Homologous recombination, one of the major mechanisms of defective DNA repair, has emerged as a bona fide therapeutic target, yet its optimal use as a biomarker for patient selection remains a clouded scientific question.
Homologous recombination, one of the major mechanisms of defective DNA repair, has emerged as a bona fide therapeutic target, yet its optimal use as a biomarker for patient selection remains a clouded scientific question.1
Tumors with mutations in BRCA1 and BRCA2, breast cancer susceptibility genes that are quintessential members of the homologous recombination repair (HRR) pathway, are highly sensitive to platinum-based chemotherapy and PARP inhibitors.2 The FDA approved the first PARP inhibitor, olaparib (Lynparza), in 2014 for patients with germline BRCA1/2 mutations who have heavily pretreated advanced ovarian cancer,3 and the agency has since broadened indications for this class of agents in terms of disease settings and biomarker status.4-6
Germline testing for BRCA1/2 mutations has been available since the 1990s and is used both to assess breast and ovarian cancer risk and to guide use of PARP inhibitors in patients with cancer.7
Now a major clinical challenge is identifying the population of patients with wild-type BRCA1/2 who nonetheless will benefit from PARP inhibition. Assays measuring homologous recombination deficiency (HRD) caused by a broader range of mechanisms than BRCA1/2 loss are the central focus. In addition to germline BRCA1/2 mutations, such tests identify somatic mutations in BRCA1/2 and other HRR-related genes and detect the presence of genomic scars indicative of HRD.8-11
HRD tests have been clinically validated in several settings in patients with ovarian cancer in phase 2 and 3 trials of various PARP inhibitors. The results have consistently demonstrated that PARP inhibitors have greater benefit in HRD-positive patients, defined as those with mutations in HRR genes.12-17
Recent approvals for olaparib and the PARP inhibitor niraparib (Zejula) encompass patients with ovarian tumors that are HRD positive as well as those with BRCA mutations. In May 2020, the FDA also approved olaparib for patients with metastatic castration-resistant prostate cancer (mCRPC) with germline or somatic HRR mutations. The agency has approved 2 HRD tests as companion diagnostics: Myriad Genetics’ myChoice CDx for olaparib and niraparib in ovarian cancer and Foundation Medicine’s FoundationOne CDx for olaparib in mCRPC.18-20
However, most clinical trial findings have shown that even patients without HRD, as assessed by the current tests, benefited from PARP inhibition.12-17 Such findings leave many open questions regarding the clinical utility of HRD testing.
How a DNA Repair Pathway Functions
The rationale for targeting HRR activity stems from the frequent occurrence of defective DNA repair mechanisms in cancer. DNA is constantly subjected to potentially damaging assaults, necessitating a complex network of molecular signaling pathways of detection and repair that maintain genomic integrity.
The HRR pathway repairs double-stranded breaks (DSBs), a particularly toxic form of DNA damage that can trigger cell death if left unrepaired. Both strands of the DNA helix are compromised, so no complementary strand is available to serve as a template for repair. Homologous recombination instead uses complementary DNA within the sister chromatid for this purpose; thus, it is largely restricted to the S and G2 phases of the cell cycle, when newly replicated sister chromatids are readily available.
The HRR pathway is mechanistically complex, involving several related subpathways. Briefly, canonical DSB repair begins with recognition and resection of the free DNA ends at damaged sites to generate short stretches of single-stranded DNA (ssDNA). This step is highly dependent on the MRN (MRE11, RAD50, and NBS1) protein complex. Other key proteins include ATM and RPA, which binds and coats the ssDNA.
Another core component of the HRR pathway is the RAD51 protein, a DNA recombinase that searches for homologous sequences within the sister chromatid and ultimately uses them as a template to repair the damaged site through strand exchange. BRCA2 helps recruit RAD51 to the RPA-coated ssDNA, where RAD51 forms filaments made of nucleic acids and proteins. Two other central players in the HRR pathway, BRCA1 and PALB2, are responsible for recruiting BRCA2.21,22
Defects Promote Cancer
Defects in the HRR pathway cause the cell to become more dependent on other DNA repair pathways, including more error-prone mechanisms of DSB repair. This can lead to the accumulation of genomic alterations and foster genomic instability, which is a hallmark of cancer cells.2,7,11
Many types of cancer harbor defects in DNA repair. The best-known example is the link between mutations in the BRCA1 and BRCA2 genes and increased risk of hereditary breast and ovarian cancers. Germline mutations in BRCA1/2 are found in around 7% of patients with breast cancer (up to 15% of triple-negative breast cancers) and 13% of ovarian cancers.2,23
BRCA1 and BRCA2 are classic tumor suppressor genes that require 2 “hits”—inactivation of both alleles—to become oncogenic; thus, individuals who inherit a defective BRCA1/2 allele are at increased risk of cancer because they have only 1 functioning allele.24 A number of genetic tests for BRCA1/2 mutations, both single-gene and comprehensive gene sequencing panels, have been developed to identify at-risk patients and guide risk mitigation.7
For patients who already have cancer, the identification of BRCA1/2 mutations also has implications for the selection of targeted therapies. The dependence of BRCA1/2-mutant tumors on alternative pathways of DNA repair has been exploited for the development of PARP inhibitors.
PARP is an enzyme involved in the repair of single-stranded breaks via the base excision repair pathway (Figure).
Figure. Targeting DNA Repair Mechanisms
BRCA1/2-mutant cells are dependent on this pathway to prevent the formation of DSBs, and loss of PARP activity induces synthetic lethality—cells can endure loss of either PARP or BRCA1/2 but not both together. Thus, PARP inhibition should trigger cell death in a tumor-specific manner.2
Clinical trial results have repeatedly shown that breast and ovarian tumors with BRCA1/2 mutations are highly sensitive to both PARP inhibition and platinum-based chemotherapy, which induces interstrand cross-links that generate DSBs.2,4
Overall, the FDA has approved 4 PARP inhibitors: olaparib for ovarian, breast, pancreatic, and mCRPC indications; niraparib for ovarian cancer; rucaparib (Rubraca) for ovarian cancer and mCRPC; and talazoparib (Talzenna) for breast cancer. All have some indications involving BRCA1/2 mutations, mostly with Myriad Genetics’ BRACAnalysis CDx as the companion diagnostic.4,25
PARP Inhibitors Expand in Scope
Although the FDA granted initial PARP inhibitor approvals for later-line treatment of ovarian cancer, these agents more recently have found a place in the frontline setting as maintenance therapy in patients responding to platinum-based chemotherapy. PARP inhibitors have also been examined as maintenance therapy in platinum-sensitive recurrent ovarian cancer. In 2017, niraparib was approved by the FDA in this setting; however, the approval was notably independent of BRCA1/2 mutation status.26
In the pivotal phase 3 NOVA trial (NCT01847274), approximately two-thirds of the 553 enrolled patients did not have a germline BRCA1/2 mutation. Although patients with germline BRCA1/2 mutations (as determined by BRACAnalysis testing) derived a greater median progression-free survival (mPFS) benefit from niraparib (HR, 0.27), patients without germline BRCA1/2 mutations also experienced improved mPFS with niraparib compared with placebo (HR, 0.45).14
The findings from this clinical trial and others have demonstrated that a population of patients with BRCA1/2 wild-type ovarian cancer would benefit from PARP inhibition. The current clinical challenge is to find ways to identify those patients.11
A genome-wide study of high-grade serous ovarian cancer samples demonstrated that HRD is not solely due to mutations in BRCA1/2. HRD was identified in approximately half of the tumor samples in this analysis.27 In addition, up to 40.0% of breast cancers, around 30.0% of prostate cancers, and 15.4% of pancreatic cancers have defects in other HRR components.2,23,28
Other mechanisms of HRD include somatic mutations in BRCA1/2 and epigenetic inactivation of HRR genes, in addition to alterations in the expression or function of HRR pathway proteins, including PALB2, RAD51, and ATM. In vitro studies have demonstrated that many of these events are also determinants of PARP inhibitor sensitivity.11
Thus, the idea of expanding the assessment of tumor HRD beyond germline BRCA1/2 mutations is highly attractive. In addition to identifying alterations in individual HRR genes, other ways exist to potentially assess HRD in cancer cells. These include identification of mutational signatures, gene expression profiling, and functional assays.11
The most clinically advanced HRD measures are so-called genomic scar assays. These single-nucleotide polymorphism–based assays quantify chromosomal abnormalities indicative of HRD irrespective of the specific cause of the DNA repair defect.11
Loss of heterozygosity (LOH) is a common event in many cancer types, reflecting the second hit to a tumor suppressor gene that already has 1 defective allele. A group of investigators generated and tested an HRD score, defined as the number of LOH regions of intermediate size (> 15 Mb but less than a whole chromosome) in a tumor genome. They found a strong correlation between elevated HRD score and defects in known HRR pathway components (BRCA1/2 and RAD51).10
Cells with HRD often attempt DSB repair using an error-prone alternative mechanism, nonhomologous end joining, which can lead to allelic imbalance as a result of aberrant chromosomal end fusions that break apart during mitosis. A telomeric allelic imbalance (TAI) assay was developed that identifies large regions of allelic imbalance that extend to the telomere.9,11
Finally, HRD status can be determined by quantitation of a type of large structural chromosomal alteration called large-scale state transitions (LSTs), defined as chromosomal breaks between 2 adjacent regions that are at least 10 Mb in size.8
Two commercial genomic scar assays have been developed; Myriad’s myChoice CDx generates a genomic instability score from combined measurement of LOH, TAI, and LSTs. The score ranges from 0 to 100, with higher scores indicating greater HRD. The cutoff for HRD positivity was set at 42 or higher, as this was the 95th percentile of HRD scores observed in tumors with BRCA1/2 deficiency.29 Meanwhile, FoundationOne CDx produces an LOH score, ranging from 0% to 100%, with a prespecified cutoff of 16% or greater regarded as LOH high.30
Both myChoice and FoundationOne are next-generation sequencing–based tests performed using formalin-fixed paraffin-embedded tumor tissue. In addition to quantifying genomic scars, they can identify both somatic and germline mutations in the BRCA1/2 genes, although they cannot distinguish between them. The FoundationOne test further detects somatic alterations in 322 other genes, including many involved in the HRR pathway (Table18-20,29,30).29,30
Table. HRD Assays Approved for Patient Selection18-20,29,30
HRD Testing Comes Into Its Own
HRD testing has been validated in a growing number of clinical trials. The single-arm, phase 2 QUADRA trial (NCT02354586) evaluated the efficacy and safety of niraparib in patients with heavily pretreated (median 4 prior lines of therapy), platinum-sensitive ovarian cancer. A total of 463 patients were enrolled and treated with niraparib 300 mg once daily.17
The primary efficacy population consisted of 47 patients with HRD-positive tumors (as determined by the myChoice test) with 3 or 4 previous lines of therapy who were PARP inhibitor naïve and sensitive to their last platinum-based agent. This group had an overall response rate of 28% compared with 10% in the overall population and an mPFS of 5.5 months.17 On the basis of the QUADRA trial results, the FDA approved niraparib for the fourth-line or later treatment of patients with HRD-positive advanced ovarian cancer. The agency simultaneously approved the myChoice test as a companion diagnostic.19
In May 2020, the FDA approved olaparib in combination with the VEGF-targetingantibody bevacizumab (Avastin) as frontline maintenance therapy for patients with ovarian cancer in complete or partial response to platinum-based chemotherapy and who have HRD-positive status as determined by myChoice CDx.18 Approval was based on the phase 3 PAOLA-1 study (NCT02477644), in which 535 patients were treated with olaparib and bevacizumab and 267 received bevacizumab alone. In the entire patient cohort, mPFS was 22.1 months with olaparib and bevacizumab versus 16.6 months with bevacizumab alone (HR, 0.59). Subgroup analyses showed mPFS benefit for patients with BRCA1/2 mutations (37.2 vs 21.7 months; HR, 0.31), with wild-type BRCA1/2 (18.9 vs 16.0 months; HR, 0.71), and with wildtype BRCA1/2 and HRD-positive tumors (28.1 vs 16.6 months; HR, 0.43); however, HRD-negative patients had no mPFS benefit (16.6 months vs 16.2 months; HR, 1.00).15
Olaparib also recently received approval for treating patients with mCRPC who have progressed following abiraterone acetate (Zytiga) or enzalutamide (Xtandi) and have HRR gene mutations, as assessed using the FoundationOne and BRACAnalysis tests.20 Approval was based on the phase 3 PROfound trial (NCT02987543), in which HRD-positive patients were randomized to receive olaparib (n = 256) or investigator’s choice of enzalutamide or abiraterone (n = 131). Patients were divided into 2 cohorts: Cohort A included patients with at least 1 alteration in BRCA1/2 or ATM, and cohort B enrolled patients with alterations in any of 12 other HRR-related genes.31
The results showed that olaparib had a significant mPFS benefit over enzalutamide and abiraterone in cohort A (7.4 vs 3.6 months; HR, 0.34; P < .001) and both cohorts combined (5.8 vs 3.5 months; HR, 0.49; P < .001).31
Trial Results Highlight Questions
Not all clinical trials evaluating HRD-positive patients have proved to be such big wins for HRD testing. Similar to the NOVA trial of niraparib,14 the ARIEL3 trial evaluated rucaparib as maintenance therapy in the recurrent ovarian cancer setting (NCT01968213).12 Both trials included cohorts of patients evaluated for BRCA1/2 mutations (BRACAnalysis) and other types of HRD (myChoice for niraparib and FoundationOne for rucaparib). In both the NOVA and ARIEL3 trials, the HRD-positive group included patients with either germline or somatic BRCA1/2 mutations.12,14
Although HRD-positive patients derived a greater mPFS benefit, efficacy occurred in ARIEL3 regardless of HRD and BRCA1/2 positivity. As with niraparib, the FDA approved rucaparib in this setting irrespective of biomarker status.32
Several other recently reported clinical trials evaluated outcomes in patients with HRD-positive ovarian cancer treated with PARP inhibitors in the frontline maintenance setting. The phase 3 PRIMA trial (NCT02655016) assessed the efficacy of niraparib in this setting using myChoice CDx to identify HRD-positive patients. Among 733 patients randomized to receive niraparib or placebo, 50.9% were HRD positive. The mPFS was 21.9 months versus 10.4 months (HR , 0.43; P < .001) in HRD-positive patients and 13.8 months versus 8.2 months (HR, 0.62; P < .001) in the overall population.16 In April 2020, the FDA approved niraparib in the frontline maintenance setting irrespective of biomarker status26; nonetheless, Myriad has submitted a supplemental premarket approval application for myChoice.33
The phase 3 VELIA study (NCT02470585) evaluated veliparib in the same frontline maintenance setting after use as induction therapy in combination with chemotherapy. Patients (N = 1140) were randomized into 3 arms: chemotherapy induction and no maintenance, chemotherapy plus veliparib induction and no maintenance, or chemotherapy plus veliparib induction and veliparib maintenance.13
Within each arm, investigators assessed outcomes in patients with BRCA1/2- mutant disease, with HRD-positive disease (BRCA1/2 mutant or a myChoice modified cutoff of ≥ 33), and across the whole population. The longest mPFS was observed with veliparib (as both induction and maintenance) compared with placebo across all patients, regardless of biomarker status, although the HRD benefit was greatest in the biomarker-defined populations.13
Although HRD testing unequivocally identifies patient populations that derive significant benefit from PARP inhibition in ovarian and prostate cancers, HRD has far to go to become an effective predictive biomarker. Current tests still miss a significant number of potential responders, and the benefit frequently observed in biomarkernegative populations leaves an open question regarding whether PARP inhibitors are appropriate for all patients with ovarian cancer. Numerous clinical trials evaluating PARP inhibitors as monotherapy and in combination with other drugs in HRD-positive patient populations are ongoing and should provide further insight.
A significant issue with the effectiveness of current genomic scar assays is that they do not capture evolutionary changes within the tumor in response to selective pressure from cancer therapy. Restoration of homologous recombination proficiency can result and is the most common cause of platinum and PARP inhibitor resistance. Genomic scars are retained even if resistance pathways emerge to reestablish homologous recombination proficiency.11
Tumor genome testing using tumor tissue has several limitations, such as intratumor heterogeneity and insufficient sample volume. Several novel blood-based HRD tests are being explored in an effort to overcome these limitations. Epic Sciences’ CTC HRD assay is designed to measure chromosomal instability in circulating tumor cells isolated from the blood to select patients for a clinical study of the investigational PARP inhibitor pamiparib in patients with mCRPC.34