Kathleen Moore, MD:
We will turn our focus now to the combination of ICI and PARP inhibition that is under active investigation in multiple tumor types, but particularly gynecologic cancers—my specialty. ICIs have been approved as single agents for endometrial and cervical cancers, whereas PARPis are approved for ovarian cancers.^46-49 This combination has not been approved by the FDA but is actively being evaluated in phase III trials, which we will discuss shortly. Our focus here will be on managing toxicities associated with this combination.

Before we get to toxicity management, I do want to discuss several hypothetical mechanisms behind why we should try combining ICIs and PARPis. The first is that by using a PARPi in tumors vulnerable to DNA damage, the resulting tumor cell death releases neoantigens that interact synergistically with T-cell activation and recruitment promoted by ICIs.⁵⁰ The second and more likely hypothesis is that when a PARPi causes DNA breaks, these result in greater amounts of unrepaired nuclear DNA fragments in the cytosol. The cytosolic DNA fragments then stimulate the STING pathway, which leads to the upregulation of interferon-γ and additional signaling that acts synergistically with ICIs.

Kathleen Moore, MD:
We do have some early clinical data on the combination of ICIs and PARP inhibition, particularly in ovarian cancers. This combination, with or without bevacizumab, has been examined in 3 general categories of ovarian cancers: first, platinum-sensitive tumors; second, those that have at least a 6-month duration from the last platinum-based treatment; and third, platinum-refractory tumors. The results have been varied, with the most promising efficacy observed in the platinum-sensitive setting.

Most data come from the multicohort MEDIOLA trial, which included cohorts of patients with ovarian cancer and, to a lesser extent, BRCA-mutated HER2-negative metastatic breast cancer. These are small groups with approximately 30 patients per cohort, but they give us an interesting signal. MEDIOLA focused on durvalumab combined with olaparib with or without the antiangiogenic agent bevacizumab.

The triplet combination of durvalumab/olaparib/bevacizumab seemed to be the most effective, yielding a median OS of 31.9 months and a 24-month OS rate of 64.5% in the cohort of 31 patients with non–gBRCA-mutated platinum-sensitive, relapsed ovarian cancer.⁵¹ With the caveat that these numbers are very small, the triplet combination was associated with an overall response rate (ORR) of 87.1% in this cohort, with tumors considered homologous recombination repair (HRR) proficient. This is striking because we really do not expect PARPis or ICI/PARPis to work in the HRR-proficient population. That signal in the HRR-proficient setting has us hopeful that at least some of the ongoing first-line phase III studies evaluating this combination will be positive.

When you look at the doublet combination of durvalumab/olaparib in other MEDIOLA cohorts, the efficacy data are still reasonable. Among 32 patients with gBRCA-mutated platinum-sensitive, relapsed ovarian cancer, the ORR was 71.9%.⁵² But I will remind you that this was from a little earlier, when patients would not have had a prior PARPi—which is now a standard first-line option—so bear in mind that these are PARPi-naive tumors. Thus, an ORR of 71.9% in a platinum-sensitive, PARPi-naive setting is good, but you would expect PARP inhibition to work there. The key question is, would this combination be more active than PARP inhibition alone? That is hard to say without a control arm.

Turning now to the setting of platinum-refractory ovarian cancer with any BRCA status, we have data on the combination of pembrolizumab/niraparib from the phase II TOPACIO study and dostarlimab/niraparib/bevacizumab from the phase II OPAL study.^53,54 In this setting, the ORR was 18% with the doublet and 17.9% with the triplet. This suggests there is some characteristic of the platinum-resistant tumor, likely involving the tumor microenvironment, that does not lend itself to efficacy of ICI/PARPi combinations. These results are consistent with other studies evaluating olaparib or rucaparib as the PARPi paired with an ICI in the platinum-resistant setting. As much as we might hope that ICIs could restore PARPi efficacy in the platinum-resistant setting, that has not been observed.

In addition to these data in ovarian cancer, we also have preliminary data suggesting that the combination of ICI/PARPi has greater activity in PD-L1–high lung cancer.^55,56 The small phase II JASPER study assigned 38 patients with advanced, previously untreated non-small-cell lung cancer (NSCLC) harboring no activating alterations in EGFR, ROS1, or ALK to a cohort with high PD-L1 expression (PD-L1 tumor proportion score ≥50%) or low PD-L1 expression (PD-L1 tumor proportion score 1%-49%). Both cohorts received pembrolizumab/niraparib. With the caveat again that these are small numbers, the investigators reported a much higher ORR of 56.3% with 12.5% complete responses in the PD-L1–high cohort and a low ORR of 20.0% with no complete responses in the PD-L1–low cohort. These results really speak to the importance of the PD-L1 biomarker here. The disease control rate, median progression-free survival, and median OS were all better in the PD-L1–high cohort. It remains to be seen how this will move forward.

That being said, these studies provide important toxicity data to help guide how we care for patients enrolled on the ongoing first-line phase III studies of ICI/PARPis, and should those read out as positive and be approved, future patients may be treated with these combinations. 

Kathleen Moore, MD:
Why are we discussing toxicity management for this investigational combination? As you can see here, there are numerous ongoing phase III trials of ICI/PARPi regimens, particularly in gynecologic malignancies. If any of these read out as positive, then these combinations may be approved—making it important to understand the toxicities.

Let’s look first at the studies in ovarian cancer, of which 3 are done accruing and should be reading out in 2023-2024. The ANITA trial is evaluating chemotherapy plus atezolizumab followed by maintenance atezolizumab/niraparib in 414 patients with recurrent ovarian, primary peritoneal, or tubal carcinoma (NCT03598270). In the United States, the indications for PARPi monotherapy in recurrent platinum-sensitive ovarian cancer have been withdrawn.⁵⁷ If ANITA is positive, it will be interesting to see whether this opens the door for using an atezolizumab/niraparib combination in this space again. However, we should bear in mind that this trial is being performed in a PARPi-naive, platinum-sensitive population, which largely does not exist anymore now that PARP inhibition is standard in the first-line setting. Applying this study to our current patients will be somewhat difficult, but ANITA is still an important trial.

The rest are first-line studies. ATHENA‑COMBO is evaluating switch maintenance, where patients who respond to their first-line platinum-based chemotherapy are then randomized to receive rucaparib monotherapy vs nivolumab/rucaparib. ATHENA-COMBO has finished accruing and should read out at the end of 2024.

DUO‑O may look like the study design for ATHENA-COMBO, but it differs in that its randomized cohorts do not harbor tumor BRCA mutations, although it does have a nonrandomized cohort with tumor BRCA mutations (NCT03737643). In DUO-O, 1317 patients with advanced ovarian cancers are receiving first-line chemotherapy plus durvalumab with or without bevacizumab followed by maintenance with durvalumab monotherapy vs durvalumab/olaparib. DUO-O has finished recruiting and is the first of these trials to read out, with results being presented at ASCO 2023.⁵⁸

The FIRST trial is an all-comers study of first-line chemotherapy plus dostarlimab followed by maintenance with dostarlimab monotherapy vs dostarlimab/niraparib in 1403 patients with advanced ovarian cancer (NCT03602859). Bevacizumab also is permitted at the investigator’s discretion. FIRST is similar to DUO‑O, but it allows BRCA mutations. This trial has finished accruing and should read out in July 2023.

Wrapping up the ovarian studies, NItCHE-MITO33 is evaluating dostarlimab/niraparib in 427 patients with recurrent ovarian, fallopian tube, or primary peritoneal cancer who are ineligible for platinum-based chemotherapy (NCT04679064). This trial is recruiting and should read out in 2025.

Turning now to endometrial cancer, I have the privilege of leading the DUO-E trial with Dr Shannon Westin at MD Anderson Cancer Center. This trial is evaluating first-line chemotherapy plus dostarlimab followed by maintenance with durvalumab/olaparib in 699 patients with advanced/recurrent endometrial cancer (NCT04269200). It will be very interesting to see these data in light of the phenomenal results reported from the phase III RUBY and NRG-GY018 trials, which demonstrated significantly improved progression-free survival when an ICI was combined with first-line chemotherapy and then continued as maintenance therapy in advanced endometrial cancer.^59,60 In the mismatch repair deficient/microsatellite instability–high cohort, the results suggested a chance of cure, with an HR of approximately 0.30 reported in both trials. There was a reasonable signal of efficacy in the pMMR cohort, although we probably can do better there. I anticipate these results leading to new indications and incorporation of ICIs with chemotherapy in the first-line setting for endometrial cancer. Returning to DUO-E, this trial is anticipated to read out in September 2023.

As you can see in this table, the ICI/PARPi combination is being examined in other solid tumors. The phase III ZEAL-1L trial is evaluating first-line pembrolizumab followed by maintenance therapy with pembrolizumab with or without niraparib in 666 patients with advanced NSCLC (NCT04475939). Finally, the KEYLYNK‑010 trial compared pembrolizumab/olaparib with abiraterone or enzalutamide in 793 patients with progressive metastatic castration-resistant prostate cancer (NCT03834519). This trial closed for futility after a planned interim analysis demonstrated no improvement in OS, which is unfortunate.⁶¹

Kathleen Moore, MD:
With some notable exceptions that we will discuss shortly, the safety profiles of PARPis and ICIs generally are nonoverlapping in both type and timing of toxicity. This means that much of your experience taking care of patients receiving monotherapy with PARPis or ICIs can be continued in patients receiving both.

Recall that the PARPis olaparib, rucaparib, and niraparib all work by inhibiting the PARP protein, which is involved in double-strand and single-strand DNA repair.⁵⁰ PARPis also can "trap” PARP—basically sticking PARP at the site of a single-strand break and causing replication fork collapse. The class toxicities for PARPis include fatigue and nausea, which are very common but usually low grade. You can get some GI toxicity, such as diarrhea, constipation, or dysgeusia. You also can have headaches. Anemia is the most common of the hematologic AEs across PARPis, and approximately 20% to 25% of patients will have grade ≥3 anemia.⁶²

There are some differences and even unique toxicities among the PARPis. Niraparib is unique in that its metabolites exert dopaminergic and serotonergic effects leading to hypertension, palpitations, and insomnia. Hypertension is reported in 10% to 15% of patients. Although insomnia has been reported only with niraparib, we do see insomnia with all the PARPis, although it is more common with niraparib.

Elevated liver enzymes are most common with rucaparib but can occur infrequently with niraparib. Elevated cholesterol is also unique to rucaparib. Elevated creatinine is, again, most common with rucaparib, although it can occur with olaparib. All of these AEs are straightforward to manage.

Dr Jonasch and I earlier discussed the safety profiles of ICIs and how they are characterized by the “-itises.” I do want to note that we will be focusing on the safety profiles of PD-1 and PD-L1 inhibitors, rather than ICIs targeting CTLA-4 or other targets, because those are the type of ICIs under investigation in combination with PARPis.

Kathleen Moore, MD:
Much of our data on the safety profile of ICI/PARPi combination therapy come from the trials I discussed earlier, such as MEDIOLA. Recall that this multicohort phase II trial evaluated durvalumab/olaparib with or without bevacizumab in cohorts with relapsed, platinum-sensitive ovarian cancer that harbored non-gBRCA mutations.⁵¹ The data from the ovarian cancer population are shown on the left, and the right table presents safety data with durvalumab/olaparib in patients with HER2-metastatic breast cancer harboring gBRCA mutations.⁶³

These are noncomparative data, but you can see that high-grade AEs were generally less frequent with the doublet in the population with breast cancer. This is consistent with other trials reporting less toxicity in breast cancer than in ovarian cancer, likely related to the amount of pretreatment with chemotherapy, the age of the patient, etc. These are different populations with different vulnerabilities to the same regimen.

Looking at the ovarian cancer cohorts, we see that the rate of grade ≥3 anemia is approximately 20%, consistent with what we expect for any PARPi monotherapy. Hypertension in the triplet arm is related to bevacizumab. Although the doublet and triplet were pretty well tolerated overall, we should note that grade ≥3 fatigue occurred in approximately 6% with either the doublet or triplet. That is pretty severe fatigue, suggesting that these regimens can be challenging in more heavily pretreated patients.

Kathleen Moore, MD:
We also have safety data on niraparib-based combinations in the platinum-resistant setting for ovarian cancer, along with some data from advanced NSCLC. The top left table here presents data from the OPAL trial of dostarlimab/niraparib/bevacizumab in 41 patients with platinum-resistant, recurrent ovarian cancer with or without BRCA mutations, and the bottom left table presents data from TOPACIO/KEYNOTE-162 on pembrolizumab/niraparib in 53 patients with recurrent ovarian cancer with or without BRCA mutations.^53-55

Niraparib has a higher risk of thrombocytopenia than other PARPis.⁶² The thrombocytopenia risk can be mitigated by individualizing the starting dose based on baseline weight and platelet count, but you still do see more grade ≥3 thrombocytopenia, which was reported in 9% to 22% of the patients on OPAL and TOPACIO. That is because of niraparib, not the combination. Grade ≥3 fatigue was relatively high with the triplet combination, occurring in 17.1% of patients. That may be because of these patients being heavily pretreated and having resistant disease, with their baseline ill health compounded by this triplet combination.

My take-home message from these data is that there is not any smoking gun for an overlapping or synergistic toxicity that occurs when we combine ICIs with PARPis and/or bevacizumab. The safety profile of these doublets and triplets is the additive toxicity of the 2-3 agents together.

Kathleen Moore, MD:
Although the safety profiles for ICIs and PARPis generally do not overlap, there are some notable exceptions: pneumonitis, elevated liver function tests (LFTs), diarrhea, fatigue/asthenia, rash, and elevated creatinine.^8,64-72

The table here presents data on the rates of these potentially overlapping toxicities from the first-line trials of PARPis in ovarian cancer, either as monotherapy (in blue) or in combination with bevacizumab (orange), along with rates from several analyses of patients with various malignancies treated with anti–PD-1 ICIs.

Kathleen Moore, MD:
To treat these potentially overlapping toxicities, you need to first determine whether the AE was more likely to have been caused by the ICI or the PARPi. Much of that relates to the timing of the AE. The AEs with PARPis—which are oral DNA damage response agents—generally are much earlier than those with ICIs, which require time for the immune system to be activated.^6-8,62,73 Dr Jonasch and I discussed earlier the kinetics of AEs related to ICI monotherapy, but I will review those briefly here to contrast with the timing of PARPi-related toxicities.

Fatigue
With PARPis, fatigue usually occurs in the first month. As we saw on the previous slide, grade ≥3 fatigue is quite uncommon with both PARPis and ICIs. If you have someone with grade ≥3 fatigue, you really should be thinking more about the ICI and the possibility of immune-mediated adrenal disorders or even myositis, where the patient reports fatigue but they actually are unable to get up because they are weak.

Diarrhea
I see constipation more often than diarrhea with PARPis, but diarrhea can occur. Both the timing and the severity of diarrhea can help you determine whether the ICI or the PARPi is more likely to be the cause. Diarrhea related to PARPi occurs pretty early, often within days of the patient starting therapy, whereas immune-related diarrhea/colitis has a median onset of 8 weeks.

Elevated Creatinine
Elevated creatinine related to immune-mediated nephritis with ICIs is uncommon. Elevated creatinine can be seen with rucaparib and olaparib, although grade ≥3 events are quite uncommon. Again, distinguishing the etiology of elevated creatinine comes down to timing. We usually see the creatinine increase in cycles 2-3 with PARPis, whereas immune-related nephritis has a median onset of 16 weeks after starting treatment.

ALT/AST Increase
Regarding liver enzyme elevations with rucaparib, those occur in cycle 1. These ALT/AST elevations with rucaparib are simply a cycle 1 event and do not recur. I will share some data illustrating this phenomenon shortly. With anti–PD-1/PD-L1 ICIs, the median onset for hepatitis is approximately 8 weeks.

Rash
Regarding rash, a photosensitive rash can occur with rucaparib more often than with niraparib or olaparib. It can be bothersome and sometimes hard to differentiate from rashes caused by the ICI, because rashes related to PARPis do occur within the same timeframe as those related to ICIs. The median onset for rash is approximately 5-6 weeks with anti–PD-1/PD-L1 ICIs.

Pneumonitis
PARPi-related pneumonitis is rare, occurring in fewer than 1% of exposed patients, and is a later event occurring as the result of cumulative toxicity. This makes it confusing to determine the etiology of pneumonitis in patients receiving ICI/PARPi combination therapy, because the median onset occurs at approximately 10 weeks for immune-related pneumonitis. Fortunately, pneumonitis is treated similarly with steroids regardless of etiology.

I have seen pneumonitis among patients enrolled on a clinical trial evaluating ICI/PARP inhibition. The cases were mostly asymptomatic, with just ground glass opacity detected on imaging. We usually attributed pneumonitis to the ICI because this event is so uncommon with PARPis.

Clinical trials—appropriately—take a very conservative approach to pneumonitis, so we administered steroids and held both agents, which resolved the imaging findings. In real-world practice, if I felt like a patient with asymptomatic grade 1 pneumonitis was really benefiting from the combination, I would consider restarting both agents.

Kathleen Moore, MD:
Unlike pneumonitis, the etiology of elevated LFTs can be sorted out by timing, with PARPis causing early events and ICIs generally causing later events.

The etiology of elevated LFTs most likely would be overlapping and confusing with rucaparib as the PARPi. Rucaparib causes a big, early bump in ALT/AST values, as you can see in these graphs presenting lab values for >20 cycles of rucaparib in the ATHENA-MONO trial.^67,68 The bump occurs around cycle 1, Day 15, and we know this only because the early clinical trials performed weekly labs.

Nowadays, when you start someone on rucaparib, you just check their labs every cycle, meaning you often miss this transient ALT/AST elevation because it resolves spontaneously. This is really key: The graphs here show that during 20 cycles of rucaparib, an ALT/AST bump occurs early on and then resolves. This contrasts with immune-related hepatitis, which has a median onset of approximately 8 weeks and is less common.

The timing, the question of whether the LFT elevation resolved, and the fact that this is very common with rucaparib vs uncommon with ICIs can help you figure out the cause and workup for a potential liver toxicity in a patient receiving rucaparib plus an ICI.

It is important to remember that elevated LFTs are not a feature of the safety profiles for olaparib and niraparib. If the patient is receiving a combination of an ICI with olaparib or niraparib, then a big bump in LFTs would have you first thinking about the ICI as the cause.

Kathleen Moore, MD:
Once you have a good idea of which agent is causing the toxicity, how do you proceed? As I mentioned earlier, the toxicity management algorithms for monotherapy can be used in patients receiving ICI/PARPi combination therapy. We already discussed the management of immune-related AEs with ICI monotherapy, so I will focus my discussion on managing PARPi-related toxicities.

Unlike with ICIs, we can use dose reductions and short-term dose interruptions with PARPis.⁶² For example, we can stop the PARPi for just a few days for nonhematologic grade 1/2 toxicities. Patients often feel much better, and sometimes you can just restart the same dose and keep them on it. If the toxicities recur, then you can dose reduce.

Discontinuations are uncommon. Approximately 9% to 13% of patients with ovarian cancer will discontinue a PARPi because of AEs.

Grade 3/4 nonhematologic AEs are uncommon with PARPis. For these high-grade events, you need to dose interrupt until resolved and then dose reduce.

Kathleen Moore, MD:
These are the dose reduction guidelines for PARPis.⁶² Beginning with olaparib, the starting dose is 300 mg twice daily. You can reduce to 250 and then 200 mg twice daily, and that is your lowest dose. Your only option for a third dose reduction is to permanently discontinue.

For niraparib, recall that the starting dose is individualized based on the patient’s “weights or plates”—their baseline weight or platelet count. Patients start at 300 mg once daily unless their baseline weight is <77 kg or their baseline platelet count is <150,000/µL, in which case they start at 200 mg once daily.⁴⁷ A patient starting at 300 mg can then have 2 dose reductions to 200 and then 100 mg once daily. However, most patients start at 200 mg once daily, so they only have 1 dose reduction available to 100 mg. Any further dose reductions mean they have to discontinue.

Rucaparib has 4 active doses, meaning 3 dose reductions are available.⁶² Patients start at 600 mg twice daily and can then decrease to 500 mg, 400 mg, or even 300 mg twice daily. I like having more dose levels available because that helps patients continue therapy and gives us more room to help manage difficult toxicities.

Kathleen Moore, MD:
Anemia is another important class effect of PARPis.^62,75 Although we cannot eliminate the risk of anemia with PARPis, I try to mitigate contributing causes that make it more severe. Because I practice in a part of the country where we have a lot of nutritional deficiencies, when I start PARPis or chemotherapy in general, I check iron, folate, and—even though it is not involved with anemia—vitamin D levels at baseline.

If a patient has iron deficiency, I usually will just replace the iron with IV ferric carboxymaltose, because someone starting on chemotherapy often cannot tolerate the AEs of oral iron (eg, constipation). I like to get my patients back up to normal iron levels and have them seeing a nutritionist so we can try to maintain them at normal levels. Once they are feeling better, if they need oral iron maintenance, they can just take a multivitamin.

If I have taken these preventive measures but still have a patient coming in with hemoglobin <8 g/dL, this is where I start thinking about transfusion. If a patient is symptomatic and/or getting close to 7 g/dL or <7 g/dL, I will transfuse and then dose modify post transfusion. You also can hold the PARPi and let the patient come back up naturally, but that usually takes 4 weeks, and I prefer to not hold therapy that long. 

I will say that if your patient develops grade 3 anemia, undergoes transfusion, and restarts the PARPi, and 2 months later the same thing happens again and then again, you should be concerned and investigate further. Repetitive anemia is not expected with PARPis, and in this situation, I would be worried about the patient’s bone marrow and would go to hematology for an evaluation. A large meta-analysis of a clinical trial cohort and a real-world pharmacovigilance cohort reported that anemia was the most common cytopenia reported before or concomitantly with the onset of myelodysplastic syndromes/acute myeloid leukemia.⁷⁹ I have sent patients to hematology, and we have detected early myelodysplastic syndromes, which led us to stop the PARPi to try to prevent transformation into acute myeloid leukemia.