Therapeutic targeting of the DNA damage response in prostate cancer

Abstract

Purpose of review

The present article highlights the most common DNA repair gene mutations, using specific examples of individual genes or gene classes, and reviews the epidemiology and treatment implications for each one [with particular emphasis on poly-ADP-ribose polymerase (PARP) inhibition and PD-1 blockade].

Recent findings

Genetic and genomic testing have an increasingly important role in the oncology clinic. For patients with prostate cancer, germline genetic testing is now recommended for all men with high-risk and metastatic disease, and somatic multigene tumor testing is recommended for men with metastatic castration-resistant disease. The most common mutations that are present in men with advanced prostate cancer are in genes coordinating DNA repair and the DNA damage response.

Summary

Although much of what is discussed currently remains investigational, it is clear that genomically-targeted treatments will become increasingly important for patients with prostate cancer in the near future and beyond.

INTRODUCTION

Prostate cancer is the most common noncutaneous cancer in men and is one of the most heritable cancers [1,2]. Mutations in the DNA repair pathways are among the most common, with approximately 8–10% of primary prostate cancers and 20–25% of metastatic prostate cancer tumors harboring mutations in DNA repair pathway genes [3]. When considering advanced castration-resistant prostate cancer (mCRPC), the lethal form of the disease, approximately 12–15% of men have germline mutations in DNA damage response genes [4,5]. Because of this, the National Comprehensive Cancer Network (NCCN) now recommends germline genetic testing for all men with node-positive or metastatic prostate cancer and also suggests considering somatic multipanel tumor testing for these patients [6]. The present article will review the most common DNA repair gene alterations found in prostate cancer and will discuss the epidemiology and what is known about response to existing treatments and novel therapies for patients with these mutations.

BRCA1, BRCA2 genes

BRCA2 mutations are the most prevalent of the DNA damage repair mutations found in prostate cancer (Fig. 1) [3]. The BRCA1 and BRCA2 proteins are members of the homologous recombination pathway to repair double-stranded DNA breaks [7]. These mutations are found more commonly in men with high Gleason grade groups, where the prevalence of somatic BRCA1/BRCA2 mutations in primary tumors with Gleason grade groups 3–5 ranges from 4 to 6% [8▪], and in metastatic castration-resistant cancers ranges from 10 to 15% [9]. Of note, in prostate cancer (unlike breast cancer), BRCA2 mutations are far more common than BRCA1 mutations. Among those with localized disease, men with germline BRCA1/2 mutations have higher-risk disease and subsequently worse outcomes [10]. There is conflicting evidence on the impact of germline BRCA1/2 mutations on response to standard systemic treatments used in prostate cancer. A retrospective review of 172 patients with metastatic castration-resistant prostate cancer (mCRPC) demonstrated improved overall survival and response to first-line novel hormonal therapy in men with BRCA1/2 and ATM mutations [11]; however, other studies have found that those patients have worse prognosis and shorter response to hormonal therapy [12,13]. The PROREPAIR-B study was a recent prospective study of 419 patients with mCRPC of whom 18 had germline BRCA1/2 mutations [14] This study demonstrated worse outcomes in germline BRCA2 carriers who received a taxane prior to novel antiandrogen therapy for mCRPC (compared to those receiving antiandrogen therapy first) and worse cause-specific survival with a median time of 17.4 months in the germline BRCA2 carriers compared to 33.2 months in noncarriers (P = 0.027) [14]. This perhaps suggests that BRCA2 carriers may do better when treated early with novel hormonal therapies rather than taxane chemotherapies.

FIGURE 1.

The prevalence and strength of evidence of novel treatment options based on genetic alterations in the DNA repair pathways in prostate cancer.

Poly-ADP-ribose polymerase (PARP) inhibitors cause cell death because of unrepairable DNA damage as the result of ‘synthetic lethality’ in tumors harboring mutations in homologous recombination repair genes [15]. There is increasing evidence of the effectiveness of PARP inhibitors in prostate cancer. In the TOPARP-A trial, all eight of the patients with BRCA1/2 pathogenic mutations had a response to olaparib, which was defined as radiographic RECIST response, PSA₅₀ response, or a decrease in circulating tumor cells to less than 5 [16]. A retrospective study of men with mCRPC treated with olaparib demonstrated a 76% PSA₅₀ response rate in the 17 men with germline/somatic BRCA1/2 mutations [17▪▪]. An interim analysis of the TRITON2 study demonstrated a 47.5% objective response rate in patients with BRCA1/2 mutations treated with the PARP-inhibitor rucaparib [18]. GALAHAD is a phase II study of the PARP inhibitor niraparib (once daily dosed medication in comparison to olaparib and rucaparib which are twice a day) [19]. This trial enrolled patients with biallelic DNA repair gene mutations, and demonstrated a 38% (5/13 patients) objective response rate and a 57% (13/23) PSA₅₀ response rate in those with biallelic BRCA1/2 mutations [19]. The largest prospective randomized phase III study of a PARP inhibitor in prostate cancer was the PROfound study [20▪▪]. This study randomized men with mCRPC who progressed on abiraterone or enzalutamide (2 : 1) to receive olaparib or the alternative novel antiandrogen. Cohort A included 245 men with BRCA1/2, or ATM mutations (57% of the cohort had BRCA1/2 mutations; 9% of the cohort had additional co-occurring mutations). The median radiographic progression-free survival time was 7.4 months in those treated with olaparib compared to 3.6 months in those treated with a novel antiandrogen (HR = 0.34; 95% CI, 0.25–0.47) [20▪▪]. This solidified the benefit of PARP inhibitors in men with mCRPC and BRCA1/2 mutations and it is likely that olaparib will gain FDA approval for use in men with mCRPC harboring BRCA1/2 mutations in the near future (Table 1). Interestingly, in a subset analysis of the PROfound study, the benefit of olaparib over androgen-directed therapy was not demonstrated in men with BRCA1 mutations (and was accentuated in those with BRCA2 mutations), potentially suggesting that BRCA2 is the key gene predicting and driving favorable response to PARP inhibitors in prostate cancer.

Table 1.

DNA repair genes mutated in prostate cancer, their role in the DNA repair process, and the therapeutic implications related to each gene

Gene(s)	Role in DNA repair process	Therapeutic implications
BRCA1, BRCA2	Homologous recombination (effector of double-strand DNA break repair)	Positive phase III clinical trial (PROFOUND study) of olaparib for mCRPC and supportive results from additional phase II trials (TRITON, GALAHAD)
ATM	Homologous recombination (sensor of double-stand DNA breaks)	Limited response to PARP inhibitors Possible benefit with ATR inhibitors - clinical trials ongoing (NCT03787680, NCT03517969, NCT04095273, NCT03188965)
CDK12	DNA damage response (via DNA replication-associated repair)	Possible benefit with immune checkpoint blockade- clinical trials ongoing (NCT03570619, NCT04104893, NCT04019964)
MSH2, MSH6, MLH1, PMS2	Mismatch repair	Pembrolizumab is FDA approved for use in MMR deficient solid tumors; possibly more responsive to other immune checkpoint inhibitors as well
POLE, POLD1	Nucleotide excision repair, base excision (DNA polymerases)	Possibly more responsive to immune checkpoint blockade
CHEK1, CHEK2	Cell-cycle checkpoint regulation	Possibly sensitive to PARP inhibition

Checkpoint inhibitors have also shown benefit in many tumor types, although unfortunately not in unselected patients with advanced prostate cancer [21–25]. Subset analyses of larger trials have potentially suggested a greater benefit to single-agent PD-1 inhibitors in men with BRCA1/2 mutations. KeyNote-199, a phase II study of pembrolizumab monotherapy, demonstrated a 12% objective response rate in men with BRCA1/2 or ATM mutations which was higher than the 4% objective response rate seen in men without those mutations [25]. In CheckMate-650, a phase II study of the combination of ipilimumab plus nivolumab in 78 men with mCRPC, three of the five men with homologous repair deficient tumors (of which BRCA1/2 mutations are likely to be the most prevalent) had objective responses [26]. Similar findings were observed in a second smaller study [27]. Although these data are promising, more research is needed to determine if and when patients with prostate cancer with BRCA1/2 mutations should be treated with immune checkpoint inhibitors.

ATM gene

ATM mutations are present as germline alterations in approximately 2–3% of men and as somatic mutations in approximately 4–8% of men with prostate cancer (Fig. 1) [3,28,29]. The ATM protein functions primarily as a sensor of double-stranded DNA breaks leading to cell-cycle arrest rather than an effector of repair like BRCA1 and BRCA2 [29]. As mentioned above, there are conflicting data on the impact of ATM mutations on prognosis and response to standard and novel hormonal therapies [11,13,14].

Although ATM-mutated patients are typically grouped together with patients with BRCA1/2 in the prostate cancer literature, there is an increasing appreciation that response to therapy may be different in this subset. This is becoming evident in the literature around response to PARP inhibitors, where it seems patients with ATM mutations are not as likely to respond to PARP inhibitors compared with patients with BRCA1/2 mutations (Fig. 1). One retrospective study of 23 consecutive men with mCRPC who were treated with off-label olaparib found no responses among the six patients with ATM mutations [17▪▪]. The TOPARP-B study of olaparib in men with mCRPC showed a 37% response rate (7/19 patients) in ATM-mutated men; however, in this study response included objective response, PSA₅₀ response, or CTC conversion [30]. In the TRITON2 study of rucaparib in men with mCRPC, none of 18 patients with ATM mutations had a PSA₅₀ response [31]. Similar findings have been reported in the GALAHAD study [19]. On the basis of these retrospective and prospective data, PARP inhibitors do not appear to be as beneficial in men with ATM mutations as they are in men with BRCA1/2 mutations. This has less to do with the notion that ATM-mutated cancers may not have frequent biallelic alterations, and probably more to do with the fact that ATM is not a mediator of homologous recombination repair (thus, not inducing synthetic lethality).

However, there may be other pathways by which ‘synthetic lethality’ can be achieved in patients with ATM mutations. One such approach is through inhibition of the ATR protein. There is early preclinical evidence that synthetic lethality can be achieved in cancers with ATM mutations by treatment with ATR inhibitors [32,33]. This has also been demonstrated in a phase I clinical trial using a clinical ATR inhibitor (Table 1) [34]. There are multiple ongoing clinical trials of ATR inhibitors in prostate cancer (NCT03787680, NCT03517969) and in all solid tumors (NCT04095273, NCT03188965). Whether patients with ATM mutations are particularly sensitive to these agents remains to be seen but is being actively explored.

CDK12 gene

CDK12-mutated cancers have emerged as a unique subset of prostate cancer with a novel genomic and immune signature that has implications for treatment. Biallelic CDK12 mutations are more prevalent in prostate cancers than in any other cancer [35▪]. The prevalence is estimated to be between 4% and 7% of mCRPC tumors and 1–2% of localized prostate cancers (Fig. 1) [35▪,36▪▪,37]. These loss-of-function CDK12 mutations result in focal tandem duplications and genomic instability which produce gene fusions and an increased burden of neoantigens and subsequent T-cell infiltration into tumors [36▪▪,37]. The increased immunogenicity of these tumors makes them potentially more susceptible to immune checkpoint immunotherapy (Table 1) [36▪▪,38].

There has been little published on the clinical outcomes of these patients in prostate cancer. One retrospective study of 58 CDK12-mutated men showed poor response to first-line treatment for mCRPC with abiraterone or enzalutamide, with a median progression free survival rate of 4.3 months (95%CI 2.6–6.0 months) and only a 47% PSA₅₀ response rate [39]. Of the 20 men in the study who received a taxane chemotherapy, median progression-free survival was 4 months (2.6–5.3 months) and only a 35% had a PSA₅₀ response [39].

CDK12 mutations are also thought to affect the DNA damage response pathway through other genes that play significant roles in DNA repair [40]. However, there is little clinical evidence that these patients with prostate cancer are susceptible to PARP inhibitors. In the largest retrospective review of patients with CDK12 mutations, none of the 11 men treated with PARP inhibitors had a response [39]. In the TRITON2 study using rucaparib, one of the 13 CDK12-altered patients had a PSA₅₀ response and none of the eight evaluable patients had objective responses [31]. Therefore, the data to date suggest a relative lack of benefit from PARP inhibitors in CDK12-mutated advanced prostate cancer.

More promising is treatment with checkpoint inhibitors for these patients. The first retrospective study to report outcomes for patients with biallelic CDK12 mutations was notable for a 50% PSA₅₀ response to anti-PD1 therapy (two of four patients) [36▪▪]. Another retrospective analysis of eight patients with CDK12-altered mCRPC showed a 38% (three of eight patients) response rate (either PSA₅₀ or radiographic objective response) to checkpoint inhibitor therapy [39]. Currently underway is the IMPACT trial: a multicenter prospective study of patients with mCRPC with biallelic CDK12 mutations treated with the combination of ipilimumab and nivolumab (NCT03570619). A second study is testing pembrolizumab monotherapy in a similar patient population (NCT04104893). A third trial is targeting CDK12-mutated men with biochemically-recurrent prostate cancer who are receiving nivolumab monotherapy (without androgen deprivation) as a noncastrating treatment strategy for their recurrent disease (NCT04019964).

Mismatch repair deficiency: MSH2, MSH6, MLH1, PMS2 mutations

MSH2, MSH6, MLH1, and PMS2 are the four key genes involved in the mismatch repair pathway that, when mutated or silenced, lead to a hyper-mutated phenotype because of microsatellite instability [41,42]. Mismatch repair deficiency (dMMR) is most commonly found in uterine and gastrointestinal cancers; however, approximately 2–4% of prostate cancers harbor mutations in these genes (Fig. 1) [4]. Microsatellite instability is thought to be more prevalent among men with ductal adenocarcinoma [43], a histology which has also been found to be associated with mutations in other DNA repair pathways [44,45], and more common in men with primary Gleason pattern 5 (5+4 = 9 or 5+5 = 10) disease [46,47]. MMR deficiency was also shown to be enriched in patients presenting with pulmonary-only metastases in one study [48]. Patients who are found to have dMMR cancers should undergo germline genetic testing for Lynch syndrome, regardless of other pertinent positives or negatives from the family or past medical history [49], because one-third will harbor a germline MMR gene alteration. However, not all men with germline (or even somatic) pathogenic MMR gene mutations will develop tumors with microsatellite instability or a high mutational load, pointing to the complexity of this genotype–phenotype association [50].

Although patients with dMMR appear to have more aggressive disease features at diagnosis, it is unclear how these patients respond to standard systemic therapies for prostate cancer. There have been two retrospective studies describing outcomes of patients with mismatch repair-deficient prostate cancer [51,52]. One study of 13 patients demonstrated very long responses to hormonal therapies – with a median time to castration resistance of 67 months and a median response duration of 26 months to abiraterone or enzalutamide [51]. The other study of 32 patients demonstrated much shorter responses to hormonal agents – a median time to castration resistance of 8.6 months and mediation duration of response to first-line abiraterone/enzalutamide therapy for mCRPC of 9.9 months [52]. What both series agreed on is that these patients do have frequent responses to PD-1 inhibitors, with two of four patients in the first series and five of 11 patients in the second series achieving a PSA₅₀ response [51,52]. It should be remembered that the PD-1 inhibitor, pembrolizumab, is FDA approved for use in cancers with mismatch repair deficiency regardless of tumor type, and is a reasonable treatment option for these patients with dMMR mCRPC (Table 1) [53,54]. Prospective studies to explore the role of PD-1 inhibitors in patients with prostate cancer with MMR deficiency or microsatellite instability are currently underway (NCT03061539, NCT04104893, NCT04019964).

POLE, POLD1 genes

POLE and POLD1 polymerases are genes encoding 3’–5’ exonucleases involved in DNA replication, repair, and recombination [55]. Somatic mutations in these genes create an ultramutated phenotype [56]. The prevalence of this ultramutator phenotype in prostate cancer (typically >100 mutations/Mb) is about 0.1–0.3% of tumors (Fig. 1) [57▪,58]. Emerging evidence suggests that these DNA polymerase gene mutations (especially those involving the exonuclease ‘proofreading’ domains) may serve as a biomarker for response to immunotherapy because of the very high mutational load which occurs in the absence of microsatellite instability [57▪]. To our knowledge, there has only been one case report of a POLE-mutated patient with metastatic prostate cancer that progressed rapidly despite androgen deprivation and chemotherapy [59]. This patient had a very high tumor mutation burden of 238 mutations/Mb, and achieved a complete and durable response to single-agent PD-1 inhibitor treatment [59]. Although patients with mCRPC with pathogenic POLD1 mutations have been reported in the literature [43], none have received immune checkpoint inhibitor therapy. There is an ongoing clinical trial treating patients with POLE and POLD1 mutations, without microsatellite instability, with a new anti-PD1 antibody, toripalimab (NCT03810339). This may represent another tumor-agnostic biomarker that may predict response to immunotherapy, albeit involving a very rare genomic subtype (Table 1).

CHEK1/CHEK2 genes

CHEK1 (Chk1) and CHEK2 (Chk2) are serine/threonine protein kinases that coordinate the DNA damage response through cell-cycle checkpoint regulation [60]. Germline mutations are more prevalent in CHEK2 than somatic mutations and are found in about 1–2% of patients who develop mCRPC [4,28], whereas germline/somatic CHEK1 mutations are even more rare (Fig. 1). Some of these germline alterations are Ashkenazi founder mutations (e.g. CHEK2 T367Mfs*15 and CHEK2 I157T) that are more common in certain ethnic groups [61]. At diagnosis, CHEK2-mutated patients seem to have more aggressive disease features, including higher Gleason scores [62]. Despite this, these CHEK2 mutations are not associated with an earlier age at death or inferior overall survival from the time of diagnosis [62]. There are no data that we are aware of on responses to approved hormonal agents in patients with these CHEK1/2-mutated cancers. Whether PARP inhibitors, or other targeted agents such as ATR inhibitors, may result in synthetic lethality in the tumors of these patients remains to be determined. However, preliminary data suggest a relatively modest response to PARP inhibitor agents in patients with CHEK2-altered mCRPC (Table 1) [20▪▪,63].

Other DNA repair genes

There are several other genes involved in DNA damage response that have been included in the previously mentioned clinical trials. For example, the PROfound study included ATM, BRCA1, BRCA2, CDK12, CHEK2 and BARD1, BRIP1, CHEK1, FANCL, PALB2, PPP2R2A, RAD51B, RAD51C, RAD51D, and RAD54L genes [20▪▪]. These additional genes were included based on their reported roles in homologous recombination DNA repair and theoretical responsiveness to PARP inhibitors. In this study, responses were observed in patients with RAD51B and RAD54L mutations. TRITON2 had a slightly different gene list that (in addition to ATM, BRCA1, BRCA2, and CHEK2) also included BARD1, BRIP1, CHEK2, FANCA, NBN, PALB2, RAD51, RAD51B, RAD51C, RAD51D, and RAD54L [18]. In this study, one patient with a BRIP1 mutation did have an objective radiographic response and a near-complete PSA response [18]. One of two patients with FANCA-mutated tumors also had a radiographic and PSA response [18]. The GALAHAD study has a shorter list of inclusive biallelic mutations of ATM, BRCA1, BRCA2, BRIP1, CHEK2, FANCA, HDAC2, and PALB2 [19]. Responses in PALB2-mutated prostate cancers have also been observed [19]. Whether all of these mutations confer a similar sensitivity to the FDA approved prostate cancer regimens or novel therapies (e.g. PARP inhibitors) has yet to be determined. Additionally, the best combinations of mutations with genomically targeted agents to achieve synthetic lethality is still being explored [64,65].

CONCLUSION

As more germline and somatic genetic testing is performed in the clinic, and more genomically targeted treatments enter clinical trials, there will be additional options for subsets of patients with prostate cancers that have certain genomic alterations. Although some prostate cancers seem to be susceptible to current therapies, newer therapies are needed for many others. Whether some of these targeted therapies can replace hormone therapy is an active area of research (NCT03413995, NCT03047135, NCT04019964), and more exploration will be needed on how to best sequence these therapies with currently approved treatment options.

KEY POINTS.

• Significant responses to PARP inhibitor therapy have been seen in men with prostate cancer and BRCA1/2 mutations.

• Men with prostate cancer and ATM mutations are less likely to respond to PARP inhibitors but clinical trials to exploit synthetic lethality with ATR inhibitors in these patients are underway.

• Prostate cancers with mismatch repair deficiency, and CDK12 or POLE/POLD1 mutations may respond to checkpoint inhibition.

• Prostate cancers with CHEK1/CHEK2 mutations may respond to PARP inhibitors and alternative strategies are also being explored for these patients.