Poly(ADP-Ribose) Polymerase Inhibitors in Pancreatic Cancer: A New Treatment Paradigms and Future Implications

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignancy. Most of the patients of PDAC present at later stages of disease and have a five-year survival rate of less than 10%. About 5–10% PDAC cases are hereditary in nature and have DNA damage repair (DDR) mutations such as BRCA 1 and 2. Besides having implications on screening and prevention strategies, these mutations can confer sensitivity to platinum-based therapies and determine eligibility for poly(ADP-ribose) polymerase inhibitors (PARPi). In the presence of DDR mutations and PARPi, the cells are unable to utilize the error-free process of homologous recombination repair, leading to accumulation of double stranded DNA breaks and cell death eventually. Various PARPi are in clinical development in PDAC in different subgroup of patients as monotherapies and in combination with other therapeutics. This review would focus on the mechanism of action of PARPi, history of development in PDAC, resistance mechanisms and future directions.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) cancer is an aggressive malignancy. In United States, it is currently the third leading cause of cancer-related mortality [1]. Most patients have advanced (locally advanced unresectable or metastatic) disease on presentation [2] and despite advances in development of multiagent cytotoxic regimens [3][4][5][6][7], the overall five-year overall survival (OS) is less than 10% [1]. There is an unmet need to develop new treatment strategies. Personalized, biomarker-based options for patients with advanced PDAC do exist, i.e., pembrolizumab for microsatellite instability-high (MSI-H)/mismatch repair deficient (d-MMR) [8] and larotrectinib for tumors with neurotrophic receptor tyrosine kinase (NTKR) gene fusions [9], but this accounts for <1% of total patient population.

BRCA Deficiency
BRCA1 (located on 17q21) and 2 (located on 13q12.3) are integral for genome stability by facilitating HR as mentioned above, and have an autosomal dominant pattern of inheritance with an incomplete penetration [30]. There are more than 1600 mutations associated with BRCA1/2, however, not all of them are considered pathogenic. The majority of the deleterious mutations leading to BRCA deficiency and subsequently HR deficiency (HDR) are insertions, frameshift mutations or nonsense mutations. Different ethnicities have a different prevalence of these mutations (i.e., most Ashkenazi Jews have one of the three founder mutations in either BRCA1 185delAG or 5382insC or BRCA2 6174delT) [37]. Cells with BRCA1/2 loss-of-function mutations cannot repair DSBs via HR but utilize NHEJ which could lead to accumulation of genetic alterations and ultimately lead to genetic instability or cell death [25]. Consequently, the presence of these mutations has been associated with increased risk of malignancies, including breast, ovarian and PDAC amongst others [38,39].

DNA Repair with BRCA-Deficient Cells in the Presence of PARPi
Utilizing PARPi in BRCA mutant malignancies is one of the first clinical applications of the age-old concept of synthetic lethality [40]. Synthetic lethality was described as a phenomenon in which a combination of two defects leads to cell death, but singularly neither of them has a detrimental effect individually [41]. As described earlier, in the presence of PARPi the cells are unable to repair SSBs with transformation to DSBs. In BRCA1/2-deficient cells, because of HRD, DSBs are repaired with either NHEJ or alternative-NHEJ, consequently leading to cycle arrest, DNA instability, and lethality ( Figure 1B) [41]. Normal cells on the other hand, utilize the functional BRCA protein and are able to repair DSBs, leading to cell survival. Therefore, PARPi are able to spare healthy cells, making them an ideal therapeutic agent in this setting. Besides the catalytic inhibitory properties, PARPi also trap PARP1 on damaged DNA and prevent its release and subsequent recruitment of DNA repair proteins, leading to formation of cytotoxic DNA complexes [42]. HRD is not limited in gBRCA1/2. A similar phenotype (BRCAness) can result from either somatic BRCA1/2 mutations or defects in other DDR genes such as PALB2, ATM, ATR, and FANC that are involved in HR [43][44][45], as well as from epigenetic BRCA silencing via promoter hypermethylation [46]. Identification of mutations beyond BRCA1/2 might be of importance as these tumors could also exhibit sensitivity to platinum-based regimens and PARPi [16,47,48]; however, the HRD score was neither predictive of a response nor survival in advanced PDAC patients treated with platinum-based therapy [49]. The significance of HRD in PDAC in terms of response to PARPi is an area of active investigation.

PARPi Pharmacology
PARP-1 is the most abundant enzyme of the PARP family and is involved in posttranslational modification of proteins involved in DNA repair. PARPi contain a carboxamide group that forms hydrogen bonds with serine hydroxyl and glycine backbone of the nicotinamide (NAD+) pocket site of PARP enzymes, mimicking NAD [50]. The adjacent benzene ring interacts with the tyrosine residue and makes π-π stacking interactions. Both PARP-1 and PARP-2 have identical NAD+ pockets and most PARPi inhibit PARP-1 and 2 enzymes similarly, with minor differences in selectivity [51]. In the presence of HRD due to BRCA1/2 mutations, the cells are unable to repair the double strand DNA (dsDNA) breaks through HRR and resort to non-conservative methods of DNA repair such as NHEJ, leading to DNA alterations and genomic instability [20]. In addition, PARPi trap PARP-1 at the site of DNA damage, preventing auto-PARylation and PARP-1 release.
Currently, six different PARPi (olaparib, rucaparib, veliparib, niraparib, talazoparib, and pamiparib) have been in clinical development at different stages. The comparisons in pharmacology are summarized in Table 1 and elaborated in this section. Single agent cytotoxicity is not directly proportional to PARylation inhibition and could be explained by differences in PARP trapping. In general, the IC 50 of PARPi falls in the low nanomolar range. Talazoparib is comparable to olaparib in catalytic activity in vitro (IC 50 : 4 vs. 6 nmol/L, respectively) and five-fold more potent compared to rucaparib (IC 50 : 21 nmol/L), however was~100 times more potent at trapping PARP and in terms of toxicity [42]. Talazoparib has superior trapping activity followed by niraparib; veliparib has the least trapping activity [42,52]. The differences in the trapping activity can be explained by differences in molecular structure with talazoparib being the largest and least flexible molecule [53,54]. The clinical and preclinical data with pamiparib is limited as of now. In a xenograft breast cancer model, it was found to be over 10 times more potent than olaparib [55].  [64], Talazoparib based on EMBRACA trial [65], Veliparib based on a Phase II trial [67], Pamiparib-Immature data from Phase I NCT 02361723 [55], Ongoing Phase III Pamiparib studies-NCT03519230, NCT03427814. Abbreviations: PARP = poly(ADP-ribose)polymerase; g = germline; s = somatic; m = mutant; CR = complete remission; PR = partial remission; Her-2 = Human epidermal growth factor receptor-2; Abd = abdominal; PFS = median progression free survival; OS= median overall survival.

Clinical Development of PARPi in PDAC
The clinical evolution of PARPi in PDAC has evolved from being used as monotherapies in refractory disease, to maintenance therapies and in combination with other classes of therapeutics ( Table 2). Contrasting the uniformly encouraging efficacy results during early development in ovarian and breast cancer between different PARPi, the reality has been different in PDAC with most patients not deriving or having a very short-term clinical benefit despite enrollment of a molecularly selected population, as we will discuss in this section of this review. Furthermore, the dual mechanism of action and the pharmacologic differences in between different agents makes evaluation of PARPi as a class challenging and does not allow easy incorporation into approved combination strategies in use in PDAC. Specifically, synergy with camptothecins appears to be mainly dependent on the inhibition of PARylation activity while for alkylating agents, the trapping activity is at least as important as PARylation inhibition [52]. In addition, due to overlapping toxicities with commonly used cytotoxic agents, the ability to use PARPi in combinations at doses that can achieve DNA trapping is limited [69].

PARPi as Monotherapy in Advanced Disease
One of the first clinical trials to evaluate the role of PARPi in metastatic PDAC patients was a phase II clinical trial involving olaparib monotherapy in patients with advanced recurrent cancers with gBRCA1/2 mutations [56]. The PDAC cohort (23 patients) had a mean two prior lines of therapy with 65% having prior platinum-based therapy. Compelling single agent activity was noted with an overall response rate (ORR) of 22% regardless of platinum exposure. Responses were durable with stable disease (SD) >/8 weeks in 35% of patients and its was well tolerated.
The activity of rucaparib as single agent was tested in the RUCAPANC phase II trial involving both somatic BRCA1/2 (n = 3) or gBRCA1/2 mutant (n = 16) advanced PDAC patients [70]. Seventy-nine percent of patients were exposed to platinum and 42% were refractory to platinum. The study did not meet its primary endpoint with ORR of 16% and disease control rate (DCR) of 32%. None of the patients with response had progressed on prior platinum-based therapy and 3 out of the 4 patients with response had only one prior line of therapy.
Veliparib was evaluated in a phase II single arm trial in advanced, pre-treated PDAC patients, with a known gBRCA1/2 or PALB2 mutation, including platinum-resistant disease [71]. No radiological response was observed; SD was seen in 25% patients at 8 weeks.
First-in-human phase Ib trial (NCT01286987) of talazoparib was conducted in gBRCA1/2 mutant patients in advanced refractory solid tumors including metastatic PDAC (n=13, median prior chemotherapy regimens and prior platinum regimens of 2.5 and 1, respectively). It was well tolerated and had an ORR of 20% and SD in 10% [76].
Two parallel phase II clinical trials in US and Israel (NCT02677038, NCT02511223) are testing olaparib in patients with >/1 lines of therapy in metastatic BRCAness PDAC, defined as either DDR deficiency beyond gBRCA1/2 mutation or a family history of BRCA-associated cancers in the absence of DDR deficiency or ATM loss by immunohistochemistry. Preliminary results (45% enrollment) showed efficacy signal only in patients with platinum-sensitive disease [72].

PARPi as Maintenance Therapy in Platinum-Sensitive Disease
Based on available evidence, the window of opportunity for benefit from PARPi in BRCA-deficient PDAC appears to be as maintenance therapy in platinum-sensitive disease. The randomized, double-blind placebo-controlled phase III POLO trial evaluated the role of maintenance olaparib in patients with gBRCA1/2 mutant metastatic PDAC, with non-progressive disease during first-line treatment with platinum-based chemotherapy for a minimum of 16 weeks [66]. Longer PFS with olaparib was observed (7.4 months) as compared to 3.8 months with placebo (HR 0.53; 95% CI 0.35 to 0.82; p = 0.004) and the study met its primary endpoint. The overall survival (OS) was similar (46% data maturity). The rate of grade 3 or higher treatment-related AEs with higher with olaparib (40%) compared to placebo (23%) with no deterioration of quality of life.
A single arm phase II study (NCT03140670) is evaluating the role of maintenance rucaparib advanced PDAC patients with either germline or somatic BRCA1/2 or PALB2 mutations, with non-progressive disease after at least 16 weeks of platinum-based regimen [73]. Early results are encouraging with a median PFS of 9 months and ORR of 37%. Clinical benefit was durable with DCR of 90% patients for at least 8 weeks.
NIRA-PANC is a phase II clinical trial (NCT03553004) assessing the role of maintenance niraparib after >1 line of therapy in metastatic PDAC patients with either germline or somatic HRD. No results have been reported so far [77].

PARP in Combination with Chemotherapy
The rationale behind combining PARPi with chemotherapeutic agents has been illustrated in Figure 2A. Oxaliplatin has been shown to have synergistic antitumor activity with PARPi in vivo.
Interestingly the combination prevented oxaliplatin-induced neurotoxicity [78]. Several clinical trials are evaluating PARPi with platinum-based therapy. A phase I/II clinical trial (NCT01489865) investigated the combination of veliparib with mFOLFOX6 (folinic acid+ 5-FU+ oxaliplatin) in metastatic, pre-treated PDAC patients [75]. Patients in phase II (n = 33) were pre-selected for germline or somatic DDR mutations (69%) or have a family suggestive of a breast or ovarian cancer syndrome (BOCS, 27%) and included both pretreated (18 patients) and untreated (15 patients). The ORR for the whole cohort was 26%. ORR was higher in platinum-naïve (33%) vs. platinum pretreated (7%), patients with BOCS (30%) vs. not (14%), and DDR mutation positive (50%) vs. negative (17%). The ORR for the platinum-naïve, DDR mutation positive, BOCS cohort was 58% highlighting the importance of patient selection. The median PFS and OS for these patients was 8.7 and 11.8 months respectively (all patients, PFS 3.7 months/OS: 8.5 months). The combination was well tolerated with grade >/3 treatment-related AEs of myelosuppression (16%) and nausea, vomiting (6%). Veliparib with oxaliplatin/capecitabine in patients with a known BRCA1/2 mutation or BRCA-related advanced malignancies was also found to be safe in a phase I clinical trial (NCT01233505) [79]. Whether the incorporation of PARPi in a platinum backbone really leads to improved outcomes (vs. using as maintenance) remains to be seen. A phase II trial (NCT01585805) is enrolling BRCA/PALB2 mutated advanced PDAC patients in the front-line setting comparing gemcitabine/cisplatin with and without veliparib, and will help answer this question [80]. Topoisomerase inhibitors (TIs) stall the RF by stabilizing the DNA complex in unrepaired state, enhancing SSBs [81]. Studies showed that PARPi can enhance the cytotoxicity of TIs at concentrations lower than the ones necessary for PARP synthesis in HRD cells [82]. SWOG S1513 (NCT02890355) randomized mPDAC patients to veliparib plus modified FOLFIRI (Folinic acid + 5-FU + Irinotecan) vs. FOLFIRI alone as second-line treatment in a biomarker unselected population [74]. Nine percent of patients had mutation in HRD gene and 20% had mutations in other DDR genes. Interim analysis showed that the combination arm did not have an OS benefit (5.1 vs. 5.9 months; HR 1.3, 95% CI 0.9-2). Similarly, there was no benefit from the addition of veliparib in patients with HRD or non-HRD mutations (OS 7.4 vs. 9.4 months in FOLFIRI alone). The incidence of grade >/3 treatment-related AEs like neutropenia, fatigue, and nausea was higher in the veliparib arm [74]. A phase I/II trial (NCT03337087) is investigating rupacarib along with liposomal irinotecan, leucovorin, 5-FU in gastrointestinal malignancies including PDAC, with preselection in phase II portion in PDAC cohort for HRD (BRCA or PALB2 mutated or HRD non BRCA/PALB2 non-mutated). Gemcitabine in combination with PARPi is more cytotoxic against PDAC in vitro compared to gemcitabine alone [83]. A phase I study (NCT00515866) is exploring the safety of olaparib with gemcitabine in the front-line setting with advanced solid tumors, including advanced PDAC, without any selection for specific mutations [84].

PARP in Combination with Radiotherapy
PARPi were found to have radiosensitizing effects in vitro [85]. Pretreatment with rucaparib followed by gemcitabine and radiation therapy increased cytotoxicity by stalling the cells in G2/M phase leading to apoptosis and cell death. Furthermore, in vitro and in vivo experiments showed that veliparib in combination with radiation led to a better tumor growth inhibition and survival, rather than single-agent activity of either of the two agents [86]. Veliparib inhibited polymerization of PAR protein with compensatory upregulation of p-ATM and PARP with enhanced DNA damage and apoptosis. The combination of PARPi and radiation therapy with or without chemotherapy is being investigated in two clinical trials. Niraparib is being tested in a phase II proof of concept trial (NCT03601923) in patients with advanced PDAC harboring HRD germline or somatic mutations (BRCA1, BRCA2, PALB2, CHEK2, or ATM mutations) in the second line setting, with exclusion of patients with progressive disease on oxaliplatin-based regimens. Palliative radiation therapy to be administered 1 week before the start of niraparib. Veliparib is being evaluated in a phase 1 clinical trial (NCT01908478) in patients with locally advanced unresectable or borderline resectable PDAC in combination with gemcitabine and intensity modulated radiation therapy in the front-line setting with no selection for mutations.

PARPi in Combination with Immunotherapy
The synergy between PARPi and immunotherapy is likely multifactorial as shown in Figure 2B. In a high-grade serous ovarian carcinoma model, HRD tumors harbored a higher neoantigen load with an increased tumor-infiltrating lymphocytes (TILs) and PD-1/PD-L1 expression as compared to non HRD tumors ( Figure 2B, 10) [87]. Besides the enhanced neoantigen load, S-phase DNA damage induces activation of PD-L1 through stimulator of interferon genes (STING) pathway [88]. Exposure of double stranded DNA in cytoplasm activates cyclic GMP-AMP synthase (cGAS) and catalyzes production of cyclic-dinucleotide (CDN) ( Figure 2B, 2-4), which activates STING with subsequent activation of either the NF-κB or the TBK1-IRF3-Type I IFN pathways ( Figure 2B, 5-8) [89]. Type I IFN have immunostimulatory effects such as upregulation of MHC and CCR7 leading to enhanced dendritic cell function [90], increase in T-helper cells chemokines (CXCL9 and CXCL10) [88] and potentiation of cytotoxic T-cell lymphocyte function ( Figure 2B, 9) [91]. Furthermore, it suppresses regulatory T-cells (T reg ) cells by downregulating cyclic AMP (cAMP) [92]. PARPi upregulated PD-L1 expression in an in vivo breast cancer model via GSK3β inactivation. Simultaneous treatment with PD-L1 blockade and olaparib led to enhanced T-cell mediated killing of tumor cells [93]. Similarly, in a BRCA1-deficient ovarian cancer model, PARPi led to activation of STING pathway and the therapeutic efficacy was improved with PD-1 blockade [91]. Interestingly, synergistic anti-tumor activity of niraparib and PD-1 blockade was seen in both BRCA-deficient and BRCA-proficient immunocompetent tumor models indicating that this combination may be active regardless of BRCA status. Furthermore, mice in complete remission after treatment rejected implanted tumor cells indicating the generation of memory T-cells [94]. Preclinical studies have shown that DSBs leads to upregulation of PD-L1 expression through ATM/ATR/Chk-1. Furthermore, treatment with radiotherapy/PARPi in BRCA2-depleted cells leads to enhancement of Chk-1-dependent PD-1 upregulation [95]. The therapeutic efficacy of PARPi and Chk-1 inhibitors with PD-L1 blockade was seen in a small cell lung cancer carcinoma model through activation of STING pathway [96]. The data of combination of PARPi with anti-CTLA-4 is limited. Therapeutic efficacy of the combination was seen in a BRCA1-deficient ovarian cancer model [97].
A large number of clinical trials are exploring the combination of immunotherapy and PARPi across various tumor types. Preliminary data from a phase II study (NCT02484404) in unselected metastatic prostate cancer showed that the combination of olaparib and durvalumab is efficacious (8/17 or 47% patients had PSA responses >50%) and is tolerable [98]. A phase II (MEDIOLA) trial in relapsed gastric cancer with a 4-week run in of olaparib, followed by combination with durvalumab failed to show efficacy (DCR at 12 weeks was 26%) due to early PD during the run-in period [99]. Combination of niraparib and pembrolizumab in a phase II trial (TOPACIO/Keynote-162) in triple negative breast cancer (TNBC) and recurrent ovarian cancer patients showed that the combination is safe and effective, irrespective of the platinum exposure or BRCA 1/2 or PD-L1 status [100,101]. A phase I study involving a combination of olaparib, durvalumab, and the vascular endothelial factor receptor (VEGFR)-1 inhibitor cediranib in ovarian/endometrial/TNC patients showed that the combination is safe; there was an efficacy signal with DCR of 67% [102]. The ongoing phase Ib/II trial PARPVAX study (NCT03404960) evaluates niraparib with either nivolumab or ipilimumab as maintenance therapy in patients with advanced, platinum-sensitive PDAC [103].

Molecular Targeted Therapy Combinations
Preclinical data has suggested synergism between PARPi and MEK inhibitors in RAS-mutant cells. A study involving PDAC cells showed that the synergy is mediated through multiple mechanisms including increased expression of FOXO3a leading to apoptosis, decrease in HR ability, increased PARP protein, enhanced PARPi induced DNA damage, and increased hypoxia through decreased vascularity [104]. The combination was effective in cells without BRCA mutations. One of the phase II feasibility studies (NCT04005690) would entail administering olaparib or MEK inhibitor cobimetinib in patients with resectable PDAC in the neoadjuvant setting, with biomarker evaluation before and after therapy. A phase Ib/II study is evaluating talazoparib in combination with avelumab and binimetinib in patients with NRAS/KRAS mutant PDAC and other advanced solid tumors after previous 1-2 lines of therapy.
Preclinical data has shown anti-angiogenic effect of PARPi through inhibition of endothelial cell migration at concentrations lacking cytotoxic effects, suggesting a novel therapeutic implication of PARPi [105]. A single arm phase II clinical trial (NCT02498613) is exploring the combination of olaparib with the VEGF inhibitor cediranib in patients with advanced solid tumors including advanced PDAC in genetically unselected population. The SMMART trial (NCT03878524) is an open-label biomarker-driven phase Ib trial in patients with PDAC, breast, prostate cancer, and acute myeloid leukemia involving administration of a combination of two drugs from 35 experimental therapeutics, including olaparib based on molecular testing.

Resistance Mechanisms
The various mechanisms through which tumor cells acquire resistance to PARPi have been illustrated in Figure 3. Several studies suggested that acquired BRCA mutations in patients with gBRCA1/2 could lead to restoration of HR and confer resistance of platinum agents and PARPi. An in vitro study by Sakai et al. [106] demonstrated that intragenic mutations in BRCA2 mutant PDAC and ovarian carcinoma cell lines restore the BRCA2 reading frame leading to both cisplatin and PARPi. Norquist et al. [107] showed that these secondary reversion mutations were found in 29% of gBRCA1/2 mutant recurrent ovarian carcinomas and were predictive of resistance to platinum containing chemotherapy and PARPi. A UK study [108] demonstrated olaparib resistance with emergence of a secondary BRCA2 mutation with restoration of its function. Figure 3. Mechanisms of resistance to PARPi. Abbreviations: RF = replication fork; MRE 11 = mitotic recombination 11; miRNA = microRNAs; HR = homologous recombination; NHEJ = non-homologous end joining; PI3K = phosphatidylinositol 3-kinases. Cells with HRD can acquire resistance through upregulation of an alternative-NHEJ error-prone pathway, known as microhomology-mediated end-joining (MMEJ) repair. MMEJ repair requires DNA polymerase POLQ for repair of DSBs [109]. Preclinical data suggests that therapeutic targeting by POLQ inhibition can have synthetic lethality effect in HRD cells [109,110].
In the absence of BRCA2 reversion mutations, tumor cells can evade lethality by PARPi through protection of RF. Chaudhuri et al. [36] showed that deficiencies of MLL3/4 in BRCA2 mutant cells inhibits the recruitment of MRE11 to the RF in vitro, thereby preventing the RF degradation. Furthermore, 53BP1 regulates the balance between HRR and NHEJ [111]. 53BP1 loss downregulates NHEJ and promotes error free HR, thereby conferring resistance of PARPi [111]. Finally, microRNAs can modulate sensitivity to PARPi. For example, miR-107 and miR-222 downregulate expression of RAD51, thereby impairing DDR by HR and enhancing sensitivity to olaparib [112]. In addition, miR-622 mediated resistance to PARPi and cisplatin in BRCA1 mutant cells by diverting the repair towards HR pathway and suppressing the NHEJ via Ku complexes [113].
DDR is dependent on the phase of cell cycle; modulating cell cycle checkpoints can potentially alter the effect of PARPi in tumor cells. Inhibition of CDK12 leads to downregulation of DDR genes [114] and reverses resistance to PARPi in breast cancer [115] and multiple myeloma in vitro [116]. PARPi-induced DNA damage can activate the ATR-mediated G2/M checkpoint facilitating DNA repair which can be reversed by ATR inhibition leading to cell death [117]. Similarly, Wee1 inhibition can allow unrepaired DNA enter mitosis via the G2/M checkpoint but the results of phase I study with olaparib with Wee1 inhibitor in unselected were disappointing with ORR of only 11% [118]. MET phosphorylates PARP1 at Tyr907, increases its enzymatic activity thereby reducing the affinity for PARPi and leads to resistance [119]. PARPi in combination with a MET inhibitor can overcome PARPi resistance in vitro [120]. The phosphoinositide 3-kinase (PI3K) signaling pathway has a role in tumorigenesis through maintenance of HR steady state [121]. A study in TNBC showed that PI3K inhibition leads to sensitization to PARPi by downregulation of BRCA1/2, increased poly ADP-ribosylation and DNA damage [122]. Finally, in BRCA-deficient cells, resistance due to upregulation of genes responsible for p-glycoprotein with subsequent increased efflux of PARPi can develop, which can be counteracted by p-glycoprotein inhibitors [123]. A next generation PARPi (AZD2461), that is not a substrate for p-glycoprotein is in development and is thought to potentially overcome p-glycoprotein-mediated olaparib resistance [124].

Discussion
Advanced PDAC is one of the most challenging malignancies to treat. PDAC is catching up with the current wave of precision medicine with the goal of improving patient outcomes. Besides two tissue-agnostic therapies (pembrolizumab for MSI-H/d-MMR and larotrectinib for NTRK fusion gene positive tumors), PARPi are an exciting addition to our armamentarium of drugs for patients with pathogenic gBRCA1/2 mutations and olaparib is endorsed by the National Comprehensive Cancer Network (NCCN) as appropriate maintenance therapy after at least 4 months of platinum-based therapy provided that there is no interim disease progression [125]. With the advent of emerging clinical data, there are several questions that remain answered.
Firstly, whether the benefit seen in the gBRCA mutant patients in POLO trial could be extrapolated to tumors with somatic BRCA mutation or in patients with other HRD or DDR mutations remains unclear. The one patient with somatic BRCA2 and the two patients with gPALB2 mutations enrolled in the maintenance rucaparib study so far (24 out of 42 planned patient accrual) attained an objective response indicating that the benefit of PARPi potentially extends beyond gBRCA1/2 [73]. Data from a large number of clinical trials (NCT02042378, NCT02677038, NCT02511223, NCT03140670, NCT02890355, NCT01489865) evaluating the role of PARPi in patients with somatic mutations would provide further clarity regarding the extent and durability of response in this group of patients. The NCCN endorses testing patients for infrequent but potentially actionable somatic alterations including BRCA1/2 mutations [125]. Whether archival or fresh tissue should be used for somatic mutation screening is unclear. Data from retrospective ovarian cancer tumor sample analysis suggests that BRCA1/2 loss occurs early in the course of the disease [126] but it is not currently known if this is true in PDAC. Validation of predictive biomarkers beyond gBRCA1/2 mutations, NTKR fusions, and MMR/MSI status is an unmet need in PDAC.
Secondly, the appropriate timing of the use of PARPi in the disease course of PDAC is unclear. As mentioned in the sections earlier, PARPi are being tested both as monotherapy (maintenance and refractory) and in combination with other agents (front-line and subsequent lines of treatment).
In the maintenance setting, olaparib led to PFS benefit in gBRCA1/2 platinum sensitive patients in POLO trial [66]. However, it is important to note that the preliminary data suggest no OS benefit and it is possible that use of PARPi in subsequent lines of therapy can be as beneficial as in the maintenance setting, but whether this is dependent on platinum sensitivity is unclear. Rucaparib in patients with somatic or germline BRCA1/2 mutations [70] and olaparib in patients with BRCAness [72] showed benefit only in platinum-sensitive disease, while with olaparib in patients with gBRCA1/2, the benefit was seen irrespective of the platinum-exposure [56]. Unfortunately, the initial report of the POLO study does not mention subsequent use of PARPi in patients treated with placebo. In addition, whether PARPi is superior to 5-FU maintenance as used in the PANOPTIMOX study is unknown [127]. There is currently no head-to-head comparison between maintenance PARPi and chemotherapy. Also, the appropriate time to switch therapy to PARPi is unclear. In the POLO trial [66], one-third of patients received platinum-based therapy for more than 6 months. On subgroup analysis for PFS, patients who received platinum-based therapy for >6 months seemed to have a greater benefit (HR 0.35, 95% CI 0.17-0.72) as compared to 4-6 months (HR 0.69, 95% CI 0.43-1.12). Finally, it is unclear if there is any benefit to platinum re-induction after progression on PARPi. Preclinical data reveal cross-resistance between PARPi and platinum chemotherapy [128] and, in an unselected patient population in the PANOPTIMOX study, reintroduction of FOLFIRINOX (folinic acid + 5-FU + irinotecan + oxaliplatin) upon disease progression led to a marginal benefit of 1.4 months in terms of median PFS [PFS1 (defined as time to first progression of disease) of 5.7 mo, PFS2 (defined as time to disease progression during FOLFIRINOX) of 7.1 mo [127]. Using an alternative alkylating agent such as mitomycin C, might be preferable to platinum re-induction but only anecdotal evidence exist so far [129]. Furthermore, there is no data to support switching from one PARPi to another at the time of disease progression.
Is it time to start using PARPi in combination with chemotherapeutic agents, be it in the first or subsequent lines of treatment? Though the mFOLFOX-6 plus veliparib combination was promising, especially in highly selected patients who are platinum-naïve and have DDR mutation and BOCS history, the lack of direct comparison to chemotherapy limits the upfront use of this strategy [75]. As a cautionary tale, SWOG S1513 showed detrimental effects for the combination, even for the subgroup of patients with DDR mutations [74]. Both trials showed an increased toxicity profile with the combination arms, indicating that the benefit of the combination (if at all) would potentially come at the cost of an increased toxicity [74,75]. The direct comparison of gemcitabine/cisplatin with and without veliparib in front-line setting in BRCA1/2/PALB2 mutated PDAC will provide further insight [80]. Several other clinical trials are evaluating PARPi in combination with different chemotherapies in front-line (NCT01282333) or subsequent lines of therapy (NCT03337087, NCT01233505, NCT00515866).
Though relatively well tolerated, PARPi do have a toxicity profile that needs to be kept in mind while making therapy decisions and it is important to understand that different PARPi differ in their potencies for catalytic inhibition, PARP binding, and inhibition of PARylation; and hence, have differing toxicity profiles and potentially efficacy and therefore, cannot be assumed that can be used interchangeably. For example, rucaparib leads to elevated creatinine based on inhibition of renal transport proteins and has the highest rate of transaminitis [63]. On the other hand, niraparib has the highest rate of bone marrow suppression and resultant hematologic toxicities [64]. In POLO study [66], the PFS benefit came at the cost of increased incidence of grade 3 AEs. Even though there was no notable deterioration in QoL, it is worth noting that majority (67%) of the patients had excellent performance status (ECOG 0). In the real world, the PFS benefit could come at the cost of increased toxicity profile and QoL deterioration, as majority of PDAC patients do not have preserved performance status.
Beyond early identification of patients with gBRCA1/2 pathogenic mutations, how can we better select patients for PARPi maintenance after platinum therapy? The development of reversion somatic BRCA2 mutations during platinum-based therapy that limits the efficacy of PARPi. Tumor mutational profiling using next generation sequencing after completion of platinum induction can assist in identifying patients with secondary mutations but can be challenging, especially for patients with locally advanced disease and no easily accessible lesions. Identification of acquired resistance mutations in circulating tumor DNA (ctDNA), collected during or after platinum induction, can be helpful. Based on available data, patients with BRCA2 reversion mutations should not be treated with PARPi maintenance; rather, 5-FU maintenance should be selected. For patients with initial benefit on PARPi maintenance, enrollment in trials with novel combinations aiming to bypass resistance, such as ATR or Wee1 inhibitors, should be encouraged.
If maintenance therapy after non-progression on platinum-based therapy remains the right place for PARPi in patients with gBRCA1/2 patients, how can we improve on the results of POLO study? The activation of antitumor immunity with PARPi and the so far documented safety in combination with immune checkpoint inhibitors begs for evaluation after initial induction chemotherapy. A study with niraparib with nivolumab/ipilimumab is currently ongoing (NCT03404960) [103]. Furthermore, PARPi combinations with MEK or VEGFR inhibitors and or radiotherapy can potentially improve outcomes.

Conclusions
In conclusion, PARPi are potential therapeutic options in advanced PDAC in select patients with gBRCA1/2 pathogenic mutations and platinum sensitivity. Combining PARPi with other classes of drugs poses the challenge of balancing the therapeutic benefit and potential overlapping toxicities. Besides careful selection of patients and drugs, testing the drugs sequentially rather than in combination could be another strategy to achieve an optimal risk vs. benefit ratio. Future efforts should entail a better understanding of the underlying mechanism of actions, resistance mechanisms, and biomarker development to achieve maximal therapeutic benefit at the cost of minimal side effects. A number of clinical trials are exploring this promising class of drugs in different patient populations.
Author Contributions: M.G.: data curation, original draft preparation, writing-review and editing; R.I.: Writing-review and editing; C.F.: Conceptualization, supervision, project administration, and writing-review and editing.