Joint analysis of germline genetic data from over 29,000 cases with suspected hereditary breast and ovarian cancer (HBOC) as part of the NASGE initiative

Abstract

As multigene panel testing is becoming routine in clinical care, there are recommendations at national and international level, as to which genes should be analyzed in the context of a hereditary breast and ovarian cancer (HBOC). However, the individual composition of gene panels offered by testing laboratories vary, resulting in a different variant diagnostic rate. Therefore, we performed a retrospective NGS dataset analysis of suspected HBOC patients who had been tested at different German diagnostic laboratories that are part of the NASGE network.

We collected 29,317 HBOC datasets and compared the diagnostic yield applying the most common panel recommendations and an internal HBOC gene panel. Additionally, we analyzed the data concerning other potential tumor risk syndromes (TRS) not caused by pathogenic variants in the core panel genes.

At least one pathogenic variant causative for an autosomal-dominant TRS was identified in 4235 datasets, resulting in an overall diagnostic yield of 14.4 %. The diagnostic yield of pathogenic variants varied depending on the applied HBOC panel (between 5 and 26 genes) from 9.0 % to 13.8 % with the internal HBOC panel having a yield of 12.7 %. Notably, in about 1 % of cases, a pathogenic variant outside the established HBOC core genes was identified, indicating the presence of other TRS.

These results are consistent with previous observations that a significant proportion of patients with HBOC predisposition were not detected by the guideline-based gene panels and suggest that expanded diagnostics compared to currently recommended multigene panels may identify additional patients at high risk for developing cancer.

Highlights

• •Expanded gene panel testing for hereditary cancer increases the diagnostic yield.
• •Disease specific gene panels overlook 1 % of patients with a high-risk TRS.
• •Joint data analysis of different German diagnostic laboratories.

1. Introduction

Worldwide, breast and ovarian cancer rank as the most common tumor types affecting women; however, only about 20 % of familial cases can be explained by variants in the primary risk genes, namely BRCA1 (MIM: 113705) and BRCA2 (MIM: 600185), associated with hereditary breast and ovarian cancer (HBOC; MIM: 114480, 167000) [1,2]. In recent decades, other genes with high and medium penetrance have been associated with HBOC beyond BRCA1 and BRCA2 [3,4]. The predisposition to HBOC can be broadly classified into three categories: 1) rare, highly penetrant variants in the genes BRCA1, BRCA2, PALB2 (MIM: 610355) and TP53 (MIM: 191170); 2) rare variants associated with lower penetrance, observed in genes such as CHEK2 (MIM: 604373) and ATM (MIM: 607585); and 3) variants that occur more frequently in the population but have only a small individual effect size (polygenic risk score - PRS) [5,6]. Despite significant advances in genetic diagnosis, the majority of suspected hereditary cases remain unexplained [7], owing to unknown pathomechanisms, familial aggregation by chance with a missing heritability, the use of overly restrictive gene panels or the misinterpretation of variants of uncertain significance (VUS) [8]. Given the accessibility of targeted therapies and screening programs for affected individuals and families, the identification of individuals carrying pathogenic variants in genes associated with HBOC is imperative [9,10].

Testing for cancer predisposition heavily relies on clinical criteria derived from national guidelines, selecting genes based on the patient's prior probability of harboring a germline alteration dictated by tumor type, age of onset, and family histories. In Germany, the German HBOC Consortium recommends analyzing a specific panel of genes as part of HBOC testing; internationally, entities such as ClinGen and the PanelApp from Genomics England suggest other genes, with only BRCA1, BRCA2, and PALB2 consistently included (summarized in Table A). The complexity is heightened by frequent changes in the content of recommended gene panels for HBOC over the past decade due to emerging gene-disease associations. In our prior study by Henkel et al., 2023 [11], we introduced an internal 14-gene HBOC gene panel and compared it with the national and diverse international recommendations regarding diagnostic yield, uncovering pathogenic variants outside the core set of genes.

In this study, we present the outcomes of a retrospective follow-up analysis on 29,317 next-generation sequencing (NGS) datasets of suspected HBOC patients tested at various German diagnostic laboratories within the NASGE network, an association of specialists in human genetics who are dedicated to rare genetic diseases (https://nasge.de/). Building upon the framework established in Henkel et al., 2023 [11], this work reinforces the advantages of analyzing a gene panel that is not overly stringent and emphasizes that the additional analysis of a comprehensive tumor panel leads to a higher diagnostic yield.

2. Methods

2.1. Patient cohort

We collected and analyzed 29,317 NGS datasets of individuals referred for HBOC testing in various German diagnostic laboratories that are part of the NASGE network (https://nasge.de/). Patients met the S3 [12] or AGO Guidelines (https://www.ago-online.de/ago-kommissionen/kommission-mamma) for HBOC testing in Germany. All individuals had genetic counseling and gave informed consent for genetic testing, which was done on a comprehensive cancer panel comprising 123 cancer predisposition genes, which includes all genes of the different panels virtually compared in this study. Patients provided their informed consent in accordance with the respective national ethical standards of Germany. The study was approved by the internal ethics committees of the participating institutions. Data processing and storage complied with the General Data Protection Regulation (GDPR).

2.2. Study design

The cohort was retrospectively analyzed as described by Henkel et al., 2023 [11]. In brief, the authors collected aggregated data from participating centers, specifically the number and type of variants identified in 123 cancer predisposition genes within their respective cohorts of individuals with suspected hereditary breast and ovarian cancer (HBOC). These data were analyzed to compare an internal HBOC gene panel with 14 genes with a national (German HBOC Consortium Panel) and various internationally recommended gene panels (German Hereditary Breast and Ovarian Cancer Consortium, ClinGen Gene Curation Expert Panel, Genomics England PanelsApp) in terms of diagnostic yield. For an overview of the gene panels and their gene content, see Table A.

3. Results

The comprehensive testing of 29,317 individuals revealed at least one variant classified following ACMG/AMP as variant of uncertain significance (VUS), likely pathogenic (LP) and pathogenic (P) in 34.9 % of cases, specifically comprising 4235 (14.4 %) LP/P variants and 5983 (20.4 %) VUS (Table 1).

Table 1.

Overview of the collected data set. VUS = ACMG/AMP variant of uncertain significance, pathogenic = ACMG/AMP (likely) pathogenic, n = 29,317, overview gene content = Table A ∗overall = Hereditary Cancer Syndromes – Comprehensive panel (123 genes).

	variant detection rate n	variant detection rate %	total %
Internal HBOC panel VUS	4.102	13.99 %	26.67 %
Internal HBOC panel pathogenic	3.718	12.68 %
overall VUS∗	5.983	20.41 %	34.85 %
overall pathogenic∗	4.235	14.44 %

In the context of different national and international HBOC gene panel recommendations (German HBOC Consortium, ClinGen Gene Curation Expert Panel, Genomics England PanelsApp), compared to the internal HBOC panel, the diagnostic yield (LP/P) varies from 9.0 % to 13.8 % (Fig. 1). Taking all variants into account (VUS/LP/P), the variant detection rate spans from 14.9 % (Genomics England PanelApp: Pertinent cancer susceptibility) to 30.8 % (Genomics England PanelApp: Familial Breast Cancer) (Fig. 1 and Table A).

Fig. 1.

Distribution of identified variants applying different gene panels: A all identified variants (variant of uncertain significance (VUS), likely pathogenic (LP) and pathogenic (P)) within the recommended HBOC gene panels and the comprehensive cancer panel. The (likely) pathogenic variants are shown as black bars, while the VUS are depicted as grey bars (see full gene list in Table A). B Descending distribution based on the number of pathogenic variants (n = 3718) in the internal HBOC gene set (n = 14).

The pathogenic variants within the internal HBOC panel (LP/P) were distributed according to Fig. 1B, with BRCA1 (n = 1193), BRCA2 (n = 1127), CHEK2 (n = 622), PALB2 (n = 242) and ATM (n = 221) being the top 5 genes. Notably, a LP/P variant in TP53 was identified in 36 individuals, representing a rate of 0.1 % in this German HBOC cohort. Concerning the distribution off VUS within the internal HBOC panel, ATM (n = 827) is followed by BRCA2 (n = 679), CHEK2 (n = 615), BRCA1 (n = 606) and PALB2 (n = 282) (Fig. 1B). A pathogenic variant outside the internal HBOC panel was identified 286 times (1 %) resulting in the diagnosis of autosomal-dominant TRS other than HBOC (Fig. 2). The most prevalent gene among these was MSH6 (n = 69 or 0.24 %), followed by MSH2 (n = 56 or 0.19 %), MLH1 (n = 35 or 0.12 %), PMS2 (n = 34 or 0.12 %) and APC (n = 14 or 0.05 %).

Fig. 2.

Distribution of variants across cancer-associated genes: identified (likely) pathogenic variants (n = 286, corresponding to 1 %) outside the HBOC core gene set associated with other autosomal-dominant TRS than HBOC.

4. Discussion

In this study, we report on a retrospective follow-up analysis of 29,317 NGS datasets of suspected HBOC patients tested at various German diagnostic laboratories within the NASGE network. While our prior publication by Henkel et al., 2023 [11] has already explored the comparison of different gene panels, albeit using a smaller data set, our key finding is that the application of a comprehensive gene panel for hereditary cancer testing significantly increases the diagnostic yield compared to guideline-based targeted testing and facilitates the identification of other TRS [11,[13], [14], [15]].

Nevertheless, the unclear clinical utility of pathogenic variants that are not related to the patient's primary disease is a frequently discussed concern associated with the use of expanded gene panels. We contend that knowledge about a TRS predisposition is crucial, allowing for the initiation of targeted surveillance examinations in the index patient and, if necessary, family members with an increased risk. However, it should be noted that tumor risks may not be accurately estimated when mutations in cancer predisposition genes, such as Lynch syndrome genes, are identified without supportive personal or family history. This option should be discussed with the patient, acknowledging that disease specific gene panels overlook 1 % (n = 286) of patients with a high-risk TRS. A limitation of this study is its multicenter nature, which not only precludes the discussion of the results at the variant level but also limits our ability to analyze individual clinical characteristics and testing indications, including specific index case presentations. Furthermore, variant classification, particularly for VUS, may vary between centers and change over time, which was not accounted for in this study. Nonetheless, our data is consistent with that of the previous work, providing valuable insights for diagnostic strategy optimization despite these limitations [11,16,17].

Furthermore, it is well known that the number of VUS detected increases proportionally to the number of genes analyzed. The burden of communicating a VUS finding to both clinicians and patients is widely accepted as a significant concern due to the associated uncertainty. We recognize that diagnostic laboratories use different policies, whether and which VUS should be included in the report. Since VUS should not be used for clinical decision making, it is important that surveillance should be based on family history. For this reason, efforts are being made to report only VUS that have multiple independent lines of evidence for pathogenicity (so-called ‘hot VUS') and, if possible, to re-evaluate these regularly as well as encourage the inclusion in research studies [18].

As others have demonstrated [11,[19], [20], [21]], regular reassessment of VUS, leveraging new data and the most recent Sequence Variant Interpretation (SVI) recommendations on the use of the ACMG/AMP classification guidelines, has the potential to dramatically enhance the clinical validity of genetic reports [9,10]. Another avenue for improvement regarding variant reassessment is the gene- or disease-specific adaptations of the ACMG codes by the Variant Curation Expert Panels (VCEPs, https://www.clinicalgenome.org/affiliation/vcep). Consequently, we anticipate a significant advancement of the variant interpretation process, leading to a substantial reduction in the number of VUS in the foreseeable future.

5. Conclusion

A significant proportion of patients with HBOC predisposition were not detected by the guideline-based gene panels. In light of HBOC testing being probably one of the most performed diagnostic tests around the world, expanded diagnostics compared to currently recommended smaller multigene panels may identify additional patients with TRS. With limited resources, it is recommendable to analyze the 14 genes of the here defined internal HBOC panel or the Genomics England PanelApp Familial breast cancer panel to improve diagnostic yield compared to other, mostly smaller panels. Whenever feasible analysis of a comprehensive cancer panel should be considered to maximize TRS identification.

CRediT authorship contribution statement

Jan Henkel: Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Andreas Laner: Writing – review & editing, Writing – original draft, Visualization, Validation, Formal analysis, Data curation. Melanie Locher: Formal analysis, Data curation. Tobias Wohlfrom: Formal analysis, Data curation. Birgit Neitzel: Formal analysis, Data curation. Kerstin Becker: Formal analysis, Data curation. Teresa Neuhann: Writing – review & editing, Supervision, Conceptualization. Angela Abicht: Writing – review & editing, Supervision, Conceptualization. Verena Steinke-Lange: Formal analysis, Data curation. Barbara Klink: Writing – review & editing. Birgit Eichhorn: Formal analysis, Data curation. Winfried Schmidt: Formal analysis, Data curation. Daniel Berner: Formal analysis, Data curation. Anna Teubert: Formal analysis, Data curation. Anne Holtorf: Formal analysis, Data curation. Sarah Heinrich: Formal analysis, Data curation. Gabriele Wildhardt: Formal analysis, Data curation. Martin Schulze: Formal analysis, Data curation. Laura von der Heyden: Formal analysis, Data curation. Konstanze Hörtnagel: Formal analysis, Data curation. Daniela Steinberger: Formal analysis, Data curation. Saskia Kleier: Formal analysis, Data curation. Peter Lorenz: Formal analysis, Data curation. Ralf Glaubitz: Formal analysis, Data curation. Saskia Biskup: Formal analysis, Data curation. Elke Holinski-Feder: Writing – review & editing, Writing – original draft, Supervision, Conceptualization.

Data availability statement

The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Ethical approval

Patients provided their informed consent compliant with the respective national ethical standards of Germany. Internal Ethics Committees approved this study.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

All the authors declare no support from any organization for the submitted work.

Appendix A. Supplementary data

The following is the Supplementary data to this article: