Highlights from the MedSeq Project: ClinGen Meeting May 27, 2015 Robert C. Green, MD, MPH director, genomes2people Research Program in Translational Genomics and Health Outcomes Division of Genetics, Department of Medicine Brigham and Women s Hospital Partners Personalized Medicine Broad Institute and Harvard Medical School
Disclosures Research Grants: Collaborations (uncompensated): Speaking (compensated): Advisory (compensated): Equity: NIH, DOD, Illumina Pathway, 23andMe Illumina Bina, Invitae, Prudential None
Biesecker and Green, NEJM, 2014
The problem and opportunity of incidental findings
Minimum list Standardized search and reporting Consistent with practice of medicine and patient expectation Green, et al., Genetics in Medicine, 2013
Examining the Impact of Genomic Medicine Medical What is the impact upon individual and public health? Behavioral What is the impact upon physician and patient behavior? Economic What is the impact upon the healthcare system?
Why Clinical Trials?
The MedSeq Project Project 2 Workflow U01 HG006500 (2012-2016) Cardiologists and their patients with cardiomyopathy Randomize each patient to receive Standard of Care + Family History Review Standard of Care + Family History Review + Genome Report Primary care physicians and their healthy middle-aged patients Randomize each patient to receive Standard of Care + Family History Review Standard of Care + Family History Review + Genome Report Physician reviews family history information and discloses results from Genome Report Patient s electronic medical record Physician & patient outcomes Medical Record Review
Who Declined?
Who Participated?
MedSeq Participant Characteristics Percentage 120 100 80 60 40 20 0 Gender Total N= 202 PCP N=102 Cardiology N=100 Females 49 59 41 Males 51 41 59 Percentage Race and Ethnicity 105 100 95 90 85 80 75 Total N= 202 PCP N=102 Cardiology N=100 Other 13 14 12 Non-Hispanic, White 87 86 88 Percentage Annual household Income 120 100 80 60 40 20 0 Total N= 202 PCP N=102 Cardiology N=100 $100,000 64 74 53 <$100,000 36 27 47 Percentage 100 80 60 40 20 0 Education Total N= 202 PCP N=102 Cardiology N=100 College or Higher 81 86 76 No College Degree 19 14 24 Characteristic Total N=202 Primary care N=102 Cardiology N=100 Age, mean years 55.4 54.9 55.9 Genetic knowledge (0-11 scale), mean 10 10 10
Variant Classification
MedSeq Genome Filtering Approach Original filters Curated Exclusion Datasets >5 million variants HGMD ClinVar >5% Novel LOF Medical exome >1% ~200-300 variants <60 A B variants 20-40 C variants 10% in WGS Cases Gene exclusions Variant exclusions 10-30 variants Data Set A 10% MAF WGS Cases Excludes common technical FPs Common indels wrong nomenclature Exceptions FV, HFE, SERPINA1 Data Set B - Gene Exclusions Evidence for gene-disease association = none, limited, or disputed Non medically relevant phenotype Not reported variants: 82% VUS, Likely Benign, Benign False positive variants Assessed 13% C 5% B 13% A 69% Data Set C - Variant Exclusions Benign interpretation LOF but LOF not disease mechanism GWAS or PGx association only 611 2 71 31 11 Pathogenic Likely Pathogenic VUS-Favor Pathogenic Other Not reported Reported variants: 18%
Rules for Reporting on the Genome Report Variant Classification Incidental Findings of Medical Significance: Monogenic Disease and Carrier Risk (All Patients) Results Relevant to Clinical Indication: Variants in Cardiovascular Genes (Cardio Patients) Pathogenic Likely Pathogenic VUS- Favor Pathogenic VUS * VUS- Favor Benign Likely Benign Benign *If evidence of pathogenicity and/or in a HCM/DCM gene
How were results delivered?
MedSeq Project Genome Report 1 Page Summary Disease causing variants Carrier variants Pharmacogenomic variants Blood groups Additional Pages Structured variant data Variant evidence Disease/inheritance Supporting references
What was disclosed?
Reported findings from analysis of variants in ~4600 genes Number of patients Mean reported variants per patient Range of reported variants per patient Mendelian Disease Risk SFs 21/100 (21%)* Carrier Status SFs 92/100 (92%) Diagnostic Findings in the Cardiology Cohort 24/50 (48%) 0.21 2.3 0.54 0-1 0-7 0-2 *1/90 (1%) by ACMG list
Reported Disease Risk Findings Gene Variant Disease Classification Inheritance Notes ELN c.1150+1g>a Supravalvular aortic stenosis Pathogenic AD LHX4 c.452-2a>c Combined pituitary hormone deficiency Pathogenic PPOX p.leu67x Variegate porphyria Pathogenic AD RDH5 p.trp95x Fundus albipunctatus Pathogenic AR Homozygous HFE p.cys282tyr Hereditary hemochromatosis Pathogenic AR 3 biallelic cases CHEK2 c.1100del CHEK2-related cancer risk Pathogenic AD F5 p.arg534gln Factor V Leiden thrombophilia Risk allele Multi-factorial 3 cases ANK2 p.glu1458gly Ankyrin-B related cardiac arrhythmia Likely pathogenic AD COL2A1 p.thr1439met Spondyloepiphyseal dysplasia congenita Likely Pathogenic EYA4 c.1739-1g>a Postlingual hearing loss Likely Pathogenic AD KCNQ1* p.ser276profsx13 Romano-Ward syndrome Likely Pathogenic AD SQSTM1 p.pro392leu Paget disease of the bone Likely Pathogenic AD 2 cases APP p.ala713thr Alzheimer s disease, late onset VUS - Favor Pathogenic AD ARSE p.gly137ala Chondrodysplasia punctata VUS Favor Pathogenic XL PDE11A p.thr58profsx41 Primary pigmented micronodular adrenocortical disease VUS Favor Pathogenic TNNT2* p.arg278cys Hypertrophic cardiomyopathy VUS Favor Pathogenic AD AD AD AD
All Genome Reports available in the EHR and GeneInsight GeneInsight Clinic Surfacing Alerts Physicians are alerted of variant reclassifications 21
Submission of Reported Variants to NCBI Variants reported in the MedSeq study will be batched and submitted to ClinVar First batch of 52 variants now live in ClinVar dbgap submission in preparation
Blood type by genome sequencing
Blood Group Typing Through Sequencing Traditional serologic phenotyping methods are: Labor intensive Costly Sometimes unreliable Reagents not always available Could the blood bank reliably predict complex blood group systems using WGS instead? Only a minor added cost Prevent adverse outcomes for patients 34 Blood Group Systems 339 Serologic Phenotypes >1,100 known alleles The first demonstration of comprehensive RBC and platelet antigen prediction using WGS data!
All 100 individuals had RBC and Platelet antigens successfully predicted. Several (~ 5) individuals with rare antigen phenotypes identified for RBC, platelet and plasma donation. Serologic confirmation done for the 22 most commonly tested RBC antigens Total of 1760 serologic confirmations with no unresolvable discrepancies.
MedSeq Outcomes
Before sequencing, patients vary in the anticipated utility of WGS results Enthusiasts, high utility (21%) Discerners, variable utility (60%) Skeptics, low utility (19%)
After disclosure, patients ascribe greater utility to WGS results
Self-Reported Health Behavior Changes at 6 Weeks
Health Behavior Changes by Number of Variants Disclosed
Examples of PCP decision-making in the first 38 WGS disclosure visits ARM PATIENT S RESULT TEST ORDERED Primary Care (023-P05) Primary Care (030-P05) Primary Care (030-P05) Primary Care (038-P11) MONOGENIC RESULT KCNQ1 c.826delt Likely Pathogenic Romano-Ward syndrome CARRIER STATUS HFE c.845g>a Pathogenic Hereditary Hemochromatosis MONOGENIC RESULT PPOX c.199delc Pathogenic Variegate porphyria CARDIOVASCULAR RISK ALLELES - Coronary heart disease - Abdominal aortic aneurysm EKG (And, referral to Cardiovascular Geneticist) Iron/ferritin studies Repeat genetic testing for variegate porphyria at Mt. Sinai to confirm findings - Exercise stress tests - Abdominal ultrasound
Tests Ordered by Clinicians
Medication Changes
Can genome sequencing be used to screen for risk of rare Mendelian conditions?
Is Opportunistic Screening the same as Population-Based Screening? Opportunistic Population Infrastructure in place Relatively cost neutral Recommendations exist Follows medical model Infrastructure not in place Adds cost No recommendations Public health model
2068 Variants Blindly Evaluated in 462 Framingham Subjects Gene Variant Amino Acid Associated Condition Phenotype Age Sex BRCA2 c.5213_5216del p.thr1738ilefs*2 Breast/ovarian cancer Breast cancer 27-60 F BRCA2 c.4398_4402del p.leu1466phefs*2 Breast/ovarian cancer Prostate cancer MUTYH c.536a>g p.tyr179cys Colon cancer No history of cancer 48-75 M 33-67 F GLA c.427g>a p.ala143thr Fabry; HCM Normal echo 26-59 F MYBPC3 c.1504c>t p.arg502trp HCM Echo showed HCM 41-71 M MYBPC3 c.26-2a>g p.? APOB c.6240t>a p.tyr2080* HCM High cholesterol/blo od pressure LDLR c.429c>a p.cys143* High cholesterol Normal echo LDL: 33 mg/dl, on no cholesterol medications LDL: 195 mg/dl, on no cholesterol medications 38-73 M 23-29 M 35-68 F
Suggestive Clinical Findings with Pathogenic Variants Gene Variant Amino Acid Associated Condition Phenotype Age Sex BRCA2 c.5213_5216del p.thr1738ilefs*2 Breast/ovarian cancer Breast cancer 27-60 F BRCA2 c.4398_4402del p.leu1466phefs*2 Breast/ovarian cancer Prostate cancer MUTYH c.536a>g p.tyr179cys Colon cancer No history of cancer 48-75 M 33-67 F GLA c.427g>a p.ala143thr Fabry; HCM Normal echo 26-59 F MYBPC3 c.1504c>t p.arg502trp HCM Echo showed HCM 41-71 M MYBPC3 c.26-2a>g p.? APOB c.6240t>a p.tyr2080* HCM High cholesterol/blo od pressure LDLR c.429c>a p.cys143* High cholesterol Gold et al., ASHG Annual Meeting, 2014 Normal echo LDL: 33 mg/dl, on no cholesterol medications LDL: 195 mg/dl, on no cholesterol medications 38-73 M 23-29 M 35-68 F
8 subjects with pathogenic variants in an ACMG gene 38% With a suggestive clinical feature (SCF) 454 subjects without a pathogenic variant in an ACMG gene 13% 62% Without a suggestive clinical feature 87% Classification of ACMG genes Subjects with pathogenic variant, + SCF Subjects without a pathogenic variant, + SCF Cancer 66.7% (2/3) 5.3% (0.16/3) Cardiovascular 60.0% (3/5) 16.8% (0.84/5) Gold et al., ASHG Annual Meeting, 2014
The MedSeq Project Collaborators Project Leadership Robert Green, MD, MPH Zak Kohane, MD, PhD Calum MacRae, MD, PhD Amy McGuire, JD, PhD Michael Murray, MD Heidi Rehm, PhD Christine Seidman, MD Jason Vassy, MD, MPH, SM Project Manager Carrie Blout, MS Project Personnel Sandy Aronson, ALM, MA Danielle Azzariti, MS David Bates, MD Jennifer Blumenthal-Barby, PhD Ozge Ceyhan-Birsoy, PhD Alexis Carere, MA, MS Kurt Christensen, MPH, PhD Allison Cirino, MS Lauren Conner Kelly Davis Project Personnel (Cont.) Lindsay Feuerman Margaret Helm, MS Carolyn Ho, MD Lily Hoffman-Andrews Peter Kraft, PhD Joel Krier, MD Sek Won Kong, MD William Lane, MD, PhD Matt Lebo, PhD Lisa Lehmann, MD, PhD, MSc In-Hee Lee, PhD Kaitlyn Lee Kalotina Machini, PhD, MS David Margulies, MD Heather McLaughlin, PhD Jill Robinson, MA Melody Slashinski, MPH, PhD Shamil Sunyaev, PhD Ellen Tsai, PhD Peter Ubel, MD Rebecca Walsh Scott Weiss, MD External Advisory Board Katrina Armstrong, MD David Bentley, DPhil Robert Cook-Deegan, MD Muin Khoury, MD, PhD Bruce Korf, MD, PhD (Chair) Jim Lupski, MD, PhD Kathryn Phillips, PhD Lisa Salberg Maren Scheuner, MD, MPH Sue Siegel, MS Sharon Terry, MA Consultants Les Biesecker, MD George Church, PhD Geoffrey Ginsburg, MD, PhD Tina Hambuch, PhD David Miller, MD, PhD J. Scott Roberts, PhD David Veenstra, PharmD, PhD Protocol Monitoring Committee Judy Garber, MD, MPH Cynthia Morton, PhD
Thank You!!! Email: rcgreen@genetics.med.harvard.edu Web: genomes2people.org Twitter: @RobertCGreen