Exploring the hereditary genetic mutational landscape of breast and ovarian cancer in Estonia.

Introduction
According to the World Health Organization, there were approximately 20 million new cancer cases and ~ 10 million deaths due to cancer in the year 2022, with breast cancer (BC) being the second most common cancer1. Ovarian cancer (OC) is the eighth most common cancer in women. In Estonia, on average, over 800 people develop BC, and ~ 150 OC per year, according to the Estonian National Institute for Health Development, Health Statistics and Health Research Database (HSHRD)2. Pathogenic or likely pathogenic genetic variants (PV) in cancer susceptibility genes can lead to cancer development. These PV are commonly inherited in an autosomal dominant manner, often leading to an earlier age at diagnosis3. Individual cancer risk assessment relies on family history, but limited information can lead to an underestimation of hereditary contributions to tumor development. In recent years, the clinical approach to hereditary cancer testing for at-risk patients and their family members has significantly advanced with the advent of multi-gene next-generation sequencing-based (NGS) testing4. Genetic testing (GT) and pretest consultations are becoming increasingly used by different medical specialists. A broader panel-based approach has become increasingly integrated into routine clinical care, facilitating earlier intervention opportunities, risk stratification, and personalized management for individuals and families affected by hereditary cancer syndromes4,5.
BRCA1/2 are the most common genes associated with HBOC6. They encode proteins that play a crucial role in tumor suppression3. Carriers of BRCA1/2 PV have an excessive risk of developing cancer and are eligible for more intensive screening and preventative management strategies3. Additionally, research has revealed that other genes (e.g., CHEK2, PALB2, RAD51C/D) contribute to the development of HBOC5,6. The emergence of novel therapeutic agents, such as poly(ADP-ribose) polymerase (PARP) inhibitors for BC and OC, alongside conventional treatments, underscores the critical need for comprehensive molecular diagnostics to identify actionable variants5,7. NGS diagnostic solutions enable the evaluation of a broad spectrum of genes associated with HBOC, facilitating the identification of patients who may benefit from these emerging treatment options8,9.
In Estonia, the approach to germline GT for HBOC has evolved significantly between the years 2007 and 2023, transitioning from single-gene analysis to targeted NGS GT. Clinical efforts to integrate GT into standard oncological care have steadily expanded, particularly in BC and OC. Despite increasing access to genetic services, there is a paucity of large-scale data on the prevalence and distribution of BRCA1, BRCA2, and non-BRCA PV in Estonia. Currently, Estonian HSHRD does not collect molecular data on patients. However, a detailed understanding of the genetic landscape is critical for optimizing GT and treatment strategies, enhancing cascade screening in at-risk relatives, and contributing to global knowledge on variant pathogenicity and penetrance.
This retrospective study characterizes the germline genetic landscape of Estonian HBOC patients by examining GT uptake and the prevalence of PV in BRCA1, BRCA2, and other cancer susceptibility genes. By integrating demographic characteristics, age at diagnosis and testing, family cancer history, tumor types, and biomarker profiles, this study characterizes the spectrum of PV across both established genes and those beyond current NCCN recommendations5. To our knowledge, this represents the most comprehensive germline molecular genetic analysis of HBOC patients in Estonia to date, supporting precision medicine approaches and informing targeted prevention and clinical management strategies.

Methods

Participants
The study subjects underwent molecular germline GT at the Genetics and Personalized Medicine Clinic, Tartu University Hospital (TUH), across a 17-year period (2007–2023). The cohort consisted of 2,856 BC and 759 OC individuals referred from TUH (n = 2,330), the North Estonian Medical Center (NEMC; n = 765), East Tallinn Central Hospital (ETCH; n = 441), and West Tallinn Central Hospital (WTCH; n = 1). Clinical information, including gender, personal cancer history, and familial cancer history, was retrospectively collected from test requisition forms completed by ordering clinicians at the time of testing. The 154 individuals with bilateral BC were counted once within the BC cohort. Additionally, 78 individuals had both BC and OC and were included in both the BC and OC groups.

Genetic testing
GT was performed using targeted testing methods such as Sanger sequencing, Arrayed Primer Extension (APEX; Asper Biotech Ltd.) microarray10, multiplex ligation-dependent probe amplification (MLPA; MRC-Holland), or NGS-based approaches. To assess the distribution of PV, individuals were grouped based on the identified PV (Fig. 3A):“BRCA1”, including patients with PV in the BRCA1 gene.

“BRCA2”, including patients with PV in the BRCA2 gene.

“other-NCCN”, patients encompassing PV in genes (ATM, BARD1, BRIP1, CDH1, CHEK2, EPCAM, MLH1, MSH2, MSH6, NF1, PALB2, PMS2, PTEN, RAD51C, RAD51D, STK11, TP53) listed in the NCCN Genetic/Familial High-Risk Assessment: Breast, Ovarian, Pancreatic, and Prostate guideline5, excluding BRCA1/2.

“non-NCCN” comprising patients with PV in genes not included in the NCCN Genetic/Familial High-Risk Assessment: Breast, Ovarian, Pancreatic, and Prostate guideline5 and have been associated with increased overall cancer risk.

“NEG”, including patients with no PV.

From 2007–2015, the detection of PV primarily used APEX microarray, targeting 66 common PV in BRCA1, BRCA2, CHEK2, and RAD51C/D. When negative, BRCA1/2 coding regions were sequenced by Sanger sequencing. For BRCA1/2 copy number variant detection, MLPA analysis was introduced in 2009. NGS panels (TruSight Cancer, TruSight One; Illumina) were implemented in 2014. From 2015, testing began with two prevalent BRCA1 PV (NM_007294.4:c.5266dup p.Gln1756Profs*74 and NM_007294.4:c.4035del p.Glu1346Lysfs*20)11, followed by NGS panels and MLPA if required. Targeted BRCA1 testing was discontinued in 2020, and only NGS panels have been used since then. In 2018, panels were updated to TruSight Hereditary Cancer (113 genes) and TruSight One Expanded (6,700 genes). Libraries were prepared using TruSight Rapid Capture or DNA Prep with Enrichment and sequenced on MiniSeq or NextSeq 500 (Illumina) with 2 × 149–151 bp paired-end reads. Detailed gene lists and kit information are provided in Supplementary Information file.

Sequencing data analysis and data interpretation
Raw sequencing data from FASTQ files were aligned to the hg19 reference genome using BWA12. Variant calling were performed using the Genome Analysis Toolkit13. Variants were annotated through an in-house variant annotation pipeline utilizing SnpSift14 and the Annovar software15, with references from RefSeq16, BRCA Exchange17, dbSNP18, ExAC19, gnomAD20, ClinVar21, and OMIM22. CNV detection was conducted using CoNIFER23 or DeCon24. Variant classification followed ACMG guidelines25, and variants were described according to the HGVS Nomenclature26. Only class 4 and 5 variants, as classified according to ACMG/AMP guidelines, were reported in the clinical workflow25. All PV are listed and named according to HGVS nomenclature in the Supplementary Information file (Supplement_table_all_PV). Variants of uncertain significance (VUS) are not reported, and no automated re-evaluation system exists. Patients with negative results are offered re-assessment after 3–5 years, particularly if hereditary cancer is suspected or based on individual clinical considerations.

Statistical analysis
Data analysis was conducted using RStudio (version 2024.12.0). Chi-square tests were used to assess the distribution of PV across age groups within the BC and OC cohorts, as well as to compare the distribution between PV carriers and non-carriers. Chi-square tests were also used to evaluate the distribution of histological subtypes, biomarker profiles, and gene variants in BC and OC cases.
Independent samples t-tests were used to examine associations between gene groups and age at diagnosis for BC and OC, and to compare mean ages at diagnosis. Additionally, t-tests were used to assess differences in the mean number of cancer-affected relatives across gene groups within both cancer cohorts.
The false discovery rate (FDR) was controlled using the Benjamini–Hochberg method to correct for multiple comparisons. Dichotomous outcomes involving gene groups, histological subtypes, and biomarker profiles were summarized using risk ratios (RR) with 95% confidence intervals (CI), and statistical significance was assessed using Fisher’s exact test. A p-value of < 0.05 was considered statistically significant for all analyses.

Ethics approval and consent to participate
This study is a hospital quality research project, authorized by TUH and the TUH Clinical Research Centre and has valid approval IDs by Research Ethics Committee of the University of Tartu (332/T-2, 21.12.2020; 341/M-2, 17.02.2021; 359/M-9, 21.02.2022; 384/M-14, 20.11.2023; 389/M-1, 15.04.2024). This was a non-experimental study, and no new biological samples were collected for its purposes. The study analyzed data from patients tested at the TUH, Genetics and Personalized Medicine Clinic, Molecular Diagnostic Laboratory, as part of the routine clinical workflow in compliance with ISO 15189:2022 standards and under the Estonian Accreditation Centre license (M005). All methods were carried out in accordance with relevant guidelines and regulations.

Results

Cohorts’ characterization
A total of 3,537 individuals were analyzed, including 2,856 BC and 759 OC patients. The general characteristics of the studied cohorts and individual data (gender, mean age at diagnosis, age at GT, and number of relatives with cancer) are summarized in Table 1. During the study period, medical geneticists ordered 1,199 (33.9%) GT, while other medical specialists ordered 2,338 (66.1%) GT.

BC cohort
The BC group includes predominantly females, with males accounting for 0.8%. The GT rate of BC patients increased nine-fold, driven by the implementation of NGS (Fig. 1). The mean age of diagnosis was 50.4 ± 12.0 years, with an average delay of 3.6 years from diagnosis to GT. During the study period, the mean age at diagnosis among BC patients increased from 47 years in 2008 to 53 years in 2023 (data from 2007 were excluded, as only two patients were tested that year) (Fig. 2). Notably, 53% of BC patients in our cohort were diagnosed before the age of 50, the age groups distribution at BC diagnosis shown in Table 2. Among males, the mean age at BC diagnosis was 63.4 ± 14.0 years, and the mean age at the time of GT was 67.0 ± 13.4 years.
Individuals aged 30–39 showed a significantly higher prevalence of PV (p = 0.00001) compared to other BC age groups. Among BC patients with PV, 482 carried a single PV, and 16 carried two (Table 2; Supplementary Table S1). BRCA1 PV carriers had the lowest mean age at diagnosis (44.8 ± 11.3 years), which was 6.1 years earlier than of individuals without PV. All BC patients with PV were diagnosed on average 3.2 years earlier (p = 3.35 × 10–8) than those without PV (Table 2). The detailed mean age at BC diagnosis for all gene groups is shown in Supplementary Table S2.
BC patients reported an average of 1.3 cancer-affected relatives, while BC patients with BRCA1 PV had a significantly higher proportion of relatives with cancer compared to patients without PV (p = 8.56 × 10⁻5; Supplement Table S3).

OC cohort
The mean age of diagnosis in our cohort with OC was 56.1 ± 14.1 years, and GT was performed on average 3.3 years later (Table 1). The mean ages of OC diagnosis and GT during the entire study period are shown in Fig. 2. Among individuals referred for GT, OC was most frequently diagnosed in those aged 60–69 years (27.7%) (Table 2). The rate of GT in OC patients exhibited a substantial increase, rising from 7.2% from 2009 to 2014 to 62.2% between 2015 and 2020 (Fig. 1). Individuals aged 40–49 and 50–59 showed a significantly higher prevalence of PV (p = 0.000005 and p = 0.0003) compared to other OC age groups. Individuals with a PV were diagnosed with OC 3.3 years earlier than those without (p = 7.02 × 10−4) (Table 2). Among those with PV, 194 had one PV, and three had two PV (Supplementary Table S4). The mean age at OC diagnosis was significantly lower in BRCA1 PV carriers (51.2 ± 9.3 years) compared to all other groups (Supplement Tables S5, S6). On average, OC patients reported one cancer-affected relative, and OC patients with BRCA1 PV had a statistically significantly higher proportion of relatives with cancer compared to patients without PV (p = 0.024) (Supplement Table S7).

Molecular findings show a broad PV spectrum
A total of 687 PV were identified in 668 individuals, with an overall diagnostic efficacy of 18.9% (668/3,537) (Supplementary Table ALL PV, Fig. 3A and Table 3). The diagnostic yield during period 2007–2014 was 22.3% (62/278) and during 2015–2023 was 18.6% (606/3,259). In the bilateral BC subgroup, the diagnostic yield was 22.1% (34/154), compared to 17.2% (464/2,702) in the unilateral BC group. In the BC&OC subgroup, 27 PV (34.6%) were identified. Notably, 88.9% of these PV involved BRCA1/2 genes, while the remaining three were detected in CHEK2 and RAD51C genes.
We identified 46 different PV in BRCA1 and 47 in BRCA2 (Fig. 3B,C, Supplement Table ALL PV). Moreover, we identified 25 novel PV across 13 genes, including nine in BRCA1 and three in BRCA2. All PV detected during the study period are listed in Supplemental Table ALL PV.
The most prevalent PV in the entire cohort was BRCA1 c.5266dup, representing 4.0% of the total cohort and 20.8% of all PV. The second most common PV, BRCA1 c.4035del, contributed 8.3% of all PV. Together, these two BRCA1 PV accounted for 66.0% of all BRCA1 reported PV. The most frequent PV in the BRCA2 gene, c.8572C > T, comprised 19.7% of all BRCA2 PV and 3.9% of all PV in the total cohort. Other BRCA1/2 PV and the detailed distributions are illustrated in Fig. 3B,C, and Table 3.
Most BRCA1 PV occurred in exon 20 (BRCT domain; 84 BC, 64 OC) and exon 11b (SCD domain; 57 BC, 43 OC). Frameshift variants predominated, followed by missense and nonsense variants (Supplementary Figures S1A-S1B). BRCA2 PV were most frequent in exons 11 (BRC domain; 26 BC, 13 OC) as well as 20 (20 BC, 7 OC), and frameshift variants were the most common (Supplementary Figures S2A and S2B).
Variants in “other-NCCN” genes accounted for 6.5% of the total cohort and 34.6% of all PV, while PV in “non-NCCN” genes were rare, observed in only 1.3% of all PV (Table 3). The full prevalence and frequencies of PV are shown in Table 3.
A total of 24 male BC cases underwent GT, of whom eight (33.3%) were found to carry PV, resulting in a total of nine PV. The majority of PV were identified in BRCA2 (four cases) and CHEK2 (three cases), with one additional PV detected in RAD51C. One individual carried two distinct CHEK2 gene PV: a frameshift PV, c.1100del, and a large genomic deletion involving exons 9 and 10.

Histological and biomarker findings
In the BC cohort, 80.0% had ductal carcinoma (DC) and 58.2% were hormone receptor-positive (HR +) (Table 1). Both HR + and HER2 + BC were significantly less frequent in the “BRCA1” gene group compared to other gene groups. In contrast, “BRCA1” was most strongly associated with triple negative breast cancer (TNBC), showing a higher prevalence compared to “BRCA2”, “other-NCCN”, and “NEG”, with 2.57, 3.84, and 3.71 times higher prevalence, respectively. Gene group and gene level analysis are shown in Table 4; Supplementary Tables S8-S12.
Among male BC patients, DC was also the most common histological type, observed in 21 individuals (87.5%). Altogether, 22 individuals (91.7%) were diagnosed with HR +, two were HER2 + (8.3%), one had TNBC, and one had unknown biomarker status.
In the OC cohort, limited referral and pathology data precluded grading of serous carcinomas. Overall, 74.4% had serous histology (Table 1), most frequent in BRCA1/2 PV carriers, while non-serous types predominated in carriers of “other-NCCN” gene PV (Table 4).

Discussion
This study provides an overview of PV frequencies among individuals with BC and OC who were tested in routine clinical diagnostics in Estonia. BC and OC GT rate have increased nine-fold, driven by greater involvement of non-geneticist specialists and continuous education on the importance of GT over the past decade. Public awareness following Angelina Jolie’s 2013 disclosure (“Jolie Effect”) has further contributed to this trend27. Mainstreamed GT is now well established, with non-geneticist clinicians accounting for approximately two-thirds of HBOC testing, consistent with global initiatives to improve access and reduce delays6,28,29. When supported by education and collaboration, mainstreaming maintains care quality30 and is cost-effective31.
In both our study cohorts, the mean interval between diagnosis and GT decreased over time, reflecting improved referral pathways. PV rates among BC patients were consistent with the known tendency of BRCA1/2-related disease to present at younger ages, while in OC, PV were most frequent in patients aged 40–59, aligning with reports of BRCA1/2 and homologous recombination repair gene enrichment in this age group5,32–34. Notably, in the ≥ 70 age group, enrichment of non-BRCA genes was found, consistent with studies from unselected populations showing enrichment of moderate-risk PV in older BC patients35,36. These findings underscore the need to offer GT to older individuals and include moderate-risk genes to enable cascade testing and risk assessment.
The overall PV detection rate across the cohort was 18.9%, comparable to other BCOC studies37,38. Stratification by testing era revealed a higher yield before the routine implementation of NGS (≤ 2014; 22.3%) compared with the NGS period (≥ 2015; 18.6%), despite improvements in technical sensitivity. This difference likely reflects strong referral and selection bias in earlier years, when testing was limited to highly selected individuals. The diagnostic yield was 17.4% in BC, 17.2% in unilateral BC, and 26.0% in OC, exceeding the rates reported by Öfverholm et al.38 (15.0%, 14.9%, and 23.8%). In comparison, the bilateral BC subgroup showed a comparable yield (22.1% vs. 22.5%)38. BRCA1/2 showed the highest PV frequency, consistent with HBOC studies37–39, with BRCA1 strongly associated with OC and BRCA2 with both cancer types4,5,33,37,38,40. The high BRCA1 PV rate in OC supports routine GT for all OC patients, regardless of family history5,41. In the BC&OC subgroup, PV detection was 34.6%, surpassing the 32.0% yield reported by Öfverholm et al.38. This subgroup demonstrates approximately twice the likelihood of harboring PV compared to unilateral BC cases, underscoring the importance of genetic testing in patients with multiple or bilateral cancers5,38,41.
The BRCA1/2 PV distribution mirrors reported mutational hotspots, particularly in exons within key functional domains4,38,40,42. BRCA1 PV clustered in exon 11, which encodes the serine‑rich central domain (SCD) involved in DNA damage response signalling, and in exon 20, which contains the BRCT domains, essential for DNA repair and cell‑cycle regulation43. In BRCA2, PV were frequent in exon 11, which harbours the BRC repeats, evolutionarily conserved motifs that mediate the interaction with RAD51 and are critical for homologous recombination44. Frameshift and nonsense variants predominated, reflecting their role in truncating proteins4,38,45 and their high representation among PV in ClinVar21.
The two most common BRCA1 PV in our cohort were c.5266dup and c.4035del, previously reported as the main PV in Estonia11,46,47. The c.5266dup PV is prevalent across Europe48, and both PV are consistently observed in Baltic cohorts. In Latvia, c.5266dup and c.4035del accounted for 57.2% and 38.8% of cases, respectively49, while in Lithuania, c.4035del was found in 44.0% and c.5266dup in 26.0%50. In our cohort, these two BRCA1 PV represented ~ 30% of all PV and 45.5% of BRCA1/2 PV, lower than in other countries42,49–52, reflecting the genetic diversity in Estonia. Thus, targeted testing for recurrent PV would miss many carriers, supporting the adoption of broad NGS-based GT in Estonia. The most frequent BRCA2 PV, c.8572C > T, was also the most common in the Estonian Biobank46,47 and reported in a few Lithuanian and Latvian families49,50. Its rarity in other populations per gnomAD20, suggest a possible Estonian founder effect.
PV in “other-NCCN” genes were identified in 6.5% of the total cohort and accounted for 34.6% of all PV, with CHEK2 being most frequent, consistent with European population data37,53–55. The OC cohort also included PV in BRIP1, RAD51C, and PALB2, genes well established in hereditary OC34,38,56. PV in “non-NCCN” genes were rare (0.25%) and typically associated with other cancer syndromes. These likely represent incidental and/or secondary findings rather than primary BCOC drivers, as such genes are excluded from HBOC testing guidelines5,6. Their presence highlights limitations of broad panels and warrants further study, while also offering opportunities for early detection and tailored surveillance in families with multiple cancer types or syndromes (e.g., MEN1 or BAP1). Notably, germline DICER1 PV, linked to DICER1 syndrome, can manifest as ovarian Sertoli-Leydig cell tumors, previously reported in OC patients57.
Histological and biomarker analyses demonstrated clear genotype–phenotype correlations. DC was the predominant subtype, consistent with prior studies and BRCA1-associated TNBC patterns4,5,33,38,58. Among OC patients, BRCA1 PV carriers had the earliest mean age at diagnosis, and serous carcinoma predominated in BRCA1/2 PV carriers, in agreement with prior reports38,59,60.
Although small, the male BC subgroup offers important insights. From 2007–2023, 97 male BC cases were registered in Estonia2, but only 24 (24.7%) underwent GT, despite clear recommendations in guidelines5,41. Uptake remains low, highlighting the need for greater awareness and risk-adapted surveillance. Among tested male BC patients, 33.3% carried PV, predominantly in BRCA2, which is strongly associated with male BC and confers increased risks for prostate, pancreatic, and other cancers5,61.
Broad germline GT enabled identification of 25 novel PV across 13 genes, with 36.0% of PV occurring in non-BRCA genes, underscoring the genetic heterogeneity of BCOC in Estonia. Limiting testing to the two common BRCA1 variants (c.5266dup and c.4035del) would have identified only 5.7% of HBOC patients. Expanding analysis to full BRCA1/2 Sanger sequencing increased the diagnostic yield to 12.4%, while NGS-based multigene panels achieved a yield of 18.9%, reflecting the aggregate diagnostic performance across the study period. These findings highlight the limitations of conventional testing strategies and support the use of multigene NGS panels, consistent with previous reports, including Henkel et al.37. The use of broad NGS-based multigene panels substantially improves diagnostic yield compared with conventional testing strategies; however, it also increases the detection of VUS, which complicates result interpretation and clinical decision-making in hereditary cancer genetics. In our routine workflow, there is a possibility to re-evaluate the GT results of individuals if they are referred to medical genetic counseling, as variant classification and clinical knowledge continue to evolve. While NGS enhances variant detection, the growing VUS burden remains a significant challenge for the clinical utility of large gene panels.
This study has several limitations. First, the retrospective design resulted in variable completeness of clinical, histological, and family history data, which may have affected the interpretation of genotype–phenotype correlations. Second, GT strategies evolved from targeted analysis to NGS-based multigene panels, leading to a possible underestimation of PV in earlier years. Third, the increased interpretative complexity introduced by large NGS panels, particularly due to the detection of VUS. In our clinical workflow, VUS are not reported, and finalized GT results are not systematically re‑evaluated.
In conclusion, this study characterized the PV spectrum of BCOC in Estonia. Our findings support NGS as the standard due to its ability to detect a broad range of PV beyond conventional methods. Increased awareness has improved GT uptake, but sustained progress requires professional education, patient engagement, interdisciplinary collaboration, and continuous refinement of national guidelines. Integrating genetic data with cancer registries and leveraging molecular diagnostics will enable real-time risk stratification, cascade testing, and personalized prevention strategies, ultimately improving outcomes through earlier diagnosis and timely intervention in Estonia.

Supplementary Information