Novel poly-ADP-ribose polymerase inhibitor combination strategies in ovarian cancer

INTRODUCTION

PARP inhibitor combination therapies reflect an interest in expanding the role of PARP inhibitors beyond breast cancer 1/2 (BRCA1/2)-associated breast and ovarian cancers. As has happened with many antineoplastic agents, our understand- ing of PARP biology and PARP inhibitor mecha- nisms has expanded concomitantly with evidence of the clinical efficacy of PARP inhibitors in select impacting the others. To discuss novel combina- tions of PARP inhibitors with other anticancer therapies, it is first important to understand PARP biology, the processes required for DNA repair, PARP inhibitor mechanisms of action, and the successes of early clinical trials. Beyond this, I will discuss the rationale behind PARP inhibitor combination strategies in current clinical trials.

The foundingmember of the PARP familyandthe best characterized is PARP-1, which has zinc-finger DNA binding domains, a nuclear localization signal, and an automodification (self-PARylation) domain [2]. PARP-2 lacks a domain to self-PARylate, but also localizes to the nucleus and can bind DNA [1]. PARP-1 and PARP-2 are activated by binding to exposed DNA, effectively acting as sensors of DNA damage, includ- ing single-strand DNA (ssDNA) breaks/nicks, double- strand DNA (dsDNA) breaks, DNA crosslinks, DNA supercoils, and stalled replication forks. On binding to DNA, PARP-1 and PARP-2 undergo conformation changes that increase their catalytic activity ~500 times, and they begin to PARylate other proteins [3]. PARylation is a temporary posttranslational modification that can turn on DNA repair processes at the site of damage. As with other posttranslational modifications that require an enzyme to catalyze a modification and a second enzyme to remove it (e.g., a kinase and a dephosphorylase for protein phos- phorylation signals or a histone acetyltransferase and a histone deacetyltransferase for acetyl groups), chains of ADP-ribose are degraded by PAR glycohy- drolase (PARG) [4] and ADP-ribosyl-hydrolase 3 [5].

POLY-ADP-RIBOSE POLYMERASE HAS ROLES IN DNA REPAIR, CELL CYCLE CHECKPOINTS, AND APOPTOSIS

PARP-1 was originally described as being part of the base excision repair (BER) pathway, but it is now known that PARP-1, PARP-2, and PARylation have many roles in DNA repair, including DNA damage recognition, histone H1 and H2B modification to relax the chromatin to allow access to DNA by repair enzymes, recruitment of DNA repair proteins, pro- gression through the G2/M checkpoint, inhibition of RNA Pol I and RNA Pol II to prevent transcription of damaged DNA [2,6]. Canonical proteins from all of the major DNA repair pathways contain PAR- binding motifs, including XPA from the Nucleotide Excision Repair (NER) pathway, MSH6 in Mismatch Repair, XRCC1 and DNA Ligase III from BER, Ku70 and DNA dependent protein kinase of non-Homol- ogous End-Joining (NHEJ), and Mre11 and ATM, which are part of DNA-double-strand break (DSB) repair pathways NHEJ and homologous recombina- tion [7,8].
PARP is also involved in cell death. In the face of overwhelming DNA damage, PARP’s PARylation activities can deplete the cell of NAD+ (an essential component of cellular respiration), contributing to a type of nonapoptotic cell death called parthanatos [2]. In contrast to death by necrosis, PARP can play an active role in apoptosis. When the apoptotic pathway is activated, caspases cleave PARP-1 at a site between the DNA binding domain and the autoPARylation domain [9]. The DNA binding domain fragment binds DNA, blocking access to DNA to prevent transcription, DNA repair, and rep- lication in a dying cell. ‘Cleaved PARP’ has become an increasingly popular marker of apoptosis in publications.

POLY-ADP-RIBOSE POLYMERASE INHIBITOR MECHANISMS

PARP inhibitors are small molecules that mimic nicotinamide binding at the NAD+ site (Fig. 1). There are many PARP inhibitors currently being evaluated in clinical trials, including olaparib (AZD-2281), veliparib (ABT-888), niraparib (MK- 4827), rucaparib (AG014699), and talazoparib (BMN-673) [7,10–15]. Although iniparib (BSI-201) was found to inhibit PARP-1 in vitro and was tested in clinical trials, it was later found to be binding to PARP-1’s zinc finger domain and not to its catalytic domain [16,17]. PARP inhibitors compete with NAD+ at its binding site [18,19], inhibiting PARy- lation and thus all of PARP’s many functions.

Beyond interfering with PARP signaling by PAR- ylation, a second mechanism has been described to account for PARP inhibitor efficacy called ‘PARP- trapping’. PARP-1 binds at the site of DNA damage and must PARylate itself on its automodification domain to undergo the conformational change nec- essary to vacate DNA [20]. If unable to self-PARylate,PARP-1 is trapped on the DNA and physically blocks access to damaged DNA by DNA repair enzymes, not to mention blocking such important cell functions as DNA replication. In fact, the replication fork can stall and collapse, resulting in a lethal DNA DSB [21]. ‘PARP-trapping’ can be induced by PARP inhibitors [22]. As discussed above, apoptotic cells inten- tionally cause PARP-trapping by caspase-mediated cleavage of the DNA-binding domains from the catalytic and automodification domains.

EARLY CLINICAL USE OF POLY-ADP- RIBOSE POLYMERASE INHIBITORS IN BRCAR PATIENTS

The roles and mechanisms of PARP-1 remain under active investigation; many of these roles were not known when PARP inhibitors were developed for clinical use. As PARP-1’s primary role was originally thought to be as a DNA repair enzyme in the BER pathway, the first in-vitro studies and clinical trials were designed to use PARP inhibitors as semitar- geted therapy for cancer cells already known to be deficient in a second DNA repair pathway called homologous recombination to take advantage of the concept of synthetic lethality. The BRCA pro- teins are involved in homologous recombination, a DNA DSB repair mechanism that utilizes a homolo- gous chromosome to faithfully recreate functional sequence at the site of damage. Although clearly homologous recombination is not essential for cell survival due to the presence of alternative DNA repair pathways, patients with germline deleterious BRCA1 or BRCA2 mutations (gBRCA+) are at high risk to develop breast and ovarian cancers at rela- tively young ages due to loss of heterozygosity (loss or mutation of the second, functional BRCA allele) in cancer cells [23–25]. Beyond gBRCA1/2+ cancers, cancers with BRCA gene silencing by promoter methylation or mutations in other members of the homologous recombination pathway (e.g., Rad51) can contribute to a ‘BRCA-like’ phenotype that can also be sensitive to poly-ADP-ribose poly- merase inhibitor (PARPi) [26,27].

OVARIAN CANCER AS A TARGET FOR POLY-ADP-RIBOSE POLYMERASE INHIBITORS

Ovarian tumors characterized by deleterious germ- line or somatic BRCA mutations are relatively com- mon, with about 8.6– 13.7% of high-grade epithelial ovarian cancer patients having germline mutations and 18.3% having somatic mutations [28]. In fact, it has become standard of practice to test all ovarian cancer patients with high-grade serous carcinoma and high-grade endometrioid cancers for BRCA mutations. The presence of an homologous recom- bination defect predicts a favorable response to platinum therapies and to PARP inhibitors [29]. Thus, PARP inhibitor research is particularly rich in this population.

Olaparib was United States Food and Drug Admin- istration-approved 19 December 2014 as monother- apy for gBRCA+ patients with ovarian cancer who have been treated with three or more prior lines of chemotherapy. In the fourth-line setting, the objec- tive response rate (ORR) was 34% with a median progression-free survival (mPFS) of 6.7 months [30]. Rucaparib was approved December 2016 as mono- therapy in the third-line setting for women with ovarian cancer with germline or somatic deleterious BRCA1/2 mutations. It was noted that patients with platinum-sensitive disease had a 66% ORR, those with platinum-refractory disease (progression <6 months after completion of platinum-based therapy) had an ORR of 25%, and there was no ORR in platinum-refractory (progression on platinum- based therapy) patients [31,32]. Niraparib was approved March 2017 as maintenance therapy after platinum-based treatment for women with recur- rent, platinum-sensitive ovarian cancer regardless of BRCA status based on the European Network for Gynecological Oncological Trial-Ovarian 16 Nira- parib Ovarian (ENGOT-OV16/NOVA) trial [33&&]. In August 2017, olaparib was also approved as main- tenance therapy for platinum-sensitive, high-grade serous or endometriod ovarian, Fallopian tube, or primary peritoneal cancers independent of BRCA status based on the placebo-controlled SOLO2/ ENGOT-OV21 and Study 19 trials, which showed significant improvement in mPFS [34–36].

CURRENT CLINICAL TRIALS

As of October 2017, there are well over 100 clinical trials registered in the ClinicalTrials.gov database investigating the use of PARP inhibitors in a diver- sity of tumor types, including leukemia, lymphoma, melanoma, sarcoma, gynecological malignancies, and cancers of the breast, head and neck, gastroin- testinal tract, bladder, prostate, and brain. As tumors inevitably evolve resistance to monotherapy with PARP inhibitors, combination therapy strategies are a natural next step. Almost 50 of these clinical trials, the majority in phase I, are combination therapy trials open to ovarian cancer patients (summarized in Table 1). Antimetabolites (fluorouracil), nucleo- side analogs (gemcitabine), microtubule inhibitors (paclitaxel), and immune checkpoint inhibitors would not necessarily be expected to synergize with PARP inhibitors but could have additive efficacies by targeting two distinct cancer liabilities. Others have been designed to take advantage of PARP biology, exploit PARP inhibitor mechanisms, or circumvent PARP inhibitor resistance patterns.

RATIONALE FOR CHEMOTHERAPY COMBINATION STRATEGIES: CAUSE DAMAGE AND INHIBIT ITS REPAIR

In the purest genetic sense, two genes are considered ‘synthetically lethal’ if cells are viable with deletion of either gene alone but inviable when both genes are deleted in the same cell. In BRCA+ patients, the concept of synthetic lethality means systemic inhi- bition of PARP’s DNA repair functions, but more selective killing of the cancer cells as they are completely deficient in homologous recombination if both copies of BRCA1 or BRCA2 are nonfunc- tional. Loss of two DNA repair pathways is not actually synthetically lethal in the purest sense of the term. In practice to see the effects of knocking- out one, two, or even three DNA repair pathways simultaneously, it is necessary to either induce DNA damage or to wait for spontaneous DNA damage to occur. In homologous recombination deficient cells, which are prone to breakage– fusion–bridge cycles, perhaps DNA DSBs are common [37,38].

Alkylating agents such as temozolomide and cyclophosphamide cause damage repaired by BER and platinum analogs (cisplatin, carboplatin, oxali- platin) result in DNA crosslinks and adducts. These types of damage may not be recognized until S phase of the cell cycle when the protective chromatin is opened to replicate DNA and PARP expression peaks. BER requires a ssDNA intermediate, and repair of interstrand crosslinks results in DNA DSBs; both of these types of damage can be recognized by PARP-1 [39]. The topoisomerase inhibitors also take advantage of cells in S-phase. As DNA is being repli- cated, topoisomerases relieve torsional strain of supercoiled DNA by cutting the DNA, unwinding the DNA, and religating the ends together. Topo- isomerase I inhibitors (irinotecan, topotecan) and topoisomerase II inhibitors (doxorubicin) block the religation step, causing persistence of ssDNA and dsDNA breaks. As PARP-1 and PARP-2 sense ssDNA breaks, dsDNA breaks, and supercoils and recruit enzymes for repair, combination of topoisomerase inhibitors with PARP inhibitors can also result in lethal DSBs that cannot be repaired [40].

Almost all of the phase I PARP– chemotherapy combination trials started with full-dose chemo- therapy and dose-escalated the PARP inhibitor with none able to reach the therapeutic dose of PARP inhibitor established in the monotherapy trials. Zev Wainberg’s phase I trial with PARP inhibitor talazoparib + temozolomide or irinotecan started with dose-escalation of talazoparib to the previously AKT, AK thymoma; ARID1A, AT-rich interaction domain 1A; ATR, ataxia telangiectasia and Rad3-related protein; CA125, cancer antigen 125 (commonly used as an ovarian tumor marker); CBR, clinical benefit rate = CR + PR + SD; CHEK1/2, checkpoint kinases 1 and 2; CR, complete response rate = proportion of patients with no measurable disease; CTCAE, Common Terminology Criteria for Adverse Events = definitions for severity of organ toxicity for patients receiving antineoplastic agents per the National Cancer Institute; CTLA4, cytotoxic T-lymphocyte associated protein 4 (cell surface receptor); DCR, disease control rate = CR + PR + SD; DLT, dose-limiting toxicity = drug-related grade 3-5 adverse events using CTCAE; DOR, duration of response = time from initial response to first documented tumor progression; HGSOC, high grade serous ovarian cancer; HSP90, heat-shock protein 90; irRE, Immune-Related Response Evaluation = rules defining tumor response, stabilization, or progression for immuno-oncology drugs, which can result in an inflammatory response that appears to be progression; KRAS, Kirsten (gene discoverer) rat sarcoma; MTD, maximum tolerated dose = one dose level below the highest dose at which 1/3 of the patients at that dose level experience a dose- limiting toxicity as defined by NCI CTCAE; mTOR, mechanistic target of rapamycin; NCI, National Cancer Institute; ORR, objective response rate = CR + PR; OS, overall survival = time from study enrollment until death; PD, pharmacodynamics = drug effect on physiology; PD1, programmed cell death protein 1 (cell surface protein); PFS, progression free survival = time from study enrollment to determination of tumor progression or death due to any cause; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; PK, pharmacokinetics = study of the absorption, bodily distribution, metabolism, and excretion of drugs; PARP, poly-ADP-ribose polymerase; PR, partial response rate = proportion of patients with favorable but incomplete response of a pre-defined amount for a pre-defined minimum time period; PTEN, phosphatase and tensin homolog; QoL, quality of life = impact of health status on physical, mental, emotional, social functioning; RECIST, Response Evaluation Criteria in Solid Tumors = rules defining tumor response, stabilization, or progression for antineoplastic agents; RP2D, recommended phase 2 dose = highest oncology drug dose with acceptable toxicity, usually defined in reference to DLT and MTD established in phase I clinical trials; SD, stable disease rate = proportion of patients without disease shrinkage or progression by RECIST criteria; S/T, safety and tolerability = number and grade of adverse events; TFST, time to first subsequent therapy = time from enrollment to the first subsequent therapy start date or death date; TNBC, triple negative breast cancer; TORC1/2, raptor-mTOR protein complex and rictor-mTOR protein complex; TRR, tumor response rate = CR + PR; TSST, time to second subsequent therapy = time from enrollment to the second subsequent therapy start date or death date; TTD, time to treatment discontinuation = time from enrollment to treatment discontinuation for any reason; TTF, time to treatment failure = time from enrollment to documentation of progression, unacceptable toxicity, or patient refusal to continue participation; TTP, time to progression = time from study enrollment to determination of tumor progression; VEGFR, vascular endothelial growth factor receptor.

Poly-ADP-ribose polymerase inhibitor combination strategies McCann established maximum tolerated dose before dose- escalating the chemotherapy, never reaching stan- dard doses of temozolomide or irinotecan due to dose-limiting hematologic toxicities [41&&,42]. PARP inhibitors could be considered chemo-sensitizing, but whether or not this will be of clinical utility remains to be explored.

Therapies that directly cause DNA damage might be the most intuitive for combination trials, but are very likely to be the most systemically toxic to normal cells. Perhaps PARP inhibitors could be better exploited as radio-sensitizers, as one recent meta-analysis suggests [43]. Radiation therapy, which causes lethal DNA breaks via oxygen free radicals, can be localized to the tumor to avoid damage to noncancer cells. Confining DNA damage to a localized area and inhibiting its repair with PARPi, which are relatively well tolerated, is worth considering.

RATIONALE FOR HEAT-SHOCK PROTEIN 90 INHIBITORS AND ANGIOGENESIS INHIBITORS: INDUCTION OF BRCA-NESS

The combination of PARP inhibitors with Heat- Shock Protein 90 (HSP90) inhibitors (e.g., onalespib) is also to induce the BRCA-ness phenotype in cells with intact homologous recombination repair. HSP90 is a chaperone protein that folds and stabil- izes BRCA1, among many other proteins [44,45].
Preclinical data suggest that antiangiogenesis therapies can induce BRCA-ness through decreased expression of BRCA1 and RAD51 in a hypoxic envi- ronment [46]. Cediranib is a small molecule vascular endothelial growth factor receptor (VEGFR) inhibi- tor, apatinib is a small molecule VEGFR2 inhibitor, and bevacizumab is a VEGF antibody already approved for use in ovarian cancer. The combina- tion of a PARP inhibitor with an angiogenesis inhib- itor is particularly attractive given they are likely to have additive effects at a minimum, could have synergism as suggested by preclinical data, and will probably be better tolerated than PARP-cytotoxic combinations.

CELL CYCLE INHIBITOR COMBINATIONS: DNA DAMAGE ACCUMULATION BY INHIBITION OF DNA DAMAGE CHECKPOINTS

CHK1 and CHK2 inhibitor prexasertib is currently being evaluated in a phase I clinical trial based on the observation that cells deficient in TP53 (a common cancer mutation, particularly in BRCA- deficient cancers and high-grade serous ovarian cancers) pushes cancer cells toward dependence on the G2/M checkpoint kinases CHK1 and CHK2 [47]. By abolishing the G2/M checkpoint, it is hoped that DNA damage will accumulate to toxic levels and force apoptosis. Similarly in concept, WEE1 inhibitor AZD1775 is meant to abrogate cell cycle arrest in the presence of DNA damage [48].

PHOSPHOINOSITIDE 3-KINASE/AK THYMOMA/MECHANISTIC TARGET OF RAPAMYCIN PATHWAY INHIBITORS: PREVENTION OF A POLY-ADP-RIBOSE POLYMERASE INHIBITOR RESISTANCE MECHANISM?

PARP inhibition seems to result in upregulation of the phosphoinositide 3-kinase (PI3K)/AK thymoma (AKT)/mechanistic target of rapamycin (mTOR) pathway [49], which has been implicated in numer- ous cancers due to its role in cell survival and proliferation. The rationale behind targeting PI3K (e.g., buparlisib, alpelisib), AKT (AZD5363), and mTOR (AZD2014) is in part to target a prosurvival pathway that may be contributing to PARP inhibitor resistance [50&&].

CONCLUSION

Although PARP inhibitor combination clinical trials are in their early stages, it is hoped that the utility of PARP inhibitors can be expanded beyond cancers with deleterious mutations in BRCA1 and BRCA2 and perhaps beyond cancers with defects in homol- ogous recombination repair. Whether or not this is possible, the results of these trials are certain to deepen our understanding of DNA repair mecha- nisms, cancer biology, and targeted therapies, thus contributing to the MK-4827 next iteration of therapeutic options for our patients.