Homologous recombination inquiry through ovarian malignancy investigations: JGOG3025 Study

Abstract The Cancer Genome Atlas (TCGA) network has clarified that ~50% of high‐grade serous ovarian cancers show homologous recombination deficiency (HRD). However, the frequency of HRD in Japanese patients with ovarian cancer remains unclear. We aimed to identify the frequency of HR‐associated gene mutations in Japanese patients with ovarian cancer. The JGOG3025 study is a multicenter collaborative prospective observational study involving 65 study sites throughout Japan. We recruited 996 patients who were clinically diagnosed with ovarian cancer before surgery from March 2017 to March 2019, and 701 patients were eligible according to the criteria. We used frozen tumor tissues to extract DNA and performed next‐generation sequencing for 51 targeted genes (including 29 HR‐associated genes) in 701 ovarian cancers (298 high‐grade serous cases, 189 clear cell cases, 135 endometrioid cases, 12 mucinous cases, 3 low‐grade serous cases, and 64 others). HRD was defined as positive when at least one HR‐associated gene was mutated. The frequencies of HRD and tumor BRCA1/2 mutations were 45.2% (317/701) and 18.5% (130/701), respectively, in the full analysis set. Next, we performed multivariate Cox proportional hazards regression analysis for progression‐free survival (PFS) and overall survival (OS). Advanced‐stage ovarian cancer patients with HRD had adjusted hazard ratios of 0.72 (95% CI, 0.55–0.94) and 0.57 (95% CI, 0.38–0.86) for PFS and OS, respectively, compared with those without HRD (p = 0.016 and 0.007). Our study demonstrated that mutations in HR‐associated genes were associated with prognosis. Further studies are needed to investigate the prognostic impact of each HR‐associated gene in ovarian cancer.


| INTRODUC TI ON
Homologous recombination deficiency has attracted attention as a new molecular biomarker in the gynecologic oncology field. 1 Homologous recombination repair is one of the repair mechanisms for DNA double-stranded breaks, in which the ends of doublestranded breaks are precisely repaired. 2 In humans, several genes, including BRCA1/2 and RAD51, are involved in HR repair, and genomic alterations in these HR-associated genes result in HRD. In HRD cells, DNA repair mechanisms other than HR repair are important, especially PARP, which is a new therapeutic target. PARP inhibition in HRD cells leads to genomic instability and cell death (synthetic lethality). 3 At present, two PARP inhibitors (olaparib and niraparib) have become clinically available for ovarian cancer in Japan. Interestingly, two phase III trials have demonstrated that niraparib significantly prolonged PFS in not only patients with BRCA1/2 mutations, but also those with HRD. 4,5 The PAOLA-1 regimen (olaparib + bevacizumab) also provides a significant PFS benefit in patients with HRD. 6 Therefore, HRD status is an important factor in determining the treatment of ovarian cancer.
Analysis based on TCGA, a national cancer genome project from the USA, revealed that approximately half of the high-grade serous ovarian cancers have genomic or epigenomic alterations in the HR-associated pathway, such as BRCA1/2 mutations, BRCA1 methylation, CDK12 mutations, and RAD51C promoter methylation. 7,8 However, the frequency of HR-associated mutations in ovarian cancer in Japan remains unclear, especially in non-serous histologic types. Because there are differences in the distribution of histologic types in ovarian cancer between the USA and Japan, the results of TCGA cannot be directly applied to Japan. In addition, the clinical significance of HRD remains unclear compared with that of germline BRCA1/2 mutations. Investigating the frequency and clinical significance of HRD in Japanese patients with ovarian cancer is an urgent issue.
Therefore, in 2017, we started the Japanese Gynecologic Oncology Group (JGOG) 3025 trial, which is a multicenter collaborative prospective observational study, to clarify the frequency of HRD in Japanese patients with ovarian cancer. We defined the presence of at least one HR-associated gene mutation as HRD in this study. We closed the enrollment of the participants at the end of March 2019 and fixed the data after 30 months of follow-up. We report the results of the final analysis of the JGOG3025 trial.

| Patients and study design
A multicenter collaborative prospective observational study (Clini calTr ials.gov, NCT03159572; UMIN 000026303) was conducted in Japan between January 2017 and September 2021. We enrolled participants for 27 months between January 2017 and March 2019, and the follow-up period was 30 months after the end of enrollment. The target sample size was 700. The study schema is shown in Figure S1.
Patients clinically diagnosed with ovarian cancer were enrolled from 65 JGOG institutions throughout Japan (Table S1). After written informed consent was obtained for the HRD test and biobanking, we collected tumor tissue samples during surgery. Tumor tissue samples were sent to the JGOG ToMMo biobank under controlled temperature. We performed an HRD test after the end of enrollment. When patients had information on germline BRCA1/2 mutations and allowed us to use the information in this study, germline BRCA1/2 mutation status was also submitted to the data center.
Treatment selection was performed according to the Japan Society 51 targeted genes (including 29 HR-associated genes) in 701 ovarian cancers (298   high-grade serous cases, 189 clear cell cases, 135 endometrioid cases, 12 mucinous cases, 3 low-grade serous cases, and 64 others). HRD was defined as positive when at least one HR-associated gene was mutated. The frequencies of HRD and tumor BRCA1/2 mutations were 45.2% (317/701) and 18.5% (130/701), respectively, in the full analysis set. Next, we performed multivariate Cox proportional hazards regression analysis for progression-free survival (PFS) and overall survival (OS). Advancedstage ovarian cancer patients with HRD had adjusted hazard ratios of 0.72 (95% CI, 0.55-0.94) and 0.57 (95% CI, 0.38-0.86) for PFS and OS, respectively, compared with those without HRD (p = 0.016 and 0.007). Our study demonstrated that mutations in HR-associated genes were associated with prognosis. Further studies are needed to investigate the prognostic impact of each HR-associated gene in ovarian cancer.

K E Y W O R D S
HITOMI study, homologous recombination deficiency, JGOG3025, ovarian cancer, targeted sequencing panel of Gynecologic Oncology guidelines for the treatment of ovarian cancer, fallopian tube cancer, and primary peritoneal cancer. 9,10 All evaluations were scheduled according to the clinical practice standards of each institution.
We collected data including patient characteristics, clinicopathological data, clinical outcomes, and adverse events. Clinical outcomes were updated prospectively every 6 months in the electronic data capture system.
The study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects. Approval from the institutional review board of each participating JGOG institution was obtained prior to the initiation of the study. All patients provided written informed consent.

| Study population
The inclusion criteria were as follows: patients who could approve informed consent and sign the form before surgery; patients who are clinically diagnosed with ovarian cancer; patients who are 20 years old and above at enrollment; patients with ECOG PS 0-2; and patients who could provide tumor tissue specimens. Ascites cytology and cell block specimens were not considered tumor tissue specimens. For patients to be treated with NAC, a tumor biopsy should be performed prior to NAC to make a pathologic diagnosis and obtain a tumor tissue specimen.
Patients were excluded based on the following criteria: patients with other clinically active malignancies (except breast cancer) that were treated within 5 years (excluding basal cell carcinoma and squamous cell carcinoma of the skin and carcinoma in situ or intramucosal carcinoma that are curable with local treatment); the principal investigator judged that enrollment of the patient in the study was inappropriate.

| Study outcomes
The primary outcome of the JGOG3025 trial was to identify the HRD frequency in Japanese patients with ovarian cancer, including fallopian tube cancer and primary peritoneal cancer. The secondary outcomes were associations between PFS/OS and HRD in ovarian cancer and between PFS/OS and germline BRCA1/2 mutation in ovarian cancer. PFS was defined as the time from registration in the trial to disease progression or death from any cause.

| Definition of HRD
We performed next-generation sequencing for 51 targeted genes using DNA extracted from cancer tissues. A list of the genes is shown in Table S2. Five genes (ARID1A, KRAS, PIK3CA, PTEN, and TP53) were included as molecular diagnostic markers in ovarian cancer, and 29 genes (ATM, ATR, ATRX, BAP1, BARD1, BLM, BRCA1, BRCA2,   BRIP1, CHEK1, CHEK2, EMSY, FANCA, FANCC, FANCL, MLE11, NBN,   PALB2, POLD1, RAD21, RAD50, RAD51, RAD51B, RAD51C, RAD51D, RAD52, RAD54L, RECQL4, XRCC2) were categorized as HR-associated genes. We used a QIAseq Targeted DNA Custom Panel (CDHS-16649Z-2451, Qiagen) and conducted 150-bp paired-end sequencing on an Illumina NextSeq sequencer. Single nucleotide variants or short InDels were detected using smCounter2 (Qiagen). 11 Variant annotation was performed using ANOVAR version 20190408. 12 To make a mutation list, we generated a mutant allele frequency (MAF) file using maftools 13  For HRD, we defined cases harboring at least one mutation of HR-associated genes as HRD and cases without any HR-associated gene mutation as non-HRD. Because MSI-high or POLE-mutated ovarian cancers showing a very high tumor mutational burden led to passenger mutations of HR-associated genes, cases with both HRassociated gene mutations and MSI-high or POLE mutations were determined to be HRD negative.

| Central pathological review
To confirm the histological diagnosis, the central pathological review was performed by three pathologists (Professor Yuko Sasajima, Professor Miki Kushima, and Dr. Masaharu Fukunaga) assigned by the JGOG using H&E-stained slide specimens from tumor tissues.
The central histological evaluation and diagnosis of ovarian cancer were based on the WHO classification of tumors of female reproductive organs.

| Statistical analysis
The analysis population was the full analysis set of all ovarian cancer patients for whom sample tumor tissue was submitted and who met study enrollment criteria, based on the intention-to-treat (ITT) principle. The frequency of HRD and the primary outcome were calculated as a percentage and 95% CI. The cumulative survival rates for PFS and OS were estimated using the Kaplan-Meier method, and the incidence rate and its 95% CI were calculated. Patients who discontinued the study without having an event were analyzed as censored cases at the time of discontinuation. Adjusted hazard ratios for HRD positive/HRD negative and its 95% CIs were calculated using a stratified proportional hazards model with age, stage, histologic type, surgery, NAC, and adjuvant chemotherapy as covariates, stratified using the FIGO classification of stage I/II and stage III/IV. Imputation of missing values was not performed for either the primary or secondary outcomes. Data are expressed as mean (SD) for continuous variables, and frequencies and percentages for discrete variables, unless specifically mentioned. The p-values were two-sided, and p-values less than 0.05 were considered significant.
The data were analyzed using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA).

| Patient characteristics
The study was conducted from June 2015 to November 2019. The enrollment period was initially planned for 2 years but was extended to 2 years and 3 months to reach the sample size needed. The patient disposition is shown in Figure 1. In total, 996 patients who were clinically diagnosed with ovarian cancer were enrolled, but 295 patients were excluded from the full analysis set according to the eligibility criteria. In total, 701 patients were included in the final analysis.
The patient characteristics are shown in Table 1

| Primary outcome
The mutation profiles of HR-associated genes per sample are shown in Figure 2A. After cases with both HR-associated gene mutations and MSI-high or POLE mutations were determined to be HRD negative, the frequency of HRD in Japanese patients with ovarian cancer was 45.2% (317/701) ( Figure 2B). The mutation rates of tumor BRCA1 and BRCA2 were 9.8% and 7.7%, respectively. Seven patients (1.0%) had both BRCA1 and BRCA2 mutations. Conversely, the mutation rate of HR-associated genes except BRCA1/2 was 26.7%.  Table 2. In patients with high-grade serous carcinoma, the frequencies of both HRD and tumor BRCA1/2 mutations were relatively higher than those in other histologic types. Although the frequency of HRD in the clear cell type was also relatively high, the tumor BRCA1/2 mutation rate was lower than that in high-grade serous carcinoma. When we focused on both stage and histologic type, there were no significant differences in PFS and OS between the HRD and non-HRD subgroups of stage III/IV high-grade serous carcinomas ( Figure S3).  Then, we focused on germline BRCA1/2 mutation status in ovarian cancer. We collected germline BRCA1/2 information from 112 participants. Of them, germline BRCA1/2 mutations were positive in 22 participants (19.6%). Nineteen of these were identified in the corresponding tumors in this study cohort. Conversely, five participants with tumor BRCA1/2 mutations did not have germline BRCA1/2 mutations. There was no significant difference between PFS and germline BRCA1/2 mutations in stage I/II ovarian cancer because only three participants harbored germline BRCA1/2 mutations (aHR, 1.14; 95% CI, 0.08-16.7; p = 0.92). By contrast, germline BRCA1/2-mutated stage III/IV ovarian cancers showed a significantly longer PFS than germline BRCA1/2 wild-type stage III/IV ovarian cancers (aHR, 0.28; 95% CI, 0.12-0.65; p = 0.003) ( Figure 6). In addition, germline BRCA1/2 mutation was significantly associated with OS in stage III/IV ovarian cancer (aHR, 0.18; 95% CI, 0.05-0.70; p = 0.013).

| Mutant allele frequency (MAF) of HRassociated genes
To examine the impact of HR-associated gene mutations in ovarian cancer, we measured the MAF per each gene ( Figure S4). Intriguingly, the median of BRCA1 MAF was 0.80, suggesting that most ovarian cancer with BRCA1 mutation occurred in the loss of function of BRCA1. BRCA2, RAD51D, and BAP1 also showed high MAF (>0.5) in a relatively large number of cases. Conversely, the MAF peaks of other HR-associated genes were located at ~0.4-0.5.

| DISCUSS ION
Our multicenter collaborative prospective observational study revealed that the frequency of HRD in Japanese patients with ovarian cancer was 45.2% and that HRD was significantly associated with PFS and OS in advanced-stage ovarian cancer.    In conclusion, HRD was detected in not only high-grade serous histologic type but also other histologic types, and HRD status may be useful in providing an effective treatment option for patients with ovarian cancer.

ACK N OWLED G M ENTS
We wish to thank all members of the JGOG. We appreciate all the women who participated in this study and their families, the staff of the Translational Research Center for Medical Innovation (TRI; the data and statistical analysis center for JGOG 3025), and the participating JGOG member institutions.

FU N D I N G I N FO R M ATI O N
This study was partially funded by AstraZeneca Externally Sponsored Research.

E TH I C S S TATEM ENT
Approval of the research protocol by an institutional review board: This study was approved by the institutional review board of each participating JGOG institution.
Informed Consent: All patients provided written informed consent.