Current and future landscape of PARPi resistance

Abstract

Purpose of the review

To highlight relevant strategies to overcome PARP inhibitor resistance and present key clinical trials.

Recent findings

The use of PARP inhibition (PARPi) for frontline maintenance offers substantial clinical benefit in patients with homologous recombination-deficient tumors. However, expanding PARPi from recurrent therapy to frontline maintenance may potentially result in more PARPi resistant tumors earlier in the treatment continuum and data for the use of PARPi after PARPi remain limited. Clinical evidence demonstrates tumors may develop resistance to PARPi through demethylation of the BRCA promoter or BRCA reversion mutations. Multiple clinical trials investigating therapeutic strategies to overcome resistance, such as combinations of PARPi with anti-angiogenic drugs, PI3K/AKT/mTOR, or MEK inhibitors have already been reported and more are ongoing. Furthermore, increasing the amount of DNA damage in the tumor using chemotherapy or cell cycle inhibitors such as ATM, ATR/CHK1/WEE1 is also under exploration.

Summary

There is increasing clinical interest to identify options to enhance PARPi efficacy and overcome adaptive resistance. PARPi represent a class of drugs that have significantly impacted the treatment and maintenance of ovarian cancer; as the use of PARPi increases, better understanding of resistance mechanisms is essential.

Introduction

Ovarian cancer is the most lethal gynecological malignancy worldwide with a 5-year relative survival of 30.2% for advanced stage disease [1]. The Cancer Genome Atlas analyses have identified key features of high-grade serous ovarian carcinoma (HGSOC) such as ubiquitous TP53 mutations, high genomic instability, and presence of homologous recombination (HR) abnormalities in up to 50% of the cases [2]. The HR pathway is the most accurate double-strand DNA break repair mechanism, therefore, tumors with HR deficiency (HRD) demonstrate increased sensitivity to DNA-damaging therapy [3].

Polyadenosine diphosphate (ADP) ribose polymerase (PARP) is a key player in the single-strand DNA break repair mechanism and its inhibition can subsequently induce double-strand DNA breaks. The cell death induced by treating HRD tumors with PARPi is referred to as synthetic lethality. Over the past decade there has been a rapid surge in development of PARP inhibiting agents, resulting in FDA approvals in both treatment and maintenance settings [4]–[11]. Therefore, the majority of patients with ovarian cancer will likely be treated with a PARPi at some point during the course of disease. However, many patients with initial response to PARPi develop resistance, and those who are treated with PARPi later in their disease course have decreased objective response rates with increasing prior lines of treatment [12]. Therefore, developing combinations to overcome resistance and sensitize tumors to PARPi beyond HRD represents an unmet need. Novel approaches for management of PARPi resistance are the main topic of this review.

Mechanisms of resistance to PARPi and their clinical significance

PARPi resistance can occur after use in either the maintenance or treatment setting, however, much of the current understanding of resistance has been obtained in the setting of treatment trials. Multiple mechanisms have been proposed for primary and acquired resistance to PARPi including inactivation of 53BP1, loss of PARP1 expression, hypomorphic BRCA mutations, and acquisition of reversion mutations, the latter of which have been shown by both preclinical and human studies (Table 1) [13]. The concept of reversion mutations gained traction following three independent studies examining samples obtained from patients at baseline and post-progression after PARPi treatment. In ovarian cancer, reversion mutations in HR genes such as BRCA2, RAD51C, RAD51D, and PALB2 have been shown to lead to the restoration of the open reading frame, thus resulting in normal DNA double strand break (DSB) repair [14], [15]. Therefore, the possibility of re-inducing HRD in patients who acquire reversion mutations has opened up options for this group of patients and those with baseline HR proficient (HRP) tumors [16].

Table 1:

Mechanisms of PARPi resistance

Mechanism of Resistance
	Homologous recombination recovery	Replication fork protection	Other
Mechanisms	Demethylation BRCA promoter	Loss of various (PTIP, MRE11, EZH2)	BRCA protein stabilization
	BRCA copy number variants		PARP mutations
	BRCA reversion mutations		Increased drug efflux
	Upregulation of hypomorphic BRCA mutations		EMT/Senescence
	Loss of negative regulators of HR
	Increase repair deficiency	Remove protection	Increase DNA damage and PARP dependency
Therapeutic approach	Angiogenesis inhibitors	EZH2 inhibitors	Chemotherapy
	PI3K/AKT/mTOR inhibitors		ATM, ATR, CHK1, Wee1 inhibition
	MEK inhibitors
	Bromodomain inhibitors
	HDAC inhibitors
	HSP90 inhibitors

Emerging opportunities to increase HR deficiency

1. Anti-angiogenic agents

Over the past two decades, extensive preclinical data have shown a relationship between DNA repair and genetic instability in hypoxic cells. Specifically, chronic exposure of human fibroblasts to hypoxia led to increased residual ɣ-H2AX foci in G0-G1 cells following exogenous damage. In this study, inactivation of ATM or ATR kinase induced alteration of DNA DSB repair, thus identifying them as key players [17]. Despite these appealing findings, clinical trials investigating the combination of bevacizumab with PARPi demonstrated conflicting results in patients with HR proficient (HRP) tumors. For example, in the PAOLA-1 trial combining olaparib and bevacizumab in upfront maintenance, an exploratory biomarker subgroup analyses revealed that HRD status was associated with PFS (HRD cohort HR 0.33; HRP cohort, HR 0.92) [11]. Conversely, AVANOVA2, a randomized phase II trial comparing bevacizumab and niraparib to niraparib alone for treatment of platinum-sensitive recurrent ovarian cancer met its primary endpoint of an improvement of PFS in an unselected population, with HR 0.35 and p<0.0001 [18]. The exploratory subgroup analysis presented a benefit irrespective of HRD status, with an impressive HR of 0.4 in the HRP cohort.

Interestingly, a randomized phase II study of olaparib combined with cediranib, another anti-antigiogenic agent, in relapsed platinum sensitive ovarian cancer showed improved PFS in the BRCAwt/unknown population with HR 0.31 and p=0.0013[19]. This discrepancy could be explained by in vitro and in vivo data that propose a second mechanism by which cediranib and consequently anti-angiogenic therapy exerts its effects on HR in tumor cells independent of hypoxia induction. Kaplan et al. determined that suppression of PDGFR signaling, activation of PP2A, and induction E2F4/p130 suppression impairs HR in BRCA wild-type tumors [20]. However, the subsequent phase III trial, in which cediranib and olaparib were compared to standard of care platinum-based therapy in relapsed platinum sensitive ovarian cancer, did not meet the primary endpoint of improved PFS [21]. This trial included 528 patients, of whom 23.7% had gBRCA mutation. In BRCAm patients the HR for PFS was 0.55 for the combination cediranib/olaparib compared with 0.63 for standard of care regimen, while in the BRCAwt group the HR was 0.97 and 1.41 for the same comparisons [21].

Although the combination of cediranib and olaparib failed to show improved clinical benefit over standard of care in a PARPi naïve population, a proof of concept study has evaluated this doublet in a PARPi pretreated population. This phase II trial enrolled 34 patients previously treated with PARPi and achieved partial response in four patients (12%) and stable disease in 18 patients (53%). This hypothesis generating study also reported on acquired genetic alterations at PARPi progression, most commonly reversion mutations in BRCA1, BRCA2, or RAD51B (19%), and also notable for ABCB1 upregulation (15%) encoding drug efflux pumps. Patients with either reversion mutations or ABCB1 upregulation had poor outcomes. This study suggests that the cediranib-olaparib combination may have efficacy in patient PARPi pretreated population and needs further investigation [22].

2. PI3K/AKT/mTOR pathway inhibitors

PTEN represents a tumor suppressor gene encoding a pivotal phosphatase in the regulation of the PI3K/AKT pathway, and its loss may lead to diminished HR functionality, demonstrated by decrease of ϒ-H2AX accumulation and reduced RAD51 and BRCA1 expression [23]. Downregulation of HR by PI3K/AKT/mTORi has opened up therapeutic opportunities based on the supporting synergism with PARPi in BRCA wild-type ovarian cancer [24]. The PI3K compound, alpelisib, in combination with PARPi was reported in a patient population with tumors enriched in de novo and acquired HR, including BRCA wild-type and platinum-resistant/refractory patients. Impressively, the combination of alpelisib and olaparib yielded a response rate of 33% in a BRCA wild-type platinum-resistant population [25].

Similarly, a trial combining the AKT inhibitor capivasertib with olaparib included 64 patients with advanced solid tumors, 25 of which had advanced ovarian malignancy [26]. In addition to dosing determination and demonstrating safety, this study reported a clinical benefit in 44.6% (CR/PR or stable disease >4 months). 11 of the patients with ovarian malignancy had clinical benefit; the majority of those with benefit had BRCA mutations (63.6%). Further preliminary results on clinical activity of capivasertib plus olaparib as part of a multi-arm study in recurrent ovarian, endometrial, and triple negative breast cancer were reported in 30 patients evaluable for response, with an overall response rate of 24% [27]. Importantly, activity in ovarian cancer was irrespective of BRCA or HRD status. Another arm of this trial combines the mTORC1/2 inhibitor, vistusertib, with olaparib in the same patient population. In the ovarian cancer cohort, response rate was 20% in a predominantly platinum resistant group and was unrelated to BRCA status. [28]. Lastly, a study investigating niraparib plus copanlisib, an FDA approved pan-PI3K inhibitor for the treatment of relapsed follicular lymphoma, aims to determine the maximum tolerated dose of the combination and anti-cancer effect in patients with recurrent ovarian and endometrial cancer (NCT03586661).

3. RAS/MAPK pathway inhibitors

MEK1/2 inhibitors are highly selective anti-cancer drugs targeting the MAPK pathway and clinical development led to their first FDA approval for BRAF-mutated melanomas and continued with approvals in other solid tumors such as colorectal and non-small cell lung cancer [29]. In an effort to identify treatment combinations that enhance PARPi efficacy, Sun and colleagues found RAS aberrant tumors are resistant to PARPi, and, conversely, MEK inhibition increases PARP levels and decreases HR repair capacity. [30]. Thus, this potential synergy is being tested in clinical trials, including selumetinib plus olaparib in a phase I trial involving patients with endometrial, ovarian, or other solid tumors with RAS pathway alterations (NCT03162627). Preliminary analysis of this trial shows promising anti-tumor activity with an overall response rate of 17%. The preclinical findings using combination therapy with a MEK inhibitor agree with previous findings showing that all patients that benefited from this therapy were BRCA wildtype [31].

4. Other Relevant Targets

New target classes for drug development to induce HRD in solid tumors are inhibitors of bromodomains (BDs) and histone deacetylases (HDAC), involved in transcriptional regulation or chromatin remodeling, and inhibitors of heat shock protein 90 (HSP90), a key regulator of proteostasis [32]. In vitro and in vivo preclinical data demonstrated that BDs, HDACs, and HSP90 downregulate HR in multiple ovarian cancer cells and xenograft models, enhancing the activity of PARPi [33]–[35]. Despite these attractive therapeutic targets, none of the three classes of inhibitors have yet achieved significant clinical promise. Pharmacokinetic and dose-finding studies investigating bromodomain extra-terminal inhibitors (BETi) in hematological cancers revealed thrombocytopenia as a major toxicity [36]. Phase III clinical trials with HDACi in patients with solid tumors are ongoing, however they have shown limited efficacy as single agents [37]. Lastly, although multiple HSP90 inhibitors have entered phase I clinical trials, none have received FDA approval [38].

Leveraging DNA damage

In the subset of tumors where HR deficiency is preserved, the synergy between DNA damage and PARP inhibition represents an attractive treatment strategy. As aforementioned, targeting base-excision repair and homologous repair (as with a BRCA mutation and PARP inhibition) results in synthetic lethality. Even in tumors which have proficient or reversion of homologous recombination, targeting of DNR repair pathways down-stream represents another strategy to overcome PARPi resistance or improve PARPi effect.

1. Wee1

Wee1 is a regulatory molecule at the G2/M checkpoint, activated by cyclin dependent kinase 1 (Chk1) kinase. When activated, it results in G2/M cell cycle arrest as well as phosphorylation of Cdk1 which impairs HR repair [39]. There is strong preclinical evidence for synergy of this combination, however, it tends to be poorly tolerated [40]. Clinically, there are ongoing trials combining agent adavosertib and PARPi, including a randomized phase II study in recurrent ovarian cancer comparing the combination with olaparib to adavosertib alone (NCT03579316). Demonstrated progression on PARP inhibitors is an eligibility criterion for this study. Additionally, a multi-arm phase II study of olaparib combinations (OLAPCO) is being performed in all solid tumors and includes an olaparib/Wee1 combination arm for patients with TP53 or KRAS mutations (NCT02576444). Sequential administration improved tolerability while preserving efficacy in xenograft and PDX model, thus, this strategy is under evaluation in the phase 1 STAR study of adavosertib and olaparib (NCT04197713).

2. Chk1

Chk1, as an activator of Wee1, is an attractive target for PARPi combination therapy. Preclinical and early clinical studies demonstrated single agent efficacy of the Chk1 inhibitor prexasertib in ovarian malignancies, notably without BRCA mutations [41], [42]. A phase I trial of olaparib and prexasertib (NCT03057145) is active and no longer recruiting, with estimated completion in 2023. This trial includes BRCA mutant patients only but allows for prior PARP therapy.

3. ATR

Ataxia telangiectasia and Rad3-related protein (ATR) is further upstream of Chk1 and, in PARPi resistant BRCA mutated cells, is essential for survival. Further, inhibition of ATR reverses resistance to PARPi in in vivo models [43]. Clinically, ATR inhibition with AZD6738 (berzosertib) improved PFS in combination with gemcitabine when compared to gemcitabine alone in a phase II study of platinum resistant ovarian cancer [44]. The CAPRI study (NCT03462342) is a phase II single arm study of combination olaparib and berzosertib, which includes both platinum sensitive and resistant disease as well as a cohort with prior PARPi treatment. The phase II ATARI (NCT04065269) trial includes patients with clear cell ovarian or endometrial cancer with ARID1A mutations, treated with berzosertib alone or in combination with olaparib. There are additional cohorts for clear cell without ARID1A mutation and rare gynecologic cancers. The previously mentioned OLAPCO study (NCT02576444) includes an arm with combination olaparib/ATR inhibition for those with HRD solid tumors. A last trial is testing BAY 1895344, an ATR inhibitor, in combination with niraparib (NCT04267939). This trial includes three experimental cohorts: 1) all solid tumors with HRD, 2) platinum resistant PARPi naïve ovarian cancer with HRD, and 3) ovarian cancer status post progression on PARPi. The results of these trials will not only contribute to improving the understanding of the mechanisms of PARPi resistance, but optimistically may change therapeutic approach to treatment in the PARPi resistant setting.

4. EZH2

Another novel approach to induce HR competent cells to adopt preference for non-homologous end-joining (NHEJ) is EZH2 inhibition. EZH2 has been demonstrated to be synthetically lethal in combination with CARM1-regulated SWI/SNF complexes [45]. CARM1 overexpression occurs in approximately 20% of HGSOC and is typically mutually exclusive with BRCA mutations. Thus potential for synergy between EZH2 and PARPi has been explored in vivo and EZH2 inhibition sensitized CARM1-high, but not CARM1-low, ovarian cancer to PARPi in both orthotopic and xenograft models [46]. Thus, in the subset of CARM1-high tumors, this strategy may have potential but requires further development.

Leveraging the tumor immune microenvironment

With the advent of immune checkpoint inhibitors and the efficacy demonstrated in some solid tumor types, there has been a recent explosion of trials utilizing these agents in the ovarian cancer population. There is expected synergy between PARPi and immune therapy via two distinct mechanisms. First, HRD results in double strand breaks with error prone repair, which leads to somatic mutations, which then may result in increased neoantigen formation [47]. Second, error prone DNA repair results in cystosolic DNA fragments which are recognized by and activates the immune system via the DNA-sensing STING pathway, and increased immune activity has been confirmed in preclinical models [48].

Therefore, the combination of immune checkpoint inhibition and PARPi is currently being tested in many different clinical studies. While a complete summary of these trials is beyond the scope of this review, there are some recently presented and ongoing trials which may significantly impact the use of this combination in ovarian malignancy (Table 2) [49]–[53]. It is important to note that these studies and others are varied in terms of BRCA status, prior PARP treatment, and additional of further therapeutic agents. Trials including the combination of PARPi and immunotherapy have reached the upfront maintenance setting as well, with hope for activity in the HR proficient patient group in addition to the expected activity in BRCA mutant patients.

Table 2:

Immunotherapy combinations

Trial	PARPi	BRCA status	Patient number and population	PFS	Response rate
NCT02734004 (MEDIOLA)	Olaparib, Durvalumab	BRCAm	Recurrent, Platinum sensitive, PARPi/IC naïve N=32 patients	Median: 11.1 mo	ORR: 71.9%, DCR: 65.6%
NCT02484404	Olaparib, Durvalumab1	All	Recurrent, Platinum resistant, prior PARPi allowed, IC naïve N=10 patients (ovarian cancer)	Not reported	ORR: 8%, DCR: 83%
NCT02657889 (TOPACIO)	Niraparib, Pembrolizumab	All	Recurrent, Platinum resistant, PARPi/IC naïve N=60 patients	Not reported	ORR: 25%, DCR: 68%

ORR = objective response rate, DCR = disease control rate. N= sample size reported, not entire intended/enrolled cohort

¹Trial additionally contained durvalumab, cediranib arm without PARPi

Future Directions

A key challenge in PARPi resistance is the selection of the potential combination agent; the choice of drug combination that may be most effective for a particular patient’s tumor remains an essential gap in knowledge. Prediction and assessment of adaptive responses, which allow cancer cells to survive until genomic or epigenetic change, may allow for rational selection of combination treatment and avoidance of acquired resistance. Proof of this principle has been demonstrated in window-of-opportunity trials; in one study, treatment with talazoparib resulted in distinct adaptive responses that could be characterized through proteomic and genomic evaluation [54]. Importantly, basal sample heterogeneity decreased in post-treatment samples, with similar interlesional responses, indicating that there is opportunity to select combination therapies which may be effective for individual patients.

Conclusions

Combinatorial strategies to overcome PARPi resistance represent key advancements in understanding PARP inhibition in ovarian malignancy. As these trials become more ubiquitous, selection of the appropriate combination or single agent strategy will be paramount. Being able to optimize therapeutic synergy based on molecular mechanisms present in a specific tumor represents the ultimate manifestation of targeted therapy, in order to improve outcomes for women with ovarian malignancy.

Key Points.

1. Ovarian cancer may develop resistance to PARP inhibition via multiple mechanisms, including homologous recombination recovery and replication fork protection.

2. Strategies to overcome PARPi resistance have included combination trials targeting increasing repair deficiency, such as with MEK or PI3K/AKT/mTOR inhibitors, or increasing DNA damage and PARP dependence, such as with ATM or ATR/CHK1/WEE1 inhibition.

3. Understanding, and overcoming, adaptive resistance to PARPi will be essential as the use of PARPi continues to increase and move into the frontline.