Clinical Impact of Next-Generation Sequencing Multi-Gene Panel Highlighting the Landscape of Germline Alterations in Ovarian Cancer Patients

BRCA1 and BRCA2 are the most frequently mutated genes in ovarian cancer (OC) crucial both for the identification of cancer predisposition and therapeutic choices. However, germline variants in other genes could be involved in OC susceptibility. We characterized OC patients to detect mutations in genes other than BRCA1/2 that could be associated with a high risk of developing OC and permit patients to enter the most appropriate treatment and surveillance program. Next-generation sequencing analysis with a 94-gene panel was performed on germline DNA of 219 OC patients. We identified 34 pathogenic/likely pathogenic variants in BRCA1/2 and 38 in other 21 genes. The patients with pathogenic/likely pathogenic variants in the non-BRCA1/2 genes mainly developed OC alone compared to the other groups that also developed breast cancer or other tumors (p = 0.001). Clinical correlation analysis showed that the low-risk patients were significantly associated with platinum sensitivity (p < 0.001). Regarding PARP inhibitors (PARPi) response, the patients with pathogenic mutations in the non-BRCA1/2 genes had worse PFS and OS. Moreover, a statistically significantly worse PFS was found for every increase of one thousand platelets before PARPi treatment. To conclude, knowledge about molecular alterations in genes beyond BRCA1/2 in OC could allow for more personalized diagnostic, predictive, prognostic, and therapeutic strategies for OC patients.


Introduction
In 2020, ovarian cancer (OC) placed 8th for both incidence and mortality among women [1]. OC is defined as a "silent killer" as it is often diagnosed at advanced stages, resulting in a high mortality rate. About 10-30% of breast and ovarian cancers show familial aggregation and 5-10% of them are considered hereditary, i.e., linked to a germline genetic variant in a cancer predisposition gene [2]. BRCA1 and BRCA2 are two of the most frequently mutated genes in high-grade serous OC, which is responsible for the majority of OC deaths [3]. Surgery represents the standard treatment, followed by platinum-taxane chemotherapy. Platinum-resistant OCs recur in nearly 25% of patients within six months and the probability of survival after five years is about 30% [4]. Investigating the BRCA1/2 mutational status in patients with OC is crucial not only for the identification of familial cancer predisposition, but also to address therapeutic choices. Indeed, patients with germline or somatic variants of BRCA1 or BRCA2 have been correlated to a better prognosis and a better response to platinum-based chemotherapy [5][6][7][8]. Moreover, inhibitors of the poly (ADP-ribose) polymerase (PARPi) have been demonstrated to be effective in germline and somatic BRCA1or BRCA2-mutated OCs [9][10][11], except for Niraparib that has been shown to be effective regardless of the presence of BRCA1/2 variants or the homologous recombination deficiency (HRD) status [12].
About 13% of high-grade serous OCs show germline variants in BRCA1 or BRCA2 [13,14] and a smaller percentage can be attributed to other germline variants. The BRCA1 and BRCA2 genes are the main actors involved in the homologous recombination (HR) DNA repair pathway, together with ATM, BARD1, NBN, and others [15]. Indeed, about 30% of patients with OC present germline and somatic variants in HR genes, of which up to 75% are in the BRCA1 and BRCA2 genes [5,15]. Therefore, the detection of new genes determining susceptibility to cancer is an urgent need. To date, other genes have been associated with inherited OC predisposition, particularly BRIP1, RAD51C, RAD51D, PALB2, STK11, and mismatch repair (MMR) genes, such as MLH1, MSH2, MSH6, and PMS2, and others [16][17][18]. Next-generation sequencing (NGS) advent has enabled the analysis of a high number of genes simultaneously with lower costs and a wider access to molecular tests for patients with suspected hereditary cancer or for eligibility for PARPi treatment [19][20][21][22].
In this study, we performed a molecular characterization of patients with OC to identify predisposing genes other than BRCA1 and BRCA2 that could permit a better selection of patients with a high risk of developing OC and allowing them to enter the most appropriate treatment and surveillance program.

Clinicopathological Features
Between January 2014 and December 2018, 219 patients were recruited in this study. All the patients had a diagnosis of OC: 182 patients developed OC alone, 16 developed OC and breast cancer, and 21 patients developed OC and another non-breast-cancer tumor. The median age at diagnosis was 60 years. One hundred sixty-two patients of the 219 patients had high-grade serous OC, 16 had endometrioid OC, and 41 patients presented other histologies. Twelve patients developed G1 OC, 4 patients developed G2 OC, and 188 patients developed G3 OC (data not available for 15 patients). Forty-three patients had stage I/II OC and 164 had stage III/IV OC (data not available for 12 patients).

NGS Analysis
The molecular analysis of the 219 patients showed a mean target coverage of 404 X and a 95.3% mean percentage of the target covered > 50 X.
We observed 42,583 variants in the exonic and splicing regions of 94 genes.

Pathogenic Variants in Other Genes
We also analyzed the other 92 genes of the panel, and we observed a total of 40,082 variants that were classified according to the ACMG guidelines in 38 pathogenic/likely pathogenic variants, 4710 VUS, and 35,334 benign variants. The 38 pathogenic/likely pathogenic variants were present in 21 genes in 36/219 patients (16.4%): PPM1D (8 variants), The mutation details are shown in Supplementary Table S2. Out of the 36 patients, two had pathogenic/likely pathogenic variants in two different genes (patient B184 in CHEK2 and EGFR, and patient B421 in FANCD2 and ALK), while four also had a pathogenic mutation in the BRCA1/2 genes (one in BRCA1 and three in BRCA2), so these four patients were considered in the BRCA1/2-mutated group.

Molecular Subgroups: Clinical and Pathological Comparison
The clinical and pathological information on the 219 patients included in this study is summarized in Table 1. Our patient cohort was grouped based on gene mutations and the presence of OC alone or OC and additional tumors ( Table 2). We observed an association between the type of tumor and the mutational status (p < 0.001). In particular, the patients with pathogenic variants in the genes other than BRCA1/2 mainly developed OC alone compared to the other groups. The patients with OC and breast cancer had predominantly BRCA1/2 pathogenic mutations, whereas only three out of 21 patients with OC and other tumors had a pathogenic variant. All the three patients who developed OC and breast cancer with a germline BRCA1 mutation had a breast cancer diagnosis (median age, 42 years) before the onset of OC (median age, 62 years). Again, in the seven patients with a germline BRCA2 mutation, the onset of breast cancer always preceded OC diagnosis, with a median age of 52 years and 66 years, respectively. The relationship between the neutrophils, lymphocytes, platelets, neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune inflammation index (SII) before treatment initiation and the mutational status was investigated. We did not observe any significant association (Supplementary Table S3).

Platinum Sensitivity
Firstly, the association between platinum sensitivity and the mutational status was investigated. No statistically significant association between platinum sensitivity and the mutational status was observed (Table 3). We then analyzed the response to platinum-based chemotherapy of 208 patients (11 patients were undetermined). The OC patients were classified as low-risk if the residual disease did not occur during the primary debulking surgery. The patients with residual disease and/or patients who underwent interval surgery were classified as high-risk. The median time from the last administration of platinum-based chemotherapy and the relapse or death was 24 We did not observe any significant correlation between the inflammatory indexes and the mutational status.

PARP Sensitivity
In our case series, 43 patients were treated with maintenance treatment in subsequent lines rather than the first one. Twelve of them presented a pathogenic mutation in BRCA1/2 (five in BRCA1 and seven in BRCA2), four patients had pathogenic/likely pathogenic variants in four genes other than BRCA1/2 (PALB2, ERCC2, ALK, and MITF), and 27 patients had no pathogenic mutations in any of the 94 genes.
The mutational status was analyzed in relation to PARP sensitivity in 43 patients. We observed that the patients without pathogenic variants had a similar outcome compared to the patients with BRCA1/2 pathogenic variants (HR = 1.15, 95% CI, 0.53-2.48, p = 0.715; HR = 0.72, 95% CI, 0.27-1.91, p = 0.511, for PFS and OS, respectively). Differently, the patients with pathogenic variants in the genes other than BRCA1/2 had a significantly worse PFS (HR = 3.56, 95% CI, 1.05-12.04, p = 0.042) and a worse OS (HR = 1.38, 95% CI, 0.15-12.13, p = 0.772), as shown in Table 4. The response to PARPi was also evaluated in association with the inflammatory indexes before PARPi treatment initiation for 29 patients. Considering NLR, PLR, and SII as continuous variables, no statistically significant association was found for both PFS and OS. However, for every one standard deviation increase in platelets, a statistically significantly worse PFS was found (HR = 1.52, 95% CI, 1.03-2.26, p = 0.037), as shown in Table 5. No association was found in relation to OS (HR = 1.18, 95% CI, 0.62-2.27, p = 0.614).

Discussion
In this study, the molecular characterization of the germline DNA of 219 OC patients was provided through a panel of 94 genes, including the genes involved in the main hereditary cancer syndromes, and the correlation with clinical characteristics was studied. The existing clinical genetic tests for OC are based only on BRCA1 and BRCA2 analysis despite new evidence of a higher number of genes eligible for testing [16]. In our cohort selected by the IRST Genetic Counseling service and the Oncology Units of the Area Vasta Romagna (AVR) catchment area, we observed a total of 72 pathogenic/likely pathogenic variants in 70/219 (32%) patients. In particular, 14 variants were found in the BRCA1 gene, 20 variants in BRCA2, and 38 pathogenetic/likely pathogenic variants were found in other 21 genes. The 38 pathogenic/likely pathogenic variants in the genes other than BRCA1/2 were observed in 36 patients, 32 of whom did not present any pathogenic/likely pathogenic variants in the BRCA1/2 genes. The most frequently mutated genes in our case series were PPM1D, MUTYH, MITF, RAD51C, BRIP1, ALK, and CHEK2. Out of the 36 patients with pathogenic/likely pathogenic variants in the other genes, four also harbored a BRCA1/2 mutation (one in BRCA1 and three in BRCA2). These four patients did not show particular features in the tumor characteristics or in the therapeutic response. Consequently, they were included in the BRCA1/2-positive group, presuming a stronger effect of BRCA1/2 mutations on the phenotype.
BRCA1 and BRCA2 are part of the BRCA-Fanconi anemia pathway, and other Fanconi genes, such as BRIP1 and RAD51C, have also been associated with an inherited risk of OC [23][24][25]. PPM1D variants have been associated with predisposition to breast and ovarian cancers [26,27] along with MUTYH [28,29] and CHEK2 [29]. Moreover, mutations in the mismatch repair genes that cause Lynch syndrome (MLH1, MSH2, MSH6, and PMS2) also confer a risk for OC [30,31]. In our case series, we detected a likely pathogenic variant in the MLH1 gene in a patient negative for BRCA1/2 mutations and a likely pathogenic variant in the MSH2 gene in a patient with a pathogenic mutation in the BRCA2 gene. Other authors also underlined that germline sequencing of BRCA1 and BRCA2 should be performed in the context of a multigene panel that also includes RAD51C, RAD51D, BRIP1, MLH1, MSH2, MSH6, PMS2, and PALB2 [32]. These data highlighted that these mutations are associated with a higher risk of OC development, so it is noteworthy to introduce a multigene panel in standard genetic analysis protocols for patients with suspected hereditary OC.
The patients enrolled in this study were selected on the basis of OC diagnosis in order to identify the patients eligible for PARP inhibitors treatment. Consequently, we do not have information on the family history of cancer for the majority of them. However, all the BRCA1/2-positive patients and the ones with pathogenic/likely pathogenic variants in the genes for which clinical guidelines are available (e.g., TP53, MSH2, MLH1) were referred to appropriate genetic counseling with the reconstruction of the family history of cancer and the extension of the genetic test to the consenting relatives.
Our molecular findings were then analyzed in relation to clinical characteristics. Our cohort of patients was divided into molecular subgroups, and we showed that patients with pathogenic/likely pathogenic variants in the genes other than BRCA1/2 developed mainly OC alone, suggesting that these genes could be specifically related to OC predisposition. Interestingly, all the patients with a germline BRCA1 or BRCA2 mutation and a second malignancy first developed a breast cancer and thereafter an OC. This observation highlights the need for a strict follow-up in order to identify OC early and/or to discuss with the patient risk-reducing salpingo-oophorectomy. This procedure was also demonstrated to reduce the risk of breast cancer in the immediate 5 years after surgery and in the longer term, especially in younger women [33,34].
Although most patients with OC initially respond to platinum-based chemotherapy, about 20% of women experience disease progression ≤ 6 months after the last cycle of a platinum-based regimen (platinum-resistant or platinum-refractory) [35]. Many efforts have been made over the years to develop predictive biomarkers of platinum sensitivity [36].
We found that patients without residual disease after the primary debulking surgery (low-risk patients) had a significantly longer median time from the last administration of platinum-based chemotherapy and the relapse or death (35.8 months) than high-risk patients (15.2 months) (p < 0.001). We also hypothesized a better clinical outcome in patients with the DNA damage response genes altered because of a worse platinum-induced DNA interstrand crosslinks repair capability [37].
A subset of 43 patients was treated with PARPi as maintenance treatment after platinum chemotherapy. We observed that the patients with pathogenic mutations in the genes other than BRCA1/2 had significantly worse PFS and OS compared to the patients with a pathogenic mutation in the BRCA1/2 genes, suggesting that this may be associated with specific biological mechanisms. However, due to the small number of cases, we could not speculate about it.
On the other hand, the patients without pathogenic/likely pathogenic variants had a similar outcome to the patients with a pathogenic variant in the BRCA1/2 genes, confirming literature data [38].
Regarding inflammation indexes correlation, we previously demonstrated that inflammatory indexes (NLR and SII) are independent prognostic factors in recurrent platinumsensitive OC patients [36]. In this analysis, only platelets were correlated with PFS, demonstrating their important role in OC not only as a poor prognostic factor [39], but also as possible predictive factor of response to PARPi. However, a validation of these easy biomarkers in a larger case series is warranted.
One of the aims of this study was to identify new genes involved in the predisposition to OC. For this reason, we chose to analyze a panel of 94 genes involved in different forms of hereditary cancer and include all the pathogenic/likely pathogenic variants in the analysis. However, this approach has some limitations. Indeed, literature statistics describe only a minor fraction of OC cases attributable to genes other than BRCA1/2 [23,40]. In particular, the eight PPM1D variants detected could be treatment-related mosaic somatic mutations, as suggested by other authors [41], but we considered them as real germline variants since one of them was present in three different patients, and among the five remaining variants, three were already reported in the literature and one of the two unreported variants was present in a patient with a BRCA2 pathogenic variant. Moreover, the VAF (variant allele frequency) was compatible with a germline nature of the mutations (~0.5). Regarding the ALK gene, one of the two mutations was present in a patient with also a FANCD2 pathogenic variant. Moreover, ALK and MITF germline alterations were associated with neuroblastoma and melanoma risk, respectively, but this finding is intriguing for possible links with other cancers such as OC. Two CHEK2 pathogenic variants were found in patients with highgrade serous ovarian cancer. The CHEK2 gene is mainly linked to ovarian cystadenomas, borderline ovarian tumors, and low-grade invasive cancers [42], but at the same time this gene is involved in DNA damage response and is associated with a moderate risk of breast cancer [18] and also a risk of other tumors [43], such as prostate [44][45][46], colorectal [47], and gastric cancers [48]. Regarding the MUTYH gene, only biallelic mutations are wellknown for their association with cancer risk [49], but an increased risk for cancer has also been reported for monoallelic carriers since the conjunction of a germline mutation with a somatic mutation may also contribute to the development of OC [28]. For these reasons, we decided to include all these patients in the group of carriers of pathogenic/likely pathogenic variants in the non-BRCA1/2 genes.

Ethics Statement
The study was performed in accordance with the Good Clinical Practice and the Declaration of Helsinki and approved by the AVR Ethics Committee (protocol No. 6326/2020). All the patients enrolled in the study signed informed consent forms for genetic analyses and for the use of the results for research purposes.

Patients and Samples
Patients with a diagnosis of OC referring to the IRST Genetic Counseling service or to the Oncology Units of the AVR catchment area in 2014-2018 were included in this study. A total of 219 patients were recruited and tested for a multi-gene panel, including the search for BRCA1/2 alterations for possible treatment with PARPi.
Peripheral blood of patients was collected and stored at −80 • C for subsequent molecular analyses. Genomic DNA was extracted with a QIAamp DNA mini kit (Qiagen, Hilden, Germany) and quantified using a Qubit dsDNA BR Assay kit (Thermo Fisher Scientific, Waltham, MA, USA).  Table S5). Sequencing was performed using the MiSeq platform (Illumina) with MiSeq Reagent Kit v2 configured for 2 × 150 cycles according to the manufacturer's instructions as previously described [50,51].

Data Analysis and Variant Calling
Paired-end sequencing reads were aligned to the reference human genome (UCSC hg19) with Burrows-Wheeler algorithm v0.7.15-r1140 [52]. The sequences around insertions and deletions (indels) were realigned locally with GATK v3.6-0 [53]. Then, Picard MarkDuplicates v2.6.0 (http://broadinstitute.github.io/picard/, accessed on 16 October 2022) was used to remove duplicate read-pairs artifacts arising during PCR amplification or sequencing. The data then underwent base quality score recalibration (BQSR) to ensure good call quality and reduce the number of false positives (again, with GATK). Variant calling was separately performed with GATK UnifiedGenotyper and freebayes v1.0.2-58 [54]; then, the resulting VCF files were merged with GATK CombineVariants. ANNOVAR v2016-02-01 was used for genomic and functional annotations of the detected variants [55], while coverage statistics were computed with the DepthOfCoverage utility of GATK and the downstream custom bash/R scripts. The resulting annotated list of variants was filtered for variants present in exonic regions or in the 20 bases flanking each exon.

Variants Classification
Genetic variants were classified into five classes according to the IARC recommendations [56]. The BRCA1/2 variants classification was performed by consulting ClinVar [57] and the main BRCA1/2 mutation databases, such as BRCA Exchange, BRCA Share, and LOVD [58][59][60]. Sequence variants in the other 92 genes were classified using ClinVar [57] and dbSNP [61]. The variants absent in any of these databases were classified using VarSome [62] in accordance with the guidelines of the American College of Medical Genetics [63].

Inflammatory Indexes
Information on the neutrophil, lymphocyte, and platelet counts from blood tests carried out at baseline was collected. The SII was calculated as the (platelet count × neutrophil count)/lymphocyte count, the NLR was obtained by dividing the absolute neutrophil count by the absolute lymphocyte count, and the PLR was calculated as the ratio of the absolute platelet count to the absolute lymphocyte count [64]. The inflammatory indexes value was multiplied by 100. The median value of the SII, NLR, and PLR was considered.

Statistical Analysis
The data were summarized as the mean ± standard deviation (SD) or median, interquartile (IQ) range and the minimum and maximum value, as appropriate, for the continuous variables and through natural frequencies and percentages for the categorical ones. The associations between the categorical variables were tested by Pearson's χ 2 test or Fisher's exact test, as appropriate, whereas those between a continuous variable and a categorical one were tested by means of Student's t-test or the F-test or analogous nonparametric tests, when appropriate.
Platinum sensitivity was defined as the time in months from the date of the end of platinumbased chemotherapy until the date of relapse or death from any cause, whichever occurred first. The alive patients without relapse were censored at the time of the last follow-up.
The prognosis of the patients treated with PARPi was investigated in terms of progressionfree survival (PFS) defined as the time in months from the date of inhibitor initiation until disease progression or death from any cause, whichever occurred first, and overall survival (OS) was defined as the time in months from the date of inhibitor initiation until death from any cause. The patients were censored at the date of the last follow-up update.
The time-to-event outcomes were analyzed by means of the Cox proportional hazards model; the effect of the biological and clinical covariates was reported in terms of hazard ratios (HRs) and the corresponding 95% confidence intervals (CIs).
All the statistical analyses were performed using the STATA 15.0 software (College Station, TX, USA).

Conclusions
To conclude, knowledge about molecular alterations in the genes beyond BRCA1/2 in OC could allow for more personalized diagnostic, predictive, prognostic, and therapeutic strategies for patients without forgetting the clinical implications for her family members.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Ethics Committee of IRST IRCCS (IRSTB112, approved on 30 July 2020).

Informed Consent Statement:
Informed consent was obtained from all the subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.
Conflicts of Interest: U.D.G. received advisory board or consultant fees from Merck Sharp & Dohme, Bristol Myers Squibb, Janssen, Astellas, Sanofi, Bayer, Pfizer, Ipsen, Novartis, and Pharmamar and institutional research grants from AstraZeneca, Sanofi, and Roche. A.F. has received personal honoraria for lectures from AstraZeneca, GSK-Tesaro, and Clovis and institutional research grants from Astellas, MSD, Bayer, advisory board fees from Janssen, AstraZeneca, and GSK-Tesaro. The other authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.