Pharmacogenomics Data Management and Application In Drug Development

1
Pharmacogenomics Data Management and Application
In Drug Development
HL7/CDISC Work Group Conference - 2003
Chuanbo Xu Senior Director, Bioinformatics
San Antonio, TX. 13 January 2003
2
Drug Development

• Future Targeted Discovery, Predictive Medicine

3
Beyond Pharmacodynamics and Pharmacokinetics
M
Y
T
X
Regulatory
4
Introducing Pharmacogenetic/Pharmacogenomics
M
M
Y
Y
T
T
X
X
Regulatory
5
Drivers for Personalized Medicine
We believe that the central issue is not
whether PGt- or PGx-guided Drug prescriptions
will happen, but when and how.
6
What Is PGt/PGx?

• Pharmacogenetics (PGt) studies the genetics
  basis of therapeutics and the individual
  reactions resulted from genotypes originally, it
  studies the effect exerted on drug ADMET
  (absorption, distribution, metabolism, excretion,
  toxicity) process by the human cytochrome
  family proteins.
• Pharmacogenomics (PGx) is the extension and
  enhancement of the PGt studies in the molecular
  sequence context of the individual genetic
  structures of the whole genome.

7
What Constitutes PGx Data?

• Key Components
• Gene, genomic structure (primary sequence and
  higher level organization) of the genes, subject
  DNA, protein, variation (SNP, INDELs, Haplotpyes,
  etc.), genotypes, gene expression profiling
• Therapeutics (compound, vaccine, antibody,
  siRNA, etc.), PK/PD profiling
• Subject demographics (age, gender, ethnicity,
  etc.), clinical measurements, phenotype,
  outcomes, statistical association analysis

8
Conservation vs. Variation
99.9 similar between individuals
.1 differences has functional consequences
9
Gene Haplotypes
Chromosome locus of gene
Exons
Promoters
SNPs
Causative Site
Haplotypes are a code for defining and tracking
the isoforms of a gene
10
Population Sample Constituted Using the
Definitions of the U.S. Census Bureau
96-well microtiter plate
11
High-Throughput Quality Control of SNPs I.
Electronic
Electronic trace analysis Phred Score gt30
Sequencing data confirmed in both directions
12
High-Throughput Quality Control of SNPs II.
Genetic

• Hardy-Weinberg Equilibrium
• Distribution frequency of heterozygotes
• must conform to frequency of
• individual alleles in ethnic group
• Example of frequencies
• if 5 for an allele, then 10
• heterozygotes and no homozygotes
• Mendelian Inheritance
• Polymorphisms are confirmed in the
• reference families
• Problems Picked Up
• Fixed heterozygosity /co-amplification
• Allele drop-out /primer sits on SNP

p2 2pqq21
Reference Families
13
Design Genaissance Bioinformatics Computing
Infrastructure (I)
14
Design Genaissance Bioinformatics Computing
Infrastructure (II)
15
Genaissance Secure Database Infrastructure
Genaissance LAN
Client Mirrors
Change tracking Audit
Change tracking Audit
Change tracking Audit
Client Users
Production System
Clinical System
CLIA Compliant HAPTyping DB
Access Control
Firewall / Domain Control
16
Genes By Functional Group
656
600
500
400
300
200
100
Enzymes
Receptors
Cytokine Receptor
Isomerase
Growth Factor
Binding Proteins
GPCR
Ligase
Hormone
Cell Cycle
Receptor Kinase
Lyase
Immunology-related
Channel
Ligand Gated Ion Channel R.
Intracellular transport
Cytokine
Kinase
Nuclear Hormone Receptor
Lipoprotein
Cytoskeletal/Cell Adhesion
Oxidoreductase
Transporter
Oncogene
Effector/Modulator
Phosphatase
Tumor Suppressor
Gene Expression
Hydrolase
Transferase
Miscellaneous
Nuclear Hormone
17
Distribution of SNPs/kb by gene region (724
genes)
18
Population Distribution of HAP Markers
U.S. Census Populations Caucasian African
American Asian Hispanic
19
MednosticsTMPharmacogenomic Trial Steps

• Define Hypothesis
• Define protocol (prospective vs. retrospective)
• Select candidate genes or SNPs
• Recruit patients (families vs. unrelated)
• Collect phenotypic data ()
• Collect blood samples (affects no. of genes
  protocol)
• Genotyping ()
• Statistical analysis (depends on all above)
• Validation

20
STRENGTH(Statin Response Examined by Genetic
HAP Markers)

• Prospective, multicenter, open-label
• Age 18 to 75
• Type IIa or IIb hypercholesterolemia
• Patients failed 6-week AHA Step I/II diet
• 4 week washout prior anti-hyperlipidemic
  medications

150 patients per each drug specific arm
21
STRENGTH Genes and Clinical Endpoints

• 175 candidate genes
• Lipid metabolism (CETP, LDLR, APOE)
• Drug Metabolism (CYP2C9, CYP2D6, CYP3A4)
• Inflammation (VCAM1, PPARG)

Clinical Endpoints

• LDL-C percent change (primary endpoint)
• HDL-C
• LDL/HDL ratios
• Total C

• triglycerides
• C-reactive protein
• Apolipoproteins
• Adverse events

22
STRENGTH I Baseline Lipids

• TC 257.8 mg/dl
• LDL-C 173.5 mg/dl
• HDL-C 48.9 mg/dl
• TG 177.1 mg/dl

23
Finding Pharmacogenetic Associations

• Gene associated with drug response will have one
  or more of its haplotypes clinically segregated
  according to outcome

No Association
Association
Average Responseper Individual
of Copies of HAP Marker
of Copies of HAP Marker
24
Finding Pharmacogenetic Associations

• Gene associated with drug response will have one
  or more of its haplotypes clinically segregated
  according to outcome

Best Responders
Partial Responders
Frequency
Haplotypes
Haplotypes
25
STRENGTH Analysis Parameters

• Statistical analysis
• ANCOVA with adjustment for multiple comparisons
• Raw p value significant markers screening
• Trial design to capture the marker of high market
  share
• Consider appropriate models
• Dominant
• Recessive
• Additive

26
STRENGTHClinical-Genetic Association Data Flow

• Define Subsets (individual statin pool)
  Endpoints Genes
• Candidate Associations
• Apply first pass comparison filtersignificance
  and marker distribution
• Visual inspection
• Biological/Medical/Literature Analysis
• Further statistical tests
• second pass multiple comparison filter
• Subset analysis (age, sex, ethnicity, alcohol)

DecoGen High throughput pipeline
27
Conclusions From STRENGTH

• Successful, first-ever comparative study using
  pharmacogenetics to

• Define populations with different response
• Differentiate between drugs in the same class

Most associations were statin-specific
Results may lead to new insights into
differential mechanisms of action for the statins
28
ADME Drug Metabolism by CYP2D6

• Central to the oxidative metabolism of gt30
  therapeutic drugs. (http//www.ncbi.nlm.nih.gov80
  /entrez/dispomim.cgi?id124030)
• Examples haloperidol, codeine, dextromethorphan,
  lidocaine, tamoxifen
• Greater than 100-fold variability in CYP2D6
  activity has been observed that can be attributed
  to genetic polymorphism
• Poor metabolizer (PM) vs. ultrarapid metabolizer
  (UM)

29
CYP2D6 Family Tree
30
Pharmacogenomics Data Standard
Defining New Standard For Drug Development
Submission Data
Genomics Data (Anonymized)
Association Data
Clinical Data (Anonymized)
31
Acknowledgements

• Medical affairs
• Genomics Sequencing and HAPTyping
• Bioinformatics and Database Management
• Software Development
• Quality Control Assurance
• Business Development and Intellectual Property
• c.xu_at_genaissance.com