Jackson Distinguished Professor of Clinical Medicine, Harvard Medical School

Transcription

1 Dennis Ausiello, M.D. Founder and Director, Center for Assessment Technology and Continuous Health (CATCH) Physician-in-Chief Emeritus, MGH Jackson Distinguished Professor of Clinical Medicine, Harvard Medical School 1

2

3

4 Promise and excitement of the human genome project CATCH 4

5 A success story for genetics: targeted therapies in oncology EGFR (NSCLC) ALK (NSCLC) Braf (Melanoma) CATCH 5

6 A success story for genetics: targeted therapies in oncology Somatic mutations Routine sampling of patient cells Unification of patient care and clinical discovery/research Histology Genomics Molecular imaging Chemical/ RNAi screens CATCH 6

7 Stratification rescues a drug Synthesis of crizotinib 2006 Phase 1 trial (without genotyping activity) 8/2007 EML4-ALK identified in NSCLC 5/2009 First report of Crizotinib activity in ALK+ NSCLC pts (ASCO 2009) 10/2010 NEJM publication of Phase 1 results 12/2007 ALK+ NSCLC Phase 1 trial begins From description of genetic mutation (EML4- ALK) in non-small cell lung cancer to FDA approval: CATCH 7

8 However, our ability to generalize this paradigm for chronic diseases remains limited! CATCH 8

9 The promise of the genetic revolution for stratification of cardiovascular disease? Accuracy predicting 10 year risk of CVD 10 yr CVD Using Risk Factors* Risk Factors Sex Cholesterol Smoking Age & family history Blood pressure Food CATCH 9

10 GWAS and the genetic revolution 10

11 The promise of the genetic revolution for stratification of cardiovascular disease? Accuracy predicting 10 year risk of CVD 10 yr CVD Using Risk Factors* Genetic Risk Score Risk Factors Sex Cholesterol Smoking Age & family history Blood pressure Food CATCH 11

12 Converging revolutions: Genetic Digital Integrative Science

13 Measurements used in medicine have largely not changed Heart rate, blood pressure Height, weight Electrocardiogram (ECG) Blood glucose, hemoglobin A 1 C Blood lipids CATCH 13

14

15 A renewed focus on phenotypes The human phenotype must be measured with the same precision that we expect of genetic data in real-time to provide a perpetual read out in the context of multiple perturbations (such as drugs and environment) and must ultimately create actionable new knowledge for healthcare performance CATCH 15

16 Not all individuals with coronary heart disease have traditional risk factors Khott et al, JAMA 2003 CATCH 16

17 Type 2 diabetes: increasing physiologic and genetic understanding 80 chromosomal loci Explains ~ 6% of variance in disease susceptibility

18 But what do we follow clinically? BLACK BOX CATCH 18

19 Unlocking what makes patients more different than the same - Data Analytics and Phenotyping 1,000 MS Patients: grouped by what they have in common 1,000 MS Patients: grouped by disease stage Grouping MS patients by what makes them different: stratified by frequency of relapses, duration of disease, and severity of flares exaptive

20 Episodic and Symptomatic Saturday, February 15, :00 Ms. Doe 8:15 8:45 9:00 20

21 Episodic and Symptomatic * Saturday, February 15, :00 Ms. Doe 8:15 8:45 9:00 21

22 Novel phenotypes, analyzed in the context of perturbations DRUGS ENVIRONMENT DIET ACTIVITY CATCH 22

23 Novel phenotypes, analyzed in the context of perturbations DRUGS ENVIRONMENT DIET ACTIVITY CATCH 23

24 The integration of data from different scales and different sources, analyzed with a broadening array of techniques. Repurposed data to yield new ideas Data collected via new methods High Dimensional Data Sets

25 Novel phenotypes, analyzed in the context of perturbations DRUGS ENVIRONMENT DIET ACTIVITY CATCH 25

26 Physiologic insights from a therapeutic device: Continuous Glucose Monitoring 24 hrs

27 CGM data clusters participants (determined by Principal Component Analysis)

28 Clusters persist across diabetes status, glucose control, medication A1C < 6 6 < A1C < 7 7 < A1C < 8 A1C > 8

29 Novel phenotypes, analyzed in the context of perturbations DRUGS ENVIRONMENT DIET ACTIVITY CATCH 29

30 Behavioral Measurements As we begin to assess health holistically we need to turn again to non-traditional technology Minimally Invasive Minimally Intrusive CATCH 30

31 Phenotypes at home: Passive and active behavioral analytics Runny nose: Late night/early morning communication sx Fever, influenza: WLAN entropy Social contacts for peer incentives: BEHAVIOR sx Sandy Pentland (MIT Media Lab) 31

32 Phenotypes at population scale Wearable Sensors, Electronic Medical Records, biospecimens and integrated analysis for discovery TO REAL-WORLD INDIVIDUALS FROM RESEARCH POPULATIONS 32 CATCH 32

33 EMRs to integrate research with clinical care and patient cells Partners EMR (~4M patients) i2b2 analysis software (phenotype extraction) Virtual patient cohort Matched, de-identified biospecimens from clinical labs (Crimson) (De-identified) Cost-effective measurement of genetic variation and phenotypes of large populations Real world patients Maintain patient privacy

34 Advanced phenotyping using data from routine clinical care Codified/Structured Data Narrative Data Carroll et al., JAMIA 2012;19:e162

35 Towards predictive models for outcomes time (years) physician notes type 2 diabetes dx coronary artery disease dx hemoglobin A1C lipid profile diabetes medications statins Genetics/epigenetics Predictive models Decision support Targeted interventions Clinical outcomes Adverse drug effects Genotype-phenotype at scale Human hypothesis testing Cohort identification

36 Stratifying natural history of autism via EMR Epilepsy Ear infections Hyperkinetic Kohane and colleagues Pediatrics (2014)

37 We have the genetic and digital revolutions helping to guide health. What about the power of social media?

38 Moving from social media and tweets. to Health!

39 Social Media for Healthcare has already arrived with

40 Novel phenotypes, analyzed in the context of perturbations DRUGS ENVIRONMENT DIET ACTIVITY CATCH 40

41 The Human Microbiome What we know about them: Microbes outnumber human cells 10 to 1 They collectively hold 8 million unique genes There are 10,000 species of microbes You carry 2-6 lbs of bacteria

42 A healthy microbiome in a healthy body Microbiome complexity & stability Birth 3 yrs Adult Elderly Genetics and environmental factors contribute to the development of the microbiome

43 Dysbiosis is implicated in diseases Healthy Microbiome complexity & stability Infectious diseases, metabolic diseases, and inflammatory disorders EARLY ONSET ADULT ONSET LATE ONSET Disease Birth 3 yrs Adult Elderly Opportunity to use this information to diagnose, predict, and treat diseases

44 Microbiome interacts with host genome, environment, disease phenotypes Faecalibacterium Loss in disease Gain in disease Demographics Disease phenotype Host genes Escherichia Meds, Diet Clinical covariates Disease activity Knights, Xavier, Gevers, Huttenhower

45 days Integrating phenotypes at many levels: chronic diseases with periodic flares Blood samples Stool samples Patient symptoms Behavioral analytics 45

46 An ecosystem outside of medicine CATCH 46

47 CATCH 47

48 A new partnership with patients Engaged, participatory Two-way information and learning exchange Patient-controlled data Unify wellness with spirit of inquiry CATCH 48

49 Convergence of multiple disciplines Data Analytics Sensors & Devices desirable patient experience people Omics Social Sciences (Behavior, Privacy) CATCH 49