Genomics and Personalized Medicine: Smoking Cessation Treatment Li-Shiun Chen, MD, MPH, ScD Washington University School of Medicine Apr 18, 2013 Genomics Informs Smoking Cessation Treatment I. What do we know about genetics of nicotine dependence? II. Are genes important for smoking cessation? Cessation success Response to pharmacotherapy III. Are these genetic associations real and useful? Genomics can lead to personalized medicine Risks Benefits Cardiovascular side effect (NRT, varenicline) Seizure, MAO-I (bupropion) Perinatal safety? Medication Cost Efficacy of cessation medication Combination vs. monotherapy E D. Green et al. Nature 2011 Chromosome 15q25 Is Important for Smoking CHRNA5-A3-B4 The Tobacco and Genetics Consortium (2010) Nature Genetics Genetics of nicotine dependence • Heritability 56%-71% • Specific genetic risks identified – CHRNA5-CHRNA3-CHRNB4 gene cluster • Association -> Function – amino acid change in nicotinic receptor (rs16969968) – CHRNA5 mRNA expression in brain/lung (rs588765) • Are genes important for nicotine dependence also relevant for smoking cessation? Does CHRNA5 Predict Smoking Cessation Success? Predicting nicotine dependence Altered nicotinic receptor function Divided evidence with cessation CHRNA5 predicts cessation success and response to medication Study Design U Wisconsin - TTURC • N=1073, European Ancestry • Pharmacotherapy arms (NRT, bupropion, combo) and 1 placebo arm • Cessation Abstinence at 60 days Time to relapse over 60 days CHRNA5-A3-B4 Haplotypes • Rs16969968 Non-synonymous coding, Amino acid change in CHRNA5 • Rs680244 CHRNA5 mRNA levels in brain and lung • Combination of 2 variants – H1 (GC, 20.8%) Low smoking quantity – H2 (GT, 43.7%) – H3 (AC, 35.5%) High smoking quantity CHRNA5 haplotypes predict cessation and response to medication 1.6 1.4 1.2 reference 1.00 OR (Abstinence) 1.0 1.13 1.11 0.98 Placebo 0.8 Treatment 0.62 0.6 0.37 0.4 0.2 0.0 H1 N=1,073 Haplotypes (rs16969968, rs680244): H1=GC(20.8%) H2=GT(43.7%) H3=AC(35.5%) H2 Haplotypes H3 Chen et al, Am J Psychiatry 2012 CHRNA5 Haplotypes predict abstinence in individuals receiving placebo medication 1.6 1.4 1.2 reference 1.00 OR (Abstinence) 1.0 1.13 1.11 0.98 Placebo 0.8 Treatment 0.62 0.6 0.37 0.4 0.2 0.0 H1 H2 Haplotypes H3 Chen et al, Am J Psychiatry 2012 CHRNA5 Haplotypes does not predict abstinence in individuals receiving active medication 1.6 1.4 1.2 reference 1.00 OR (Abstinence) 1.0 1.13 1.11 0.98 Placebo 0.8 Treatment 0.62 0.6 0.37 0.4 0.2 0.0 H1 H2 Haplotypes H3 Chen et al, Am J Psychiatry 2012 Smokers with the high risk haplotypes are 3 times more likely to respond to pharmacotherapy 1.6 1.4 1.2 reference 1.00 OR (Abstinence) 1.0 1.13 1.11 0.98 Placebo 0.8 Treatment 0.62 0.6 0.37 0.4 0.2 0.0 H1 H2 Haplotypes H3 Chen et al, Am J Psychiatry 2012 Smokers with the low risk haplotypes do not benefit from pharmacotherapy 1.6 1.4 1.2 reference 1.00 OR (Abstinence) 1.0 1.13 1.11 0.98 Placebo 0.8 Treatment 0.62 0.6 0.37 0.4 0.2 0.0 H1 H2 Haplotypes H3 Chen et al, Am J Psychiatry 2012 A Significant Genotype by Treatment Interaction 1.6 1.4 1.2 reference 1.00 OR (Abstinence) 1.0 1.13 1.11 0.98 Placebo 0.8 Treatment 0.62 0.6 0.37 0.4 0.2 0.0 H1 H2 Haplotypes H3 The interaction of haplotypes and treatment is significant (X2=8.97, df=2, p=0.011). Chen et al, Am J Psychiatry 2012 Number Needed to Treat (NNT) Varies with Haplotypes NNT: # of patients to treat for 1 to benefit 1.0 0.9 0.8 NNT=7 0.7 0.6 H1 Abstinence0.5 >1000 H2 0.4 H3 0.3 4 0.2 0.1 0.0 Placebo H1=GC(20.8%) H2=GT(43.7%) H3=AC(35.5%) Treatment Chen et al, Am J Psychiatry 2012 Genetics can predict prognosis & inform treatment • Smokers with the low risk haplotype (H1/GC) – quit more successfully without medication – do not benefit from medication • Smokers with the high risk haplotype (H3/AC) – have more difficulty quitting without medication – benefit from medication Does CYP2A6 Predict Smoking Cessation Success? Predicts smoking quantity Encodes the primary nicotine metabolism enzyme Fast metabolizers have more withdrawal CYP2A6 predicts response to medication Faster metabolism (n=501) Slower metabolism (n=208) Placebo Active Treatment A significant interaction (wald=7.15, df=1, p=0.0075) Chen, Bloom, et al, Under review Medication effect (NRT, Not bupropion) differs by metabolism Faster metabolism Slower metabolism Nicotine Replacement Therapy Buproprion Placebo Active Treatment Time to relapse over 90 days A significant interaction between NRT and CYP2A6 (wald=4.84, df=1, p=0.028). No interaction between bupropion and CYP2A6 (wald=0.036, df=1, p=0.85). NRT Bupropion Placebo Faster metab olism 363 157 58 Slower metabolism 149 96 21 Combine CHRNA5 and CYP2A6 Independent Additive Nicotine replacement therapy (NRT) vs. placebo effect varies with the combined effects of CYP2A6 and CHRNA5 90.0% 80.0% 70.0% 60.0% Abstinence 50.0% placebo 40.0% NRT 30.0% 20.0% 10.0% 0.0% CYP2A6: CHRNA5: Placebo Medication Low risk Low risk n=6 n=50 Low risk High risk n=14 n=90 High risk Low risk n=23 n=134 High risk High risk n=33 n=221 NNT >1000 16.6 3.7 2.6 A significant interaction (wald=7.44, df=1, p=0.0064) Chen, Bloom, et al, Under review Are these results real and useful? Validation in different samples (PNAT) Validation in special populations (myocardial infarction) Validation in natural cessation in observational studies Replication by PNAT Consortium CHRNA5 decreases abstinence with PLACEBO but not with NRT Less likely to quit N=2,633; 8 RCTs PNAT, Bergen et al, 2013, Pharmacogenetics and genomics Replication in Smokers Hospitalized with Myocardial Infarction, CHRNA5 predicts quitting Cessation before Admission 100% 90% 80% 70% % Abstinence 60% 50% 40% 30% 20% 10% 0% Cessation at 1 Year 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% GG GA AA GG CHRNA5 (rs16969968) Predictors Age Sex Genotype (rs16969968) N=1,450; TRIUMPH Consortium GA AA CHRNA5 (rs16969968) Having Quit Smoking at Baseline Admission for MI OR 95% C.I. P 1.10 (1.08-1.11) <0.0001 0.59 (0.45-0.77) 0.0001 0.81 (0.68-0.97) 0.0201 Abstinence at 1 Year Follow-up after Admission OR 95% C.I. P 1.06 (1.05-1.08) <0.0001 0.67 (0.48-0.44) 0.0197 0.77 (0.62-0.96) 0.0199 Chen et al, Under review Replication in NCI/GAMEON meta-analysis CHRNA5 rs16969968 (A) delays age of quitting smoking Cox regression models adjusted for age, sex, and lung cancer status for lung cancer /ILCCO studies 26 CHRNA5 rs16969968 delays quitting by 2-4 years (age 41->45 at first quartile, 54->56 at median) Proportion Having Quit rs16969968 genotype + AA + GA + GG Age of Quitting Smoking AGE at Cessation 27 Quit early, live longer Jha et al, 2013, NEJM Quit delay is clinically significant Quit by 40 • Both smoking quantity and quit age affect risk • Quit by 40 avoided nearly all the excess risk • Quit age delay of 2-4 years Genetic Effect Genetic Effect Ongoing International Collaboration on Smoking Research Acknowledgement • Cross-Population Meta-Analyses International Consortium of Smoking, PHASE I Washington U Nancy Robert Alison Sarah Thomas John Linus Jen Hong Laura MD Anderson Chris Margaret Sanjay Younghun MSTF Ming Jennie Thomas WSU Ann Angie UM Nicole Stephen Braxton Yu-Ching MGS Alan R. Jubao Jianxin Douglas F. Pablo V. Sharon WHI Andrew Sean Charles Helena Saccone GENOA Culverhouse Goate Hartz Przybeck HyperGen Rice SchwantesAn UCSF Wang Xian Bierut Amos Nanjing/Beijing, China Spitz Shete Han Li Ma Korea Payne Schwartz Wenzlaff Dueker Japan Kittner Mitchell Taiwan Cheng Sanders Duan Shi GenSalt, China Levinson Gejman Kardia Bergen David ARIC Eaton Furberg Thomas Jennifer Yan Steve DC Yun Ju Mosley Smith Sun Hunt Rao Sung John Helen Paige Margaret Jin Hongbing Zhibin Dongxin Chen Dankyu Taesung Young Jin Yoon Shin Taskashi Jun Chien-Hsiun Jer-Yuarn Ying Ting Fuu-Jen Treva Jiang Dongfeng Hongyan Jiang investigators Wiencke Hansen Bracci Wrensch Guangfu Shen Hu Lin Wu Yoon Park Kim Cho Kohno Yokota Chen Wu Chen Tsai Rice He Gu Huang He • Special acknowledgement to COGEND CTRC KL2 NIDA Louis Fox Sherri Fisher Hilary Davidson collaborators and staff KL2 RR024994 P01 CA89392 International Cross-Population Consortium CHRNA5 rs16969968 is consistently associated with heavy smoking across three populations (Phase I Finding) Bin A rs16969968* European ancestry Sub-bin A-AS1: rs16969968* Asian ancestry Sub-bin A-AA1: rs16969968 African American ancestry Chen et al. 2012, Genetic Epidemiology PHASE II: Meta-Analysis with Imputed Data Cross-Population Meta-Analyses International Consortium Smoking and Chromosome 15q25 European ancestry COGEND MD Anderson MSTF WSU GEOS MGS GENOA HyperGEN ARIC Marchini Oxford samples WTCCC-CAD QIMR UK UK lung cancer Northern Finland Birth Cohort Germany Finnish Study NAG Young Finns Study SHIP NFBC66 Croatian Cohorts Dental Study COGA CADD NYSFS Sardinia Netherland Twin Registry (NTR) SMOFAM Yale study Total- European ancestry N=50,000 Asian ancestry African American ancestry Nanjing COGEND Beijing MD Anderson KARE (Korea) MSTF Tokyo WSU SC (Taiwan) UCSF T2D (Taiwan) GEOS GenSalt (China) AGEN-Chen Peng/Singapore (Malay, Indian, Chinese) MGS AGEN-Ying Wu CLHNS China ARIC AGEN-Jaeseong Korea WHI AGEN-Huaixing China MESA AGEN-Xiao-Ou, China CARDIA, CFS, JHS Wuhan study Dental Study PROMIS Pakistani COGA ABNET's study Total- African American ancestry Total-Asian ancestry N=39,000 N=109,000 GENOA HyperGEN N=20,000 Conclusion on Personalized Medicine • It matters – Minimize medication risk and cost – Target high risk patients – Optimize treatment matching for improved effectiveness • It works – Addiction/Smoke/Onco chip Acknowledgement Washington U In St. Louis Laura Bierut Rich Grucza Sarah Hartz Alison Goate Joseph Bloom Jen Wang Nancy Saccone Rob Culverhouse John Rice Robert Carney Sharon Cresci Richard Bach U Wisconsin Timothy Baker Megan Piper Steven Smith U Utah Dale Cannon Robert Weiss Harvard U Pete Kraft Nancy Rigotti Darmouth Christopher Amos RTI Eric Johnson Michigan State U Naomi Breslau U Minnesota Dorothy Hatsukami U Bristol Marcus Munafo Cross-population Consortium on Genetics of Smoking chenli@psychiatry.wustl.edu Extra Slides Smoking Cessation and Psychiatric Disorders • Patients with psychopathology are less likely to quit • Quitting failure-> decreased mental health • Patients with anxiety have decreased response to treatment • Introducing genetics: – Hypothesis: Negative affect decrease cessation in subjects with high genetic risk. Smoking Cessation Trial (TTURC) Fast Metabolizers benefit from NRT Fast metabolizers (n=409) Cigarettes per day (CPD) 10 8 placebo 6 lozenge 4 patch patch+lozenge 2 0 1 2 3 4 5 6 7 8 Post-quit Treatment Weeks Slow metabolizers (n=145) Cigarettes per day (CPD) 10 8 placebo 6 lozenge patch 4 patch+lozenge 2 0 1 2 3 4 5 6 Post-quit Treatment Weeks 7 8 What is new • PNAT – Patch: slow metabolizers quit better – Spray: no difference – Placebo: slow metabolizers quit better – Bupropion: no difference • We confirm placebo and bupropion • New – PNAT: It was unknown if NRT vs placebo differ by NMR – we find NRT vs placebo effect differ with CYP2A6 (like their spray substracting placebo effect if it exists) – Combo is better than mono Genes, Environment, and Clinical Prediction We know genetic (G) risk is modified by treatment Is environmental (E) risk modified by G? Does treatment alter G by E risks? Smoking Pregnant Women Partner Smoking: Partner Smoking Is Worse in Individuals with CHRNA5 Risk (G*E) 3 Testing G Cig per day 2 2 GG GA 1 AA 1 0 Testing G *E Cig per day CPD0 CPD1 Time CPD2 3 GG/no partner smoking 2 GG/partner smoking 2 GA/no partner smoking 1 GA/partner smoking 1 AA/no partner smoking 0 CPD0 CPD1 Time CPD2 AA/partner smoking Interaction of rs16969968 and partner smoking on quitting (decrease of smoking quantity over time) is significant (n=869, t=2.60, p=0.017 in ALSPAC, and n=104, t=2.97, p=0.0033 in TTURC) Partner Smoking: Environmental Effect Is Stronger in Individuals with CHRNA5 Risk Alleles (G*E) Smoking Pregnant Women Cig per day Testing G Testing G *E 3 3 2 2 2 GG2 1 GA 1 AA 1 0 0 1 CPD0 CPD1 Time GA/no partner smoking GA/partner smoking AA/no partner smoking CPD0 CPD1 Time CPD2 35 AA/partner smoking 30 GG/partner smoking 25 25 GG/no partner smoking 20 GG20 15 GA15 10 AA 10 5 5 0 0 30 CO level GG/partner smoking CPD2 35 Cessation Trial Placebo GG/no partner smoking GA/partner smoking GA/no partner smoking AA/partner smoking CO1 CO2 CO3 CO4 CO5 Time CO6 CO7 Time CO1 CO2 CO3 CO4 CO5 CO6 CO7 Interaction of rs16969968 and partner smoking on quitting (decrease of smoking quantity over time) is significant (n=869, t=2.60, p=0.017 in ALSPAC, and n=104, t=2.97, p=0.0033 in TTURC) AA/no partner smoking Genetic Effects (main G and G*E) in the placebo group can be neutralized by medication Testing G 35 35 CO level 30 25 Placebo N=104 30 GG/partner smoking 25 GG/no partner smoking 20 GG 20 GA/partner smoking 15 GA 15 GA/no partner smoking AA 10 10 AA/partner smoking 5 5 0 0 CO1 CO2 CO3 CO4 CO5 Time CO6 CO7 CO1 CO2 CO3 CO4 CO5 CO6 CO7 35 35 CO level 30 Treated N=765 Testing G *E 25 AA/no partner smoking Time 30 GG/partner smoking 25 GG/no partner smoking 20 GG 20 GA/partner smoking 15 GA 15 GA/no partner smoking AA 10 10 AA/partner smoking 5 5 0 0 CO1 CO2 CO3 CO4 Time CO5 CO6 CO7 Medication neutralizes the G effect (n=869, t=2.60, p=0.0093) Medication neutralizes the G*E effect (n=869, t=3.59, p=0.00034) CO1 CO2 CO3 CO4 CO5 CO6 CO7 Time AA/no partner smoking Combination of G and E informs who will benefit from treatment • Most cessation is unassisted – during pregnancy or post-MI • In unassisted cessation, there is a G*E interaction on quitting – accentuated E effect with risk G, or – expression of G effect with risk E • Medication neutralizes both the main effect of G and G*E Future Goals • Generalize to diverse populations • Design mechanism-specific treatments • Develop treatment algorithm incorporating multiple G, E, and other predictors • Conduct cost benefit analysis of random vs. genotype-based treatment Response to Treatment Differs by Haplotype a. Haplotype H1 (GC) RH=0.83, p=0.36 b. Haplotype H2 (GT) RH=0.48, p=2.7*10-8 c. Haplotype H3 (AC) RH=0.48, p=9.7*10-7 Placebo Active Treatment Chen et al, Am J Psychiatry 2012 The CHRNA5 genetic effect does not differ by type of pharmacotherapy 1.0 0.9 0.8 0.7 0.6 Abstinence 0.5 H1 0.4 H2 H3 0.3 0.2 0.1 0.0 Placebo Buproprion only NRT only Combined No difference in haplotypic risks on cessation across medication groups (wald=1.16, df=3, p=0.88) Chen et al, Am J Psychiatry 2012 Fast metabolizers on placebo treatment have a significantly faster escalation into heavy smoking over time Cigarettes per day (CPD) 10 8 Fast metabolizer on placebo (n=72) 6 Slow metabolizer on placebo (n=27) 4 Fast metabolizer on active medication (n=521) Slow metablizer on medication (n=224) 2 0 wk 1 wk 2 wk 3 wk 4 wk 5 wk 6 Post-quit Treatment Weeks A significant interaction t=3.13, df=1, p=0.0020. wk 7 wk 8 Phase II goals • Genotyped data -> imputed data – Because some variants were not genotyped – Can impute insertions and deletions • Expanded smoking behavior phenotypes – Heavy smoking phenotype – Age of quitting • Scientific questions – Refinement of association signals – Identify additional new loci – Identify consistent (or unique), and biologically significant associations 51 CHRNA5 rs16969968 delays smoking cessation Proportion Having Quit rs16969968 genotype + AA + GA + GG Age of Quitting Smoking AGE at Cessation 52 Smoking quantity and age of quitting are both important for risk of lung cancer and COPD Lung Cancer Risk COPD Risk Thun et al, 2013, NEJM