Multi-Gene Next-Generation Sequencing Panel for Analysis of BRCA1/BRCA2 and Homologous Recombination Repair Genes Alterations Metastatic Castration-Resistant Prostate Cancer

Despite significant therapeutic advances, metastatic CRPC (mCRPC) remains a lethal disease. Mutations in homologous recombination repair (HRR) genes are frequent in mCRPC, and tumors harboring these mutations are known to be sensitive to PARP inhibitors. The aim of this study was to verify the technical effectiveness of this panel in the analysis of mCRPC, the frequency and type of mutations in the BRCA1/BRCA2 genes, as well as in the homologous recombination repair (HRR) genes. A total of 50 mCRPC cases were analyzed using a multi-gene next-generation sequencing panel evaluating a total of 1360 amplicons in 24 HRR genes. Of the 50 cases, 23 specimens (46.0%) had an mCRPC harboring a pathogenic variant or a variant of uncertain significance (VUS), whereas in 27 mCRPCs (54.0%), no mutations were detected (wild-type tumors). BRCA2 was the most commonly mutated gene (14.0% of samples), followed by ATM (12.0%), and BRCA1 (6.0%). In conclusion, we have set up an NGS multi-gene panel that is capable of analyzing BRCA1/BRCA2 and HRR alterations in mCRPC. Moreover, our clinical algorithm is currently being used in clinical practice for the management of patients with mCRPC.


Introduction
Prostatic cancer is the most common male neoplasia in Europe, the Americas, Australia, and Africa, with a total of 19,292,789 new cases worldwide and a world mortality rate in 2020 equal to 7.7 per 100,000 [1].
RAD51D, or RAD54L (cohort B) [16]. Rucaparib received approval based on the phase two TRITON2 trial (ClinicalTrials.gov identifier NCT02952534) in patients with BRCAmutant mCRPC [23]. On the basis of the interim results from the phase two GALAHAD trial (ClinicalTrials.gov identifier NCT02854436), niraparib received breakthrough therapy designation for the treatment of BRCA-mutant mCRPC. Niraparib demonstrated clinical activity in patients with treatment-refractory mCRPC who showed durable responses, particularly in biallelic BRCA mutation carriers [24].
The next-generation sequencing panel can detect different genetic aberrations, point mutations, indels, and copy number variations (CNVs) in a single test. Even if commercial NGS panels are commonly available, the NGS gene panels may also be customizable and provide flexibility to select therapeutically actionable genes for specific testing purposes of germline and tissue testing [25]. For example, laboratory-developed panels may be designed for analyzing only BRCA1/BRCA2 and the other HRR genes clinically relevant in prostate cancer [6].
Though a higher prevalence of HRR mutations was obtained from the metastatic tissue samples, less than 5% of metastatic samples were from bone tissue, emphasizing the inaccessibility of bone metastatic tissue [6,16]. Although a higher prevalence of mutations is observed in metastatic tissue, obtaining a sample from a metastatic site is difficult in mCRPC, as the most frequent site of metastasis is the bone. The heterogeneity of the tumor tissue itself is a restriction because it may not accurately reflect the biology of the tumor and, as a result, its overall genetic mutation profile. Improper fixation of tumor samples poses specific challenges to the integrity of DNA. Fresh-frozen samples are a feasible sample type for genetic analysis. However, in clinical settings, it may not always be possible to perform a rebiopsy, and the determination of tumor content may also be a challenge before proceeding with NGS-based HRR gene testing. In such cases, archived samples are often used, with FFPE samples being the most preferred option [26][27][28]. Tumor content evaluation of FFPE samples is critical for identifying successful genetic alterations through NGS gene panel testing. If the tumor content is inadequate, it is advisable to obtain microdissected target tissue by a trained pathologist to enrich the tumor content [29].
Given the importance of mutational analysis and PARP inhibitors in the treatment of prostate cancer, the aim of this study was the development of a next-generation sequencing panel for the molecular characterization of prostate cancer. In particular, with this study, we wanted to verify the technical effectiveness of this panel, the frequency and type of mutations in the BRCA1/BRCA2 genes, as well as the possible presence of mutations or mutations in the HRR (homologous recombination repair) genes, and the significance of possible VUS (variants of uncertain significance) in mCRPC.

NGS Panel Performance
The mean coverage of the whole panel was 1291.5 (ranging from 297 to 3334.7), whereas the mean coverage of the amplicons covering BRCA1/BRCA2 regions was 1684.6 (ranging from 468 to 4162) ( Figure 1).  In three of the fifty samples (6.0%), the mean coverage of the whole panel was below 500× (ranging from 297.0 to 460.6) (Figure 2A), while in two of the fifty samples (4.0%) the mean coverage of the BRCA1/BRCA2 regions was below 500× (ranging from 468.3 to 480.5) ( Figure 2B). Of the three samples with a mean coverage below 500×, two were biopsy specimens and one was a surgical resection, all from primary prostatic lesions. Of these three In three of the fifty samples (6.0%), the mean coverage of the whole panel was below 500× (ranging from 297.0 to 460.6) (Figure 2A), while in two of the fifty samples (4.0%) the mean coverage of the BRCA1/BRCA2 regions was below 500× (ranging from 468.3 to 480.5) ( Figure 2B). In three of the fifty samples (6.0%), the mean coverage of the whole panel was below 500× (ranging from 297.0 to 460.6) (Figure 2A), while in two of the fifty samples (4.0%) the mean coverage of the BRCA1/BRCA2 regions was below 500× (ranging from 468.3 to 480.5) ( Figure 2B). Of the three samples with a mean coverage below 500×, two were biopsy specimens and one was a surgical resection, all from primary prostatic lesions. Of these three Of the three samples with a mean coverage below 500×, two were biopsy specimens and one was a surgical resection, all from primary prostatic lesions. Of these three speci-

Mutational Analysis in Neoplastic Tissue (Somatic Alterations)
A total of 23 patients (46.0%) harbored a pathogenic variant or a VUS, whereas in 27 samples (54.0%), no variants were detected (wild-type tumors) (Supplementary Table S1, Figure 4A). Of the twenty-three mutated samples, sixteen harbored at least one pathogenic mutation, whereas in seven, only VUS variants (but not pathogenic ones) were detected ( Figure 4B). BRCA2 was the most commonly mutated gene (14.0% of patients, 12.0% if only pathogenic/likely pathogenic variants were considered), followed by ATM (12.0%, but only 2.0% if VUS alterations were not included), and BRCA1 (6.0%) ( Table 1). Other altered genes were found in one to two patients (Table 1). In three cases (#11, #19, and #27, Supplementary Table S1) the BRCA1/BRCA2 mutations were confirmed using another NGS panel (Oncomine BRCA1/BRCA2, Thermo Fisher Scientific, Waltham, MA, USA), and in all cases the mutations were confirmed. For the other BRCA1/BRCA2 mutated samples, not enough material was available to perform another test.

Mutational Analysis in Neoplastic Tissue (Somatic Alterations)
A total of 23 patients (46.0%) harbored a pathogenic variant or a VUS, whereas in 27 samples (54.0%), no variants were detected (wild-type tumors) (Supplementary Table S1, Figure 4A). Of the twenty-three mutated samples, sixteen harbored at least one pathogenic mutation, whereas in seven, only VUS variants (but not pathogenic ones) were detected ( Figure 4B).

Mutational Analysis in Neoplastic Tissue (Somatic Alterations)
A total of 23 patients (46.0%) harbored a pathogenic variant or a VUS, whereas in 27 samples (54.0%), no variants were detected (wild-type tumors) (Supplementary Table S1, Figure 4A). Of the twenty-three mutated samples, sixteen harbored at least one pathogenic mutation, whereas in seven, only VUS variants (but not pathogenic ones) were detected ( Figure 4B). BRCA2 was the most commonly mutated gene (14.0% of patients, 12.0% if only pathogenic/likely pathogenic variants were considered), followed by ATM (12.0%, but only 2.0% if VUS alterations were not included), and BRCA1 (6.0%) ( Table 1). Other altered genes were found in one to two patients (Table 1). In three cases (#11, #19, and #27, Supplementary Table S1) the BRCA1/BRCA2 mutations were confirmed using another NGS panel (Oncomine BRCA1/BRCA2, Thermo Fisher Scientific, Waltham, MA, USA), and in all cases the mutations were confirmed. For the other BRCA1/BRCA2 mutated samples, not enough material was available to perform another test. BRCA2 was the most commonly mutated gene (14.0% of patients, 12.0% if only pathogenic/likely pathogenic variants were considered), followed by ATM (12.0%, but only 2.0% if VUS alterations were not included), and BRCA1 (6.0%) ( Table 1). Other altered genes were found in one to two patients (Table 1). In three cases (#11, #19, and #27, Supplementary Table S1) the BRCA1/BRCA2 mutations were confirmed using another NGS panel (Oncomine BRCA1/BRCA2, Thermo Fisher Scientific, Waltham, MA, USA), and in all cases the mutations were confirmed. For the other BRCA1/BRCA2 mutated samples, not enough material was available to perform another test.  Overall, a total of 35 pathogenic/likely pathogenic/VUS variants were detected in the HRR genes. BRCA2 was the most commonly mutated gene, accounting for nine of the thirty-five mutations (25.7%), followed by ATM (six mutations, 17.1%), and BRCA1 (three mutations, 8.6%) ( Table 1, Supplementary Figure S1). In all the other genes, one or two variants were detected (Table 1, Supplementary Figure S1).
In seven cases, concomitant mutations were detected ( Intriguingly, after the first "round" of analysis, the total number of VUS variants was equal to 32. After revising the significance of the variants at the end of the study, sixteen variants were reclassified from VUS to "benign" or "likely benign" in the following genes: three variants in ARID1A, four in ATM, three in BRIP1, two in CDK12, two in CHEK2, and one in RAD51B.

Correlation between HRR Mutations and Age or the Gleason Score
No statistically significant differences were observed in age between BRCA1/2 mutated patients (mean age: 67.2 years) and BRCA1/BRCA2 WT patients (mean age: 63.4 years) (p = 0.2451, Mann-Whitney test) ( Figure 5A). Likewise, no differences in age were observed also in HRR mutated patients' group (mean age: 64.9 years) and HRR WT patients (mean age: 62.9 years) (p = 0.2711, Mann-Whitney test) ( Figure 5B), and between BRCA1/2 mutated patients (mean age: 67.2 years) and HRR-BRCAness mutated patients (mean age: 63.1 years) (p = 0.4729, Mann-Whitney test) ( Figure 5C). We did not find any differences when considering or excluding VUS mutations from the groups of mutated patients. Correlation between HRR Mutations and Age or the Gleason Score No statistically significant differences were observed in age between BRCA1/2 mutated patients (mean age: 67.2 years) and BRCA1/BRCA2 WT patients (mean age: 63.4 years) (p = 0.2451, Mann-Whitney test) ( Figure 5A). Likewise, no differences in age were observed also in HRR mutated patients' group (mean age: 64.9 years) and HRR WT patients (mean age: 62.9 years) (p = 0.2711, Mann-Whitney test) ( Figure 5B), and between BRCA1/2 mutated patients (mean age: 67.2 years) and HRR-BRCAness mutated patients (mean age: 63.1 years) (p = 0.4729, Mann-Whitney test) ( Figure 5C). We did not find any differences when considering or excluding VUS mutations from the groups of mutated patients. No statistically significant differences were observed in BRCA1/BRCA2 mutation frequencies according to the Gleason Score (p = 0.6754, Chi-square test) (Table 2). Similarly, no significant differences were found in HRR gene mutations according to the Gleason Score (p = 0.5652, Chi-square test) ( Table 2).   No statistically significant differences were observed in BRCA1/BRCA2 mutation frequencies according to the Gleason Score (p = 0.6754, Chi-square test) (Table 2). Similarly, no significant differences were found in HRR gene mutations according to the Gleason Score (p = 0.5652, Chi-square test) ( Table 2).  No statistically significant differences were observed in the frequencies of BRCA1/BRCA2 (p = 0.5349, Chi-square test) or HRR gene mutations (p = 0.9221, Chi-square test) frequencies in the different Grade Score groups (Table 3).

Germline Analysis
In five patients, the detected variants required genetic counseling and evaluation at the germline level (Table 4). For the other detected mutations, genetic counseling was not recommended due to low VAF (<20%) or the mutated gene. Regarding the five patients for whom genetic counseling was recommended, three have been analyzed, and two will be in the near future. Two variants, BRCA2 c.4131_4132ins TGAGGA and MLH1 c.794G>A, were confirmed in the germline DNA. However, the latter is classified as "likely pathogenic" according to ACMG (and Tier II for AMP), but as a VUS according to ClinVar. The germline DNA from one patient harboring the BRCA2 c.4913_4915delinsTTC and BRCA2 c.4983T>G variants is still under evaluation.

Discussion
In a molecular analysis of 333 primary prostate cancers, the Cancer Genome Atlas (TCGA) study showed a 19% prevalence of alterations in several DNA repair genes, including BRCA2, BRCA1, ATM, CDK12, FANCD2, and RAD51C [30].
The recent PROfound study represents the largest analysis currently available on DNA repair defects in prostate cancer [16]. This phase three clinical study evaluated the efficacy of the PARP inhibitor "Olaparib" in patients with metastatic castration-resistant prostate cancer (mCRPC) and evaluated 2792 biopsies for the presence of aberrations in fifteen genes involved in DNA repair [31].
It has been demonstrated that the pathogenic variants (PV) of BRCA1/BRCA2 genes, whether germline or somatic, represent a predictive biomarker of greater sensitivity to treatment with inhibitors of the enzyme poly (ADP-ribose) polymerase (PARP), which intervenes in DNA repair damaged single-strand prostate cancer in patients with hormoneresistant metastatic prostate cancer. The efficacy of PARP inhibitors as a therapeutic option in prostate cancer is achieved through a "synthetic lethality" process with a simultane-ous loss of double-stranded DNA repair function by homologous recombination (HR), in which BRCA1/BRCA2 proteins play an essential role [32]. In October 2020, clinical studies led to the registration by the European Regulatory Agency EMA (European Medicines Agency) of the PARP inhibitor olaparib "indicated, in monotherapy, for the treatment of adult patients with castration-resistant metastatic prostate cancer with BRCA1/BRCA2 gene mutations (germline PV and/or somatic PV), progressing after previous treatment including a new hormonal agent" (https://www.ema.europa.eu/en/documents/product-information/lynparza-eparproduct-information_it.pdf (accessed on 30 March 2023)). Patients must have confirmation of a PV in the BRCA1/BRCA2 prostate cancer susceptibility genes (germline or somatic) before initiating treatment with olaparib. The assessment of BRCA1/BRCA2 mutation status should be performed in a specialized laboratory using a validated test method [32][33][34][35]. Based on this evidence, a referral for BRCA1/BRCA2 testing was proposed for men with metastatic prostate cancer.
The analysis of BRCA1/BRCA2 and HRR genes is, therefore, important today both for the therapeutic management of the patient and for undertaking a possible path of oncogenic counseling in family members in order to identify high-risk carriers, to whom to propose targeted programs of early diagnosis of tumors associated with BRCA-related hereditary-familial transmission syndromes and strategies aimed at reducing the risk.
Molecular analysis of prostate samples is particularly tricky. In the PROFOUND study, FFPE tumor tissue samples were used for genetic testing. Of the 4047 samples available, the reasons for test failure in 31% of samples was pathology review failure (6.8%), DNA extraction failure (13.2%), and failure after DNA extraction (6.9%) [16].
Multiple gene targets can be tested using multi-gene NGS cancer panels. The benefits of using multi-gene panels are: (i) a large number of targets can be tested starting from the same amount of nucleic acid input (usually~10-20 ng of DNA); (ii) costs are reduced to the minimum possible, and the price of analyzing a specific target gene is equal to the overall cost divided by the number of genes in the panel.
Moreover, the sequences of genes not initially sought by the clinician remain in lab databases. This enables quick data recovery in the event of a need (e.g., updating of recommendations or the discovery of additional predictive/prognostic indicators) without having to reextract the nucleic acid and resequence the specimen. Repeating these analyses would be difficult in samples with low amounts of biological material, such as prostate biopsies. The custom-designed multi-gene panels also have the flexibility to be quickly updated at any moment in response to the identification of unique new biomarkers or guideline revisions, irrespective of the selection and timing of commercial diagnostic tools.
In this study, we validated a laboratory-developed custom-designed multi-gene NGS panel allowing us to analyze BRCA1/BRCA2 genes together with other 22 HRR and mismatch repair involved genes. Our custom-designed multi-gene NGS panel is reliable, with a relatively low overall percentage of cases with low coverage (i.e., <500×) amplicons.
Considering that, in mCRPC, material from the metastatic site may not always be available or of poor quality/quantity (e.g., in bone metastases), the NGS panel must be successfully performed on archival prostate material. By also using our NGS panel on archived material from the primitive site, we have demonstrated that this academic panel can successfully be used in patients where metastatic specimens are unavailable, and the archived prostate sample must be used for BRCA/HRR analysis.
Our data are consistent with those reported in the literature: BRCA1 and BRCA2 genes were confirmed as the more frequent HRR genes altered in mCRPC (20% of analyzed samples), demonstrating that the primary prostate site can be successfully used for the analysis of BRCA/HRR genes in mCRPC patients. However, it should be considered that taken together, the other HRR genes were observed to be altered in 26% of mCRPCs. These data lead to the hypothesis that it could be useful to investigate other HRR genes other than BRCA1/BRCA2 to obtain a more reliable molecular profile of mCRPCs. Considering that only 20% of specimens harbored BRCA1/BRCA2 alterations, it would be more profitable to use a unique multi-gene panel, including BRCA1/BRCA2 genes together with other HRR markers, rather than using a sequential approach, analyzing BRCA1/BRCA2 as a first step and continuing with HRR analysis only in BRCAness samples. Another useful observation from the data that we have obtained is the importance of revaluing the VUS alterations. In fact, in our study, 50% of VUS variants identified have been reclassified as "benign" or "likely benign" at a subsequent evaluation (from 6 to 12 months later than the first evaluation).
The limitations of this study are due to the limited number of samples and the fact that the molecular analyses were predominantly performed on archival primary biopsies rather than metastatic samples. Even if it has been demonstrated that primary prostate tissue accurately reflects the mutational status of metastatic prostate tissue [36]. Considering that the molecular status of metastases may differ from that of the primary lesion, whenever possible (i.e., when the material is available and the nucleic acid is of good quality), it may be preferable to perform BRCA/HRR analysis on the metastatic specimen. However, if the metastatic sample is not available or the extracted DNA is of poor quality, as can be the case for bone biopsies, then it would be possible to perform NGS analysis on the primary lesion. Indeed, in light of the poor quality of nucleic acid obtained from bone specimens, the ESMO guidelines emphasize that the sample used for this test "should possibly not belong to bone metastasis" [37]. In the present study, it has been demonstrated that analysis of archival samples in mCRPC samples is feasible using this NGS panel, and the percentage of mutations observed in BRCA1/BRCA2 and HRR genes is overlaps with that reported in the literature [8,9,38]. Another possible alternative in case of unavailable metastatic material could be to perform the test on liquid biopsy (e.g., plasma or urine [39,40]), although the application of this method is challenging due to the lack of standardized analysis methods, the need for high analytical sensitivity, and expertise in data analysis [39].
Using our analysis protocol, it was possible to create a clinical pipeline for the management of patients with mCRPC. The clinician may request the typing of BRC1/BRCA2 and HRR genes on somatic tissue when necessary. The analysis is then performed on the available lesion or the most representative one. The result is then communicated to the oncologist for therapeutic management, and if a pathogenic/VUS mutation with a VAF greater than 20% is identified, the genetic counseling process is activated, thanks to the signing of the informed consent by the patient at the time of the oncological visit ( Figure 6). In our series, five of fifty patients (10.0%) needed genetic counseling, three patients due to BRCA1/BRCA2 mutations, and two patients due to genes connected to the mismatch repair system (MLH1, PMS2). Therefore, the algorithm is currently in use in clinical practice for the management of patients with mCRPC. Moreover, from a future perspective, the panel could also be further implemented by adding the HRD (Homologous Recombination Deficiency) evaluation, as already happens for ovarian cancers.

Case Selection
A total of 50 mCRPC cases were analyzed in the Molecular Laboratory of Solid Tumors, IRCCS Policlinico di S. Orsola. All cases were formalin-fixed and paraffin-embed-

Case Selection
A total of 50 mCRPC cases were analyzed in the Molecular Laboratory of Solid Tumors, IRCCS Policlinico di S. Orsola. All cases were formalin-fixed and paraffin-embedded (FFPE), 40 cases (80.0%) were bioptic specimens, and 10 (20.0%) were surgical resections ( Table 5). The age range was between 40 and 80 years old (mean 63.8 years). All 50 specimens were from metastatic CRPC. The sources of the analyzed specimens were as follows: 43 specimens (86.0%) were from primary prostatic lesions, 3 (6.0%) were from lymph nodal metastases, 2 (4.0%) were from lung metastases, 1 (2.0%) was from liver metastasis, and 1 (2.0%) was from a bladder metastatic site ( Table 5). All the specimens were obtained from patients who showed disease progression after undergoing androgen deprivation therapy. In cases where the primitive prostate site was analyzed and no mutations were found in BRCA1/BRCA2 or other HRR genes (n = 26), it was not possible to perform a biopsy of the metastatic site for molecular analysis.
All patients signed an informed consent to the genetic test performed on tumoral tissue and, in case the analysis revealed potentially hereditary genetic variants, to authorize to communicate data to the Cancer Genetics Clinic (IRCCS Policlinico di S. Orsola) and the latter to contact the patient to schedule a cancer genetic consultation (Figure 7). the Grade Group, 8 tumors (16.0%) had a grade equal to 3, 13 (26.0%) had a grade score of 4, and 26 (52.0%) had a grade score equal to 5 (Table 5). Neoplastic cell enrichment ranged from 20% to 80% (mean, 37.8%).
All patients signed an informed consent to the genetic test performed on tumoral tissue and, in case the analysis revealed potentially hereditary genetic variants, to authorize to communicate data to the Cancer Genetics Clinic (IRCCS Policlinico di S. Orsola) and the latter to contact the patient to schedule a cancer genetic consultation (Figure 7).

NGS Analysis
DNA from the FFPE block was extracted using starting from two to four 10 μm FFPE sections under microscopic guidance according to the more representative area marked

NGS Analysis
DNA from the FFPE block was extracted using starting from two to four 10 µm FFPE sections under microscopic guidance according to the more representative area marked by a pathologist on a haematoxylin and eosin control-stained slide. Extracted DNA was quantified using a fluorometer (Qubit, Thermo Fisher Scientific, Waltham, MA, USA).
In the event of detection of pathogenic variants or VUS with allelic load ≥ 20% in the BRCA1/BRCA2, MLH1, MSH2, MSH6, PMS2, RAD51, BAP1 genes or any other genes present in the panel to be reported ad hoc (potentially hereditary variants), the report of the Metropolitan Laboratory of Molecular Pathology is sent to the requesting clinician as well as to the Oncological Genetics Clinic for further case management. Institutional Review Board Statement: All the experimental procedures were carried out in accordance with the general authorization to process personal data for scientific research purposes from "The Italian Data Protection Authority" (http://www.garanteprivacy.it/web/guest/home/docweb/ -/docwebdisplay/export/2485392 (accessed on 29 January 2021)). All information regarding human material was managed using anonymous numerical codes, and all samples were handled in compliance with the Helsinki Declaration (https://www.wma.net/fr/news-post/en-matierede-transfertdes-taches-la-securite-des-patients-et-la-qualitedes-soins-devraient-etre-primordiales/ (accessed on 29 January 2021)). The study did not affect the clinical management of the involved patients' samples.