transforming the practice of medicine

TRANSFORMING THE PRACTICE OF MEDICINEJune 7-9th 2015, UBC, Vancouver, BC

Be part of the Personalized Medicine Revolution

personalizedmedsummit.com#PMSummit2015

http://www.personalizedmedsummit.com/

1

Lee Hood UBC P4 Medicine Lecture

6-8-15

Systems Medicine and P4 Medicine: Transforming Healthcare through

Wellness—a Personal View

2

Companies I have co-founded and will mention in this lecture

•  Integrated Diagnostics—a blood protein diagnostic company (2012)

•  Indi Molecular—a peptide protein-capture company that will replace monoclonal antibodies (2014)

•  Arivale—a consumer-based scientific wellness company (2014)

3

The grand challenge for biology and medicine is deciphering biological complexity

4

6 Blind Men and an Elephant Systems Approach

5

Paradigm Changes Drive Radical Changes in Science

6

I Participated in Five Paradigm Changes in Biology Dealing with Complexity over 45 Years Leading to Systems Medicine and P4 Medicine

•  Brought engineering to biology –  Developed 6 instruments that led to high-throughput biology and big data in

biology (1970 - present) •  The Human Genome Project

–  Invented enabling technology, advocate, participant, applying genomics to P4 medicine (1990-2003)—complete parts list human genes

•  Cross-disciplinary biology –  Created 1st cross-disciplinary department: enabled technology development to

be driven by biology (1992-2000) •  Systems biology

–  Created 1st systems biology institute: deciphering the complexities of biology and disease (2000 – present)

•  Systems medicine / emergence of proactive P4 medicine –  Early advocate and pioneer of a P4 medicine that will transforming healthcare

(2001 – present) –  Pioneered systems driven technologies and strategies for P4 –  100,000 person wellness project (2013—present)

7

Central features of systems medicine

8

Dynamical, dense personalized clouds of billions of data points—capturing both genetic and environmental contributions to wellness and disease

Transac9onal

11010100010101010110101010100100

0

Phenome

Na143 K 3.7 BP 110/70

HCT32 BUN 12.9 Pulse 110 PLT150 WBC 92

GCGTAG ATGCGTAGGCATGCATGCCATTATAGCTTCCA

Genome

Proteome

arg-‐his-‐pro-‐gly-‐leu-‐ser-‐thr-‐ala-‐trp-‐tyr-‐val-‐met-‐phe-‐

Transcriptome

UUAGUG AUGCGUCUAGGCAUGCAUGCC

Epigenome

110101000101010101101010101001000101101010001

Single Cell

11010100010101010110101010100100

iPS Cells

11010100010101010110101010100100

Social Media

110101000101010101101010101001000101101010001

TeleHealth

110101000101010101101010101001000101101010001

9

Systems Medicine Biological networks carry information and mediate development, physiology and aging Disease-perturbed networks mediate disease

•  Integration of patient data will reveal biological networks that specify health and are altered in disease

•  Understanding differences in normal and disease-perturbed networks will provide fundamental insights into disease mechanisms

•  These insights are essential for developing more effective diagnostic and therapeutic approaches

10

Evolution of the Vision for Systems Medicine and P4 Medicine 2001 - Present

•  By 2006, the vision of Systems Medicine/P4 medicine had been clearly articulated by ISB –  Key question: How to bring P4 to the healthcare system?

•  In 2008, ISB formed a 5 year $100M strategic partnership with Luxembourg pushed us to a tipping point for medicine –  Developed about 10 new systems-driven technologies and strategies –  Placed P4 medicine in 2013 at a tipping point for transforming the

practice of healthcare

•  In 2013, ISB first proposed the P4 pilot project to carry out a longitudinal, high-dimensional data study 100,000 well people –  Bringing P4 medicine to the contemporary healthcare system –  Initiated a 100 person wellness project (2014)—100 Pioneers

•  Obama’s precision medicine initiative—presents a unique funding opportunity for the 100 K wellness project

11

Systems-driven Novel and Emerging Technologies

•  3rd generation DNA sequencing –  Nanopore/nanochannel, single molecule, electronic

detection (long reads, detect epigenetic marks, simple sample preparation, throughput)

•  Peptide protein-capture agents –  For sensitive protein quantification to replace antibodies

•  Global mass spectrometry –  Proteome analyses (SWATH) and the ability to identify

isoforms arising from mutation, splicing, editing, processing and chemical modifications

•  Targeted mass spectrometry (SRM) –  Analyze 100 proteins from complex samples like blood

•  Single-cell highly multiplexed omic and phenotypic analyses –  Detecting quantized cell populations and tipping points

12

Systems-driven strategies are transforming healthcare

•  Provide fundamental insights into dynamical disease-perturbed networks –  Enable mechanistic insights, diagnosis, therapy and prevention for the individual patient

•  Transform blood into a window to distinguish health from disease –  Disease diagnostics, assess drug toxicity, assess wellness –  Human examples: lung cancer, PTSD, liver toxicity, liver hepatitis

•  Family genome sequencing—identifying disease genes –  Identify disease, wellness genes and drug-intolerant genes. For the identification for

each individual’s 300 or more actionable genes •  Stratify diseases into their distinct subtypes

–  For impedance match with appropriate drugs –  Human example: various cancers

•  Stratify patients drug adverse reactions, modifier genes to disease mechanisms, eg, early and late onset of Huntington’s disease, Variant genes increase mercury susceptibility in kids

•  Enable a new computational approaches to pioneering drug reuse and drug target discovery –  Re-engineer disease-perturbed networks to normalcy with drugs, Repurpose drugs. faster and cheaper, drugs that prevent networks from becoming disease-perturbed Longitudinal, individual high-dimensional data clouds, Framingham-like clinical trials for preterm birth (Inova), cardiovascular disease, wellness, neurodegeneration etc.

13

Systems-driven strategies: dynamic approaches to prion-induced neurodegeneration in mice

•  George Carleson McLaughlin Institute, Great Falls, MT

•  Inyoul Lee—ISB •  Dawhee Haung—Seoul •  Hyuntae Yoo—Dallas Southwestern

14

Global and Subtractive Brain Transcriptome Analysis Differentially Expressed Genes (DEGs)

Uninfected brain

Prion infected brain

Inoculate w/ Prions

Time-‐course array analysis: subtra7ve analyses to DEGs

Mouse Genome array: 45,000 probe sets

~22,000 mouse genes.

RNA from brain

homogenate

Prion strains: •  RML •  301V

Mouse strains: §  C57BL/6J §  FVB/NCr §  BL6.I §  FVB/B4053

§  C57BL/6J-‐RML: 12 9me points

§  FVB/NCr-‐RML: 11 9me points

§  BL6.I-‐301V: 9 9me points

§  FVB/B4053-‐RML: 8 9me points

•  7400 DEGs -‐ Signal to noise issues, biological/technical, deep biology

•  300 DEGs -‐ Encode the prion neurodegenera9ve response

15

Neuropathology Identifies 4 Major Disease-Perturbed Networks for Prion Disease PrP replica7on/accumula7on Microglia/astrocyte ac7va7on

The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.

Synap7c degenera7on Normal Infected

Nerve cell death

16

Prion accumula9on

Glial Ac9va9on

Synap9c Degenera9on

Neuronal Cell Death

Cholesterol transport

Sphingolipid synthesis

Lysosome proteolysis

Reac9ve Astrocytes Leukocyte

extravasa9on

Na+ channels Cargo

transport

Caspases

*Arachidonate metab./Ca+ sig.

Clinical Signs

Sequential Disease-Perturbation of the Four Major Networks of Prion Disease

0 wk 18~20 wk 22 wk 7 wk

17

10 Disease-Perturbed Dynamical Networks in Prion Disease Explain Virtually all of the Pathophysiology of the Disease in Mice

18

Mouse Model Diseases

•  Prion-induced neurodegeneration •  Frontal temporal dementia •  Huntington’s disease •  Post traumatic stress disorder (PTSD) •  Liver toxicity •  4 models of glioblastoma—mimic Grade II, III,

and IV human disease

Examples that have been studied dynamically to reveal early disease-‐perturbed networks correla9ng with pathophysiology

19

Making blood a window to distinguish health from disease

20 20

High-throughput technologies set the stage for information-rich systems medicine

•  Key issues include:"– Many published, highly

promising results that don’t hold up for conversion to useful clinical assays"

– Need for systems analysis for extracting signal from noise--knowledge from data"

– Genetic diversity means must carry out assays in different geographical locations"

•  Report on best-practices released in 2012"

21

Keys to Blood as a Window to Health and Disease Systems decoding health and disease signals (reflections) from the body and the proper technologies

–  Blood is the key window: bathes all organs –  Longitudinal analyses –  Multiparameter panels –  Quantitative analyses –  Proteins may be the most effective

biomarker—closest to biology –  Systems strategies address signal to noise

•  Organ-specific blood proteins •  Systems filtering of biologically relevant blood

proteins for biomarker identification •  Others

22

Using organ-specific blood proteins to detect disease perturbations

•  Hyuante Yoo—Dallas Southwestern •  Yong Zhou--ISB •  Shizen Qin—ISB •  Gustavo Glusman—ISB

23

APLP1,SNAP25,LGI1,NACM1, CLSTN2

KINESIN,MAP1B,SYT3,CTN

ND1

CAMKII,PCLO,GRIA4,GLUR3,NSF,ANK2,ENO2,DOCK3

,SCG3,

L1CAM,CTF1, ARF3,ANK3,

MAP3K12,CTNNA2,KIF3A,GFAP,CNTN1,ENC1,

CRMP2, SYNAPSIN1

NEUROMODULIN,HUC,CAMKIIA,RIN,SYNAPSIN1,RGS4,PEA15,RASGRF1,NR1

GNAO1, GNA13, GABBR1, GLUR1,GRI

A1

MAP1A,SPTBN,

SPTBN4,FOXG1,EPHA5,N

CAM2, ELAVL3

TAU,MAP2, CAMKII, EPHA5,

UCHL1,NCAM1

RGS4,PEA15,CAMKII,RASGRF1,

NR1

Synap7c vesicle transport

Calcium mediated signaling

Synap7c Transmission

Neurogenesis

Cell surface receptor signaling

GPCR signaling

Cellular differen7a7on

Anatomical structure

development

Nerve growth factor signaling

200 Brain-Specific Blood Proteins Reflect Key Networks (SRM assays)

24

15 Brain-Specific Blood Proteins Reflect the early Initiation and Progression of Prion Disease-Perturbed Networks

Prion accumula9on

Glial Ac9va9on

Synap9c Degenera9on

Neuronal Cell Death

Cholesterol transport

Sphingolipid synthesis

Lysosome proteolysis

Reac9ve Astrocytes Leukocyte

extravasa9on

Na+ channels Cargo

transport

Caspases

*Arachidonate metab./Ca+ sig.

Apod* Scg3*

Cntn2* Ttc3*

Gria3* Gfap* L1cam*

Mapt* Snap25* Myo5a* Kif5a*

Gria1* Bcas1*

Grin1* Prkar1b*

Clinical Signs 0 wk 18~20 wk 22 wk

* indicates brain-specific blood proteins

25

A systems approach to blood diagnostic for identifying benign lung nodules in human lung cancer Integrated Diagnostics—Paul Kearney, Xiao-jun Li, Nathan Price, etc. X. Li et.al. Science Translation Medicine: 5, 207, 2013

26

Lung Cancer: Indeterminate Pulmonary Nodules

Integrated Diagnostics

Is this cancer?

~3 million cases annually in the USA

Patrick Nana-‐Sinkham, MD Ohio State University

27

Lung cancer blood biomarker 13-protein panel

•  Rule out for surgery about 40% of the benign nodules with 90% specificity—prevent 1/3rd of unnecessary surgeries

•  Save the healthcare system in US about $3.5 billion per year

28

Panel of 13 human blood proteins that distinguished benign from neoplastic lung

nodules: 12/13 proteins map to 3 networks

29

Why Blood Biomarker Panels for Detecting Disease—to detect Seven Features

•  Distinguish normal individuals from diseased individuals

•  Early diagnosis •  Follow progression •  Follow response to therapy •  Follow the reoccurrence of disease •  Reveal disease-perturbed networks which suggest

mechanisms of disease and candidate drug targets •  Stratification of disease into different subgroups for

impedance match against effective drugs—and proper prognosis

30

Peptide protein-capture agents will replace antibodies

Jim Heath Caltech

31

N 3 Protein catalyst library of alkynes

Very large (>106) azide library

“Click”

Protein target

C≡CH

Protein selec9vely couples only those pep9de library elements that fit onto its surface in just the right fashion

Circular 5-‐mer D amino acid pep9des are posi9oned on a protein and joined together with click chemistry

Form dimers or trimers—high affinity Very stable In vitro synthesis with Cu catalysis-‐-‐digital

33

64Cu – PCC <hPSA> Biodistribution and Tolerability—PET Scanning

1 min 40 min 120 min

PET images superimposed over whole body CT of normal mouse

Enables in vivo dynamics—temporal and spa9al—to be studied

34

Data •  reveals ~10:1 selec9vity for G12D vs wt •  PCC Agent is cell-‐penetrant •  PCC Agent is an inhibitor of RAS •  Selec9vity for G12D variant unknown •  Xtal structure of KRASG12D with PCC underway •  Med-‐chem improvements of PCC underway

IndiPCC agent drugs RAS!

Nature Chemistry (2015). DOI: 10.1038/NCHEM.2223.

Earlier genera7on related approach: selec7vely drugging AktE17K

Ras

Α-‐Tubulin

RAS inhibi9on data in pancrea9c cancer cell line PCC Agent structural analogue 7b5

PCC variants

Cell recovery with proteosome inhibitor

35

Peptide Protein Capture Agents--Features

•  Stable—send to Africa in an envelop •  Sensitive—each monomer a log increase in sensitivity •  Digital–synthesize unlimited quantities •  May be precisely directed at epitopes—hence avoids

cross-reactivity problems that plague antibodies •  May be adapted to large-scale production—easy to

produce •  Functions

–  In vitro diagnosis –  In vivo diagnosis –  Therapeutic reagents—possibly lacking cross reactivities

•  Prediction—will replace monoclonal antibodies with 10-15 years

•  Co-founded Indi Molecular in 2014 with Jim Heath

36

Three major thrusts from different fields started converging in 2012 or so leading to a more clearly defined P4 medicine that is predictive, preventive, personalized and participatory

37

The Emergence of P4 Medicine Predictive, Preventive, Personalize, Participatory

Converging Megatrends Driving the transforma9on of healthcare for pa9ents

Systems Biology & Systems Medicine

Consumer-‐Driven Social Networks

P4 MEDICINE

Digital Revolu9on Big Data

38

P4 medicine vs. contemporary medicine

•  Proactive vs. reactive •  Focus on individual vs. populations •  Focus on wellness and disease vs. just

disease •  Generate personalized data clouds •  Use personalized data clouds rather than

averaged populations of patients for clinical trials (N-1 experiments)

•  Employ patient (consumer) activated social networks for education, crowd sourcing and advocacy

39

Conceptual Themes of P4 Medicine

Disease Demys9fied Wellness Quan9fied

P4 Medicine Predic<ve Preven<ve Personalized Par<cipatory

Wellness Industry Disease Industry

40

Understanding Wellness is Key Developed World

If the trend of the last 10 years of increases in life

expectancy continue, more than half of all children

born today in developed countries can expect to

celebrate their 100th birthday.

Christensen, Ageing Populations: The Challenges Ahead, Lancet , 2009

41

A Framingham-like digital-age study of wellness in 100,000 (100K Project) patients longitudinally -- 20-30 years

2014 P4 Pilot Project Hundred Person Wellness Project (March-‐-‐December 2014) Nathan Price—ISB-‐-‐CoPI

42

Assays / Measurements

Gut Microbiome 3x Con7nual self-‐tracking & lifestyle monitoring

Whole Genome Sequencing

Detailed lab tests 3x (blood, urine, saliva)

Database of actionable possibilities that will grow exponentially over time

Crea9ng a dense, dynamical individual data cloud

43

Actionable Traits, Coaches and Positive Reinforcement •  Integration and modeling of each individual dynamic data cloud will identify actionable possibilities for the individual.

•  Actionable possibilities permit individuals to optimize wellness or avoid disease

•  Actionable possibilities from single or integrated data types

•  Coaches with MD advisors – Bringing actionable opportunities to each individual

•  Positive reinforcement / immediate gratification –  Individuals can see improvement of their high-

dimensional data clouds within a three-month period (from one blood draw or other sample to the next)

44

Health: What do we really want to understand from analyzing 100,000 well patients?

Wellness

Time

Wellness

Disease transi9on

45

Nature, Feb 2014

107 Individuals

46

Preliminary stories about actionable possibilities for the 107 Pioneers Nathan Price--ISB

47

New Discoveries: Five types of transitions from single and integrated data

Whole Genome

Sequencing

Lab Tests Blood, saliva,

urine

Gut Microbiome

Quan7fied Self/Traits

DISCOVERY Data

Integra7on & Correla7ons

Iden9fying five early transi9ons to op9mize wellness and avoid disease: •  Less to more wellness •  More to less wellness •  Wellness to disease •  Disease to wellness •  Changes in response to ac9onable

possibili9es

48

Initial Clinical Labs Discovery: High Rate of Actionable Clinical Lab Results

•  The 107 “well” par9cipants had a high rate of ini9al abnormal lab results

•  100% of the par9cipants had ac9onable recommenda9ons from their blood results

Baseline Blood Results

Cardiovascular

59%

Inflamma9on

68%

Nutrient Abnormali9es

91%

Diabetes Risk

54%

48

49

Actionable possibilities from two or more integrated data types

50

Disease to wellness transition: hemachromatosis

51

•  Blood + Gene9cs illuminated the effects of increasing copies of the Hemochromatosis variant •  Let untreated, this disorder could lead to car9lage damage, liver cancer, diabetes, and heart

disease: Easily treated by regular blood dona9ons to reduce the iron stores •  One par9cipant ALREADY had car9lage damage from his undiagnosed disease •  Subsequent family gene9c tes9ng detected other family members at risk 51

0.0

50.0

100.0

150.0

200.0

250.0

Zero copies of rare variant

(86 individuals)

One copy of rare variant

(12 individuals)

Two copies of rare variant

(2 individuals)

Ferri7n levels

Baseline 3 months

Genetics and Clinical Labs: Hemochromatosis Detected risk of a deadly disease in two participants

52

Vitamin D deficiency: arising from both genetics and environment

•  Vitamin D—90/107 Pioneers are low •  Six genetic variants from 3 genes block

Vitamin D absorption •  Risks associated with low Vitamin D

•  Ricketts—improper bone mineralization •  Increased risk of death from cardiovascular

disease •  Cognitive impairment in older adults—Alzheimer’s •  Severe asthma in children •  Cancer

53

Presence of risk alleles in vitamin D binding proteins is negatively correlated with vitamin D levels

28.75

31.64

35.71 36.25

41.07

43.85

25.00

30.00

35.00

40.00

45.00

50.00

Vitamin D vs. risk alleles in GC, DHCR7, CYP2R1 Round1 Round2

Wang, TJ et al. Common genetic determinants of vitamin D insufficiency: a genome-wide association study. Lancet 2010. 376(9736): 180-8.

4 or 5 risk alleles 8 par9cipants

3 risk alleles 15 par9cipants

1 or 2 risk alleles 15 par9cipants

25-‐OH D Co

ncen

tra9

on (n

g/mL)

54

Vitamin D is significantly increased across our participants

Round 1

Roun

d 2

μR1 = 33.6 ng/mL μR2 = 44.0 ng/mL psignedrank = 1.45E-‐9

Vitamin D

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

Vitamin D was the most significantly changed metabolite out of 189 for which data are available.

55

We can determine the individual risk for more than 50 diseases for which there are good statistical GWAS markers and these risks correlate with disease phenotypes in all 5 of the first studied disease examples

56

Es9mated risk for the disease or trait rela9ve to a popula7on

Enormous potential for genetics (GWAS variants) to estimate the genetic risk for each individual

Variant rs6827

Variant rs8572

Variant rs68883

Variant rs0994

Variant rs9769

Variant rs14445

Variant

Variant rs111393

Variant rs6837

Variant rs5837

Variant rs68279

Variant rs59583

Variant rs1352

Variant rs6827

Variant rs68883

Variant rs9769

Variant rs14445

Variant rs5837

Detrimental Variant Beneficial Variant

0

50

100

150

Cumula7ve Risk

Below average

Distribu9on from 2000 Genomes ADHD COPD Myopia

Alzheimer's disease Crohn's disease Obesity Anorexia Esophageal cancer Osteoarthri9s Asthma Gout Osteoporosis

Atrial fibrilla9on Grave's disease Ovarian cancer Breast cancer Hematocrit Pancrea9c cancer Bipolar disorder Hypertension Parkinson's disease Blood pressure Hypothyroidism Primary biliary cirrhosis

Bone mineral density Inflammatory bowel disease Prostate cancer Inflamma9on Iron levels Psoriasis

Calcium Lung Cancer Rheumatoid arthri9s Cardiovascular disease Lupus Schizophrenia

Celiac disease Macular degenera9on Stroke Cholesterol levels Magnesium levels Type 1 Diabetes

Chronic kidney disease Metabolic syndrome Type 2 Diabetes Colorectal cancer Migraine Ulcera9ve coli9s

Coronary heart disease Mul9ple sclerosis Urate levels

Ever increa

sing list of g

ene9c con

di9ons with

risk distribu

9ons =

Huge oppo

rtunity to im

prove welln

ess and pre

vent disease

57

Discovery and refinement of frequent genetic variants (GWAS) associated with lipid levels

Global Lipids Genetics Consortium. Discovery and refinement of loci associated with lipid levels. Nat Genet. 2013. 45(11): 1274-83.

•  Study examined 188,577 individuals using genome-wide and custom genotyping arrays. – 94,595 in initial discovery set – 93,982 in validation set

•  The study associates 72 loci with total cholesterol levels.

58

Individuals contain different subsets of the 72 variants that affect cholesterol levels

Variant rs68275

Variant rs8572

Variant rs68883

Variant rs099485

Variant rs97693

Variant rs14445

Variant rs65313

Variant rs111393

Variant rs86837

Variant rs6837

Variant rs583785

Variant rs68279

Variant rs59583

Variant rs13523

Each individual harbors a subset of the universe of possible variants that affect a trait. Although each variant alone has only a small effect, the cumula9ve effect of an individual’s variant set can add up to significant differences between individuals.

Small increase in overall cholesterol levels

Small decrease in overall cholesterol levels

59

•  Expected blood measurements can be computed from genomics

4000

6000

8000

10000

12000

14000

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

HDL Large Par9cle Co

ncen

tra9

on (n

mol/L)

Cumula9ve gene9c effect for each individual (std. dev.)

HDL Large Par7cle Concentra7on vs. Individual Gene7c Profile

0

50

100

150

200

250

300

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 LD

L Sm

all Par9cle Con

centra9o

n (nmol/L)

Cumula9ve gene9c effect for each individual (std.dev.)

LDL Small Par7cle Concentra7on vs. Individual Gene7c Profile

Low risk

59

Genetics and Clinical Labs Discovery Coaching to one’s genetic potential for HDL/LDL levels

High risk

Low risk

Correlation = 0.56 Correlation = -0.85, p<0.01

High risk

60

4000

6000

8000

10000

12000

14000

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6


ncen

tra9

on (n

mol/L)



0

50

100

150

200

250

300

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 LD

L Sm

all Par9cle Con

centra9o

n (nmol/L)



Low risk

60


High risk

Low risk

Correlation = -0.85, p<0.01

•  Genes are NOT des9ny – behavior can modify drama9cally

High risk

Correlation = 0.56

61

4000

6000

8000

10000

12000

14000

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6


ncen

tra9

on (n

mol/L)



0

50

100

150

200

250

300

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 LD

L Sm

all Par9cle Con

centra9o

n (nmol/L)



High risk

Low risk

61


High risk

Low risk

Correlation = -0.85, p<0.01

•  Genomes personalize proper interpreta9on of blood results

Correlation = 0.56

62

There are several par9cipants on sta9ns that drag down the LDL cholesterol average. Removing these individuals allows us to see the overall trend that correlates strongly with the gene9c cumula9ve risk. In fact, this may be highly predic9ve of who will need to be prescribed medica9on!

LDL cholesterol in our participants shows monotonic relationship with ‘genetic risk’

70#

80#

90#

100#

110#

120#

130#

140#

150#

160#

0#

10#

20#

30#

40#

50#

60#

Very#Low# Low# Medium# High# Very#High#

LDL#Ch

oleseterol#(m

g/dL)#

Num

ber#o

f#par8cipan

ts#in#th

is#risk#ra

nge#

Gene8c#Risk#

LDL#Cholesterol#Levels#vs#Gene8c#Risk#(59#variants)#Gene=c#Risk# Total#Cholesterol# Total#Cholesterol#(No#Medica=ons)#

63

Match your genetic risk against those of 2000 normal individuals

Individual genome

Mul9-‐genome reference models

2000 genomes Preliminary analysis

High-‐quality analysis

64

0

20

40

60

80

100

120

140

9.5 10

10.5

11

11.5

12

12.5

13

13.5

14

14.5

15

15.5

16

16.5

17

17.5

18

18.5

19

19.5

20

20.5

21

21.5

22

22.5

Gen

ome Co

unt

Cumula7ve Odds Ra7o

Risk for high cholesterol levels

Cumulative sum for an individual can give an estimate of risk relative to a population of 2000 normal individual genomes

Variant rs099485

Variant rs97693

Variant rs111393

Variant rs6837

Variant rs583785

Variant rs68279

Variant rs14445

Calculate the effects of all common variants within an individual

Es9mate the risk for high cholesterol rela9ve to a popula9on.

Below average

Above average

65

Relative disease risk in 2000 normal genomes Crohn’s

Obesity

Alzheimer’s (late onset)

Psoriasis

rela9ve risk

individu

als

44 GWAS 166 GWAS

57 GWAS 31 GWAS

66

The 100 Pioneer project has been a lens to view the “dark matter of human biology”

67

Sta9s9cal Correla9ons among Five Major Types of Pioneer Data—more than 35,0000 correla9ons—revealing the dark ma}er of human biology

choleostasis

68

The power of integrative analysis on multiple quadrants where N=100

We have calculated more than 30 thousand statistically-significant correlations between all of our five data quadrants for 107 individuals.

We have demonstrated that genome data can successfully predict risk for traits such as high LDL cholesterol, Crohn’s disease, vitamin D deficiency and type 2 diabetes. We believe it can predict risk for all GWAS diseases.

We have begun to identify the next generation of early-stage disease biomarkers and the first metrics for wellness.

75

85

95

105

115

125

135

0

5

10

15

20

25

30

35

40

Very Low Low Medium High Very High

LDL Ch

olesterol

Gene7c Risk

69

107 Pioneers insights

•  We are all less well than we think. All 107 Pioneers had multiple actionable possibilities.

•  Your genome does not control your destiny—rather just your potential.

•  “I can take control of my health with the proper data.”

•  Almost all of the Pioneers want to continue the longitudinal wellness study in its next phase.

•  70% of the actionable possibilities were acted upon

70

100K Project: Transforming Healthcare •  Identify vast array of actionable possibilities •  Analytics to optimize wellness and avoid (reduce) disease for each

individual patient—optimize human potential •  Create a data base of wellness measurements to mine for the

“multiparameter wellness metrics”—define fundamental human features of wellness—physiological and psychological

•  Generate a data base from individuals that will allow us to follow early disease mechanisms in the transitions from wellness to disease for major diseases—diabetes, cardiovascular, cancer, neurodegeneration and diseases of pregnancy—enable early transition back to wellness

•  Drive the development of improved old and new assays and analytics—parallelize, miniaturize, increase throughput, reduce cost, point of contact—digitalization through smart phone format

•  Database of wellness and disease transitions catalyze innovation for the wellness industry industry

•  Bring P4 medicine into the healthcare system –  Improving the quality of healthcare –  Decreasing the cost of healthcare –  Promoting innovation in Healthcare

71

The ISB Wellness Project Has Two Directions

•  The 100,000 person wellness project—academic—discovery science—pioneer assays (to the smart phone) and pioneer the integrative and modeling analytics—seek Congressional funding—focus on both wellness and early disease transitions

•  Arivale-a wellness sciences company—consumer directed—generate 10,000 individual data clouds in next 18 months—may really lead the large-scale adoption of P4 medicine and the democratization of healthcare

72

Three big ideas

•  Dynamical, dense personalized data clouds can be use to optimize individual human potential.

•  Tens of thousands of personal data clouds will transform how pharma, biotech, nutrition, and diagnostic companies practice their art.

•  The digitalization of medicine, the optimization of human potential and the identification and reversal of the earliest disease transitions for individuals will strikingly reduce the cost of healthcare—eventually leading to the democratization of healthcare for the poor as well as the rich.

73

Project Leadership

•  Leroy Hood, MD, PhD PI

•  Nathan Price, PhD, Co-PI

•  Sean Bell, Business Director

ISB Hundred Person Wellness Project – Team

Data Analytics

•  Nathan Price, Analytics Lead

•  Gustavo Glusman, Genomics

•  Andrew Magis, Multi-Omics

•  John Earls

Project Management

•  Sean Bell, Business Director

•  Kristin Brogaard, Project Manager

•  Sara Mecca, Project Assistant

•  Mary Brunkow, Project Coordinator

Medical Advisory Board

•  Robert Green, M.D.

•  Michael Raff, M.D.

•  Sarah Speck, M.D.

External Relations

•  Gretchen Sorensen, Consultant

Participant Engagement

•  Jennifer Lovejoy, VP Clinical Affairs

•  Sandi Kaplan, Wellness Coach

•  Craig Keebler, M.D., Study Physician

75

The gut microbiome

76

Distribution of microbial phyla in our participants

Ra9o of bacteroidetes to firmicutes phyla spans a wide range in only 100 individuals. Some individuals have excess growth of proteobacteria and verrucomicrobia, which may be linked to excess inflamma9on or a result of recent an9bio9c treatment.

100%

75%

50%

25%

0% Each bar is one individual—no dis9nct entroptypes

77

Genetics and Microbiome Discovery: Link between Gut Microbiome and Genetic Risk for Crohn’s Disease

0"

10"

20"

30"

40"

50"

60"

Very%Low% Low% Medium% High%% Very%High%

Num

ber%o

f%par6cipan

ts%in%th

is%risk%ra

nge%

Gene6c%Risk%

Microbial%Composi6on%vs.%Crohn's%Disease%Gene6c%Risk%(91%variants)%

77

78


0"

10"

20"

30"

40"

50"

60"


Num

ber%o

f%par6cipan

ts%in%th

is%risk%ra

nge%

Gene6c%Risk%


Verrucomicrobia 0%

Proteobacteria 3%

55% Bacteroidetes

42% Firmicutes

Verrucomicrobia 1%

Proteobacteria 4%

56% Bacteroidetes

39% Firmicutes

Verrucomicrobia 2%

Proteobacteria 2%

54% Bacteroidetes

42% Firmicutes Verrucomicrobia

3%

Proteobacteria 3%

54% Bacteroidetes

42% Firmicutes

Verrucomicrobia 7%

Proteobacteria 10%

41% Bacteroidetes

42% Firmicutes

78

Verrucomicrobia and Proteobacteria are pro-inflammatory bacteria

79

0"

10"

20"

30"

40"

50"

60"


Num

ber%o

f%par6cipan

ts%in%th

is%risk%ra

nge%

Gene6c%Risk%


Verrucomicrobia 0%

Proteobacteria 3%

55% Bacteroidetes

42% Firmicutes

Verrucomicrobia 1%

Proteobacteria 4%

56% Bacteroidetes

39% Firmicutes

Verrucomicrobia 2%

Proteobacteria 2%

54% Bacteroidetes

42% Firmicutes Verrucomicrobia

3%

Proteobacteria 3%

54% Bacteroidetes

42% Firmicutes

Verrucomicrobia 7%

Proteobacteria 10%

41% Bacteroidetes

42% Firmicutes


One par7cipant in study previously diagnosed with Crohn’s disease was iden7fied with high gene7c risk 79

Verrucomicrobia and Proteobacteria are pro-inflammatory bacteria

transforming the practice of medicine

Documents