Potential advantage of clinical exome sequencing in BRCA1/2-negative families: a retrospective study of a cohort of 500 patients at a high-risk for hereditary cancers

Abstract

Background

Hereditary cancer syndromes (HCSs) account for approximately 5–10% of all cancers and are frequently associated with pathogenic variants (PVs) in genes such as BRCA1 and BRCA2. Nevertheless, a substantial proportion of individuals with strong familial cancer aggregation remain genetically unexplained after standard multigene panel testing. Clinical exome sequencing (CES) may overcome this limitation by enabling a broader exploration of cancer susceptibility genes.

Methods

This retrospective study included 500 high-risk patients who previously tested negative for BRCA1/2 PVs. All samples were analysed using next-generation sequencing technology. Initially, a 60-gene filtered panel recommended by the American Society of Clinical Oncology (ASCO) guidelines was applied, followed by a 102-gene filtered panel derived from CES to identify additional PVs or likely PVs (LPVs) beyond current diagnostic panels. Variant classification was performed according to the American College of Medical Genetics and Genomics/Association for Molecular Pathology guidelines, incorporating ClinGen and gene-specific expert curation recommendations when available.

Results

Within the ASCO panel, PV/LPVs were identified in 9% of patients while 35% carried variants of uncertain significance. CES identified additional PVs/LPVs in 5% of patients in genes not currently included in ASCO-recommended surveillance panels, such as RAD50, BLM, WRN, PMS1 and FANCA. These variants may represent candidate susceptibility loci requiring further clinical and functional validation.

Conclusions

CES may provide additional exploratory genomic information in selected high-risk BRCA1/2-negative patients, particularly when standard panel testing is uninformative. Although many of the additional findings identified are not yet clinically actionable, CES represents a valuable resource for future reinterpretation as gene–disease evidence evolves. Overall, our findings support the use of CES as a second-tier exploratory approach in carefully selected high-risk cases, while emphasising the need for cautious interpretation and continued gene–disease curation.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Genetic testing for hereditary cancer syndromes is typically based on targeted multigene panels, yet a proportion of high-risk patients remain negative despite strong clinical suspicion.

WHAT THIS STUDY ADDS

• This study shows that clinical exome sequencing can identify additional PVs/LPVs beyond standard panels, although many findings—particularly in non-guideline genes or recessive conditions—have uncertain clinical utility.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These results support the use of broader sequencing approaches in selected high-risk patients while highlighting the need for cautious interpretation and further studies to clarify clinical relevance.

Introduction

Hereditary cancer syndromes (HCSs) account for approximately 5–10% of all cancers and typically follow an autosomal dominant inheritance pattern, resulting in a 50% risk of transmission to offspring.¹ Although each syndrome exhibits distinct clinical manifestations, several shared features characterise HCSs: (1) early cancer onset, (2) occurrence of multiple primary tumours in the same individual and (3) a positive family history of cancer.^{2 3} Genetic testing plays a crucial role in precision oncology, allowing the identification of individuals eligible for targeted therapies and enabling preventive strategies for high-risk families.⁴ Germline testing is increasingly recommended for patients with cancer, as inherited pathogenic variants (PVs) contribute to a growing proportion of diagnosed cases.⁵ A detailed family history is essential for selecting appropriate germline tests: for some patients, testing eligibility is determined by familial aggregation, whereas for others, it is indicated regardless of family background. Moreover, family history often guides the inclusion of specific genes in testing panels, complementing those dictated by the patient’s personal tumour history. Germline PVs have significant implications for targeted therapies, such as the use of PARP inhibitors in BRCA1/2 carriers or immunotherapy in tumours with microsatellite instability or mismatch repair deficiency.⁶ Carriers of these variants benefit from specific therapeutic and surveillance protocols, representing one of the most successful examples of precision medicine. Likewise, relatives of individuals carrying PVs should undergo cascade testing and, if positive, be included in personalised surveillance programmes distinct from general population screening. BRCA1 and BRCA2, discovered in the mid-1990s, remain the major susceptibility genes for hereditary breast and ovarian cancers.⁷ Nevertheless, numerous additional genes with low, moderate or high penetrance have since been identified,⁸ indicating the importance of broader and more cost-effective molecular analyses.⁹ The advent of next-generation sequencing (NGS) technologies and multigene panels has enabled simultaneous analysis of multiple samples and genes, significantly improving diagnostic efficiency.^{9 10} Although BRCA1/2 variants account for roughly 25% of hereditary cancer cases,^{8 11} several non-BRCA genes also contribute to cancer predisposition, as highlighted in the National Comprehensive Cancer Network (NCCN) guidelines. These guidelines stress the importance of genetic testing, family history and management strategies to identify high-risk individuals and recommend active surveillance or, in selected cases, prophylactic surgery for early diagnosis and prevention across multiple cancer types, including breast, ovarian, colorectal, prostate and gastric cancers.¹² The American Society of Clinical Oncology (ASCO) also provides evidence-based recommendations for hereditary cancer management, supporting informed clinical decisions and optimised patient outcomes.¹³ ASCO currently recommends the use of a 60-gene panel, beyond BRCA1/2, for clinical surveillance, as shown in table 1. However, high-risk patients testing negative for variants in ASCO-recommended genes may still harbour PVs in additional cancer susceptibility loci. Clinical exome sequencing (CES) offers a comprehensive diagnostic strategy for these individuals, expanding variant detection and identifying potentially novel genes involved in cancer predisposition.¹⁴ Nevertheless, CES has limitations, including the detection of variants of uncertain significance (VUS), incomplete clinical actionability and ethical considerations regarding incidental findings. In this study, we applied CES to a cohort of 500 high-risk patients who previously tested negative for BRCA1/2 PVs/likely PVs (LPVs) and presented with a strong family history of cancer. The aims of this study were: (1) to investigate the prevalence of PV or LPV in 60 ASCO-recommended cancer susceptibility genes beyond BRCA1/2 and (2) to assess the utility of CES in identifying additional variants in genes not currently included in clinical surveillance programmes.

Table 1. Genes recommended for testing and inclusion in multigene panels for selected cancers.

Gene	RefSeq MANE	OMIM	Gene	RefSeq MANE	OMIM
BRCA1	NM_007294.4	113705	KIT	NM_000222.3	164920
BRCA2	NM_000059.4	600185	PDGFRA	NM_006206.3	173410
PALB2	NM_024675.4	610355	SDHA	NM_004168.3	600857
CDH1	NM_004360.3	192090	SDHAF2	NM_001282.3	615674
PTEN	NM_000314.6	601728	SDHB	NM_003000.3	185470
STK11	NM_000455.6	602216	SDHC	NM_003001.3	602413
TP53	NM_000546.6	191170	SDHD	NM_003002.3	602414
ATM	NM_000051.3	607585	RET	NM_020975.3	164761
BARD1	NM_001024.5	601593	EGFR	NM_005228.4	131550
CHEK2	NM_001005735.3	604373	MEN1	NM_001080.3	613733
RAD51C	NM_058216.3	602774	CDKN2A	NM_000077.5	600160
RAD51D	NM_002878.3	602775	CDK4	NM_000075.4	600681
NF1	NM_001042492.2	162200	MC1R	NM_002387.3	155555
APC	NM_000038.5	611731	MITF	NM_000248.3	156845
EPCAM	NM_002354.3	608089	POT1	NM_006261.3	608485
MLH1	NM_000249.3	120436	TERT	NM_198253.3	187270
MSH2	NM_000251.3	120436	BRIP1	NM_032043.3	603358
MSH6	NM_000179.3	600678	FH	NM_000143.3	136850
MUTYH	NM_001128425.2	604933	MAX	NM_002382.3	155550
NTHL1	NM_001130383.2	613989	TMEM127	NM_001039.3	613667
PMS2	NM_000535.7	600259	VHL	NM_000551.3	608537
POLD1	NM_002691.3	174761	EGLN1	NM_001429.3	604273
POLE	NM_006231.3	174760	EPAS1	NM_001430.3	604274
BMPR1A	NM_004329.3	601299	KIF1B	NM_001080.3	603850
SMAD4	NM_005359.4	600993	MET	NM_000245.3	164860
AXIN2	NM_003535.3	601143	HOXB13	NM_001144.3	601213
MBD4	NM_004830.3	603506	BAP1	NM_004656.3	603089
GREM1	NM_001080.3	607296	FLCN	NM_001456.3	607273
MSH3	NM_002439.3	600678	TSC1	NM_000368.3	191092
RNF43	NM_001277.3	611232	TSC2	NM_000548.3	191091
CTNNA1	NM_001801.3	601149	RB1	NM_000321.3	614041

MANE, Matched Annotation from NCBI and EMBL-EBI; OMIM, Online Mendelian Inheritance in Man.

Materials and methods

Study population and samples collection

From January 2023 to October 2025, a total of 500 patient samples referred to the Medical Genetic Service of the University ‘G.d’Annunzio’ of Chieti-Pescara–Center of Advanced Studies and Technologies were collected for genetic analysis. All patients had previously undergone targeted sequencing of the BRCA1 and BRCA2 genes using NGS. Multiplex ligation-dependent probe amplification analysis was conducted to detect large deletions and duplications in the BRCA1 and BRCA2 genes.

Subsequently, CES was performed on each sample to extend the genomic investigation and identify additional variants of potential clinical relevance. Patients were stratified into subgroups according to tumour type, allowing comparative analyses across different cancers. Eligible participants met at least one of the following familial cancer criteria: (1) multiple first-degree or second-degree relatives with breast cancer; (2) breast cancer diagnosed before age 40 or bilateral breast cancer before age 50; (3) male breast cancer or ovarian cancer at any age; and (4) two or more cases of prostate cancer among first-degree or second-degree relatives. All participants received pretest genetic counselling jointly conducted by a clinical geneticist and a psychologist. During this session, patients were informed about the purpose, scope and limitations of the genetic analyses. Written informed consent was obtained from each participant. Personal and family medical histories were collected to support the interpretation of genetic results. Post-test counselling was provided to explain the outcomes and discuss clinical implications.

DNA extraction

Blood samples or buccal swabs were obtained from all patients. Using the MagPurix system and the Forensic DNA Extraction Kit (Cat. No. ZP01001) or the Blood DNA Extraction Kit 200 (Cat. No. ZP02001) from Zinexts Life Science (Taipei, Taiwan), genomic DNA was extracted in accordance with the manufacturer’s guidelines.

Clinical exome sequencing

Sample analysis was performed using NGS technology. The analytical workflow began with the preparation of NGS libraries using the ClinEX-4bases clinical exome kit (4bases SA, CH-6928 Manno, Switzerland). This kit targets over 10 000 genes listed in the Online Mendelian Inheritance in Man (OMIM) database as being associated with known diseases and is based on a capture-based sequencing protocol. The resulting libraries were normalised, pooled and loaded onto a flow cell for sequencing. Sequencing was carried out using the NextSeq 550DX platform (Illumina, San Diego, California, USA).

Data analysis

Sequencing data analysis was conducted using the Geneyx software (Geneyx, USA), a dedicated platform for the interpretation and clinical annotation of NGS results. The bioinformatics pipeline includes sequence alignment to the human reference genome (GRCh19), variant calling and annotation based on public and proprietary databases. For each patient, a targeted filtering strategy was applied focusing on a curated panel of 60 genes, shown in table 1, selected in accordance with the ASCO guidelines.

These genes are clinically relevant and frequently implicated in HCSs, reported in table 2.

Table 2. List of genes used in the filtered panel.

Cancer type	Associated genes
Breast cancer	BRCA1, BRCA2, PALB2, CDH1, PTEN, STK11, TP53, ATM, BARD1, CHEK2, RAD51C, RAD51D, NF1
Colorectal cancer	APC, EPCAM, MLH1, MSH2, MSH6, MUTYH, NTHL1, PMS2, POLD1, POLE, BMPR1A, SMAD4, STK11, TP53, AXIN2, CHEK2, MBD4, GREM1, MSH3, PTEN, RNF43
Endometrial cancer	EPCAM, MLH1, MSH2, MSH6, PMS2, PTEN, STK11
Gastric cancer	APC, CTNNA1, EPCAM, MLH1, MSH2, MSH6, PMS2, BMPR1A, CDH1, SMAD4, STK11
Gastrointestinal stromal tumours	KIT, PDGFRA, SDHA, SDHAF2, SDHB, SDHC, SDHD, NF1
Medullary thyroid carcinoma	RET
Non-small cell lung cancer (EGFR-related)	EGFR, STK11, TP53
Adrenocortical tumours	APC, EPCAM, MEN1, MLH1, MSH2, MSH6, PMS2, TP53
Cutaneous melanoma	CDKN2A, CDK4, BAP1, MC1R, MITF, POT1, TERT, PTEN
Uveal melanoma	BAP1
Ovarian cancer (epithelial)	BRCA1, BRCA2, BRIP1, EPCAM, MLH1, MSH2, MSH6, PALB2, PMS2, RAD51C, RAD51D, ATM
Pancreatic adenocarcinoma	ATM, BRCA1, BRCA2, CDK4, CDKN2A, EPCAM, MLH1, MSH2, MSH6, PALB2, PMS2, STK11, TP53, APC
Pheochromocytomas and paragangliomas	FH, MAX, RET, SDHA, SDHB, SDHC, SDHD, TMEM127, NF1, VHL, EGLN1, EPAS1, KIF1B, MET, SDHAF2
Prostate cancer	BRCA1, BRCA2, EPCAM, HOXB13, MLH1, MSH2, MSH6, PMS2, ATM, CHEK2, PALB2
Renal cell carcinoma	BAP1, FH, FLCN, MET, SDHA, SDHAF2, SDHB, SDHC, SDHD, PTEN, VHL, TSC1, TSC2
Sarcoma (soft tissue/osteosarcoma)	TP53, NF1, RB1

For patients testing negative on the ASCO panel, analysis was extended to 102 additional genes reported by the Human Phenotype Ontology (HPO) (https://hpo.jax.org/) as linked to hereditary cancers, despite lacking official clinical validation, as shown in table 3.

Table 3. List of genes used in the second filtered panel of 102.

102 Filtered genes (HPO)
AAGAB	AIP	AKT1	ALK	ANKRD26	ARMC5	ATR	BLM	BUB1	BUB1B
BUB3	C1S	CASP8	CDC73	CDKN1B	CDKN1C	CEBPA	CEP57	CHEK1	COL14A1
CTNNB1	CYLD	DDX41	DICER1	DIS3L2	DLST	DROSHA	ERBB2	ESR1	ETV6
EXT1	EXT2	FAN1	FANCA	FANCC	FANCM	FGFR2	FOXE1	GALNT12	GATA2
GNAS	GPC3	HABP2	HMMR	HRAS	IDH1	IDH2	KLLN	KRAS	LZTR1
MDM2	MEFV	MINPP1	MLH3	MRE11	NBN	NF2	NQO2	OPCML	PALLD
PHB1	PHOX2B	PIK3CA	PMS1	PPM1D	PRKAR1A	PRKN	PTCH1	PTCH2	RABL3
RAD1	RAD50	RAD51	RAD54L	RB1CC1	RECQL	RECQL4	RECQL5	REST	RPS19
RUNX1	SMAD9	SMAD9L	SEC23B	SH3KBP1	SLC22A18	SLC6A17	SMAD7	SMARCA4	SMARCB1
SMARCE1	SRP72	SUFU	TERC	TGFBR2	TP53I3	TRIP13	TWIST1	WRN	WT1
XRCC2	XRCC3

HPO, Human Phenotype Ontology.

Genetic variant classification

The nomenclature of all genetic variants was standardised according to the guidelines of the Human Genome Variation Society (https://www.hgvs-nomenclature.org, accessed 30 October 2025). To interpret their clinical relevance, data were cross-referenced using multiple curated databases, including ClinGen (https://clinicalgenome.org), ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), for selected cancer susceptibility genes, LOVD-InSIGHT (https://www.insight-group.org/variants/databases/, accessed 30 October 2025) and interpretation tools like Varsome (https://varsome.com) and Franklin by Genoox (https://franklin.genoox.com).

Variant classification was performed according to the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) 2015 guidelines for germline variant interpretation, using the five-tier classification system: benign (C1), likely benign (C2), VUS (C3), likely pathogenic (C4) and pathogenic (C5). ClinGen Variant Curation Expert Panel specifications and gene-specific recommendations were considered to support variant interpretation. In this study, particular attention was given to LPVs and PVs due to their potential relevance in hereditary cancer risk assessment and preventive oncology.

Results

In our study, a total of 500 BRCA1/2 negative patients diagnosed with various malignancies underwent comprehensive molecular profiling through CES. Patients were stratified by sex and tumour type. The age of first cancer diagnosis ranged from 34 to 70 years, with a mean and median age of 56 and 54 years, respectively.

Overall, 64 were male and 436 were female. The distribution by tumour type was as follows: 370 cases of breast cancer (74%), 31 cases of multiple primary tumours (6%), 19 melanomas (4%), 17 thyroid cancers (4%), 14 prostate cancers (3%), 12 ovarian cancers (2%), 10 colon cancers (2%), 7 pancreatic cancers (1%) and 4% of other tumours (renal, endometrial, rectal, sigmoid, biliary tract, paraganglioma and basal cell cancers) as shown in figure 1 (Supplementary material).

Figure 1. Distribution of different types of tumours in the cohort of 500 patients analysed with CES. CES, clinical exome sequencing.

The high number of breast cancer cases explains the higher number of women than men in our genetic tests. Before applying filters, a maximum of 48 thousand variants were identified.

After the selection of 60-gene filtered panel, we obtained over 300 variants. The figure 2 (Supplementary materials) illustrates the study design and the steps taken during the analysis of data.

Figure 2. Study design of analysis process. ASCO, American Society of Clinical Oncology.

After the analysis of BRCA1/2 negative patients, in the second step of our analysis, we focused on the canonical genes predisposed to HCSs, and CES analysis was conducted using the filtering of 60 genes recommended by ASCO guidelines.

Application of this panel identified PVs/LPVs in 9% of cases, as shown in figure 3 (Supp. materials), in particular in canonical genes associated with autosomal dominant conditions like POLE (13%), APC (8%), CHEK2 (8%), ATM (6%), MSH6 (6%), BRIP1 (4%), EPCAM (4%), PALB2 (4%), PMS2 (4%), SDHA (4%), MLH1 (3%), POLD1 (3%), RAD51C (3%) and SDHAF2 (3%) and in canonical genes associated with autosomal recessive conditions like MUTYH (15%), MSH3 (6%) and NTHL1 (6%). Of these PVs/LPVs in autosomal recessive genes, most of them were occurring as heterozygous variants, except for MUTYH c.1187G>A variant in homozygous state. All identified POLE variants were loss-of-function variants (non-sense, frameshift or canonical splice-site alterations). No missense variants affecting the exonuclease domain were detected. VUSs were identified in 35% of the patients, most of them in APC, ATM, CDH1, PALB2 and other genes, and wild-type profiles are 51% of the samples.

Figure 3. All PVs/LPVs found in ASCO recommended genes. ASCO, American Society of Clinical Oncology; LPVs, likely pathogenic variants; PVs, pathogenic variants.

Genotype–phenotype correlation in ASCO recommended genes

Our analysis revealed several noteworthy genotype–phenotype correlations within the cohort. In 30 patients (6%), PV/LPVs were identified in genes consistent with the diagnosed tumour type. Notably, patients with breast cancer frequently carried PV/LPVs in ATM, CHEK2 and PALB2, all of which are recognised contributors to hereditary breast cancer susceptibility. Similarly, variants in colorectal cancer-associated genes such as MSH6 and MLH1 were detected in individuals affected by colorectal tumours. Furthermore, a PV in SDHA was identified in a patient diagnosed with a gastrointestinal stromal tumour, and a LPV in SDHAF2 was found in a patient with paraganglioma, both reflecting established gene–disease associations.^{15 16} Conversely, in 15 patients (3%) who tested positive on the 60-gene filtered panel, no apparent correlation was observed between the identified variants and the individual’s personal or family cancer history. Specifically, several cases involved PV/LPVs in genes typically unrelated to the observed tumour phenotype. For instance, five different monoallelic variants of MUTYH gene were detected in patients affected by breast, thyroid and prostate cancers. In this context, heterozygous variants in MUTYH are more appropriately interpreted as incidental carrier status, as biallelic PVs are required to cause MUTYH-associated polyposis. In our previous work, MUTYH was identified as the gene with the highest PVs/LPVs percentage in the studied group.¹⁰ Notably, one patient with breast cancer carried the c.1187G>A variant in a homozygous state, representing a more clinically relevant example of a pathogenic genotype identified in a patient with a tumour type not classically associated with this condition. Similarly, monoallelic PV/LPVs in MSH3 and NTHL1 were detected in patients with breast, prostate and thyroid cancers. As PVs in these genes are known to cause colorectal cancer predisposition only in the biallelic state, these findings are also more consistent with incidental carrier status rather than atypical genotype–phenotype correlations.^{17 18}

Interestingly, a patient with triple-negative breast cancer harboured a PV in RAD51C, a gene involved in homologous recombination repair and known to be associated with hereditary breast cancer, together with a PV in EPCAM, the c.93del, whose role in this tumour type is unclear.

The coexistence of PVs in two cancer predisposition genes (double heterozygosity) is considered relatively infrequent in the general population but is increasingly observed with the widespread use of multigene sequencing panels.³ This coexistence highlights the complexity of genotype–phenotype correlations and raises the possibility of secondary or modifying genetic effects beyond established cancer predisposition genes.

Analysis of non-ASCO filtered genes

In the next step of our analysis, to further explore potentially relevant genomic alterations beyond the standard diagnostic scope, a research-based secondary panel filtered referred to as ‘non-ASCO PVs/LPVs’ was subsequently applied. This list included 102 genes not currently reportable under ASCO recommendations but suggested by HPO. CES study was extended also to gene ASCO carriers, as individuals with PVs in two cancer susceptibility genes account for 0.2% of all cases. Analysis of non-canonical HCS genes revealed that 23 patients (5%), previously classified as wild type, harboured PVs/LPVs potentially associated with their primary tumour type, but not reportable due to exclusion from standard clinical gene panels. Variants were classified as PV or LPV based on literature and database data. The study identified 9 missense variants, 13 non-sense variants and 3 changing canonical splicing site variants in genes with limited clinical evidence of HCS, reported in table 4. In some patients we found correlation between clinical manifestation and genotype association. For example, a patient with metastatic breast cancer resulted positive at the variant c.306-1954_306-1953del (ClinVar ID: 2072614) in the CASP8 gene, generally indicated that could have a role in breast cancer susceptibility and progression, as it encodes caspase-8, a key protein in apoptosis, as reported in literature.^{19 20}

Table 4. Table of PVs/LPVs observed in genes not addressed by the ASCO guidelines.

Gene	cDNA	Protein	MANE transcript
BLM	c.98+2del	/	NM_000057.4
BUB1B	c.1302del	p.Ser435Valfs*19	NM_001211.6
CASP8	c.358_359del	p.Gln120fs*33	NM_001228.4
FAN1	c.1597C>T	p.Gln533*	NM_014967.5
FANCA	c.3761_3762del	p.Glu1254Glyfs*23	NM_000135.4
FANCA	c.306del	p.Ser103Glnfs*35	NM_000135.4
FANCA	c.3788_3790del	p.Phe1263del	NM_000135.4
FANCA	c.1715+1G>A	/	NM_000135.4
FANCM	c.2354del	p.Leu785Tyrfs*34	NM_020937.4
LZTR1	c.628C>T	p.Arg210*	NM_006767.4
LZTR1	c.401-2_401-1del	/	NM_006767.4
NBN	c.2238C>A	p.Tyr746*	NM_002485.5
PMS1	c.829C>T	p.Arg277*	NM_000534.
PTCH2	c.1296_1300del	p.Val433Profs*36	NM_003738.5
PTCH2	c.35del	p.Pro12Argfs*15	NM_003738.5
RAD50	c.2165del	p.Lys722Argfs*14	NM_005732.4
RUNX1	c.968-2A>C	/	NM_001754.5
RUNX1	c.1277_1278inscccccccc	p.Arg427Profs*170	NM_001754.5
SEC23B	c.993+2T>A	/	NM_006363.5
SEC23B	c.325G>A	p.Glu109Lys	NM_006363.6
SEC23B	c.400C>T	p.Arg14Trp	NM_006363.6
SEC23B	c.1489C>T	p.Arg497Cys	NM_006363.6
TGFBR2	c.914T>C	p.Leu305Pro	NM_003242.6
WRN	c.1578del	p.Leu528Cysfs*29	NM_000553.6
WRN	c.1105C>T	p.Arg369*	NM_000553.6
BLM	c.98+2del	/	NM_000057.4

ASCO, American Society of Clinical Oncology; LPVs, likely pathogenic variants; MANE, Matched Annotation from NCBI and EMBL-EBI; PVs, pathogenic variants.

In a patient with colon cancer with liver and peritoneal metastases, the missense variant c.914T>C (ClinVar ID: 2747583) was found in the TGFBR2 gene (OMIM *190182). This variant involves the substitution of a thymine with a cytosine at position 914 of exon 7 of TGFBR2, which could alter the amino acid sequence of the protein, potentially compromising its function. PV/LPV in the TGFBR2 gene are common in microsatellite instability (MSI-H) colon cancers, which are characterised by defects in the mismatch repair (MMR) system.²¹ These PVs/LPVs lead to loss of receptor function, reducing the cell’s response to TGF-β-mediated growth inhibition. As a result, tumour cells proliferate uncontrollably, evade apoptosis and may acquire invasive properties. Although not reported in the ASCO recommendations, TGFBR2 is closely related to the patient’s diagnosis.²² However, in these two situations, we were able to establish a clinical correlation between the tumour and the genetic result.

An interesting finding was the presence of two individuals with two PVs. Specifically, biallelic RUNX1 PVs were observed in a woman with multiple tumours. The variants c.1277_1278insCCCCCCCC and the c.968-2A>C were found in RUNX1, a gene that (OMIM *151385) encodes a Runt-related transcription factor, which is part of the RUNX gene family and is associated with autosomal dominant acute myeloid leukaemia.²³ Although two PVs were identified in this gene, it was not possible to determine their phase; therefore, it is unclear whether they are located in cis or in trans. Biallelic germline PVs in RUNX1 are extremely rare, as reported in the literature. In fact, a study by Preudhomme et al²⁴ describes 19 out of 20 families with monoallelic germline mutations in RUNX1. Biallelic RUNX1 mutations have been found generally in somatic tissue, as reported in the literature.^{23 25} In addition, in a study by Brown et al²⁶ their findings suggest a general model of RUNX1 as a tumour-suppressor gene, in which RUNX1 operates as both a haploinsufficient tumour suppressor—with one mutation only required—and a classic Knudsonian tumour suppressor, where two mutations are required. Furthermore, this indicates that alterations in other pathways may, in some cases, phenocopy the effect of a second RUNX1 mutation or alternatively contribute to leukemogenesis through distinct mechanisms. In our context, the absence of haematological manifestations and the patient’s advanced age might suggest that the two variants are more likely located on the same allele (in cis), although this cannot be confirmed given the impossibility of segregation due to the death of the patient’s parents. The personal and family history did not reveal neoplasms typically associated with RUNX1-related predisposition. In the second subject, a woman with spitzoid melanoma harboured two variants in the SEC23B and BUB1B genes, respectively.

The SEC23B gene (OMIM *610512) is an essential component of coat protein complex II-coated vesicles that transport secretory proteins from the endoplasmic reticulum to the Golgi complex and it would seem to be associated with autosomal dominant Cowden Syndrome. The BUB1B gene (OMIM *602860) encodes a protein that plays a critical role in regulating the spindle-assembly checkpoint and appears to be associated with somatic colorectal cancer,²⁷ but the authors did not determine whether BUB1B PVs functionally alter the gene product. Thomas Rio Frio et al, describe a case of homozygous PV of BUB1B and susceptibility to gastrointestinal cancer in a patient with adenocarcinoma of the ampulla of Vater and subsequently colon and stomach cancers.²⁸

About this subject, a strong family history of melanomas was present, but our data did not explain the correlation between the genotype and the clinical manifestations. This study was constrained by the inability to analyse the deletion and duplication status of cancer susceptibility genes in all germline-negative patients evaluated for CES.

The additional genes affected included BLM, FAN1, FANCA, FANCM, LZTR1, NBN, PMS1, PTCH2, RAD50 and WRN.

These findings highlight the complexity and the potential clinical relevance of incorporating broader genomic panels in selected cases, as they may uncover actionable or biologically significant variants not captured by conventional testing approaches.

Discussion

In this retrospective study including 500 high-risk patients who tested negative for BRCA1 and BRCA2 PVs, CES enabled the identification of PV/LPVs in additional cancer susceptibility genes. While many of these findings currently have uncertain or limited clinical actionability, they contribute to a broader genomic characterisation of hereditary cancer susceptibility and may inform future gene–disease associations as knowledge evolves.^{29 30}

A key issue in hereditary cancer genetics is whether to employ targeted multigene panels or broader approaches such as CES. While multigene panels remain cost-effective and guideline-driven, they are inherently limited by predefined gene lists.³¹ In patients with persistent clinical suspicion after negative panel testing, CES allows the exploration of a wider spectrum of genes that may contribute to cancer susceptibility and familial clustering of tumours. In this context, CES may provide an additional layer of genomic information that can help refine the molecular characterisation of these patients. In this study, a 60-gene panel based on ASCO recommendations, including genes with high, moderate and low penetrance, was used as the primary analytical approach. CES was subsequently applied in selected cases with a strong clinical suspicion of hereditary cancer predisposition despite negative first-tier testing. While CES enabled the identification of additional PV/LPVs beyond those included in standard panels, the clinical utility of many of these findings remains uncertain at present. In several cases, these variants may represent incidental carrier status or alterations in genes for which cancer risks and management recommendations are not yet well defined. Therefore, rather than providing an immediate diagnostic advantage, CES currently offers a broader exploratory assessment of cancer susceptibility. Importantly, these findings should be considered within a research context, and further studies are needed to validate their clinical relevance and to determine their potential impact on patient management and diagnostic strategies over time. In our cohort, applying the ASCO filter identified PV/LPV in 9% of cases and VUS in 35%, while extension to a 102-gene research filter derived from the CES dataset uncovered additional PV/LPVs in 5% of patients that are not currently included in ASCO/NCCN clinical surveillance panels. Rather than representing a validated increase in diagnostic yield, these additional findings should be considered exploratory observations requiring further clinical and functional validation. Although some variants showed potential genotype–phenotype correlations, the current evidence supporting the role of several of these genes in hereditary cancer predisposition remains limited or controversial. Importantly, although CES allows the analysis of a broad range of genes, it is important to recognise that not all genes contribute equally to hereditary cancer risk. In clinical practice, genetic testing is typically guided by tumour type and established recommendations, with specific high-penetrance genes being preferentially associated with defined cancer syndromes (eg, BRCA1/2 and PALB2 for breast cancer, and MMR genes such as MLH1 and MSH6 for colorectal cancer). Therefore, a targeted and clinically driven approach remains essential for appropriate test selection and interpretation.

Our results highlight two notable observations. First, expected genotype–phenotype concordance was observed for a subset of variants identified within the ASCO gene panel (see section: “Genotype-phenotype correlation in ASCO recommended genes”). For example, established breast cancer susceptibility genes such as ATM, CHEK2 and PALB2^32–34 were detected among patients with breast tumours, while MMR genes (MSH6 and MLH1) were identified in individuals with colorectal cancer.³⁵ Although these associations are well established, their presence in our cohort supports the analytical validity of our CES-based approach and indicates that, when interpreted through an ASCO-filtered framework, CES reproduces the expected gene–disease relationships observed with standard hereditary cancer panels.

Second, CES identified a series of PV/LPVs in genes that are not currently included among those recommended for clinical reporting under ASCO guidelines (see section: “Analysis of non-ASCO filtered gene”), including RAD50, BLM, WRN, PMS1, FANCA, CASP8 and TGFBR2. While the clinical implications of variants in some of these genes remain to be fully defined, their detection illustrates the broader genomic scope of CES and its potential to capture additional candidate susceptibility loci beyond currently recommended panels. For several of these non-ASCO genes we observed phenotype correlations suggestive of biological plausibility (eg, a CASP8 variant³⁶ in a metastatic breast cancer case, and a TGFBR2 missense alteration in a young patient with MSI-related colorectal disease). Although these associations remain provisional, they highlight candidate loci that merit further functional validation and systematic curation.

However, caution is warranted when interpreting these findings. Particular caution is warranted when interpreting POLE variants. Current evidence indicates that hereditary cancer predisposition associated with polymerase proofreading-associated polyposis (PPAP) is primarily linked to specific missense variants affecting the exonuclease domain. In contrast, all POLE variants identified in our cohort were loss-of-function variants, for which the relationship with hereditary cancer susceptibility remains uncertain. Therefore, although these variants were classified according to ACMG/AMP criteria, their role in cancer predisposition should be interpreted cautiously and cannot currently be considered equivalent to established PPAP-associated variants. For several of the identified variants, particularly those occurring in genes typically associated with autosomal recessive cancer predisposition syndromes such as MUTYH, the observed alterations may reflect incidental carrier status rather than a direct causal relationship with the tumour phenotype. Given the relatively high population frequency of heterozygous variants in some of these genes, their presence in patients with cancer does not necessarily imply a pathogenic contribution to disease development and should therefore be interpreted within the broader clinical and genetic context. Additionally, cases exhibiting multiple PVs in the same individual were observed, such as biallelic RUNX1 variants detected in one subject, and instances of dual variants in both SEC23B and BUB1B in another.

The phenomenon of double heterozygosity and the occurrence of unexpected zygosity states contribute to complexities in interpretation, influencing factors like penetrance and phenotypic expressivity. Such cases necessitate thorough family segregation analysis and, where possible, functional assessment to elucidate the implications of these genetic variants. Incidental findings in CES represent an important ethical and clinical challenge, particularly when such findings are unrelated to the indication for testing. Monoallelic PVs were identified in genes typically associated with autosomal recessive cancer predisposition syndromes, including MUTYH, NTHL1 and MSH3. According to current NCCN guidelines, these findings should not be considered causative for the tumour phenotype when present in the heterozygous state, as biallelic alterations are required to confer a significantly increased cancer risk. Therefore, such variants are more appropriately interpreted as incidental carrier status rather than atypical or clinically actionable findings in the context of cancer predisposition. However, their identification may still be relevant for reproductive counselling, as they are associated with autosomal recessive conditions and could have implications for offspring risk. The additional variants identified outside established clinical panels primarily involve genes lacking validated surveillance recommendations or definitive gene–disease associations. Consequently, these findings are currently not directly applicable to routine clinical management and should be interpreted cautiously within a research-oriented framework.

However, these variants are recognised as a potential resource; as evidence further develops and the process of variant curation evolves, some of these findings may eventually shift from purely research-based insights to actionable clinical knowledge. Nonetheless, these variants are recognised as a potential resource; as evidence further develops and the process of variant curation evolves, some of these findings may eventually shift from purely research-based insights to actionable knowledge.

To facilitate this transition and address the translational gap, this work underscores the necessity of implementing structured reanalysis pipelines and promoting data sharing through variant databases. Additionally, participation in international curation consortia is emphasised as vital for the advancement of genomic medicine.

Practical implications for diagnostics are also mentioned, particularly in relation to high-risk patients. For those who yield negative results on initial first-tier panels yet exhibit significant personal or family histories (criteria detailed in the Methods section), it is proposed that CES serves as a reasonable second-tier approach. This strategy supports the ongoing refinement of genetic testing processes to better serve patients in critical healthcare contexts.

Wherever CES is employed, the integration of both pre-test and post-test genetic counselling is crucial, as outlined in our study protocol. This counselling serves to establish expectations, discuss possible outcomes particularly VUS and incidental findings and to devise strategies for cascade testing. VUS presents significant challenges in interpretation, particularly in the context of family cancer histories. In situations where there is a notable familial predisposition to cancer, segregation analysis could provide valuable insights, potentially leading to a reclassification of the VUS as likely pathogenic or likely benign. Despite this potential for clarification, the clinical interpretation of VUS remains challenging. Although segregation analysis can contribute to variant classification under ACMG/AMP criteria (eg, PP1 or BS4), its practical application in clinical settings is often limited by the availability of informative family members and by the complexity of interpreting segregation patterns in small pedigrees.³⁷ Consequently, VUS frequently remain unresolved despite additional family studies. To ensure ongoing monitoring and updates regarding their VUS status, patients are advised to contact our laboratory every 6 months.

This regular engagement facilitates the continuous evaluation of their genetic variants, allowing patients to stay updated on any alterations or new discoveries regarding their genetic data.

Unlike targeted panels, CES offers the advantage of retrospective data reanalysis, enabling the discovery of novel gene–disease associations over time. This adaptability allows CES to function as a long-term genomic archive, which can lead to more accurate molecular diagnoses for families who may have previously received negative results. Additionally, the genotype–phenotype correlations identified in the assessed cohort bolster the biological plausibility of various novel associations, thereby enhancing the clinical implications of this genomic approach. Unlike targeted panels, CES offers the advantage of retrospective data reanalysis enabling the discovery of novel gene–disease associations over time.³⁸ Moreover, the genotype–phenotype correlations observed in our cohort reinforce the biological plausibility of several novel associations, thereby supporting the clinical implications of this genomic approach. Economically, the use of CES is becoming more sustainable due to a significant reduction in sequencing costs. Although there is a slight rise in laboratory expenses in comparison to extensive multigene panels, this rise is justified by the considerable diagnostic and research benefits it offers. Particularly noteworthy is the potential for future reanalysis of samples without necessitating additional sampling, which enhances the long-term value and applicability of CES in various scientific and clinical contexts.

Looking forward, the continued evolution of sequencing technologies may further improve the diagnostic evaluation of HCSs. In selected cases with strong clinical suspicion and negative clinical or whole exome sequencing results, whole genome sequencing could represent a valuable next step, as it enables the detection of non-coding variants, structural rearrangements, and other genomic alterations that may not be captured by CES or WES approaches.^{39 40} Nevertheless, several limitations persist. In the interpretation of VUS and variants located in genes not yet included in clinical guidelines. These challenges present both significant ethical and practical concerns. To address these issues, it is critical to engage in meticulous multidisciplinary evaluations and provide unnecessary anxiety or misinterpretation of the findings. Additionally, fostering collaborative efforts for data sharing and variant curation is essential to close the existing gap between discovery and clinical translation. In summary, our findings suggest that CES can expand the exploratory genomic assessment of hereditary cancer susceptibility in selected BRCA1/2-negative families. Although many of the identified gene variants, particularly those outside ASCO and NCCN guidelines, are not yet clinically applicable, the additional findings obtained through CES may represent a useful resource for future research and reinterpretation as evidence evolves. At present, our data support the role of CES primarily as a second-tier exploratory approach in carefully selected high-risk patients rather than as a broadly applicable first-line diagnostic strategy.

Data availability statement

Data are available upon reasonable request.