Somatic mutations in BRCA1&2 in 201 unselected ovarian carcinoma samples – single institution study

Introduction

Ovarian cancer is the most lethal among gynecologic malignancies with 238700 new cases and 151900 deaths reported annually worldwide [1]. In Europe alone the statistics points towards 65500 new cases and 42700 deaths each year [2]. As the cancer initially develops without any specific symptoms, only 20-30% of women are diagnosed at early stages (I-II) which translates into 90% 5-year survival rates. Unfortunately, in around 70% of cases cancer is diagnosed at late stages (III-IV) which decreases the 5-year survival rate to 25%. Despite numerous research efforts, the combination of low specificity symptoms and lack of reliable biochemical markers makes the early detection of ovarian cancer one of the main challenges for health care systems [3].
The standard of care in ovarian cancer is debulking surgery followed by chemotherapy regimens based on platinum salts [3]. In 2014 this standard was enriched by first registration of poli-ADP-ribose Polymerase inhibitors. Since 2014 several other indications of PARP inhibitors became available for OC patients with germline or/and somatic mutations in BRCA1 & BRCA2 genes, including maintenance therapy in the 1st line setting translating to spectacular progression-free survival (PFS) difference [4].
The prevalence of mutations in BRCA1 & BRCA2 genes in OC patients population is approximately 14-18% and 7-8% for germline and somatic setting, respectively [5, 6]. In Poland, until recently the dominant constitutive mutation detection system was based on founder mutations screening, focusing on 3-5 most frequent loci [7, 8]. Dramatic cost reduction of Next Generation Sequencing (NGS) and advent of PARP inhibitors have been the main driving forces for the founder screening system to be gradually replaced by full genes sequencing of DNA isolated either from leukocytes (germline mutation) or tumour specimen (germline and somatic mutation detection). Initial studies clearly showed that founder mutations cover as little as 48-65% of the germline mutation spectrum further reinforcing the testing paradigm shift towards NGS-based approach [9, 10, 11]. In Poland, germline mutations are now detected in 15% of unselected ovarian cancer cases and as many as 28% for testing from tumour specimen [10, 12, 13]. Introduction of NGS-based tumour testing has the advantage of detection both germline and somatic mutations yet poses several challenges in the patients pathway and lab pipeline [14]. Here we present the results of analysis of 201 unselected ovarian cancer cases specimens that were tested for BRCA1/BRCA2 mutations using Next Generation Sequencing of tumour-derived DNA.

Material and methods

The material for the study derives from 201 unselected patients with ovarian carcinomas operated in 2015-2017 at the Department of Gynecologic Oncology Clinic Holycross Cancer Center in Kielce (Table I). Mean and median age of patients at diagnosis were 58.5 and 59 years, respectively. Formalin-fixed paraffin-embedded tissue blocks stored at the Department of Pathology were used for the study. In terms of histopathology, the majority of cases (78%) in the study group were serous and 11% were endometrioid. The other histopathological types were clear cell, mucosal, undifferentiated and mixed. Among serous carcinomas, the high grade (HG) subtype constituted 78,7% and the low grade (LG) –21.3 (6/33; 38%). According to FIGO (International Federation of Gynecology and Obstetrics) classification, 76% of patients were in stage III/IV, 20% in I/II and for 4% no data were available.
All of the study procedures were approved by the Institutional Review Board at the Holycross Chamber of Physicians in Kielce (10/2016) and performed according to the Declaration of Helsinki. All patients gave their informed consent and all methods were performed in accordance with the relevant guidelines and regulations.

DNA isolation

The pathologist marked the area containing OC tumor cells on a hematoxylin and eosin-stained slide. In all studied cases in the marked area tumor cell content was more than 70%, range (10-100%). Then, the tumor tissue on matched unstained slides was deparaffinized and the selected area was transferred to a tube for DNA isolation using the Maxwell 16 and Maxwell® 16 FFPE Tissue LEV DNA Purification Kit, according to the manufacturer’s instructions (Promega, USA). The isolated DNA concentration was measured by using NanoDrop (TkBiotech, Warsaw, Poland). Mean concentration of isolated DNA was 100ng/µl and purity for 260/280 ratio ranged between 1.8-2.0.

Next Generation Sequencing Library preparation

The DNA was diluted to 10 ng/µl. The libraries were prepared using the Oncomine™ BRCA Research Assay, Manual Library Preparation and the Ion Xpress Barcode Adapters Kit (Thermo Fisher Scientific), according to the manufacturer’s instructions (Thermo Fisher Scientific). Oncomine™ BRCA Research Assay is two pool Ion AmpliSeq™ design containing 265 amplicons. Two separate polymerase chain reaction (PCR)-multiplex reactions were performed for each of the samples tested, and the samples were combined into one. The resulting multiplex PCR products were subjected to partial enzymatic digestion to remove primer sequences. Next, adapters for multiplex PCR products were enzymatically attached using the Ion Xpress Barcode Adapters Kit (Thermo Fisher Scientific). One of the adapters contains barcodes that allow identification of sequences from a given patient among a mixture of libraries. The prepared libraries were cleaned using Agencourt AMPure XP (Beckman Coulter Genomics) according to the manufacturer’s instructions (Ion AmpliSeq Library Kit 2.0 – Thermo Fisher Scientific).

Preparation of clonally amplified template for sequencing – emulsion PCR (emPCR) for S5 using IonChef

The concentration of libraries was measured by quantitative PCR with real-time detection (qRT-PCR) using the Ion Library TaqMan™ Quantitation Kit (Thermo Fisher Scientific) on a Rotor-Gene Q instrument (Qiagen). Based on the values obtained with qRT-PCR, all prepared libraries were diluted to a concentration of 100pM. Then, with Ion Chef (Thermo Fisher Scientific) and Ion 520 & Ion 530 Kit-Chef and Ion 530 ™ Chip Kit (Thermo Fisher Scientific), emPCR was performed, enrichment and two 530 chips were loaded (16-24 samples per chip, cov x1000).

Sequencing

Sequencing was performed on an Ion S5 sequencer (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Bioinformatic analysis

The raw data generated during sequencing was processed using the Torrent Server Suite 5.6(TSS) (Thermo Scientific, USA). The obtained sequences were matched (mapped) to the reference sequence of the human genome (hg19). Searching for different variants (single nucleotide polymorphism [SNP], deletions, insertions) was carried out using the Variant Caller 5.6 program which is a part of Torrent Server Suite 5.6. The following basic parameters of the variants were used: minimum allele frequency – SNP = 0.01 / INDEL = 0.05, minimal quality – 10, minimal sequencing depth – 10. Variant Caller is compatible with the IGV genomic browser – Integrative Genomics Viewer (Broad Institute), which enables fast visualization of sophisticated variants. To annotate the detected variants with the TSS, the wANNOVAR software (www.wannovar.usc.edu) was used. Additionally, Torrent Server Suite 5.6 generated FASTQ files that were used for analysis using the CLC Biomedical Workbench 5.0 (Qiagen). The basic parameters used in the analysis using CLC software were: minimum allele frequency – 0.01, minimal quality – 10, minimal sequencing depth – 100. Detected mutations, SNP, insertions and deletions of the coding regions of the analyzed genes were filtered to detect pathogenic mutations by COSMIC base, dbSNP database (to discard hereditary polymorphisms) and population base of the 1000GENOMES project. Only variants with minimal 5% allelic frequency were reported.

Classification of pathogenicity of mutations.

Detected mutations were classified based on the information deposited in the ClinVar database and according to the American the College of Medical Genetics and Genomics (ACMG) recommendations [15]. In addition, in the case of variant of unknown significance or conflicting results we have performed in silico analysis using Varsome (https://varsome.com/) which integrates useful algorithms and databases, frequency in the populations etc. and literature search [16, 17].

Sanger Sequencing

Thanks to the close cooperation with our Genetic Clinic, mutations detected in tumor tissue were verified (germline vs. somatic) using DNA isolated from blood sample when it was available from the same patient. The Sanger Sequencing reaction used PCR amplification products that were purified using 10 U of exonuclease I (EN 0582) and 1 U of phosphatase Fast-AP (EF 0651) (both from ThermoFisher Scientific, Waltham, MA). The reaction was incubated for 15 min at 37°C, followed by 20 min at 80°C. Sequencing reactions were performed using forward and reverse sequence-specific primers and the ABI PRISM Big Dye Terminator Kit, version 3.1 (catalogue number 4337450, Applied Biosystems/ThermoFisher Scientific), according to the manufacturer’s instructions. The sequencing results were analyzed using the 3130 Capillary Sequencer (Applied Biosystems/ThermoFisher Scientific). The generated sequences were compared to the reference sequence using the NCBI Blast Nucleotide program.

Statistical analysis

Categorical data were summarized by frequencies and percentages and compared by Fisher’s exact test. Numerical variables were presented as mean and standard deviations as well as median and quartiles (Q1-Q3); differences between groups were analyzed (due to non-normality) by Mann-Whitney U-test. A two-tailed p-value < 0.05 was considered as statistically significant. All statistical analyses were performed using R software (version 3.1.2; The R Foundation for Statistical Computing, Vienna, Austria).

Results

NGS sequencing of BRCA1 and BRCA2 was successful in 98% (197/201) of the cases studied, while in four cases (2%) good quality DNA sequencing results were not obtained (Fig. 1). In 36 (18%) tissue samples, pathogenic mutations were detected in the BRCA1 gene and in 13 (6%) – in the BRCA2 gene. In total, pathogenic mutations in both genes were detected in 24% (49/201) of the ovarian cancer cases tested (Fig. 1, Table II). In 11 (11/49; 22%) cases, pathogenic mutations in BRCA1: 4x p.Cys61Gly, 7x p.Gln1756ProfsTer74, which are well known as a founder for the Polish population, were detected (Table II). In addition, in 4 (4/49; 8%) cases we detected a pathogenic mutation (BRCA1: p.Arg1751Ter) which was recognized as recurrent in the Polish population. The above-mentioned mutations were most often detected mutations in our cohort. Due to close cooperation with our Genetic Clinic, for 41 patients results of testing of DNA isolated from blood sample were available. Thirty five (17%; 35/201) mutation were germline origin, whereas 6 (3%; 6/201) mutations were somatic. We were unable to reveal germline vs. somatic status in 8 (4%; 8/201) patients (Fig. 2). However, in this group of unknown status two (1%; 2/201) mutations (BRCA1: one p.Cys61Gly and one p.Gln1756ProfsTer74) are founder for Polish population.
Variants of unknown significance (VUS) were detected in 12 (6%) tissue samples (Table III). In the remaining 136 (68%) cancer tissue samples no mutations in the BRCA1/2 genes were detected. Considering the histopathological type of ovarian cancer, pathogenic mutations in BRCA1/2 were detected in 29% (45/155) cases of serous ovarian cancer. In the serous HG subtype, 32% (39/121) of pathogenic mutations were detected and in the serous LG ovarian cancer, pathogenic mutations were detected in 18% (6/33) of cases. In endometroid ovarian cancer, in 10% (2/21) of cases pathogenic mutations in BRCA1/2 were detected. In clear cell ovarian carcinoma, in 1 out of 9 cases pathogenic mutation was detected and in 1 out of 4 cases of mixed ovarian (serous and endometroid) cancer. In the mucinous and undifferentiated subtypes mutations were not detected (Table IV).
The analysis indicated that patients with pathogenic mutations in ovarian cancer were on average 3 years younger compared to patients without mutations (p = 0.041). The presence of pathogenic mutation in BRCA1/2 correlated with the onset of serous histological type (p = 0.003). There is no relationship between the presence of pathogenic BRCA1/2 mutation and FIGO stage (p = 0.95; Table V). In the subgroup with only serous histology (n = 151), pathogenic mutations in BRCA1/2 were detected in patients on average 4 years younger than patients without mutation. In the case of serous subtypes and FIGO stage no statistically significant correlation was found (Table VI).

Discussion

As there are no reliable markers that would support early detection of ovarian cancer, around 70% of patients are diagnosed when the disease has already spread (FIGO stage III/IV). A well established and optimal therapeutic approach usually consists of extensive cytoreductive surgery followed by chemotherapy regimen based on platinum salts [3]. In 2014 this armamentarium was enriched by the advent of PARP inhibitors with indication for use in patients with somatic or germline mutation in BRCA1 & BRCA2 genes [18]. Recently published phase III study of Olaparib as maintenance therapy in newly diagnosed ovarian cancer patients with BRCA mutation showed a 70% reduction in the risk of disease progression or death [19]. Therefore, it has become of utmost importance to utilize the most efficient testing tools in order to maximize the pool of patients who could benefit from this treatment. As sequencing of DNA derived from tumor specimen allows to detect both germline and somatic mutations, NGS-based analysis of FFPE ovarian cancer samples becomes the standard of care [14].
NGS-based analysis of large genes such as BRCA1 and BRCA2 from FFPE sample has proved to be more challenging than similar testing performed on blood-derived DNA. First of all, formalin fixation modifies bases in the DNA which are the source of artefacts that can blur identification of pathogenic variants. Additionally, formalin-based fixation should be performed strictly to the protocol as DNA is prone to break and shear in spots of formaldehyde-introduce crosslinks. Overfixation of sample is a major reason of excessive fragmentation of DNA which is not compatible with downstream analysis steps, such as library generation [20, 21].
In the current study, with in-house derived and processed material, only 2% of samples gave inconclusive test result due to insufficient quality of DNA which is lower than published from other centers [22]. We attribute this low testing failure rate for FFPE samples to an excellent crossdisciplinary cooperation model between surgery unit, pathology and molecular diagnostics laboratory we have set up in our center to ensure optimal pipeline of sample acquisition, sample fixation & processing and downstream molecular analysis.
Secondly, the availability of PARP inhibitors (PARPi), now also in newly diagnosed ovarian cancer patients, has remodelled BRCA testing referral pattern. In order to ensure that BRCA1, BRCA2 genes status is known in time for first line treatment decisions, patients should be referred for molecular testing at the time of cytoreductive surgery by the operating surgeon. In our center, a vast majority of patients are automatically referred for testing upon debulking surgery and confirmation of the diagnosis by the pathologist. This ensures on-time identification of patients qualifying for PARPi treatment and allows to minimize the number of tests performed on archival FFPE samples that could be subject to degradation [23].
In this study we identified pathogenic BRCA mutation in 24% of unselected ovarian cancer cases. Among patients with serous subtype, mutation was found in 29% of samples. These results are in line with similar analyses of smaller groups performed in the Polish population (28%) by other investigators, as well as studies performed by Italian or Chinese groups, 28% and 26%, respectively [13, 24, 25].
Many countries with strong founder effect in BRCA1 & BRCA2 mutation epidemiology such as Poland have originally set up their testing system on a panel of most frequent founder loci analysis to optimize cost effectiveness. In this study we show that in our region only 30% (16/49) of identified mutations belong to the founder/recurrent spectrum, further reinforcing the need to shift from founder to NGS-based approach (Table II) [9]. Thanks to cooperation with the Genetic Counselling Unit, for 41 (84%) of identified patients it was possible to determine somatic vs. germline mutation status and provide extensive genetic counselling. The analysis showed that 17% of the mutations were germinal and only 3% were somatic.
The frequency of germline mutations (17%) detected in this work is similar to those reported in the literature (14-23%) and could be higher if we verified the status (germinal vs. somatic) in 8 (4%) cases (Fig. 2) [6, 10, 12]. According to published data, prevalence of somatic BRCA1 and BRCA2 mutations varies between 3% and 10%. Our results (3%) fall almost below this range most likely due to the sample size and inability to verify somatic vs. germline status for 8 (4%) patients in the cohort. Additionally, in our bioinformatic pipeline we apply a minimum 5% allelic frequency cut-off which might have contributed to lower detection of somatic variants.
In subtypes other than serous pathogenic BRCA1 and BRCA2 variants were also detected but with significantly lower frequencies, this observation is in line with available data that points towards mutations in other genes (PIK3CA, PTEN, CTNNB1, ARIDA1, KRAS) as more frequent aberrations in non-serous ovarian cancer [26].
Age analysis revealed that ovarian cancer patients with BRCA1, BRCA2 mutations are on average 3 years younger than patients without pathogenic variant and mutations more often occur in the serous type in comparison to other histological types. As previously observed, we also did not observe any effect of FIGO stage between the groups [5]. However, in the group of serous carcinomas (155) only age at diagnosis remained a parameter that differentiated cancers with pathogenic mutation in BRCA1/2 in comparison to non-mutant cancers (p = 0.018).
To further optimize identification of ovarian cancer patients with BRCA1/2 mutations we plan to extend the molecular analysis of large rearrangements that may be found in up to 5% of Polish population [27]. Additionally, we also consider to introduce a broader gene panel for ovarian cancer patients that would include, apart from BRCA1 and BRCA2, a number of other genes involved in homologous recombination-mediated repair of DNA double strand breaks that have been linked with familial ovarian cancer, with which we may detect additional 6% of carriers [12].

Conclusion

The availability of efficient, newly developed PARP inhibitors in BRCA-mutated OC patients and NGS technologies becoming an affordable testing tool, have driven a dynamic shift in the way BRCA1/BRCA2 mutation testing is performed. This paradigm change from test-to-prevent to test to-to-treat implied some important challenges in the inter-disciplinary OC patients care. From the experience of our center, we identified the following key touch points: 1) need of efficient communication between the pathologist and the surgeon performing debulking surgery that ensures correct acquisition and fixation of adequate, representative cancer specimen for histopathology report and that is compatible with downstream molecular analysis; 2) robust cooperation between molecular laboratory and pathologists securing on-time and quality sample processing; 3) working out format of test result report that meets expectations of oncologists to support informed therapeutic decisions and, lastly; 4) strict cooperation between molecular laboratory and Genetic Counselling Unit that would ensure all OC patients to receive genetic counselling and, if needed, further molecular analysis to determine germline or somatic background of detected variants. 
The authors declare no conflict of interest.
The study was supported by the Holycross Cancer Centre.

References

1. Torre LA, Bray F, Siegel RL, et al. Global cancer statistics, 2012. CA Cancer J Clin. 2015; 65: 87-108.
2. Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, et al. Cancer incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 2013; 49: 1374-1403.
3. Narod S. Can advanced-stage ovarian cancer be cured? Nat Rev Clin Oncol 2016; 13: 255-261.
4. Ledermann J, Harter P, Gourley C, et al. Olaparib maintenance therapy in patients with platinum-sensitive relapsed serous ovarian cancer: a preplanned retrospective analysis of outcomes by BRCA status in a randomised phase 2 trial. Lancet Oncol 2014; 15: 852-861.
5. Alsop K, Fereday S, Meldrum C, et al. BRCA mutation frequency and patterns of treatment response in BRCA mutation-positive women with ovarian cancer: a report from the Australian Ovarian Cancer Study Group. J Clin Oncol 2012; 30: 2654-2663.
6. Pennington KP, Walsh T, Harrell MI, et al. Germline and somatic mutations in homologous recombination genes predict platinum response and survival in ovarian, fallopian tube, and peritoneal carcinomas. Clin Cancer Res 2014; 20: 764-775.
7. Górski B, Jakubowska A, Huzarski T, et al. A high proportion of founder BRCA1 mutations in Polish breast cancer families. Int J Cancer 2004; 110: 683-686.
8. Szwiec M, Jakubowska A, Górski B, et al. Recurrent mutations of BRCA1 and BRCA2 in Poland: an update. Clin Genet 2015; 87: 288-292.
9. Kowalik A, Siolek M, Kopczynski J, et al. BRCA1 founder mutations and beyond in the Polish population: A single-institution BRCA1/2 next-generation sequencing study. PLoS One 2018; 13: e0201086.
10. Ratajska M, Krygier M, Stukan M, et al. Mutational analysis of BRCA1/2 in a group of 134 consecutive ovarian cancer patients. Novel and recurrent BRCA1/2 alterations detected by next generation sequencing. J Appl Genet 2015; 56: 193-198.
11. Wojcik P, Jasiowka M, Strycharz E, et al. Recurrent mutations of BRCA1, BRCA2 and PALB2 in the population of breast and ovarian cancer patients in Southern Poland. Hered Cancer Clin Pract 2016; 14: 5.
12. Koczkowska M, Krawczynska N, Stukan M, et al. Spectrum and Prevalence of Pathogenic Variants in Ovarian Cancer Susceptibility Genes in a Group of 333 Patients. Cancers (Basel) 2018; 10: pii: E442.
13. Koczkowska M, Zuk M, Gorczynski A, et al. Detection of somatic BRCA1/2 mutations in ovarian cancer – next-generation sequencing analysis of 100 cases. Cancer Med 2016; 5: 1640-1646.
14. Weren RD, Mensenkamp AR, Simons M, et al. Novel BRCA1 and BRCA2 Tumor Test as Basis for Treatment Decisions and Referral for Genetic Counselling of Patients with Ovarian Carcinomas. Hum Mutat 2017; 38: 226-235.
15. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med 2015; 17: 405-424.
16. Eccles DM, Mitchell G, Monteiro AN, et al. BRCA1 and BRCA2 genetic testing-pitfalls and recommendations for managing variants of uncertain clinical significance. Ann Oncol 2015; 26: 2057-2065.
17. Kopanos C, Tsiolkas V, Kouris A, et al. VarSome: The Human Genomic Variant Search Engine. Bioinformatics 2018; doi: 10.1093/bioinformatics/bty897.
18. Dougherty BA, Lai Z, Hodgson DR, et al. Biological and clinical evidence for somatic mutations in BRCA1 and BRCA2 as predictive markers for olaparib response in high-grade serous ovarian cancers in the maintenance setting. Oncotarget 2017; 8: 43653-43661.
19. Moore K, Colombo N, Scambia G, et al. Maintenance Olaparib in Patients with Newly Diagnosed Advanced Ovarian Cancer. N Engl J Med 2018; 379: 2495-2505.
20. Do H, Dobrovic A. Sequence artifacts in DNA from formalin-fixed tissues: causes and strategies for minimization. Clin Chem 2015; 61: 64-71.
21. Ellison G, Huang S, Carr H, et al. A reliable method for the detection of BRCA1 and BRCA2 mutations in fixed tumour tissue utilising multiplex PCR-based targeted next generation sequencing. BMC Clin Pathol 2015; 15: 5.
22. Ellison G, Ahdesmaki M, Luke S, et al. An evaluation of the challenges to developing tumor BRCA1 and BRCA2 testing methodologies for clinical practice. Hum Mutat 2018; 39: 394-405.
23. Ledermann J, Harter P, Gourley C, et al. Olaparib maintenance therapy in patients with platinum-sensitive relapsed serous ovarian cancer: a preplanned retrospective analysis of outcomes by BRCA status in a randomised phase 2 trial. Lancet Oncol 2014; 15: 852-861.
24. Chao A, Chang TC, Lapke N, et al. Prevalence and clinical significance of BRCA1/2 germline and somatic mutations in Taiwanese patients with ovarian cancer. Oncotarget 2016; 7: 85529-85541.
25. Mafficini A, Simbolo M, Parisi A, et al. BRCA somatic and germline mutation detection in paraffin embedded ovarian cancers by next-generation sequencing. Oncotarget 2016; 7: 1076-1083.
26. Teer JK, Yoder S, Gjyshi A, et al. Mutational heterogeneity in non-serous ovarian cancers. Sci Rep 2017; 7: 9728.
27. Ratajska M, Brozek I, Senkus-Konefka E, et al. BRCA1 and BRCA2 point mutations and large rearrangements in breast and ovarian cancer families in Northern Poland. Oncol Rep 2008; 19: 263-268.