This discussion focuses on PARP inhibitors specifically and targeted therapy more generally. In the past 6 years, there have been 13 new FDA-approved indications for ovarian cancer; 9 are for PARP inhibitors and 3 are for bevacizumab-containing regimens for first-line treatment and treatment of recurrent disease (both platinum resistant and sensitive). Here, I discuss when to use PARP inhibitors and targeted agents in treating ovarian cancer, including combinations and sequencing, as well as which biomarkers can help inform those decisions.

2020 has been an exciting year in ovarian cancer treatment. The use of PARP inhibitors has expanded to earlier lines of therapy. The combination of olaparib plus bevacizumab was approved for the maintenance treatment of patients with advanced ovarian cancer with HRD (either BRCA mutation or genomic instability) who are in CR or PR to frontline platinum-based chemotherapy based on findings from PAOLA-1.¹ The most recent PARP approval in the United States was niraparib in April 2020 for the maintenance treatment of patients with advanced ovarian cancer also in the frontline setting,² but unlike the combination of olaparib plus bevacizumab,¹ it is does not require BRCA mutation or other biomarker testing. In the EU, the EMA approved niraparib in first-line monotherapy maintenance treatment of ovarian cancer in September 2020, also regardless of BRCA mutation or biomarker status.

This table shows the current FDA indications for PARP inhibitors in ovarian cancer.^1-3 As mentioned, in the EU, olaparib plus bevacizumab has been approved, and the first-line use of olaparib maintenance therapy in the BRCA‑mutated population is generally available around the world. The latest approval from the FDA and EU of niraparib in first-line monotherapy maintenance therapy is for patients regardless of biomarker status. 

Despite the current enthusiasm for PARP inhibitors and bevacizumab, these agents do not remove the need for systemic chemotherapy, which remains important in the treatment of ovarian cancer. In the phase III ICON8 study (N = 1566), weekly dose-dense first-line chemotherapy was shown to be similarly effective as standard every-3-week carboplatin/paclitaxel and did not increase toxicity.⁴

Specifically, in this study, median PFS was 20.8 months with weekly paclitaxel vs 21.0 months with weekly carboplatin/paclitaxel vs 17.7 months with standard every-3-week treatment. Mean survival times for all 3 groups were approximately 25 months.

Cells require intact DNA to survive and have multiple strategies to repair both single-stranded and double-stranded DNA breaks. The role of PARP is to detect and repair single‑stranded DNA breaks and the role of the BRCA proteins is to repair double-stranded DNA breaks through homologous recombination repair (HRR). In HRR, broken DNA is aligned with its intact “sister” chromatin (ie, the other chromosome) and repaired, which is called homologous recombination. What this all means is that in a cell with deficient homologous recombination, such as with mutated BRCA, the use of PARP inhibitors inhibits repair of single-stranded breaks, which will turn into unrepaired double-stranded DNA breaks and result in cell death.⁵

It has been noted that patients with mutated BRCA2 are easier to treat than those with mutated BRCA1, as suggested by these preclinical cell survival graphs.⁶ These figures show a greater separation between mutated and wild-type cells with a higher concentration of PARP inhibitors, and the most benefit is in cells with BRCA2 mutations vs BRCA1 mutations. We have seen this clinically when treating patients with ovarian cancer, and a similar dynamic was seen following the approval of PARP inhibitors in the setting of prostate cancer.⁷

Approximately one half of patients with ovarian cancer have deficiency in double‑stranded DNA repair, that is, they are HRD.⁸ As demonstrated in this graphic, much of this comprises BRCA alterations (~ 25%), but also includes mutations or alterations in BRCA‑like genes (~ 10%) as well as epigenetic abnormalities (~ 15%).

The clinical relevance is that mutational analysis alone is insufficient, because it will miss the epigenetic phenomena. This is why the Myriad myChoice assay is recommended in the guidelines for biomarker assessment in ovarian cancer⁹—it assesses the entire genomic landscape.

As shown on the left in this figure, patients with mutated BRCA are the easiest to treat, in that they require the lowest PARP inhibitor concentration.⁶ Patients with HRD are in the middle, and the patients with the difficult-to-treat disease are those without any molecular signature. The 3 approved agents differ in their bioavailability, half‑life, metabolism, and other important ways.

Niraparib is the only PARP inhibitor dosed once daily, has higher tissue distribution (which is why it is studied in BRCA wild-type tumors¹⁰), and is the only PARPi with a personalized dose. But that does not mean that the other PARP inhibitors are a poor choice, just that they are different. In patients who are not candidates for bevacizumab, the first-line maintenance treatment option is limited to niraparib. Patients who are candidates for first-line maintenance bevacizumab therapy, if they also have HRD by the Myriad test, are candidates for olaparib plus bevacizumab maintenance therapy.

Although the doses of the 3 approved PARP inhibitors are different and the way they are metabolized is different, patients are monitored similarly regardless of the specific agent they are receiving.^1-3 This includes weekly complete blood counts. If the platelet count is < 100,000/µL, treatment should be paused to let the platelet count recover. In our clinic, we do this with every PARP inhibitor even though the only one that requires it is niraparib. We also routinely monitor vital signs in patients with ovarian cancer receiving PARP inhibitors.

These are oral medications, which means it is easy to pause treatment due to toxicity, which allows the patient to recover. By contrast, long‑acting intravenous medications have effects long after stopping treatment; for example, even after stopping paclitaxel, a patient’s hair does not grow back immediately and neuropathy and bone pain do not resolve quickly. This table shows recommended dosing modifications and monitoring for the 3 approved PARP inhibitors.

These are the key randomized clinical trials supporting PARP inhibitors as maintenance therapy in recurrent ovarian cancer: Study 19 (olaparib),¹¹ SOLO-2 (olaparib),¹² NOVA (niraparib),¹³ and ARIEL-3 (rucaparib).¹⁴ As shown here, median PFS with the PARP inhibitors ranged from 8.4-21.0 months vs 4.3-5.5 months with placebo. The NOVA study is the only clinical trial of a PARP inhibitor to evaluate efficacy in the germline BRCA–mutant and difficult-to-treat non-germline BRCA patient populations. In the germline BRCA group, PFS was 21.0 vs 5.5 months, respectively, whereas in the non-germline BRCA cohort PFS was 8.7 vs 4.3 months.

The PFS curves in SOLO2, NOVA, and ARIEL-3 are fairly similar, making it challenging to make decisions in overlapping indications for PARP inhibitors based on efficacy in BRCA-mutated ovarian cancer. As shown here, median PFS by blinded independent central review was 30.2 months for olaparib, 21.0 months for niraparib, and 26.8 months for rucaparib, with HRs of 0.20-0.27.^10,12-14-15

Clinicians need to base their choices on other issues such as patient convenience, clinician familiarity, toxicity profiles and patient comorbidities, and cost/insurance factors.

Of  importance, PARP inhibitors as maintenance therapy can even benefit patients with ovarian cancer who do not have mutated BRCA or other HRD. In phase II Study 19, the median PFS with olaparib was 7.4 months vs 5.5 for placebo in this group of patients.¹⁶ Likewise, in NOVA median PFS was 6.9 vs 3.8 months for niraparib vs placebo in these patients,¹⁰ and in ARIEL-3, median PFS was 9.7 months for rucaparib vs 5.4 months for placebo.¹⁷ For all 3, the HRs were approximately 50%.

With the standard cross-trial comparison caveat, the safety profiles are different among the 3 approved PARP inhibitors.^12,18,19 Of note, more patients discontinued or had dose reductions with rucaparib than olaparib or niraparib. Also, the niraparib safety is slightly misrepresented here as these data are for the fixed dose, which few clinicians use. Instead, in our clinic, we more often use the personalized dose for niraparib, which is 200 mg in underweight patients or patients with low baseline platelet counts.

In this retrospective analysis of the NOVA study of niraparib maintenance in recurrent ovarian cancer, most patients were started at the standard dose of 300 mg, but by Month 2, 200 mg was the most common dose.²⁰

SOLO1 was a randomized phase III trial that compared olaparib 300 mg twice daily to placebo in patients with newly diagnosed stage III/IV ovarian cancer and a CR or PR to platinum-based chemotherapy (N = 391).²¹ Patients were treated for up to 2 years or until no evidence of disease. Those with a PR could continue treatment beyond 2 years. The primary endpoint was PFS.

At 5 years, the median PFS was 56.0 months with olaparib maintenance vs 13.8 months with placebo (HR: 0.33; 95% CI: 0.25-0.43) with a median olaparib treatment duration of 24.6 months. In December 2018, the FDA approved olaparib in this setting.

However, despite these strongly positive results, olaparib in this study was only evaluated in patients with a BRCA mutation, and most ovarian cancer is not associated with a BRCA mutation. As a result, the PRIMA and PAOLA-1 studies were conducted to evaluate PARP inhibition in patients with ovarian cancer regardless of BRCA or HRD status.

PRIMA is an ongoing, randomized, double-blind phase III trial of niraparib 300 mg once daily vs placebo in patients with newly diagnosed ovarian cancer and a response to 6-9 cycles of first-line chemotherapy, and who are at high risk of recurrence (N = 733).^19,22 This includes stage III disease with residual tumor after debulking, inoperable stage III disease, or any stage IV disease.

Of importance, PRIMA is an “all‑comers” study, in which if the primary endpoint of PFS in HRD-positive patients was met, PFS in the overall population can be evaluated. Patients are stratified according to previous neoadjuvant chemotherapy, best response to first platinum therapy, and HRD status.

PRIMA enrolled 487 patients to the niraparib arm and 246 to the placebo arm. Patients mostly had a good performance status, with approximately 70% being Eastern Cooperative Oncology Group performance status 0. Approximately 66% had received previous adjuvant chemotherapy, and approximately 70% had achieved a CR.

The study included high-risk patients with poor prognoses: 85% had residual disease after debulking, 35% had stage IV disease, and many had persistent disease or only a PR after first‑line therapy. In other words, the high-risk group looked a lot like the patients that we see every day in the clinic who are in trouble. 

The HRD score is assessed using a 3‑pronged approach which looks not only at genetic mutations but also the DNA damage caused by those mutations.² Specifically, the genomic instability score includes loss of heterozygosity (LOH), which has been characterized as a genomic scar. The telomeres (chromosome ends) are very susceptible to DNA damage and telomeric allelic imbalance (TAI) can be assessed. Big chunks of DNA can be absent and not identified by LOH testing, and these are assessed as large‑scale state transitions (LST) in the chromosome. Taken together, the tumor BRCA status plus the composite genomic instability score (LOH plus TAI plus LST) provides the myChoice score. Of note, this test is available globally, not just in the United States. Of importance, if the score is > 42, this indicates that patients do not have a loss of heterozygosity, do have HRD, and experience similar improvement as patients with BRCA mutations.

In PRIMA, the median PFS was 21.9 months with niraparib vs 10.4 months with placebo, a significant difference corresponding to a 57% reduction in risk of relapse or death with niraparib (HR: 0.43; 95% CI: 0.31-0.59; P < .001).²² This was reflected in the PFS rates over time, with 59% in the niraparib arm free of progressive disease or death at 18 months vs 35% in the placebo arm.

By contrast, the HR for PFS with niraparib in SOLO-1 was 0.33. I would point out that HRD is very similar to BRCA in terms of helping guide treatment with PARP inhibitors, and I think it would be inappropriate not to consider PARP inhibitor therapy in the first-line setting if the patient has HRD. Clinicians may question the need to test all patients with ovarian cancer for HRD if they plan to use niraparib across the board. Particularly in Europe, molecular tumor testing is less available. That said, if it is reimbursed, I believe that clinicians should do HRD testing if possible.

PFS was significant in both the overall population and HRD subpopulation. A median PFS of 13.8 months vs 8.2 months with placebo (HR: 0.62; 95% CI: 0.50-0.76; P < .001) was seen in the overall population. In the HRD subpopulation, the median PFS was 21.9 months vs 10.4 months for niraparib and placebo, respectively.

PRIMA showed no new safety signals for niraparib, and the toxicity profile was similar to NOVA.²² Nearly all patients receiving niraparib experienced a treatment-emergent adverse event (AE), and 70% were at least grade 3. The most frequent grade 3 or worse AEs included anemia (31%), thrombocytopenia (29%), platelet decrease (13%), and neutropenia (13%). However, AEs led to treatment discontinuation in only 12% of niraparib-treated patients and no patients died of treatment-related AEs.

PRIMA prospectively demonstrated the benefit from being able to choose either the fixed starting dose of niraparib or a lower individualized starting dose. The individualized dose reduced the incidence of thrombocytopenia from 39% to 22%, anemia from 31% to 23%, and neutropenia from 21% to 15%.¹⁹ In fact, this was so effective that the 200-mg/day dose is now the approved starting dose for niraparib as first-line maintenance therapy for ovarian cancer.

Patient compliance with their medication regimen was assessed using 4 instruments for patient-reported outcomes: functional ovarian symptom index (FOSI), EORTC QLQ-C30, EORTC QLQ-OV28, and EQ-5D-5L.²³ Due to the ability to personalize the dose and manage the AEs, patient compliance rates remained above 80% throughout the trial. Overall, the message here is that the AEs from niraparib were manageable and did not significantly affect quality of life.

In PRIMA, the PFS with niraparib was very similar between patients with mutations in BRCA1 vs BRCA2, with HRs of 0.39 and 0.35, respectively.²⁴ This is important because in the BRCA‑mutated ovarian cancer patient population, clinicians must decide whether to use olaparib as in SOLO-1 (assuming bevacizumab is not an option) or niraparib, as in PRIMA.

With the usual caveats regarding cross-trial comparisons, it seems to me that when looking at SOLO-1 and PRIMA, niraparib may be more effective in the BRCA1-mutated subset of patients. However, what if I have a patient with a BRCA1 mutation who is not receiving bevacizumab? Either olaparib or niraparib are available, and these data suggest niraparib may be preferable. Of note, BRCA1‑mutated ovarian cancer is approximately 3 times more common than BRCA2‑mutated ovarian cancer and more difficult to treat.

Of importance, niraparib is similarly beneficial in patients with both deficient or proficient homologous recombination (HRD and HRP).²⁴ The HRs for HRD with mutated vs unmutated BRCA were 0.40 and 0.50, respectively, vs 0.68 for HRP. Of note, this slide shows 2 very similar versions of the PFS curves in each subpopulation, with one version adjusted based on imbalances in stratification factors.

Although the OS data were not yet mature at the time of presentation at ESMO 2019, a preplanned interim analysis did show a numerical superiority for niraparib over placebo at 2 years (11% mature).²⁵ In the overall population, OS rates were 84% vs 77%, respectively. In the HRD group (7% mature), OS was 91% vs 85%, and in the HRP group (16% mature), OS was 81% vs 59%.

As presented by Han and colleagues²⁶ at the 2020 Society for Gynecology Oncology annual meeting, niraparib improved time to first subsequent treatment in all patients on PRIMA. However, as with OS, these data are not yet mature. In the HRD group (38% mature), the HR was 0.46, vs 0.64 in the HRP group (59% mature) and 0.65 in the overall population (48% mature).

Likewise, PFS2 results were similar across groups and also not yet mature. In the HRD group (15% mature), the HR was 0.84, vs 0.56 in the HRP group (27% mature) and 0.81 in the overall population (20% mature). 

The PRIMA study included an exploratory analysis of the time between PFS and PFS2 (determined by subtracting the initial PFS time from the second PFS interval).²⁶ Of importance, niraparib had no detrimental affect on the efficacy of subsequent therapy, which has been a controversial subject. Overall, it does not look like resistance to a PARP inhibitor in ovarian cancer will affect response to the next therapy.

In 2020, based on results from the phase III PRIMA study, the FDA and EMA approved niraparib for the maintenance treatment of adult patients with advanced epithelial ovarian, fallopian tube, or primary peritoneal cancer who are in a CR o PR to first-line platinum-based chemotherapy.

PAOLA-1 was a randomized phase III trial evaluating the addition of olaparib to bevacizumab as maintenance therapy in patients with newly diagnosed, FIGO stage III-IV, high-grade serous/endometrioid ovarian, fallopian tube, or primary peritoneal cancer (N = 806).²⁷ PAOLA-1 differs from the other PARP inhibitor trials due to its active control arm. The study was conducted entirely in Europe because of access at the time to bevacizumab. All patients had previously responded to platinum- or taxane-based chemotherapy and at least 3 cycles of bevacizumab. The primary endpoint is PFS, with secondary endpoints including time to first subsequent therapy (TFST), PFS2, OS, and safety. Olaparib was given at 300 mg twice daily for 2 years, and bevacizumab was given at 15 mg/kg every 3 weeks for 15 months (including bevacizumab during chemotherapy).

As this graphic shows, the incidence of HRD in the PAOLA-1 study population was similar to the general population. However, olaparib in combination with bevacizumab was FDA approved as first-line maintenance therapy only for patients with ovarian cancer who have HRD.¹

I would personally not use olaparib with bevacizumab in this setting unless the patient definitely had HRD biomarkers. One question is what to do with a patient who is already receiving bevacizumab and then develops a BRCA mutation. Should the bevacizumab be stopped and niraparib started considering data from SOLO-1 and PRIMA? Or should bevacizumab be continued with added olaparib? In our clinic, we typically choose the latter so that patients can obtain the maximum benefit.

The PFS results from PAOLA-1 highlight that there was much greater benefit from the addition of olaparib to bevacizumab in patients with HRD²⁸: Median PFS with the combination in patients with HRD including tumor BRCA mutations was 37.2 months (HR: 0.33) and PFS with HRD but without tumor BRCA mutations was 28.1 months (HR: 0.43), whereas PFS in HRD negative/unknown was 16.9 months (HR: 0.92). Median PFS with bevacizumab plus placebo was approximately 17 months in all 3 groups.

The AEs seen in PAOLA-1 are hard to interpret, as at the time the trial was conducted the use of olaparib was not yet widespread in Europe and the management of AEs is better now. This is evident in that the AEs for bevacizumab (plus placebo) were less common and less severe than in ICON7²⁹ or GOG-218³⁰ because clinicians had become better at managing them.

In PAOLA-1, the most common AEs of any grade in the olaparib plus bevacizumab arm were fatigue (53%), nausea (53%), hypertension (46%), and anemia (41%). The most common grade ≥ 3 AEs included hypertension (19%), anemia (17%), and lymphopenia (7%).

As seen in this graphic, in this study, there was no difference in quality of life with the addition of olaparib to bevacizumab, which is important.²⁸

In 2020, based on results from the phase III PAOLA-1 study, the FDA approved olaparib plus bevacizumab for the maintenance treatment of adult patients with advanced epithelial ovarian, fallopian tube, or primary peritoneal cancer who are in a CR or PR to first-line platinum-based chemotherapy and whose cancer is HRD positive.

Currently in ovarian cancer, there are 9 indications for PARP inhibitors: 3 in first-line maintenance treatment, 3 in maintenance for recurrent disease, and 3 in later-line recurrent disease.^1-3 This class of drugs represents a wonderful advance for patients with ovarian cancer. As a physician, my goal in treating ovarian cancer is to use a PARP inhibitor for every patient ideally as early as possible to maximize the chance of success.

This treatment algorithm represents how I think about decision making in ovarian cancer, and reflects expert consensus.^{22, 28,31-33} First, every patient should receive germline testing, ideally a panel, because there are more than 30 hereditary breast and ovarian cancer genes that are important. If no germline BRCA mutation is identified, the tumor should undergo HRD testing. If the patient has a germline BRCA mutation, there is no reason to do an HRD test. Of note, it is not optimal to start with an HRD test as it will miss many mutations in hereditary breast and ovarian cancer genes.

After genetic testing, the next decision is whether to administer neoadjuvant chemotherapy or have primary debulking surgery. A laparoscopic assessment can be useful here. Many patients with ovarian cancer will receive first-line bevacizumab, unless they are early stage or if primary debulking left zero residual disease. At least one half of our patients with ovarian cancer are candidates for neoadjuvant treatment, including patients with stage IV disease and those with some residual disease after frontline debulking.

As mentioned, the majority (85% to 90%) of the patients with ovarian cancer we see in our practice will receive bevacizumab, including the settings of stage IV, neoadjuvant, and residual disease with primary debulking. Patients who have received bevacizumab should receive a HRD test, and those with BRCA mutations can receive olaparib in the maintenance phase (per PAOLA-1). If patients are HRP, continue treatment with bevacizumab alone.

If patients are not candidates for bevacizumab, they will very likely receive niraparib. However, most clinicians will still do an HRD test to confirm that patients are candidates for PARP inhibitors.

In patients with HRP, it is still my preference to use niraparib because there is an indication and some patients will be exceptional responders. Very few patients should receive observation only. My overall interpretation is that first-line maintenance treatment in ovarian cancer is part of the 2020 paradigm and is here to stay.

ASCO recently published guidelines for the use of PARP inhibitors in patients with ovarian cancer.⁹ I would like to highlight a few key points from these guidelines:

• Essentially all patients should be offered PARP inhibitor first-line maintenance therapy with niraparib.
• Patients with a BRCA mutation should be offered first-line maintenance olaparib.
• The combination of olaparib and bevacizumab may be offered as first-line maintenance therapy to patients with a BRCA mutation or HRD confirmed by the Myriad myChoice companion diagnostic.
• Avoid combining PARP inhibitors with chemotherapy and avoid retreatment with PARP inhibitors.

These are practical, helpful recommendations from an international panel of experts who treat patients with ovarian cancer. 

Today, many patients with ovarian cancer have been treated with PARP inhibitors and have become resistant to them. Managing that resistance can include restoration of homologous recombination activity, mitigation of the replication stress, or overcoming other resistance mechanisms (eg, transport).³⁴ These are exciting opportunities to overcome both primary resistance and developed resistance.