Real-World Efficacy and Safety of PARP Inhibitors in Recurrent Ovarian Cancer Patients With Somatic BRCA and Other Homologous Recombination Gene Mutations

Abstract

Background:

Real-world data regarding the use of poly (ADP-ribose) polymerase (PARP) inhibitors in recurrent ovarian cancer patients with non-BRCA homologous recombination (HR) mutations or somatic BRCA mutations are lacking.

Objective:

The purpose of our study is to evaluate the response rate, duration of treatment, time to progression (TTP), and toxicities of olaparib, niraparib, and rucaparib in somatic BRCAm and non-BRCA HR-mutated patients.

Methods:

This was a retrospective study using the electronic medical record to identify patients across our health system who were initiated on a PARP inhibitor for ovarian cancer between December 2014 and December 2019. Patients were screened for the presence of a somatic BRCA1/2 mutation or a mutation in non-BRCA HR genes. Data were collected via chart review.

Results:

For the efficacy analysis, 8 patients had somatic BRCA mutations and 12 patients had HR mutations. The overall response rate (ORR) was 50% for BRCA-mutated (BRCAm) patients and 9.1% for non-BRCA HR-mutated (non-BRCA HRm) patients. 72.7% of patients with non-BRCA HR mutations had stable disease. The duration of therapy ranged from 2 to 66 months. The median TTP was 9.5 months. Overall, 66.7% of patients in the entire cohort started on a reduced dose of PARP inhibitor. Dose reductions due to AEs were observed in 52.4% of patients, while AEs requiring treatment interruption occurred in 61.9%.

Conclusion and Relevance:

We found that PARP inhibitors provided stable disease in a high proportion of recurrent ovarian cancer patients who had pathogenic HR mutations, with toxicities comparable to major trials. Patients with non-BRCA HR and somatic BRCA mutations could benefit from PARP inhibitors.

Introduction

Ovarian cancer is the second most common gynecologic cancer in the United States and the leading cause of death among female reproductive cancers.¹ Mutations in the BRCA genes and deficiencies in homologous recombination (HR) predispose women to ovarian cancer but can also be used as treatment targets. BRCA gene mutations lead to loss of DNA repair function, resulting in genomic instability and increased risk of breast, ovarian, prostate, and pancreatic cancers.² Homologous recombination is a vital process in the maintenance and repair of chromosomes that uses a variety of proteins such as RAD51, ATM, CHEK2, and BRIP1, among others.³ Similar to BRCA mutations, mutations in HR genes impair the normal function of HR pathways and can be exploited to incite synthetic lethality when paired with poly (ADP-ribose) polymerase (PARP) inhibitors.

Since 2014, the PARP inhibitors (PARPis)—olaparib, niraparib, and rucaparib—have been approved for use in ovarian cancer in various settings, including BRCA-mutated (BRCAm) and Myriad’s homologous recombination deficiency (HRD) disease. The efficacy of PARPis in BRCAm or HRD cells stems from the synthetic lethality induced when both single-stranded and double-stranded repair mechanisms are blocked. Poly (ADP-ribose) polymerase inhibitors prevent the repair of single-stranded breaks typically conducted by PARP-base excision repair pathways. When PARPis are used in BRCAm cancer cells, single-stranded breaks accumulate, leading to double-stranded breaks that cannot be restored in the setting of BRCA deficiency, causing the cells to undergo apoptosis.

Germline BRCA mutations, which are hereditary, occur in approximately 10% to 15% of ovarian cancer patients, whereas somatic BRCA mutations are nonhereditary and occur in about 7%.⁴ All women diagnosed with epithelial ovarian cancer should undergo germline and somatic BRCA mutation testing.⁴ Next-generation sequencing has also revealed a significant proportion of genetic variants of uncertain significance (VUS), which are variants that have not yet been clearly defined or classified as pathologic.⁵

To identify patients who could respond to PARPis, there are several ways to test for BRCA mutations or HRD. While BRCA mutations are easily delineated in tumor profiling reports, HRD is not as straightforward. Currently, the methods for identifying HRD consist of Myriad’s myChoice HRD and FoundationOneCDx, which are 2 different next-generation sequencing tests. Myriad’s testing assesses for BRCA1/2 variants and genomic instability determined by loss of heterozygosity (LOH), telomeric allelic imbalance, and large-scale state transitions, while FoundationOne HRD is defined as BRCA1/2 mutations and/or high LOH.^6,7 Separately, FoundationOne also reports pathogenic mutations in proteins that partake in HR or are part of the Fanconi anemia pathway, which may also classify a tumor as HRD. The lack of standardization in defining HRD complicates not only the interpretation of tumor profiling studies but also the decision to treat with PARPi. In this study, our definition of “HRD” specifically focuses on the presence of pathogenic mutations in HR genes; these patients will be indicated as “non-BRCA HR-mutated.”

While BRCA mutations exist in approximately a quarter of ovarian cancer patients, germline and somatic HR deficiency, which includes BRCA mutations, is even more common, occurring in roughly 50% of high-grade serous ovarian cancers (HGSOC).⁸ Although HR deficiency exists in half of HGSOC, real-world data regarding the use of PARPis in patients with non-BRCA HR somatic gene mutations and somatic BRCA mutations are lacking. The purpose of our study is to evaluate the best overall response, duration of therapy, time to progression (TTP), and toxicities of PARPis in somatic BRCAm patients and non-BRCA HR-mutated (non-BRCA HRm) patients in a real-world setting.

Materials and Methods

This retrospective study was approved by the hospital’s institutional review board. Electronic medical records were used to identify patients with ovarian cancer across our health system who were initiated on a PARPi between December 2014 and December 2019. Patients’ records were reviewed for the presence of a deleterious somatic BRCA1/2 mutation or a pathogenic mutation in HR genes through central genomic testing of tumor tissue as performed by Foundation Medicine or Myriad’s myChoice (Table 2).^9–15 Patients were also reviewed for VUS and LOH status and to confirm whether germline testing was conducted. Exclusion criteria included: germline BRCA or HR mutation; PARPi clinical trial; PARPi for maintenance therapy with no BRCA mutations or HRD; PARPi for first-line maintenance; or prior treatment with PARPi.

Table 2.

List of Pathogenic Mutations and Efficacy Endpoints.

Patient number	PARP	Treatment or maintenance	Gene(s)	Mutation name	VUS	Duration of therapy (mo)	Best overall response	Time to progression (mo)
1^	Rucaparib	Maintenance	AKT2	amplification		3	P	3
2	Niraparib	Maintenance	ARID1A	S1868fs*14		36	SD	36
3	Niraparib	Maintenance	FANCA	A586T		5	SD	5
4	Niraparib	Maintenance	FANCC	splice site 844–1G>C	ATM K1964E	19	SD	19
5	Olaparib	Maintenance	RAD21	RAD21 amplification	BRCA2 K3326*	19	SD	No data
6	Rucaparib	Maintenance	RAD21, CDK12	RAD21 amplification		5	SD	5
7^	Niraparib	Maintenance	RAD21, CDK12	CDK12 Y319fs*l, RAD21 amplification		20	PR	18
8	Olaparib	Maintenance	BRCA2	K2613fs*4		66	SD	No data
9^	Rucaparib	Treatment	ARID1A	ARID1A G1137fs*24	BRCA2 K3326*	2	P	1
10	Olaparib	Treatment	ATM	Y2954C		13	SD	11
11	Olaparib	Treatment	BRIP1	BRIP1 Y19fs*2	FANCA T724M	3	SD	No data
12	Olaparib	Treatment	FANCA	Q271*		5	No data	No data
13	Rucaparib	Treatment	RAD51D	RAD51D G105*, Q107*	PARPI K22N, CDKN2B D86N	8	SD	6
14	Rucaparib	Treatment	BRCA1	E1660*		5	P	5
15	Rucaparib	Treatment	BRCA1	E1060*		4	PR	No data
16	Olaparib	Treatment	BRCA1	rearrangement intron 10		16	PR	16
17	Niraparib	Treatment	BRCA1	T1051fs*11		16	SD	16
18	Olaparib	Treatment	BRCA2	W2725*		32	PR	No data
19	Niraparib	Treatment	BRCA2	BRCA2 R2949fs*27		21	SD	No data
20	Rucaparib	Treatment	BRCA2	R2520*		31	PR	No data
21	Niraparib	Treatment	HRD+		BRCA2 c.2347G>A (p. Va178311e)	7	SD	5

Description of pathogenic BRCA and non-BRCA HR mutations as well as efficacy endpoints including duration of therapy, best tumor response, and time to progression.

Abbreviations: HR, homologous recombination; HRD, homologous recombination deficiency; PARP, PARP, poly (ADP-ribose) polymerase; PR, partial response; SD, stable disease; VUS, variants of uncertain significance.

^{^}Loss of hetereozygosity (LOH)-high.

Treatment with PARPi was defined as the use of PARPi due to progression of disease on prior regimen based on staging scans and/or most recent CA125. Maintenance PARPi was defined as use of PARPi after complete response (CR) or partial response (PR) to platinum-based chemotherapy, specifically in the recurrent setting. Patients were followed through July 31, 2020.

Cancer staging was recorded using the International Federation of Gynecology and Obstetrics (FIGO) staging system.¹⁶ Best response was defined by RECIST (Response Evaluation Criteria in Solid Tumors) criteria if imaging was available.¹⁷ The duration of therapy was defined as the amount of time in months that the patients received PARPi. Overall response rate (ORR) was calculated as the number of patients who obtained PR or CR out of the total number of patients in the group. Patients receiving PARPi for maintenance in the recurrent setting were included in the response assessment as these patients could have had disease response on PARPi. Time to progression was calculated as the time from treatment initiation to the time of first progression based on scans. Toxicity assessments were conducted via chart review for all patients who received ≥1 dose of any PARPi. Adverse events (AEs) of any grade were recorded for patients exhibiting a worsening shift from baseline after initiation of PARPi. Descriptive statistics were used to analyze data.

Results

A total of 268 patients were prescribed a PARPi during the study period. In all, 247 patients did not meet the inclusion criteria. The most common reasons for exclusion were the following: 62 patients carried a germline BRCA mutation; 61 patients were men; 48 patients were on a PARPi for maintenance therapy without a BRCA or HR mutation; and 41 patients had a nongynecologic cancer (Figure 1). The final safety population included 21 patients (Table 1). Germline testing was performed in all patients except 2. Twenty patients underwent testing with FoundationOne, while 1 patient received Myriad’s testing. The patient with Myriad’s testing (on niraparib) was identified as “HRD-positive” but did not have a specific HR mutation reported; therefore, she was not included in the efficacy analysis but was included in the safety analysis. The final efficacy population consisted of 20 patients with somatic BRCA or non-BRCA HR mutations. Patients 1, 7, and 9 also had LOH-high disease.

Figure 1.

Flow chart for inclusion and exclusion of patients.

Abbreviations: PARP, poly (ADP-ribose) polymerase; PARPi, PARP inhibitor.

Table 1.

Baseline Characteristics.

	Overall	Somatic BRCA mutation			HRD mutation
Characteristics	N = 21	Olaparib (n = 3)	Niraparib (n = 2)	Rucaparib (n = 3)	Olaparib (n = 4)	Niraparib (n = 5)	Rucaparib (n = 4)
Mean age, y (range)	70.0 (57.7–87.0)	72.3 (59.1–87.0)	70.5 (69.6–71.3)	71.2 (63.4–82.2)	66.8 (57.7–70.4)	66.8 (62.7–77.4)	74.2 (67.0–81.6)
ECOG performance status, No. (%)
0	7 (33.3)	0	1 (50)	0	2 (50)	3 (60)	1 (25)
1	9 (42.9)	3 (100)	0	0	2 (50)	2 (40)	2 (50)
2	5 (23.8)	0	1 (50)	3 (100)	0	0	1 (25)
Cancer stage, No. (%)a
I or II	1 (4.8)	0	0	0	0	1 (20)	0
III	13 (61.9)	2 (66.7)	1 (50)	2 (66.7)	4 (100)	4 (80)	0
IV	7 (33.3)	1 (33.3)	1 (50)	1 (33.3)	0	0	4 (100)
Platinum sensitivity at the start of PARP, No. (%)
Platinum-sensitive	17 (81.0)	2 (66.7)	1 (50)	3 (100)	3 (75)	5 (100)	3 (75)
Platinum-resistant/refractory	4 (19.0)	1 (33.3)	1 (50)	0	1 (25)	0	1 (25)
Median number of prior chemotherapies (range)
0–3 prior lines of therapy, No. (%)	16 (76.2)	3 (100)	1 (50)	1 (33.3)	3 (75)	5 (100)	2 (50)
4–7 prior lines of therapy, No. (%)	3 (14.3)	0	0	1 (33.3)	1 (25)	0	2 (50)
>7 prior lines of therapy, No. (%)	2 (9.5)	0	1 (50)	1 (33.3)	0	0	0
Type of treatment
Maintenance	8 (38.1)	1 (33.3)	0	0	1 (25)	4 (80)	2 (50)
Treatment	13 (61.9)	2 (66.7)	2 (100)	3 (100)	3 (75)	1 (20)	2 (50)
BRCA status, No. (%)
BRCA1	4 (19.0)	1 (33.3)	1 (50)	2 (66.7)		N/A
BRCA2	4 (19.0)	2 (66.7)	1 (50)	1 (33.3)
Race, No. (%)
Caucasian	20 (95.2)	3 (100)	2 (100)	2 (66.7)	4 (100)	5 (100)	4 (100)
African American	1 (4.8)	0	0	1 (33.3)	0	0	0
Cancer type, No. (%)
Ovarian carcinoma	15 (71.4)	3 (100)	2 (100)	2 (66.7)	3 (75)	3 (60)	2 (50)
Fallopian tube carcinoma	5 (23.8)	0	0	1 (33.3)	1 (25)	1 (20)	2 (50)
Primary peritoneal carcinoma	1 (4.8)	0	0	0	0	1 (20)	0
Histological classification, No. (%)
Serous	15 (71.4)	2 (66.7)	2 (100)	2 (66.7)	3 (75)	3 (60)	3 (75)
Mixed	4 (19.0)	1 (33.3)	0	1 (33.3)	0	1 (20)	1 (25)
Clear cell carcinoma	1 (4.8)	0	0	0	0	1 (20)	0
Mucinous	1 (4.8)	0	0	0	1 (25)	0	0

Abbreviations: ECOG, Eastern Cooperative Oncology Group; HRD, homologous recombination deficiency; N/A, not applicable; PARP, poly (ADP-ribose) polymerase.

^aStaging based on International Federation of Gynecology and Obstetrics.

Overall, the average age at the start of PARPi was 70 years, and most patients were Caucasian. In all, 38.1% (8/21) of patients had somatic BRCA mutations, 57.1% (12/21) had non-BRCA HR somatic mutations, and 4.8% (1/21) were identified as HRD-positive (Table 2). Almost all patients had FIGO stage III or IV disease. The most common histology was serous ovarian cancer. Most patients were platinum-sensitive and had received 0–3 prior lines of treatment.

For the efficacy population (n = 20), 60% (12/20) of patients were on a PARPi for treatment, while 40% (8/20) were on a PARPi for maintenance therapy in recurrent disease. Of the 12 patients who received PARPi for a treatment indication, 7 (58.3%) had BRCA mutations and 5 (41.7%) had non-BRCA HR mutations. In the maintenance group, most patients (87.5%) had non-BRCA HR mutations; only 1 had a BRCA mutation.

Best Overall Response

Nineteen of 20 patients were evaluated for best overall response, which included all patients in the efficacy arm except for 1 patient who refused scans (Table 2). Overall, 26.3% (5/19) of patients had a best response of PR, 57.9% (11/19) had stable disease (SD), and 15.8% (3/19) had progression.

In BRCAm patients, the ORR was 50% (4/8) compared with 9.1% (1/11) in non-BRCA HRm patients. Most (4/5) of the PRs were seen in patients with somatic BRCA mutations. The single patient who had a PR in the non-BRCA HRm group was receiving niraparib for maintenance therapy. Eight of 11 patients (72.7%) with non-BRCA HR mutations had SD. The 3 patients with progression had BRCA1, ARID1A, and AKT2 mutations.

In terms of the specific PARPi, 2 of 6 patients receiving olaparib had a best response of PR and 4 of 6 had a best response of SD. Of the 6 patients on niraparib, 1 had PR while 5 had SD. Finally, 2 of 7 patients on rucaparib had PR, 2 of 7 had SD, and 3 of 7 had progression. All 3 patients who progressed in this study had received rucaparib.

For patients receiving PARPis for treatment, the ORR was 36.3% (4/11) compared with 12.5% (1/8) for recurrent maintenance. In BRCAm patients, the ORR for the treatment group was 57.1% (4/7). In non-BRCA HRm patients, 3 of 4 patients had SD and 1 had progression. Figure 2 shows the evolution of imaging results for each patient over the course of treatment. Five patients were monitored primarily with CA-125 levels rather than scans.

Figure 2.

Evolution of imaging results for each patient throughout course of treatment.

Abbreviations: CT, computed tomography; MRI, magnetic resonance imaging; PARP, poly (ADP-ribose) polymerase; PARPi, PARP inhibitor; PET, positron emission tomography.

Duration of Therapy

Figure 3 shows the duration of therapy in months for each patient regarding their BRCA1, BRCA2, or non-BRCA HR mutation status. The duration of therapy ranged from 2 to 66 months. Of note, 1 patient moved away from our institution after 31 months of treatment on rucaparib. The median duration of therapy overall was 14.5 months, while the most frequent duration was 5 months. For BRCAm patients, the median duration was 18.5 months and for non-BRCA HRm patients, the median duration was 6.5 months. The patient with the longest duration of PARPi received olaparib for 66 months (for maintenance), and this patient had a BRCA2 mutation. Notably, all 4 patients with BRCA2 mutations were on PARPi for 21 months or longer. For patients on maintenance PARPis, the median duration was 19 months overall (66 months for BRCA2 patients and 19 months for non-BRCA HRm patients). For those on treatment PARPis, the median duration was 10.5 months overall (16 months for BRCA patients and 5 months for non-BRCA HRm patients). At the point of data collection cutoff, 5 patients were still on treatment, including 3 on olaparib and 2 on niraparib (Figure 2). Three of these patients had BRCA2 mutations, while 2 had RAD21 mutations.

Figure 3.

Duration of PARPi treatment for each patient, categorized by BRCA or HRD status.

Abbreviations: HRD, homologous recombination deficiency; PARPi, PARP inhibitor.

Time to Progression

For most patients, TTP was the same amount of time as the duration of treatment, but 4 patients continued treatment for various reasons despite progression, resulting in a longer duration of therapy compared with TTP (Table 2). The median TTP for all evaluable patients was 9.5 months. For BRCAm patients, the median TTP was 16 months. For non-BRCA HRm patients, the median TTP was 6 months. One patient who had an isolated liver lesion received liver ablation and continued niraparib for 2 months after the procedure until data cutoff. One patient continued olaparib for 2 months after progression until another plan was finalized. Another patient had been on rucaparib for 1 month when her CA-125 started rising with mild progression on scans, so it was decided to continue treatment for another month, at which point she progressed further and rucaparib was discontinued. One patient on rucaparib had slight progression at 6 months, but CA-125 was stable at the time and she had been off therapy for 1 month due to thrombocytopenia; thus, she continued rucaparib for another 2 months before switching therapy. Time to progression could not be calculated in several patients because they were not followed with routine imaging or because the PARPi was discontinued for tolerability reasons rather than for progression.

Safety

A total of 21 patients were included in the safety analysis (olaparib [n = 7], niraparib [n = 7], rucaparib [n = 7]). Overall, 33.3% (7/21) of patients initiated PARPi at full dose, and 4 of those patients remained on full dose until data cutoff (Table 3). In total, 66.7% (14/21) of patients started on a reduced dose of PARPi due to poor performance status and 5 of those patients were subsequently titrated up in dose after initial dose reduction. Dose reductions due to PARPi-related AEs were observed in 52.4% of patients, while AEs requiring treatment interruption occurred in 61.9% of the population.

Table 3.

PARPi Dose Adjustments and Most Common AEs.

Summary of AEs	Overall population N = 21 (%)	Olaparib n = 7 (%)	Niraparib n = 7 (%)	Rucaparib n = 7 (%)
Started at full dose	7 (33.3)	3 (42.9)	—	4 (57.1)
Remained on full dose	4	1	—	3
Started at reduced dose due to comorbidities/anticipated tolerability	14 (66.7)	4 (57.1)	7 (100)	3 (42.9)
Dose uptitrated after initial dose reduction	5	2	—	3
AEs requiring dose reduction	11 (52.4)	2 (28.6)	5 (71.4)	4 (57.1)
AEs requiring treatment interruption	13 (61.9)	5 (71.4)	3 (42.9)	5 (71.4)
AEs requiring treatment discontinuationa	2 (9.5)	2 (28.6)	—	—
AEs resulting in deathb	0	—	—	—
Tolerated PARP inhibitor without reported AE	3 (14.3)	—	—	3 (42.9)
Most common AEs (>10%)
Fatigue	10 (47.6)	4 (57.1)	4 (57.1)	2 (28.6)
Nausea	8 (38.1)	6 (85.7)	1 (14.3)	1 (14.3)
Thrombocytopenia	6 (28.6)	—	2 (28.6)	4 (57.1)
Anemia	5 (23.8)	2 (28.6)	2 (28.6)	1 (14.3)
Decreased appetite	4 (19)	1 (14.3)	1 (14.3)	2 (28.6)
Cytopenias requiring transfusions	4 (19)	2 (28.6)	2 (28.6)	—
Neutropenia	3 (14.3)	—	2 (28.6)	1 (14.3)
Constipation	3 (14.3)	2 (28.6)	1 (14.3)	—
Increased creatinine	2 (9.5)	1 (14.3)	1 (14.3)	—

Characteristics of PARPi dose adjustments and treatment interruptions due to toxicity as well as the most common AEs from PARPis.

Abbreviations: AE, adverse event; PARP, poly (ADP-ribose) polymerase; PARPi, PARP inhibitor.

^aExcludes patients requiring discontinuation of PARPi due to progression of disease.

^bExcludes patients who died while on PARPi due to progression of disease, unrelated to toxicity profile of PARPi.

The most common causes for dose reduction/interruption included anemia, thrombocytopenia, acute kidney injury (AKI), and gastrointestinal (GI) disturbances. Two patients discontinued treatment due to AEs (AKI and profound anemia in the setting of olaparib). Three of 21 patients tolerated PARPi without any reported AEs. All 3 were on rucaparib. Adverse event leading to death was not reported in any patients. In the overall population, the most common (≥10%) PARPi-related AEs were fatigue (47.6%), nausea (38.1%), thrombocytopenia (28.6%), anemia (23.8%), decreased appetite (19%), neutropenia (14.3%), constipation (14.3%), and increased creatinine (9.5%). Profound cytopenia requiring transfusions was observed in 4 patients (2 patients each on olaparib and niraparib).

Discussion

In this article, we report on the use of PARPis in recurrent ovarian cancer patients with somatic BRCA mutations or non-BRCA HR somatic mutations. While most of the major PARP trials include patients with HR defined by Foundation Medicine LOH or Myriad’s myChoice HRD, they do not delineate treatment responses according to specific non-BRCA HR pathogenic mutations. In addition, the studies evaluating PARPis in BRCA mutations primarily enrolled patients with germline mutations. Across 10 major studies, there were 100 patients with somatic BRCA mutations included, compared with more than 1300 patients with germline BRCA mutations.^18–27 ARIEL3 and the combined analysis of Study 10 and ARIEL2 were the only trials that conducted subgroup analyses of somatic BRCAm patients, and both studies reported benefit of rucaparib in this population. Importantly, in all studies, none described non-BRCA HR somatic mutations. Thus, there is a need for better understanding regarding somatic BRCA and non-BRCA HR somatic mutations and the use of all 3 PARPis in these patients. Given the paucity of information on PARPis in HR mutations or somatic BRCA mutations, our study describes the experience of a single health system on using PARPis in non-BRCA HR mutations and somatic BRCA mutations.

Our results suggest that PARPis are beneficial in patients with somatic BRCA mutations, as most BRCA patients (75%) received therapy for 16 months or longer, including a patient on treatment for 66 months. Moreover, there was an ORR of 50% (4/8) in these patients. Furthermore, most patients (72.7%) with non-BRCA HR mutations on PARPis had SD, highlighting the efficacy of a PARPi in a patient without known LOH or HRD but a non-BRCA HR mutation.

In terms of a strictly defined BRCA mutation, Mohyuddin et al²⁸ found that response rates to PARPis were similar between somatic and germline BRCAm patients in a meta-analysis of multiple cancer types. Our findings support the use of PARPis in somatic BRCAm patients. Interestingly, the patients with somatic BRCA2 mutations had the longest treatment durations (except 1 patient with ARID1A mutation). Previous studies have also described results in which BRCA2-mutated cancers responded better to PARPis compared with BRCA1-mutated, but these results are derived primarily from patients with germline mutations.^29–31 Notably, similar findings in prostate cancer have led authors to hypothesize that BRCA1 alterations may be less susceptible to PARPis as BRCA1 mutations are less often biallelic and BRCA1-mutated cancers tend to have more genomic co-alterations compared with BRCA2-mutated cancers.³² Overall, more studies are needed to confirm whether there is truly a significant difference in response to PARPis between BRCA1-mutated and BRCA2-mutated cancers.

In terms of non-BRCA HR mutations, we show that 72.7% of patients (8/11) had a best response of SD, and 1 patient with both RAD21 amplification and CDK12 mutation had PR. All these patients were receiving PARPis in the setting of recurrent disease (treatment or maintenance). This is an important finding as it demonstrates the utility of PARPis in heavily pretreated ovarian cancer patients without BRCA1/2 mutations.

Major trials evaluating PARPis in HRD-positive patients reported that patients with BRCAm experienced longer progression-free survival (PFS) compared with HRD-positive patients.^23,27,33 However, HRD positivity in these trials was defined by Myriad’s myChoice, which assigns HRD based on LOH, telomeric allelic imbalance, and large-scale state transitions. None of the major trials delineated specific pathogenic mutations of non-BRCA HR genes. Thus, we chose to describe the specific pathogenic mutations in genes involved in HR to provide as much information as possible. Hodgson et al³⁴ reported that patients with loss-of-function HR mutations had a greater PFS benefit with olaparib compared with patients without BRCA or HR mutations. Our results further emphasize the potential benefit of using PARPis in non-BRCA HRm patients by way of maintaining SD. With HR deficiency being so prevalent in HGSOC, many patients may have additional treatment options if screened for HR somatic mutations. Future studies should be conducted to test HRD as a possible predictive biomarker for achieving positive outcomes with PARPis in ovarian cancer.

In our study, only 25% (3/12 patients) of non-BRCA HRm patients had LOH-high disease. Two of these patients received rucaparib and progressed while 1 patient had PR on niraparib. Due to the small number of patients in this subset, however, we are unable to draw definitive conclusions on the correlation between PARPi efficacy and non-BRCA HR mutation status in this setting.

Of note, all 3 patients in our study population who progressed had received rucaparib (2 for treatment and 1 for maintenance). Recently, Clovis Oncology withdrew rucaparib’s indication for treatment of BRCAm ovarian cancer after ≥2 chemotherapies based on results from ARIEL4 demonstrating shorter survival in patients treated with rucaparib compared with chemotherapy (19.6 months vs 27.1 months).³⁵ Although our numbers are small, we observed a similar trend of worse outcomes with rucaparib.

Six patients had VUS in addition to their pathogenic mutations. Three of these VUS were related to BRCA2, while others were related to HR mutations. Eighty-three percent of patients (5/6) with VUS had SD; 1 patient had progression. Although certain variants are classified as VUS at the present time, it is possible that some of these variants may eventually be revealed to be benign or pathogenic.

Our safety analysis indicates that despite significant AEs, many patients were able to continue on PARPi (Table 3). PARPi-associated lab abnormalities and AEs were managed with appropriate monitoring, supportive care measures, dose modifications, and, in rare cases, treatment discontinuation. The most common supportive care measures were antiemetic agents and blood transfusions. Discontinuation of PARPi due to treatment-related/intolerable AEs occurred in 2 patients. In most cases, discontinuation of PARPi was due to progressive disease rather than toxicity. While most patients reported some degree of AE related to PARPi, 3 patients on rucaparib were able to tolerate treatment without any reported AEs. Of note, 5 of 14 patients were able to be titrated up in dose after a lower initial starting dose. Four patients remained on full dose throughout the entire treatment course.

In our study, 6 patients started niraparib at a reduced dose of 200 mg once a day and 1 patient started niraparib at 100 mg once a day. Recently, a phase III trial concluded that patients initiated on niraparib 200 mg had better tolerability and maintained the same efficacy as patients who started on 300 mg.³⁶ In our study, only 1 patient remained on niraparib 200 mg for the entire treatment period; 5 other patients had their doses modified for tolerability. In our efficacy analysis, of the 6 patients on niraparib (not including patient reported as “HRD-positive”), 1 had PR, while 5 had SD. The patient with PR had been dose-reduced from niraparib 200 mg to 100 mg. It is worth noting that despite 66.7% of our patients starting on reduced doses of all 3 PARPis, 28.6% (4/14) of these patients achieved PR. Seven of 14 patients starting on reduced doses had a best response of SD.

We recognize various limitations of our study due to its retrospective nature and small number of patients. Our data were based on chart review only; thus, there could be inaccuracies in patient adherence to medication and overall underreporting of toxicities. We also did not conduct independent review of imaging. Moreover, the small number of patients limited our ability to perform robust statistical analyses for efficacy outcomes. There were also a few patients who did not receive routine imaging and could not be included in the response analysis. Finally, treatment and maintenance indications were combined in our analysis, which could have skewed outcomes. Despite these limitations, our study reports important findings on the use of PARPis in recurrent ovarian cancer patients with somatic BRCA and non-BRCA HR mutations in clinical practice.

Conclusion and Relevance

Overall, we found that PARPis provided stable disease in a high proportion of recurrent ovarian cancer patients who had pathogenic HR somatic mutations, with toxicity rates comparable to major trials. While the non-BRCA HRm patients had stable disease, the BRCAm patients had the best responses and longer durations of therapy. The PARPis therefore remain a viable option not only in BRCAm patients but also in patients with unknown LOH and HRD but known non-BRCA HR somatic mutations. Our findings may increase awareness about the use of PARPis in a wider population of ovarian cancer patients (the non-BRCA HR-mutated group) and help guide treatment decisions in this patient population.