Clinical Pharmacogenetics David A Flockhart MD, PhD Chief, Division of Clinical Pharmacology Professor of Medicine, Genetics and Pharmacology Indiana University School of Medicine

Pharmacogenetics, 1990

Pharmacogenomic Journals, 2006

SNP Variability in The Human Genome December billion base pairs ~22,000 genes 1.7% of the genome codes for protein 3.3% of the genome is as conserved as the 1.7% that codes for protein On average 1 SNP/1.2kb million SNPs that occur at > 1% frequency ~450,000 SNPs in MCS (Multiply Conserved Regions)

SNP Variability In Exons ~150,000 SNPs in known exons 48,451 non-synonymous SNPs 1113 introduce a stop codon 104 disrupt an existing STOP

Hierarchy of Pharmacogenetic Information from Single Nucleotide Polymorphisms (SNPs)

Multiple Copies of human DNA can result in significant inter-individual variability 3,654 autosomal segmental regions, 800 at a frequency of at least 3%. In 95 individuals analyzed, the two most diverse genomes differed by at least 9 Mb in size or varied by at least 266 loci in content. 14 copied regions harbor 21 of the known human microRNAs, raising the possibility of the contribution of microRNAs to phenotypic diversity in humans. i.e. We are NOT 99 % the Same Am J Hum Genet Jan;80(1):

Why Pharmacogenomics? To divine mechanism To predict therapeutic response

Current Clinical Ability to Predict Response Clinical Value of a Pharmacogenetic Test Pharmacogenetic Principle 1: Value Decreases when Current Predictive Ability is High Meyer UA and Flockhart DA, 2005 Azathioprine/TPMT β-blockade/β Receptor Cancer Chemotherapy Antidepressants/5HTR

Pharmacogenomic Methods Candidate gene approach Pathway (multiple candidate gene) approach Whole genome approach Combinations of the above

Drug Response Pathways can Involve many Candidate Genes

Methods in Pharmacogenetics Population distribution analysis with Normit plots using a valid probe to detect phenotypic polymorphism (> 1% of population) Identification of gene and variants Family and twin studies to confirm genetic characteristics (dominant, recessive, Mendelian, maternal etc.) Development of a genetic test for DNA variants Correlation between genotype and phenotype Application in Clinical Practice

Polymorphic Distribution Antimode

The Value of Normit Distribution Plots: Population Distribution of CYP2C19 phenotype Flockhart et al: Clin Pharmacol Ther 1995;57:

Skewed Distribution

Example 1 of a Skewed Distribution: Heterogeneity in response to Inhaled Corticosteroids Weiss ST et al. Hum Molec Genetics 2004; 13:

Using the extremes of a phenotypic distribution as a strategy to identify pharmacogenetic predictors Example: Iressa™ and the EGF receptor

EGFR mutations in tumors that are sensitive or insensitive to kinase inhibitors. Minna JD et al. Science 2004;304: All activating variants are in exon 9 and code for 4 specific αα changes

EGF Receptor Variant predicts Iressa™ Response : The “oncogene addiction model”

Lessons Germline genetic variation matters, but so do somatic mutations in tumors Extremes of phenotype are often viewed as “discardable data”, but outliers should be viewed as important research stimuli Patients who are “outliers” matter

Genetics and Drug Absorption

Enterocyte GI Lumen ATP ADP P-gp Transport Passive Diffusion Digoxin Transport across the GI lumen 

Fig. 3. Correlation of the exon 26 SNP with MDR-1 expression. The MDR- phenotype (expression and activity) of 21 volunteers and patients was determined by Western blot analyses. The box plot shows the distribution of MDR-1 expression clustered according to the MDR-1 genotype at the relevant exon 26 SNP. The genotype-phenotype correlation has a significance of P = (n = 21). P-Glycoprotein Pharmacogenetics : Effect of a “wobble” (no coding change) SNP in exon 26 Eichelbaum et al. Proc Nat Acad Sci March, 2000.

Eichelbaum et al, Proc Nat Acad Sci, 2000:March 0.25 mg of digoxin po at steady state

Brain Blood ATP ADP P-gp Transport Passive Diffusion Digoxin Transport across the Blood-Brain Barrier 

Genetics and Drug Elimination

Cytochrome P450 2D6 Absent in 7% of Caucasians Hyperactive in up to 30% of East Africans Catalyzes primary metabolism of: propafenone codeine  -blockers tricyclic antidepressants Inhibited by: fluoxetine haloperidol paroxetine quinidine

CYP2D6 Pharmacogenetics

New CYP2D6 variants continue to appear…. From: Zanger et al: Clin Pharmacol Ther Aug;76(2):

CYP2D6 Alleles 51 as of October, alleles have no activity 6 have decreased activity *1, *2, *4 and others have copy number polymorphisms The *2 variant can have 1,2,3,4,5 or 13 copies i.e increased activity

Oligonucleotide array for cytochrome P450 genotesting From: Flockhart DA and Webb DJ. Lancet End of Year Review for Clinical Pharmacology, 1998.

From: Dalen P, et al. Clin Pharmacol Ther 63: , 1998.

Paroxetine and CYP2D6 genotype change the plasma concentrations of endoxifen Stearns, Flockhart et al. J Natl Cancer Inst. 2003;95(23):

Lessons from CYP Pharmacogenetics Multiple genetic tests of one gene may be needed to accurately predict phenotype Gene duplication in the germline exists The environment in the form of Drug Interactions can mimick a genetic change

Vitamin K Carboxylase and CYP2C9 Genotypes altered Warfarin Dose Rieder et al. N. Eng J. Med 2005;352:

Genetic alterations in Phase 2 enzymes with clinical consequences UGT1A1 NAT-2 SULT1A1 COMT TPMT

UGT1A1 TA repeat genotype alters irinotecan neutropenia/activity /66/77/7 P=0.007 UGT1A1 genotype % grade 4/5 neutropenia N= /66/77/7 UGT1A1 genotype Objective response (%) P=0.045 McLeod H. et al, 2003.

N-Acetylation Polymorphism NAT-2 Late 1940’s : Peripheral Neuropathy noted in patients treated for tuberculosis : Genetic factors influencing isoniazid blood levels in humans. Trans Conf Chemother Tuberc 1959: 8, 52–56.

NAT-2 substrates (All have been used as probes) Caffeine Dapsone Hydralazine Isoniazid Procainamide

Incidence of the Slow Acetylator NAT-2 phenotype 50% among Caucasians 50% among Africans 20% among Egyptians 15% among Chinese 10% among Japanese

Onset of Positive ANA Syndrome with Procainamide. Woosley RL, et al. N Engl J Med 298: , 1978.

Thiopurine Methyl Transferase Homozygous mutants are 0.2% of Caucasian Populations Heterozygotes are ~ 10% Homozygous wild type is 90% –Metabolism of Azathioprine –6-Mercaptopurine

Thiopurine Methyl Transferase Deficiency From: Weinshilboum et al. JPET;222:

Germline TPMT genotype associates with incidence of toxicity

Examples of Human Receptors shown to be genetically polymorphic with possible alterations in clinical phenotype G-proteins Angiotensin II receptor and angiotensinogen Angiotensin converting enzyme  2 receptor Dopamine D 4 receptor Endothelial NO synthase 5HT 4 receptor

SerArg SerGly Arg Gly Ser Arg SerArg Ser Gly SerArg Gly Arg SerArg Gly SerArg Gly SerGly Arg SerGly Ser Gly SerGly Arg Gly Arg GlyArg HaplotypesDiplotypes Ying-Hong Wang PhD, Indiana University School of Medicine Gly 2SNPs: 10 possible hapoltypes

Observed  1 AR Haplotypes in Caucasians and African American Women (WISE study) Terra et al. Clin. Pharmacol. Ther. 71:70 (2002)

Ser Arg SerArg Ser Gly SerArg Gly Arg SerArg Gly Arg SerGly SerArg Gly SerGly Ser Gly SerGly Arg Gly Arg GlyArg HaplotypesDiplotypes SR/SRSR/SGSR/GR SG/GR SerArg SerGly Arg Gly Ying-Hong Wang PhD, Indiana University School of Medicine Of 10 theoretical diplotypes, only 4 were present in the study population

Johnson et al. Clin Pharmacol & Ther. 2003,74: Diplotype predicts Beta-blocker effect

Lesson: Diplotype may be a better predictor of effect than Genotype A SNP that tags a Haplotype (tagSNP) may be an economical means of screening

Lanfear DE et al. JAMA September 28 th, 2005;294: β 2 receptor Gln27Glu (79CG) genotype predicts mortality during β -blockade after MI.

Non-synonymous coding region polymorphisms in long QT disease genes SCN5A I Na I Ks I Kr KvLQT1 minK HERG MiRP1 K897T (10%) T8A (1.5%) D85N (1.5%) S38G (38%) H558R (24%) Vanderbilt R34C (4%) P1090L (4%) G643S (9%) Japan only P448R (20%[?]) Dan Roden MD, October 2003.

Hierarchy of Pharmacogenetic Information from Single Nucleotide Polymorphisms (SNPs)

A whole genome approach can work:

Current Methods for genetic testing By phenotype: metabolic probe drug or Western blot By PCR with mutation-specific endonuclease By PCR and allele-specific hybrization By oligonucleotide chip hybridization By laser lithography - guided oligonucleotide chip hybridization. By rapid throughput pyrosequencing Taqman probe screening Whole genome chips Custom chips Koch WH. Technology platforms for pharmacogenomic diagnostic assays. Nat Rev Drug Discov Sep;3(9):

Homepage of

Browsing PharmGKB

Example of the PharmGKB gene variant browser

Estimated cost to the patient of Genetic Tests in Clinical Practice By simple PCR for one mutation: ~$10 For 50 mutations: ~$150 By Chip for ~ 20 mutations: ~ $70 By Chip for 100 mutations: ~ $250

Ethical, legal and policy issues within pharmacogenetics Risk of Loss of Patient Confidentiality –Need for anonymized DNA storage systems Risk that existing patents will stifle progress –Need for careful interpration of Bayh-Dole Untangling the relationship between genetics and self-described race and ethnicity

Role Models for Pharmacogenetics Concorde? Nuclear Power? The Longitude Problem?

Clinical Pharmacogenetics Summary A good phenotyping probe is critical Genetic tests need validation just as any other tests A potent inhibitor can mimick a genetic polymorphism Not all genetic polymorphisms have a phenotypic correlate, or clinical effect The clinical relevance of genetic polymorphisms is greatest with drugs of narrow therapeutic range, but not confined to them The cost of genetic testing is not likely to be limiting

Medication History: AVOID Mistakes Allergies? : Is there any medicine that we should not give you for any reason? Vitamins and Herbs? Old drugs? …..as well as current Interactions? Dependence? Mendel: Family Hx of benefits or problems with any drugs?

Pharmacogenetics Websites The SNP consortium: The Human Genome: CYP alleles: Drug Interactions: