In Silico Prediction of BRCA1 and BRCA2 Variants with Conflicting Clinical Interpretation in a Cohort of Breast Cancer Patients

Germline BRCA1/2 alteration has been linked to an increased risk of hereditary breast and ovarian cancer syndromes. As a result, genetic testing, based on NGS, allows us to identify a high number of variants of uncertain significance (VUS) or conflicting interpretation of pathogenicity (CIP) variants. The identification of CIP/VUS is often considered inconclusive and clinically not actionable for the patients’ and unaffected carriers’ management. In this context, their assessment and classification remain a significant challenge. The aim of the study was to investigate whether the in silico prediction tools (PolyPhen-2, SIFT, Mutation Taster and PROVEAN) could predict the potential clinical impact and significance of BRCA1/2 CIP/VUS alterations, eventually impacting the clinical management of Breast Cancer subjects. In a cohort of 860 BC patients, 10.6% harbored BRCA1 or BRCA2 CIP/VUS alterations, mostly observed in BRCA2 sequences (85%). Among them, forty-two out of fifty-five alterations were predicted as damaging, with at least one in silico that used tools. Prediction agreement of the four tools was achieved in 45.5% of patients. Moreover, the highest consensus was obtained in twelve out of forty-two (28.6%) mutations by considering three out of four in silico algorithms. The use of prediction tools may help to identify variants with a potentially damaging effect. The lack of substantial agreement between the different algorithms suggests that the bioinformatic approaches should be combined with the personal and family history of the cancer patients.


Introduction
Breast cancer (BC) represents the most frequent malignancy worldwide and the most lethal in women, with 2 million new cases diagnosed each year [1][2][3].In addition to the biological features, including BC subtypes, tumor grading and other transversal biomarkers that directly and independently correlate with tumor aggressiveness, inherited genetic alteration plays a role in BC hereditary predisposition [4].Different epidemiological studies identified a set of eight genes (ATM, BARD1, BRCA1, BRCA2, CHEK2, PALB2, RAD51C and RAD51D) mostly responsible for hereditary BC.Among these, the two main susceptibility genes correlated with a higher risk to develop breast and ovarian carcinoma are BRCA1 and BRCA2 [5][6][7][8].
Germline BRCA1 and BRCA2 alterations increase the probability to develop BC and other tumor types, including ovarian, pancreatic, prostate, colorectal cancer and melanoma [9].Over the years, the availability of next generation sequencing (NGS) technologies led to an increase of BRCA1 and BRCA2 genetic testing requests to improve diagnoses, prognostic information, research and clinical practice.The currently accepted method for the BRCA1 and BRCA2 variants' classification is based on the Evidence-based Network for the Interpretation of Germline Mutant Alleles (ENIGMA) consortium classification, supported by the ClinVar database, according to the International Agency for Research on Cancer (IARC)'s recommendations [10,11].This classification system includes five classes, as follows: (i) benign (class I); (ii) likely benign (class II); (iii) variant of uncertain significance (VUS, class III); (iv) likely pathogenic (class IV) and (v) pathogenic (PV, class V).A VUS or conflicting interpretation of pathogenicity (CIP) variant consists of an alteration in the gene sequence, with unknown consequences on the function of the gene product or on the potential risk for disease development.The identification of VUS is evaluated as inconclusive and clinically not actionable for the patients' and unaffected carriers' management.
To date, the frequency of VUS reporting is about 10-20% in women who have undergone BRCA analysis [12,13].This frequency depends on the testing prevalence and/or population ancestry [14].Several studies reported a VUS frequency of 21% in the African-American population, 5-6% in the European ancestry population and up to 15% in European subjects [15][16][17].
The significant rate of CIP/VUS retrieval furthers a claim for an alternative approach of their annotation or re-classification as benign or pathogenic alterations.With the advances of bioinformatics biology, different groups have improved high-throughput pipelines and in silico tools to identify functionally germline CIP/VUS alterations [18][19][20][21].Among them, Polymorphism Phenotyping v.2 (PolyPhen-2) [22] and Sorting Intolerant From Tolerant (SIFT) [23] are two algorithms used to assess the functional impact of missense alterations, whereas the Mutation Taster (MT) [24] and Protein Variation Effect Analyzer (PROVEAN) [25] can be utilized also to improve potential deleterious effects of synonymous or intronic and indel mutations.
In this study, we evaluated the in silico predictions of BRCA1 and BRCA2 CIP/VUS alterations in a cohort of BC patients, according to the ENIGMA or ClinVar database, using PolyPhen-2, SIFT, MT and PROVEAN tools.By doing so, we investigated whether these algorithms could predict the clinical effect and significance of CIP/VUS alterations, eventually impacting the clinical management of subjects diagnosed with BC.

Study Population
A retrospective collection of molecular data from a total of 1027 patients with breast and ovarian cancer, melanoma, pancreatic tumor or prostate carcinoma was carried out at the "Center of Experimental Oncology and Hematology" of the Hospital Policlinico "G.Rodolico-San Marco" of Catania from January 2017 to May 2023.The patients were referred to our Molecular Diagnostics Laboratory for BRCA1 and BRCA2 genetic testing.The present study was conducted in accordance with the Declaration of Helsinki and in accordance with the local legislation and institutional review board.All participants provided their written informed consent before undergoing molecular analysis and to participate in this study.

DNA Extraction from Patients
Each patient provided 20 milliliters (mL) of peripheral blood within a single blood draw, which was collected into EDTA tubes (BD Biosciences, Franklin Lakes, NJ, USA).After the blood draw, fenomic DNA was isolated from 0.7 mL of whole blood samples using the Qiasymphony DSP DNA Midi kit Isolation Kit (QIAGEN, Hilden, Italy), and quantified using a Qubit ® 3.0 or Qubit ® 3.0 fluorometer (Thermofisher Scientific, Waltham, MA, USA), according to the manufacturer's instructions.

Next Generation Sequencing Analysis for BRCA1/2 Genes
Target enrichment and library preparation was performed using an Oncomine™ BRCA Research Assay Chef.After the preparation, the samples were loaded into an Ion Am-pliSeq™ Chef Reagents DL8 cartridge for automated libraries preparation, as previously published [7].The kit contains two multiplex PCR primer pools, proficient to study all BRCA1 (NM_007300.3) and BRCA2 (NM_000059.3)genes.In brief, 10 ng of each DNA sample were placed in the barcode plate for library preparation.The plate, with all reagents and consumables, was loaded on the Ion Chef™ Instrument.An automated library preparation and pooling of barcoded sample libraries was then performed on the Ion Chef™ Instrument.The quantity of pooled libraries was evaluated using a Qubit ® 3.0 or 4.0 fluorometer.Lastly, the pooled libraries were mixed in an equimolar ratio in the Ion Chef™ Library Sample Tube and loaded onto the Ion Chef™ Instrument.The sequencing was performed with an Ion 510 or 520 Chip, using an Ion Gene Studio S5 Plus System instrument (Thermofisher Scientific, Waltham, MA, USA).Analysis of data was performed with Amplicon Suite (SmartSeq s.r.l., Alessandria, Italy) and Ion Reporter Software (v.5.20.2.0).

Data Analysis and Genetic Classification
All alterations' nomenclature followed current guidelines of the Human Genome Variation Society, available online (HGVS, http://www.hgvs.org/mutnomen,accessed on 11 July 2024).The clinical significance of BRCA1 and BRCA2 alterations was characterized using the classification of the consortium ENIGMA (Evidence-based Network for the Interpretation of Germline Mutant Alleles, https://enigmaconsortium.org/, accessed on 11 July 2024), and by consulting several databanks, such as ClinVar, ARUP, BRCA Exchange, IARC_LOVD and UMD.The classification includes the following five distinct classes of risk: benign (class I), likely benign (class II), variant of uncertain significance (VUS, class III), likely pathogenic (class IV) and pathogenic (class V).The alterations were classified with a conflicting interpretation of pathogenicity, and variants of uncertain significance were defined as CIP/VUS, respectively.The effect of alterations on protein structure and function was also analyzed by VarSome, an informatic tool that allows access to about 30 databases, as previously described [7,26].

Sanger Sequencing
The presence of each CIP/VUS mutation was verified by Sanger sequencing.A specific primer forward and reverse were designed for each of the detected alterations by using the BRCA1 and BRCA2 gene reference sequences (NG_005905.2,NM_007294.3and NG_012772.3,NM_000059.3,respectively).After, a PCR specific to the target sequence was performed, followed by Sanger sequencing.
PROVEAN is a bioinformatic tool for the screening of alterations in order to determine an amino acid substitution, nonsynonymous or indel variants that are predicted to have an impact on protein function.
PolyPhen-2 is a tool which predicts the effect of the amino acid change on a human protein function and structure.The algorithm identifies protein ID numbers and entry names using the UniProtKB database.The tool generates a report in which the alteration may be classified into three categories, as follows: benign, possibly damaging and probably damaging.A result of "benign" (score ≤ 0.5) suggests that the mutation is not likely to affect protein function.A result of "possibly damaging" (0.5 < score ≤ 0.9) points to a possible effect on protein function.A result of "probably damaging" (score > 0.9) indicates that the alteration is likely to impact protein function.The algorithm uses an amino acid alteration or a protein sequence as input, then performs a BLAST search to recognize homologous sequences and generates scores.The variants are classified into a functional category, either deleterious or neutral, based on a pre-set threshold, with a default threshold value fixed at −2.5.Therefore, variants with scores below this threshold are categorized as deleterious.
MutationTaster is an online tool able to investigate the potential effect of DNA sequence variants on the gene products.The algorithm works on both protein and DNA level by testing substitutions of single amino acids, as well as synonymous or intronic variant.The alteration is predicted as one of four possible types: (i) disease causing (probably deleterious), (ii) disease causing automatic (variation known to be deleterious), (iii) polymorphism (probably neutral) and (iv) polymorphism automatic (known to be neutral).
SIFT is a bioinformatics program that predicts whether an amino acid alteration may have an impact on protein function.The software analyzes nonsynonymous polymorphisms and missense mutations based on sequence homology and the physical properties of amino acids.Therefore, SIFT chooses related proteins and elaborates an alignment of these proteins with the query.Finally, the software estimates the probability that the toleration of an amino acid at a position is conditional on the most frequent amino acid being tolerated.The score can range from 0 to 1, and substitutions with scores < 0.05 are considered deleterious.
The CIP/VUS alterations predicted as "possibly/probably damaging/intolerant" were classified as a "damaging" variant, whereas the mutations predicted as "benign moderate or strong/tolerant" were classified as a "neutral" variant.

Population Characteristics
A total of 1027 patients were screened for germline BRCA1/2 mutations between January 2017 and May 2024.Molecular analyses were performed at the Center of Experimental Oncology and Hematology of the Hospital Policlinico "G.Rodolico-San Marco" of Catania, according to Sicilian guidelines (http://www.gurs.regione.sicilia.it/Indicep1.htm,accessed on 11 July 2024).Among the 1027 recruited, 860 subjects had breast cancer, 48 had ovarian cancer, 60 had pancreatic cancer, 47 had prostate cancer and 12 had melanomas.Patient distribution according to cancer type and data results is shown in Figure 1.

Description and Localization of BRCA1 and BRCA2 CIP/VUS Variants
Next, we evaluated the type of BRCA1 and BRCA2 CIP/VUS variants.Among the ninety-two patients harboring CIP/VUS, two subjects presented the same alteration (c.5333A > G) in the BRCA1 sequence; therefore, we found ten different BRCA1 CIP/VUS alterations.Nine of these alterations were single nucleotide variants (SNVs), whereas only one was a deletion (c.4063_4065delAAT) (Table 1).

Description and Localization of BRCA1 and BRCA2 CIP/VUS Variants
Next, we evaluated the type of BRCA1 and BRCA2 CIP/VUS variants.Among the ninety-two patients harboring CIP/VUS, two subjects presented the same alteration (c.5333A > G) in the BRCA1 sequence; therefore, we found ten different BRCA1 CIP/VUS alterations.Nine of these alterations were single nucleotide variants (SNVs), whereas only one was a deletion (c.4063_4065delAAT) (Table 1).Concerning BRCA2, the genetic analysis showed a total of 55 different CIP/VUS mutations in 81 individuals.Furthermore, it is of interest that the c.3509C > T alteration was found in five subjects, the c.7871A > G mutation was found in seven patients, four individuals presented the c.8850G > T variation, the c.9586A > G variant was found in three patients and the c.9613_9614delGCinsCT del/ins was found in three subjects.Each of the other eight alterations (c.28A > G, c.442T > C, c.599C > T, c.1054T > C, c.3070A > G, c.4769A > G, c.5870T > C and c.9698G > A) were attained in two individuals.Moreover, in the case of the BRCA2 sequence, fifty-three of fifty-five alterations were SNVs, while only one was a deletion (c.9052_9057delAGTAAA) and another one a deletion/insertion (c.9613_9614delGCinsCT) (Table 2).Specifically, the c.7871A > G alteration at the protein level, found in the majority of our patients, results in the change of a tyrosine to a cysteine located in the helical domain.It is of interest that the BRCA2 c.9052_9057delAGTAAA alteration deletes six nucleotides from exon 23 of the BRCA2 mRNA, and it is predicted to result in an in-frame deletion.Furthermore, the c.9613_9614delGCinsCT alteration replaces alanine with leucine at the codon 3205 of the BRCA2 protein.The alanine amino acid is moderately conserved, and it has similar physiochemical characteristic to leucine.
Subsequently, we mapped the BRCA1 and BRCA2 CIP/VUS alterations through the proteins' binding regions and functional domain (Figure 3).In the BRCA1 gene, the 10 CIP/VUS variants were distributed along the entire gene sequence.Among these, the c.4063_4065delAAT deletion was located inside the coiled-coil domain, which binds to the WD 40 domain of the PALB2 gene.Among the BRCA2 gene, 30.1% variants were located in the BCCR regions and 36.4% of CIP/VUS alterations in the OCCRs.For BRCA2 protein, we found that two CIP/VUS alterations (c.7479G > A, c.7871G > A) were mapped in the DNA Binding Domain.
Next, we wanted to evaluate the agreement rate of the four in silico tools.To this end, we focused this analysis only on the 55 BRCA2 CIP/VUS alterations, due to the larger sample size.The analysis of all CIP/VUS variants by the four approaches showed that 42/55 alterations were predicted as damaging at least one in silico used tool.Prediction agreement was achieved in 25/55 (45.5%)CIP/VUS mutations, with 23 (92%) and 2 (8%) CIP/VUS alterations predicted as neutral and damaging, respectively (Figure 5).Lasty, we assessed the consensus in prediction outcomes for the 42 tested CIP/VUS alterations with an estimated damaging effect.We found a concordance for all four in silico tools in two alterations (c.7871A > G, c.8939C > A) (Figure 6).Moreover, the highest consensus was obtained in 12/42 (28.6%) mutations by considering three out four in silico tools.In detail, a concordance between SIFT, PolyPhen-2 and MT was observed for eight variants, whereas two CIP/VUS alterations were in agreement with PROVEAN, PolyPhen-2 and MT, or PROVEAN, SIFT and PolyPhen-2.Notably, 17/42 (40.5%) of the predicted variants were unique for being damaging in in silico tools alone (Figure 6).Lasty, we assessed the consensus in prediction outcomes for the 42 tested CIP/VUS alterations with an estimated damaging effect.We found a concordance for all four in silico tools in two alterations (c.7871A > G, c.8939C > A) (Figure 6).Moreover, the highest consensus was obtained in 12/42 (28.6%) mutations by considering three out four in silico tools.In detail, a concordance between SIFT, PolyPhen-2 and MT was observed for eight variants, whereas two CIP/VUS alterations were in agreement with PROVEAN, PolyPhen-2 and MT, or PROVEAN, SIFT and PolyPhen-2.Notably, 17/42 (40.5%) of the predicted variants were unique for being damaging in in silico tools alone (Figure 6).

Discussion
Accurate classification of molecular alterations with regard to their pathogenic potential is of pivotal importance, especially for patients presenting a familial cancer history.Recently, the development of high-throughput technologies determined

Discussion
Accurate classification of molecular alterations with regard to their pathogenic potential is of pivotal importance, especially for patients presenting a familial cancer history.Recently, the development of high-throughput technologies determined unprecedented progress in BRCA1 and BRCA2 genetic testing.To this aim, the molecular analysis of BRCA1 and BRCA2 allows the identification of patients with high probability to develop different tumor types, including breast, ovarian and prostate cancer, among others.At the present time, approximately 20.000 BRCA1/2 variants have been recognized and categorized according to the American College of Medical Genetics [29] and ENIGMA systems [7,30,31].With the recent increase in the use of NGS technology, the spectrum of missense and spicing alteration, described as CIP and VUS alterations, were increased in the BRCA1 and BRCA2 sequence by a genetic test.It was estimated that up to 15% of VUSs were identified among European subjects, with higher rates in African-American and Hispanic populations [15,16,32].To this regard, the useful approaches that measure the impact of CIP or VUS on biological processes represent an alternative method for pathogenic prediction and clinical annotations.
In this study, we used four in silico prediction tools in order to predict the effect and potential significance of the BRCA1 and BRCA2 CIP/VUS alterations for the patient's clinical management.First, we characterized the CIP/VUS alterations according to the ENIGMA and ClinVar databases.Of the 860 BC patients analyzed, 10.6% harbored BRCA1 and BRCA2 CIP/VUS mutations.This frequency was concordant with previous data reported in the literature [16].The highest numbers of CIP/VUS were distributed in the BRCA2 sequence, and similar findings were observed in published studies [33][34][35].Our analysis showed a CIP/VUS landscape that is differently distributed in the studied population.Indeed, while some CIP/VUSs were detected only in a single subject, others were repeated in more individuals, such as the c.3509C > T and c.7871A > G, which were observed in two different groups of seven patients.For c.3509C > T, although this variant is still defined as a CIP mutation, several findings, including co-occurrence, have a lack of segregation with the disease of association in case-control studies, thus supporting its classification as benign [36][37][38][39].Moreover, it occurs at a poorly conserved position in BRCA protein.Concerning the c.7871A > G alteration, it was previously described in a cohort of breast and ovarian cancer.At the protein level, it causes the substitution of a conservative amino acid that results in a reduced stability of BRCA2 protein [40,41].Furthermore, two del/ins were identified in the BRCA2 gene.The first is represented by the deletion of c.9052_9057delAGTAAA, observed in a subject and previously reported in individuals with personal or family history of cancer related to BRCA [42].However, to date, its clinical significance remains unclear.The second del/ins is the c.9613_9614delGCinsCTdel/ins, also previously detected in both breast and prostate cancer subjects [43][44][45].This variant rises at a non-conserved position and in a domain of unknown function.
Subsequently, we mapped the BRCA1 and BRCA2 CIP/VUS alterations along the functional domains of the protein and the gene putative OCCR and BCCR regions, which were previously defined by Rebbeck et al. [46].These regions were defined as dangerous for developing breast and ovarian cancer, respectively.With regards to the gene localization of CIP/VUS, the c.4063_4065delAAT deletion interested the coiled-coil domain of the BRCA1 protein.This domain is crucial for the interaction with PALB2 that binds to BRCA2 and acts by recruiting BRCA2 to the chromatin and promoting the homologous recombination.Loss of this interaction results in an increase of tumorigenesis [47].The BRCA2 CIP/VUSs were distributed along the entire sequence.Among them, the c.7479G > A alteration is mapped in the DNA Binding Domain (DBD), as required for homologous recombination [48].This domain is located in the C-terminal region encompassed Helical domain, Tower domain and three OB folds.The DBD and RAD51 interacting sites of BRCA2 are able to promote the assembly of RAD51 toward single-strand DNA and single-strand DNA/double-strand DNA connection [49].
To understand whether the CIP/VUS variants may potentially be damaging, we interrogated the four in silico prediction tools, PROVEAN, SIFT, PolyPhen-2 and Mutation Taster.Since CIP/VUSs are mostly a change of amino acid with residues with the same properties or in-frame ins/del alteration, their impact on protein function is often complex for the interpreter.The in silico interrogation is predicted as damaging few of the CIP/VUS alterations.Although a similar performance was observed, the PolyPhen-2 and SIFT tools were mostly predicted to damage alterations (47.6% and 39%), compared to the other two in silico approaches (MT = 26.1% and PROVEAN = 19%).Interestingly, both the c.7871A > G and c.9839C > A mutations were predicted to have a damaging effect by the four algorithms.The c.7871A > G alteration has been described in several breast and ovarian cancer patients.This mutation mapped at codon 2624 of the BRCA2 protein (p.Tyr2624Cys) in the helical domain.This domain binds the 70 amino acids deleted in split-end/split food syndrome (DSS1) protein.This variation results in a significant decrease in structural stability, suggesting a damaging effect on protein function [41].To date, supporting evidence on its function is conflicting, because the clinical significance of this alteration is still unclear [40,50,51].Furthermore, the c.9839C > A alteration, located in exon 26 of the BRCA2 sequence, has been shown to modify a poorly conserved amino acid.Its clinical interpretation is still controversial, and a functional test suggested its neutral effect on protein function, whereas in silico approaches predicted a damaging impact, as previously reported [31,52,53].
A limitation of the current study was the small size of the dataset of CIP/VUS alterations detected in the BRCA1 sequence.Consequently, we focused our next analysis on BRCA2 CIP/VUS mutations.When we considered the four approaches to predict the effect of the variants on the clinical outcome, a low consensus was achieved for the CIP/VUSs classified as damaging (8%), compared to neutral.The absence of agreement between the in silico tools can generate confusion in the interpretation of these variants.To this regard, there are studies with controversial results.Some of them suggested that the combination of different in silico algorithms may help to improve the prediction performance.Other researchers reported opposite data [17,54,55].In our study, a higher concordance was obtained (28.6%) when we used three of four in silico tools, suggesting that this approach may be more efficient than the previous one.

Conclusions
The inherent uncertainty related to the nature of CIP/VUS does not allow us to classify these alterations as deleterious or benign, potentially influencing patients' management.In this context, the use of bioinformatic in silico tools may help to identify variants with a potentially damaging effect.Still, the lack of a substantial agreement between the different algorithms suggests that these bioinformatic approaches should be combined with an accurate assessment of the family and clinical history of the proband.Institutional Review Board Statement: The present study was conducted in accordance with the Declaration of Helsinki and in accordance with the local legislation and institutional review board.All participants provided their written informed consent before undergoing molecular analysis and to participate in this study.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
interpretation of pathogenicity; DEL: deletion; ENIGMA: evidence-based network for the interpretation of germline mutant alleles; HGVS: human genome variant society; NI: not investigated; NYR: not yet reviewed; VUS: variant of uncertain significance.

Figure 3 .
Figure 3. Representations of BRCA1 and BRCA2 proteins and localization of conflicting interpretations of pathogenicity or variants of uncertain significance.The figure shows the distribution of BRCA1 (A) and BRCA2 (B) CIP/VUS amino acid changes in breast cancer patients.The bar heights show the number of cases.BRCA1 and BRCA2 proteins are painted with their functional domains.(A) The BRCA1 protein includes a RING domain (RING) and a nuclear localization sequence (NLS), a coiled-coil domain, an SQ/TQ cluster domain (SCD) and BRCA1 Cterminus domains (BRCT).(B) The BRCA2 protein contains 8 BRC repeats, a DNA binding domain with a helical domain (Helical), a 3 oligonucleotide/oligosaccharide binding (OB) fold, a tower

Figure 3 .
Figure 3. Representations of BRCA1 and BRCA2 proteins and localization of conflicting interpretations of pathogenicity or variants of uncertain significance.The figure shows the distribution of BRCA1 (A) and BRCA2 (B) CIP/VUS amino acid changes in breast cancer patients.The bar heights show the number of cases.BRCA1 and BRCA2 proteins are painted with their functional domains.(A) The BRCA1 protein includes a RING domain (RING) and a nuclear localization sequence (NLS), a coiled-coil domain, an SQ/TQ cluster domain (SCD) and BRCA1 C-terminus domains (BRCT).(B) The BRCA2 protein contains 8 BRC repeats, a DNA binding domain with a helical domain (Helical), a 3 oligonucleotide/oligosaccharide binding (OB) fold, a tower domain (T) and an NLS at the C-terminus.Regions referred to as breast cancer cluster regions (BCCRs) and ovarian cancer cluster regions (OCCRs) are indicated at the bottom of the figure.CIP = conflicting interpretations of pathogenicity; VUS = variants of uncertain significance.

Figure 4 .
Figure 4. Distribution of in silico prediction outcomes on BRCA1 and BRCA2 CIP/VUS alterations.The histograms show the distribution of CIP/VUS alterations in BRCA1 (A) and BRCA2 (B) genes predicted as neutral (light blue) and damaging (light red) by PROVEAN, SIFT, PoliPhen-2 and MT.The number reported inside the box indicated the number of CIP/VUS alterations obtained by each in silico tools analysis.CIP = conflicting interpretations of pathogenicity; MT = Mutation Taster; PROVEAN = Protein Variation Effect Analyzer; SIFT = Sorting Intolerant From Tolerant and VUS = variants of uncertain significance.

Figure 6 .
Figure 6.Concurrence in prediction outcomes between PROVEAN, SIFT, PoliPhen-2 and MT for BRCA2 CIP/VUS alterations.Venn diagram represents the in silico prediction on 42 CIP/VUS alterations with predicted damaging effects.Variants were interrogated by PROVEAN (light yellow), SIFT (light blue), PoliPhen-2 (light red) and MT (light pink).Number inside the circle indicated the number of alterations with an agreement prediction effect.CIP = conflicting interpretations of pathogenicity; MT = Mutation Taster; PROVEAN = Protein Variation Effect Analyzer; SIFT = Sorting Intolerant From Tolerant; VUS = variants of uncertain significance.

Figure 6 .
Figure 6.Concurrence in prediction outcomes between PROVEAN, SIFT, PoliPhen-2 and MT for BRCA2 CIP/VUS alterations.Venn diagram represents the in silico prediction on 42 CIP/VUS alterations with predicted damaging effects.Variants were interrogated by PROVEAN (light yellow), SIFT (light blue), PoliPhen-2 (light red) and MT (light pink).Number inside the circle indicated the number of alterations with an agreement prediction effect.CIP = conflicting interpretations of pathogenicity; MT = Mutation Taster; PROVEAN = Protein Variation Effect Analyzer; SIFT = Sorting Intolerant From Tolerant; VUS = variants of uncertain significance.

Table 1 .
BRCA1 gene variants with conflicting clinical interpretations of pathogenicity.

Table 1 .
BRCA1 gene variants with conflicting clinical interpretations of pathogenicity.
CIP: conflicting interpretation of pathogenicity; DEL: deletion; ENIGMA: evidence-based network for the interpretation of germline mutant alleles; HGVS: human genome variant society; NI: not investigated; NYR: not yet reviewed; VUS: variant of uncertain significance.

Table 2 .
BRCA2 gene variants with conflicting clinical interpretations of pathogenicity.