Molecular Mechanisms of PARP Inhibitors in BRCA-related Ovarian Cancer

Abbreviations: PARP: Poly(Adp-Ribose) Polymerase; SSBS: Single Strand Breaks; DSBS: Double Strand Breaks; BER: Base Excision Repair; NER: Nucleic Acid Excision Repair; MMR: Mismatch Repair; HR: Homologous Recombination; NHEJ: Non-Homologous End Joining; DNA-PKCS: DNA-Dependent Protein Kinase; ORR: Objective Response Rate; PFS: Progression-Free Survival; PLD: Pegylated Liposomal Doxorubicin; OS: Overall Survival


Introduction
Ovarian cancer still represents the main cause of death in women with gynaecological cancers, counting in the United States about 22,280 estimated new cancer cases in 2012 and about 15,500 estimated deaths. The prevalence of ovarian cancer among gynaecologic malignancies is rising; unfortunately, most of patients are diagnosed at advanced stages with consequently worse prognosis. Thus, overall survival is the poorest of all gynaecologic malignancies, with a five-year relative survival rate of 44% for all stages [1].
Currently, the standard treatment in advanced disease remains optimal surgical debulking followed by a chemotherapy regimen based on taxane and platinum [2,3]. Despite surgical cytoreduction and chemotherapy, a large proportion of patients are at high risk for recurrent disease and are candidates for a second-line treatment. Recurrent ovarian cancer is currently classified according to sensitivity to platinum-based chemotherapy. Patients with a complete response after a platinum-based treatment who achieve a platinum-free interval more than 6 months before recurrence are classified as having platinum-sensitive disease (partially platinum-sensitive if platinumfree interval is between 6 and 12 months) and should be treated again with platinum-derived combinations. Women who progress during chemotherapy or experience a response of less than 6 months duration should be classified as having chemorefractary or chemoresistant (platinum-resistant) disease, respectively, and should be treated with a non-platinum single agent. However, cancer recurrences show low chemo-sensitivity and poor prognosis, thus new treatment strategies are urgently needed to improve outcomes [4,5].
The various histological subtypes of ovarian cancer are determined by different molecular alterations. Understanding the tumour molecular biology and identifying predictive indicators of outcome and response to therapy are essential steps in selecting the novel treatment strategies. The wide knowledge of molecular mechanisms in ovarian cancer pathogenesis allowed to identify several molecular targets, thus several agents targeted at these molecules are now entering in clinical practice [6]. The family of poly(ADP-ribose) polymerase (PARP) inhibitors represents a widely investigated and promising alternative for the targeted therapy of ovarian malignancies.

Ovarian Cancer Pathogenesis and Hereditary Cancer Syndromes
For a long period of time, ovarian cancer has been defined as one single disorder. Nowadays, we know that ovarian cancer is a heterogeneous disease that includes various biological behaviour from a clinical and molecular point of view. Epithelial ovarian cancer is characterized by four main histotypes that show differentiation resembling normal tissues of genital apparatus. Serous ovarian cancer seems to derive from the cells that line the fallopian tube, endometrioid tumors from endometrium, mucinous tumors from endocervix and clear cell tumors from the vagina epithelium. Even from a molecular point of view, the genetic profile of each histotype is similar to that of the histological counterparts in normal cells [7]. On this basis, Kurman et al. has recently reconsidered the function of the ovarian surface epithelium in tumorigenesis of epithelial ovarian cancer. The author emphasises the role of the fimbriae of fallopian tube in the pathogenesis of serous ovarian carcinomas and foci of endometriosis in endometrioid and clear cell ovarian cancers [8].
Regarding genetic pathogenesis, sporadic ovarian cancer is characterized by marked genetic instability caused by the modulation of several gene expressions. At the present time, a total of 16 tumor suppressor genes, a total of 15 oncogenes and three imprinted tumor suppressor genes have been described (Table 1) [9,10]. Depending on the gene expression profile, two diverse types of ovarian cancer have been described. Type I ovarian cancer includes low-grade and borderline serous cancers, endometrioid, mucinous and clear-cell tumors. The most frequent mutations in type I tumors involve PTEN, PI3K catalytic subunit-α (PIK3CA), KRAS, BRAF and b-catenin (CTNNB1) genes. On the other hand, high-grade serous carcinomas, mixed malignant mesodermal tumors, carcinosarcomas and undifferentiated cancers are included in type II ovarian cancers. Type II tumors express high genomic instability and in up to 80% of patients TP53 is affected by the mutation. Moreover, this type of tumor is characteristic of BRCA1 and BRCA2 mutated patients and mostly arises from the fallopian tubes and the peritoneum [11].
BRCA1 and BRCA2 mutation carriers have an increased lifetime risk of developing breast and ovarian cancer (up to an 85% for breast cancer and up to a 54% for ovarian cancer), and other cancer types as pancreatic and prostate [12][13][14][15]. About 10-15% of all ovarian cancers have been associated to hereditary DNA repair defects, and in about 90% of hereditary cancers the repair defect is caused by a germline mutation in BRCA genes. However, several other DNA repair genes have been linked to hereditary breast and gynaecological cancers, such as TP53, PTEN, BARD1, CHEK2, RAD51 and PALB2 [16][17][18]. At least 16 genes, mostly involved in the DNA repair pathways, have been showed to play a role in hereditary ovarian tumorigenesis [10]. Nevertheless, several hereditary ovarian malignancies are currently associated to unknown mutations and thus they cannot be detected by specific tests.
The identification and management of women at high risk for hereditary ovarian cancer should be carried out in a specialized family cancer center. Family-based care programs provide genetic counseling in order to inform women and their families about primary and secondary cancer prevention. To date, in healthy women carrying a BRCA mutation, surveillance programs for ovarian cancer have not been proven to be effective. Empirical ovarian cancer screenings are based on annual or semi-annual gynecological exams, transvaginal ultrasound, and evaluation of serum CA 125 concentrations. Prophylactic salpingo-oophorectomy (removal of both ovaries and fallopian tubes) is strongly recommended by the age of 35 or 40 years, even before the natural menopause, as primary prevention for ovarian and fallopian tube cancer. Alternatively, women at increased risk should be informed about the opportunity to join prevention clinical trials, such as chemoprevention trials. In particular, oral contraceptives could play an important role as chemopreventive agents for young women carrying a BRCA1 mutation who refused risk-reducing salpingooophorectomy (RRSO). Furthermore, cyclo-oxygenase inhibitors, retinoids, analgesic drugs, peroxisome proliferator-activated receptor gamma ligands and vitamin D are currently under investigation and represent the most promising future chemopreventive agents for cancer prevention [19].

DNA Repair Mechanisms
Due to the high frequency of replication and their genetic profile, tumor cells have high genomic instability with increased probability of DNA mutations. Several DNA repair mechanisms are employed to remove single-strand breaks (SSBs) and double-strand breaks (DSBs) (Figure 1). The single strand break repair is accomplished by base excision repair (BER), nucleic acid excision repair (NER) and mismatch repair (MMR). BER is important for removing damaged bases by a DNA glycosylase and it is involved in the damage induced by radiation and alkylating agents. MMR recognizes and corrects mismatched bases that can result from DNA replication and recombination. NER removes short single-stranded DNA segment around the lesion and repairs mutations resulting from UV light and hydrocarbons.
Poly(ADP-ribose) polymerase (PARP) is a crucial enzyme involved in BER pathway ( Figure 2). PARP has been described for the first time in 1963 and in 1980 his modulation has been proposed to increase the efficacy of alkylator chemotherapy [20,21]. Seventeen structurally similar proteins compose the PARP family. PARP proteins play several roles in different biological pathways, from DNA damage repair to differentiation and cell death. Particularly, research on PARP enzyme as target for cancer treatment has focused on PARP1, the best characterized protein of the family. Consequently to SSBs, PARP1 detect DNA strand interruptions and promote the synthesis of poly(ADP-ribose) (PAR) using NAD + as a substrate. Poly (ADP-ribosylation) of histones and their release from DNA permit chromatin relaxation to facilitate the access of more repair components ( Figure 3). PARP1 account for more than 90% of ADP-ribosylation in cells, while PARP2 is only responsible for 15% of the cell's PAR production and its precise functions remains to be explained [22,23]. Furthermore, some PARP1 polymorphisms have been associated with increased risk of developing solid tumours, such as germ cell tumour, breast cancer, bladder cancer, lung cancer, gastric cancer and prostate cancer. Particularly, previous in vitro and in vivo clinical trials highlighted that Val762Ala in the catalytic domain might influence clinical outcome in ovarian cancer [24].
Several exogenous agents, such as alkylating drugs or ionizing radiations, and endogenous processes, for example resulting from an error in SSB repair, may produce double-strand breaks. DSBs are corrected by the homologous recombination (HR) and nonhomologous end joining (NHEJ). Homologous recombination provides accurate recombination using a sister chromatid as a template, maintaining genomic stability. However, due to the need for a sister chromatid, HR is limited to the S-phase and G2-phase of cell cycle. Several proteins are largely involved in the HR pathway, such as BRCA 1/2, ATM, CHEK2, RAD51 and Fanconi's anemia proteins ( Figure 4) [25]. The BRCA 1 and BRCA 2 proteins play crucial roles in promoting the repair by HR. Particularly, DSBs activates the kinases ATM, ATR and CHK2, which in turn phosphorylate BRCA1 on several different residues modulating its function. The role of BRCA1 consist in DNA repair and in cell cycle regulation, causing G1-S, S or G2-M phase arrest depending on the residues phosphorylated. BRCA1 forms a complex with BARD1, a protein with structural similarity, important for BRCA1 stability. Lately, the BRCA1-BARD1 complex Although less accurate, NHEJ plays a crucial role in minimizing DNA damage in both G0 and G1 phases of cell cycle, when HR cannot be supplied. Moreover, when a defect occurs in one of the enzymes involved in HR, the DSBs are repaired from error prone mechanisms,    mostly NHEJ, resulting in increased risk of new chromosomal defects and thus the development of cancer [23]. In the first step of NHEJ the heterodimer Ku70/Ku80 breaks the DNA ends and improves the stability of the NHEJ enzymes at the DNA termini. Two Ku70/Ku80 heterodimers recruit DNA-dependent protein kinases (DNA-PKcs) to the DNA ends. The resulting complex of DNA-PKcs and his substrate Artemis has shown an endonuclease activity, thus it processes the DNA termini in order to prepare them for the intervention of XRCC4-Ligase IV. The nuclease functions of Artemis seem to be accomplished by the complex of RAD50, MRE11 and NBS1, which in vitro models interacts also with Ligase IV and Ku homologues ( Figure 5) [29].
Finally, also PARP1 is involved in the two principal mechanisms of DSB repair: HR and NHEJ. Particularly, PARP prevents NHEJ components from binding to site of DNA damage [30].

Synthetic Lethality and PARP Inhibitors Trials
Synthetic lethality occurs when a combination of different events, which singularly are not lethal, causes cell death. Particularly, if BER is impaired, through the inhibition of PARP, single strand breaks, e.g. caused by alkylant agents, can not be correct and become double strand breaks. In patients with HR defects, such as a BRCA mutation carrier, this damage causes the cancer cell death since PARP inhibitors induce aberrant activation of NHEJ ( Figure 6). Thus, tumor cells with defective HR are highly sensitive to blockade of the BER pathway by PARP inhibitors when associated with alkylant agents [31]. In fact, in 2005 two seminal preclinical studies pointed out that BRCA-mutated cell are more sensitive to PARP inhibitors than heterozygous mutant and wild-type cells, highlighting the promising role of PARP inhibition in treatment of BRCA-mutated patients [32,33].
To date, several PARP inhibitors have been investigated and mentioned in literature. These molecules act binding to the donor site of the catalytic domain and causing reversible inhibition of PARP enzyme. In clinical trials, the most widely studied reversible PARP inhibitors are AZD2281 (Olaparib) and ABT-888 (Veliparib). BSI-201 (Iniparib), initially considered as a PARP inhibitor, has still unclear mechanism of activity and it does not seem to inhibit PARP enzymes at the clinically used dose [34].
In 2009 a phase 1 trial of Olaparib in BRCA mutation carriers has been published. The authors enrolled and treated 60 patients with different doses of Olaparib and analyzed the pharmacokinetic and pharmacodynamic characteristics of the agent. The maximum tolerated dose was established at 400 mg twice daily. The dose of 200 mg twice per day showed a favourable tolerability with an objective antitumor activity in BRCA1 or BRCA2 patients [35]. The successive expansion cohort study confirmed these notable results, highlighting that the clinical benefit rate is significantly associated with platinumfree interval and increases through platinum-refractory, resistant and sensitive subgroups (23%, 45%, and 69% respectively). These data suggested that PARP inhibitors anti-tumor activity is effective even in platinum-resistant disease but sensitivity to these agents' decreases with the raising resistance to platinum [36].
In 2010 an international, multicentre, phase 2 study with a cohort sequential design compared the continuous administration of Olaparib at the dose of 400 mg twice a day to Olaparib at 100 mg twice a day, in BRCA1 or BRCA2 mutated patients with recurrent ovarian cancer. Both cohorts of patients showed a significant antitumor efficacy with an objective response rate (ORR) of 33% and median response duration of 9.5 months at the dose of 400 mg twice a day, and an ORR of 13% with median response duration of 8.8 months at the dose of 100 mg twice a day. The tolerability profile and related adverse events were quite similar between the two cohorts of patients with nausea, fatigue  and anaemia (all events mostly grade 1 or 2) in patients who assumed Olaparib at the dose of 400 mg and nausea and fatigue (mostly grade 1 or 2) in the other cohort [37].
The role of BRCA mutation status in patients treated with PARP inhibitors has long been discussed. A randomized double-blind placebo-controlled phase 2 study enrolled patients with platinumsensitive high-grade serous ovarian cancer to investigate the role of Olaparib maintenance therapy. Two hundred and fifty patients with objective complete response to the last platinum-based treatment were randomized to receive Olaparib or placebo until progression. The results highlighted that progression-free survival (PFS) in the Olaparib arm improved significantly compared to placebo [38]. In 2012, the last interim analysis of the study was published and confirmed that Olaparib as maintenance treatment significantly increased PFS (from 4.8 to 8.4 months) among patients with platinum-sensitive, relapsed, high-grade serous ovarian cancer, regardless of the BRCA gene mutation. When the interim analysis was published, there was no evidence of overall survival benefit [39]. These data confirmed the results of a Canadian multicentric study, in which 55 high-grade serous ovarian cancer patients received Olaparib 400 mg twice daily. The study included BRCA carriers and women with unknown BRCA status and concluded that the efficacy of Olaparib is not related to the BRCA genes mutation status [40]. These results suggested that there is a specific phenotype of BRCA negative tumor (BRCAness) with a defect in the HR system, thus with features and behaviour similar to BRCA-related cancers even if BRCA mutation negative.
The successive step in the clinical research has been the study of PARP inhibitors as second-line treatment in BRCA-related ovarian cancers. In 2012 a phase 2 multicenter three-arm study compared two diverse dosage of Olaparib (200 and 400 mg twice per day continuously) to Pegylated liposomal doxorubicin (PLD) 50 mg/ m 2 by IV infusion every 4 weeks, in 97 BRCA1 or BRCA2 mutation carriers affected by partially platinum-sensitive or platinum-resistant ovarian cancer. Median PFS was 6.5 months for the Olaparib 200 mg, 8.8 months for the Olaparib 400 mg and 7.1 months for PLD group. The difference in PFS between the Olaparib and PLD group was not statistically significant. To conclude, the activity of Olaparib in this study showed to be consistent with previous research whereas PLD has proven to be more effective than previously described [41]. Three possible explanations for these negative results have been listed by Konstantinopoulos et al. [42]. First, in the PLD group there was a relatively higher frequency of platinum-sensitive ovarian cancers (57.6%) than in Olaparib groups (46.9% in 400 mg dose and 43.8% in 200 mg dose). Considering the higher efficacy of Olaparib in platinumsensitive disease [36], this unbalanced distribution could have led to an underestimation of Olaparib activity. Furthermore, in 2011 an observational study of multidimensional genomics and clinical data on 316 high-grade serous ovarian cancer patients investigated the relationships between BRCA1/2 mutations and overall survival (OS), progression-free survival (PFS) and chemotherapy response. Interestingly, BRCA2 mutation status in ovarian cancer patients has proven to be an independent predictive factor for OS, while BRCA1 mutation status was not significantly associated with increased survival. No differences in PFS between BRCA1 mutation carriers and wild-type BRCA patients were found, while BRCA2 mutation carriers showed significantly longer PFS than the other two groups. Finally, analyses of chemotherapy response revealed that BRCA2-mutated ovarian cancer were more chemo-sensitive and showed longer platinum-free intervals than BRCA1-mutated and wild-type BRCA diseases [43]. In the study comparing different dosage of Olaparib to PLD, the higher proportion of BRCA1-mutated cases over BRCA2-mutated in each group might be another plausible explanation for the negative results. Finally, the predominance of more heavily treated patients in the Olaparib 400 mg group than PLD one (78.2% vs. 51.5%) could have contributed to the development of subsequent somatic mutations that, restoring BRCA1/2 functions, could have conferred resistance to Olaparib [44]. In the same year, a phase 1 trial evaluated the role of Veliparib in association to metronomic Cyclophosphamide in patients with refractory solid tumors and lymphoid malignancies. Of the 35 patients enrolled, 11 had ovarian cancer and 12 had breast cancer. The maximum dose tolerated was defined as Veliparib 60 mg plus Cyclophosphamide 50 mg once a day. Seven cases, 5 of which were BRCA 2-related ovarian cancers, achieved partial responses; additional 6 patients, one of which was BRCA 2-related ovarian cancer, achieved stable disease for at least six cycles. The study showed promising activity of the combination in particular in the subgroup of BRCA mutation carriers [45]. Currently, the combination compared to Cyclophosphamide monotherapy is under investigation in a phase 2 trial enrolling BRCA-related ovarian cancers, triple-negative breast cancers, and low-grade lymphomas.
Recently, the association between PARP inhibition and antiangiogenic strategies has been analyzed in a phase 1 trial. This study investigated the combination of Cediranib with Olaparib in patients with recurrent ovarian cancer and breast cancer. ORR was achieved in 44% of ovarian cancer cases, and the clinical benefit rate (defined as ORR plus stable disease >24 weeks) was 61%. Conversely, no clinical response was observed in the seven evaluable breast cancer cases. In conclusion, this study showed promising evidence of activity of the combination Cediranib and Olaparib in ovarian cancer patients [46].
In 2011, two single-arms phase 2 trials investigating the combination of Iniparib (BSI-201) with Gemcitabine/Carboplatin in patients with platinum-sensitive and platinum-resistant ovarian cancer has been presented at the ASCO Annual Meeting. In the first trial in platinum-sensitive disease, analysis from the first 17 patients demonstrated an increase in ORR (70.6%) compared with previous data. In the preliminary analysis, no significant association between BRCA mutation status and objective response rate has been observed, and no unexpected toxicities have been reported [47]. On the other hand, in platinum-resistant ovarian cancer, the combination showed promising evidence of response (ORR 31.6%) and median PFS substantially improved (5.9 months) [48]. In 2013, a Phase 1/1b study analyzing the combination of Olaparib and Carboplatin in BRCA1/2-related breast and ovarian cancer has been presented in the poster discussion session of ASCO Annual Meeting. The analysis of results concluded that Olaparib 400 mg twice daily with Carboplatin AUC5 every three weeks is active and tolerable in BRCA mutated patients despite interactive marrow suppression. Moreover, exploratory translational studies indicated FOXO3 and NFkB1 as possible predictive factors for response to therapy, requiring a prospective validation [49].
As previously mentioned, defects in the BER system have particular impact on the repair of the damage induced by alkylating agents and ionizing radiation. On this basis, PARP-inhibitors have been studied in association to alkylating agents as a potential approach to increase cytotoxicity of radiotherapy. Recently, Veliparib has been investigated combined with radiotherapy and temozolomide in glioblastoma, showing clinically significant benefit particularly in MGMT-unmethylated tumors [50]. The role of PARP inhibitors in association with chemotherapy as radiosensitizers has been analysed in several other settings where radiotherapy represents a crucial tool for the control of the disease. For instance, Rucaparib has been studied in BRCA-2-deficient and wild type pancreatic cancer cells [51], Olaparib has been evaluated in nasopharyngeal carcinoma cells [52], in nonsmall cell lung carcinoma [53], and in Ewing Sarcoma [54] while Veliparib has been investigated in colorectal cancer cells [55].  To date, other novel PARP inhibitors have been proposed and are being studied in preclinical and clinical setting. For instance, in preclinical tumor models with defects in BRCA and PTEN function, Niraparib (MK4827) has been shown to inhibit selectively PARP-1 and PARP-2 inducing synthetic lethality. In a phase 1 study that enrolled patients affected by advanced stage solid tumors, Niraparib (maximum tolerated dose of 300 mg/day) showed antitumor activity in eight of 20 patients with BRCA-related ovarian cancer and in two of four patients with BRCA-related breast cancer. Anti-tumor efficacy was also observed in sporadic high-grade serous ovarian cancer, non-small-cell lung cancer, and prostate cancer [56]. Moreover, a recent preclinical study investigated growth inhibitory effects of the PARP inhibitor Rucaparib in a set of 39 ovarian cancer cell lines [57].
In conclusion, Table 2 lists and describes current studies of PARP inhibitor as mono-therapy or combined with different agents in ovarian cancer patients (

Conclusion
In the recent years, the increasing knowledge of molecular mechanisms in ovarian cancer pathogenesis allowed to identify several targeted agents that are now entering in clinical practice. Nowadays, the family of poly(ADP-ribose) polymerase (PARP) inhibitors represents a widely investigated and promising alternative for the targeted therapy of ovarian malignancies. PARP inhibitors exploit the synthetic lethality concept to prevent the DNA damage repair, causing cancer cell death. Several agents have already been identified and studied in phase 1 and 2 trials and others are still under investigations in preclinical and clinical studies. The first published phase 1 and 2 studies analyzed the role of PARP inhibitors as single agent in recurrent ovarian cancer. However, to date available data in literature and ongoing trials [58] are mostly related to the association of PARP inhibitors and chemotherapy. This trend suggests a future prevalent role for PARP inhibitors as combination rather than monotherapy, probably confining the use of PARP inhibitors as single agent for the maintenance therapy.
Despite the enrolment of an adequate number of participants in order to obtain significant statistical power could be a challenge, randomized phase 3 trials are urgently needed to compare PARP inhibitors to standard therapies. The evidence of BRCAness represents a resource to extend the amount of patients who might benefit from PARP inhibitors activity. However, while genetic testing helps to find BRCA mutation carriers, to date we still need tests to allow identifying BRCAness patients, carrying dysfunctions in HR pathway. Future research should be directed to define the cases that may truly benefit from the synthetic lethality approach and thus from PARP inhibition strategy.
Moreover, the resistance mechanisms to PARP inhibitors still represent a crucial issue for the proper development of these promising agents. To date, several mechanism have been described including restoration of BRCA function, up regulation of NHEJ system, induction of P-glycoprotein efflux pump expression and the loss of the protein 53BP1 which avoids HR proceeding in DNA repair [59].
Another important issue must be discussed in order to ensure an effective use of these agents. The long-term effects of inhibition of PARP enzymes in combination with DNA-damaging agents should be evaluated in animal models, in order to define the risk of secondary malignancies.
Finally, further research should focus on the structure, mechanism and function of PARP enzymes other than PARP1. Improved understanding of the function of the other PARP enzymes may lead to better interpret the wider consequences of PARP inhibition and may help to suggest novel treatment approaches, identifying further molecular targets.