriate due to different reasons. Despite the overlapping donors, the total number of samples was not identical between the two studies. Moreover, independent DNA extractions were performed for each study, resulting in different cells being probed, and different DNA qualities across samples. Nevertheless, the fraction of PCBM patients (12/44 = 27.3%) with qualifying alterations in the PROfound HRR genes identified using the FoundationOne®CDx assay was in agreement with the WES study (10/51 = 19.6%).

All together, by using the FDA approved companion diagnostic FoundationOne®CDx assay, we could determine predictive molecular alterations in the PCBMs of 13 patients (29.5%). Of them, six patients (13.6%) might have benefited from ICI and PARPi combination therapy, six patients (13.6%) from PARPi therapy, and one patient (2.3%) from ICI therapy. We posit that routinely molecular diagnostic of PCBM in the context of mCRPC could identify up to a third of patients with targetable alterations, who might benefit from personalized novel therapies.

Author contributions

AR-C, AB and MAR designed the study and the experiments. AR-C and AB performed experiments and analysis of the results. AB performed the bioinformatic analysis and generated the figures. AR-C, AB, and MAR developed the concept. AR-C and MAR developed and performed the pathology review and collected the clinical data. DA provided a clinical perspective to the results. SM, HM, MZ, and MN performed and coordinated the FoundationOne®CDx analyses. MAR provided administrative, technical and material support. AR-C, AB, DA and MAR wrote the initial draft of the manuscript and all authors contributed to the final version.

Funding

The study was supported by Roche Pharma (Schweiz) AG, Basel, Switzerland. A. R.-C was supported by a Prostate Cancer Foundation (PCF) Young Investigator award, Stiftung für klinisch-experimentelle Tumorforschung (SKET), Protected Research Time Grant (DLF, University Hospital of Bern and University of Bern). D.A. was supported by Stiftung für klinisch-experimentelle Tumorforschung (SKET), Protected Research Time Grant (DLF, University Hospital of Bern and University of Bern).

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

All analyses followed protocols approved by the Ethical Committee Bern (Project ID: 2019–00328). The study was performed in accordance with the Declaration of Helsinki. Patients qualified for inclusion in this study if they were informed about the conditions in the general consent and written general consent was signed. Inclusion also applied for patients whose prostate cancer tissue was collected prior to official use of general consent, and/or no written or documented verbal refusal was available at any time (Human Research Act, HRA, Swiss Confederation; Art. 34).

Consent for publication

According to the above mentioned ethical approval, we may publish the results after completion of the research project. Publications will not contain any references to specific individuals.

Competing interests

Costs of FoundationOne®CDx analyses were covered by Roche Pharma (Schweiz) AG, Basel, Switzerland. D.A. has received honoraria for lectures, presentations or advisory boards from BMS, Janssen, and Bayer; conference/travel support from Janssen, Ipsen, and Bayer. H.M. has received research funding to University Zurich from F. Hoffmann-La Roche Ltd., and personal consultancy fees as advisory board member from Astra Zeneca and Stemline Therapeutics, Bayer, Amgen, Astella, MSD. M.A.R. is a co-inventor on patents in the area for diagnosis and therapy in prostate cancer for ETS fusions (University of Michigan and the Brigham and Women’s Hospital), SPOP mutations (Cornell University), and EZH2 (University of Michigan). The authors declare no competing interests.