Early Progress and Findings from a Genetically Informative Investigation of the PROSPER Prevention Project H. Harrington Cleveland Human Development Family Studies The Pennsylvania State University Overview • • • • • • Issues in candidate gene research The PROSPER Study How gPROSPER leverages PROSPER for G x E Data collection Progress Cleaning and analyzing gPROSPER Early Findings Issues with Candidate G X E Research • Candidate Gene Research – especially G x E research (e.g. Caspi et. al., 2003) – has been aggressively criticized. • Some (e.g. meta-analysis by Risch et.al., 2009) claim that findings are consistent with chance. • Others say that Risch et. al. (2009) is incorrect due to inclusion of weak studies (see Uher & McGuffin, 2010, and Rutter et al., 2010). • Common problems with adding candidate genes to existing studies – Poor measurement – few items, single reporters – Retrospective data – recall bias – Passive large scale data sets are vulnerable to rGE: • reducing exposure to full range of environments across genotypes (reduces power) • undercutting interpretation of findings More about rGE confounds The problem: Families, peers, experiences are all genetically influenced The result: G x E “effects” might be due to genetic selection of environments/experiences Minimum controls: investigate associations between genes and presumptive “environmental moderator” Better solutions: Random assignment Quasi-experimental solutions - school level interventions (e.g., PROSPER) - regression discontinuity designs - state level differences in welfare policies Another Problem: Population Stratification A 3rd variable confound: 1. Samples may be made up of distinct genetic groups or they have been genetic mixing of groups in recent past (i.e., racial admixture). 2. Groups may differ in allelic distributions and outcomes 3. Creating spurious associations between alleles and outcomes 4. Classic Study: Knowler et al. General Finding – Gm haplotype of negatively associated with Type 2 diabetes in Amer Ind./euro sample. But: Gm haplotype more common among European descent cases Analyses within Am. Indian and Euro groups found no association Citations: Kwowler, et al. (1988).GM and Type 2 Diabetis Mellitus: An Association in American Indians with Racial Admixture. American Journal of Human Genetics. 43. 520526. Cardon & Palmer (2003). Population stratification and spurious allelic association. Lancet, 361, 598-604 The PROSPER Parent Study • Prospective Adolescent data collections: From 2002 through 2009, over 8,000 took part in 8 annual data collections. Average participation rate was 89% across waves. In-school data collections provide primary measures of family and peer environment and substance use outcomes for grant aims. • Randomized community design: PROSPER is organized around 28 rural communities in Iowa and Pennsylvania randomized into 14 control and 14 intervention units. • Interventions: Communities chose one of three similar school-based program: Life Skills Training Program (Botvin, 2000), All Stars Program (McNeal et al., 2004), and Project Alert (Ellickson et al., 2003). All programs target social norms, personal goal setting, decision-making, and peer group affiliation. Interventions were implemented during required 7th grade health classes. • Social network data: Social network data collected across adolescence provide intricate measures of peer micro-environments from 6th through 12th grade. Fully coded forgrades 6 to 9, rhese measures allow investigation of if and how peer experiences mediate and moderate genetic influences on substance use outcomes. • In-Home sub-sample: Detailed parent and child reports of family processes, including video-taped interactions (n=600 with genotyped DNA): 5 waves of measures for family environment, parenting practices, relationships, from beginning of 6th grade to end of 9th grade. • Early adult data collection: A follow-up sample is currently participating in two annual telephone/webbased surveys, at ages 19 and 20. These data collections oversample high-risk participants. Primary Environmental Measures Community Environments Community intervention status Community crime Community alcohol and tobacco availability SES and community disorganization Peer Environments Peer substance use Unsupervised time w/ substance using peers Substance use by clique Degree of social isolation Family Environments Parent-child affective quality Parent-child joint activities Parental monitoring Parental discipline Substance Use and Abuse Measures Adolescent Measures. Set 1: Ever Used, Past Month, and Past Year: Alcohol, Cigarettes, Other Drugs (depending on time frame: marijuana, inhalants, methamphetamine, ecstasy, other people‘s prescription drugs, and prescription pain killers). Early Adult Measures. Over 45 items across a broad range of substances. • From tobacco and alcohol to marijuana and other illegal drugs (e.g., cocaine, inhalants, methamphetamine, ecstasy, hallucinogens, other people‘s prescription drugs, and prescription pain killers). • Alcohol use detail: including frequency, volume, getting drunk, and different thresholds of binging. • Alcohol abuse and dependency diagnostic criteria (DIS). Peer substance use. Peer Measures Focus on reciprocated nominations (reciprocation rates are about 50%). In addition: ‘Sent‘ and ‘received‘ networks, as well as best friends’ behaviors. Unsupervised time with substance using friends. Item: How often students “hang out (without adults around)” outside of school with each nominated friend. Allows indices of unsupervised and unstructured exposure to substance using peers (Osgood et al., 1996; Osgood, 2010). Clique substance use. Cliques provide the proximate social context for information about group norms. PROSPER network project (Osgood) provides clique substance use levels. Degree of social isolation. Core clique members: 77% Peripheral clique members: 16% Non-group members (including isolates): 7% Family Measures Parent-child affective quality. Parental Affective Quality toward Child and Child’s Affective Quality toward Parents. Across the 8 waves of in-school assessment, Alphas range from .86-.95 (mother subscale) and .91-.96 (father subscale). Parent-child joint activities. The amount of time and range of activities shared by parents and adolescents is assessed with 6 items, such as “work on homework or a school project together”. Alphas range from .87 to.88. Parental monitoring. Assessed with 5 items measuring parents’ knowledge of adolescent’s whereabouts and activities. Contributing items include “During the day, my parents know where I am.” Alphas for this scale ranged from .78 to .88. Parental discipline. Parents‘ consistent discipline was measured with four items, including “My parents discipline me for something some times, and then other times don‘t discipline me for same thing.”‖Alphas range from .71 to .82. Community Measures Community intervention status. 14 coded as intervention. 14 coded as comparison communities. Community crime. Community violent crime (Alpha = .79; murder, rape, robbery, and aggravated assault), property crime (Alpha = .96; arson, burglary, larceny, and motor theft), and narcotic-related crime. Community alcohol and tobacco availability. Number of outlets each community; Number divided by community size; Distance from school. SES and community disorganization. Percentage students receiving free and reduced lunches, ranging from 11% to 53%; percentage of households in poverty; residential instability (ranging from 25.4 -45.5%); and percent single female-headed households. gPROSPER Overview Implications of Genetic Variance for Substance Use Interventions in Adolescence Funding Source NIDA Bo Cleveland, The Pennsylvania State University: M-PI David Vandenbergh, The Pennsylvania State University: M-PI Mark Feinberg, The Pennsylvania State University: Co-Principal Investigator Mark Greenberg, The Pennsylvania State University: Co-Investigator Jenae Neiderhiser, The Pennsylvania State University: Co-Investigator Richard Spoth, Iowa State University: Consortium PD/PI Particular strengths of PROSPER for GxE work o Detailed measurement of environment and outcomes: Developed to assess early adolescent interventions designed to operate through family and peer contexts, a primary focus of PROSPER is measuring processes within these environmental domains. o Randomized prevention trial design: Community-level randomized of interventions removes person-level selection –both general and rGE– of intervention-related experiences. o Prospective cohort design: Reduces perception-related rGE by not using long-term retrospective report of environmental factors (Jaffe & Price, 2007). o Large sample size: 1,500-1,800 for adolescent and young adult analyses 1,800-2,200 for adolescent-only analyses Opportunity for self-replication by splitting sample into exploratory and confirmatory subsamples 600 with hyper-measured family relationship data Conceptual model: How genes may influence substance use and transact with intervention, family, and peer factors gPROSPER Aims Specific Aim 1: Examine impact of specific genetic risk on modifying the effects of preventive interventions on the development of substance use behaviors. Specific Aim 2: Characterize the interplay between proximal environmental measures (parent-child relationship quality, parental monitoring, peer micro-environments) and specific genes on substance use and abuse during adolescence and early adulthood. Specific Aim 3: Examine how GxE and rGE processes involved in substance use outcomes change from early adolescence to early adulthood and from initiation through addictive use. gPROSPER Data Collection Framework PROSPER and Follow-up Adolescent PROSPER Adolescent In-School Surveys N = 8,000 gPROSPER Supplement DNA collection N = 1900 Supplemental Early Adult Interviews N = 900 Follow-up Study In-Home Sample N = 538 gPROSPER Supplement (in-home) Sample w/ both DNA and adolescent interviews N = 2,200 Analysis Data Sets Follow-up Early Adult Interview N =1,000 * Adolescent and Early Adult Interviews w/ DNA N = 15,00-1,800 Data Collection Progress • 530 pre-existing in-home DNA samples • Years 1 and 2 adding 1200 more DNA samples – From both follow-up and supplemental data collections • Year 3: Add 500 more DNA only participants Candidate Genes Initial list Additions: Oxytocin and Vasopressin markers Adopted approach (split sample approach): • 600 In Home “richly phenotyped” samples – 318k SNP array of coding exons – Commonly studied VNTRs (e.g., DRD4, 5-httplr) • 1400-1800 (early adult subsample) – 30-35 candidate genes (approx 300-500 SNPs): – Selected by function (e.g., dopamine related genes) and/or subsample GWAS findings Recruiting PROSPER participants into GPROSPER • Follow-up sample: • DNA recruitment after completing survey • Supplement (gPROSPER) • Recruit into DNA and survey at outset – Get DNA sample in lab, then do send survey • Recruit into DNA-only data collection • Sample Size Issues: – Is 2,200 small? • Depends on study design – Passive correlational – Assignment with manipulation/intervention » See Brody Research – Controlled environments (social psych experiments) • Measurement Quality – – – – Retrospective Report vs. Prospective Design Omnibus Study vs. Targeted Study Single SNPs vs. Haplotypes Single vs. Multi-Locus Constructions Part 4: Early Findings Based on 560 In Home cases Preliminary work - Dropped overly related individuals - Developed racial stratification scheme - Analyses were done: 1. On all cases 2. Controlling for dimensions of racial stratification (PC1 and PC2) 3. Dropping all non-euros Preliminary Cleaning: Finding and Dropping Related Individuals 319k markers vs. 41k markers Racial Stratification in Prosper Racial Stratification in Prosper 1 sd from mean of pc 1 among self-identified non-euros Combined DRD4 by Intervention Effects on Association between Maternal Activities with Child at 6th grade and Child Alcohol Use by 9th grade DRD4: Dopamine related VNTR gene • Coding: 7 repeats vs. others: DRD4 7+ vs. DRD4 7DRD4 7 + linked to: 1. ADHD, substance use, addiction, novelty seeking 2. Developmental susceptibility to environment (linked to attentional processes) 3. Experimental studies link drd4 7+ with: - greater sensitivity to confederate drinking - greater perception of social bonding when drinking gPROSPER Early Findings Based on 560 In-Home cases 1. 2. Intervention Effects on Associations between Maternal Activity and Alcohol Initiation among DRD4 7+ Adolescents Moderation of Maternal Closeness and Resistance of Peer Pressure by DRD4 Population Stratification Checks: 1. Preliminary: Examine PC associations with DV & environmental variable 2. Run full model 3. Repeat with PC as statistical controls 4. Repeat with dropping non-euros allelic distance Predicting Wave 5 Alcohol Initiation by Wave I Mother Activity and Intervention Status Three-way Int*MCA*D4 -.47(.21)* Sensitivity Analysis Intervention Effects of the Association between 6th grade Maternal Activities and Alcohol Use Initiation by 9th Grade among DRD4 7+ Adolescents Thanks to Gabriel Schlomer Association between Maternal Closeness and Belief in Ability to Resist Negative Peer Influence as a Function of DRD4 0.25 Resist Negative Peer Influence 0.2 0.15 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2 -1SD MEAN +1SD Closeness DRD4 7- 0.045 -0.02 -0.085 DRD4 7+ -0.14 0.033 0.205 Note: Effect goes away with control for population stratification