Poly(ADP-ribose) polymerase (PARP) inhibitors are life prolonging for patients with metastatic castration-resistant prostate cancer (mCRPC) harboring homologous recombination repair (HRR) gene alterations [1,2,3,4,5,6]. PARP is a DNA repair enzyme involved in the repair of single-strand breaks. When PARP is inhibited, single-stranded-DNA-breaks accumulate, leading to the accumulation of double-strand breaks (DSBs). A set of repair mechanisms are then activated, including homologous recombination repair (HRR). DSB repair relies primarily on HRR. Hence, aberrations within the HRR pathway, particularly germline or somatic deleterious mutations in Breast cancer gene-1 (BRCA1) BRCA1 or Breast cancer gene-2(BRCA2), can result in synthetic lethality in the presence of PARP inhibitors [7].

In recent years, PARP inhibitors have been combined with androgen receptor signaling inhibitors (ARSIs) in frontline mCRPC therapy with clear benefits for a subset of patients identified via genomic profiling. To identify those patients who are most likely to benefit, somatic and germline testing has been recommended as a standard of care by many society guidelines, including the National Comprehensive Cancer Network (NCCN) and American Urological Association (AUA) [8,9,10].

In this two-part expert opinion-based guide, we provide an expert consensus opinion on the utilization of germline and somatic testing to detect HRR alterations in patients with mCRPC. This guide was developed by a multidisciplinary expert panel that convened in 2023–2024, including representatives from medical oncology, urology, radiation oncology, pathology, medical genomics, and basic science. We argue for the widespread adoption of germline testing for all patients at the time of prostate cancer (PCa) diagnosis. Patients with metastatic hormone-sensitive PCa (mHSPC) or mCRPC would subsequently qualify for somatic tumor genetic testing to determine eligibility for potential life-prolonging precision therapies. Tumor genetic testing is potentially not necessary for men with only localized PCa.

In this second part, we highlight how genetic testing can lead to improved, life-prolonging mCRPC therapeutic strategies based on a review of the recent phase III trials and subsequent regulatory approvals for PARP inhibitors in mCRPC.

Implications of genetic testing results on clinical outcomes – the devil is in the details

Based on the comparison of primary prostate tumors and subsequent metastases at the time of castration resistance, DNA damage repair (DDR) mutations appear to occur early in the evolution of most cases of advanced PCa [11,12,13]. These mutations have profound implications for a patient’s course and response to therapies.

However, there are many questions regarding genetic testing results that are often overlooked by the broader community. These include:

1.

Does the presence of a BRCA versus non-BRCA HRR mutation impact treatment outcomes differently?

2.

Does the origin of a mutation (i.e. somatic vs. germline) affect outcomes?

3.

Does it matter whether one (monoallelic) or two (biallelic) alleles for a DDR gene are mutated?

4.

Does the nature of that mutation (e.g. deletions versus other types) affect treatment response?

5.

How do co-occurring additional genetic alterations impact response and resistance to treatment?

Such variables create a myriad of potential testing outcomes. However, patient numbers for any given testing outcome are often limited. Therefore, one must exercise caution when drawing conclusions for how any specific mutation signature may affect long-term outcomes.

First we will review the therapeutic implications of each issue in the context of the natural history of PCa and response to more conventional PCa therapies (i.e. androgen receptor signaling inhibitors, ARSIs/taxanes) [14]. We will then explore these issues in the context of PARP inhibitors.

HRR mutations associated with inferior outcomes BRCA versus non-BRCA mutations

Numerous studies have documented that BRCA2-disrupted tumors represent a unique and clinically relevant molecular subtype of aggressive PCa [15]. Given the aggressiveness and genomic complexity/heterogeneity of BRCA2-mutated disease, one may expect the presence of a BRCA2 mutation to lead to worse patient outcomes [3, 16, 17].

Recent next-generation-sequencing (NGS) studies have revealed that pathogenic germline mutations in DDR genes are present in 8%-12% of patients with metastatic PCa [18]. The most frequently mutated gene is BRCA2 (~5–6%), with BRCA1 and other DDR genes (e.g. Partner and localizer of BRCA2 (PALB2), Checkpoint kinase 2 (CHEK2), Ataxia-telangiectasia-mutated (ATM)each occurring in <1% of patients.

PROREPAIR-B [18] was the first prospective study to estimate how germline homologous recombination repair (HRR) alterations impact PCa-specific survival (PCSS). PCSS was halved in patients who were germline BRCA2 carriers compared with noncarriers (17.4 vs. 33.2 months P = 0.027). By contrast, there was not a significant difference in PCSS comparing ATM/BRCA1/CHEK2/PALB2carriers to noncarriers (23.3 versus 33.2 months, P = 0.264). Germline BRCA2 mutation is a particularly poor marker for patient outcomes.

A recent multicenter observational study of 729 PCa patients (“CAPTURE”) compared the outcomes of patients both with/without somatic and germline HRR mutations [14]. When comparing three different groups – BRCA-mutated vs. HRR-mutated-non-BRCA vs. non-HRR mutated – the radiographic progression-free survivals (rPFSs) were 7.1 vs. 9.0 vs. 10.7 months, with all differences significant. Similar significant trends were found in overall survival (OS) (18.4 vs. 21.9 vs. 29.1 months, respectively). These data reiterate the inferior outcomes of patients with BRCA mutations.

Somatic versus germline mutations

While patients with germline BRCA2 mutations are known to have a more aggressive disease course [13, 14, 19,20,21], most reports of HRR alterations in PCa comparing the prognostic implications of somatic versus germline alterations have been somewhat inconclusive. Many studies do not draw conclusions about how mutation origin impacts patient outcomes.

In CAPTURE, somatic mutations were more common than germline mutations in HRR genes overall (23.2% versus 7.4%). In both the BRCA cohort and the HRR-mutated non-BRCA cohort, somatic mutations outnumbered non-somatic mutations by ~3 to 1. Similarly, in the non-BRCA, HRR mutated cohort, somatic-only mutations outnumbered germline mutations (13.4% vs. 4.0% respectively). Of note, somatic BRCA1/2 mutations were significantly more common than germline BRCA1/2 mutations (9.8% vs. 3.4% respectively) [22]. Interestingly, patients with alterations in BRCA1/2 were more likely to have co-occurring non-HRR somatic or germline mutations (28.1%) compared to patients with other gene alterations (10.2%), likely indicative of genomic instability.

In sum, this data suggests the origin of a mutation (i.e. somatic tumor tissue versus germline) may does not have a major impact on its prognostic significance in metastatic prostate cancer.

Patients with germline BRCA2 vs. BRCA 1 mutations

Important differences exist between BRCA2 and BRCA1 mutated patients. In one study the prevalence of germline BRCA2 mutations was nearly 5x the rate of BRCA 1 (5.3 vs. 0.9% respectively) [23]. In the largest pan-cancer analysis of BRCA1/2 alterations published to date [24] (including 7186 prostate cancers), it was shown that germline mutations were present in 35% of patients with BRCA1-altered mCRPC compared with 50% of patients with BRCA2 altered mCRPC.

HRR gene zygosity

Two (biallelic) or one (biallelic) loss of function (LOF) alterations are possible for a given DDR gene. These alterations can come in many forms including allelic loss/deletion, frameshift insertions/deletions, and deleterious missense or truncating mutations. In the context of a pathogenic germline mutation impacting one allele, the tumor generally acquires a second oncogenic somatic ‘hit’ in the wild-type allele typically through deletion of the wild-type allele or duplication of the mutation, a process called loss of heterozygosity (LOH) which results in biallelic loss and presumed DNA repair deficiency. BRCA1 mutations are more likely to be bystander mutations in the context of high tumor mutational burden (TMB)/Microsatellite-high (MSI-high) disease and did not lead to HRR deficiency and as a result were more resistant to PARP inhibition [22, 25]. By contrast, BRCA2 mutations are more likely biallelic, leading to loss of function in HRR, and more responsiveness to PARP inhibition [25].

Zygosity is highly dependent on tumor purity and difficult to discern. The most careful studies found an 80% rate of biallelic BRCA inactivation in PCa [26]. Thus drawing conclusions based on zygosity is difficult. This is further complicated by the fact that most multi-institutional studies rely on the provision of archival tissue and most commercial NGS testing does not report zygosity or loss of heterozygosity.

Despite these limitations, in CAPTURE, biallelic mutations (14.7%) were approximately equal in prevalence to monoallelic mutations (15.9%). While biallelic alterations were slightly more frequent than monoallelic in the BRCA subgroup (56% versus 44% respectively), in the non-BRCA subgroup most patients presented with monoallelic mutations (58.3%). Despite some small differences in PFS and OS, in no study was there a significant difference observed for those with biallelic versus monoallelic alterations [14]. This suggests, with extreme caution about the interpretation of the results, that the zygosity of an HRR mutation may not have a major impact on responses to ARSIs/taxanes. However, the lack of difference in this study and others may be a function of inadequate identification of zygosity and testing artifacts rather than a real biological finding.

Prospective Clinical Trials

Clinical trials have confirmed that BRCA-mutated tumors have worse outcomes. Exploratory analyses of the control arms (i.e. ARSI only arms) of the three phase III trials evaluating the benefit of first-line ARSIs combined with PARP inhibitors (PROPEL [2], MAGNITUDE [4] and TALAPRO-2) [27] demonstrated that patients with BRCA-mutated tumors had a lower rPFS and OS than patients with non-BRCA HRR mutated and non-HRR mutated tumors. In addition, patients with all HRR-mutations, including BRCA patients, had worse responses to ARSI-monotherapy than non-HRR tumors [2, 4, 27].

In a study of 319 patients in which 7.5% were found to be DDR deficient, patients with germline DDR defects exhibited attenuated responses to AR-targeted therapy, with a median PFS of 11.8 months after starting ADT and a median time to PSA progression on first-line AR targeted therapy of 3.3 months (95% CI 2.7–3.9 months) [28]. Importantly, whether the mutations were germline/somatic or mono- versus biallelic did not alter these conclusions. Of note, patients with mutations outside of BRCA2 had similar treatment responses as patients without any HRR mutations [29, 30].

Therapeutic Sequencing

Studies have also tried to determine the optimal sequence of ARSIs and taxanes in the mCRPC treatment paradigm for HRR mutations. In PROREPAIR-B, PCSS for germline BRCA2 carriers was superior for those treated first with ARSIs followed by taxanes (P = 0.014) than the reverse. Of note, for non-HRR germline patients, both sequences were equivocal.

These data imply that optimal first-line mCRPC treatment options may be heavily informed by genetic testing, particularly for patients with BRCA2 mutations. Thus, not only do germline BRCA2 mutations have a detrimental impact on mCRPC outcomes, but BRCA2 mutations may benefit from exposure first to ARSIs followed by taxanes.

Summary of Implications for ARSIs and taxanes

Overall, patients with BRCA mutations had significantly worse treatment outcomes than non-BRCA HRR mutated or non-HRR tumors, regardless of whether mutations present were somatic or germline alterations. Suboptimal responses to ARSIs or taxanes may be a clue that a BRCA mutation (specifically BRCA2) is present and may prompt genetic testing.

PARP inhibitors in mCRPC

PARP inhibitors are the only agents that have a demonstrated survival benefit specifically for patients with HRR mutations, particularly BRCA2 mutated carriers. It is critical to identify these patients early so that PARP inhibitors can be made available to the patient.

Evidence for PARP inhibitors in later stage mCRPC

Multiple phase III clinical trials have demonstrated the efficacy of PARP inhibitor monotherapy in patients with mCRPC with HRR mutations after progression on ARSIs (Table 1). Rucaparib is FDA-approved to treat patients with mCRPC with deleterious germline or somatic BRCA1 or BRCA2 mutations and who had previously received an ARSI and taxane-based chemotherapy. Olaparib is FDA-approved for the treatment of mCRPC in a broader panel of 1 of 14 deleterious germline or somatic mutations in HRR genes after previously receiving an ARSI.

Table 1 Completed trials of PARPi monotherapy in mCRPC.

In the phase III PROfound trial [1, 31] olaparib was associated with longer progression-free survival (PFS) and improved clinical responses in patients with mCRPC who previously progressed on enzalutamide or abiraterone and had an HRR gene alteration, when compared to physician’s choice of enzalutamide or abiraterone. Patients were required to have somatic or germline HRR gene mutations and were allocated to one of two cohorts: Cohort A with BRCA1/2 or ATM mutations and Cohort B comprised of patients with a mutation in at least one of 12 other prespecified HRR genes.

At the time of the trial’s initial reporting in the entire population (cohorts A + B), rPFS was significantly longer in the olaparib arm (HR 0.49, 95% confidence interval (CI) 0.38–0.63, P < 0.001). While OS was also significantly improved with olaparib (HR 0.69, 95% CI 0.50–0.97, P = 0.02), it was only statistically significant for cohort A, but trended towards improved mOS in the overall population (HR 0.79, 95% CI 0.61–1.03) [1]. Of note, superior clinical benefit with olaparib versus control was found regardless of whether patients previously received taxanes, and the benefits were largely observed in the BRCAm subgroup, not the ATMm subgroup [32].

At the final analysis, updated PROfound data demonstrated that olaparib was associated with a significantly longer rPFS (HR 0.22) and OS (HR 0.63) than the control group [32]. This was particularly true for patients with BRCA2 homozygous deletions who experienced prolonged responses to olaparib (n = 16, median rPFS 16.6 months), with similar responses in both germline or somatic mutation [32].

TRITON3 found similar benefits when the PARP inhibitor rucaparib was compared to physician’s choice of chemotherapy or second-generation ARSI in chemotherapy naïve mCRPC patients with BRCA1/2 or ATM alterations. Radiographic PFS was significantly improved with rucaparib (HR 0.61, HR 0.47–0.80, P < 0.001), with a trend towards improved OS, but again largely observed in patients with mCRPC and BRCAm rather than ATMm [33, 34]. This data supports the use of a PARPi prior to a taxane in BRCAm mCRPC.

The NCCN guidelines recommend olaparib for patients with HRR-mutated mCRPC. Similarly, NCCN recommends rucaparib for patients with BRCA mutated mCRPC who have progressed on prior novel hormone therapy and/or chemotherapy. However, only olaparib has been shown to improve survival in a phase 3 trial and is thus the preferred PARPi for monotherapy use.

BRCA2 patients are most responsive to PARP inhibition

The specific HRR gene that is mutated impacts response to PARP inhibition. Efficacy in PROfound was primarily driven by patients with BRCA mutations, specifically BRCA2. BRCA2’s prevalence dramatically outweighs BRCA1 (90.6% versus 9.4% respectively of BRCA mutations). Patients with BRCA2 mutations experienced deep rPFS benefits (HR 0.22, 95% CI 0.15–0.32, P < 0.001) and OS benefits from olaparib (HR 0.59, 95% CI 0.37–0.95) whereas the HR for OS in all patients was not statistically significant (HR 0.79, 95% CI 0.61–1.03) [32]. The survival benefit was also not statistically significant in ATM patients. The small number of patients with mutations other than BRCA1/2 or ATM limits the trials’ ability to draw conclusions about these groups.

PARP inhibitor efficacy is diminished in BRCA1 versus BRCA2 altered mCRPC [22, 25]. One study pooled the efficacy data for a range of PARP inhibitors for mCRPC and compared the outcomes in those with deleterious BRCA1 versus BRCA2 mutations. Rucaparib’s activity seemed to be generally greater in patients with BRCA2 alterations than patients with BRCA1 alterations, as evidenced by PSA50 (60% versus 15% respectively), ORR (45% versus 33%) and median rPFS (9.7 vs. 8.7 months). This apparent discrepancy in PARP inhibitor sensitivity between patients with BRCA2 and BRCA1 mutated disease is not restricted to rucaparib but is a class effect of all PARP inhibitors.

This observed difference has been hypothesized to be explained by several factors:

BRCA2 alterations are more likely to be biallelic than BRCA1: In TRITON2, only 40% of BRCA1 mutations that were evaluable for zygosity status were biallelic alterations, whereas 85% of evaluable BRCA2 alterations were biallelic [35]. In a large pan-cancer BRCA1/2 analysis, biallelic inactivation occurred in only 50% of BRCA1 mutants versus 90% of BRCA2 mutants. Therefore, biallelic mutations in PCa seem to involve BRCA2 more than BRCA1. BRCA1 mutations are more likely to be passenger events in the setting of MSI-H, TMB-high disease [22, 25].

BRCA1 mutated cancers have more genomic co-alterations (e.g. in TP53 or PTEN) than BRCA2: TP53 alterations portend a worse prognosis in most malignancies. In TRITON2, patients with BRCA1 mutations had more frequent co-alterations in TP53 (62%) and PTEN (69%) compared with those who had BRCA2 mutations (42 and 29% respectively). Deleterious TP53 alterations were significantly more frequent in BRCA1 mutated versus BRCA2 mutated PCa.

Sensitivity to PARP inhibitors for non-BRCA HRR Mutants

PARP inhibition is less effective in patients with non-BRCA HRR gene mutations (Table 1). The non-BRCA non-ATM cohort in PROFOUND had neither a significant rPFS benefit nor OS benefit from olaparib compared with control [1, 31]. Similar non-significant results for non-BRCA patients were seen in other PARP inhibitor monotherapy trials, including TRITON-3 (rucapaib), GALAHAD (niraparib), TALAPRO-1 (talazoparib), and TOPARP-B (olaparib) [33, 36,37,38]. Similarly, patients with CDK12 mutations have also been found to have aggressive disease, with high rates of metastases and shortened OS [39, 40]. These patients have also been shown to not respond well to hormonal therapy, PARP inhibitors, or taxanes. However, the therapeutic implications of CDK12 mutations are still an active area of investigation [22, 24, 25].

Outside of some cases of sensitivity in PALB2 and RAD51 mutations, these trials all found very low response rates, short PFS and low PSA declines from PARP inhibitor monotherapy in non-BRCA2-mutated patients, including ATM, CHEK2, and CDK12. Expectations for PARP inhibitors are tempered for patients with such mutations.

In addition, there may be other mutations not currently identified in the canonical HRR gene sets. These include RNASEH2 loss MMS22L loss, and others [41, 42]. These genes and their therapeutic implications are active areas of investigation.

Effect of other variables on PARP inhibitor monotherapy efficacy

TRITON2 examined the activity of rucaparib according to genetic origin (germline or somatic), zygosity status (monoallelic or biallelic) and mutation type (homozygous deletion or other deleterious mutation). The efficacy of rucaparib was generally greater (although not always statistically superior) in patients with germline compared with somatic BRCA1/2 mutations (PSA50 response of 61% versus 51% with similar ORR estimates). Efficacy was also higher in biallelic versus monoallelic mutations (PSA50 response 75% versus 11% respectively; ORR 52% vs. 50%) and in patients with homozygous deletions versus other deleterious mutations (PSA50 response 81% versus 53%; ORR 60% versus 38%). Some of these trends, albeit non-significant, were also born out in TRITON3 [33].

Similarly, a recent post hoc analysis of PROfound attempted to determine whether these same issues affected efficacy outcomes. Olaparib was associated with prolonged rPFS and higher confirmed ORR for patients with either germline (rPFS HR 0.08) or somatic (HR 0.16) BRCA alterations compared with control. The numbers of evaluable patients were underpowered to make OS comparisons. Of note, similar results have been reported in the TOPARP-B study [38] of olaparib, and the rucaparib studies TRITON2 and TRITON3 [33, 35, 43]. Hence, although germline status identification is relevant for estimating patient and relatives’ cancer risk, whether a mutation was germline or somatic was not found to materially impact responses to olaparib in PROfound.

Of note, in PROfound patients with homozygous BRCA2 deletions (n = 16), which may include those that have both one germline and then a second-hit somatic mutation, experienced a significantly prolonged response to olaparib (16.6 m) compared to those with BRCA1 and/or ATM gene mutations. This is consistent with observations from TOPARP-B. This result suggests exceptional responses to PARP inhibition may be seen in patients with “two hit” events (inheriting a germline BRCA2 alteration and then acquiring a second somatic BRCA2 alteration later or possessing two aberrant somatic alleles). Similarly, patients with deletions, where secondary BRCA2 reversion mutations cannot emerge because of complete gene absence, may be exceptional responders to PARP inhibition [17, 44, 45].

However, response to olaparib was not restricted to biallelic BRCA mutations. While the ORR in patients with biallelic BRCA mutations was 60.7%, ORR in the heterozygous group (i.e. no evidence of a second hit loss) was still 44.4%. Tumor responses were also observed among the small number of patients with heterozygous loss and those where zygosity could not be determined (36% of cases). This data supports the idea that, regardless of zygosity, identification of any pathogenic BRCA alteration, particularly BRCA2, is sufficient to identify patients who may benefit from PARP inhibition – even without clear evidence of second HRR mutation inactivation events. These results are consistent with previous studies of BRCA alterations in other BRCA-driven tumor types, regardless of whether patients still benefited from PARP inhibitors regardless of whether BRCA alterations were biallelic or heterozygous “one hit” events [26, 35, 46].

A similar observation was made in TRITON-2 for individuals with both germline and somatic HRR mutations (so-called “double hit” patients). TRITON-2 evaluated rucaparib monotherapy in later-stage mCRPC [