The clinical challenges, trials, and errors of combatting PARPi resistance

Abstract

The use of poly(ADP-ribose) polymerase inhibitors (PARPi) exploits synthetic lethality in solid tumors with homologous recombination repair (HRR) defects. Significant clinical benefit has been established in breast and ovarian cancers harboring BRCA1/2 mutations, as well as tumors harboring characteristics of “BRCAness”. However, the durability of treatment responses is limited and emerging data have demonstrated the clinical challenge of PARPi resistance. With the expanding use of PARPi, the significance of PARP therapy in patients pre-treated with PARPi remains in need of significant further investigation. Molecular mechanisms contributing to this phenomenon include restoration of HRR function, replication fork stabilization, BRCA1/2 reversion mutations and epigenetic changes. Current studies are evaluating the utility of combination therapies of PARPi with cell cycle checkpoint inhibitors, anti-angiogenic agents, PI3K/AKT pathway inhibitors, MEK inhibitors, and epigenetic modifiers to overcome this resistance. In this review, we address the mechanisms of PARPi resistance supported by preclinical models, examine current clinical trials applying combination therapy to overcome PARPi resistance, and discuss future directions to enhance the clinical efficacy of PARPi.

The significance of PARP inhibition, synthetic lethality and beyond

Poly(ADP-ribose) polymerase (PARP) 1 is a member of a family of enzymes with a critical role in the base excision repair (BER) mechanism of single stranded DNA breaks (SSB), a component of the larger DNA damage repair (DDR) machinery (1–3). Upon sensing and binding to SSB, the catalytic domain of PARP1 undergoes “auto-PARylation”, forming poly(ADP-ribose) complexes that ultimately recruit SSB repair proteins, promote chromatin remodeling, and induce repair of DNA damage (1,2,4,5). Dysfunction in PARylation leads to the accumulation of SSB causing replication-dependent DNA double strand breaks (DSB), as well as “PARP trapping”, whereby PARP1 is unable to dissociate from sites of DNA damage (6–8). PARP inhibitor (PARPi)-induced PARP trapping causes replication fork collapse and an insurmountable accumulation of DSB for cellular viability, leading to clinically relevant cytotoxicity (8–10).

To restore normal DNA architecture at sites of DSB, cells rely on one of two mechanisms: homologous recombination (HR) or non-homologous end joining (NHEJ). While healthy cells can depend on BRCA1/2 proteins and the error-free HR mechanism of DSB repair, cancer cells with germline BRCA1/2 (gBRCA1/2) mutations or HR deficiency (HRD) must resort to the lower fidelity NHEJ repair mechanisms, leading to gene alterations and deletions. The subsequent chromosomal instability promotes overwhelming levels of DNA damage and replication stress (1,11) In BRCA1/2 mutant tumors, defective DSB mechanisms force a reliance on PARP-induced SSB repair mechanisms. The opportune use of PARPi in the setting of synthetic lethality is thus the rationale supporting the clinical application and success of PARPi in this subset of exquisitely sensitive tumor cells compared to wild-type counterparts (4,12).

Clinically, patients with gBRCA1/2 mutations have demonstrated an increased propensity for breast, ovarian, and prostate cancers (13,14). Among patients with ovarian cancer, 20% harbor BRCA1/2 mutations, the majority of which are germline mutations (15,16). With more granular analysis, half of patients with high grade serous ovarian cancer have HRD tumors, exhibiting sensitivity to PARPi comparable to their BRCA1/2-mutated counterparts (17). While this “BRCAness” phenotype has been explained by hypermethylation of the BRCA1 gene promotor, loss of heterozygosity of BRCA1/2, and loss of function mutations or methylation of other HRR genes (BRCA1/2, RAD51C, RAD51D, PALB2), HRD is a dynamic process and there are likely additional relevant mechanisms that have yet to be defined (18).

A number of clinical trials have led to FDA approval for the use of olaparib, niraparib and rucaparib with specific clinical indications (Table 1). FDA approval for PARPi use in the maintenance setting of ovarian cancer following platinum-based chemotherapy, for example, stems from the Phase III PAOLA-1 trial demonstrating progression free survival (PFS) benefit of 22.1 months with olaparib and bevacizumab compared to 16.6 months with bevacizumab alone (HR 0.59; 95% CI 0.49–0.72; p<0.001) (19). Furthermore, 12 BRCA2-mutated ovarian cancer patients had a preliminary disease control rate of 92% with use of niraparib and bevacizumab compared to niraparib alone in the phase 1 ENGOT-ov24 clinical trial (20).

Table 1:

PARP Inhibitor Use by FDA Designation and Associated Clinical Trial

Primary Disease
Agent	Treatment Indication	Upfront Indication	Clinical Trial
Olaparib (Lynparza)	gBRCA/sBRCAm ovarian cancer	Ovarian, fallopian tube, peritoneal cancer after partial or complete response to initial platinum-based therapy	SOLO-1 (NCT01844986)
	HRD ovarian cancer(in combination with bevacizumab)	Ovarian, fallopian tube, peritoneal cancer after partial or complete response to initial platinum-based therapy	PAOLA-1 (NCT03737643)
	gBRCAm metastatic pancreatic adenocarcinoma	Maintenance treatment for disease without progression on at least 16 weeks of first-line platinum-based chemotherapy	POLO (NCT02184195)
Niraparib (Zejula)	Any advanced ovarian, fallopian tube, peritoneal cancer	Ovarian, fallopian tube, peritoneal cancer after partial or complete response to initial platinum-based therapy	PRIMA (NCT02655016)
Secondary Disease
Agent	Treatment Indication	Recurrence Indication	Clinical Trial
Olaparib (Lynparza)	gBRCAm ovarian cancer	Ovarian cancer treated with three or more prior lines of chemotherapy	Study 42 (NCT01078662)
	gBRCAm, HER2-negative metastatic breast cancer	Metastatic breast cancer treated with chemotherapy in the neoadjuvant, adjuvant or metastatic setting	OlympiAD (NCT02000622)
	Recurrent HRD mCRPC	mCRPC with progression following prior treatment with enzalutamide or abiraterone	PROFOUND (NCT02987543)
Rucaparib (Rubraca)	gBRCA/sBRCA mutant ovarian cancer	Ovarian, fallopian tube, peritoneal cancer treated with two or more prior lines of chemotherapy	ARIEL2 (NCT01891344) STUDY 10 (NCT01482715)
	BRCA1/2-mutatant mCRPC	Prostate cancer treated with both androgen deprivation therapy and chemotherapy	TRITON2 (NCT02952534)
Niraparib (Zejula)	gBRCA/sBRCAm ovarian cancer OR genomic instability and who have progressed more than six months after response to the last platinum-based chemotherapy	Ovarian, fallopian tube, peritoneal cancer treated with three or more prior lines ofchemotherapy	QUADRA (NCT02354586)
Talazoparib (Talzenna)	gBRCAm, HER2-negative locally advanced or metastatic breast cancer		EMBRACA (NCT01945775)
Agent	Treatment Indication	Second Line Maintenance Indication	Clinical Trial
Olaparib (Lynparza)	Any patient with recurrent ovarian, fallopian tube, peritoneal cancer	Platinum-sensitive ovarian, fallopian tube, peritoneal cancer after partial or complete response to platinum-based therapy	SOLO-2 (NCT1874353)
Rucaparib (Rubraca)	Any patient with recurrent ovarian, fallopian tube, peritoneal cancer	Platinum-sensitive ovarian, fallopian tube, peritoneal cancer after partial or complete response to platinum-based therapy	ARIEL3 (NCT01968213)
Niraparib (Zejula)	Any recurrent ovarian, fallopian tube, peritoneal cancer	Platinum-sensitive ovarian, fallopian tube, peritoneal cancer after partial or complete response to platinum-based therapy	ENGOT-OV16/NOVA (NCT01847274)

Mechanisms of de novo and adaptive PARPi resistance

Clinical trials have demonstrated an acquired drug resistance to PARPi after prolonged exposure, likely contributing to the observed treatment failure rate of 40% of patients (21). Proposed mechanisms of resistance include restoration of the HRR machinery and stabilization of the replication fork, components of which contribute to the DNA replication stress response necessary to maintain genomic integrity (22). Additionally, BRCA1/2 reversion mutations and epigenetic modifications have also been associated with PARPi resistance. Given FDA approval of olaparib, niraparib and rucaparib in the treatment and maintenance setting of recurrent ovarian cancer, as well as current studies exploring the benefit of PARPi use in the frontline setting, understanding the mechanisms of PARPi resistance has become critical for effective patient selection and to produce durable clinical responses (23).

Restoring the HRR machinery

Of the two coordinated cellular repair mechanisms for DNA DSB, the HR machinery has higher fidelity due to use of the sister chromatid as a template in the repair process. Resection of DSB ends into single-strand DNA (ssDNA) is initiated in the S/G2 phase by the MRE11-Rad50-Nbs1 (MRN) complex, C-terminal binding protein interacting protein (CtIP), and other nucleases such as EXO-1 and DNA2; this DNA end resection induces the HR pathway (24,25). The ssDNA overhanging sections are coated by replication protein A (RPA), leading to activation of gH2AX by ataxia-telangiectasia mutated (ATM) and ataxia-telangiectasia and Rad3 related (ATR) kinases, part of the phosphatidylinositol 3-kinase (PI3K) family that respond to DNA replication stress (26). Upon further signal transduction, BRCA1 and p-53 binding protein 1 (53BP1) are recruited to the sites of DSB and activate CtIP (27). Functional CtIP creates 3’ overhangs at which a complex of BRCA2, PALB2, and Rad51 works in conjunction to form a nucleoprotein filament D-loop, protecting further DNA degradation and ultimately completing HRR (28,29).

Restoring HR function through alteration of deficient HR-related proteins or inhibitor molecules, thus allowing DSB repair and decreasing replication stress, can lead to PARPi resistance (23). One proposed mechanism of PARPi resistance involves cyclin-dependent kinases (CDKs), which dictate DNA end resection of HR through phosphorylation of CtIP and nucleases, as well as activation and regulation of ATR/ATM (25,30). In vitro studies using OVCAR-8 cells demonstrate that impaired CDK12 function in ovarian cancer cells leads to decreased BRCA1 levels, compromised HR function, and sensitivity to the PARPi, veliparib (16). Furthermore, the chromatin binding protein 53BP1 is a negative regulator of DNA end resection, by promoting binding of the Shieldin effector protein complex onto ssDNA overhangs, leading to molecular alterations in the Shieldin genes (23,31). Decreased 53BP1 activity allows for DNA end resection in the absence of BRCA1, recruitment of RAD51 and, thus, HR restoration. Altered 53BP1 function is thus associated with decreased PARPi efficacy in BRCA1-mutated breast and ovarian cancers (32,33). Interestingly, HR restoration due to loss of 53BP1 does not occur in BRCA2-deficient tumors (34).

Stabilization of the replication fork

Functional PARP1 and BRCA1/2 proteins play key roles in stabilizing the DNA replication fork and maintaining genomic integrity (35). PARP1 regulates replication stress and BRCA1/2 proteins protect DNA strands from degradation by the MRN complex (36,37). Following the concept of synthetic lethality, PARPi-induced PARP trapping leads to an overwhelming accumulation of DSB and clinically significant cytotoxicity in the absence of BRCA1/2-dependent prevention of DNA degradation and replication fork protection (23). Emerging evidence points to the protection of the DNA replication fork as the driver of PARPi resistance. Decreased EZH2 activity inhibits MUS81 nuclease, which has been associated with PARPi resistance in BRCA2-mutant breast cancer cells through replication fork stabilization rather than HR (38,39). BRCA1/2-mutant breast and ovarian tumors have been shown to have FANCD2 overexpression and increased activity, also associated with fork stability and PARPi-resistance in these cell lines (40). The loss of FANCD2 in these tumors lead to increased cytotoxicity, implying a central role in replication fork protection (40). Furthermore, PTIP, a MLL3/4 recruitment protein complex that works in conjunction with 53BP1 to inhibit MRE11-initiated DNA strand degradation, is also shown to have a protective role in BRCA-deficient cells. Inactivation of PTIP in B-lymphocytes inhibited MRN complex recruitment to stalled replication forks, and thus prevented further DNA degradation without affecting HR activity (41). Replication fork stability is thus a feasible mechanism by which BRCA-deficient tumors evade cytotoxicity and acquire drug resistance.

BRCA1/2 reversion mutations and the restoration of BRCAness

Secondary BRCA1/2 reversion mutations are an established mechanism for PARP1 resistance (42). These somatic mutations may take the form of amino acid insertions or deletions to restore a functional, wild-type protein, or restore the open reading frame (ORF) of BRCA1/2 to ultimately allow DSB by HR (42,43). The exact etiology of these reversion mutations—sporadic or drug-induced—is unclear. The frequency of reversion mutations is also unknown, though the intentional use of ctDNA and liquid biopsy in clinical trials offers some insight, suggesting that reversion mutations may vary between tumor types (43). Among 24/1534 patients with metastatic castration-resistant prostate cancer (mCRPC) with gBRCA1/2 mutations and history of PARPi or platinum chemotherapy exposure, the frequency of BRCA2 reversion mutations was estimated to be 40% (44). Within a cohort of 97 patients with high grade serous ovarian cancer and either germline or somatic BRCA1/2 mutations undergoing treatment with rucaparib on clinical trial, 8.2% patients had BRCA1/2 mutation reversion prior to treatment and 10.3% exhibited reversion post-progression. The presence of BRCA1/2 reversion mutations was directly associated with decreased clinical benefit from rucaparib (45).

Interestingly, reversion mutations have also been described in other HRR genes. RAD51 has been implicated in PARPi resistance among triple negative breast cancer (TNBC) patients (46). In the ARIEL2 trial of rucaparib in platinum-sensitive, recurrent HGSOC, RAD51C/D reversion mutations were associated with PARPi resistance in cohort analysis (47,48).

Epigenetic modifications

Prolonged exposure to PARPi treatment has been associated with epigenetic modifications to PARP1, BRCA1/2, and other proteins essential to DDR—including post-translational histone modifications, chromatin remodeling, DNA methylation, and RNA-mediated processes—affecting protein function and altering sensitivity to PARPi. For example, the demethylation of BRCA1 has been shown to restore BRCA1 expression, reinstate the HR machinery, and thus impart PARPi resistance to BRCA1-mutated and BRCA1-methylated patient-derived xenograft (PDX) models of breast cancer (49). In the phase II ARIEL 2 study of rucaparib treatment for platinum-sensitive, relapsed ovarian cancer, analysis of pre-treatment and post-treatment tumor biopsy samples revealed that heterozygous demethylation of BRCA1 was specifically attributed to PARPi resistance (48,50). This mechanism of resistance has been demonstrated clinically in the recurrent setting of both PARPi and platinum-based chemotherapy treatments (48,51). Quantification of BRCA1 methylation prior to treatment has potential to determine the benefit and utility of PARPi therapy (50).

Other epigenetic modifications associated with PARPi resistance includ de-ubiquitination of the BARD1 BRCT1 domain by the USP15 enzyme (52). The functional N-terminal RING domain of BRCA1 dictates its function and forms a heterodimer with BARD1 to regulate DNA end resection, forming a single-stranded template upon which the essential recruitment of BRCA2/PALB2 and RAD51 occurs to carry out HR (53). As the BRCA1/BARD1 complex is essential to RAD51 stimulation, de-ubiquitination of the BARD1 domain and consequent retention of BRCA1/BARD1 complex at sites of DSB promotes recruitment of downstream tumor suppressor proteins, leading to PARPi resistance (52). Post-translational modifications to PARP1 have also been demonstrated to cause resistance to PARPi treatment in preclinical studies of TNBC. The phosphorylation of PARP1 protein by MET in the setting of oxidative DNA damage conferred increased PARP1 enzymatic activity, leading to reduced binding and activity of PARPi. MET inhibitors re-sensitized breast and lung cancer xenografts to PARPi therapy (23). Furthermore, the acetylation of 53BP1 creates a cellular environment analogous to 53BP1 loss, which inhibits NHEJ, restores HR function, and thus causes resistance to PARP inhibition in BRCA1-mutated tumors (54,55). This 53BP1-mediated rescue of BRCA1 deficiency has been demonstrated in BRCA-mutated, TNBC and suggests a clinically relevant mechanism of PARPi resistance (34).

Combination therapy strategies to circumvent PARPi resistance

The established association of prolonged PARPi exposure with resistance to therapy necessitates efforts to circumvent PARPi resistance, especially given increasing use of PARPi as maintenance and treatment of multiple solid tumors. Preclinical and early stage clinical trials are utilizing rational combination strategies to re-sensitize tumor cells to PARP inhibition, as well as induce sensitivity to PARPi in BRCA1/2 wildtype tumors or those with HR proficiency (23). In targeting the various pathways and functional steps of HR restoration, replication fork stabilization, and BRCA1/2 reversion mutations, these strategies may manipulate tumor cells’ ability to evade DNA damage (Table 2).

Table 2.

A Summary of PARPi combination Clinical Trials

	Combination therapy	Trial	Phase	Study popoulation
PARPi + WEE1 inhibitor	Olaparib + AZD1775	NCT03579316	II	Recurrent fallopian tube, ovarian and primary peritoneal cancers
		NCT04197713	I	Advanced solid tumors with selected mutations and PARP resistance
		NCT02576444	II	Tumors harboring TP53 or KRAS mutations
		NCT02511795	I	Refractory solid tumors, Relapsed SCLC
PARPi + ATR inhibitor	Olaparib + AZD6738	NCT02576444 (OLAPCO)	II	Solid tumors haboring mutations leading to dysregulation of PI3K/AKT pathway
		NCT04065269 (ATARI)	II	Gynecologic cancers
		NCT03787680	II	Prostate cancer
		NCT03878095	II	IDH1 and IDH2 mutant tumors
		NCT03462342 (CAPRI)	II	High grade serous ovarian cancers
		NCT03428607	II	SCLC
		NCT03682289	II	Clear cell renal cell cancer; metastatic renal cell cancer; metastatic urothelial cancer; metastatic pancreatic cancer; locally advanced pancreatic cancer
	Niraparib + BAY1895344	NCT04267939	Ib	Advanced solid tumors and ovarian cancer
PARPi + CHK1 inhibitor	Olaparib + prexasertib	NCT03057145	I	Advanced solid tumors
PARPi + anti-angiogenics	Olaparib + bevacizumab (PAOLA-1)	NCT02477644	III	High grade serous ovarian cancers
	Niraparib + bevacizumab	NCT02354131	II	Platinum-sensitive recurrent ovarian cancer
	Olaparib + cediranib (EVOLVE)	NCT02681237	II	HGSOC with progression post-PARPi
	Olaparib + cediranib v olaparib alone	NCT0111648	II	Relapsed platinum-sensitive ovarian cancer of high grade serous or endometrioid histology or deleterious gBRCA1/2 mutation
	Olaparib + cediranib v olaparib alone v standard platinum-based chemo	NCT02446600	III	Recurrent platinum-sensitive ovarian cancer
PARPi + PI3K inhibitor	Olaparib + BKM120 or BYL719	NCT01623349	Ib	Recurrent TNBC or HGSOC
	Olaparib + vistusertib (AZD 2014) or olaparib + capivasertib (AZD5363)	NCT02208375	Ib/II	Recurrent endometrial, TNBC, ovarian, primary peritoneal or fallopain tube cancer
	Olaparib + capivasertib	NCT02338622	Ib	BRCA1/2 and non-BRCA1/2 mutant cancers
	Niraparib + copanlisib	NCT03586661	Ib	Recurrent endometrial, ovarian, primary peritoneal or fallopain tube cancer
PARPi + MAPK pathway inhibitor	Olaparib + selumetinib	NCT03162627	I/II	Endometrial, ovarian and other solid tumors with Ras pathway alterations and ovarian tumors with PARP resistance
PARPi + IO	Olaparib + durvalumab	NCT03594396	I/II	Resectable stage II/III TNBC
		NCT03772561 (MEDIPAC)	I	Advanced or metastatic solid tumors
		NCT03842228	I	DDR-mutated unresectable, advanced or metastatic solid tumors
		NCT03775486	II	Metastatic NSCLC
		NCT03579784	II	Unresectable or recurrent gastric carcinoma
		NCT03810105	II	Recurrent prostate cancer
		NCT04053322 (DOLAF)	II	Locally advanced or metastatic ER+ HER2- breast cancer
		NCT03851614 (DAPPER)	II	Locally advanced or metastatic MMR proficient CRC, pancreatic adenocarcinoma or leiomyosarcoma
		NCT03951415 (DOMEC)	II	Recurrent, refractory or metastatic endometrial cancer or carcinosarcoma of the endometrium
		NCT03167619 (DORA)	II	Locally advanced or metastatic platinum-treated TNBC
		NCT02734004 (MEDIOLA)	II	gBRCAm platinum-sensitive ovarian cancer; gBRCAm HER2- breast cancer; platinum-resistant relapsed gastric cancer
		NCT03801369	II	Metastatic TNBC
		NCT03459846 (BAYOU)	II	Advanced or metastatic platinum-ineligible urothelial carcinoma
		NCT02484404	II	Advanced, recurrent or metastatic ovarian, TNBC, lung, prostate, CRC or solid tumors
		NCT03534492 (NEODURVARIB)	II	Resectable urothelial cancer
		NCT03737643 (DUO-O)	III	Newly diagnosed advanced ovarian, fallopian tube or primary peritoneal carcinoma or carcinosarcoma
		NCT04269200 (DUO-E)	III	Newly diagnosed advanced or recurrent endometrial carcinoma
	olaparib + pembrolizumab	NCT02861573 (KEYNOTE-365)	Umb	Previously treated mCRPC
	olaparib + tremelimumab	NCT02571725	I	BRCA1/2 mutated recurrent ovarian cancer
	olaparib + tremelimumab + durvalumab	NCT02953457	I/II	DDR mutated recurrent ovarian, fallopian tube or primary peritoneal cancer
	niraparib + pembrolizumab	NCT02657889 (TOPACIO-KEYNOTE 162)	II	Advanced or metastatic TNBC or platinum-resistant ovarian cancer
	niraparib + PD-1 inhibitor	NCT03308942	II	Locally advanced or metastatic NSCLC
	niraparib + TSR-042	NCT03307785	I/II	Solid tumors
	niraparib + TSR-042 + platinum-based therapy	NCT03602859	III	Nonmucinous epithelial ovarian cancer
	niraparib + atezolizumab	NCT03522246 (ANITA)	III	Recurrent ovarian, fallopian tube, or primary peritoneal carcinoma
	rucaparib + nivolumab	NCT03572478	I/II	mCRPC or recurrent endometrial cancer
		NCT03522246 (ATHENA)	III	Front line ovarian cancer maintenance
		NCT03338790 (CheckMate 9KD)	III	mCRPC
	rucaparib + atezolizumab	NCT04276376 (ARIANES)	II	DDR-deficient or platinum sensitive solid tumors
		NCT03694262 (EndoBARR)	II	Recurrent progressive endometrial carcinoma
	talazoparib + avelumab	NCT03330405 (Javelin PARP Medley)	Ib/II	Previously treated advanced solid tumors
		NCT03565991 (Javelin BRCA/ATM)		BRCA/ATM-mutant solid tumors
		NCT03637491 (Javelin BRCA/ATM)		Ras-mutant solid tumors
		NCT03642132 (Javelin Ovarian PARP 100)	III	Front line ovarian cancer
	BGB-A317 + BGB-290	NCT02660034	Basket	Ovarian cancer, TNBC, mCRPC, bladder cancer, SCLC, HER2 (−) gastric cancer, pancreatic cancer, other solid tumors
	veliparib + atezolizumab	NCT02849496		HER2 (−), BRCA-mutant TNBC
	pamiparib + tislelizumab	NCT02660034	I	Previously treated advanced solid tumors

Combining PARPi + ATR, CHK1, and Wee1 inhibitors

Targeting the downstream effector molecules and signaling pathways of HR has yielded promising clinical benefit regardless of HR status. To ensure cellular survival amidst DNA damage, cells activate cell cycle checkpoints to halt the cell cycle and stimulate DNA damage repair (DDR) mechanisms to repair the SSB and accumulated DSB through PARPi and replication fork collapse in S phase. Due to reliance on p-53 dependent S phase checkpoint, tumors harboring p53 mutations, namely ovarian cancers, must rely instead on G2 checkpoints to induce DDR (56). Specifically targeting ATR, Chk1, and Wee1 has been shown to restore HR and replication fork stability, thereby re-sensitizing BRCA1/2 deficient tumors to PARP inhibition (23,57).

Inhibition of ATR prevents G2/M cycle arrest, disrupts the recruitment of PALB2/BRCA2 and RAD51 onto sites of DNA damage, and thus creates an environment of HRD and replication fork instability (58). The synergism of DDR and cell cycle regulation in the combination of PARPi and ATRi has resulted in tumor suppression BRCA2-mutant PDX models (59). The CAPRI phase II single arm study combining olaparib with ceralasertib in HGSOC is currently recruiting and includes cohorts based on platinum sensitivity and prior PARPi exposure (NCT03462342). Another phase II study is also currently recruiting patients with gynecologic cancers, including clear cell ovarian or endometrial cancers with ARID1A mutations, and is analyzing the effects of single-agent berzosertib versus combination with olaparib (NCT04065269). Additionally, a phase Ib trial combines ATRi (Bay 1895344) with niraparib and includes three experimental cohorts: 1) HRD solid tumors 2) platinum resistant, HRD, PARPi naïve HGSOC 3) PARPi-resistant HGSOC (NCT04267939). These studies will not only provide further information on mechanisms of PARPi resistance, but also help to delineate which populations of patients will optimally benefit from combination treatment of ATRi and PARPi.

Activated by ATR and regulated by transcription factor E2F7, Chk1 interacts with downstream Cdc25a and Cdc25c molecules and ultimately regulates RAD51 expression and recruitment (60). Chk1 inhibition in combination with PARPi has shown similar benefit to ATRi in combination with PARPi in preclinical models, regardless of BRCA1/2 status (59,61). In a phase 1 study of the Chk1 inhibitor prexasertib combined with olaparib in HGSOC and other solid tumors, four of the 18 treated patients with BRCA1-mutant, PARPi resistant HGSOC achieved partial response (PR). Moreover, increased expression of phosphorylated histone-2AX (γH2AX), a molecular marker of DNA damage and replication stress, and pRPA, as well as reduction in RAD51 was noted in tumor biopsy samples post-treatment, further supporting mechanisms of PARPi resistance and potential markers for response (62).

Wee1 is an additional modulator of the G2/M checkpoint and is activated by Chk1 (57). There are several phase I and II trials evaluating the combination of PARPi with Wee1 inhibitors in ovarian cancer, solid tumors with PARPi resistance, and tumors harboring TP53 or KRAS mutations. The phase II EFFORT trial combining olaparib and adavosertib has shown clinical benefit in PARPi-resistant ovarian cancers (NCT03579316). The ORR for single agent adavosertib versus combination treatment was 23% and 29%, respectively, with CBR 63% and 89%, respectively (63).

Combining PARPi + Immunotherapy

Preclinical studies have provided evidence for synergy between PARPi and immunotherapy (IO) agents via two mechanisms: 1) PARPi-induced accumulation of overwhelming DNA damage and cytosolic DNA, and 2) PARPi-induced T cell modulation and PD-L1 upregulation (64–67). These molecular changes ultimately prime the tumor microenvironment and, in setting of HR deficiency, create an opportune role for the use of immune checkpoint blockade (ICB) in combination with PARP inhibition to enhance the clinical efficacy of each agents (68–71).

As demonstrated in HR deficient and/or BRCA1/2-mutated tumors, the inhibition of PARP1 leads to a catastrophic accumulation of cytotoxic DNA DSB and collapse of the replication fork (1). This overwhelming tumor mutational burden (TMB) is recognized by the immune microenvironment through damage-associated molecular patterns (DAMP), specifically cytosolic DNA-cycling GMP-AMP synthase complex (cGAS) (72). Subsequent recruitment of stimulator interferon genes (STING) and activation of the cGAS-STING pathway leads to downstream infiltration of T cells, increased type 1 IFN expression and signaling, and paracrine stimulation of dendritic cells (73,74). Both talazoparib and olaparib have shown efficacy in activating the cGAS-STING pathway and inducing T cell infiltration in BRCA1-deficient murine models (70,75). Preclinical models using breast cancer lines also demonstrate PARPi-induced upregulation of PD-L1 expression through the inactivation of GSK3β (76). Murine models of breast, ovarian and colorectal murine models have demonstrated promising antitumor activity with the combination of anti-PD-1/L1 with olaparib, talazoparib and rucaparib (67). Downstream from PARP inhibition, there is further evidence that CTLA-4 suppresses the antitumor immune via increased T-cell infiltration and IFN-y production; preclinical models of BRCA1-deficient mice with ovarian cancer demonstrate improved survival with the combination of veliparib and anti-CTLA-4 (77).

Currently, two phase II trials combining PARPi and IO have shown promising results. The phase II MEDIOLA basket trial of olaparib and durvalumab in advanced solid cancers, including TNBC, ovarian, and uterine cancers (NCT02734004), demonstrated objective response rate (ORR) and disease control rate (DCR) of 63% and 81% at 12 weeks for patients with germline BRCA1/2 mutations and platinum-sensitive disease, respectively. Patients pre-treated with prior lines of chemotherapy had an ORR of 68% (68,78). In the open-label, phase II TOPACIO trial (NCT02657889), the effects of niraparib in combination with pembrolizumab were evaluated in women with advanced or metastatic TNBC or recurrent ovarian carcinoma, irrespective of BRCA1/2 status. Those with BRCA1/2 mutations had improved ORR (45%) and disease control rates (73%) compared to the entire study cohort (25% and 68%, respectively). Interestingly, there was also clinical benefit in BRCA1/2 wildtype and HR-proficient patients (ORR 24% and 27%, respectively), suggesting benefit beyond mutational status (69). Additionally, with preclinical and phase 1 results suggesting therapeutic effect with the combination of PARPi and anti-CTLA-4 therapy in heavily-pretreated patients with gBRCA recurrent ovarian cancer (NCT02571725), a phase II study is evaluating olaparib versus the combination of olaparib and tremelimumab in platinum-sensitive, recurrent ovarian cancer (NCT04034927) (79).

Combining PARPi + anti-angiogenic agents

Preclinical data suggest a synergistic role between PARP inhibition and anti-angiogenic agents through two mechanisms: 1) inhibition of BRCA1/2 expression and 2) exacerbation of cellular hypoxia leading to genetic instability (80). In a model of BRCA2-mutated ovarian cancer stemlike cancer cells with confirmed BRCA2 reversion, chemosensitivity was restored with inhibition of VEGFR3 (81). While it is established that tumoral hypoxia promotes genetic instability with altered HR protein expression, the use of anti-angiogenics has been associated with increased hypoxia, impaired HR, and sensitivity to PARPi treatment (82,83).

Clinical trials combining the anti-VEGF agent, bevacizumab, with PARPi present contradictory findings based on HRD status. In exploratory subgroup analysis of olaparib in combination with bevacizumab in ovarian cancer patients enrolled in the PAOLA-1 trial (NCT02477644), HR deficiency was associated with improved progression free survival (PFS) with HR 0.33 compared to 0.92 in the HR proficient patient group (19). Conversely, in the randomized phase II ENGOT-ov24 trial (NCT02354131) of bevacizumab and niraparib in combination versus niraparib alone in the treatment of platinum-sensitive recurrent ovarian cancer, clinical benefit was demonstrated irrespective of HR status. In subgroup analysis, the HR proficient cohort had a HR of 0.4 (20).

A randomized phase II study of olaparib in combination with cediranib, an oral VEGF inhibitor (VEGFi), versus niraparib alone demonstrated improved PFS in recurrent platinum-sensitive ovarian cancers with combination therapy (16.5 v 8.2 months, HR 0.50; p=0.007). In post-hoc exploratory analyses, while PFS and OS were similar in the two arms of gBRCAm patients, significant PFS and OS benefit was observed in the gBRCA1/2 wild-type cohort with combination therapy (23.7 v 5.7 months, HR 0.31; p=0.0013; 37.8 v 23 months, HR 0.44; p=0.047) (84). In the subsequent phase III trial comparing combination therapy to standard platinum-based chemotherapy in patients with recurrent, platinum-sensitive HGSOC, there was no clinical benefit to olaparib + cediranib compared to standard of care (SOC) therapy and the primary endpoint of PFS was not met (85). Patients with gBRCA1/2 mutations accounted for 23.7% of the total study population and HR for PFS was 0.55 with combination therapy and 0.63 for single-agent olaparib versus SOC; for gBRCA1/2 wild-type patients, a HR of 0.97 and 1.41 was observed for combination therapy and SOC therapy, respectively (85). An open-label single arm clinical and translational phase II trial (EVOLVE, NCT02681237) exploring the use of cediranib and olaparib for HGSOC in the setting of post-PARPi progression demonstrated varying ORR and clinical benefit according to platinum-sensitivity—0% in the platinum sensitive (55% PFS rate at 16 weeks), 20% in the platinum-resistant (50% PFS), and 8% in the exploratory (39% PFS) cohorts, respectively—and mechanism of PARPi resistance. Whole exome and RNA sequencing analysis reported 19% of the acquired genomic alterations were BRCA1, BRCA2, or RAD51B reversion mutations (86). Further investigation is needed regarding clinical benefit of combination therapy not only according to overall HRD status, but within cohorts of different PARPi resistance mechanisms.

Combining PARPi + PI3K/AKT pathway inhibitors

The PI3K/AKT pathway is responsible for cell growth, survival, and regulation of metabolism, as well as serving as a sensor for DNA DSB within the DDR machinery (87). Aberrant activation is found in several tumor types, including TNBC. In preclinical models of BRCA wildtype, PTEN-mutant TNBC, silencing of PIK3CA ultimately led to an accumulation of phosphorylated histone-2AX (γ-H2AX), a molecular marker of DNA damage and replication stress. High levels of γ-H2AX support the concept that PIK3CA loss offers a potential synergy with PARPi (88,89). Investigators further demonstrated that the inhibition of PI3K impairs BRCA1/2 expression and can sensitize BRCA1/2-proficient TNBC to PARPi (90). Phase 1 trials combining PI3Ki and PARPi have also suggested that altering the PI3K pathway can sensitize a population of HR-proficient and BRCA1/2 wildtype ovarian cancer patients to PARPi (91,92). In a phase I dose escalation trial of buparlisib and olaparib (NCT01623349), PR was observed in 12/46 BRCA1/2 wildtype ovarian cancer patients (91). Similarly, the phase Ib trial combining alpelisib and olaparib showed similar rates of PR in both gBRCA1/2 wildtype and gBRCA1/2m ovarian cancer patients (92). A phase I trial combining olaparib with AKT inhibitor capivasertib further suggested that combination therapy may re-sensitize tumors to PARPi treatment. Of the 64 patients with solid tumors, 25 patients had epithelial ovarian cancer and 11 of these patients achieved clinical benefit. Four of the patients with clinical benefit were considered PARPi resistant (93). It has been suggested that decreased mTOR activity is associated with response in ovarian and TNBC, while high tyrosine kinase receptor activity and mTOR activation is associated with resistance (94). However further exploration should refine markers of treatment response and the specific populations that may benefit from combination therapy.

Combining PARPi + Ras/Raf/MEK/MAPK pathway inhibitors

While KRAS mutations are an established driver of low grade rather than high grade ovarian tumors, the Ras/Raf/MEK/MAPK (MAPK) pathway is activated in ~25% of HGSOC due to high mutational burden and gene copy number variation (CNV) (95,96). This signaling pathway is ultimately responsible for tumor cell proliferation, differentiation, and survival (95). Preclinical models of ovarian cancer cells lines treated with PARPi demonstrated enhanced activation of the MAPK pathway and downregulation of its targets, including FOXO3a and BIM1; these targets were found to be restored in cell lines with adaptive PARPi resistance. The use of MEK inhibitors (MEKi) was found to re-sensitize RAS-mutant pancreatic and ovarian cancer cell lines to PARPi with induction of FOXO3a, leading to alteration of PARP1 expression and function, as well as decreased expression of the MRN complex and BRCA1/2 (97). The downregulation of these key components of HR in RAS-mutant tumors by PARP inhibition ultimately lead to impaired DDR and genomic instability, thereby suggesting another potential synergistic mechanism of overcoming resistance to treatment (96). There is currently a phase I/II trial (NCT03162627) evaluating the use of olaparib with selumetinib in RAS-mutated solid tumors and ovarian tumors with PARPi resistance (23).

Combining PARPi + Epigenetic modifiers (BET and HDAC inhibitors)

Targets involved in transcriptional regulation and/or chromatin remodeling represent another class of drugs in development to induce HRD in solid tumors. Preclinical data including cell line and xenograft ovarian cancer models suggest that inhibition of bromodomains (BDs) and histone deacetylases (HDACs) may enhance the activity of PARPi (98–100). Mechanisms proposed for this synergy include enhanced PARP-related DNA damage, decreased BRCA1, and increased genomic instability via disruption of replication fork progression (100). However, the clinical promise of these agents has been limited by toxicity. Initial dose-finding studies performed in patients with hematologic malignancy demonstrated that BET inhibition was strongly correlated to thrombocytopenia (101–103). A recent Phase I trial of ODM-207 (a BETi) in select solid tumors demonstrated comparable toxicities and narrow therapeutic window (104).

HDAC inibitors as single agents have shown limited efficacy in solid tumors, and, given the parallel involvement at the transcriptional level, HDACi results in similar side effect profile to BETi (105). Therefore, there is significant interest in combination strategies. Ongoing clinical work investigating HDACi with PARPi includes early phase trials, such as a phase I/II trial of entinostat in combination with olaparib for women with recurrent ovarian cancer (NCT03924245), and belinostat in combination with talazoparib for multiple solid tumor types (NCT04703920).

Summary and future directions

With FDA approval of several PARP inhibitors both in the frontline and secondline maintenance and treatment settings, as well as proven clinical benefit across several solid tumor types, the increasing use of PARP inhibitors has yielded the clinical challenge of PARPi resistance. Restoration of the HR machinery, stabilization of the DNA replication fork, BRCA reversion mutations, and epigenetic molecular modifications are established mechanisms contributing to PARPi resistance. Preclinical models have offered insight into these various mechanisms yet analyses of paired tumor samples and clinical response in early phase clinical trials have suggested that multiple mechanisms may be inherent to progression on PARPi treatment. The use of cfDNA has allowed for more in-depth investigation into the adaptive genetic changes that may occur while on treatment and should be expanded to future clinical trial analyses.

Early clinical trials are utilization rational combination strategies to evade PARPi resistance, including: PARPi + ATR/Chk1/Wee1 inhibitors, PARPi + IO, PARPi + anti-angiogenic agents, PARPi + PI3K/AKT pathway inhibitors, PARPi + Ras/Raf/MEK/MAPK pathway inhibitors, and PARPi + epigenetic modifiers. While preliminary analyses have suggested promising response and clinical benefit, durability of treatment response is unclear and combination treatment is not without additive toxicity profiles. Further exploration of meaningful molecular markers of response are needed to delineate which patient cohorts will benefit most from each combination therapy. Consideration of employing ctDNA analyses to assess for BRCA and other HRR reversion mutations may be beneficial to help tailor both treatment choice, as well as optimize treatment duration. Dosing schedules must also be further investigated to reduce toxicity while maintaining benefit beyond that expected with each monotherapy.