Homologous recombination repair deficiency (HRD) testing in newly diagnosed advanced-stage epithelial ovarian cancer: A Belgian expert opinion

Ovarian cancer (OC) has a poor prognosis as most patients present with non-specific symptoms and the disease is mostly diagnosed at advanced stages. Approximately 90% of cases are classified as epithelial OC (EOC), a category comprising histologically and molecularly distinct tumours. Identifying reliable biomarkers and employing personalised therapies in OC subgroups is crucial for battling the disease. EOCs are often characterised by homologous recombination repair deficiency (HRD), frequently caused by inactivation of the breast cancer susceptibility (BRCA) genes. These findings have led to the development of poly- (adenosine diphosphate [ADP])- ribose polymerase inhibitors (PARPi), which are synthetically lethal to HRD tumour cells. Both patients with HRD and non-HRD tumours can benefit from PARPi therapy in the recurrent setting. Moreover, recent phase III trials in patients with newly diagnosed advanced-stage OC have demonstrated greater clinical benefit from PARPi in treating HRD than non-HRD tumours. These findings offer new opportunities for the use of PARPi as maintenance therapy after first-line chemotherapy based on the presence of HRD. In the current article, we provide recommendations for HRD testing and treatment of patients with newly diagnosed advanced-stage EOC.


Introduction
Ovarian cancer (OC) is among the most lethal gynaecological cancers in the United States and Europe. Because symptoms are non-specific, OC diagnosis is usually delayed at the more advanced stages (International Federation of Obstetrics and Gynaecology [FIGO] stage III and IV), (Colombo et al., 2019;Vergote et al., 2020). OC comprises a heterogeneous group of diseases with various histological subtypes, differentiation grades, and molecular characteristics (Lheureux et al., 2019). Approximately 90% of OCs are of epithelial origin (EOC), the most common type of which is high grade serous carcinoma (HGSOC) (Vergote et al., 2020).
EOC is genetically heterogeneous (Lheureux et al., 2019). Chromosomal instability and inactivation of tumour suppressor genes are common in HGSOC (also referred to as tubo-ovarian HGSOC) (The Cancer Genome Atlas Research Network, 2011), HOMOLOGOUS RECOMBINATION REPAIR DEFICIENCY (HRD) -VERGOTE et al.

Homologous recombination DNA repair deficiency (HRD)
The HRR pathway is a low-error mechanism to repair double stranded DNA breaks (DSBs) caused by endogenous (e.g., DNA replication defects) or exogenous (e.g., chemotherapeutic agents) factors (Fuh et al., 2020). Cell-culture-based experiments indicate that dozens of DSBs are likely to occur in human cells daily (Fuh et al., 2020). The HRR pathway is active during the synthesis and gap 2 cell cycle phases, when an intact homologous chromosome is available as a repair template. Following DSB detection, several proteins are recruited to the damaged DNA site to correctly repair the break (Figure 1). Among the best known and characterised proteins are breast cancer susceptibility type 1 and 2 (BRCA1 and BRCA2), ataxia-telangiectasia mutated (ATM), the MRN complex (consisting of meiotic recombination 11 [MRE11], RecA-like protein 50 [RAD50] and Nijmegen breakage syndrome protein 1 [NBS1]), and RAD51 (Fuh et al., 2020). Of particular interest in EOC is also poly-(adenosine diphosphate [ADP])-ribose polymerase (PARP), involved in single strand DNA break repair (Ray Chaudhuri and Nussenzweig, 2017).
The consequences of HRD are varied (Fuh et al., 2020), as illustrated in Figure 1B and 1C. If HRR is impaired, the damaged DNA is repaired by more error-prone mechanisms. This leads to genomic instability reflected in genetic alterations of variable sizes, including a specific set of single nucleotide polymorphisms (SNPs), insertions and deletions (collectively termed indels, up to 1 kilobase [kb] in size) flanked by short tandem repeats, overlapping microhomologies (short, identical sequence stretches at DNA breakpoints) (Nik-Zainal et al., 2012), and copy number variations (CNVs; indels larger than 1 Kb that alter gene expression) (Haunschild and Tewari, 2021). Larger rearrangements also occur and include loss of heterozygosity (LOH), telomeric allelic imbalance (TAI) and large-scale state transitions (LST), jointly referred to as "genomic scars" (Figure 1). LOH occurs if an entire allele is lost due to faulty DNA repair (e.g., through a larger indel), while TAI and LST involve loss of larger chromosomal regions (≥10─15 mega base [Mb]) (Watkins et al., 2014;Haunschild and Tewari, 2021). Genomic instability and scarring patterns are specific for the defective DNA repair pathway.

Molecular tests of HRD positivity
HRD testing should ideally be performed to establish the genetic profile of the tumour, estimate patient prognosis, and guide appropriate therapy (Haunschild and Tewari, 2021). Germline mutations of HRR-related genes, including BRCA1 and BRCA2, are established hereditary risk factors for developing multiple malignancies. Germline and tumour BRCA1/2 status are highly concordant (i.e., most tBRCAmut alterations are germline) (Vergote et al., 2020), although BRCA1/2 mutations detected in 5%─7% patients with HGSOC are identified in the tumour, but not through germline testing (Callens et al., 2021). Moreover, BRCA1/2   (Konstantinopoulos et al., 2015). (Ceccaldi et al., 2016). HRD scores based on functional RAD51 foci assays have been reported to correlate with tumour sensitivity to chemotherapy, PARPi response, and overall survival (Fuh et al., 2020), but have not yet been reported in major prospective randomised phase III studies. Limitations of the RAD51 foci assays are the high technicality, lack of automation for foci counting, and the complex HRD score calculation. Two recent review articles provide a detailed overview of additional but clinically still nonvalidated methods, which include gene expression profiling, promoter methylation, and replication fork stalling assays (Fuh et al., 2020;Haunschild and Tewari, 2021).

Genomic HRD tests available in Belgium
Several tests were recently recommended by the Personalised Medicine Commission (ComPerMed) (Belgian Cancer Registry, 2018) and are summarised in Table I. The commercially available assays (from Myriad Genetics and Foundation Medicine) are available worldwide but are not yet reimbursed in the Belgian health system and were so far conducted only in the context of clinical trials.

Clinical utility of HRD testing
Current HRD tests measure a genotype indicative of HRD. A correlation between HRD test scores and PARPi treatment benefit is a common criterion to evaluate if a particular test score is clinically meaningful (Miller et al., 2020;Haunschild and Tewari, 2021). Until recently, most established data on the use of BRCA status for PARPi treatment decisions came from studies in relapsed OC setting. In platinumsensitive relapsed OC, PARPi treatment is active as maintenance monotherapy also in patients with tBRCAwt tumours, although with lower benefit than in tBRCAmut carriers (Mirza et al., 2016;Coleman et al., 2017). Two phase II studies in relapsed setting (ARIEL2 and QUADRA) found that tumour HRD status, determined by either the genomic LOH score (FoundationOne ® CDx) or combined GIS (Myriad myChoice ® CDx), was a predictor of OS with PARPi treatment (Swisher et al., 2017;Moore et al., 2019).
Furthermore, more recent clinical trials provide evidence that HRD genomic profiling is an important determinant of PARPi therapy response in newly diagnosed advanced stage compared to relapsed OC. These trials also highlight the need for reliable and standardised profiling of advanced-stage OC patients, to identify those who may significantly benefit from targeted PARPi therapy. The four published randomised phase III clinical trials investigating PARPi treatment in newly diagnosed mutations are known predictors of OC response to PARPi (Miller et al., 2020), while the evidence for other HRR-related gene mutations in conferring PARPi sensitivity is still controversial, especially in the first-line setting. Additionally, it is essential to identify good predictors of the OC biological status as HRD or non-HRD (Haunschild and Tewari, 2021). Various genomic and biochemical tests have been developed to check for HRD positivity in human tissue ( Figure 1D).

Testing approaches
Genomic HRD tests detect germline and tumour mutations of HRR-related genes, as well as genomic scars indicative of HRD (Fuh et al., 2020;Miller et al., 2020). Specific tests are needed to provide a readout of the genomic scars, such as genome wide LOH (e.g., FoundationOne ® CDx) or a multicomponent genomic instability score (GIS) (e.g., Myriad myChoice ® CDx). According to genomic tests, tumour samples are classified as HRD if they bear a BRCAmut and/or are positive for genomic scars (i.e., genomic instability score above a pre-defined threshold). The gain from sequencing readouts is thus two-fold as both the genetic causes and consequences of HRD may be identified. For clinical purposes, several next generation sequencing (NGS) tests have been validated. The NGS method enables parallel sequencing of thousands of predefined genomic loci with high sensitivity and accuracy (Haunschild and Tewari, 2021).
Non-neoplastic tissue (e.g., blood, saliva) is used for germline tests, while tumour tests are conducted on freshly frozen or formalin-fixed and paraffinembedded (FFPE) tumour tissue. Tumour-based assays are technically more challenging compared to germline testing due to histological (e.g., low tumour cell content) and clonal heterogeneity of tumour tissue samples. Major factors for the success of sequencing approaches are sampling of sufficient high-quality material, and optimal tumour cell content (usually at least 30%) (Miller et al., 2020). Also, OC tumours are known to genetically change over time, in terms of tBRCAmut status and possible BRCA1/2 reversion mutations, but their larger genomic scar patterns detected by genomic instability testing remain quite stable over time (Patel et al., 2018).
Aside from the sequencing based HRD profiling tests, different molecular, biochemical, and cytological assays have been described to further characterise tumour HRD status. One of these is the RAD51 foci assay, which provides a functional readout of HRD. RAD51 normally accumulates at DSBs during functional HRR, so the impaired formation of RAD51 foci is the reflection of HRD   (Table II). A common outcome of these trials was the statistically significant increase in median PFS of patients with an advanced-stage EOC and a confirmed HRD status (presence of BRCAmut and/or genomic scars) after PARPi treatment.
Additional analyses on patient tumour samples from the PAOLA-1-ENGOT-ov25 trial using the Myriad myChoice ® CDx assay showed that mutations in non-BRCA HRR-related genes were present in 3.7%-9.8% of patients (Pujade-Lauraine et al., 2021a; Pujade-Lauraine et al., 2021b). Unlike the HRD status, none of the tested non-BRCA gene panels were predictive of prolonged PFS in patients treated with olaparib and bevacizumab (BEV) (Pujade-Lauraine et al., 2021a). Mutations in some of these genes (e.g., BRIP1, RAD51C, RAD51D, PALB2) can result in HRD and are known to increase the risk of developing OC . However, despite being useful for preventive familial screening and clinical research, they cannot be used as predictive markers of response to PARPicontaining therapies (Pujade-Lauraine et al., 2021a; Vergote et al., 2022).
Importantly, the PRIMA-ENGOT-ov26, PAOLA-1-ENGOT-ov25, and VELIA trials confirmed that gBRCA, tBRCA, and HRD status (as determined by the multicomponent GIS) are relevant determinants of PARPi response. Of note, GIS-based HRD scoring may need to be adapted due to overlapping hazard ratios in subgroup analyses. Alternatively, new HRD tests with adapted cut-offs, such as the recently developed assay of KU Leuven and collaborators, may relate a patient's HRD status with predicted response to targeted therapy (Loverix et al., 2022). All these data are in favour of reimbursing BRCA1/2 and genomic HRD testing to identify patients who can benefit most from the PARPi therapy.
Aside from PARPi, it is important to consider inclusion of BEV in the treatment algorithm of OC. In clinical trials, BEV has demonstrated benefits in terms of PFS when administered in first line with standard chemotherapy or in relapsed disease regardless of platinum sensitivity (Vergote et al., 2020). BEV-associated improvements in OS are currently restricted to the ICON7 and GOG-218 trial findings in high-risk and advanced-stage patients, respectively (Perren et al., 2011;Norquist et al., 2018). There are still no reliable predictive biomarkers to inform decisions on which patients should or should not receive BEV (Colombo et al., 2019).

ESMO and ESGO guidelines on HRD testing
Supported by recent data in patients with newly diagnosed advanced-stage EOC (Coleman et al., 2019;González-Martín et al., 2019;Moore et al., 2019;Ray-Coquard et al., 2019), the European Society for Medical Oncology (ESMO) and European Society of Gynaecological Oncology (ESGO) concede that BRCA1/2 status is a good predictor of patients' response to PARPi and the extent to which a patient can benefit from such a treatment (Colombo et al., 2019;Miller et al., 2020). Furthermore, it was advised that all patients with HGSOC should be tested for mutations in BRCA1/2, and possibly other HRR-related genes (e.g., RAD51C, RAD51D, BRIP1, and PALB2). In the setting of first-line maintenance treatment, ESMO in 2020 recommended that gBRCAmut and tBRCAmut, as well as genomic scars should be routinely tested to identify HGSOC patients who may benefit from PARPi treatment (Miller et al., 2020).
Belgian consensus on HRD testing and treatment of newly diagnosed advanced-stage EOC Given its relevance to the therapeutic response, we hereby advocate for early and reliable HRD testing in EOC patients, to ensure optimal and timely treatment decisions. We focus on clinical and practical recommendations for Belgian physicians and provide a consensus expert opinion on the decision-making process for newly diagnosed stage III─IV EOC (Figure 2). The presented recommendations are based on the available clinical research evidence and outline an ideal-case scenario for a maximised benefit of HRD testing and proposed therapies, acknowledging that not all testing and treatment options may be accessible for each patient. The clinical experience and the presented consensus opinion are similar to European guidelines and expert panel outcomes published in the last years (Colombo et al., 2019;Miller et al., 2020;Vergote et al., 2022).
As per recent Belgian regulatory decisions, PARPi reimbursement criteria were not extended to accommodate HRD testing outcomes ("Substantiated final proposal from the Medicines Allowance 2021/155-0154124504 Commission for an application to amend Lynparza's reimbursement modalities"). The implementation of below recommendations will therefore depend on the patient's individual status and diagnosis and future available treatment options.  PARPi (olaparib, rucaparib, niraparib) and BEV are approved by the European Medicines Agency (EMA) as targeted therapies for OC (Vergote et al., 2020). Details on reimbursement policies of these medicines for OC treatment in Belgium have been extensively summarised elsewhere (Belgian Cancer Registry, 2018;Vergote et al., 2020). EMA approved the combination of olaparib and BEV for use in patients with a confirmed HRD status, defined by BRCAmut and/or GIS (Myriad myChoice ® CDx or another validated GIS) on FFPE tumour tissues (Belgian Cancer Registry, 2018;Vergote et al., 2020). Olaparib is reimbursed as a maintenance monotherapy in BRCAmut patients with newly diagnosed advanced-stage carcinoma, who partially or completely responded to platinum-on multiple steps: pathology diagnosis, referral waiting times for genetic counselling, shipment duration, and lab-specific test turnaround times (Haunschild and Tewari, 2021). The entire duration of the process can thus take two months on average (ORPHA, 2021). As for maintenance treatment, PARPi are recommended for all except BRCAwt non-HRD patients who had received BEV during first-line chemotherapy (Figure 2). Observation only is an option for BRCAwt non-HRD patients who could not receive BEV in the first-line setting.  (Vergote et al., 2020).
Information on different aspects of HRD, including HRR gene mutations, GIS, LOH score, is complementary as it points to different patient subsets (Belgian Cancer Registry, 2018). NGS testing of non-BRCA1/2 mutations is not reimbursed in Belgium but can be accessed through certain trials. As discussed before (Vergote et al., 2020), it would be practical to implement "reflex" (guaranteed) BRCA1/2 genetic testing for all EOC patients as a standard pathology procedure.
Ideally, both gBRCA and tBRCA should be simultaneously tested at diagnosis. Therefore, NGSbased testing for tBRCA genotype and genomic instability scores should be conducted routinely for all advanced-stage EOC patients as soon as possible following diagnosis. This approach is also beneficial in view of current reimbursement policies, whereby simultaneous BEV and PARPi use is not reimbursed. Preferably, the genomic HRD test results should be known by the second cycle of chemotherapy (after 3 weeks of treatment) for patients undergoing primary debulking surgery (before the eventual start of BEV) or latest by the end of the sixth chemotherapy cycle (after 18 weeks of treatment) for all patients.
Collecting sample of sufficient quantity and quality and subjecting it to timely testing is key. Pathologists can prepare multiple samples following debulking surgery or tissue biopsy. It is important to consider the variable time between sample collection and complete test results, which depends * a genetic testing results should be known before the 2nd cycle of chemotherapy for patients who underwent primary debulking and who are candidates for bevacizumab treatment; b reimbursed only for stage IV carcinoma in Belgium (situation in November 2021); cPARPi reimbursement criteria were not extended to accommodate HRD testing outcomes (Belgium, November 2021). BEV, bevacizumab; BRCA, breast cancer susceptibility gene; BRCAmut, genotype without a functional BRCA1 and/or BRCA2 gene;BRCAwt, genotype; HRD, homologous recombination DNA repair deficiency; IV Q3W, intravenous administration every three weeks; NACT, neo-adjuvant chemotherapy; OC, ovarian cancer; PARPi, poly-(adenosine diphosphate [ADP])-ribose polymerase inhibitor. The referenced clinical trials are SOLO-1 (Moore et al., 2018), PRIMA (González-Martín et al., 2019), PAOLA-1 (Ray-Coquard et al., 2019), GOG 218 (Burger et al., 2011, GOG 262 (Chan et al., 2016), and ICON7 (Perren et al., 2011). AstraZeneca, Eli Lilly, Novartis, Amgen, Tesaro; grants from Research Funding institutional Roche; and other fees (travel, accommodations, expenses) by Pfizer, Roche, PharmaMar, Teva, AstraZeneca; all of these outside the submitted work. Dr Gennigens reports personal fees for advisory boards from AstraZeneca, BMS, GSK GlaxoSmithKline Pharmaceuticals, Pfizer, MSD, and Ipsen (including travel); all of these outside the submitted work.
Dr Vergote reports personal fees for consulting (paid to his university) from Deciphera Pharmaceuticals, Verastem Oncology, MSD Belgium, Roche NV, Genmab A/S -Genmab B.V. -Genmab US, F. Hoffman-La Roche Ltd, Pharmamar-DoctaforumServicios SL, Millennium Pharmaceuticals, Clovis Oncology, AstraZeneca NV, AstraZeneca UK Ltd, AstraZeneca Belux, Tesaro Inc.,Tesaro Bio GmbH, Oncoinvent AS, Immunogen Inc, Sotio a.s., Amgen Europe, Carrick Therapeutics, Debiopharm International SA, GSK GlaxoSmithKline Pharmaceuticals, Medical University of Vienna, and Octimet Oncology NV. He also reports grants from Amgen and Roche; other fees (contracted research) from Oncoinvent AS and Genmab; and fees to cover accommodation and travel expenses from Takeda Oncology, Pharma Mar, Genmab, Roche, AstraZeneca, Tesaro, Clovis, and Immunogen. All the Potential Conflicts of Interest reported by Dr Vergote are outside the submitted work. Dr Van de Vijver and Dr Kerger have nothing to declare. Dr De Greve reports personal fees as a member of the AstraZeneca Foundation board and institutional clinical trial grants for IIS trials from Roche and AstraZeneca.
Dr Altintas reports personal fees for advisory boards from Eli Lilly, Pfizer, MSD, Astra Zeneca, Novartis, Bayer; grants from Roche for her breast cancer research; and other fees (travel, accommodations) by Pfizer, Roche, PharmaMar, Novartis, AstraZeneca, Bayer.