Opportunities and Challenges in the Design of a Longitudinal Gene x School-Based Intervention Study

Recent experimental evidence of gene x intervention (G x I) interactions suggests the importance of assessing genetic influences on intervention outcomes in randomized trials (Brody et al., 2009; Cleveland et al., in press).  However, while whole-school preventive interventions have demonstrated postive impacts on children’s social, emotional and academic development (Durlak et al, 2011), only one school-based intervention study to date has examined whether intervention impacts may be modified by genetic factors associated with children’s differential sensitivity to environments (Musci et al, 2013).  This presentation addresses design and practical issues related to the collection of saliva-based DNA samples as part of a final wave of assessment in a longitudinal study of approximately 625 youth originally assessed in the fall and spring of 3^rd, 4^th and 5^th grades as part of a school randomized trial of a social and emotional learning intervention involving 18 New York City elementary schools.  Post-intervention follow-up assessments with these youth were also conducted during their 7^th or 8^th grades (middle school) and 9^th or 10^th grades (high school). Mental health indicators were assessed at each wave including aggression/violence, depression, ADHD, substance use (starting in grade 5) and extensive data was collected on classroom-level stressors and supports.  The elementary school intervention was a 3-year school-wide program involving professional development of teachers to promote children’s social, emotional, and academic skills and prevent or slow the emergence of later health risk behaviors.  Positive impacts have been observed after two years on depression and aggressive behavior, and on math and reading achievement for high behavioral risk children (Jones et al, 2011). The current study will test whether intervention impacts on youth mental health outcomes differ based on candidate gene polymorphisms that have been linked in prior association or intervention studies to these outcomes. Specifically, polymorphisms in three different candidate genes (MAOA, 5-HTTLPR, DRD4) will be tested as moderators of the intervention’s impact on aggression and depression. Design issues will focus on (i) potential risk-based (including genetic) selection bias in reconsenting the sample for DNA collection, (ii) consenting for a limited and identified set of candidate genes or an extensive microarray for derivation of relevant polygenic sensitivity, (iii) implications of prevalence rates of key allelic variants in the general population and across demographic subgroups for testing G x I effects, and (iv) design limitations in assessing epigenetic variation that may follow directly from intervention exposure and systematically influence targeted outcomes.