Use of PARP inhibitors in ovarian cancer

PARP inhibitors are the first class of drugs to exploit a new concept in oncology – synthetic lethality. Alexandra Leary explains the rationale, reviews the trial evidence and clinical experience, and looks to their future possible use in subsets of ovarian cancers without the BRCA germline mutation.

This grandround was first presented by Alexandra Leary, from the Gustave Roussy Cancer Centre, Villejuif, France, as a live webcast for the European School of Oncology. Margaret Hutka, from St George’s University Hospital NHS Foundation Trust, London, posed questions raised during the presentation It was edited by Susan Mayor. The webcast of this and other e-sessions can be accessed at e-eso.net.

Ovarian cancer is a rare disease, and yet it is the fourth leading cause of cancer-related death among women, after breast, lung and colon cancers, which are much more common. The reason for the high mortality rate is that ovarian cancers tend to be picked up at an advanced stage. One of the enigmas associated with ovarian cancers is that they are initially very sensitive to chemotherapy, with response rates to first-line platinum of 70–80%. However, the prognosis is poor, with half of patients relapsing within two years. Until recently, the only ‘targeted’ therapy available was the anti-angiogenic agent bevacizumab, for which there are no predictive biomarkers.

An important point about ovarian cancers is that 12–15% are associated with germline mutations in either BRCA1 or BRCA2. The Cancer Genome Atlas (TCGA) has provided more information, showing that, beyond these two germline mutations there are also somatic mutations that are acquired uniquely in tumours, occurring in 5–7%. The BRCA1 and BRCA2 genes can also be lost due to epigenetic silencing via hypermethylation (11–13%). In total, this means that around 30% of ovarian cancers may have alterations in BRCA1 or BRCA2 (Nature 2011, 474:609–15).

This finding is important because BRCA1 and 2 are key effectors in DNA repair, as downstream proteins involved in the repair of double-strand breaks in DNA, mainly via homologous recombination. BRCA1- or BRCA2-mutated tumours lose one of these proteins, and their homologous recombination DNA repair system no longer works effectively, so double-strand breaks accumulate in the genome of the tumour cells. The impaired DNA repair mechanism probably explains their sensitivity to platinum-induced DNA damage. It also affects the genomic profile: the figure above contrasts the comparative genome hybridisation (CGH) profile of a genomically stable ovarian cancer with the multiple gains and losses throughout the genome in genomically unstable high-grade serous ovarian cancer.

BRCA mutations result in loss of expression or function of BRCA, affecting its role as a major DNA repair effector of homologous recombination. This poses the question of how to target the loss of a protein to treat a cancer. In other cancers, drugs have been developed to inhibit EGFR mutations, but these are oncogenic gain-of-function mutations, in contrast to the loss of function in BRCA-mutated ovarian cancers.

Ovarian cancer is the first cancer to exploit a new concept in oncology – synthetic lethality – which targets the loss of an entity. This is illustrated in the mechanism of action of PARP inhibitors in BRCA-mutated tumours (see figure p35). Cells have many ways of repairing their DNA. Homologous recombination is one of the major DNA repair mechanisms, but there are others, such as base-excision repair. This explains how mutated cancer cells survive, because they switch to another mechanism to repair their DNA.

PARP1 is a major mediator of base-excision repair. Blocking PARP1 in normal cells has no effect, because they switch to homologous recombination to repair DNA. Cells with BRCA mutations have lost homologous recombination, so blocking base-excision repair with a PARP1 inhibitor results in loss of DNA repair and cell death, or synthetic lethality. PARP inhibitors are the first example of drugs that target the loss of a gene suppressor. BRCA-mutated tumours are dependent on other DNA repair pathways, so PARP inhibition becomes synthetically lethal in the context of an inactivating BRCA mutation.

Olaparib in ovarian cancer

Olaparib was one of the first PARP inhibitors to be developed. Early studies in BRCA-mutated ovarian cancer demonstrated olaparib’s efficacy, with objective response rates (Lancet 2010, 376:245–51). Phase I and II studies showed clear olaparib activity, principally in ovarian cancers that were BRCA-mutated and/or tumours showing platinum-sensitive relapses (>6 months). These findings prompted the first large trial in patients with high-grade serous ovarian cancers with platinum-sensitive relapse responding to re-challenge with platinum-based chemotherapy (NEJM 2012, 366:1382–92). They were randomised at this point to maintenance olaparib or placebo. Overall results were positive, regardless of BRCA status, with a median progression-free survival (PFS) of 8.4 months with olaparib compared to 4.8 months with placebo (HR=0.35, P<0.001) (see figure overleaf top).

A subgroup analysis of patients with BRCA-mutated tumours (germline or somatic) showed even greater benefit with olaparib (median PFS 11.2 months vs 4.3 months, HR=0.18, P<0.0001) (see figure overleaf bottom). It was previously very rare to see such a dramatic impact with drug treatment in ovarian cancer. These results led to approval by the European Medicines Agency of olaparib for patients with platinum-sensitive relapsed high-grade serous ovarian, fallopian tube or primary peritoneal cancer associated with a deleterious BRCA1 or 2 mutation, which can be germline or somatic. One of the remarkable consequences was that somatic mutation analysis entered routine practice. Olaparib was the first targeted therapy associated with a genomic predictive biomarker approved in gynaecological cancers, representing a major step forward.

Olaparib is given as monotherapy at a dose of 400 mg twice daily, starting within eight weeks of the last platinum chemotherapy, avoiding too short an interval, to reduce the risk of cumulative toxicity. Patients should have monthly blood cell counts for the first 12 months and then periodically.

Olaparib is relatively well tolerated. The main side-effects are fatigue, nausea and vomiting, and anaemia. One factor that is problematic is that patients have to take eight capsules twice a day, giving a total of 16 per day. The first dose reduction is to 200 mg twice daily, followed by a second reduction to 100 mg twice daily.

An important message to give patients is that, unlike chemotherapy, the side-effects with olaparib are at their worst during the first three months of treatment, and then often improve. They are not cumulative. When patients start olaparib they may experience nausea and not feel very well, but reassuring them that these side effects should improve can help them continue with treatment, which is important for a drug that is taken long term. Efficacy is dose-sensitive, so it is important to support patients to maintain dose intensity.

The European approval of olaparib for use in platinum-sensitive, relapsed high-grade ovarian cancers in patients with BRCA mutations was given on the basis of a phase II study, conditional on completion of a second, larger phase III confirmatory study.

This was recently achieved in the SOLO 2 trial, which randomised patients with a germline or tumour BRCA mutation to olaparib or placebo maintenance therapy for two years following response to platinum-based chemotherapy.

Results reported at the Society of Gynecologic Oncology meeting (12–15 March, 2017) showed this much larger study confirmed previous findings, with a significant 14-month increase in median progression-free survival with maintenance olaparib, compared to placebo, based on investigator assessment (see figure p 37). The central radiological review showed even more positive results.

When should we test for BRCA mutations?

In the past, BRCA testing was used only for hereditary cancers and testing family members for mutations. This has changed following the recognition that BRCA status can have a therapeutic implication. We now know that 20% of high-grade ovarian cancer is associated with germline or somatic BRCA mutations.

Germline testing

BRCA germline testing was previously carried out based on family history, but would now be recommended in all high-grade ovarian cancers regardless of family history. We know that women without family history can be the index case, and have a BRCA germline mutation. It is important to test as early as possible, ideally at diagnosis, because we know that once the patients relapse, their BRCA status could have therapeutic implications.

Tumour BRCA testing

Tumour BRCA testing is beginning to be introduced into routine care. It may become a therapeutic emergency to test patients with confirmed BRCA germline wild-type status who relapse. Academic centres now do targeted next generation sequencing for BRCA on tumours. This should be considered in patients without germline mutations who relapse, to look for the 7% who have somatic mutations.

PARP inhibitors: beyond BRCA-mutated ovarian cancer?

Clinicians now have a drug that is rationally designed to target patients with ovarian cancer who have a deficiency in homologous recombination, identified by a mutation in BRCA. The next question is whether these agents could be used in patients with ovarian cancer beyond those with BRCA mutations. Impetus for this comes from early studies showing responses to olaparib in BRCA wild-type ovarian cancer. Some of the studies, although small, showed an overall response of 25%, but the question remains as to how to identify these patients.

The Cancer Genome Atlas also showed that, in addition to the 30% of patients with high-grade serous ovarian cancer who lose BRCA due to germline or somatic mutations or hypermethylation, there is a further subset of around 20% who have rare alterations in various other members of the homologous recombination DNA repair pathway, such as EMSY amplification (5–17%) or RAD51 loss (3–5%). Although these are all quite rare, together they account for a further 20% of high-grade serous ovarian cancers that could have deficiency in homologous recombination.

This may mean that the germline BRCA mutation is just the tip of the iceberg. We have now proved that somatic mutations occur as well, and now there is a wide range of other rare mutations that we can identify.

There are two methods of identifying BRCA wild-type ovarian cancers with homologous recombination deficiency: looking for rare mutations by carrying out targeted sequencing, or assessing the DNA damage scar of a tumour as a reflection of homologous recombination deficiency regardless of cause. A tumour that is unable to repair single- and double-strand DNA breaks accumulates DNA damage and has a very erratic genomic profile, in contrast to a tumour that is homologous recombination competent.

Identifying this profile would reveal tumours unable to repair DNA with deficiency in homologous recombination, which might respond to a PARP inhibitor regardless of the underlying cause.

Several studies have investigated this approach. ARIEL-2 set out to identify patients with BRCA wild-type ovarian cancer with defective homologous recombination that was sensitive to PARP inhibitors. Patients with relapsed high-grade ovarian cancer, regardless of BRCA mutation status, were biopsied before being treated with the PARP inhibitor rucaparib. The biopsy tissue was scored for homologous recombination deficiency, and the scores in patients responding to PARP inhibitors were compared with those in patients showing no response.

Results showed highest progression-free survival in patients with BRCA mutations (see figure p 38). There was some benefit in BRCA wild-type tumours with homologous recombination deficiency, with a response rate of 30%, but this was lower than that seen in BRCA-mutated cancers. Lowest response was found in those with low homologous recombination deficiency and BRCA wild-type tumours.

A second study with similar design, NOVA, also looked at this question with another PARP inhibitor, niraparib, in two groups of patients: those with germline BRCA mutations and those without (NEJM 2016, 375:2154–64). Results showed a benefit in progression-free survival of more than 15 months in patients with germline BRCA-mutated cancers treated with niraparib compared to those randomised to placebo (median PFS 21.0 vs 5.5 months, P<0.0001). Patients without germline BRCA mutations showed a six-month benefit (median PFS 9.3 vs 3.9 months, P<0.0001). Overall, the study showed that non-germline-mutated ovarian cancer also benefited from PARP inhibitors.

The remaining question is how to identify the subset of non-germline-mutated patients who benefit most from PARP inhibitors. An exploratory analysis of the NOVA study showed that, among patients with homologous recombination deficiency (as assessed by the Myriad HRD test), those with a somatic BRCA mutation had a progression-free survival hazard ratio of 0.27 with niraparib, compared with 0.38 in those who were BRCA wild type.

Interestingly, even patients without homologous recombination deficiency benefited, although the hazard ratio was less marked, at 0.58 (NEJM 2016, 375:2154–64). The authors came to the provocative conclusion that all ovarian cancer patients benefit from niraparib maintenance therapy, regardless of BRCA mutation status or HRD status (as measured by currently available tests).

In fact, the results of the NOVA trial suggest that expensive genomic characterisation may not be needed, as platinum sensitivity is a valid predictor of benefit from PARP inhibitors. The FDA has recently approved niraparib in this indication.

Summing this up, homologous recombination deficiency occurs in 50% of patients with high-grade serous ovarian cancers. The development of PARP inhibitors has resulted in the first targeted therapy in this cancer associated with a genomic biomarker – germline or somatic BRCA mutations – with a response rate of 50% to 80%.

Several PARP inhibitors have now been shown to be active in ovarian cancers: olaparib, rucaparib and niraparib. Recent studies have suggested PARP inhibitor activity is not limited to BRCA-mutated ovarian cancers, and work is underway to identify the subset of patients with BRCA wild-type and homologous recombination deficiency who could benefit from these agents, with potential factors being: high homologous recombination deficiency score, mutations in non-BRCA homologous recombination genes and platinum sensitivity.

PARP inhibitors in combination?

PARP inhibitors have been tried in combination with chemotherapy, but this proved difficult because of cumulative toxicities. Remarkable activity has been seen combining a PARP inhibitor plus an anti-angiogenic agent. A small phase II study of olaparib in combination with the VEGF inhibitor cediranib in patients with platinum-sensitive ovarian cancer showed high response rates and a median progression-free survival of 17.7 months with the combination, compared to 9.0 months with olaparib alone (Lancet Oncol 15:1207–14; see figure p 38 bottom). Subgroup analysis suggested the benefit of the combination was greatest in BRCA wild-type patients, indicating some synergy between these two approaches.

A large phase III European study, PAOLA 1, is currently investigating first-line maintenance therapy with bevacizumab alone or in combination with olaparib, regardless of BRCA status, in patients with high-grade ovarian cancer following first-line treatment with surgery and chemotherapy plus bevacizumab. More than 600 patients have been recruited and results are eagerly awaited.

Question & Answer session with Alexandra Leary

Margaret Hutka from St George’s Hospital, London posed the questions.

Question: How do you think testing for homologous repair deficiency (HRD) will be used in the future? Will this take over from BRCA testing?

Answer: For now, I don’t think the HRD scores that have been tested in studies are convincing enough to go into routine practice. While they are somewhat discriminating, they are probably not discriminating enough between responders and non-responders.

Q: Based on the ARIEL-2 study, considering HRD scores, how will we predict the effect of PARP inhibitors? What type of testing would you envisage at this point, as we have various options – BRCA, HRD or no testing at all?

A: I think we have to continue working on this in the current randomised studies that include a PARP inhibitor. Testing has to be both sensitive and specific. We want to make sure we include all potential responders. We need to analyse our data and compare tests until we find one that’s good enough for practice. For now we don’t have anything apart from BRCA testing that we can use in clinical practice.

Q: If a patient progresses while on a PARP inhibitor, do you stop the PARP inhibitor or continue?

A: Stop, as there is no data to support continued PARP inhibition with a subsequent line of chemotherapy, especially given overlapping toxicity concerns. The real question is whether there would be value to the re-introduction of a PARP inhibitor as maintenance in a patient previously exposed to a PARP inhibitor. A clinical trial opening very soon – OREO – will be asking exactly this question. Continuing treatment should only be considered in a patient who has shown a response initially. I wouldn’t re-use a PARP inhibitor in a patient who has progressed within three or six months. It will be interesting to know whether you can resensitise a patient with chemotherapy and then re-introduce a PARP inhibitor.

Q: What would you add to a PARP inhibitor after progression? Maybe an antiangiogenic?

A: Without a doubt, the antiangiogenic combination with a PARP inhibitor is very encouraging. We only have results from one study, but they were very positive. Is there a rationale for combining a PARP inhibitor with immunotherapy? There probably is. We don’t yet know whether or not BRCA mutated tumours are going to be the ones that are most sensitive to immunotherapy, but biologically there are some suggestions this could be possible, because they frequently demonstrate lymphocytic infiltration, are genomically unstable and may produce more neoantigens. Combining a PARP inhibitor with immunotherapy might enhance the antigenicity of a tumour; there might be a rationale for combining PARP inhibitors with immune checkpoint inhibitors.

Q: What about PD-L1 testing?

A: For now I wouldn’t consider PD-L1 expression as a predictor of response in ovarian cancer.

Q: Can you comment on predictive markers for antiangiogenic therapy?

A: I don’t think we have any. We keep searching for them and there has been a lot of work on circulating biomarkers, but they haven’t given reproducible results. We may have biomarkers for the combinations but that’s a different question.

Q: Thinking beyond ovarian cancer, endometrial cancer is increasing in incidence, it seems interesting to have results with PARP inhibitors in this tumour site?

A: Given the homology in the genomic profile of high-grade serous ovarian cancer, triple-negative breast cancer and serous endometrial cancer or certain grade 3 serous-like endometrial cancers, some endometrial cancers probably have HD deficiency and may respond to PARP inhibitors. We know that endometrial cancer can be associated with BRCA mutations. I think the next step will be to consider PARP inhibitors in the treatment of serous-like endometrial cancers.