Calibration of computational tools for missense variant pathogenicity classification and ClinGen recommendations for PP3/BP4 criteria.

Recommendations from the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) for interpreting sequence variants specify the use of computational predictors as "supporting" level of evidence for pathogenicity or benignity using criteria PP3 and BP4, respectively. However, score intervals defined by tool developers, and ACMG/AMP recommendations that require the consensus of multiple predictors, lack quantitative support. Previously, we described a probabilistic framework that quantified the strengths of evidence (supporting, moderate, strong, very strong) within ACMG/AMP recommendations. We have extended this framework to computational predictors and introduce a new standard that converts a tool's scores to PP3 and BP4 evidence strengths. Our approach is based on estimating the local positive predictive value and can calibrate any computational tool or other continuous-scale evidence on any variant type. We estimate thresholds (score intervals) corresponding to each strength of evidence for pathogenicity and benignity for thirteen missense variant interpretation tools, using carefully assembled independent data sets. Most tools achieved supporting evidence level for both pathogenic and benign classification using newly established thresholds. Multiple tools reached score thresholds justifying moderate and several reached strong evidence levels. One tool reached very strong evidence level for benign classification on some variants. Based on these findings, we provide recommendations for evidence-based revisions of the PP3 and BP4 ACMG/AMP criteria using individual tools and future assessment of computational methods for clinical interpretation.

Assessment of the evidence yield for the calibrated PP3/BP4 computational recommendations.

PURPOSE: To investigate the number of rare missense variants observed in human genome sequences by ACMG/AMP PP3/BP4 evidence strength, following the calibrated PP3/BP4 computational recommendations. METHODS: Missense variants from the genome sequences of 300 probands from the Rare Genomes Project with suspected rare disease were analyzed using computational prediction tools able to reach PP3_Strong and BP4_Moderate evidence strengths (BayesDel, MutPred2, REVEL, and VEST4). The numbers of variants at each evidence strength were analyzed across disease-associated genes and genome-wide. RESULTS: From a median of 75.5 rare (≤1% allele frequency) missense variants in disease-associated genes per proband, a median of one reached PP3_Strong, 3-5 PP3_Moderate, and 3-5 PP3_Supporting. Most were allocated BP4 evidence (median 41-49 per proband) or were indeterminate (median 17.5-19 per proband). Extending the analysis to all protein-coding genes genome-wide, the number of PP3_Strong variants increased approximately 2.6-fold compared to disease-associated genes, with a median per proband of 1-3 PP3_Strong, 8-16 PP3_Moderate, and 10-17 PP3_Supporting. CONCLUSION: A small number of variants per proband reached PP3_Strong and PP3_Moderate in 3,424 disease-associated genes, and though not the intended use of the recommendations, also genome-wide. Use of PP3/BP4 evidence as recommended from calibrated computational prediction tools in the clinical diagnostic laboratory is unlikely to inappropriately contribute to the classification of an excessive number of variants as Pathogenic or Likely Pathogenic by ACMG/AMP rules.

Two integrated and highly predictive functional analysis-based procedures for the classification of MSH6 variants in Lynch syndrome.

PURPOSE: Variants in the DNA mismatch repair (MMR) gene MSH6, identified in individuals suspected of Lynch syndrome, are difficult to classify owing to the low cancer penetrance of defects in that gene. This not only obfuscates personalized health care but also the development of a rapid and reliable classification procedure that does not require clinical data. METHODS: The complete in vitro MMR activity (CIMRA) assay was calibrated against clinically classified MSH6 variants and, employing Bayes' rule, integrated with computational predictions of pathogenicity. To enable the validation of this two-component classification procedure we have employed a genetic screen to generate a large set of inactivating Msh6 variants, as proxies for pathogenic variants. RESULTS: The genetic screen-derived variants established that the two-component classification procedure displays high sensitivities and specificities. Moreover, these inactivating variants enabled the direct reclassification of human variants of uncertain significance (VUS) as (likely) pathogenic. CONCLUSION: The two-component classification procedure and the genetic screens provide complementary approaches to rapidly and cost-effectively classify the large majority of human MSH6 variants. The approach followed here provides a template for the classification of variants in other disease-predisposing genes, facilitating the translation of personalized genomics into personalized health care.

Determining the clinical validity of hereditary colorectal cancer and polyposis susceptibility genes using the Clinical Genome Resource Clinical Validity Framework.

PURPOSE: Gene-disease associations implicated in hereditary colorectal cancer and polyposis susceptibility were evaluated using the ClinGen Clinical Validity framework. METHODS: Forty-two gene-disease pairs were assessed for strength of evidence supporting an association with hereditary colorectal cancer and/or polyposis. Genetic and experimental evidence supporting each gene-disease relationship was curated independently by two trained biocurators. Evidence was reviewed with experts and assigned a final clinical validity classification. RESULTS: Of all gene-disease pairs evaluated, 14/42 (33.3%) were Definitive, 1/42 (2.4%) were Strong, 6/42 (14.3%) were Moderate, 18/42 (42.9%) were Limited, and 3/42 (7.1%) were either No Reported Evidence, Disputed, or Refuted. Of panels in the National Institutes of Health Genetic Testing Registry, 4/26 (~15.4%) contain genes with Limited clinical evidence. CONCLUSION: Clinicians and laboratory diagnosticians should note that <60% of the genes on clinically available panels have Strong or Definitive evidence of association with hereditary colon cancer or polyposis, and >40% have only Moderate, Limited, Disputed, or Refuted evidence. Continuing to expand the structured assessment of the clinical relevance of genes listed on hereditary cancer testing panels will help clinicians and diagnostic laboratories focus the communication of genetic testing results on clinically significant genes.

A functional assay-based procedure to classify mismatch repair gene variants in Lynch syndrome.

PURPOSE: To enhance classification of variants of uncertain significance (VUS) in the DNA mismatch repair (MMR) genes in the cancer predisposition Lynch syndrome, we developed the cell-free in vitro MMR activity (CIMRA) assay. Here, we calibrate and validate the assay, enabling its integration with in silico and clinical data. METHODS: Two sets of previously classified MLH1 and MSH2 variants were selected from a curated MMR gene database, and their biochemical activity determined by the CIMRA assay. The assay was calibrated by regression analysis followed by symmetric cross-validation and Bayesian integration with in silico predictions of pathogenicity. CIMRA assay reproducibility was assessed in four laboratories. RESULTS: Concordance between the training runs met our prespecified validation criterion. The CIMRA assay alone correctly classified 65% of variants, with only 3% discordant classification. Bayesian integration with in silico predictions of pathogenicity increased the proportion of correctly classified variants to 87%, without changing the discordance rate. Interlaboratory results were highly reproducible. CONCLUSION: The CIMRA assay accurately predicts pathogenic and benign MMR gene variants. Quantitative combination of assay results with in silico analysis correctly classified the majority of variants. Using this calibration, CIMRA assay results can be integrated into the diagnostic algorithm for MMR gene variants.

Modeling the ACMG/AMP variant classification guidelines as a Bayesian classification framework.

PURPOSE: We evaluated the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) variant pathogenicity guidelines for internal consistency and compatibility with Bayesian statistical reasoning. METHODS: The ACMG/AMP criteria were translated into a naive Bayesian classifier, assuming four levels of evidence and exponentially scaled odds of pathogenicity. We tested this framework with a range of prior probabilities and odds of pathogenicity. RESULTS: We modeled the ACMG/AMP guidelines using biologically plausible assumptions. Most ACMG/AMP combining criteria were compatible. One ACMG/AMP likely pathogenic combination was mathematically equivalent to pathogenic and one ACMG/AMP pathogenic combination was actually likely pathogenic. We modeled combinations that include evidence for and against pathogenicity, showing that our approach scored some combinations as pathogenic or likely pathogenic that ACMG/AMP would designate as variant of uncertain significance (VUS). CONCLUSION: By transforming the ACMG/AMP guidelines into a Bayesian framework, we provide a mathematical foundation for what was a qualitative heuristic. Only 2 of the 18 existing ACMG/AMP evidence combinations were mathematically inconsistent with the overall framework. Mixed combinations of pathogenic and benign evidence could yield a likely pathogenic, likely benign, or VUS result. This quantitative framework validates the approach adopted by the ACMG/AMP, provides opportunities to further refine evidence categories and combining rules, and supports efforts to automate components of variant pathogenicity assessments.

HGVS Recommendations for the Description of Sequence Variants: 2016 Update.

The consistent and unambiguous description of sequence variants is essential to report and exchange information on the analysis of a genome. In particular, DNA diagnostics critically depends on accurate and standardized description and sharing of the variants detected. The sequence variant nomenclature system proposed in 2000 by the Human Genome Variation Society has been widely adopted and has developed into an internationally accepted standard. The recommendations are currently commissioned through a Sequence Variant Description Working Group (SVD-WG) operating under the auspices of three international organizations: the Human Genome Variation Society (HGVS), the Human Variome Project (HVP), and the Human Genome Organization (HUGO). Requests for modifications and extensions go through the SVD-WG following a standard procedure including a community consultation step. Version numbers are assigned to the nomenclature system to allow users to specify the version used in their variant descriptions. Here, we present the current recommendations, HGVS version 15.11, and briefly summarize the changes that were made since the 2000 publication. Most focus has been on removing inconsistencies and tightening definitions allowing automatic data processing. An extensive version of the recommendations is available online, at http://www.HGVS.org/varnomen.

Universal Versus Targeted Screening for Lynch Syndrome: Comparing Ascertainment and Costs Based on Clinical Experience.

BACKGROUND: Strategies to screen colorectal cancers (CRCs) for Lynch syndrome are evolving rapidly; the optimal strategy remains uncertain. AIM: We compared targeted versus universal screening of CRCs for Lynch syndrome. METHODS: In 2010-2011, we employed targeted screening (age < 60 and/or Bethesda criteria). From 2012 to 2014, we screened all CRCs. Immunohistochemistry for the four mismatch repair proteins was done in all cases, followed by other diagnostic studies as indicated. We modeled the diagnostic costs of detecting Lynch syndrome and estimated the 5-year costs of preventing CRC by colonoscopy screening, using a system dynamics model. RESULTS: Using targeted screening, 51/175 (29 %) cancers fit criteria and were tested by immunohistochemistry; 15/51 (29 %, or 8.6 % of all CRCs) showed suspicious loss of ≥1 mismatch repair protein. Germline mismatch repair gene mutations were found in 4/4 cases sequenced (11 suspected cases did not have germline testing). Using universal screening, 17/292 (5.8 %) screened cancers had abnormal immunohistochemistry suspicious for Lynch syndrome. Germline mismatch repair mutations were found in only 3/10 cases sequenced (7 suspected cases did not have germline testing). The mean cost to identify Lynch syndrome probands was ~$23,333/case for targeted screening and ~$175,916/case for universal screening at our institution. Estimated costs to identify and screen probands and relatives were: targeted, $9798/case and universal, $38,452/case. CONCLUSIONS: In real-world Lynch syndrome management, incomplete clinical follow-up was the major barrier to do genetic testing. Targeted screening costs 2- to 7.5-fold less than universal and rarely misses Lynch syndrome cases. Future changes in testing costs will likely change the optimal algorithm.

Clinically relevant variants - identifying, collecting, interpreting, and disseminating: the 2013 annual scientific meeting of the Human Genome Variation Society.

The dramatic advances in genetic sequencing technologies used in research laboratories are now entering the clinic, and applications of whole-genome and whole-exome sequencing to disease diagnosis, predisposition, and treatment will soon be commonplace. However, the standards and methods for identifying clinically relevant variants are currently being debated and defined. Multiple agencies worldwide have recognized that we have reached an exciting and critical transition point into the clinic, and many important issues are being discussed that impact how genetic variation data in the clinic will be interpreted and used. The 2013 annual scientific meeting of the Human Genome Variation Society (HGVS) had as its main theme the discovery, interpretation, and dissemination of clinically relevant DNA variants. The meeting featured the continuously developing technology of databasing genetic variation and computational tools for allelic variant discovery. Attention was given to curating and integrating these data with clinical findings, including approaches to distinguish between functional alleles underlying clinical phenotypes and benign sequence variants and making data sources interoperable and functional for clinical diagnostic utility, citing examples in specific diseases.

Phenotype correlations with pathogenic DNA variants in the MUTYH gene.

MUTYH -associated polyposis (MAP) is an autosomal recessive disorder where the inheritance of constitutional biallelic pathogenic MUTYH variants predisposes a person to the development of adenomas and colorectal cancer (CRC). It is also associated with extracolonic and extraintestinal manifestations that may overlap with the phenotype of familial adenomatous polyposis (FAP). Currently, there are discrepancies in the literature regarding whether certain phenotypes are truly associated with MAP. This narrative review aims to explore the phenotypic spectrum of MAP to better characterise the MAP phenotype. A literature search was conducted to identify articles reporting on MAP-specific phenotypes. Clinical data from 2109 MAP patients identified from the literature showed that 1123 patients (53.2%) had CRC. Some patients with CRC had no associated adenomas, suggesting that adenomas are not an obligatory component of MAP. Carriers of the two missense founder variants, and possibly truncating variants, had an increased cancer risk when compared to those who carry other pathogenic variants. It has been suggested that somatic G:C>T:A transversions are a mutational signature of MAP, and could be used as a biomarker in screening and identifying patients with atypical MAP, or in associating certain phenotypes with MAP. The extracolonic and extraintestinal manifestations that have been associated with MAP include duodenal adenomas, duodenal cancer, fundic gland polyps, gastric cancer, ovarian cancer, bladder cancer and skin cancer. The association of breast cancer and endometrial cancer with MAP remains disputed. Desmoids and Congenital Hypertrophy of the Retinal Pigment Epithelium (CHRPEs) are rarely reported in MAP, but have long been seen in FAP patients, and thus could act as a distinguishing feature between the two. This collection of MAP phenotypes will assist in the assessment of pathogenic MUTYH variants using the American College of Medical Genetics and the Association for Molecular Pathology (ACMG/AMP) Variant Interpretation Guidelines, and ultimately improve patient care.

Assessment of the evidence yield for the calibrated PP3/BP4 computational recommendations.

Purpose: To investigate the number of rare missense variants observed in human genome sequences by ACMG/AMP PP3/BP4 evidence strength, following the calibrated PP3/BP4 computational recommendations. Methods: Missense variants from the genome sequences of 300 probands from the Rare Genomes Project with suspected rare disease were analyzed using computational prediction tools able to reach PP3_Strong and BP4_Moderate evidence strengths (BayesDel, MutPred2, REVEL, and VEST4). The numbers of variants at each evidence strength were analyzed across disease-associated genes and genome-wide. Results: From a median of 75.5 rare (≤1% allele frequency) missense variants in disease-associated genes per proband, a median of one reached PP3_Strong, 3-5 PP3_Moderate, and 3-5 PP3_Supporting. Most were allocated BP4 evidence (median 41-49 per proband) or were indeterminate (median 17.5-19 per proband). Extending the analysis to all protein-coding genes genome-wide, the number of PP3_Strong variants increased approximately 2.6-fold compared to disease-associated genes, with a median per proband of 1-3 PP3_Strong, 8-16 PP3_Moderate, and 10-17 PP3_Supporting. Conclusion: A small number of variants per proband reached PP3_Strong and PP3_Moderate in 3,424 disease-associated genes, and though not the intended use of the recommendations, also genome-wide. Use of PP3/BP4 evidence as recommended from calibrated computational prediction tools in the clinical diagnostic laboratory is unlikely to inappropriately contribute to the classification of an excessive number of variants as Pathogenic or Likely Pathogenic by ACMG/AMP rules.

Systematic large-scale application of ClinGen InSiGHT APC -specific ACMG/AMP variant classification criteria substantially alleviates the burden of variants of uncertain significance in ClinVar and LOVD databases.

Background: Pathogenic constitutional APC variants underlie familial adenomatous polyposis, the most common hereditary gastrointestinal polyposis syndrome. To improve variant classification and resolve the interpretative challenges of variants of uncertain significance (VUS), APC-specific ACMG/AMP variant classification criteria were developed by the ClinGen-InSiGHT Hereditary Colorectal Cancer/Polyposis Variant Curation Expert Panel (VCEP). Methods: A streamlined algorithm using the APC -specific criteria was developed and applied to assess all APC variants in ClinVar and the InSiGHT international reference APC LOVD variant database. Results: A total of 10,228 unique APC variants were analysed. Among the ClinVar and LOVD variants with an initial classification of (Likely) Benign or (Likely) Pathogenic, 94% and 96% remained in their original categories, respectively. In contrast, 41% ClinVar and 61% LOVD VUS were reclassified into clinically actionable classes, the vast majority as (Likely) Benign. The total number of VUS was reduced by 37%. In 21 out of 36 (58%) promising APC variants that remained VUS despite evidence for pathogenicity, a data mining-driven work-up allowed their reclassification as (Likely) Pathogenic. Conclusions: The application of APC -specific criteria substantially reduced the number of VUS in ClinVar and LOVD. The study also demonstrated the feasibility of a systematic approach to variant classification in large datasets, which might serve as a generalisable model for other gene-/disease-specific variant interpretation initiatives. It also allowed for the prioritization of VUS that will benefit from in-depth evidence collection. This subset of APC variants was approved by the VCEP and made publicly available through ClinVar and LOVD for widespread clinical use.

Getting ready for the Human Phenome Project: the 2012 forum of the Human Variome Project.

A forum of the Human Variome Project (HVP) was held as a satellite to the 2012 Annual Meeting of the American Society of Human Genetics in San Francisco, California. The theme of this meeting was "Getting Ready for the Human Phenome Project." Understanding the genetic contribution to both rare single-gene "Mendelian" disorders and more complex common diseases will require integration of research efforts among many fields and better defined phenotypes. The HVP is dedicated to bringing together researchers and research populations throughout the world to provide the resources to investigate the impact of genetic variation on disease. To this end, there needs to be a greater sharing of phenotype and genotype data. For this to occur, many databases that currently exist will need to become interoperable to allow for the combining of cohorts with similar phenotypes to increase statistical power for studies attempting to identify novel disease genes or causative genetic variants. Improved systems and tools that enhance the collection of phenotype data from clinicians are urgently needed. This meeting begins the HVP's effort toward this important goal.

Calibration of multiple in silico tools for predicting pathogenicity of mismatch repair gene missense substitutions.

Classification of rare missense substitutions observed during genetic testing for patient management is a considerable problem in clinical genetics. The Bayesian integrated evaluation of unclassified variants is a solution originally developed for BRCA1/2. Here, we take a step toward an analogous system for the mismatch repair (MMR) genes (MLH1, MSH2, MSH6, and PMS2) that confer colon cancer susceptibility in Lynch syndrome by calibrating in silico tools to estimate prior probabilities of pathogenicity for MMR gene missense substitutions. A qualitative five-class classification system was developed and applied to 143 MMR missense variants. This identified 74 missense substitutions suitable for calibration. These substitutions were scored using six different in silico tools (Align-Grantham Variation Grantham Deviation, multivariate analysis of protein polymorphisms [MAPP], MutPred, PolyPhen-2.1, Sorting Intolerant From Tolerant, and Xvar), using curated MMR multiple sequence alignments where possible. The output from each tool was calibrated by regression against the classifications of the 74 missense substitutions; these calibrated outputs are interpretable as prior probabilities of pathogenicity. MAPP was the most accurate tool and MAPP + PolyPhen-2.1 provided the best-combined model (R(2)  = 0.62 and area under receiver operating characteristic = 0.93). The MAPP + PolyPhen-2.1 output is sufficiently predictive to feed as a continuous variable into the quantitative Bayesian integrated evaluation for clinical classification of MMR gene missense substitutions.

Classifying variants of CDKN2A using computational and laboratory studies.

Variants in the CDKN2A tumor suppressor are associated with Familial Melanoma (FM), although for many variants the linkage is weak. The effects of missense variants on protein function and pathogenicity are often unclear. Multiple methods (e.g., laboratory, computational, epidemiological) have been developed to analyze whether a missense variant is pathogenic or not. It is not yet clear how to integrate these data types into a strategy for variant classification. We studied 51 CDKN2A missense variants using a cell cycle arrest assay. There was a continuum of results ranging from full wild-type effect through partial activity to complete loss of arrest. A reproducible decrease of 30% of cell cycle arrest activity correlated with FM association. We analyzed missense CDKN2A germline variants using a Bayesian method to combine multiple data types and derive a probability of pathogenicity. When equal to or more than two data types could be evaluated with this method, 22 of 25 FM-associated variants and 8 of 15 variants of uncertain significance were classified as likely pathogenic with >95% probability. The other 10 variants were classified as uncertain (probability 5-95%). For most variants, there were insufficient data to draw a conclusion. The Bayesian model appears to be a sound method of classifying missense variants in cancer susceptibility genes.

Deciphering the colon cancer genes--report of the InSiGHT-Human Variome Project Workshop, UNESCO, Paris 2010.

The Human Variome Project (HVP) has established a pilot program with the International Society for Gastrointestinal Hereditary Tumours (InSiGHT) to compile all inherited variation affecting colon cancer susceptibility genes. An HVP-InSiGHT Workshop was held on May 10, 2010, prior to the HVP Integration and Implementation Meeting at UNESCO in Paris, to review the progress of this pilot program. A wide range of topics were covered, including issues relating to genotype-phenotype data submission to the InSiGHT Colon Cancer Gene Variant Databases (chromium.liacs.nl/LOVD2/colon_cancer/home.php). The meeting also canvassed the recent exciting developments in models to evaluate the pathogenicity of unclassified variants using in silico data, tumor pathology information, and functional assays, and made further plans for the future progress and sustainability of the pilot program.