Poly(ADP‐ribose) polymerase inhibition in pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) is a lethal disease with limited treatment options. Recently, the poly(ADP‐ribose) polymerase inhibitor (PARPi) olaparib has been approved for maintenance therapy after successful platinum‐based chemotherapy in patients with germline mutations in BRCA1 and BRCA2. Approval was based on the POLO study that has shown a significant improvement in progression‐free survival for patients with metastatic PDAC after at least 4 months of platinum‐based chemotherapy. Hopefully, this first biomarker‐directed targeted therapy for a relevant subgroup of pancreatic cancer patients is only the beginning of an era of personalized therapy for pancreatic cancer. The potential role for PARPi in improving survival in patients with pancreatic cancer containing somatic tumor mutations has yet to be established. Multiple studies investigating whether PARPi therapy might benefit a larger group of pancreatic cancer patients with homologous recombination repair deficiency and whether combinations with chemotherapy, immunotherapy, or small molecules can improve efficacy are currently underway. We here review the molecular basis for PARPi therapy in PDAC patients and recent developments in clinical studies.

on the POLO study that has shown a significant improvement in progression-free survival for patients with metastatic PDAC after at least 4 months of platinum-based chemotherapy. Hopefully, this first biomarker-directed targeted therapy for a rele- Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal disease that has a 5 year survival of less than 10%, one of the worst outcomes of all major cancers. 1 According to the GLOBOCAN data, 458 918 new cases and 432 242 deaths worldwide have been estimated for 2018. 2 In contrast to most other cancer types, improvement of overall survival over the last decades has been very limited. 3 The main reason for the dismal survival is the fact that PDAC can only be cured by complete resection, but more than 80% of patients are diagnosed at a late disease stage when primary surgery is no longer possible. The ESPAC-4 4 and PRODIGE-24 5 studies have markedly improved adjuvant therapy and outcome after resection, but most patients still have recurrence after surgery. The mainstay of treatment for patients that cannot be resected or relapse after surgery is systemic chemotherapy. 6 In contrast to other cancer types like breast cancer or colon cancer, PDAC is a relatively chemotherapy-resistant disease, and for most patients, only a small survival benefit can be achieved. Patients with metastatic disease treated with the currently most effective chemotherapy protocols, FOLFIRINOX and nab-paclitaxel/gemcitabine, have a median survival of only 11.1 months 7 and 8.5 months, 8 respectively. Immunotherapy, now an important treatment option in multiple cancers, has not yet been equally successful in PDAC. 9 Targeted therapy is also rarely possible since more than 90% of PDAC patients have an activating mutation in the Kirsten rat sarcoma (KRAS) gene, 10 which is the most important driver mutation in this cancer. 11 There is no effective targeted therapy for KRAS mutated cancers, 12 with the novel exception of the p.
G12C mutation. 13 Although this variant is quite commonly found in lung cancer it is very rare in PDAC. 14 In the small subgroup of patients with KRAS wild type, addressable genomic alterations including structural variants are increasingly found and thus opportunities for targeted therapy arise, for example, in patients with NTRK-, NRG-1-, RET-, ROS1and ALK-fusions or BRAF mutations. [15][16][17] There are promising attempts to use transcriptional PDAC subtypes to guide systemic therapy, but this approach is not yet ready for clinical practice. 18,19 The poly(ADP-ribose) polymerase inhibitor (PARPi) olaparib has recently been licensed for PDAC patients with germline BRCAmutations based on the randomized phase III POLO study. 20 The option to treat a relevant fraction of PDAC patients with biomarkerguided targeted therapy is an exciting novelty for this notoriously difficult to treat cancer type and the focus of this review.

| HOMOLOGOUS RECOMBINATION REPAIR DEFICIENCY IN PANCREATIC CANCER
Genome instability caused by defects in the cellular DNA repair machinery has been recognized as a fundamental "enabling characteristic" of cancer. 21 Homologous recombination repair (HRR) of DNA double-strand breaks is a crucial component of the DNA damage response. 22 Germline mutations in genes involved in HRR including BRCA1/2 have long been known to increase tumor risk and lead to hereditary tumor syndromes. 23 Loss-of-function mutations in BRCA1/2 can lead to homologous recombination repair deficiency (HRD) with accumulation of chromosomal rearrangements and copynumber alterations. Interestingly, tumors in patients with pathogenic germline BRCA mutations can also arise without signs of genomic instability, possibly since inactivation of the wild type allele might be a requirement for HRD. 24,25 HRD can also be caused by somatic mutations in BRCA1/2 and other HRR genes in patients without pathogenic germline mutations. Finally, tumor cells can show characteristics of HRD without mutations in BRCA genes or other HRR genes. The term "BRCAness" was initially coined for tumors with HRD characteristics in the absence of germline BRCA mutations but is now also used in a broader sense for all tumors with HRD. 26,27 While the detection of mutations in HRR genes in tumor tissue or, for germline mutations, in leukocytes, is commonly used in clinical routine, the validation of biomarkers and tests that directly detect HRD is still developing and an active area of research (reviewed in Reference 28).
Germline mutations in BRCA1/2 result in an increased lifetime risk for PDAC (2-to 4-fold for BRCA1 and 3-to 8-fold for BRCA2). 29 In three large US studies, the prevalence of BRCA1 and BRCA2 germline mutations in PDAC patients was 0.3% to 2.4% and 1.4% to 5.7%, respectively. [30][31][32] Other HRR genes with germline mutations were ATM (1.1%-2.88%) and PALB2 (0.2%-0.4%). Higher rates of germline BRCA mutations have been reported for Ashkenazi Jewish PDAC patients, 33 ranging from 5.5% in one study with 137 resected patients 34 to 21.6% in another single-site study with 37 patients. 35 HRD can also arise during PDAC carcinogenesis without predisposing germline mutations. It is difficult to estimate the prevalence of HRD mutations or "BRCAness" in PDAC patients since studies that report mutation frequencies have mostly been performed in selected populations, for example, resected stage I/II patients or patients from certain geographical regions. It is also conceivable that patients with a family history of pancreatic cancer have been preferentially referred to genetic profiling, introducing a selection bias that might overestimate the prevalence of germline mutations associated with familiar pancreatic cancer. Furthermore, many variants in genes related to HRR have not yet been clinically validated, and their relevance for therapeutic interventions remains unclear.
In the study by Wadell et al, 36 whole-genome sequencing and copy number variation analysis was performed on 100 patients with primary operable, nonpretreated PDAC. Based on variations in chromosomal structure, four PDAC subtypes were described: stable, locally rearranged, scattered, and unstable. The genomic unstable subtype (14%) correlated with germline and somatic mutations in HRR genes and a mutational BRCA signature. 37 Seven percent of patients in this study had germline mutations in BRCA2 (n = 4) and PALB2 (n = 3), and further 7% of patients had somatic mutations in BRCA1/2 and PALB2. Altogether, 24% of patients had a BRCA signature and/or and unstable genome and therefore signs of HRD, some of them without an unequivocal causative mutation. In a large study with genomic profiling of 3594 PDAC samples, 14% of patients had mutations in HRR-related genes. 17 Clinical information for this cohort is sparse, including the tumor stage and the trigger for molecular testing. There was no parallel germline sequencing, but it was estimated with a computational method that approximately half of the patients with mutations in HRR-related genes had germline alteration. In a detailed study on transcription phenotypes with microdissected tumor tissue from 206 resected patients and 111 patients with advanced disease from the COMPASS trial, 7% of patients had a unique mutational signature indicating HRD. 19 Interestingly, HRD was not associated with any of the transcription subtypes. In a recent study with 62 patient-derived PDAC cell lines, predictive biomarkers for HRD and response to platinum derivates and olaparib were described that also identify patients without mutations in the key HRR genes. 38 These results indicate that the number of patients that might benefit from PARPi therapy is possibly not limited to patients with germline BRCA1/2 mutations. Further clinical studies are needed to evaluate the benefit for a larger group of patients, and to define the optimal testing strategy. 39

| THERAPEUTIC CONSEQUENCES OF HRD IN PANCREATIC CANCER
PARPi are drugs that exploit HRD to kill tumor cells based on a concept termed "synthetic lethality." 40 43,44 In the clinic, PARPi are currently used as single agent therapy, but combination treatments may increase efficacy. Interestingly, sequential therapy involving PARPi and cell cycle checkpoint inhibitors can work not only in HRD tumors but also in tumors without HRD. 45 Tumor cells with high-endogenous replication stress are more susceptible to therapy with cell cycle checkpoint inhibitors. 38 In PDAC, the basal/squamous subtype has increased endogenous replication stress relative to the classical subtype, and basal/squamous tumors are therefore more likely to be susceptible to cell cycle checkpoint inhibitors irrespective of HR deficiency status. Thus, there might even be a role for PARPi combinations in HRR competent PDAC.
PARPi-induced DNA damage is known to trigger T-cell infiltration with potential anti-tumor effects, but also adverse inflammatory responses and increased immune checkpoint signaling. [46][47][48] Furthermore, HRD in PDAC can be associated with a high-mutational burden. 49 There is therefore a rationale for the combination of PARPi and immune checkpoint blockade. 50 Whether such combination approaches will help to overcome the resistance of pancreatic cancer to immunotherapies has to be determined.
In addition to conferring sensitivity to PARPi, the inability to repair double-strand DNA breaks also increases the sensitivity of cancer cells to certain classes of chemotherapy, including platinum derivates. Interestingly, differences in the mode of action have been described between individual platinum-derived agents: cisplatin and carboplatin cause DNA cross-linking and induce a DNA damage response, while oxaliplatin kills tumor cells by inducing ribosome biogenesis stress. 51 An increased sensitivity of patient-derived cell lines and PDX models with HRD to platin derivates has been demonstrated. 38 In the whole-genome sequencing study discussed above, hints for an association of unstable genome PDACs with successful platinum response were found, but this was based on only eight patients. 36 In a retrospective analysis of resected patients after neoadjuvant chemotherapy with FOLFIRINOX the complete pathological response rate was higher in patients with germline BRCA mutations (44.4% vs 10%) which was also reflected in a better overall survival. 52 Further evidence for an increased susceptibility of HRD PDAC to platinum derivates has been reviewed in Reference 33. In a recent trial with cisplatin and gemcitabine for germline BRCA/PALB2-mutated stage III and IV PDAC patients, an exceptional response rate of 65.2% has been reported. 53 The majority of patients in the study (56%) had one of the Ashkenazi Jewish "founder" mutations. It remains unclear if cisplatin/gemcitabine is superior to FOLFIRINOX in HRD PDAC patients since randomized trials are lacking and a response rate for FOLFIRINOX in PDAC patients with HRD has not yet been reported.
The triple combination of cisplatin, gemcitabine, and nab-paclitaxel that had a response rate of 70% in unselected stage IV PDAC patients 54 might also be a promising protocol in HRD PDAC patients.
Notably, exceptional responses in HRD PDAC have also been reported for the alkylating agent mitomycin C 55 and the topoisomerase inhibitor irinotecan. 56

| COMPLETED CLINICAL STUDIES
To date, 14 clinical trials of PARPi involving patients with pancreatic cancer have been reported (Table 1)  advanced solid tumors also included patients with pancreatic cancer (Table 1).

| Olaparib
Olaparib was the first PARPi to enter clinical trials as a single agent cancer therapy. 59 Along the clinical pipeline in pancreatic cancer, olaparib is also the most advanced candidate. The landmark phase 1 study of olaparib in BRCA1/2 mutation carriers included two patients with pancreatic cancer in the safety cohort, however, no data on antitumor activity were published for these patients. 59 Two subsequent phase 1 to 2 studies of olaparib included patients with pancreatic cancer, and the majority in both trials were germline BRCA1/2-positive (

| Rucaparib
Rucaparib has been evaluated in completed trials both as a monotherapy in pretreated patients and as a maintenance strategy for platinum-sensitive pancreatic cancer.
A phase 2 trial (RUCAPANC) tested single agent rucaparib (600 mg twice daily) in patients with pretreated locally advanced or metastatic pancreatic cancer and either germline (n = 16) or somatic (n = 3) BRCA1/2 mutations. 65 The ORR was 16%, including at least one CR, and the observed DCR (disease control rate) was 32%.
Recruitment was stopped based on an unmet efficacy threshold among the first 15 cases. Patients had received a maximum of two prior lines of therapy, predominantly including platinum-based regimens (78.9%). Of note, all patients experiencing a response to rucaparib had platinum-sensitive disease.

| Veliparib
Veliparib has been extensively studied in pancreatic cancer, either as single agent or in combination with existing regimens (

| Combination therapy
Anti-tumor effects of PARPi not only depend on HRD, but also on the rate at which DNA damage is accumulated. A number of existing cancer therapies increase genotoxic stress at the cellular level. 74,75 It is hoped that the co-administration of PARPi with some of these agents will further improve clinical response rates and efficacy. Beyond

| Combination with chemotherapies
Based on the phase 1 data obtained in advanced solid tumors, a randomized phase 2 study of second-line veliparib plus modified FOLFIRI vs FOLFIRI alone is ongoing in pancreatic cancer (NCT02890355).
Only participants without previous irinotecan exposure during firstline therapy were included. An interim analysis of 108 patients did not suggest superior PFS with veliparib at 35% data maturity. 76  In clinical studies completed to date, the addition of PARPi to regimens including irinotecan or oxaliplatin has been characterized in pretreated patients. However, many patients with pancreatic cancer are being exposed to both agents early in the disease course as part of either adjuvant or first-line therapy. In an ongoing phase 1/2 trial, T A B L E 2 (Continued) The primary endpoint for both trials will be the DCR. Hopefully, combining PARPi therapy with other modalities like chemotherapy, immunotherapy, and small molecules will lead to further advances in therapeutic results.

| Combination with other targeted therapies
There is little doubt that progress in these research areas will be made in the near future-a promising new perspective for patients with this devastating disease.

ACKNOWLEDGEMENT
Open Access funding enabled and organized by ProjektDEAL.

DATA AVAILABILITY STATEMENT
n.a.