PARP inhibitors in ovarian cancer: Sensitivity prediction and resistance mechanisms

2.2. Potentially relevant gene aberrations and combination strategies for PARP inhibitor sensitization

To identify which gene aberrations might induce synthetic lethality with PARP inhibitors, researchers studied the five major DNA damage repair pathways operational in cells.3, 34 The combination of NER deficiency and PARP inhibition has been demonstrated not to be synthetically lethal.26 However, although NER deficiency does not lead to synthetic lethality, the combination of PARP inhibition and deficiencies in other DNA repair pathways such as BER, HR, non‐homologous end joining (NHEJ) and even mismatch repair (MMR) has been demonstrated to be synthetically lethal.35 Since the HR pathway is a high‐fidelity repair pathway for accurately repairing DSBs and involves the BRCA1/2 proteins, researchers have focused on the HR pathway.

Approximately 50% of high‐grade serous ovarian cancers (HGSOCs) are deficient in HR because of germline or somatic mutations in BRCA1/2 (20%), the epigenetic silencing of BRCA1 (11%), the amplification or mutation of EMSY (8%), the deletion of PTEN (7%), the hypermethylation of RAD51C (3%), mutations in ATM or ATR (2%) or mutations in Fanconi anemia genes (5%).36 The key genes and modulators of the HR pathway that might induce synthetic lethality with PARP inhibitors are under investigation. Figure ​Figure11 shows the aberrations in HR genes or potentially relevant HR genes and Figure ​Figure22 presents the molecular process of DNA damage repair associated with PARP inhibitor sensitivity and resistance.

Open in a separate window

Figure 1

According to The Cancer Genome Atlas (TCGA), ovarian cancers can be classified as HR‐proficient and other (34%), possibly HR‐deficient (10%) (including those with alterations in 42 other potentially relevant HR genes, such as BLM, ERCC4, TP53BP1, RPA1 and XRCC3) and HR‐deficient (56%) (including those with alterations of BRCA1/2; the amplification or mutation of EMSY; the deletion of PTEN, Fanconi anemia genes [including FANCA, FANCB, FANCC, FANCD2, FANCE, FANCF, FANCG, FANCI, FANCL, FANCM, PALB2 and C19orf40], core HR RAD genes [including RAD50, RAD51, RAD51C, RAD51L1, RAD51L3, RAD52, RAD54B and RAD54L] or HR‐related DNA damage response genes [including ATM, ATR, CHEK1 and CHEK2])27, 36

Open in a separate window

Figure 2

The molecular process of DNA damage repair associated with PARP inhibitor sensitivity and resistance. 1) During the repair of DSBs by NHEJ, damage is recognized and bound by Ku70‐Ku80 heterodimers, DNA‐PKcs and Artemis are activated, and XRCC4 and DNA ligase‐IV are recruited to complete DNA end joining. NHEJ occurs throughout the cell cycle and directly ligates the ends of a DSB. The loss of the abovementioned crucial proteins in the NHEJ pathway might induce resistance to PARP inhibitors. 2) During the repair of DSBs by HR, DNA lesions are recognized by the MRE11‐NBS1‐RAD50 (MRN) complex, and DNA end resection is initiated. In a PALB2‐dependent fashion, BRCA2 is recruited, which loads RAD51 onto the DNA to mediate strand invasion on the homologous sister chromatid. HR occurs mainly in the S and G2 phases. The switch between HR and NHEJ depends on the activity of S phase CDKs, which phosphorylate CtIP in order to activate the MRN complex and stimulate DNA end resection, which is regulated by 53BP1, REV7 and RIF1. The abovementioned crucial proteins in the HR pathway might thus affect PARP inhibitor sensitivity. 3) In addition, the cell cycle checkpoint pathway and the FA‐BRCA1 pathway are involved in HR, and crucial proteins in these pathways are associated with PARP inhibitor sensitivity27, 46

To identify tumours with a deficiency in HR genes, targeted capture and the massively parallel genomic sequencing of multiple‐gene panels would be required. A 60‐gene panel, referred to as the BRCAness profile, was developed to distinguish BRCA‐like tumours from non‐BRCA‐like tumours. The BRCAness profile accurately predicted platinum sensitivity in 8 of 10 ovarian tumours and accurately identified PARP inhibitor sensitivity and resistance in all four cancer cell lines tested.37 A 30‐gene panel was developed to target all known breast and ovarian cancer genes, including BRCA1 and BRCA2 as well as 11 HR‐related genes.21 A 21‐gene panel was developed to target germline mutations in tumour suppressor genes, including BRCA1, BRCA2, and other genes known to cause inherited breast or ovarian carcinoma.38 These multiple‐gene panels are prospective for identifying tumours that are sensitive to PARP inhibitors.

Germline or somatic BRCA1/2 mutations have been demonstrated to significantly affect tumour sensitivity to PARP inhibitors.39 The epigenetic silencing of BRCA1 represented 15% of sporadic ovarian cancers in a population‐based study. In addition, the epigenetic silencing of BRCA1 was demonstrated to be excluded in hereditary ovarian cancers, and cancers with epigenetic BRCA1 silencing did not exhibit BRCA1 protein expression by immunohistochemistry.40 Furthermore, BRCA1 hypermethylation conferred the same degree of sensitivity to PARP inhibitors as did BRCA1/2 mutations in a breast cancer cell line. In addition, the growth of xenograft tumours with epigenetically silenced BRCA1 was reduced by PARP inhibitor treatment.41, 42 However, in a panel of 39 ovarian cancer cell lines representing all histologic subtypes, as well as in a panel of 21 breast cancer cell lines, methylation of the BRCA1 promoter was not detected.35, 43 Furthermore, according to The Cancer Genome Atlas (TCGA), unlike BRCA1/2 mutations, ovarian cancers with epigenetic silencing of BRCA1 through promoter hypermethylation appear not to respond as favourably to platinum and not to exhibit improved survival. This raises the question whether patients with epigenetic silencing of BRCA1 may respond more favourably to PARP inhibitors.36

The DNA recombinase RAD51 is a crucial downstream protein in the HR repair pathway.44 RAD51 is relocalized within the nucleus in response to DNA damage to form distinct foci, which are thought to represent assemblies of proteins at these sites of HR repair. PARP inhibition leads to the failure of SSB repair via the BER pathway, and persistent SSBs result in stalled replication forks and subsequently develop into DSBs. In HR‐proficient cells, these DSBs can be repaired by error‐free HR accompanied by an increase in RAD51 foci formation, whereas in HR‐deficient cells, repair by error‐prone NHEJ results in genomic instability, chromosomal aberrations, cell cycle arrest and ultimately cell death.45 Hence, RAD51 nuclear foci can be a surrogate biomarker for HR repair functionality and can be detected by a functional RAD51 immunofluorescence assay. The absence of RAD51 foci formation represents a functional deficiency of HR repair.46 The RAD51 foci formation assay successfully predicted the in vitro response to PARP inhibitors in primary cultures from the ascitic fluid of ovarian cancer patients.22, 45 Using RAD51 foci as a biomarker, PARP inhibitor cytotoxicity was observed in 93% of HR‐deficient tumours but in no HR‐proficient tumours.45 Furthermore, irradiation‐induced RAD51 foci formation assay accurately predicted which patient‐derived xenograft (PDX) models were sensitive to PARP inhibition in vitro and in vivo.47 However, in a study by AlHilli et al, the decreased formation of RAD51 foci failed to predict a response to PARP inhibitors in HGSOC PDX models.19

Five RAD51 paralogs exist: RAD51B, RAD51C, RAD51D, XRCC2 and XRCC3. Germline mutations in both RAD51C and RAD51D were identified in families with ovarian cancer. Loss‐of‐function mutations in RAD51D predispose individuals to ovarian cancer but not to breast cancer.44 Germline RAD51D mutations were found in 0.8% of subjects with ovarian cancer not selected for a family history of ovarian cancer.48 When treated with RNA interference targeting RAD51D, tumour cells exhibit a sensitivity to olaparib similar to that achieved by the silencing of BRCA2.44 RAD51C acts sequentially with RAD51 at the DNA damage site to repair DNA damage, and RAD51C depletion leads to impaired RAD51 foci formation, resulting in impaired DNA repair. RAD51C promoter methylation has been studied thoroughly. RAD51C was found to be densely methylated in both cancer cell lines and tumour tissue samples, resulting in low RAD51C expression. RAD51C‐deficient cancer cells were more sensitive to olaparib than were RAD51C‐proficient cancer cells, and the silencing of RAD51C in olaparib‐resistant cell lines could enhance olaparib sensitivity. Olaparib significantly suppressed RAD51C‐deficient tumour growth in a xenograft model.49 In addition, RAD51C promoter methylation was demonstrated to be associated with PARP inhibitor response in a PDX model.19

BCCIP is an important cofactor for BRCA2 in tumour suppression. BRCC is a holoenzyme complex containing BRCA1, BRCA2, RAD51, BRCC36 and BRCC45. EMSY binds BRCA2 within exon 3 of BRCA2 and is capable of silencing the activation potential of this exon. The protein kinase Aurora‐A (AURKA) inhibits RAD51 recruitment to DNA DSBs, decreases DSB repair by HR and sensitizes cancer cells to PARP inhibitors. Hence, the low expression of BCCIP or BRCC3 and the amplification of EMSY or AURKA were demonstrated to be significantly associated with in vitro PARP inhibitor sensitivity in a series of 39 ovarian cancer cell lines.35 Whether PTEN deficiency causes HR deficiency (HRD) in the repair of DNA DSBs remains controversial.50, 51, 52 PTEN deficiency might alter multiple cell cycle checkpoints, possibly decreasing the time available for DNA damage repair.52 PTEN deficiency has been observed to sensitize cancer cells to PARP inhibitors both in vitro and in vivo.50, 51 However, in the aforementioned 39 ovarian cancer cell lines, PTEN mutations were not significantly associated with in vitro sensitivity to PARP inhibitors.35

ATR and its downstream effector CHK1 are two important kinases in the replication checkpoint signaling pathway. The ATR‐CHK1 pathway can be activated by DNA damage and replication stress; the activity of this pathway arrests cell cycle progression and helps cells survive these genotoxic stresses. In addition, this pathway stabilizes replication forks and prevents the collapse of replication forks into DNA DSBs. The inactivation of ATR/CHK1 results in substantial DSBs, the loss of the G2‐M checkpoint, and the premature entry of cells with high‐level DSBs into mitosis, resulting in chromosome aberrations and cell apoptosis. Furthermore, ATR participates in the HR pathway by phosphorylating and regulating BRCA1 and other important proteins.53, 54 The depletion of ATR but not CHK1 using siRNAs or small molecular inhibitors has been demonstrated to sensitize ovarian cancer cells (including HR‐proficient cells) to PARP inhibitors. In addition, ATR inhibition disrupted the function of HR repair and further sensitized cells with HRD to PARP inhibition.53 Combined treatment with a PARP inhibitor and an ATR/CHK1 inhibitor decreased the viability and colony‐forming ability of both BRCA mutant and HR‐proficient cells. The PARP‐ATR inhibitor combination caused the complete regression of BRCA‐mutated ovarian cancer PDX tumours, while the PARP‐CHK1 inhibitor combination led only to tumour suppression rather than to tumour regression. Nevertheless, compared with the PARP inhibitor alone, both of the combinations showed significantly improved efficacy.54 The phase II trial combining an ATR inhibitor (AZD6738) with a PARP inhibitor (olaparib) to treat recurrent ovarian cancers is ongoing (CAPRI: {"type":"clinical-trial","attrs":{"text":"NCT 03462342","term_id":"NCT03462342"}}NCT 03462342).

ATM and CHK2 are also important cell cycle checkpoint proteins.55 ATM phosphorylates CHK2 and promotes CHK2 activation. In turn, CHK2 rapidly hyperphosphorylates BRCA1 at multiple sites, switching error‐prone NHEJ to error‐free HR56, 57 CHK2 also helps BRCA1 control NHEJ and reduces the mutagenic potential of NHEJ.58 ATM deficiency induces both DNA DSB repair defects and cell cycle checkpoint deficiencies. Checkpoint signals resulting from the detection of DNA damage cannot induce cell cycle arrest, thus preventing the opportunity for cellular repair systems to respond and resulting in cell apoptosis. Hence, ATM defects impair HR‐mediated DSB repair.59 Low ATM expression was demonstrated to be significantly associated with PARP inhibitor sensitivity both in vitro and in vivo.35, 60 Consistent with the above findings, a deficiency in proteins integral to HR (RAD51, RAD54, DSS1 and RPA1), Fanconi anemia (FANCA, FANCC and FANCD2) or DNA damage signaling (ATM, ATR, CHK1, CHK2 and NBS1) was demonstrated to be associated with sensitivity to PARP inhibitors.18 These results demonstrated that gene aberrations directly or indirectly affecting the DNA damage response (DDR) or DNA repair determine the sensitivity of the tumour to PARP inhibitors. However, the pathways involved are complex and interacting.

Many studies aim to expand the utility of PARP inhibitors to BRCA‐proficient tumours by combining PARP inhibitors with other surrogates to induce “synthetic lethality” in these tumours. The aforementioned gene aberrations that induce HR pathway deficiency can be exploited using small‐molecule inhibitors or RNA interference. The phosphoinositide 3‐kinase (PI3K) pathway is the most frequently altered pathway in human tumours and regulates a wide range of cancer cell processes, including cell cycle progression, survival, metabolism, motility and genomic instability.61 PI3K inhibitors have shown significant anti‐tumour activities in ovarian cancers both in vitro and in vivo.62 The PI3K inhibitor BKM120 was shown to impair DNA repair by HR and sensitize breast cancer cells to PARP inhibition regardless of the BRCA mutation status.63, 64 Moreover, this same PI3K inhibitor was demonstrated to decrease PI3K/AKT/mTOR signaling activity, impair DNA repair by HR and sensitize ovarian cancer cells to PARP inhibition regardless of the status of the PIK3CA or BRCA genes. In addition, BRCA downregulation was observed to be a potential biomarker for an effective response to the proposed combination treatment.65, 66 Cyclin‐dependent kinase (CDK)1 is a core component of the cell cycle and promotes cell cycle progression. CDK1 phosphorylates BRCA1, which is necessary for BRCA1 to form foci at sites of DNA damage. CDK1 inhibition impairs BRCA1 function, compromises the cellular capacity to repair DNA by HR and sensitizes cells to PARP inhibitors both in vitro and in vivo.67 CDK12 is one of the 9 significantly mutated genes in HGSOC and regulates the transcription of BRCA1 and several other DNA repair genes.36 In HGSOCs, CDK12 mutations impaired CDK12 kinase activity, reduced HR gene (including BRCA1, ATR, FANCI and FANCD2) levels, and decreased RAD51 foci formation, thus disrupting HR repair and predicting sensitivity to PARP inhibitors.68 A genome‐wide synthetic lethality screen for candidate PARP inhibitor sensitivity genes identified CDK12 deficiency as a clinically relevant biomarker for PARP inhibitor sensitivity.69 The phase I study of an oral PI3K inhibitor (BKM120 or BYL719) combined with olaparib in patients with recurrent triple‐negative breast cancer or HGSOC ({"type":"clinical-trial","attrs":{"text":"NCT 01623349","term_id":"NCT01623349"}}NCT 01623349); the phase Ib study of olaparib combined with either an oral mTORC1/2 inhibitor (AZD2014) or an AKT inhibitor (AZD5363) in patients with recurrent endometrial, triple‐negative breast, ovarian, primary peritoneal or fallopian tube cancer ({"type":"clinical-trial","attrs":{"text":"NCT 02208375","term_id":"NCT02208375"}}NCT 02208375); and the phase I study of veliparib combined with a CDK inhibitor (dinaciclib) in patients with advanced solid tumours ({"type":"clinical-trial","attrs":{"text":"NCT 01434316","term_id":"NCT01434316"}}NCT 01434316) are ongoing.

Vascular endothelial growth factor receptor 3 (VEGFR3) inhibition in ovarian cancer cells led to decreased levels of BRCA1/2 and restored chemosensitivity in resistant cell lines with wild‐type BRCA2.70 In a randomized phase II trial ({"type":"clinical-trial","attrs":{"text":"NCT 01116648","term_id":"NCT01116648"}}NCT 01116648), the combination of olaparib with cediranib (a highly potent inhibitor of VEGFR 1‐3) significantly improved the survival of patients with BRCA wild‐type, platinum‐sensitive recurrent high‐grade ovarian cancer to a level equivalent to that of their BRCA mutant counterparts, thus overcoming the requirement for BRCA mutations in mediating PARP inhibitor sensitivity.71 Histone deacetylase (HDAC) enzymes are important for the repair of DNA DSBs by HR. The inhibition of HDAC enzymes led to a significant reduction in the transcription of HR genes, including BRCA1 and RAD51, and sensitized HR‐proficient ovarian cancer cells to PARP inhibition both in vitro and in vivo.72, 73 Furthermore, combining PARP inhibition with metformin enhanced the anti‐tumour effects of PARP inhibitors in both BRCA‐deficient and BRCA‐proficient ovarian cancer cells.74 DNA methylation plays an essential role in regulating normal biological processes as well as carcinogenesis. DNA methyltransferase 1 (DNMT1) belongs to a family of enzymes responsible for maintaining cellular DNA methylation patterns. The combination of a DNMT inhibitor with a PARP inhibitor impaired the BRCA‐mediated DDR and sensitized PARP inhibitor‐resistant ovarian cancer to PARP inhibitor therapy regardless of BRCA status.75

This study was supported by Beijing Natural Science Foundation (no. 7172077), and Beijing Municipal Administration of Hospitals Clinical Medicine Development of special funding ‐Yang Fan Project (ZYLX201713).