If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Human Genetics, Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, Nijmegen, the NetherlandsDepartment of Pathology, Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, Nijmegen, the Netherlands
Probability of finding germline BRCA1/2 PVs varies widely among histological subtypes of ovarian carcinoma (OC).
•
Germline BRCA1/2 PVs are most frequently detected in high-grade serous OC patients.
•
Limiting testing to high-grade serous histology will be insufficient to identify all OC patients with germline BRCA1/2 PVs.
Abstract
Background
Histology restricted genetic predisposition testing of ovarian carcinoma patients is a topic of debate as the prevalence of BRCA1/2 pathogenic variants (PVs) in various histological subtypes is ambiguous. Our primary aim was to investigate the proportion of germline BRCA1/2 PVs per histological subtype. Additionally, we evaluated (i) proportion of somatic BRCA1/2 PVs and (ii) proportion of germline PVs in other ovarian carcinoma risk genes.
Methods
PubMed, EMBASE and Web of Science were systematically searched and we included all studies reporting germline BRCA1/2 PVs per histological subtype. Pooled proportions were calculated using a random-effects meta-analysis model. Subsets of studies were used for secondary analyses.
Results
Twenty-eight studies were identified. The overall estimated proportion of germline BRCA1/2 PVs was 16.8% (95% CI 14.6 to 19.2). Presence differed substantially among patients with varying histological subtypes of OC; proportions being highest in high-grade serous (22.2%, 95% CI 19.6 to 25.0) and lowest in clear cell (3.0%, 95% CI 1.6 to 5.6) and mucinous (2.5%, 95% CI 0.6 to 9.6) carcinomas. Somatic BRCA1/2 PVs were present with total estimated proportion of 6.0% (95% CI 5.0 to 7.3), based on a smaller subset of studies. Germline PVs in BRIP1, RAD51C, RAD51D, PALB2, and ATM were present in approximately 3%, based on a subset of nine studies.
Conclusion
Germline BRCA1/2 PVs are most frequently identified in high-grade serous ovarian carcinoma patients, but are also detected in patients having ovarian carcinomas of other histological subtypes. Limiting genetic predisposition testing to high-grade serous ovarian carcinoma patients will likely be insufficient to identify all patients with a germline PV.
Recognition of heredity in ovarian carcinoma (OC) patients is crucial to reduce cancer risks among patients and family members and it may facilitate treatment decisions. About 20–25% of all OCs are caused by an underlying heritable tumor risk syndrome [
]. Germline PVs in RAD51C, RAD51D, BRIP1, PALB2, ATM, and the mismatch repair genes (MLH1, MSH2, MSH6, PMS2) also confer a moderately increased risk for OC [
Large-scale meta-analysis of mutations identified in panels of breast/ovarian cancer-related genes - providing evidence of cancer predisposition genes.
]. Tumor testing detects both germline PVs and somatic PVs (present in tumor DNA but absent in blood). The increasing importance of tumor DNA testing is underlined by developments in treatment options. For example, poly ADP ribose polymerase inhibitor (PARPi) therapy has proven to be beneficial for patients with a tumor BRCA1/2 PV [
], either germline or somatic. Tumor DNA testing (i.e. a Tumor-First approach) detects individuals eligible for treatments options, and can simultaneously function as a prescreen to tailor genetic counseling and germline testing to patients at higher risk [
]. However, OC is a heterogeneous disease and histological subtypes display varying molecular genetic landscapes and distinct precancerous lesions. Selection of histological subtypes for germline testing and tumor DNA testing (as prescreen) to reduce costs and optimize recognition of hereditary OC is still a topic of debate. Former studies and reviews have demonstrated that high-grade serous OC is the cancer associated with germline BRCA1/2 PVs [
]. This is supported by the detection of serous tubal intraepithelial carcinomas (STICs) during prophylactic risk reducing salphingo-oophorectomy in individuals with a BRCA1/2 PV [
]. This association with distinct histology raises the question whether genetic predisposition testing could be executed more efficiently by restricting testing to certain histological subtypes of OC.
Histology restricted genetic predisposition testing is highly dependent on the accuracy of histology typing and the proportion of PVs detected per histological subtype. In 2014 and 2020, the World Health Organization (WHO) published new criteria for histological subclassification of OC [
]. Therefore, the proportion of PVs per histological subtype needs re-evaluation. Here, we performed a systematic review and meta-analysis of recent literature (>2015) with the primary aim to investigate the proportion of germline BRCA1/2 PVs per histological subtype of OC. Secondarily, we evaluated (i) proportion and histology of somatic BRCA1/2 PVs and (ii) proportion of germline PVs in moderate risk genes for OC (BRIP1, RAD51C, RAD51D, ATM, PALB2, MLH1, MSH2, MSH6, PMS2) and OC histology.
2. Methods
This systematic review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [
Databases PubMed/Medline, EMBASE and Web of Science were systematically searched for studies published from 1 January 2015 to 5 November 2020. A comprehensive search strategy was constructed using medical subject headings (MeSH), Embase subject headings (Emtree), keywords and synonyms related to three aspects: (i) ovarian carcinoma, (ii) BRCA, and (iii) germline/tumor testing. The complete search strategies are provided in supplement S1. Searches were restricted to English language and timeframe of publication 2015 (after introduction of WHO 2014 histology classification system) till “current”. All references were uploaded in Endnote reference management program (Endnote™ X9). Manual removal of duplicates and selection of articles was performed by two reviewers (VW and MvB) independently, achieving agreement after discussion or by consultation of a third reviewer (NH).
2.2 Eligibility criteria
Selection of articles was performed according to predefined inclusion and exclusion criteria. Articles were included if all information required for computing the prevalence of germline BRCA1/2 PVs per histological subtype of OC was provided. Germline BRCA1/2 PVs were defined as class 4 and 5 variants, and OC was defined by the WHO 2014 and 2020 guidelines [
]. Articles were excluded when the population did not consist of OC patients, when the number of OC patients was unclear, when no germline testing was performed, when testing was restricted to pre-specified (founder) mutations, or when the information on histology was insufficient to compute proportions per subtype. Solely articles written in English language and investigating human subjects were included. In case of overlapping cohorts, only the study with most patients was included. Review articles, case-reports, opinion pieces and letters to editors were excluded, similar to conference abstracts.
2.3 Critical appraisal
The quality of selected studies was rated using an adapted version of the critical appraisal tool for prevalence studies from the Joanne Briggs Institute [
]. The standard appraisal tool consisting of nine categories was adapted to enable scoring specifically for this systematic review. The adapted version is provided in supplement S2. Here, six (out of 13) items were considered to be essential (‘answered with yes’) to be included in the quantitative analysis: 1) sample frame broader than serous ovarian carcinomas, 2) total population size >50, 3) serous histology subdivided in high- and low-grade, 5) histological subgroup ‘non- high-grade serous’ specified, 5) germline and somatic PVs are distinguishable, and 6) variants of uncertain significance (VUS) and pathogenic variants are distinguishable. Scoring was performed by two reviewers (VW and MvB) independently, achieving agreement after discussion. The total critical appraisal score was the number of items answered with ‘yes’, which had no further consequences.
2.4 Outcomes and data-extraction
Our primary outcome was defined as the proportion of germline BRCA1/2 PVs per histological subtype of OC. Our secondary outcomes were: (i) proportion of somatic BRCA1/2 PVs and (ii) proportion of germline PVs in other risk genes for OC (BRIP1, RAD51C, RAD51D, PALB2, ATM, MLH1, MSH2, MSH6, PMS2). We defined somatic BRCA1/2 PVs as variants that are present in tumor DNA but absent in normal tissue or blood. We extracted data from eligible studies using a data extraction table consisting of predefined topics: bibliographical data, population data, methodological data, and outcome data. Data on histology of tested population and data on histology of BRCA1/2 positive cases were essential items in data extraction. We recorded whether tumor testing was performed in addition to germline testing and which genes other than BRCA1/2 were tested. Data extraction was split between two reviewers, who cross-checked each other's work. In case any discrepancies in original articles were identified during data extraction, we considered data from tables to be most reliable.
2.5 Data analysis and statistics
We performed meta-analyses of the proportion of germline and somatic BRCA1/2 PVs in all OCs. Additionally, we performed meta-analyses of the proportion of germline BRCA1/2 PVs per histological subtype of OC: high-grade serous (HGS), endometrioid, clear cell, low-grade serous (LGS), mucinous, carcinosarcoma, and ‘other’. The group ‘other’ was a merge of the histological types seromucinous, transitional cell, Brenner, undifferentiated, mixed, and other. We did not calculate an average proportion for (ovarian) carcinoma not specified, adenocarcinoma not specified and serous carcinoma not specified. Data analysis was performed at study level.
Pooled proportions were calculated by a random intercepts logistic regression model (GLMM) using a maximum likelihood estimation (ML) [
]. Heterogeneity across studies was estimated using the I2 statistic (<25% low level of heterogeneity, 50% moderate level of heterogeneity, >75% high level of heterogeneity) [
]. Subgroup analysis was undertaken based on ethnicity (country where study was performed) to assess potential differences.
Data was examined for the presence of outliers, defined as studies in which the individual confidence interval (CI) did not overlap with the meta-analysis CI. Outliers were not excluded from analyses, but these articles were screened for potential reasons for variation. In addition, the data was examined for influential studies, defined as studies for which exclusion leads to changes in result of the meta-analysis [
]. Also, removing studies one-by-one was performed and the effect on pooled proportion was evaluated to assess sensitivity of the meta-analysis model.
All analyses were conducted using statistical software R version 3.6.2 (2019-12-12) using the packages “meta” and “metafor”.
3. Results
3.1 Study selection
Database searches generated a total of 4756 records, of which 2941 remained after removal of duplicates. Exclusion based on screening of title and abstracts (n = 2708) and full text (n = 135) resulted in 98 articles who were subjected to a critical appraisal. Then, another 69 articles were excluded which resulted in a total of 29 articles. During data extraction two articles were merged as they presented results of the same cohort [
]. The critical appraisal scores of the included articles are presented in supplement S3. In total, this systematic review and meta-analysis is based on 28 studies (Fig. 1).
3.2 Characteristics of the included studies
Table 1 presents an overview of the study characteristics and main outcome data of the included studies. The included studies were conducted either in Asian or European countries, with the exception of one study which was conducted in the United States [
]. Details on selected OC patients are also provided; most studies included all OC patients, with mucinous ovarian carcinoma being the predominant exclusion criterium. The number of included OC patients in individual studies ranged from 56 to 1915 (patients with known germline or somatic mutation status and known histology of OC). In total, we include 11,351 OC patients from 28 studies. The individual study results on total number of BRCA1/2 PVs and BRCA1/2 PVs per histological subtype of OC are presented in Table 1. Table 1 also marks that nine studies performed somatic tumor testing in addition to germline testing, this subset was used to analyze the proportion and histology of somatic BRCA1/2 PVs. Another subset of nine studies could be used to evaluate proportion and histology of other risk genes for OC as they tested for these in addition to BRCA1/2.
Fig. 1Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart of study selection. After several rounds of selection, 28 articles remained from the initial database searches.
Number of patients with BRCA PVs in total population and per histological subtype: HGS, E, CC, LGS, M, CS, other. Number of patients with BRCA PVs (n) per number of tumors tested (N). ‘other’ is combined group from: seromucinous, transitional cell, undifferentiated, mixed and other. — indicates histological subtype is not present in this cohort. Excluded from this table are: (ovarian) carcinoma not specified, serous not specified and borderline tumors.
Test: only germline testing (G), germline and tumor testing (B), or tumor testing and subsequent germline testing when test was positive (S). *marks germline testing not performed for all patients with positive tumor test, these patients were excluded from our analysis.
Prevalence of BRCA1 and BRCA2 mutations in patients with primary ovarian cancer - does the German checklist for detecting the risk of hereditary breast and ovarian cancer adequately depict the need for consultation?.
The first Japanese nationwide multicenter study of BRCA mutation testing in ovarian cancer: CHARacterizing the cross-sectionaL approach to ovarian cancer geneTic TEsting of BRCA (CHARLOTTE).
Germline and somatic BRCA1/2 gene mutational status and clinical outcomes in epithelial peritoneal, ovarian, and fallopian tube cancer: over a decade of experience in a single institution in Korea.
Implementing NGS-based <em>BRCA</em> tumour tissue testing in FFPE ovarian carcinoma specimens: hints from a real-life experience within the framework of expert recommendations.
BRCA1 and BRCA2 mutations in ovarian cancer patients from China: ethnic-related mutations in BRCA1 associated with an increased risk of ovarian cancer.
Abbreviations: UK: United Kingdom, USA: United States of America, OC: ovarian carcinoma, MGP: multi-gene panel, MA: meta-analysis, PV: pathogenic variant, g: germline, s: somatic, HGS: high-grade serous, (HG)E: (high-grade) endometrioid, CC: clear cell, LGS: low-grade serous, M: mucinous, CS: carcinosarcoma.
a Test: only germline testing (G), germline and tumor testing (B), or tumor testing and subsequent germline testing when test was positive (S). *marks germline testing not performed for all patients with positive tumor test, these patients were excluded from our analysis.
b Number of patients with BRCA PVs in total population and per histological subtype: HGS, E, CC, LGS, M, CS, other. Number of patients with BRCA PVs (n) per number of tumors tested (N). ‘other’ is combined group from: seromucinous, transitional cell, undifferentiated, mixed and other. — indicates histological subtype is not present in this cohort. Excluded from this table are: (ovarian) carcinoma not specified, serous not specified and borderline tumors.
3.3 Proportion of germline BRCA1/2 PVs per histological subtype
Meta-analysis of 28 studies resulted in an estimated proportion of 16.8% (95% CI 14.6 to 19.2) for germline BRCA1/2 PVs in a population of OC patients as is illustrated in Fig. 2A. Considerable (high level) heterogeneity was present (I2 = 88%), representing differences in results among studies. Subgroup analysis for studies conducted in Asian versus European/American countries (predominantly Asian versus predominantly Caucasian women) revealed no differences in proportions and heterogeneity between these two groups (supplement S4). More specifically, the estimated proportion of germline BRCA1 PVs was 10.7% (95% CI 8.8 to 12.9, I2 = 88%) and the proportion of BRCA2 PVs was 5.5% (95% CI 4.7 to 6.3, I2 = 51%).
Fig. 3A presents a pie chart of the OC histological subtypes of all patients and Fig. 3B presents a pie chart of the OC histological subtypes of women with BRCA1/2 PVs. It appears that OC patients with a germline BRCA1/2 PV are relatively more likely to develop HGSOC compared to the general OC population in this meta-analysis: in women with BRCA1/2 PVs this percentage is 91% (1738 / 1907), whereas around 75% (7914 / 10,487) of all OCs are of HGS histology.
Fig. 2Forest plots presenting meta-analyses of proportions of germline (A) and somatic (B) BRCA1/2 PVs in all histological subtypes of ovarian carcinoma combined. Somatic BRCA1/2 PVs are defined as those that are present in tumor DNA but absent in normal tissue or blood.
Fig. 3Pie-charts representing the presence of histological subtypes in all ovarian carcinoma patients (A), patients with germline BRCA1/2 PV (B), patients with somatic BRCA1/2 PV (C), patients with germline PV in BRIP1, RAD51C, RAD51D, PALB2 and ATM (D) and patients with germline PV in MLH1, MSH2, MSH6, PMS2 (E). This excludes the presence of the following histological subtypes: ovarian carcinoma not specified, adenocarcinoma not specified, serous not specified, borderline & unknown (excluded: A n = 864, B n = 198, C n = 2, D n = 41, E n = 6).
Proportions of germline BRCA1/2 PVs per histological subtype of OC are presented in Table 2 and Fig. 4. Presence of germline BRCA1/2 PVs varied substantially among patients having various histological subtypes of OC. The proportion of germline BRCA1/2 PVs was highest in patients with HGSOC (22.2%, 95% CI 19.6 to 25.0), and also the proportion in ovarian carcinosarcoma patients was found to be relatively high (11.9%, 95% CI 5.8 to 22.6). The probability of detecting a germline BRCA1/2 PV was lower in patients having endometrioid OC (5.8%, 95% CI 3.3 to 9.9), LGSOC (5.2%, 95% CI 2.3 to 11.3), clear cell OC (3.0%, 95% CI 1.6 to 5.6) or mucinous OC (2.5%, 95% CI 0.6 to 9.6). Thus, for all OC patients, irrespective of the histological subtype, there is a probability that the patient is carrying a germline BRCA1/2 PV, but the number of patients needed to test to identify one BRCA1/2 PV vary substantially as is presented in Table 2.
Table 2Meta-analyses of proportion germline BRCA1/2 PVs per histological subtype of OC.
Fig. 4Meta-analysis of pooled proportion of germline BRCA1/2 PVs per histological subtype of ovarian carcinoma. The pooled proportion including 95%CI is presented, and individual study results.
Heterogeneity was low in all non-HGS histological subtypes, indicating here that the CIs of individual study estimates overlap, potentially due to rarity of these subtypes. We assessed the sensitivity of the meta-analysis by removing studies one-by-one and analyzing their influence on the pooled proportion. The pooled proportion of germline BRCA1/2 PVs in patients with non-HGSOC were more sensible to the effect of removing single studies compared to HGS. Therefore, the pooled proportion of germline BRCA1/2 PVs in patients with non-HGSOC are more uncertain, especially for carcinosarcoma, LGS, and mucinous OC. This uncertainty in the pooled proportion is also visible from the wider CIs of the pooled proportions. Despite this uncertainty in the estimate proportion, germline BRCA1/2 PVs were detected in all histological subtypes of OC. Also, in seven studies that incorporated an extra round of pathological revision in their study design, germline BRCA1/2 PVs were identified in all histological subtypes [
The first Japanese nationwide multicenter study of BRCA mutation testing in ovarian cancer: CHARacterizing the cross-sectionaL approach to ovarian cancer geneTic TEsting of BRCA (CHARLOTTE).
Implementing NGS-based <em>BRCA</em> tumour tissue testing in FFPE ovarian carcinoma specimens: hints from a real-life experience within the framework of expert recommendations.
BRCA1 and BRCA2 mutations in ovarian cancer patients from China: ethnic-related mutations in BRCA1 associated with an increased risk of ovarian cancer.
3.4 Proportion and histology of somatic BRCA1/2 PVs
Meta-analysis of a subset of nine studies that performed tumor testing in addition to germline testing indicated that the estimated proportion of somatic (non-germline) BRCA1/2 PVs in a population of OC patients was 6.0% (95% CI 5.0 to 7.3), as illustrated in Fig. 2B. Heterogeneity does not seem to play a role here (I2 = 0%), indicating no major differences among studies. Proportions of somatic BRCA1 and BRCA2 PVs were more or less similar; 3.5% (95% CI 2.7 to 4.5, I2 = 21%) and 2.7% (95% CI 2.0 to 3.6, I2 = 0%), respectively. Fig. 3C presents a pie-chart of the histological subtypes of tumors with a somatic BRCA1/2 PV. Tumors with a somatic BRCA1/2 PV are predominantly HGS and endometrioid carcinomas, and rarely clear cell, mucinous, carcinosarcoma and other carcinomas, more or less similar to the general OC population in our study.
3.5 Proportion and histology of germline variants in other ovarian carcinoma risk genes
Our systematic review included nine studies that investigated the prevalence of germline PVs in other OC risk genes (BRIP1, RAD51C, RAD51D, PALB2, ATM, MLH1, MSH2, MSH6, PMS2). Individual study results are presented in supplement S5. Rough estimates of the prevalence of germline PVs in these genes are: BRIP1 0.9% (42/4658), RAD51C 0.8% (44/5257), RAD51D 0.7% (34/5195), PALB2 0.6% (27/4658), ATM 0.3% (14/4658), MSH6 0.3% (14/4658), PMS2 0.2% (7/3538), MLH1 0.2% (7/4658), MSH2 0.2% (7/4658) (as presented in supplement S6). In total, the combined probability of detecting a germline PV in BRIP1, RAD51C, RAD51D, PALB2, or ATM is around 3.3% in OC patients. The probability of detecting a germline PV in a mismatch repair gene (MLH1, MSH2, MSH6 or PMS2) was <1% in total.
Fig. 3D and E present pie-charts that illustrate the OC histological subtypes of patients with a germline PV in the genes BRIP1, RAD51C, RAD51D, PALB2, ATM and the mismatch repair (MMR) genes, respectively. HGS was the predominant histological subtype for both groups of genes. Furthermore, germline PVs in the genes BRIP1, RAD51C, RAD51D, PALB2 and ATM were detected in all histological subtypes, whereas germline PVs in MMR genes were predominantly detected in HGS and endometrioid OCs.
4. Discussion
In this systematic review and meta-analysis, we have shown that germline BRCA1/2 PVs were detected in 16.8% (95% CI 14.8 to 19.2) of the patients with OC. The probability of detecting a germline BRCA1/2 PV varied widely among the various histological subtypes of OC; ranging from 22.2% (95% CI 19.6 to 25.0) in patients with HGSOC to 3.0% (95% CI 1.6 to 5.6) in patients with clear cell OC and 2.5% (95% CI 0.6 to 9.6) in patients with mucinous OC. Unlike the generally accepted assumption that BRCA1/2 PVs are exclusively related to HGS histology [
], our meta-analysis indicated that BRCA1/2 PVs are also found in all other histological subtypes of OC (endometrioid, clear cell, LGS, mucinous, carcinosarcoma and ‘other’). Therefore, limiting genetic predisposition testing to HGSOC patients will likely be insufficient to identify all patients with an underlying germline pathogenic variant.
The overall estimated probability of finding a germline BRCA1/2 PV in OC patients (16.8%, 95% CI 14.8 to 19.2) seemed somewhat higher than reported in a previous systematic review (12.7%, 95% CI 9.5 to 15.9) [
]. Importantly, proportions of germline BRCA1/2 PVs differed substantially among included studies. The differences could not be explained by ethnicity, since subgroup analysis revealed no difference between Asian (predominantly Asian ethnicity) and European/American countries (predominantly Caucasian ethnicity). This suggests differences are caused by more specific characteristics of the included populations. Outliers were screened and we identified potential explanations: variation in the presence of founder mutations [
]. Unfortunately, as there were no non-European/American/Asian studies identified during our search, we are uncertain if our results apply to other populations. Furthermore, germline BRCA1 PVs were more frequently observed than germline BRCA2 PVs, probably related to the overall lifetime risk of developing OC that is more than twice as high for women with a BRCA1 PV than for those with a BRCA2 PV [
]. Somatic (non-germline) BRCA1 and BRCA2 PVs were present in comparable proportions, detected with a total estimate proportion of approximately 6% in all histological subtypes.
Our result that BRCA1/2 PVs were detected in patients having endometrioid OC, clear cell OC, LGSOC, or mucinous OCs does not fit within the classical hypothesis of the origin of these carcinomas. Endometrioid and clear cell OCs are thought to develop from endometriosis [
]. The development of non-HGSOC in women with BRCA1/2 PV suggests another carcinogenic pathway, but precursor lesions of these histological subtypes are not known to be more frequently observed during prophylactic RRSO in women with BRCA1/2 PV. In addition, the identification of germline BRCA1/2 PVs in these patients does not necessarily indicate a causal relationship with tumor development. However because the incidence of BRCA1/2 PV is higher than in the general population, the results of our meta-analysis challenge the hypothesis that germline BRCA1/2 are exclusively related to HGSOC.
We should however take into account the potential risk of misclassification of histological subtypes and its influence on our results. Seven of our included studies had performed pathological revision, and in these specific studies BRCA1/2 PVs were identified in non-HGS OC as well. However, it is impossible to reliably evaluate quality of pathological assessment in a literature review, as this measure is partly subjective depending on experience of pathologist and histological criteria used which are both seldomly reported. We included studies published after 2015 as the WHO2014 OC histological classification system is demonstrated to be more reproducible compared to previous systems [
], but patient inclusion and pathological assessment were sometimes performed before 2014. In addition, studies rarely indicated which classification system was used and whether immunohistochemistry was used to support histologic classification. Therefore, when interpreting our results, one should consider that these resemble evidence from most recent literature, but they do not represent the most recent diagnostic criteria. Prospectively investigating the effect of assessing histology by experienced gynecological pathologists on the probability of detecting BRCA1/2 PVs in non-HGSOC would be essential in future research.
The presence of germline BRCA1/2 PVs in patients having non-HGSOC clearly requires additional investigation to elucidate true presence (and potential carcinogenic pathway) or misclassification. The rarity of some non-HGS histological subtypes will likely complicate this investigation. Our meta-analysis was limited by the rarity of some non-HGSOC as well. We demonstrated that BRCA1/2 PVs were detected the least frequent in mucinous OC. However, conclusions on this data should be interpreted with caution as several studies (8 out of 28) specifically excluded mucinous OC. Because of scarce data of some histological subtypes, we used a GLMM meta-analysis model to estimate all pooled proportions. This model has been recommended as good alternative for conventional two-step methods [
]. The sensitivity of our model was assessed by removing studies one-by-one, which demonstrated that the probability of detecting a germline BRCA1/2 PV was more uncertain in patients having non-HGSOC (specifically carcinosarcoma OC, LGSOC and mucinous OC) compared to patients having HGSOC. This is also visible from the wide confidence intervals. Future research on the presence of germline BRCA1/2 PVs in patients with non-HGSOC should consider their power and sample size to detect variants in these histological subtypes.
Additionally, as wider panels are expected to become increasingly common because of decreasing costs, assessment of the added clinical value of expanding the BRCA1/2 test panel to include moderate risk genes for OC is important. In comparison to BRCA1 and BRCA2, the detected proportions of germline PVs were considerably lower in BRIP1 (0.9%), RAD51C (0.8%), RAD51D (0.7%), PALB2 (0.6%), ATM (0.3%), MLH1 (0.2%), MSH2 (0.2%), MSH6 (0.3%), and PMS2 (0.2%). Noticeably, our systematic search did not focus on identification of these articles specifically. Previous studies have more elaborately investigated presence and cancer risks of germline PVs in these genes [
Large-scale meta-analysis of mutations identified in panels of breast/ovarian cancer-related genes — providing evidence of cancer predisposition genes.
]. Besides testing for BRCA1/2, we encourage testing all patients with OC for germline PVs in RAD51C, RAD51D, BRIP1 and PALB2, and where possible also ATM. However, testing all OC patients for MLH1, MSH2, MSH6 and PMS2 is debatable given the rare occurrences and the fact that testing for MLH1, MSH2, MSH6 and PMS2 can be restricted to those with additional reasons to do so, for example a family history with Lynch syndrome associated cancers [
In conclusion, germline BRCA1/2 PVs are being detected in all histological subtypes of OC, and most frequently in HGSOC. Limiting genetic predisposition testing to HGSOC will likely be inadequate to identify all patients with an underlying germline pathogenic variant. Future research (e.g. focusing on cost-effectiveness) might shed new light on the issue. However, based on current literature, we strongly encourage to test all OC patients for germline BRCA1/2 PVs, irrespective of their histological subtype. These considerations will contribute to optimize recognition of heredity in ovarian carcinoma patients.
Funding
This work was supported by a grant from the Dutch Cancer Society (KUN2019–12732).
Author contributions
All authors critically revised the manuscript and provided final approval of the version to be published. VW and MvB performed data collection and wrote the original draft of the manuscript. VW, MvB and JV were involved in analyzing the data. VW, MvB, ML, TB and NH contributed to the study design and data interpretation. NH supervised the study.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We thank Dorien Hermkens, Yvonne Smolders and Jozé Braspenning for useful discussions and the Radboudumc Medical library for assistance in establishing comprehensive search strategies.
Appendix A. Supplementary data
The following are the supplementary data related to this article.
Critical appraisal score list. An adjusted version of the Joanna Briggs critical appraisal list for prevalence studies. Six items (2, 4, 5a, 5b, 5c, 5d) were essential to be answered with ‘yes’ to be included in our meta-analysis. The other items were scored without any consequences.
Sub-group analysis based on ethnicity (country where study was conducted). Proportion of germline BRCA1/2 PVs in study countries with predominant Caucasian ethnicity and in study countries with predominant Asian ethnicity.
Number of pathogenic variants identified per study. Total number of ovarian carcinoma patients included in meta-analysis (N meta) and the number of pathogenic variants in various genes identified per study.
Proportion of germline PV in other ovarian carcinoma risk genes: genes involved in homologous recombination (HR) and mismatch repair (MMR), and histology of positive cases.
References
Norquist B.M.
et al.
Inherited mutations in women with ovarian carcinoma.
Large-scale meta-analysis of mutations identified in panels of breast/ovarian cancer-related genes - providing evidence of cancer predisposition genes.
Prevalence of BRCA1 and BRCA2 mutations in patients with primary ovarian cancer - does the German checklist for detecting the risk of hereditary breast and ovarian cancer adequately depict the need for consultation?.
The first Japanese nationwide multicenter study of BRCA mutation testing in ovarian cancer: CHARacterizing the cross-sectionaL approach to ovarian cancer geneTic TEsting of BRCA (CHARLOTTE).
Germline and somatic BRCA1/2 gene mutational status and clinical outcomes in epithelial peritoneal, ovarian, and fallopian tube cancer: over a decade of experience in a single institution in Korea.
Implementing NGS-based <em>BRCA</em> tumour tissue testing in FFPE ovarian carcinoma specimens: hints from a real-life experience within the framework of expert recommendations.
BRCA1 and BRCA2 mutations in ovarian cancer patients from China: ethnic-related mutations in BRCA1 associated with an increased risk of ovarian cancer.
Large-scale meta-analysis of mutations identified in panels of breast/ovarian cancer-related genes — providing evidence of cancer predisposition genes.
01516-X&id=gr1.jpg){kind=link}
01516-X&id=gr2.jpg){kind=link}
01516-X&id=gr3.jpg){kind=link}
01516-X&id=gr4.jpg){kind=link}