Clinical Application of Poly(ADP-Ribose) Polymerase Inhibitors in High-Grade Serous Ovarian Cancer

To identify fully the ovarian cancer patient population that would benefit from poly(ADP-ribose) polymerase (PARP) inhibitors, predictive biomarkers are required. Additionally, better understanding of the toxicity profile is needed if PARP inhibitors are to be used in the curative, rather than the palliative, setting. The development of PARP inhibitors in phase I–III clinical trials, approval for these agents, mechanisms of resistance, and outstanding issues were reviewed.

Abstract

High-grade serous ovarian cancer is characterized by genomic instability, with one half of all tumors displaying defects in the important DNA repair pathway of homologous recombination. Given the action of poly(ADP-ribose) polymerase (PARP) inhibitors in targeting tumors with deficiencies in this repair pathway by loss of BRCA1/2, ovarian tumors could be an attractive population for clinical application of this therapy. PARP inhibitors have moved into clinical practice in the past few years, with approval from the Food and Drug Administration (FDA) and European Medicines Agency (EMA) within the past 2 years. The U.S. FDA approval of olaparib applies to fourth line treatment in germline BRCA-mutant ovarian cancer, and European EMA approval to olaparib maintenance in both germline and somatic BRCA-mutant platinum-sensitive ovarian cancer. In order to widen the ovarian cancer patient population that would benefit from PARP inhibitors, predictive biomarkers based on a clear understanding of the mechanism of action are required. Additionally, a better understanding of the toxicity profile is needed if PARP inhibitors are to be used in the curative, rather than the palliative, setting. We reviewed the development of PARP inhibitors in phase I–III clinical trials, including combination trials of PARP inhibitors and chemotherapy/antiangiogenics, the approval for these agents, the mechanisms of resistance, and the outstanding issues, including the development of biomarkers and the rate of long-term hematologic toxicities with these agents.

Implications for Practice:

The poly(ADP-ribose) polymerase (PARP) inhibitor olaparib has recently received approval from the Food and Drug Administration (FDA) and European Medicines Agency (EMA), with a second agent (rucaparib) likely to be approved in the near future. However, the patient population with potential benefit from PARP inhibitors is likely wider than that of germline BRCA mutation-associated disease, and biomarkers are in development to enable the selection of patients with the potential for clinical benefit from these agents. Questions remain regarding the toxicities of PARP inhibitors, limiting the use of these agents in the prophylactic or adjuvant setting until more information is available. The indications for olaparib as indicated by the FDA and EMA are reviewed.

Introduction

Ovarian cancer remains the most lethal of the gynecological malignancies in the developed world and an important cause of female cancer-related deaths. Typically, patients with ovarian cancer present at an advanced stage. Despite a 70%–80% response rate to first-line treatment with platinum-based combination chemotherapy, most patients will relapse and eventually develop platinum-resistant disease with a poor overall prognosis [1]. The most common subtype of ovarian cancer, high-grade serous ovarian cancer (HGSOC), is characterized by p53 mutation (96%) and genomic instability. Approximately one half of all HGSOCs display defects in the genes involved in the DNA repair pathway homologous recombination (HR), with mutations of BRCA1 and BRCA2 identified in 23% of cases [2]. Sporadic nonhereditary tumors with defects in HR display similar features to BRCA1/2 mutant tumors and have been described as exhibiting “BRCAness” [3]. Both BRCA1/2 mutation-associated tumors and BRCA-like tumors have higher response rates to platinum-based chemotherapy than non-BRCA ovarian cancers, with longer treatment-free intervals and improved overall survival rates, most likely owing to their inability to repair DNA damage [4, 5].

In view of the high prevalence of HR defects, HGSOCs might be the ideal population for DNA repair-targeted therapy with poly(ADP-ribose) polymerase (PARP) inhibitors. We reviewed the current evidence for the clinical application of PARP inhibitors in ovarian cancer treatment.

BRCA, BRCAness, and PARP Inhibition

It has been estimated that a single cell could experience up to 100,000 injuries to its DNA daily, arising spontaneously during normal DNA replication or from external environmental factors [6]. Faithful DNA replication is essential for life, and a number of DNA repair pathways exist to protect the integrity of the genome. Recognizing the importance of these pathways, the 2015 Nobel Prize in Chemistry was awarded to three scientists for their work in DNA repair: Thomas Lindahl, Paul Modrich, and Aziz Sancar, for their work on base excision repair (BER), mismatch repair, and nucleotide excision repair (NER), respectively [7]. Other pathways that are key in DNA repair include HR, nonhomologous end-joining (NHEJ), and translesion DNA synthesis (TLS). HR is responsible for repairing double-strand DNA breaks in the synthesis phase (S phase) of the cell cycle, where it uses the sister chromatid as a template to repair DNA. Therefore, HR is an error-free pathway. When the HR pathway is lost, by malfunction of BRCA1/2 for example, cells will then rely on the more error-prone NHEJ pathway [8].

Mutations of the Fanconi anemia (FA) pathway occur frequently in cancer and have been reported in 46.6% of ovarian cancer cases [9]. The FA/BRCA pathway is required for the repair of stalled DNA replication, coordinating key repair pathways of HR, NER, and TLS [10, 11] (Fig. 1). These pathways act together to excise damaged areas of DNA, such as cisplatin-induced crosslinks, and repair the resulting gap to allow replication to begin again. Loss of HR is therefore associated with sensitivity to DNA-crosslinking agents such as platinum agents and mitomycin C. Unrepaired DNA crosslinks give rise to double-strand breaks in response to injuries from endogenous reactive oxygen species or from exogenous ionizing radiation and chemotherapeutic agents such as anthracyclines and bleomycin [12]. The FA/BRCA pathway is active in the S phase of the cell cycle, with FANCD2 deubiquitination occurring at the entry to G₂ and colocalization with BRCA1 and RAD51 nuclear foci in S phase [13]. The activation of the FA complex in response to stalled replication forks has been shown to depend on the DNA-damage response kinase ataxia telangiectasia and Rad-3–related (ATR) and its binding partner ATR-interacting protein [14, 15]. Both BRCA1 and BRCA2 have roles in restarting stalled replication forks [16, 17].

Figure 1.

The Fanconi anemia/BRCA repair pathway. Following DNA damage, ataxia telangiectasia and Rad3-related kinase (ATR) and its binding partner ATR interacting protein (ATRIP) are activated, and in turn activate Fanconi Anemia complex 1. This ubiquinates FANCD2/FANCI, which then colocalize with other key repair proteins on the damaged DNA. Following repair of DNA, FANCD2/FANCI are deubiquinated allowing replication to proceed.

Abbreviations: ATR, ataxia telangiectasia and Rad3-related kinase; ATRIP, binding partner ATR interacting protein; Ub, ubiquinated.

In contrast, PARP is involved in the repair of single strand breaks (SSBs). Of the 17 members of the PARP protein family, PARP-1 is the most well-characterized [18]. In some cases, it has been reported that PARP-1 is required for BER. However, the loss of key BER proteins (e.g., APE1, XRCC1, or Polβ) are embryonically lethal in mouse models. In contrast, PARP-1^−/− mice are viable [19]. Detailed study of the mechanism of PARP-1 function demonstrated that PARP-1 was not required for BER to proceed; however, with the use of PARP inhibitors, BER was inhibited by trapping PARP-1 onto SSBs, leading to stalling of the replication fork [20].

PARP-1 binds to stalled replication forks [21] and has been postulated to be hyperactivated in the context of HR deficiency [22]. It was therefore proposed that in the context of HR deficiency, by the loss of BRCA1/2 function or other methods, PARP would be required for cell survival. Consistent with this idea, PARP inhibitor treatment (targeting PARP-1 and PARP-2) alone was found to result in cell death and tumor shrinkage specifically in BRCA1/2-deficient cancers, a process termed “synthetic lethality,” in which the loss of function of each protein is not fatal in itself but with the absence of function of both proteins, cell death results [23, 24] (Fig. 2). This concept has moved from the field of genetics to oncology, with the goal of discovering novel therapies and minimizing toxicity [25]. The finding of synthetic lethality in BRCA1/2-mutant tumors using PARP1/2 inhibitors was a landmark finding in oncology treatment, targeting cancer cells with the expectation of sparing normal tissue.

Figure 2.

Synthetic lethality between PARP inhibition and HR deficiency (e.g., in BRCA1/2-mutant cancer cells). PARP inhibition results in an inability to repair single strand breaks, and the Fanconi anemia/BRCA pathway is required to repair the resulting stalled replication forks. In the absence of functional HR, DNA cannot be repaired, and cell death results. In an HR competent cell, DNA is repaired normally, resulting in cell survival.

Abbreviations: HR, homologous recombination; PARP, poly(ADP-ribose) polymerase; SSB, single strand break.

PARP Inhibitors in Ovarian Cancer

PARP inhibitors in development are listed in Table 1. Iniparib was developed as a PARP inhibitor and demonstrated clinical responses combined with gemcitabine and carboplatin in triple-negative breast cancer [26]. A phase III trial was therefore pursued of metastatic triple-negative breast cancer, comparing gemcitabine/carboplatin combination chemotherapy with the addition of iniparib or placebo, with no significant difference found in overall or progression-free survival [27]. During the course of this trial, new data emerged, showing that iniparib did not inhibit PARP [28, 29]. Phase II studies in platinum-resistant ovarian cancer and squamous non–small-cell lung cancer proved negative [30, 31]. The disappointing results for iniparib might have negatively affected the class as a whole and delayed PARP inhibitor development [32].

Table 1.

PARP inhibitors currently in development

Selected Phase I Studies

Phase I studies of BRCA1/2-mutant tumors supported the antitumor activity of PARP inhibition [33, 34]. In the initial phase I study using the PARP inhibitor olaparib, the population enrolled was not restricted to BRCA1/2-mutant tumors, although the expansion phase was limited to germline BRCA1/2-mutation carriers. Patients with ovarian cancer constituted most of the enrolled patients, with 16 of 21 having a BRCA1/2 mutation. Responses to olaparib were seen only in patients with BRCA-mutant tumors, all of whom had received at least one previous line of chemotherapy [33]. The expansion phase of the trial then showed responses in germline BRCA1/2 patients in both platinum-sensitive and platinum-resistant disease (61.5% in the platinum-sensitive cohort and 41.7% in the platinum-resistant cohort). In those with platinum-refractory disease, no radiologic responses were seen [34]. Although clinically tolerated, the toxicities of olaparib suggested it was not as specifically targeted to tumor cells alone as hoped, with side effects common to DNA-damaging agents such as nausea (48% with any grade), fatigue (44% with any grade), and myelosuppression (8% with grade ≥3).

Before the recognition of the specific role of PARP inhibition in BRCA1/2-mutant tumors, PARP inhibitors had been recognized as chemo- and radiopotentiating agents [35]. It was demonstrated that the addition of PARP inhibitors could improve the response to temozolomide in the treatment of cerebral melanoma, lymphoma, and glioblastoma multiforme in preclinical studies [36]. The combination of PARP inhibition with doxorubicin or platinum agents has also been reported to increase sensitivity [37, 38].

However, the addition of a PARP inhibitor to standard dose chemotherapy has proved challenging. Combining olaparib with carboplatin/paclitaxel in standard doses led to high toxicity rates, with 71% of patients experiencing myelosuppression in a phase I study [39]. This myelosuppression was presumably due to PARP inhibitor-mediated sensitization of the bone marrow to chemotherapeutic agents or the additive cytotoxic effects of combining these agents. Long treatment delays due to myelosuppression made the combination difficult to deliver. With a lower dose of olaparib, however, it could be delivered safely in combination with weekly paclitaxel (which does not directly induce DNA damage) or using intermittent scheduling with lower doses of carboplatin/paclitaxel.

Combining olaparib with carboplatin/paclitaxel in standard doses led to high toxicity rates, with 71% of patients experiencing myelosuppression in a phase I study. This myelosuppression was presumably due to PARP inhibitor-mediated sensitization of the bone marrow to chemotherapeutic agents or the additive cytotoxic effects of combining these agents.

Selected Phase II Studies

Promising activity in the phase I arena led to several phase II trials of epithelial ovarian cancer. A randomized phase II study of olaparib monotherapy compared with pegylated liposomal doxorubicin (PLD) in germline BRCA1/2-mutant platinum-resistant ovarian cancer did not show any difference in median progression-free survival (PFS) when comparing the three arms of 200 mg of olaparib b.i.d., 400 mg of olaparib b.i.d., and PLD 50 mg/m² every 28 days [40]. The response rates were higher in patients receiving olaparib (25% of the 200-mg cohort, 31% of the 400-mg cohort) than in those receiving PLD (18%); however, this difference was not statistically significant. The higher than expected response to PLD was likely due to the BRCA1/2-mutant status of the patients enrolled in that trial, consistent with the improved sensitivity to anthracyclines reported elsewhere in germline BRCA1/2 mutations owing to an inability to repair anthracycline-induced DNA damage [41]. Patients receiving olaparib had fewer grade ≥3 toxicities (in particular, rash, palmar-plantar erythrodysesthesia, diarrhea, and stomatitis) and a higher quality of life score, supporting PARP inhibitor treatment in this population.

Having gained experience in combining olaparib with chemotherapy, a randomized phase II trial was designed using lower doses of chemotherapy (carboplatin area-under-the-curve 4 mg/ml/min) and short courses of reduced-dose olaparib (200 mg b.i.d. on days 1–10) combined with standard paclitaxel (175 mg/m²), followed by standard-dose olaparib (400 mg b.i.d.) maintenance therapy, compared with standard carboplatin/paclitaxel chemotherapy for patients with germline BRCA-mutant or sporadic platinum-sensitive HGSOC [42]. Even with the adjusted dosing schedule, the toxicities remained elevated in the olaparib arm (43% grade ≥3 neutropenia vs. 35% of patients receiving standard chemotherapy). However, a significant improvement in PFS in the olaparib combination arm compared with standard chemotherapy was noted (12.2 months vs. 9.6 months). The improvement in PFS with the addition of olaparib was more marked in germline BRCA-mutant disease, with the median PFS not reached in this subgroup at the time of reporting (compared with 9.7 months in the chemotherapy-alone arm). Despite the improved PFS, no significant difference in overall survival was detected between the treatment arms in the overall trial population or the BRCA-mutant subgroup.

A phase II study of the combination of the antiangiogenic agent cediranib with olaparib compared with olaparib monotherapy showed significant improvement in median PFS in the combination arm (17.7 months vs. 9.0 months in the olaparib-only arm) [43]. Again, that trial enrolled patients with platinum-sensitive disease and was not limited to BRCA-mutant tumors. The effect of cediranib was more marked in the BRCA wild-type tumors, which could partly be explained by a poorer response to olaparib single-agent treatment (median PFS 5.7 months vs. 16.5 months for the combination arm). In the BRCA-mutant tumors, the difference in PFS between the two arms was not statistically significant (19.4 months for cediranib/olaparib vs. 16.5 months for the olaparib-alone arm). However, that analysis was performed retrospectively in the BRCA-mutant subgroup, and the study was not powered to detect significance in this group. Given the toxicities involved with the addition of cediranib (23% grade ≥3 diarrhea, 41% grade ≥3 hypertension vs. 0% with olaparib monotherapy and 27% vs. 11% grade ≥3 fatigue), the activity of the cediranib/olaparib combination in BRCA-mutant tumors might not be substantial enough to justify the increased toxicity. It could be argued that the survival benefit in the patients with BRCA-mutant is derived from olaparib and from cediranib in the BRCA wild-type tumors. To address this, a stratification approach would be needed and the inclusion of a cediranib monotherapy arm for patients with BRCA wild-type disease.

Phase II Studies Resulting in Food and Drug Administration and European Medicines Agency Approval

In October 2014, the European Medicines Agency (EMA) granted marketing approval for olaparib as maintenance monotherapy for patients with platinum-sensitive BRCA-mutant (both germline and somatic) ovarian cancer. This was based on the findings of a large phase II randomized trial (study 19) of olaparib maintenance in the patients with ovarian cancer sensitive to platinum treatment who had received two or more previous lines of chemotherapy. The results were encouraging for PFS (median 8.4 vs. 4.8 months in the placebo group, as measured from completion of chemotherapy). However, importantly, the results did not show an overall survival advantage [44]. A preplanned retrospective subgroup analysis showed a greater improvement in PFS for patients with germline BRCA mutation (11.2 vs. 4.3 months in the placebo group) [45]. However, an overall survival advantage for olaparib was not found in the analysis, regardless of BRCA status.

Olaparib monotherapy was well tolerated, with the most common toxicities being fatigue (7% of patients experiencing grade ≥3), anemia (5%), and neutropenia (4%). Nausea was markedly more common in patients receiving olaparib (73% with any grade nausea vs. 32% in the placebo group). Given the high rates of nausea and the significant pill burden of olaparib (16 capsules daily), with possible compliance issues, attempts have been made to address this using a newer formulation of olaparib in a number of phase III trials [46].

The same study 19 data that resulted in EMA approval was not accepted by the U.S. Food and Drug Administration (FDA) for approval of olaparib maintenance treatment. The Oncology Drugs Advisory Committee (ODAC) in June 2014 had voted against approval of olaparib (11 to 2). The ODAC briefing document cited concerns regarding the rate of development of myelodysplastic syndrome/acute myeloid leukemia (MDS/AML) of 2.2%. (The incidence of MDS/AML following standard-dose adjuvant epirubicin/cyclophosphamide has been reported to be 0.37% within 8 years [47]. However, these data might not be relevant for comparison, because a greater incidence of MDS/AML has previously been reported in carriers of BRCA1/2 mutations [48].) They also cited concern regarding the pill burden of olaparib (and the untested nature of the newer formulation) and reported that the 7-month difference seen in PFS had resulted from an underperforming control arm [49]. This led AstraZeneca to amend their application, citing a new indication.

Subsequently, the FDA granted approval for olaparib in December 2014, in this case in the setting of olaparib monotherapy for patients with germline BRCA1/2-mutant advanced ovarian cancer who had received at least three previous lines of chemotherapy. The phase II trial (study 42) supporting this approval enrolled patients with germline BRCA-mutant tumors, including ovarian, breast, pancreatic, and prostate cancer [50]. Patients with ovarian cancer in that study were defined as platinum-resistant (although FDA approval is not limited to platinum-resistant disease). The median PFS in this heavily pretreated BRCA-mutant ovarian cancer population was 7.0 months. This was a single-arm trial; therefore, no direct comparisons were made with other potential treatments. However, options are limited in platinum-resistant disease; therefore, new treatments are welcomed in this setting. The toxicities associated with olaparib monotherapy in this population remained similar to those previously described: 18.7% with grade ≥3 anemia and 61.7% with nausea of any grade.

Although olaparib has been the first PARP inhibitor granted FDA approval, rucaparib has been granted breakthrough status by the FDA. This might allow rucaparib to be fast-tracked to FDA approval. One of two phase II trials on which this breakthrough status was granted studied rucaparib monotherapy in patients with germline BRCA-mutant and platinum-sensitive HGSOC after two to four previous lines of chemotherapy [51]. Of the patients enrolled in the study, 65% had a complete or partial response to treatment (using the Response Evaluation Criteria in Solid Tumors [RECIST]), and the tolerability profile of rucaparib was similar to that of olaparib, although with transient transaminase increases noted in 40% of patients.

A further phase II trial, ARIEL2 (A Study of Rucaparib in Patients With Platinum-Sensitive, Relapsed, High-Grade Epithelial Ovarian, Fallopian Tube, or Primary Peritoneal Cancer), studied rucaparib monotherapy in patients with platinum-sensitive ovarian cancer (both HGSOC and endometrioid cancer) [52]. That study took the unique approach of using a biomarker (loss of heterozygosity [LOH]) to select sporadic BRCA-wild-type tumors with a defect in HR and detect tumors with somatic BRCA mutations (the presence of a BRCA mutation in the tumor but not in normal tissue). A response rate of 32% (using the RECIST) was reported in the LOH-high, HR-deficient group (compared with 66% in the germline and somatic BRCA-mutant tumors). In the LOH-low group, a response rate of only 11% was seen. Consistent with previous data, about one half of the BRCA wild-type tumors were LOH high. This finding has the potential to expand the population who would benefit from PARP inhibitors and is discussed in more detail below.

Phase III Studies

Hoping to strengthen the position of PARP inhibitors and maintain FDA and EMA approval of these agents, a number of phase III trials of interest have been started. Two large randomized placebo-controlled phase III trials (studies of olaparib in ovarian cancer [SOLO1 and SOLO2]) have completed recruitment, assessing the performance of olaparib maintenance monotherapy for platinum-sensitive germline BRCA-mutant ovarian cancer [53]. SOLO1 enrolled patients who had received one previous line of chemotherapy and SOLO2 after at least two previous lines of chemotherapy. These trials used a newer formulation of olaparib, a tablet form at 150 mg rather than the 50-mg capsule previously available. The pharmacokinetic profiles differ between these two formulations, and a dose of 300 mg b.i.d. was determined as the bioequivalent dose from two phase I trials [54, 55]. A third trial, SOLO3, assesses olaparib compared with the physician’s choice of therapy (single-agent therapy, nonplatinum). Patients eligible for SOLO3 will have germline BRCA mutations, platinum-sensitive disease, and received at least two previous lines of platinum-based chemotherapy (ClinicalTrials.gov identifier, NCT02282020).

Predictive Biomarkers for PARP Inhibitors

Although the application of PARP inhibition to germline BRCA1/2-mutant tumors has a clear rationale, it was proposed that tumors exhibiting a BRCA-mutant-like phenotype (BRCAness) could also be sensitive to PARP inhibition. BRCAness can arise independently of germline mutations of BRCA1/2, by somatic mutations of BRCA1/2 in the tumor tissue [56], by promoter hypermethylation leading to epigenetic silencing of BRCA1 (reported in up to 31% of ovarian cancers) [2, 57–59], or by defects in key components of the Fanconi anemia/BRCA repair pathway, including members of the FA complex [60, 61]. Additionally, amplification of EMSY, resulting in silencing of BRCA2, has been identified in 17% of sporadic ovarian cancers and is associated with genomic instability and a BRCAness phenotype [62, 63]. Given the varying methods by which function of the FA/BRCA pathway can be lost, no single mutation test will detect all patients with potential benefit from PARP inhibitors. Clinically relevant biomarkers designed to detect loss of function of HR and “BRCAness” are therefore of interest.

Given the varying methods by which function of the FA/BRCA pathway can be lost, no single mutation test will detect all patients with potential benefit from PARP inhibitors. Clinically relevant biomarkers designed to detect loss of function of HR and “BRCAness” are therefore of interest.

The Foundation Medicine LOH assay (Foundation Medicine, Inc., Cambridge, MA, http://www.foundationmedicine.com; in collaboration with Clovis Oncology, Inc., Boulder, CO, http://www.clovisoncology.com) referenced above in the ARIEL2/3 trials is promising—using the idea that tumors deficient in HR rely on the error-prone NHEJ pathway for repair, resulting in large-scale LOH. DNA is extracted from formalin-fixed paraffin-embedded (FFPE) tissue and next-generation sequencing performed. By assessing >3,500 single nucleotide polymorphisms (SNPs) and sequencing coverage, a LOH score is generated [64]. This assay can also differentiate between germline and somatic BRCA1/2 mutations using a computational approach [65]. Although this is promising in its ability to detect somatic mutations and loss of HR, it should be noted that the response rate was lower in those with BRCA wild-type/LOH-high tumors (32% vs. 66%), suggesting this does not reflect the loss of functional HR as accurately as BRCA1/2 mutation status.

A study of rucaparib as maintenance treatment and prospective validation of the LOH biomarker is ongoing in ARIEL3 (ClinicalTrials.gov identifier, NCT01968213). This trial is enrolling patients both with and without BRCA-mutant, platinum-sensitive ovarian cancer (including endometrioid tumors) who have received two or more previous lines of platinum-based chemotherapy to either rucaparib maintenance treatment or placebo. The LOH assay will be used as a stratification factor within randomization, but patients will not be selected for rucaparib on the basis of the findings from this assay. The primary endpoint of this trial is PFS as assessed in each of the subgroups (BRCA-mutant, BRCA wild-type LOH-high, and BRCA wild-type LOH-low). Therefore, the trial will not only assess rucaparib monotherapy but also the performance of the LOH biomarker and might enable the assay to meet the requirements for FDA approval.

Another biomarker in the pipeline is Myriad’s HRD assay (Myriad Genetics, Salt Lake City, UT, http://www.myriad.com). This is a combination of three scores using DNA extracted from FFPE samples—the first is again based on LOH analysis and profiling of BRCA1/2 [66]. A second score, telomeric allelic imbalance uses the idea of increased reliance on the NHEJ repair pathway as a model for allelic imbalance extending to the telomeres [67]. The third component measures large-scale state transitions—the determination of large chromosome breaks based on information gained from SNP array data [68]. The combination of the three scores, the HRD assay, has been developed from a breast cancer cohort [69], although further prospective validation studies are awaited.

The approaches taken by both these assays do not take account of possible reversion mutation of BRCA1/2, a secondary mutation that can restore function. These reversion mutations occur in up to 28% of previously treated ovarian cancer cases and are strongly associated with the development of resistance to platinum chemotherapy and PARP inhibitors [70]. Therefore, identification of a gene-expression pattern associated with functional HR deficiency could be of benefit. Taking this approach, the 44-gene DNA damage response deficiency assay was developed from a cohort enriched for BRCA1/2-mutant breast cancer and samples from Fanconi anemia patients [71]. This signature represents a specific immune response to abnormal S-phase DNA and has been validated as predicting the response to neoadjuvant and adjuvant anthracycline-based chemotherapy in breast cancer and the response to platinum-based treatment in ovarian and esophageal cancer [72, 73] using formalin-fixed samples. Validation of this assay in relation to PARP inhibitors is awaited. Another approach has been the development of functional assays that measure the ability of patient-derived tumor cells to develop DNA damage-inducible nuclear repair foci (a marker of HR-mediated DNA repair) following radiation treatment in the laboratory [74]. This has the advantage of providing a direct measure of tumor repair capacity but might be limited by the practicality of collecting fresh tissue and analyzing it using this relatively complex approach in a general hospital setting.

Mechanisms of Resistance

It was hoped that PARP inhibitors, being a new class of targeted therapy, would not be subject to the same mechanisms of resistance as other chemotherapeutic agents. However, this has proved not to be the case. Reversion mutations of BRCA1 and BRCA2 that restore HR-mediated DNA repair, referenced above, result in cross-resistance to platinum-based chemotherapy and PARP inhibitors [70, 75]. In addition, work using mouse models of BRCA1-deficient cancer (with complete absence of BRCA1 and therefore not subject to reversion mutations) has shown that long-term treatment with olaparib resulted in resistance, with upregulation of permeability glycoprotein efflux pumps (encoded by MDR1) identified as a mechanism [76]. This mechanism has previously been described in relation to doxorubicin and docetaxel [77]. Previous attempts to overcome this mechanism of resistance using small molecular inhibitors to the efflux pumps have proved disappointing in the clinic [78].

In addition to these two mechanisms, the loss of proteins involved in NHEJ, including 53BP1 and REV7, has been shown to be a mechanism of PARP inhibitor resistance in preclinical models by the restoration of functional HR [79–81]. In addition, tumors with loss of 53BP1 and acquired resistance to PARP inhibitors remained sensitive to cisplatin, offering an explanation for the lack of complete correlation between cisplatin sensitivity and sensitivity to PARP inhibitors in the clinic. A further mechanism, found in conjunction with 53BP1 loss, is the stabilization of a partially functional BRCA1-mutant protein through the inhibition of heat shock protein 90 [82]. Although these mechanisms of resistance to PARP inhibitors have been well described in preclinical models, their clinical incidence and relevance remains unknown at this point.

A newer PARP inhibitor, talazoparib, has been reported to be 100-fold more potent at trapping PARP-DNA complexes than other PARP inhibitors in preclinical studies [83]. It is hoped that this will result in a reduction of the acquired resistance mechanisms and in activity in tumors that have become resistant to other agents.

Conclusion

Although PARP inhibitors are clinically tolerated, they are not free of toxicities. In particular, the reported rate of 2.2% incidence of MDS/AML initially appears high. Phosphorylated-gamma-H2AX foci (a marker of double strand DNA breaks) have been identified in the normal tissue of patients receiving PARP inhibitor monotherapy [33], indicating these agents can cause untargeted DNA damage similar to other cytotoxic agents. It is important to recognize that an underlying BRCA1/2 mutation might predispose to the development of MDS/AML [48], and AML has been reported in association with BRCA2-mutant families [84]. A postmarketing requirement for annual reporting of all cases of MDS/AML related to olaparib has been mandated by the FDA. However, this should be balanced by the reporting of all cases developing in patients receiving placebo, which will provide a more accurate picture of the true underlying risk of MDS/AML in BRCA1/2 carriers who have received chemotherapy, and the effect of PARP inhibitors on this risk.

In a BRCA-mutant population with sensitivity to existing DNA-damaging agents, it might be that the decision to use a PARP inhibitor should be based on the toxicity profile compared with the toxicity of alternative available treatments rather than the supposed targeted nature of this therapy. In addition, we require more data on the extended cancer population that might benefit from these agents. Increasing evidence has shown that patients beyond those with germline BRCA mutations should receive these agents; however, biomarkers to identify this population require further validation. Overall, it is clear that PARP inhibitors represent an important addition to the current arsenal of antineoplastic therapies for the oncologist; however, better approaches for patient selection and an understanding of the long-term toxicities of these agents are required for their impact to be maximized in the clinic.

Author Contributions

Conception/Design: Eileen E. Parkes, Richard D. Kennedy

Collection and/or assembly of data: Eileen E. Parkes, Richard D. Kennedy

Manuscript writing: Eileen E. Parkes, Richard D. Kennedy

Final approval of manuscript: Eileen E. Parkes, Richard D. Kennedy

Disclosures

Richard D. Kennedy: Almac Diagnostics (E). The other author indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board