clinical & basic research Sultan Qaboos University Med J, May 2015, Vol. 15, Iss. 2 pp. e218–225, Epub. 28 May 15 Submitted 11 Sep 14 Revision Req. 4 Nov 14; Revision Recd. 13 Nov 14 Accepted 3 Dec 14 Department of Diagnostic Genetics, LabPLUS, Auckland City Hospital, Auckland, New Zealand *Corresponding Author e-mail: donaldl@adhb.govt.nz BRCA2 و BRCA1 توقع امكانية حدوث األمراض للمتغريات اجلينية ل احملددة يف الفحص اجليين السريري كلري بروك�ض، �ستيال الي، الني دوروثي، دونالد لوف abstract: Objectives: Missense variants are very commonly detected when screening for mutations in the BRCA1 and BRCA2 genes. Pathogenic mutations in the BRCA1 and BRCA2 genes lead to an increased risk of developing breast, ovarian, prostate and/or pancreatic cancer. This study aimed to assess the predictive capability of in silico programmes and mutation databases in assisting diagnostic laboratories to determine the pathogenicity of sequence-detectable mutations. Methods: Between July 2011 and April 2013, an analysis was undertaken of 13 missense BRCA gene variants that had been detected in patients referred to the Genetic Health Services New Zealand (Northern Hub) for BRCA gene analysis. The analysis involved the use of 13 in silico protein prediction programmes, two in silico transcript analysis programmes and the examination of three BRCA gene databases. Results: In most of the variants, the analysis showed different in silico interpretations. This illustrates the interpretation challenges faced by diagnostic laboratories. Conclusion: Unfortunately, when using online mutation databases and carrying out in silico analyses, there is significant discordance in the classification of some missense variants in the BRCA genes. This discordance leads to complexities in interpreting and reporting these variants in a clinical context. The authors have developed a simple procedure for analysing variants; however, those of unknown significance largely remain unknown. As a consequence, the clinical value of some reports may be negligible. Keywords: Genes, BRCA1; Genes, BRCA2; HBOC Syndrome; In Silico. امل�سببة والطفرات .BRCA2 و BRCA1 جينات يف الطفرات فح�ض عند �سائعة ب�سورة املغلوطة املتغريات تكت�سف امللخ�ص: الهدف: لالأمرا�ض يف جيني BRCA1 و BRCA2 توؤدي ايل زيادة خطورة حدوث �رشطانات الثدي واملباي�ض والربو�ستاتا والبنكريا�ض. وهدفت هذه الدرا�سة ايل تقييم القدرة التنبوؤية لقواعد بيانات وبرامج "انسيليكو" مل�ساعدة املعامل الت�سخي�سية يف حتيد قدرة الطفرات الت�سل�سلية علي الت�سبب يف االأمرا�ض. الطريقة: مت حتليل 13 متغري مغلوط يف جينات ال BRCA يف الفرتة مابني يوليو 2011 ايل ابريل 2013 يف عينات برنامج 13 التحليل يف وا�ستخدم .BRCA ال جينات لتحليل )Northern Hub( النيوزيالندية ال�سحية اخلدمات ايل املحولة املر�سي "انسيليكو" التنبوؤى للربوتني، واثنني برنامج "انسيليكو" للتحليل الن�سي وفح�ض ثالثة قواعد بيانات جلني ال BRCA. النتائج: اأظهرت نتائج هذه تف�سري يف الت�سخي�سية املعامل تواجه التي التحديات يو�سح وهذا املتغريات. ملعظم "انسيليكو" تف�سري يف ملحوظا اختالفا التحليل النتائج. اخلال�صة: لالأ�سف فانه يحدث الكثري من االأختالف عند ت�سنيف املتغريات املغلوطة يف جني ال BRCA باأ�ستخدام قواعد بيانات ال�سبكة العنكبوتية لتنفيذ حتليل "انسيليكو". وقد اأدي هذا االأختالف ايل حدوث تعقيدات كبرية يف تف�سري هذه املتغريات يف ال�سياق ال�رشيري. وقد طور الباحثون طريقة ب�سيطة لتحليل املتغريات ولكن بالرغم من ذلك فانه اليزال هناك الكثري منها غري معلوم الداللة. وبالتايل فاأن القيمة ال�رشيرية لبع�ض هذه التقارير ميكن اإهمالها. مفتاح الكلمات: اجلينات، BRCA1؛ واجلني، BRCA2؛ متالزمة HBOC؛ "انسيليكو". Predicting the Pathogenic Potential of BRCA1 and BRCA2 Gene Variants Identified in Clinical Genetic Testing Clare Brookes, Stella Lai, Elaine Doherty, *Donald R. Love Advances in Knowledge - The analysis of sequence-detectable variants in the BRCA1 and BRCA2 (BRCA1/2) genes is critical in establishing if these variants are disease-causing. - The analysis presented here shows the challenges posed by in silico programmes. - Diagnostic laboratories may therefore have to rely on familial segregation studies or the development of better in silico programmes possibly based on advanced neural network modelling requiring phenotypic as well as genotypic data. Appication to Patient Care - The analysis in this study shows the advantages and disadvantages of database searching and in silico analyses in predicting the pathogenicity of gene variants. - In the case of BRCA1/2 gene variants, evolving analytical tools offer an improved outcome for guiding counselling of patients at risk of hereditary breast and ovarian cancer. Clare Brookes, Stella Lai, Elaine Doherty and Donald R. Love Clinical and Basic Research | e219 Pathogenic mutations in the BRCA1 and BRCA2 (BRCA1/2) genes predispose patients to an increased risk of developing breast, ovarian, prostate and/or pancreatic cancer; these genes are two of the genes most commonly tested for cancer predisposition. In the USA, a known pathogenic mutation is detected in approximately 10–15% of patients who undergo sequencing of the entire coding regions of the BRCA genes.1 However, a variant of uncertain significance (VUS) is detected in more than 5% of patients, with higher frequencies seen in less commonly tested ethnic groups.2 Patients with known pathogenic BRCA gene mutations are offered preventative strategies including enhanced surveillance, chemoprevention and irreversible surgical interventions. A study of patients in the USA, surveyed two years after being given either an uninformative (UN) BRCA gene-negative or VUS result by trained genetic counsellors, found that a VUS result did not result in excessive surgeries, exaggerated distress or increased risk perception compared to patients with a UN result.3 The risk- reducing mastectomy rate was 7% in both groups and the oophorectomy rate was 5% for VUS patients and 3% for UN patients.3 A pathogenic mutation refers to a genetic variant that has been shown to cause or contribute to disease. A benign variant does not significantly impact on the function of the protein or increase disease risk, and it includes polymorphisms which are seen in over 1% of the general population. A VUS is a variant where the effect on protein function and disease risk is unknown.4 In the case of the BRCA genes, VUS are largely missense substitutions where a single nucleotide change results in an altered amino acid. The terms, VUS and unclassified variant (UV) are often used interchangeably in the literature; however, they have slightly different interpretations. The term UV is suggestive of an unstudied variant, whereas a VUS may or may not have been studied but still has unknown clinical relevance.5 Providing a clear interpretation of a VUS is a complex challenge for a diagnostic laboratory. Common methods used to predict pathogenicity can include literature and database searches, in silico analyses, segregation analyses and functional studies. The requesting clinician may be faced with the difficult task of deciphering the ambiguity of the VUS and communicating the result to the patient along with clinical recommendations. Furthermore, it is imperative that the classifications of VUS are regularly checked and any changes to their classifications are relayed to the patients and their family. The majority of missense mutations in the BRCA genes are classified as VUS. The exceptions include missense mutations that lie within the highly conserved BRCA1 RING and the BRCA1 carboxyl-terminal domains.6 Known pathogenic missense mutations in the BRCA2 gene are less common but may occur in the DNA-binding domain.7 The difficulty in the interpretation of missense variants in the BRCA genes arises due to the discordance in the classification of variants in the breast cancer databases and the variety of predictions based on in silico analyses. Recently, Lindor et al. used a quantitative posterior probability model to reclassify VUS in the BRCA1/2 genes into five classes as defined by the International Agency for Research on Cancer (IARC) Working Group on Unclassified Genetic Variants.8 These classes range from class 1 (not pathogenic) through to class 5 (definitely pathogenic). This reclassification attempts to combine a range of information regarding each VUS in the literature and convert this into a useful posterior probability. This study analysed nine BRCA1 and four BRCA2 gene missense variants identified in the Diagnostic Genetics LabPLUS, Auckland City Hospital, Auckland, New Zealand, where the interpretation was hampered by the diversity of classifications in international databases and online in silico predictions. Methods This study was carried out between July 2011 and April 2013 and included 20 patients referred to Genetic Health Services New Zealand (Northern Hub) for BRCA1/2 gene mutation screening. DNA was extracted from peripheral blood samples in ethylenediaminetetraacetic acid (EDTA) using the Gentra Puregene DNA Extraction kit (Qiagen GmbH, Hilden, Germany). Genomic DNA from 20 patients were subjected to BRCA1/2 gene sequencing as described elsewhere;9 any identified variants were subsequently confirmed by exon-targeted polymerase chain reaction amplification and bi-directional Sanger-based sequencing.10 Sequence traces were analysed using KB Basecaller Version 1.4 (Applied Biosystems Inc., Foster City, California, USA), on Variant Reporter™ Software Version 1.0 (Applied Biosystems Inc.), with a minimum trace score of 35, which corresponds to an average false base-call frequency of 0.031%. The analysis of sequence data and the subsequent investigation of databases and bioinformatic programmes used the relevant Reference Sequence (RefSeq) transcript, RefSeq protein and Uniprot accession numbers for the BRCA1 (NM_007294.3; NP_009225.1; P38398) and Predicting the Pathogenic Potential of BRCA1 and BRCA2 Gene Variants Identified in Clinical Genetic Testing e220 | SQU Medical Journal, May 2015, Volume 15, Issue 2 BRCA2 (NM_000059.3; NP_000050.2; P51587) genes. All variants were checked for splicing effects using two in silico splice prediction programmes: the Splice Site Prediction by Neural Network online tool of the Berkeley Drosophilia Genome Project and the Alternative Splice Site Predictor (ASSP) tool.11–14 All of the patients included in the study gave informed consent. The New Zealand Multi-Region Ethics Committee has ruled that cases of patient management do not require formal ethical approval from a committee. Results The missense BRCA gene variants identified are shown in Table 1. These variants were checked for pathogenicity in six databases (three of which were specific to the BRCA genes)[Figure 1 and Table 2].8,15‒19 The missense variants were also scored for predic- ted pathogenicity using 13 online in silico protein analysis programmes [Figure 2 and Table 3].20‒43 When all variants were checked for splicing effects using the two aforementioned in silico splice prediction programmes, both of the programmes predicted that each variant would have no effect on splicing (data not shown). Apart from four of the variants, the results of the in silico protein analysis programmes varied depending on which programme was used. The frequency with which variants were predicted to be pathogenic varied significantly between programmes [Table 4].22‒43 A total of 13 missense BRCA gene mutations were identified and only one was identified as probably pathogenic (BRCA1: c.140G>A) based on the combined results achieved from databases and in silico programmes. However, this variant was predicted to be benign using the Polymorphism Phenotyping, Version 2 (PolyPhen-2), HumVar database and the Protein Variation Effect Analyzer (PROVEAN).20,37 In addition, a further three variants appeared to be probably benign (BRCA1: c.2612C>T, c.3548A>G and BRCA2: c.2971A>G). However, the remaining nine variants could not be interpreted even though minor allele frequencies on the Database of Single Nucleotide Polymorphisms (dbSNP) ranged from 0.01 to 0.327.19 Table 1: Missense BRCA gene mutations identified in the DNA of 20 patients Mutation Predicted amino acid change Detection frequency BRCA1 c.140G>A p.Cys47Tyr 0.05 c.1067A>G p.Gln356Arg 0.14 c.2077G>A p.Asp693Asn 0.10 c.2315T>C p.Val772Ala 0.05 c.2612C>T p.Pro871Leu 0.38 c.3113A>G p.Glu1038Gly 0.48 c.3119G>A p.Ser1040Asn 0.05 c.3548A>G p.Lys1183Arg 0.43 c.4837A>G p.Ser1613Gly 0.43 BRCA2 c.865A>C p.Asn289His 0.05 c.1114A>C p.Asn372His 0.52 c.2971A>G p.Asn991Asp 0.05 c.8149G>T p.Ala2717Ser 0.05 Figure 1: Diagrammatic representation of the pathogenicity calls of 13 BRCA gene missense variants in six databases, including the (1) Human Gene Mutation Database Professional 2013;15 (2) Breast Cancer Information Core;16 (3) Universal Mutation Database;17 (4) Leiden Open Variation Database;18 (5) International Agency for Research on Cancer,8 and (6) Database of Single Nucleotide Polymorphisms.19 Clare Brookes, Stella Lai, Elaine Doherty and Donald R. Love Clinical and Basic Research | e221 Table 2: Database listings for BRCA gene missense mutations Variant HGMD class BIC clinically important LOVD summary UMD biological significance IARC class dbSNP dbSNP MAF BRCA1 c.140G>A DM Not listed Not listed 5 – Causal Not listed Not listed - c.1067A>G DP Unknown Mixed 1 – Neutral 1 - not path rs1799950 C = 0.028 c.2077G>A DP No Mixed 1 – Neutral 1 - not path rs4986850 T = 0.039 c.2315T>C DM Unknown Neutral 1 – Neutral 1 - not path rs80357467 - c.2612C>T DFP1 No Mixed 1 – Neutral 1 - not path rs799917 A = 0.483 c.3113A>G DP2 No Mixed 1 – Neutral 1 - not path rs16941 C = 0.303 c.3119G>A DM? Unknown Neutral 1 – Neutral 1 - not path rs4986852 T = 0.012 c.3548A>G DP1 No Mixed 1 – Neutral 1 - not path rs16942 C = 0.324 c.4837A>G DM? No Mixed 1 – Neutral 1 - not path rs1799966 C = 0.327 BRCA2 c.865A>C DP1 No Mixed 1 - Neutral Not listed rs766173 C = 0.058 c.1114A>C DFP Listed as C>A Neutral Listed as C>A Listed as C>A rs144848 C = 0.240 c.2971A>G DM? No Neutral Polymorphism Not listed rs1799944 G = 0.062 c.8149G>T DM? No Neutral 1 - Neutral 1 - not path rs28897747 T = 0.001 HGMD = Human Gene Mutation Database Professional 2013;15 BIC = Breast Cancer Information Core database;16 LOVD = Leiden Open Variation Database;18 UMD = Universal Mutation Database;17 IARC = International Agency for Research on Cancer;8 dbSNP = Database of Single Nucleotide Polymorphisms;19 MAF = minor allele frequency; DM = disease-causing mutation; DP = disease-associated polymorphism; path = pathogenic; DFP = disease-associated polymorphism with additional supporting functional evidence; 1 = associated with a decreased risk; 2 = comments included “polymorphism”; DM? = potential disease-causing mutation. Figure 2: Diagrammatic representation of the pathogenicity calls of 13 BRCA gene missense variants using 13 online in silico analysis programmes (all used in default online mode). These prediction programmes included: both the (1) HumDiv and (2) HumVar predictions of Polymorphism Phenotyping, Version 2;20,21 (3) Mutation Assessor, release 2;22,23 (4) I-Mutant, Version 3.0, for the prediction of disease-associated single point mutations from protein sequence;24,25 (5) MutPred, Version 1.2;26,27 (6) SNPs&GO;28,29 (7) Protein Analysis Through Evolutionary Relationships Evolutionary Analysis of Coding SNPs, Version 6.1;30,31 (8) Align-Grantham Variation Grantham Deviation used with the supplied BRCA1 and BRCA2 alignments;32,33 (9) SNAP;34,35 (10) Predictor of Human Deleterious Single Nucleotide Polymorphisms;36 (11) Protein Variation Effect Analyzer, Version 1.1.3, and Sorting Intolerant from Tolerant;37–39 (12) Sorting Intolerant from Tolerant BLink,40,41 and (13) Mutation Taster.42,43 Predicting the Pathogenic Potential of BRCA1 and BRCA2 Gene Variants Identified in Clinical Genetic Testing e222 | SQU Medical Journal, May 2015, Volume 15, Issue 2 Ta bl e 3: O nl in e in si lic o pr ed ic ti on s* r eg ar di ng th e pa th og en ic it y of B R C A g en e m is se ns e m ut at io ns V ar ia nt P ol yP he n 2 H um D iv P ol yP he n 2 H um V ar M ut A ss I- M ut an t M ut P re d SN P s& G O PA N T H E R A li gn - G V G D SN A P P hD - SN P SI F T P R O V E A N M ut T as Sc or e Sc or e FI s co re R I Sc or e R I su bS PE C C la ss R I, A cc ur ac y R I Se qR ep Sc or e Pr ob ab ili ty B R C A 1 c. 14 0G >A Po ss ib ly 0. 67 7 B en ig n 0. 22 3 H ig h 4. 02 5 D 2 D 0. 98 9 D 10 -4 .6 20 3 C 65 N on -N 6, 9 3% D 7 N ot to le ra te d 0. 23 N -1 .5 88 D is ea se 0. 97 5 c. 10 67 A >G Pr ob ab ly 0. 99 8 Pr ob ab ly 0. 98 8 H ig h 3. 63 5 N 6 N 0. 23 2 D 7 -1 .9 15 24 C 0 N on -N 3, 7 8% D 4 N ot to le ra te d 0. 98 D - 3. 30 2 Po ly 0. 99 9 c. 20 77 G >A B en ig n 0. 00 0 B en ig n 0. 01 0 N eu tr al 0. 55 N 5 N 0. 13 5 D 3 -1 .6 36 02 C 0 N on -N 2, 7 0% N 6 N ot to le ra te d 1. 00 N 0. 03 2 Po ly 0. 99 9 c. 23 15 T >C Po ss ib ly 0. 84 8 Pr ob ab ly 0. 92 8 M ed iu m 2. 63 N 4 D 0. 84 7 D 8 -2 .2 58 07 C 0 N on -N 4, 8 2% D 3 N ot to le ra te d 1. 00 D - 3. 61 2 Po ly 0. 99 9 c. 26 12 C >T B en ig n 0. 00 0 B en ig n 0. 00 0 N eu tr al -3 .3 95 N 4 N 0. 23 2 N 6 -2 .7 18 72 C 0 N 0, 5 3% N 8 To le ra te d 1. 00 N 5. 74 2 Po ly 0. 99 6 c. 31 13 A >G Po ss ib ly 0. 93 6 Po ss ib ly 0. 60 6 M ed iu m 2. 54 N 3 N 0. 19 2 D 2 -2 .4 87 8 C 0 N on -N 4, 8 2% N 3 N ot to le ra te d 0. 98 D -5 .6 87 Po ly 0. 99 9 c. 31 19 G >A Pr ob ab ly 0. 97 4 Po ss ib ly 0. 83 1 M ed iu m 2. 78 5 N 6 N 0. 12 3 D 7 -3 .4 63 89 C 0 N on -N 4, 8 2% N 3 To le ra te d 0. 98 N -1 .6 83 Po ly 0. 99 9 c. 35 48 A >G B en ig n 0. 00 0 B en ig n 0. 00 1 N eu tr al -1 .0 85 N 8 N 0. 09 6 N 6 -0 .8 26 66 C 0 N 4, 8 5% N 7 To le ra te d 1. 00 N 0. 39 8 Po ly 1 c. 48 37 A >G B en ig n 0. 25 5 B en ig n 0. 03 8 Lo w 1. 04 N 5 N 0. 12 7 D 2 -1 .5 78 16 C 0 N on -N 2, 7 0% N 6 N ot to le ra te d 0. 26 N -0 .5 09 Po ly 0. 99 9 B R C A 2 c. 86 5A >C B en ig n 0. 27 8 B en ig n 0. 03 4 Lo w 1. 44 5 N 0 N 0. 19 4 D 10 Er ro r C 0 N on -N 4, 8 2% N 4 To le ra te d 0. 37 N -0 .5 09 Po ly 0. 99 9 c. 11 14 A >C C al le d as H a t 3 72 H re si du e N 4 N 0. 05 8 Er ro r Er ro r Er ro r N on -N 0, 5 8% N 9 To le ra te d 0. 33 N -0 .5 99 Po ly 0. 99 9 c. 29 71 A >G B en ig n 0. 00 0 B en ig n 0. 00 0 N eu tr al -1 .1 5 N 0 N 0. 32 5 D 7 -0 .4 52 25 C 0 N 3, 7 8% N 9 To le ra te d 0. 99 N 2. 12 7 Po ly 0. 99 9 c. 81 49 G >T Po ss ib ly 0. 95 5 Po ss ib ly 0. 76 3 M ed iu m 1. 96 5 D 1 D 0. 78 3 D 9 -2 .3 92 9 C 0 N 1, 6 0% N 8 To le ra te d 0. 37 N -0 .4 41 Po ly 0. 96 1 Po ly Ph en 2 = Po ly m or ph ism P he no ty pi ng , V er sio n 2, p re di ct io n an d sc or e s ho w n; 20 ,2 1 M ut A ss = M ut at io n A ss es so r, re le as e 2 , p re di ct ed F I ( hi gh o r m ed iu m ) o r n on -F I ( lo w o r n eu tr al ) a nd co m bi ne d sc or e i s s ho w n; 22 ,2 3 I -M ut an t = V er sio n 3. 0, p re di ct io n of d is ea se -a ss oc ia te d sin gl e p oi nt m ut at io n fr om p ro te in se qu en ce a nd R I s ho w n; 24 ,2 5 M ut Pr ed = V er sio n 1. 2, p re di ct io n of a n am in o ac id su bs tit ut io n as D o r N a nd sc or e s ho w n; 26 ,2 7 S N Ps & G O = p re di ct io n of h um an d is ea se - re la te d m ut at io ns in p ro te in s w ith fu nc tio na l a nn ot at io ns , p re di ct ed e ffe ct a nd R I s co re fr om 0 (u nr el ia bl e) to 1 0 (r el ia bl e) sh ow n; 28 ,2 9 P A N T H ER = P ro te in A na ly sis Th ro ug h Ev ol ut io na ry R el at io ns hi ps , V er sio n 6. 1, e vo lu tio na ry a na ly sis o f co di ng si ng le n uc le ot id e p ol ym or ph ism s, su bS PE C sc or es o f c on tin uo us v al ue s f ro m 0 (n eu tr al ) t o -1 0 (m os t l ik el y to b e d el et er io us ) s ho w n* *;3 0, 31 A lig n- G V G D = A lig n- G ra nt ha m V ar ia tio n G ra nt ha m D ev ia tio n, u sin g s up pl ie d al ig nm en ts , cl as s s ho w n (c la ss C 65 m os t l ik el y an d C 0 le ss li ke ly ;32 ,3 3 S N A P = pr ed ic tio n of e ffe ct o f n on -s yn on ym ou s p ol ym or ph ism s o n fu nc tio n, R I a nd e xp ec te d ac cu ra cy sh ow n; 34 ,3 5 P hD -S N P = Pr ed ic to r o f H um an D el et er io us S in gl e N uc le ot id e Po ly m or ph ism s, pr ed ic tio n an d RI sh ow n; 36 S IF T = S or tin g I nt ol er an t f ro m T ol er an t B Li nk , t ol er at ed o r n ot to le ra te d pr ed ic tio n an d Se qR ep sc or e s ho w n; 40 ,4 1 P RO V EA N = P ro te in V ar ia tio n Eff ec t A na ly ze r, V er sio n 1. 1. 3, u sin g t he H um an Pr ot ei n Ba tc h to ol , p re di ct io n an d sc or e ( sc or e o f ≤ -2 .5 si gn ifi ed a “d el et er io us ” e ffe ct a nd sc or e o f > -2 .5 si gn ifi ed a “n eu tr al ” e ffe ct ) s ho w n; 37 –3 9 M ut Ta st = M ut at io n Ta st er , p re di ct io n an d pr ob ab ili ty (u sin g P v al ue s, w ith a v al ue cl os e t o 1 in di ca tin g t he h ig h se cu ri ty o f t he p re di ct io n; th e P v al ue u se d he re is n ot th e p ro ba bi lit y of e rr or a s u se d in t- te st st at is tic s) sh ow n; 42 ,4 3 F I = fu nc tio na l i m pa ct ; R I = re lia bi lit y in de x; S eq Re p = fra ct io n of se qu en ce s t ha t c on ta in o ne o f t he b as ic am in o ac id s, w he re a lo w fr ac tio n in di ca te s t he p os iti on is e ith er se ve re ly g ap pe d or u na lig na bl e a nd h as li ttl e i nf or m at io n, a p oo r p re di ct io n is e xp ec te d at th es e p os iti on s; D = d is ea se -a ss oc ia te d; N = n eu tr al ; P ol y = po ly m or ph ism ; H = hi st id in e. *A ll pr og ra m m es w er e u se d in d ef au lt on lin e m od e. ** A sc or e o f - 3 is th e p re vi ou sly id en tifi ed cu to ff po in t f or fu nc tio na l s ig ni fic an ce . Clare Brookes, Stella Lai, Elaine Doherty and Donald R. Love Clinical and Basic Research | e223 Discussion The results presented here illustrate a major problem in interpreting missense BRCA1/2 gene variants. The classifications from various databases and the predictions from a variety of online in silico analysis programmes can vary widely. This highlights the risk of relying on information obtained from just one database or from using only a few in silico programmes when reporting missense variants, as the outcome can affect clinical surveillance and prevention decisions. Of the 13 missense BRCA gene mutations identified, only one was shown to be probably pathogenic, although the same variant was predicted to be benign by the PolyPhen-2 HumVar database and PROVEAN.20,37 Lindor et al. classified nine of the variants in their study as IARC class 1 (not pathogenic).8 Their reclassification uses a model based on prior probabilities derived from evolutionary predictions combined with a likelihood component from segregation information, co-occurrence in ‘trans’, personal and family history and a histopathology profile to give a posterior probability of causality. The outcome of this analysis is based on combining a wide range of information, which is clearly different from the predictions made from individual databases and single in silico programmes, and again highlights the importance of an over-reliance on one source of information to determine the disease causality of a variant. The Clinical Molecular Genetics Society (CMGS) in the UK states in their 2007 guidelines for interpre- ting and reporting UVs that it is unacceptable to rely solely on in silico predictions to assign pathogenicity to a previously unclassified variant.44 Furthermore, the Association for Clinical Genetic Science states in their 2013 practice guidelines for the reporting of sequence variants in clinical molecular genetics that “the classification generated from the prediction tools must not be considered definitive”.45 The American College of Medical Genetics (ACMG) guidelines state that all variants of unknown clinical significance must be included in a laboratory’s report and be followed by an interpretation of their likely clinical significance.46 ACMG recommend categorising uncertain sequence variants as either “previously unreported and of the type which may or may not be causative of the disorder” or “previously unreported and probably not causative of disease”.46 The CMGS 2007 guidelines also state that it is “essential to report all UVs where the clinical significance is uncertain” and furthermore that it is “essential that reports of UVs should be issued to appropriately trained clinicians”.44 The European Molecular Genetics Quality Network’s best practice guidelines for genetic analysis in hereditary breast ovarian cancer recommend that the identification of BRCA gene VUS do not “provide a basis for changing the clinical management of the patient or for offering predictive testing to at risk relatives”.6 The protocol which the authors have established for interpreting BRCA gene missense variants includes: (1) Checking the Breast Cancer Information Core (BIC)and IARC databases;8,16 (2) Checking the dbSNP for classification and minor allele frequency;19 (3) Undertaking splice site predictions using the online Splice Site Prediction by Neural Network and ASSP tools,11,13 and (4) Undertaking in silico protein analysis using the Grantham score PolyPhen SIFT BLink, SNPs&GO and PROVEAN.2,20,28,32,40 In the event that a database search is conflicting, or there is no entry, the authors recommend that dbSNP and splice site/in silico protein analysis programmes are also used. Apart from the BIC and IARC databases, other databases are not as comprehensive, or provide little value in assigning benign/disease-causing status to a missense variant. The in silico programmes use a variety of approaches to achieve a prediction: sequence and Table 4: Percentage of BRCA1 and BRCA2 gene missense variants predicted to be pathogenic using online in silico analysis programmes Programme % predicted to be pathogenic SNPs&GO 83 SNAP 69 PolyPhen - HumDiv 50 SIFT 46 PolyPhen - HumVar 42 MutPred 23 PhD-SNP 23 PROVEAN 23 PANTHER 18 MutAss 17 I-Mutant 15 Align-GVGD 8 MutTas 8 SNP&GO = predicts human disease-related mutations in proteins with functional annotations;28,29 SNAP = predicts effect of non-synonymous polymorphisms on function;34,35 PolyPhen = Polymorphism Phenotyping, Version 2;20,21 SIFT = Sorting Intolerant from Tolerant BLink;40,41 MutPred = Version 1.2, classifies an amino acid substitution as disease-associated or neutral;26,27 PhD-SNP = Predictor of Human Deleterious Single Nucleotide Polymorphisms;36 PROVEAN = Protein Variation Effect Analyzer, Version 1.1.3;37–39 PANTHER = Protein Analysis Through Evolutionary Relationships, Version 6.1;30,31 MutAss = Mutation Assessor programme, release 2;22,23 I-Mutant = Version 3.0;24,25 Align-GVGD = Align-Grantham Variation Grantham Deviation;32,33 MutTast = Mutation Taster.42,43 Predicting the Pathogenic Potential of BRCA1 and BRCA2 Gene Variants Identified in Clinical Genetic Testing e224 | SQU Medical Journal, May 2015, Volume 15, Issue 2 evolutionary conservation-based methods, protein sequence and structure-based methods and machine learning methods. The data from this study support using a number of programmes to achieve a consensus prediction rather than relying on only one programme. The authors suggest that, when results are uncertain, a report of the cascade approach used should be recorded and a detailed work-up should be archived for the clinician to refer to if necessary. The authors recommend that their conclusions are reviewed by clinicians to determine their continuing validity. The predictions have varying levels of confidence, but are considered as an aid to clinical interpretation, although the work described here shows that the value of these predictions may be largely ambivalent at best, or misleading at worst. The authors recommend that the testing of additional family members and a correlation with clinical findings would be helpful to determine the significance of the result. This recommendation for segregation analysis is not entirely fool-proof, especially in light of the predominance of breast cancer in families with BRCA1/2 gene mutations, and that cancer risk may involve an appreciation of familial context rather than a population-based calculation.47 Critically, the authors suggest only accepting referrals from trained genetic counsellors or clinicians with a sufficient understanding of interpreting complicated genetic results. In the event of BRCA gene missense mutations that are reliably benign (stated as such in BIC/IARC databases or when all in silico predictions agree), then these should be relegated to an ancillary table in the report with a footnote indicating how the benign status was determined. This study highlights the complexity of interpreting and reporting missense BRCA1/2 gene variants where the results will be used in genetic counselling, screening and disease prevention. It demonstrates that some BRCA gene missense variants cannot be clearly interpreted with the tools and data available today; however, these variants must be included in laboratory reports so that if future information becomes available regarding their classification then this can be passed on to the patient and their family. This future information could be provided by international developments under the auspices of the Enhancing Neuro Imaging Genetics through Meta Analysis Consortium which is involved in coordinating the development of algorithms for the classification of variants in the BRCA1/2 genes.48 Recent work reported by this consortium has embraced functional assays of BRCA2 gene variants and has attempted to translate functional outcomes into a probability of pathogenicity.49 Conclusion The findings of this study show that there is significant discordance in the classification of some missense variants in the BRCA genes when using online mutation databases and carrying out in silico analyses. This discordance leads to complexities in interpreting and reporting these variants in a clinical context. As such, it is vital that laboratories have agreed guidelines for determining the pathogenicity of a given variant based on a wide range of information and for reporting an uncertain result to the referring clinician. Importantly, the complexity of interpreting and communicating VUS findings highlights the importance of sequencing results being conveyed to patients in a specialist genetic counselling environment. c o n f l i c t o f i n t e r e s t The authors declare no conflicts of interest. References 1. Vallée MP, Francy TC, Judkins MK, Babikyan D, Lesueur F, Gammon A, et al. Classification of missense substitutions in the BRCA genes: A database dedicated to Ex-UVs. Hum Mutat 2012; 33:22–8. doi: 10.1002/humu.21629. 2. Saam J, Burbidge LA, Bowles K, Roa B, Pruss D, Schaller J, et al. Decline in Rate of BRCA1/2 Variants of Uncertain Significance: 2002-2008. Poster. 27th Annual Education Conference of the National Society of Genetic Counselors, Los Angeles, USA, 24–28 Oct 2008. 3. Culver JO, Brinkerhoff CD, Clague J, Yang K, Singh KE, Sand SR, et al. Variants of uncertain significance in BRCA testing: Evaluation of surgical decisions, risk perception, and cancer distress. Clin Genet 2013; 84:464–72. doi: 10.1111/cge.12097. 4. Lindor NM, Goldgar DE, Tavtigian SV, Plon SE, Couch FJ. BRCA1/2 sequence variants of uncertain significance: A primer for providers to assist in discussions and in medical management. Oncologist 2013; 18:518–24. doi: 10.1634/ theoncologist.2012-0452. 5. Iversen ES Jr, Couch FJ, Goldgar DE, Tavtigian SV, Monteiro AN. A computational method to classify variants of uncertain significance using functional assay data with application to BRCA1. Cancer Epidemiol Biomarkers Prev 2011; 20:1078–88. doi: 10.1158/1055-9965. 6. Larsson N, Borg Å, Hodgson S, Sinilnikova O, Loman N, McDevitt T, et al. EMQN Best Practice Guidelines for Molecular Genetic Analysis in Hereditary Breast/Ovarian cancer, 2008. From: www. emqn.org/emqn/Best+Practice Accessed: Oct 2013. 7. Easton DF, Deffenbaugh AM, Pruss D, Frye C, Wenstrup RJ, Allen-Brady K, et al. A systematic genetic assessment of 1,433 sequence variants of unknown clinical significance in the BRCA1 and BRCA2 breast cancer-predisposition genes. Am J Hum Genet 2007; 81:873–83. doi: 10.1086/521032. 8. Lindor NM, Guidugli L, Wang X, Vallée MP, Monteiro AN, Tavtigian S, et al. A review of multifactorial probability-based model for classification of BRCA1 and BRCA2 variants of uncertain significance (VUS). Hum Mutat 2012; 33:8–21. doi: 10.1002/humu.21627. 9. Lai S, Brookes C, Prosser DO, Lan CC, Doherty E, Love DR. Diagnostic screening for mutations in the BRCA1 and BRCA2 genes. Sultan Qaboos Univ Med J 2015; 15:58‒66. Clare Brookes, Stella Lai, Elaine Doherty and Donald R. Love Clinical and Basic Research | e225 10. Doherty E, Marquis-Nicholson R, Brookes C, Love JM, Prosser D, Love DR. From Primer Design to Sequence Analysis: A pipeline tool for use in a diagnostic genetics laboratory. In: Ivanov O, Ed. Applications and Experiences of Quality Control. Rijeke, Bosnia and Herzegorvina: InTech, 2011. Pp. 257–72. doi: 10.5772/612. 11. Berkeley Drosophilia Genome Project. Splice Site Prediction by Neural Network online tool. From: www.fruitfly.org/seq_tools/ splice.html Accessed: Oct 2013. 12. Reese MG, Eeckman FH, Kulp D, Haussler D. Improved splice site detection in Genie. J Comput Biol 1997; 4:311–23. doi: 10.1089/ cmb.1997.4.311. 13. Wang Computing. Alternative Splice Site Predictor (ASSP). From: www.wangcomputing.com/assp/index.html Accessed: Oct 2013. 14. Wang M, Marín A. Characterization and prediction of alternative splice sites. Gene 2006; 366:219–27. 15. BIOBASE Biological Databases. Human Gene Mutation Database (HGMD), Professional 2013.3 Database. From: www.biobase- international.com/product/hgmd Accessed: October 2013. 16. National Human Genome Research Institute. Breast Cancer Information Core (BIC). From: research.nhgri.nih.gov/bic/ Accessed: Oct 2013. 17. Aix Merseille University, Inserm. The Universal Mutation Database (UMD). From: www.umd.be/ Accessed: Oct 2013. 18. Leiden Open Variation Database (LOVD). From: chromium.liacs. nl/LOVD2/cancer/home.php Accessed: Oct 2013. 19. National Center for Biotechnology Information. Single Nucleotide Polymorphism Database (dbSNP). From: www.ncbi.nlm.nih.gov/ projects/SNP/ Accessed: Oct 2013. 20. Department of Genetics, Harvard University, Brigham & Women’s Hospital. Polymorphism Phenotyping v2.2.2 (PolyPhen-2) HumVar database. From: genetics.bwh.harvard.edu/pph2/ Accessed: Oct 2013. 21. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P, et al. A method and server for predicting damaging missense mutations. Nat Methods 2010; 7:248–9. doi: 10.1038/ nmeth0410-248. 22. Computational Biology, Memorial Sloan Kettering Cancer Center. Mutation Assessor, release 2. From: www.mutationassessor.org Accessed: Oct 2013. 23. Reva B, Antipin Y, Sander C. Predicting the functional impact of protein mutations: Application to cancer genomics. Nucleic Acids Res 2011; 39:e118. doi: 10.1093/nar/gkr407. 24. Biocomputing Group, University of Bologna. I-Mutant Suite, I-Mutant 3.0. From: gpcr2.biocomp.unibo.it/cgi/predictors/I- Mutant3.0/I-Mutant3.0.cgi Accessed: Oct 2013. 25. Capriotti E, Fariselli P, Casadio R. I-Mutant 2.0: Predicting stability changes upon mutation from the protein sequence or structure. Nucleic Acids Res 2005; 33:W306–10. doi: 10.1093/nar/gki375. 26. Mooney Laboratory, Buck Institute, Indiana University. MutPred v1.2. From: mutpred.mutdb.org Accessed: Oct 2013. 27. Li B, Krishnan VG, Mort ME, Xin F, Kamati KK, Cooper DN, et al. Automated inference of molecular mechanisms of disease from amino acid substitutions. Bioinformatics 2009; 25:2744–50. doi: 10.1093/bioinformatics/btp528. 28. Biocomputing Group, University of Bologna. SNPs&GO. From: snps-and-go.biocomp.unibo.it/snps-and-go/ Accessed: Oct 2013. 29. Calabrese R, Capriotti E, Fariselli P, Martelli PL, Casadio R. Functional annotations improve the predictive score of human disease-related mutations in proteins. Hum Mutat 2009; 30:1237– 44. doi: 10.1002/humu.21047. 30. Mooney Laboratory, Buck Institute, Indiana University. Protein ANalysis THrough Evolutionary Relationships (PANTHER) v6.1. From: www.pantherdb.org/tools/csnpScoreForm.jsp Accessed: Oct 2013. 31. Thomas PD, Campbell MJ, Kejariwal A, Mi H, Karlak B, Daverman R, et al. PANTHER: A library of protein families and subfamilies indexed by function. Genome Res 2003; 13:2129–41. doi: 10.1101/ gr.772403. 32. International Agency for Research on Cancer. Align-Grantham Variation Grantham Deviation (AGVGD). From: www.agvgd.iarc. fr/about.php Accessed: Oct 2013. 33. Tavtigian SV, Deffenbaugh AM, Yin L, Judkins T, Scholl T, Samollow PB, et al. Comprehensive statistical study of 452 BRCA1 missense substitutions with classification of eight recurrent substitutions as neutral. J Med Genet 2006; 43:295–305. doi: 10.1136/jmg.2005.033878. 34. Rost Lab, Columbia University. Screening for Non-Acceptable Polymorphisms (SNAP). From: www.rostlab.org/services/snap/ Accessed: Oct 2013. 35. Bromberg Y, Rost B. SNAP: Predict effect of non-synonymous polymorphisms on function. Nucleic Acids Res 2007; 35:3823–35. doi: 10.1093/nar/gkm238. 36. BioFold Unit, University of Alabama at Birmingham. Predictor of human Deleterious Single Nucleotide Polymorphisms (PhD-SNP). From: snps.biofold.org/phd-snp/phd-snp.html Accessed: Oct 2013. 37. J. Craig Venter Institute. Protein Variation Effect Analyzer (PROVEAN) database. From: provean.jcvi.org Accessed: Oct 2013. 38. Choi Y, Sims GE, Murphy S, Miller JR, Chan AP. Predicting the functional effect of amino acid substitutions and indels. PLoS One 2012; 7:e46688. doi: 10.1371/journal.pone.0046688. 39. J. Craig Venter Institute. Sorting Intolerant from Tolerant (SIFT). From: sift.jcvi.org/ Accessed Oct 2013. 40. J. Craig Venter Institute. Sorting Intolerant From Tolerant (SIFT) BLink. From: sift.jcvi.org/www/SIFT_BLink_submit.html Accessed: Oct 2013. 41. Ng PC, Henikoff S. Predicting deleterious amino acid substitutions. Genome Res 2001; 11:863–74. doi: 10.1101/gr.176601. 42. Charité–Universitätsmedizin Berlin. Mutation Taster. From: www.mutationtaster.org/ Accessed: Oct 2013. 43. Schwarz JM, Rödelsperger C, Schuelke M, Seelow D. MutationTaster evaluates disease-causing potential of sequence alterations. Nat Methods 2010; 7:575–6. doi: 10.1038/nmeth0810-575. 44. Bell J, Bodmer D, Sistermans E, Ramsden SC. Practice Guidelines for the Interpretation and Reporting of Unclassified Variants (UVs) in Clinical Molecular Genetics. From: cmgsweb.shared. hosting .zen.co.uk/BP Gs/p dfs%20cur rent%20bpgs/U V%20 GUIDELINES%20ratified.pdf Accessed: Oct 2013. 45. Wallis Y, Payne S, McAnulty C, Bodmer D, Sistermans E, Robertson K, et al. Practice Guidelines for the Evaluation of Pathogenicity and the Reporting of Sequence Variants in Clinical Molecular Genetics. From: www.acgs.uk.com/media/774853/ e v a l u at i o n _ a n d _ r e p o r t i n g _ o f _ s e q u e n ce _ v a r i a nt s _ b p g s _ june_2013_-_finalpdf.pdf Accessed: Oct 2013. 46. Richards CS, Bale S, Bellissimo DB, Das S, Grody WW, Hegde MR, et al. ACMG recommendations for standards for interpretation and reporting of sequence variations: Revisions 2007. Genet Med 2008; 10:294–300. doi: 10.1097/GIM.0b013e31816b5cae. 47. Evans DG, Shenton A, Woodward E, Lalloo F, Howell A, Maher ER. Penetrance estimates for BRCA1 and BRCA2 based on genetic testing in a Clinical Cancer Genetics service setting: Risks of breast/ ovarian cancer quoted should reflect the cancer burden in the family. BMC Cancer 2008; 8:155. doi: 10.1186/1471-2407-8-155. 48. Spurdle AB, Healey S, Devereau A, Hogervorst FB, Monteiro AN, Nathanson KL, et al. ENIGMA—Evidence-based network for the interpretation of germline mutant alleles: An international initiative to evaluate risk and clinical significance associated with sequence variation in BRCA1 and BRCA2 genes. Hum Mutat 2012; 33:2–7. doi: 10.1002/humu.21628. 49. Guidugli L, Carreira A, Caputo SM, Ehlen A, Galli A, Monteiro AN, et al. Functional assays for analysis of variants of uncertain significance in BRCA2. Hum Mutat 2014; 35:151–64. doi: 10.1002/ humu.22478.