FormalPara Key Points

This study provides a detailed description of the mutation profile of 15 genes involved in the HRR pathway in Chinese ovarian cancer patients. These findings are compared with previous research in Asian and European populations.

The study reveals that mutations in the BRCA1/2 genes are associated with the prognosis of EOC patients. However, mutations in other non-BRCA genes within the 15-gene panel of the HRR pathway do not exhibit the same predictive value.

1 Introduction

Epithelial ovarian cancer (EOC) is a heterogeneous and often fatal disease in women worldwide and there is currently no effective general population screening tool for ovarian cancer, which has economic implications. Treatment costs for ovarian cancer are the highest among all cancer types [1]. EOCs can be classified into two types. Type I EOCs are relatively slow growing and genetically stable tumors that often develop from identifiable precursor lesions. On the other hand, type II EOCs are aggressive tumors from the start, with a tendency to metastasize even from small primary lesions. The most common type of EOC, high-grade serous ovarian cancer (HGSOC), follows the type II pathway and often exhibits p53 and BRCA mutations. It also includes ovarian clear cell carcinoma (OCCC) and endometrioid ovarian cancer (EnOC) [2, 3]. Maintenance therapy with poly (adenosine diphosphate-ribose) polymerase inhibitors (PARPis) has revolutionized the treatment strategy of EOC [4,5,6]. PARPis can result in synthetic lethality in homologous recombination repair (HRR)-deficient cancer cells [7]. Combining PARP inhibitors with immunotherapies, such as anti-CTLA-4 and PD-1/PD-L1, is based on the hypothesis that BRCA1/2 mutations and HRR-deficient (HRD) tumors without BRCA1/2 mutations have more neo-antigens, which can result in a stronger anti-tumor immune response. BRCA deficiency is known to activate an innate immune response through STING, leading to the production of type I interferons and pro-inflammatory cytokines. Clinical studies have shown that PARP inhibition can affect GSK3, increase PD-L1 expression, suppress T-cell activation, and promote cancer cell apoptosis [8]. Thus, PARPis are recommended for the maintenance treatment of EOC patients with HRD status [9]. However, it is difficult to identify the HRD status for most EOC patients in China due to a lack of diagnostic certified companies and technology.

It is more objective to identify the HRD status by using gene mutation information involved in the HRR pathway. Nevertheless, there is currently no global consensus on the optimal HRR gene panel to serve as a marker for identifying the HRD status and predict patients’ prognosis. BRCA1/2 are well-characterized genes involved in the HRR pathway, and loss of function of the genes results in HRD [10, 11]. Deficiencies in other genes, such as RAD51D, BRIP1, and the FANC gene family, have also been shown to be hypersensitive to treatment with PARPis [12,13,14]. Homologous recombination proficiency likewise requires multiple other genes [15, 16]. It is unclear whether non-BRCA HRR variation can explain the gap between BRCA mutation (BRCAm) and HRD status. To date, there is no consensus regarding the exact constituents of the HRR gene panel [17,18,19,20].

Progression-free survival (PFS) was not associated with the non-BRCA HRR mutation (HRRm) assessed using six different HRR gene panels, including the predefined 15-gene panel in the exploratory PAOLA-1 study [17]. The ORZORA trial and the PROfound trial were also conducted based on the 15 prespecified genes involved in HRR [21, 22], and a Korean study recently detected HRRm in prostate cancer with the same 15-gene panel [23]. Therefore, we conducted an analysis of HRR gene variants based on the same 15-gene panel mentioned above in the Chinese population [17, 21,22,23]. In this study, we revealed the 15-gene panel profile and its relationship with initial clinical characteristics. We also assessed the association of prognosis and the 15-gene panel HRRm in the multicenter retrospective study, which was the first to be presented in the Chinese population.

2 Material and Methods

2.1 Study Design

We conducted a retrospective multicenter study including the data of patients with EOC that were collected from six gynecological centers: The First Affiliated Hospital of USTC; Zhejiang Cancer Hospital; Fudan University Shanghai Cancer Center; Zhongda Hospital Southeast University; The Obstetrics and Gynecology Hospital of Fudan University; Sun Yat-sen Memorial Hospital, Sun Yat-sen University. The study was performed on 308 patients who had received HRR genetic testing from January 2014 to June 2022.

The inclusion criteria were as follows: (a) The patients were diagnosed with high-grade serous ovarian cancer (HGSOC), ovarian clear cell carcinoma (OCCC), or endometrioid ovarian cancer (EnOC) for the first time followed by platinum-based chemotherapy; and patients who underwent both primary debulking surgery (PDS) and interval debulking surgery (IDS) at any stages (I–IV) were included; (b) the data had complete clinicopathologic characteristics and prognostic information; and (c) the patients had a defined 15-gene panel HRR testing. Patients who were previously diagnosed with other kinds of cancer, suffered from HIV or other serious chronic diseases, or received nonstandard treatment were excluded. The baseline information on clinical characteristics and HRRm readouts were collated through a review of the electronic medical records and telephone interviews (Fig. 1).

Fig. 1
figure 1

Schematic representation of the determination of patients and analysis process. FUOG The Obstetrics & Gynecology Hospital of Fudan University, FUSCC Fudan University Shanghai Cancer Center, SUZH Zhongda Hospital Southeast University, SYSU Sun Yat-sen Memorial Hospital, Sun Yat-sen University, USTC-FAH The First Affiliated Hospital of USTC, ZJCH Zhejiang Cancer Hospital

Full size image

This study was conducted with the approval of the Institutional Review Board of the first Affiliated Hospital of USTC (approval no. 2022-KY113) and was conducted in accordance with the principles of the Declaration of Helsinki.

2.2 HRR Gene Panel Detection Methods and Data Collection

All tumor samples and peripheral blood were detected by targeted next-generation sequencing (NGS) technology to assess the following defined HRR gene panel: ATM, BARD1, BRIP1, BRCA1, BRCA2, CDK12, CHEK1, CHEK2, FANCL, PALB2, PPP2R2A, RAD51B, RAD51C, RAD51D, and RAD54L (15-gene panel). The NGS methods and analysis approaches utilized by the six centers are largely consistent. Briefly, genomic DNA was extracted, purified, and measured from formalin-fixed paraffin-embedded samples and blood cells, followed by DNA library construction. Then capture-based targeted sequencing was performed. Libraries were sequenced on an Illumina NextSeq 500 (Illumina, Inc., San Diego, CA, USA) with paired-end reads. The raw data were pre-processed and mapped to human genome (hg19) by using Burrows-Wheeler Aligner 0.7.10 and Genome Analysis Toolkit 3.2 (Broad Institute, Cambridge, MA, USA). Variant calling was performed by using VarScan and variants were annotated with ANNOVAR and SnpEf v3.6.

Pathologic/likely pathologic mutations were defined as deleterious HRR gene mutations based on the American College of Medical Genetics and Genomics guideline 2015 [24]. BRCAm was defined as a deleterious mutation in the BRCA1 and/or BRCA2 genes. Non-BRCA HRRm was defined as the presence of a deleterious mutation in one or more other genes involved in the 15-gene panel and the absence of a deleterious mutation in BRCA1/2. Relapses occurring 6 months after platinum initial therapy are considered platinum-sensitive [25]. All patients were followed up via clinical records and telephone interviews, which were completed in October 2023. PFS was defined as the time from primary surgery or biopsy to the date of disease progression or the last follow-up date without relapse [26]. The follow-up time of the patients was longer than 12 months.

2.3 Statistical Analysis

Statistical analysis was performed using R 4.3.1 and GraphPad Prism v9.0 software. Categorical variables are described as frequencies and percentages, and continuous variables are expressed as medians and ranges. The chi-squared or Fisher’s exact test was used for categorical variables, while the Mann–Whitney U test was used for continuous variables. PFS was by Kaplan–Meier methods with log-rank or Tarone-Ware tests. Survival curves were generated with univariate analysis, while Cox proportional hazards regression models with hazard ratios (HRs) and 95% confidence intervals (95% CIs) were used to identify independent predictors of PFS. All variables were subjected to multivariate analysis with p values of < 0.1 in univariate analysis and selected with the forward selection likelihood ratio. All two-sided p values < 0.05 were considered statistically significant.

3 Results

3.1 Demographic Characteristics and Initial Clinical Features of Patients

In total, 466 patients were recruited, and 308 patients were eligible for analysis (Fig. 1). A summary of clinicopathologic characteristics is shown in Table S1 (Online Supplemental Material (OSM)). The median age at diagnosis was 54 years (interquartile range: 49–61), and 69.8% were over 50 years of age. The distribution of the age is shown by the 15-gene panel HRR gene, and the median age of patients with different HRR gene mutations ranged from approximately 48 to 59 years old (Fig. 2a). The distribution of age in patients with germline gene mutations was also delineated, including BRCA1, BRCA2, RAD51C and RAD51D genes, because they were present in a higher proportion in this study (Fig. S1a, OSM). Patients who received PARPis made up 29.1% (37/127) and 50.5% (46/91) in the HRR wild-type (HRRwt) group and HRRm group at stages IIIC-IV, respectively (Fig. 2b and Table S1, OSM). The majority of subtypes were HGSOC (89.0%), while the others accounted for the minority of our group (OCCC: 8.1%, EnOC: 2.9%). In addition, no BRCA1/2 mutations were observed in patients with OCCC, and no non-BRCA HRRm was detected in patients with EnOC (Fig. 2c and Table S1, OSM). Patients with advanced-stage disease (FIGO IIIC-IV) predominated in the study, and the HRRm rate did not differ between early- and advanced-stage patients (Table S1, OSM). There was no significant difference between the HRRwt and HRRm groups among the initial characteristics and the result was the same between patients with and without germline HRRm (Table S1, OSM).

Fig. 2
figure 2

HRR gene mutation profile and its relationship with clinical characteristics. (a) Distribution of age stratified by HRR gene using ridgeline plot. Genes (PPP2R2A, RAD54L, PALB2, and FANCL) with no or fewer than three mutation sites are not shown. (b) The proportion of patients who received PARPis. (c) The proportions of mutations in different pathology types. BRCAm BRCA mutation, EnOC endometrioid ovarian cancer, HGSOC high-grade serous ovarian cancer, HRR homologous recombiantion repair, HRRm HRR mutation, HRRwt HRR wild-type, NA not available, OCCC ovarian clear cell carcinoma, PARPi poly (adenosine diphosphate-ribose) polymerase inhibitor

Full size image

3.2 HRR Mutation Profile of the 15-Gene Panel

The variants of 15-gene panel HRRm are provided in detail in Table S2 (OSM). A total of 43.2% (133/308) of patients were ascertained to carry 144 deleterious HRRm. Nine patients carried more than one mutation; among these six patients had two mutations in different genes and one patient had mutations in BARD1, BRCA1, and BRIP1. One patient carried two mutations in the same gene (ATM: c.2413C>T and c.3840_3841del), and another mutation site in BCRA2 (c.1813_1814ins and c.6566_6567ins) co-occurred in CHEK1. The ages of the nine patients ranged from 47 to 68 years, and the rates of recurrence and platinum sensitivity were randomly distributed (Table S3). Of the 144 deleterious mutations, 68.1% (98/144) were germline mutations, which accounted for 31.8% of all patients (Table S4, OSM), and 70.1% (101/144) were BRCA1/2 mutations, accounting for 32.8% of the group (Fig. 3a and Table S4, OSM).

Fig. 3
figure 3

HRR gene mutation characteristics. (a) The frenquency of HRR gene mutations. (b) The mutation types in different HRR genes. (c) Mutation sites of BRCA1 gene using a lollipop plot. (d) The gene mutations of 15-gene-panel involved in HRR pathway were shown in the waterfall plot by oncoprint. g germline, s somatic, Del deletion, Indel insertion and deletion, Ins insertion, Mis missense, Non nonsense, Spl splicing, Other one patient carried the CHEK1 mutation co-occurring with two BRCA2 mutations and one patient had two ATM mutations

Full size image

BRCA1 gene mutations were predominantly found in 24.0% (74/308) of patients, followed by BRCA2 gene mutations in 8.8% (27/308) of patients, and non-BRCA HRRm presented a low frequency (Fig. 3a and Table S4, OSM), as reported previously [17, 18]. In addition, RAD51D accounted for the majority of variants (2.6%, 8/308) in the non-BRCA HRRm (Table S4, OSM). The deleterious variants represented different types of mutation, including frameshift (53.5%), nonsense (21.5%), missense (15.3%), and splice site variants (9.7%) (Fig. 3b and Table S2, OSM). Fifteen mutation sites were observed to occur more than twice (Table S2, OSM). Moreover, we presented the distribution of variant sites in BRCA1 and BRCA2 using a lollipop plot, and no hotspots were observed in BRCA1/2 genes (Fig. 3c and Fig. S1b, OSM). A concurrent analysis described the HRRm profile and revealed the distribution of the prognostic indicators related to HRRm (Fig. 3d).

3.3 Prognosis Analysis

The rate of follow-up was 93.2%, and the survival results of patients who provided complete prognostic information were analyzed. The median follow-up time was 27.0 months (interquartile range: 19–35 months). These prognostic characteristics were from stages IIIC–IV patients mainly because of their poorer prognosis compared with early-stage patients. A total of 89.7% (78/87) of patients showed platinum sensitivity with HRRm, which was significantly higher than the 78.3% (94/120) of patients with HRRwt (p = 0.032). There was no significant difference in platinum sensitivity between patients with and without germline HRRm. Notably, the recurrence rates were 24.1% (21/87) and 50.0% (60/120) in patients with and without HRRm, respectively, which presented a significant difference (p < 0.001), and the result was consistent between patients with and without germline HRRm (Table S5, OSM).

The median PFS was significantly longer in the HRRm group than in the HRRwt type group among all patients (not reached vs. 43.0 months; HR for PFS, 0.42; 95% CI: 0.28–0.64; p = 0.0001) (Fig. 4a), while the result among IIIC-IV patients was the same (not reached vs. 27 months; HR for PFS, 0.42; 95% CI: 0.27–0.65; p = 0.0003) (Fig. 4b). In the subgroup analysis, the difference between the BRCAm group and the HRRwt group was significant, and the estimated HR was 0.39 (95% CI: 0.26–0.60; p = 0.0001) in all patients and 0.33 (95% CI: 0.24–0.60; p = 0.0002) in IIIC–IV patients (Fig. 4c, d). PFS in patients receiving PARPis was significantly better than in those not receiving PARPis in IIIC–IV patients (HR for PFS, 0.48; 95% CI: 0.31–0.74; p = 0.0028) (Fig. 4e). For the subgroups of stages IIIC–IV patients who did not use PARPis, the prognosis of patients with HRRm was better than that of patients in the HRRwt group (HR for PFS: 0.41; 95% CI: 0.24–0.68; p = 0.0034) (Fig. 4f). Subsequently, the impact of different variables on recurrence was explored among IIIC–IV patients. In univariate analysis, significant predictors of relapse included non-residual status (HR: 0.62; 95% CI: 0.39–0.98; p = 0.0287), 15-gene panel HRRm (HR: 0.42; 95% CI: 0.28–0.64; p = 0.0001), platinum sensitivity (HR: 0.07; 95% CI: 0.02–0.22; p < 0.001) and maintenance therapy (PARPi) (HR: 0.48; 95% CI: 0.31–0.74; p = 0.0028) (Fig. 5). In multivariate Cox analysis, non-residual (HR: 0.63; 95% CI: 0.40–0.99; p = 0.046), 15-gene panel HRRm (HR: 0.49; 95% CI: 0.28–0.88; p = 0.016) and platinum sensitivity (HR: 0.02; 95% CI: 0.01–0.05; p < 0.001) remained independent prognostic factors (Fig. 5). The results for all patients are shown in Fig. S2 (OSM). The variables stage (IIIC–IV) (HR: 2.26; 95% CI: 1.22–4.19; p = 0.009), 15-gene panel HRRm (HR: 0.47; 95% CI: 0.28–0.80; p = 0.005), and platinum sensitivity (HR: 0.02; 95% CI: 0.01–0.35; p < 0.001) were associated with survival.

Fig. 4
figure 4

Kaplan-Meier curves of PFS stratified by HRR gene mutation (a/b/f), BRCA1/2 mutation (c/d), maintain therapy (PARPi) (e). CI confidence interval, PFS progression-free survival. # Tarone-Ware test

Full size image
Fig. 5
figure 5

Univariate and multivariate analysis of factors associated with progression-free survival of patients (stages IIIC–IV). DDS the disease dissemination score, HR hazard ratio, P+LN metastasis lymph and peritoneum metastasis, SCS surgical complexity score

Full size image

3.4 Mutation Characteristics of Our Study and Previously Published Studies

In the subset of 96 patients with germline mutations, there were 25.0% BRCAm and 6.8% non-BRCA HRRm in our study, while the frequencies of somatic mutations were 7.8% and 7.1%, respectively. We compared the data with previously published results, which presented different gene panels. Six different panels were assessed with the data from the PAOLA-1 study, and the frequency of non-BRCA HRRm ranged from 3.7% to 9.8% [17]. Using the data from the NOVA study, the 20-gene panel was presented and analyzed with 14.5% BRCAm and 12.4% non-BRCA HRRm [27]. These studies using patients with Asian ethnicity included 13-gene, 24-gene, and 15-gene panels and presented the frequencies of mutations, in which 12.3–28.3% of patients carried BRCAm and 2.0–16.7% of patients had non BRCA mutations [18, 20, 28]. Furthermore, 3.8–10.0% patients with non-BRCA HRRm were shown in the other studies to originate from ethnic Europeans and Americans [15, 22, 29,30,31] (Fig. 6).

Fig. 6
figure 6

The BRCAm and non-BRCA HRRm in different gene panels in our study and previous published study. BRCAm BRCA mutation, HRRm HRR mutation, G+S germline and somatic mutations

Full size image

4 Discussion

Six primary pathways have been identified for DNA damage repair (DDR), which involve double-strand DNA breaks (DSB) and single-strand DNA breaks (SSB) caused by various injury mechanisms. Homologous recombination (HR) and nonhomologous end joining (NHEJ) are the main pathways for DSB repair. HR is active in the S/G2 phase when a sister chromatid is available, while NHEJ repairs DSB throughout all cell cycle phases except M phase. NHEJ is faster and mainly occurs in the G1 phase [32]. Error-free repair of DSBs is highly dependent on HR in response to various types of DNA damage [33]. It is recommended to carry out the detection of BRCA1/2 or genomic “Scar”, which is used to measure genomic instability (including genome-wide loss-of-heterozygosity, telomeric allelic imbalance and large-scale state transitions), to identify HRD status [34]. BRCA1/2 are the key components in the HRR pathway, and other genes are also essential for this process, such as ATM, which encodes a sensor that recognizes damage and RAD51 paralogs assembled at the site of DSB to launch the repair process [35]. Deficiencies in these non-BRCA HRR genes confer sensitivity to PARPis and were reported to be related to ovarian cancer risk [14, 36, 37]. The PI3K pathway plays a significant role in DDR and is particularly important in EOC. Upregulation of the PI3K pathway is commonly observed in EOC and contributes to chemoresistance and genomic stability. It is involved in DNA replication and cell cycle regulation. Inhibiting PI3K can result in genomic instability and mitotic catastrophe by reducing the activity of Aurora kinase B, a protein critical for the spindle assembly checkpoint. This can lead to an accumulation of lagging chromosomes during prometaphase [38]. However, controversy remains when trying to define the gene panels for HRR. The efficacy of the prespecified 15-gene panel has been demonstrated by many research groups, including The PROfound trial and The ORZORA trial [21, 23]. Pujade-Lauraine also presented data from six HRR gene panels, including the 15-gene panel, when analyzing the data from the PAOLA-1 study [17]. Currently, the 15-gene panel might be the most common panel when we discuss the HRR pathway or cancer risk associated with HRD.

The mutation profile of the 15-gene panel, which was not previously explored in the Chinese population, was presented in detail in our study. We also showed the germline mutations of the 15-gene panel that were not investigated in the PROfound and PAOLA-1 studies. In conclusion, as reported in a previous study, we found that germline BRCA1/2 mutations were still the most common and that the frequencies of germline non-BRCA mutations, such as BRIP1, RAD51C and RAD51D, were higher than those in previous data [39,40,41]. This may be due to differences in ethnicity. HRRm is more likely to be hereditary and shows familial aggregates [42]. The National Comprehensive Cancer Network (NCCN) recommends that risk-reducing salpingo-oophorectomy be considered for women with germline HRRm at a certain age [43]. Given that BRCA1, BCRA2, RAD51C, and RAD51D had a higher proportion, the distribution of age in patients with germline 15-gene panel mutations was similar to previous results [44].

Patients with 15-gene panel HRRm in this study had longer PFS than patients without mutations. This finding is consistent with some previous studies. Yao reported a 13-gene for HRRm in Chinese EOC and found that 13-gene HRRm predicted longer prognosis, while other non-BRCA HRR gene mutations did not show a better PFS; the limitations of this study were the short follow-up time and PARPi usage information [18] (Fig. 6). However, the results in our study were inconsistent with some other reports when considering the prognosis of the non-BRCA HRRm group [30, 31] (Fig. 6). They also found that patients with HRRm had a better prognosis, while the difference was that patients with non-BRCA HRRm also showed a better prognosis. Norquist et al. defined HRRm as a 16-gene panel [30]. The patients, mostly non-Hispanics and Whites, were enrolled from the GOG 218 study at stage III-IV without using PARP inhibitors, and 25.4% of patients were identified as having HRR mutations. The authors found that HRR mutations, including non-BRCA genes, significantly prolonged PFS and overall survival (OS) [30]. Pennington et al. reported that patients with 13-gene panel HRRm had a benefit in OS and platinum responsiveness [31]. In addition, the percentage of BRCA or HRR gene mutations in our study was higher than other previous studies. All these inconsistent results above might be ascribed to different HRR gene panels, treatment strategies and ethnicities.

In the subgroup of patients not using PARPis, we found that the patients with HRRm showed longer PFS than those with HRRwt. PAPR inhibitors can achieve excellent OS and PFS with advanced EOC regardless of BRCA, BRCAwt, HRD, or HRP and unknown gene status [45,46,47]. Owing to the small sample size of PARPis usage in this study, we cannot analyze the PFS difference in the HRRm patients with or without PARPis. We also indicated that the independent prognostic factors were platinum sensitivity, disease nonresidual and 15-gene panel HRRm, in accordance with previous reports. In the latter study, we will provide overall survival data and more maintenance treatment results.

It should be noted that this study has some limitations: First, samples of retrospective data from different centers may lead to potential recall bias. Second, the prognosis was analyzed without OS data.

5 Conclusions

In summary, our study depicted the gene mutation profile involved in HRR pathway and revealed that BRCA1/2 mutations, rather than the non-BRCA HRRm, appear to predict the prognosis of EOC in a Chinese clinical study. Careful consideration is necessary when expanding the gene panel for genetic testing in EOC patients. Further rigorous research is still required for the treatment of patients with epithelial ovarian cancer.