1
DETERMINING THE CLINICAL SIGNIFICANCE OF BRCA1 AND BRCA2 INTRONIC AND EXONIC SPLICING VARIANTS Karla R. Bowles, Nanda Singh, Ulrike P. Kappes, Erin Mundt, Bryan Warf, Brian Morris, Dmitry Pruss, Benjamin B. Roa, Richard Wenstrup Myriad Genetic Laboratories, Inc., Salt Lake City, UT, USA BACKGROUND Nucleotide changes which are predicted by in silico splicing analysis to affect mRNA splicing are sometimes identified during diagnostic sequencing analysis of the BRCA1 and BRCA2 genes. In silico splicing analysis tools should be used with caution as their use can result in incorrect variant interpretation, potentially leading to inappropriate medical management decisions. We describe the algorithms used by our laboratory to determine possible pathogenicity of intronic and exonic variants predicted by in silico analysis to result in abnormal mRNA splicing. METHODS Clinical germline testing for BRCA1 and BRCA2 sequencing mutations was performed on extracted patient genomic DNA, after informed consent was obtained. Sanger or next generation sequencing analyses of BRCA1 and BRCA2 identified nucleotide changes predicted by in silico splicing analysis to result in abnormal mRNA splicing. Using our variant classification and reclassification processes, which include multiple methodologies for variant evaluation and interpretation (Figure 1, Table 1), we further investigated the pathogenicity of these putative splicing variants. RESULTS INTRONIC VARIANTS: Sequencing analysis of >1 million patients identified ~1500 unique intronic variants in BRCA1 and BRCA2. In silico splice site analysis accurately identified pathogenic mutations lying within the consensus splice junctions at the +1/+2 or -1/-2 intronic positions as pathogenic splicing mutations (Tables 2-3, Figure 2). In silico splice site analysis of 456 benign intronic variants indicated that ~4.6% of these variants may negatively affect normal mRNA splicing, resulting in a 25%-75% estimated decrease in wild type donor or acceptor strength (Figure 3). However, analysis of some of these variants, such as BRCA1 c.81-13C>A and BRCA2 c.9501+3A>T, using other classification methodologies confirms their benign classifications. EXONIC VARIANTS: In silico splice site analysis accurately identified pathogenic mutations lying at the consensus last nucleotide of certain exons, such as BRCA1 c.4185G>A and BRCA2 c.9117G>A, as pathogenic splicing mutations. In silico splice site analysis of 3380 benign exonic variants indicated that ~4% of these variants have the potential to result in the creation of or significant strengthening of (> 0.10 absolute score increase with a final score >0.50) an alternative donor or alternative acceptor. However, variants such as BRCA1 c.3699A>G (p.Lys1233Lys) and BRCA2 c.9876G>A (p.Pro3292Pro) have proven to be benign using other classification evidence. BRCA1 c.4185G>A 0 5000 10000 15000 20000 -15 -10 -5 0 5 10 BRCA2 c.9117G>A 0 5000 10000 15000 -40 -20 0 20 40 BRCA1 c.3699A>G (p.Lys1233Lys) 0 5000 10000 15000 20000 -30 -20 -10 0 10 20 BRCA2 c.9876G>A (p.Pro3292Pro) 0 5000 10000 15000 -40 -20 0 20 40 0 5000 10000 15000 20000 25000 -100 -50 0 50 BRCA1 c.81-13C>A 0 5000 10000 15000 20000 25000 -100 -50 0 50 BRCA1 c.135-1G>T 0 5000 10000 15000 20000 -40 -20 0 20 40 BRCA2 c.9501+3A>T -2000 3000 8000 13000 18000 -40 -20 0 20 40 BRCA2 c.7977-1G>C CONCLUSIONS We have developed and implemented a robust classification and reclassification program for variants predicted by in silico analysis to result in abnormal mRNA splicing. In silico mRNA splicing analysis may indicate that particular variants negatively affect mRNA splicing. However, many of these predictions are either inaccurate or splicing effects are small and do not result in significant increases in cancer risk. In silico predictors should be used with caution and results rigorously verified by other independent methods before being used in the clinical setting in order to ensure correct test result interpretation and appropriate clinical management. Figure 1. Basic algorithm used to classify potential mRNA splicing variants Novel Intronic or Exonic Variant Which May Affect Splicing Identified in Clinical Patient Evaluate Data Determine variant frequencies in cases and controls Perform in silico splice site analysis Review literature Review other available data Classify Variant Pathogenic Benign Send Patient Report Likely Pathogenic Likely Benign Variant of Uncertain Significance (VUS) Variant Reclassification Program Automated Statistical Analysis Additional proband identified Automated History Weighting analysis Automated Mutation Co-occurrence (MCO) analysis Data compiled Statistical threshold reached - Evidence for pathogenic or benign Statistical threshold not reached - Hold for more probands Family Analysis Offer free testing to appropriate family members Run lab-developed segregation algorithm in trans co-occurrence identified - Evidence for benign Segregation data significant in multiple families - Evidence for pathogenic or benign Segregation data insignificant - Gather more family members • Identify new relevant literature • Performed weekly Monitor Literature Monitor variant frequencies in cases and controls Variant reclassified and amended reports issued Evidence for Pathogenic • Found at higher frequency in cancer cases than in controls Evidence for Benign • Found in >2% of controls • Found at equal frequencies in cancer cases and controls Uninformative • Found in <2% of controls Evidence for Pathogenic • Predicted to disrupt splicing with definitive biochemical confirmation Evidence for Benign • Not predicted to disrupt splicing Uninformative • Predicted to disrupt splicing without definitive biochemical confirmation Evidence for Pathogenic • Segregates with cancer in multiple families • Biochemical confirmation of splicing defect • Biochemical confirmation of complete transcript loss Evidence for Benign • in trans with pathogenic mutation • Does not segregate with cancer in multiple families • Biochemical confirmation of normal transcription Uninformative • Segregation not concordant between families • Biochemical analysis shows partial effect on splicing • Biochemical analysis shows partial effect on transcription Table 2. BDGP Splice Site Analysis Results Gene Variant Wild Type Score Variant Score BRCA1 c.135-1G>T 0.97 (Acceptor) <0.10 (Acceptor) BRCA1 c.81-13C>A 0.52 (Acceptor) 0.25 (Acceptor) BRCA1 c.3699A>G (p.Lys1233Lys) <0.10 (Alternate Donor) 0.98 (Alternate Donor) BRCA1 c.4185G>A 0.95 (Donor) 0.38 (Donor) BRCA2 c.7977-1G>C 0.98 (Acceptor) <0.10 (Acceptor) BRCA2 c.9501+3A>T 0.99 (Donor) 0.41 (Donor) BRCA2 c.9117G>A 0.57 (Donor) <0.10 (Donor) BRCA2 c.9876G>A (p.Pro3292Pro) 0.10 (Alternate Acceptor) 0.85 (Alternate Acceptor) Table 3. Variants determined by in silico splicing analysis (BDGP) to possibly result in abnormal splicing Gene Variant Transcript Analysis # Myriad Probands # Probands with a Pathogenic Mutation (in trans) History Weighting Algorithm Call Mutation Co-occurrence Algorithm Call Final Interpretation BRCA1 c.135-1G>T Abnormal splicing 1 164 0 (0) Pathogenic No Call Pathogenic BRCA1 c.81-13C>A No Data 168 11 (1) Benign Benign Benign BRCA1 c.3699A>G (p.Lys1233Lys) No Data 47 3 (1) Benign No Call Benign BRCA1 c.4185G>A Abnormal Splicing 2 11 0 (0) No Call No Call Pathogenic BRCA2 c.7977-1G>C Abnormal Splicing 3 118 0 (0) Pathogenic No Call Pathogenic BRCA2 c.9501+3A>T No Data 316 18 (5) Benign Benign Benign BRCA2 c.9117G>A Abnormal Splicing 4 150 0 (0) Pathogenic No Call Pathogenic BRCA2 c.9876G>A (p.Pro3292Pro) No Data 136 11 (1) Benign Benign Benign 1. Tesoriero AA, et al. Hum Mutat 26:495, 2005. 2. Claes K, et al. Genes, Chromosomes & Cancer 37:314-320, 2003. 3. Myriad Genetic Laboratories, Inc. Internal data. 4. Peelen T et al. British J Cancer 82:151-6, 2000; Bonatti F et al. Cancer Genet Cytogenet 170:93-101, 2006. Table 1. Additional descriptions of select variant analysis methods Method Description/ Rationale for Use In silico splice site prediction Multiple splice site analysis programs, which estimate the impact of a particular variant on mRNA splicing, are available for public use. Myriad uses both publicly available and internally developed programs. BDGP results are provided for the variants discussed. 1 History Weighting analysis This statistical technique, developed and validated by Myriad, is based upon the premise that disease associated mutations will be observed more often in individuals at high risk for carrying a mutation, as determined by the severity of personal and family history, but the observation of benign variants should be independent of personal and family history. 2 Mutation co- occurrence analysis This statistical technique, developed and validated by Myriad is based on the observation that the primary genetic cause of disease in a family is usually attributable to a single pathogenic mutation. Therefore, variants found to co-occur with a pathogenic mutation in the same individual are less likely to be pathogenic themselves. 3 In trans co- occurrence and homozygosity analyses Homozygous or compound heterozygous BRCA1 and BRCA2 pathogenic mutations are generally presumed to be embryonically lethal (BRCA1/2) or to result in severe phenotypes such as Fanconi anemia (BRCA2), although some exceptions have been identified. Therefore, a homozygous observation of a variant or, an in trans co-occurrence of a particular variant with a pathogenic mutation, provides evidence that a variant may be benign. 3 1. Berkeley Drosophila Genome Project (BDGP). www.fruitfly.org/seq_tools/splice.html 2. Pruss D et al. Breast Cancer Res Treat, Epub ahead of print, 2014. 3. Eggington JM, et al. Clin Genet. 2014 Sept;86(3):229-37. PubMed PMID: 24304220. Figure 2: Raw history weighting algorithm graphs illustrating classification calls for select BRCA1 and BRCA2 variants. Deleterious (red) and benign (green) control distributions with corresponding variant-specific history weighting scores (blue) are indicated for each variant. The Log History Weighting Score is plotted on the x-axis and the Number of Control Variants is plotted on the y-axis. Figure 3: A) In silico splice site analysis of 456 benign intronic variants indicated that ~4.6% of these variants may negatively affect normal mRNA splicing, resulting in a 25%-75% estimated decrease in wild type donor or acceptor strength. B) In silico splice site analysis of 3380 benign exonic variants indicated that ~4% of these variants have the potential to result in the creation of or significant strengthening of (> 0.10 absolute score increase with a final score >0.50) an alternative donor or alternative acceptor. < 0.1 absolute increase in alternative donor/acceptor score or not predicted to affect splicing > 0.1 absolute increase in alternative donor/acceptor score, final score <0.25 > 0.1 absolute increase in alternative donor/acceptor score, final score 0.25 to 0.49 > 0.1 absolute increase in alternative donor/acceptor score, final score 0.50 to 0.74 > 0.1 absolute increase with alternative donor/acceptor score, final score > 0.75 No effect or positive effect on donor/acceptor Negative effect < 25% Negative effect 25% to 49.9% Negative effect 50% to 74.9% 423 92.8% 12 2.6% 16 3.5% 5 1.1% 2999 88.7% 137 4.1% 75 2.2% 73 2.2% 96 2.8% A B Presented at NSGC - September 18, 2014
