Characteristics of BRCA2 Mutated Prostate Cancer at Presentation

Genetic alterations of DNA repair genes, particularly BRCA2 in patients with prostate cancer, are associated with aggressive behavior of the disease. It has reached consensus that somatic and germline tests are necessary when treating advanced prostate cancer patients. Yet, it is unclear whether the mutations are associated with any presenting clinical features. We assessed the incidences and characteristics of BRCA2 mutated cancers by targeted sequencing in 126 sets of advanced prostate cancer tissue sequencing data. At the time of diagnosis, cT3/4, N1 and M1 stages were 107 (85%), 54 (43%) and 35 (28%) samples, respectively. BRCA2 alterations of clinical significance by AMP/ASCO/CAP criteria were found in 19 of 126 samples (15.1%). The BRCA2 mutated cancer did not differ in the distributions of TNM stage, Gleason grade group or histological subtype compared to BRCA2 wild-type cancers. Yet, they had higher tumor mutation burden, and higher frequency of ATM and BRCA1 mutations (44% vs. 10%, p = 0.002 and 21% vs. 4%, p = 0.018, respectively). Of the metastatic subgroup (M1, n = 34), mean PSA was significantly lower in BRCA2 mutated cancers than wild-type (p = 0.018). In the non-metastatic subgroup (M0, n = 64), PSA was not significantly different (p = 0.425). A similar trend was noted in multiple metastatic prostate cancer public datasets. We conclude that BRCA2 mutated metastatic prostate cancers may present in an advanced stage with relatively low PSA.

It has been known that some PCs present with low serum PSA but high-grade histology, which often rapidly progresses to lethal disease [15,16]. Reportedly, the PSA-low tumors are characterized by aberrant histology such as neuroendocrine, ductal/intraductal, or sarcomatoid features [15,17,18]. In particular, intraductal and cribriform histology are frequently found in BRCA2 and other HRR gene-mutated PCs [19][20][21].
We hypothesized that BRCA2 mutation-either somatic or germline-is a special subgroup of PC and may be clinically or pathologically distinguished from those wild-type tumors. In this paper, we reviewed a consecutive series of targeted gene panel sequencing data of high-risk PCs-Gleason grade group (GGG) 4-5, cT3/4 stage, N1 or M1 stage at diagnosis. We identified those with BRCA2 mutations, and compared their clinical, pathological and genomics characteristics with the wild-type tumors.

Results
We evaluated 126 sets of consecutive prostate cancer tissue NGS data (from June 2019 and April 2021). There were 26 patients with 44 BRCA2 genetic alterations (42 mutations, 1 amplification and 1 heterozygous loss) (Table S1). Among the 26 patients, we identified 19 (15.1%) patients with BRCA2 alterations that were classified as "Pathogenic" or "Likely Pathogenic" by ACMG classification and "Tier I" or "Tier II" by AMP/ASCO/CAP classification (Table S2). Of note, the 1 hetloss was considered as an alteration whereas the 1 amplification was considered as wild-type, since BRCA2 is a tumor-suppressor gene.

Clinical Characteristics of BRCA2 Mutated Prostate Cancers
We compared the clinical characteristics of the 19 cases of BRCA2 oncogenic alterations with the remaining 107 wild-type cases. Patient age or TNM stage distribution were not significantly different between BRCA2 mutated (BRCA2mut) and wild-type (BRCA2 WT ) tumors (Table 1). Interestingly, mean PSA at diagnosis of BRCA2mut group tended to be lower than that of the BRCA2 WT group (94.7 ng/mL vs. 242.7 ng/mL, p = 0.101, Tables 1 and S3).

Serum PSA Level Validation in Public Datasets
To validate our findings, we assessed four large public datasets of metastatic castrationresistant (n = 444, 123) [6,22], metastatic castration-sensitive (n = 424) [23], and localized prostate cancers (n = 187) with serum PSA data available [24]. The prevalence of BRCA2 alterations in the studies are summarized in Table 3. In the TCGA-localized prostate cancers, BRCA2 genetic alterations (mutations or deep deletion, somatic/germline) were found in 7 (4%) patients. In this group, Serum PSA of BRCA2-mutated tumors was not significantly different from the wild-types (26.4 ng/mL vs. 10.4 ng/mL, p = 0.107, t-test with Welch's correction, Figure 1), consistent with our data. Median mutation count was higher in the BRCA2-mutated group, yet did not reach statistical significance (

Discussion
We found that BRCA2-mutated metastatic PCs present with relatively low serum PSA. In localized tumors, however, serum PSA of BRCA2 mutated PCs was not lower than the wild-type PCs. Also, we validated our findings with multiple publicly available mCSPC and mCRPC datasets. To our knowledge, this is first report that suggests BRCA2mutated metastatic tumors differ from the wild-type tumor by PSA level at diagnosis. Velho et al. reported that germline DNA-repair gene mutation-positive prostate cancer patients (52% BRCA1 or BRCA2) had lower median PSA levels at diagnosis than mutation-negative patients. The patient cohort analyzed by Velho et al. consisted of 53-62% of T3/4 and 14-23% M1 disease at diagnosis [25]. Based on our analysis, we speculate that the PSA-low advanced stage tumors are enriched by BRCA2 mutation carriers. For metastatic PCs, it is now recommended by the NCCN, AUA or EAU guidelines to run both.
Serum PSA is a highly sensitive biomarker for prostate cancer diagnosis as well as monitoring for recurrence and drug response. Yet, its specificity is not high enough for metastatic disease monitoring, leaving cases with clinical or radiographic progression without PSA elevation [26]. Our findings speculate that BRCA2-altered prostate cancer

Discussion
We found that BRCA2-mutated metastatic PCs present with relatively low serum PSA. In localized tumors, however, serum PSA of BRCA2 mutated PCs was not lower than the wild-type PCs. Also, we validated our findings with multiple publicly available mCSPC and mCRPC datasets. To our knowledge, this is first report that suggests BRCA2-mutated metastatic tumors differ from the wild-type tumor by PSA level at diagnosis. Velho et al. reported that germline DNA-repair gene mutation-positive prostate cancer patients (52% BRCA1 or BRCA2) had lower median PSA levels at diagnosis than mutation-negative patients. The patient cohort analyzed by Velho et al. consisted of 53-62% of T3/4 and 14-23% M1 disease at diagnosis [25]. Based on our analysis, we speculate that the PSA-low advanced stage tumors are enriched by BRCA2 mutation carriers. For metastatic PCs, it is now recommended by the NCCN, AUA or EAU guidelines to run both.
Serum PSA is a highly sensitive biomarker for prostate cancer diagnosis as well as monitoring for recurrence and drug response. Yet, its specificity is not high enough for metastatic disease monitoring, leaving cases with clinical or radiographic progression without PSA elevation [26]. Our findings speculate that BRCA2-altered prostate cancer progresses rapidly to metastatic and castration-resistant disease not necessarily accompanied by increased serum PSA level. Molecular studies suggest that BRCA2 mutation combined with RB1 alteration, which is located closely on chromosome 13q14, can generate an aggressive PC model with castration resistance [27].
Our study shows that BRCA2 mutation is associated with lower PSA at the metastatic state, but not in the localized setting. We suggest some explanations, but these are not yet definitive. First, the major source of serum PSA in patients with metastatic prostate cancer is the cancer cells themselves. However, in a localized setting, theoretically, PSA released from adjacent normal cells may contribute more significantly than in a metastatic setting. Secondly, genetic characteristics of BRCA2 wild-type localized cancer may be different from those of BRCA2 wild-type metastatic cancer. As BRCA2 mutation is enriched in metastatic prostate cancer, TP53 mutation or RB1 deletions are also enriched in metastatic disease, whereas SPOP mutation is more frequent in localized disease [6,21,22,28,29]. Lastly, the metastatic microenvironment may pressure cancer cells to express PSA more or less than the primary site. In other words, a BRCA2-mutated cancer cell at the primary site (prostate) may express high PSA while a cell with same genetic background but located in bone marrow may express less PSA.
Among the various DNA-repair genes of HRR or mismatch repair, we focused on BRCA2 alteration because it is the most frequently mutated HRR gene in advanced PCs [1], and represents a subset of patients that are likely to respond to PARP inhibitors [12,13]. While earlier studies combined ATM mutation in this subgroup [4], recent evidence shows that PARP inhibitor alone is not enough to treat the ATM mutated PCs [30]. Indeed, we found that ATM mutation frequently accompanies BRCA2, which may explain the earlierdocumented benefit of PARP inhibitors in ATM-mutated cases. ATM germline mutation increases the likelihood of PC development [31], and ATM mutation or protein loss are enriched in high-grade cancer [32]. In response to DNA double strand break, the ATM protein modulates diverse aspects of the cell's responses such as cell cycle checkpoint arrest, apoptosis-not limited to DNA repair itself [33][34][35]. This may lead to varying responses to DNA-damaging agents such as PARP inhibitors or radiations in patients with ATM somatic or germline alterations unlike BRCA1 or BRCA2 [36][37][38].
We point to some limitations in this study. Interpretation of our data is limited due to the study's retrospective design. Serum PSA data was available in only a subset of the datasets analyzed in this study, which may increase selection bias. We did not perform survival analysis as the consecutive treatments after sequencing analysis were rather heterogenous. Particularly, some of the patients with HRR mutations underwent clinical trials involving PARP inhibitors which would deviate survival outcome. Lastly, we tried setting some arbitrary cut-off points (100 ng/mL, 10 ng/mL) of PSA threshold or values that could possibly be considered suspicious and warrant genetic testing in metastatic PC at presentation, which were not successful. We argue that somatic and germline genetic testing is recommended in patients with metastatic PC no matter what their PSA level is.
It has been known that there is a clinically distinctive aggressive variant prostate cancer (AVPC), presenting with low-serum PSA, neuroendocrine small cell or ductal histology, which frequently metastasizes to visceral organs and rapidly develops castration resistance. Yet there is substantial genomic and transcriptomic heterogeneity that exists within these tumors [39]. Our paper addresses the contention that BRCA2-mutated metastatic PC fits in the criteria of AVPC, providing further understanding on this disease entity.

Materials and Methods
Patient data: The YUHS Big-Data team identified cases of prostate cancer patients who had performed next generation sequencing on their achieved prostate cancer tissues (prostatectomy, biopsy or transurethral resection specimen) between January 2019 and May 2021. Associated clinical and pathological information was retrieved from electrical medical records. The study design was approved by Severance Hospital Institutional Review Board (IRB #4-2020-0812). Patient consent was waived due to the retrospective nature of the study. All data underwent deidentification process.
Panel Sequencing of Prostate Cancer Tissues: Targeted DNA and RNA sequencing were performed using TruSight Tumor 170 or 500 (Illumina, San Diego, CA, USA). Briefly, 40 ng of formalin-fixed paraffin-embedded (FFPE) tissue-derived DNA and RNA were extracted using QIAGEN AllPrep DNA/RNA FFPE Kit (Qiagen, Hilden, Germany). After hybridization capture-based target enrichment, paired-end sequencing (2 × 150 bp) was performed using a NextSeq sequencer (Illumina) according to the manufacturer's instructions. Variants with a total depth of at least 100× were included for analysis. Variant interpretation was based on recommendations from the Association for Molecular Pathology (AMP), American Society of Clinical Oncology (ASCO) and College of American Pathologists (CAP) [40]. AMP/ASCO/CAP Tier 1 variant included level 1 and level 2 genetic alterations that are FDA-approved biomarkers and standard of care. Tier 2 variant included alterations with compelling clinical or preclinical evidence to drug response. Variant interpretations followed the "Standards and guidelines for the interpretation of sequence variants" published by the American College of Medical Genetics and genomics (ACMG) [41]. Each variant was initially queried using Varsome (Varsome.com) which annotates variants by combining data resources including AACT Clinical Trials from clinicaltrials.gov, Pharmacogenomics Knowledge Base (PharmGKB), COSMIC by the Sanger Institute, Polyphen-2 scores by Harvard University, CADD scores by the University of Washington, Clinical Pharmacogenetics Implementation Consortium (CPIC), the Drug-Gene Interaction Database (DGIdb) and OMIM [42]. Output variant annotation was manually reviewed in OncoKB website (http://www.OncoKB.org) (last accessed on 30 September 2022).
Statistical Methods: Cases were grouped by their germline or somatic alteration status of BRCA2 (mutation, deletion). Serum PSA at diagnosis and tumor mutation count were compared between BRCA2-altered group and wild-type from each study. Baseline characteristics, including Gleason grade group, TNM stage and other gene mutations were compared between the two groups using Pearson's chi-squared test or Fisher's exact test. Continuous variables of age at diagnosis and PSA at diagnosis were compared using t test with Welch's correction. Tumor mutation burden or total mutation count were compared using Mann-Whitney test. All p values were two-sided and p < 0.05 was defined as statistically significant. Statistical analysis was performed using SPSS 26.0 and GraphPad Prism 9.2.0.  Institutional Review Board Statement: The study design was approved by Severance Hospital Institutional Review Board (IRB #4-2020-0812).
Informed Consent Statement: Patient consent was waived due to the retrospective nature of the study. All data underwent deidentification process.
Data Availability Statement: All processed genomic, pathologic and clinical data are presented in Supplementary Tables. Original genomic data will be provided upon request. The link and location of public data used in this paper is described in method section.

Conflicts of Interest:
The authors declare no conflict of interest.