Transcript Personalized Medicine Background and Challenges Geoffrey S

Personalized Cardiovascular Medicine:
Where We Stand Now,
and The Road Ahead
Kiran Musunuru, MD, PhD, MPH
Jeffrey S Berger, MD, MS
Geoffrey S Ginsburg, MD, PhD, FACC
Personalized Medicine:
Definition
“Personalized medicine is the use of diagnostic and
screening methods to better manage the individual
patient’s disease or predisposition toward a disease….
“Personalized medicine will enable risk assessment,
diagnosis, prevention, and therapy specifically tailored to
the unique characteristics of the individual, thus enhancing
the quality of life and public health.”
– NHLBI Strategic Planning, Theme #10
Disclaimer
Personalized medicine remains a research
concept – it is not yet ready for clinical
practice!
Addressing the Complexity of
Cardiovascular Disease
Metabolic pathways
Gene expression profile
Standard Biochemical
Pathway
GCCCACCTC
CGGGTGGAG
GCCCACCUC
DNA (genome)
RNA (transcript)
metabolites
protein
(e.g., enzyme)
A New Biomarker Toolbox:
Shift to Personalized Care
Human Genome Sequence
(Genomics)
Polymorphisms ~ 10,000,000
Gene Expression Profiles
(Transcriptomics)
Microarrays of ~ 25,000 gene
transcripts
Proteome
(Proteomics)
Protein arrays of specific protein
products ~ 100,000
Metabolome
(Metabolomics)
Small molecule metabolites
~ 5000
The Challenge
Individuals
with personal
profile A
Two different
“phenotypes”
– i.e., survival rates
Individuals
with personal
profile B
Example: Gene Expression
Profiling
Concept of “Biosignature”
Cardiovascular Insult or Therapy
Patient A
Patient B
“Biosignatures”
Gene chips
Outcome A
Outcome B
The Path to Personalized
Medicine
Standard of Care
(A or B)
Random Selection
A or B
Outcome
“Omics”-guided
A or B
Outcome
A
B
Health and Economic Outcomes
Genomics – Definitions




Polymorphism – an area of DNA sequence that varies from
person to person
“Single nucleotide polymorphism” (SNP) – a polymorphism
in which a single base in the DNA differs from the usual base
at that position
“Copy number variant” (CNV) – a polymorphism in which the
number of repeats of a DNA sequence at a location varies
from person to person
“Indel” – a polymorphism in which a DNA sequence is either
present (insertion) or absent (deletion) at a location, varying
from person to person
Types of Polymorphisms
GCCCACCTC
CGGGTGGAG
Single Nucleotide Polymorphism (SNP)
GCCCGCCTC
CGGGCGGAG
GCCCACCTCCTC
CGGGTGGAGGAG
Copy Number Variant (CNV)
GCC CTC
CGG GAG
“Indel” Polymorphism
More Definitions



“Locus” is the local area on a chromosome around a SNP
“Genotype” is the identity of the base at a SNP position for
each of the two alleles (since humans have paired
chromosomes) – a genotype is always two letters, unless the
SNP is on the X or Y chromosome in a man (XY)
“Haplotype” is a combination of SNPs at multiple linked loci
that are usually transmitted as a group from parent to child
Chromosomal Locus (1)
X
A
B C
D
E
Y
gene
hotspot



hotspot
Groups of SNPs are separated by recombination “hotspots”
– the SNPs tend to be passed from parents to children as a
unit, called a “linkage disequilibrium” block
Thus, SNPs A, B, C, D, E tend to stay together as a group –
the bases at these SNPs make up a “haplotype” – whereas
SNPs X and Y are not linked (not in linkage disequilibrium)
Because SNPs A–E are all linked, only one of these SNPs is
needed to act as a “tag” SNP for the whole group
Chromosomal Locus (2)
X
A
B C
D
E
Y
gene
hotspot



hotspot
Even though (in this example) SNPs A, B, C, and E are not
in the gene, each is in “linkage disequilibrium” with the gene
and remains associated with the gene as it is passed from
parents to children
If the gene causes a particular outcome (e.g., higher risk for
a disease), SNPs A–E will be associated with that outcome
This is the basis for genome wide association studies
Genome Wide Association
Studies (GWAS)


“HapMap” is an international consortium that has identified
millions of SNP locations in the human genome
Genome wide association studies (GWAS) in the HapMap
era:
 Identify an optimal set of ~ 300,000 “tag” SNPs (to
adequately cover genome of 3,000,000,000 bases)
 Collect > 1000 cases (e.g., a disease, a successful
response to a therapy, etc.) and > 1000 controls
 Genotype all case/control DNAs for all tagging SNPs
 600 million – not 6 trillion for full genomes – genotypes
 @ $0.005/genotype = ~ $3 million for each disease
(but cost is falling rapidly)
GWAS to Find OutcomeAssociated SNPs (1)
Outcome 1 (cases)
A1
A1
A2
A1
A1
G1
G1
G2
G1
G1
Outcome 2 (controls)
A2
A2
A2
A1
A2
G2
G2
G2
G1
G2
Association between SNP variant A1 / gene variant G1 and outcome 1
GWAS to Find OutcomeAssociated SNPs (2)
Outcome 1 (cases)
B1
B1
B2
B1
B2
G1
G1
G2
G1
G2
Outcome 2 (controls)
B1
B2
B1
B1
B2
G1
G2
G1
G1
G2
No association between SNP variant B1 / gene variant G1, and outcome 1
Uses for the Results of GWAS
human genome
Genome to
genes
~3,000,000,000 bases
chromosomal
loci – SNPs
~100,000 bases
causal genes
risk prediction
(epidemiology)
1
3
function
(basic science)
pharmacogenomics
2
4
therapy (drug
development)
SNPs for Risk Prediction
SNP 1
GCCCGCCTC
= AA (+0) vs. GA (+1) vs. GG (+2)
GCCCACCTC
SNP 2
SNP 3
.
.
.
SNP n
Total risk score
.
.
.
.
.
.
.
= ?? (+0) vs. ?? (+1) vs. ?? (+2)
= ?? (+0) vs. ?? (+1) vs. ?? (+2)
.
.
.
= ?? (+0) vs. ?? (+1) vs. ?? (+2)
= X (low risk vs. medium vs. high)
Genetic Risk Score for
Cardiovascular Disease


A genetic risk score calculated with 9 SNPs associated with
LDL or HDL cholesterol (score from 0-18) is associated with
cardiovascular disease
However, the score does not add to traditional risk factors
for CVD risk prediction
From Kathiresan et al., N Engl J Med, 2008: 358,1240-1249
SNP Panels for Risk
Prediction – Pitfalls





Several companies are marketing SNP panels to the
general public, charging hundreds to thousands of $$$
The premise for these panels is that they will let patients
know if they are at higher risk for particular diseases
None of these panels have yet been shown to add value to
traditional risk factor algorithms, and they should not be
recommended to patients at this time
The panels do not include rare mutations that cause
disease
Because all genome-wide studies to date have been done
in Caucasian populations, the SNP panels are not relevant
to non-Caucasian individuals
Pharmacogenomics –
Gene-Based Clinical Trial
Randomization
Control
group
CYP2C9 genotype
VKORC1 haplotype
Endpoints
Usual practice of
prescribing warfarin
Time to
adequate INR
Warfarin dose adjusted
to genotype/haplotype
Complications
– bleeds,
hospitalization
Warfarin Resistance and Sensitivity
Pharmacogenomics –
Mixed Success for Warfarin
Formula for dosing:
Estimated weekly coumadin dose = 1.64 + expe[3.984 + *1*1(0) + *1*2(-0.197)
+ *1*3(-0.360) + *2*3(-0.947) + *2*2(-0.265) + *3*3(-1.892) + Vk-CT(-0.304) +
Vk-TT(-0.569) + Vk-CC(0) + age(-0.009) + male sex(0.094) + female sex(0) +
weight in kg(0.003)]
where expe is the exponential to base e; *1, *2, *3 refer to CYP2C9 wild-type
(*1) or variant (*2, *3) genotypes, respectively; and Vk refers to VKORC1 with
variants CT, TT, or CC

In an early clinical trial, use of this formula improved the
accuracy and efficiency of warfarin initiation, though it did
not significantly reduce out-of-range INRs
From Anderson et al., Circulation, 2007: 116,2563-2570
Pharmacogenomics –
Mixed Success for Warfarin



The International Warfarin Pharmacogenetics Consortium
(IWPC) used a large retrospective study of warfarin users to
develop an algorithm to predict weekly dosing of warfarin
The algorithm includes age, height, weight, race, CYP2C9
genotype, VKORC1 haplotype, and use of interacting
medications (amiodarone, statins, azoles, sulfa drugs)
The IWPC tested the algorithm in a (retrospective) validation
cohort of warfarin users, comparing with a fixed-dose
approach and a clinical algorithm (i.e., no genetics
information)
From The International Warfarin Pharmacogenetics Consortium, N Engl J Med, 2009: 360,753-764
Pharmacogenomics –
Mixed Success for Warfarin



For most patients, there was no predictive advantage to
the pharmacogenetic algorithm
However, for outliers—patients requiring low or high doses
of warfarin to maintain stable INR—the pharmacogenetic
algorithm was significantly better
Algorithms such as this one need to be validated in
prospective clinical trials
From The International Warfarin Pharmacogenetics Consortium, N Engl J Med, 2009: 360,753-764
Pharmacogenomics –
Response to Clopidogrel




The cytochrome P-450 2C19 enzyme converts clopidogrel
into its active metabolite
CYP2C19 was genotyped in subjects of 3 large studies of
post-ACS patients receiving clopidogrel
In all 3 studies, carriers of reduced-function CYP2C19 alleles
had increased death, subsequent MI, and stroke
Will this be clinically useful?
Pharmacogenomics –
Statin-Induced Myopathy



GWAS of statin-induced myopathy found SNP in
SLCO1B1 gene associated with the condition
Individuals with CC genotype have 17 times higher risk of
myopathy than those with TT
May be useful for predicting risk before starting statin
therapy—needs to be tested in clinical trial
From The SEARCH Collaborative Group, N Engl J Med, 2008: 359,789-799
Candidate Genes for Lipid
Traits from Genomic Studies
GCCCGCCTC
lipid level
GCCCACCTC
lipid level
LDL
HDL
Triglycerides
APOB
ABCA1
APOA cluster
APOE cluster
CETP
ANGPTL3
LDLR
LIPC
MLXIPL
HMGCR
LIPG
GCKR
PCSK9
LPL
TRIB1
CSPG3
GALNT2
SORT1
MVK
From: (1) Willer et al., Nat Genet 2008: 40,161-169; (2) Kathiresan et al., Nat Genet 2008: 40,189-197; (3) Kooner et al., Nat Genet
2008: 40,149-151; (4) Sandhu et al., Lancet 2008: 371,483-491
Candidate Genes for CAD/MI
from Genomic Studies

Myocyte Enhancer Factor 2A (MEF2A)
Wang et al., Science 2003: 320,1578-81

5-lipoxygenase activating protein (FLAP)
Helgadottir et al., Nat Genet 2004: 36, 233-239

GATA-2
Connelly et al., PLOS Genet 2006: 2, 1265-1273

Chromosome 9p21 (gene not known)
Samani et al., N Engl J Med 2007: 357, 443-453
What to Expect




Although there are no applications of personalized medicine
being routinely used in cardiology yet, examples from other
specialties suggest what’s on the horizon
E.g., HLA-B*5701 allele testing is now done before abacavir
therapy to reduce the risk of hypersensitivity reactions
Genetic counselors will play an increasingly important role
in patient management as genetic information becomes
incorporated into everyday clinical practice
Referrals to genetic counselors are encouraged; if no
counselors are available at one’s institution, local
counselors can be found through the website: www.nsgc.org
Conclusion
We look to a future in which
medicine will be predictive,
preventive, preemptive and
personalized