Genetic and Environmental Risk Factors in the Development of Obesity

Lifestyle behaviors and genetic variation have clear and distinguishable effects on obesity risk; however, the
pattern of disease occurrence within and between populations that differ in their genetic and environmental
underpinnings suggests obesity is caused in part by the interaction between adverse lifestyle behaviors and the
genetic risk profile of an individual. This thesis aims to investigate the joint effects of genetic and environmental
(specifically lifestyle) risk factors for obesity and its comorbidities using cross-sectional and longitudinal
epidemiological cohorts and clinical trials. Characterizing interactions may help optimize prevention and treatment
strategies by identifying risk groups of people for targeted interventions. The work in this thesis was conducted in
the cross-sectional European ancestry sample of 111,421 individuals from the GLACIER, MDC, INTER99,
HEALTH2006, HPFS, NHS, WGHS, InterAct, MESTSIM, TwinGene cohorts (paper I), 14,131 Pakistani adults
from the PROMIS cohort (paper II), 3,541 adult from the prospective GLACIER Study (paper III) and in 5,730
participants of the DPP and Look AHEAD clinical trial (paper IV). In paper I, we reported that physical activity,
assessed by the Cambridge physical activity index, diminishes the genetic risk of obesity predisposed by 12 BMIassociated
genetic variants (Pinteraction=0.015). In sensitivity analyses, the interaction was only evident in the
Northern American (N= 39,810) but not the European (N= 71,611) cohorts. In paper II, by employing genomewide
heterogeneity of variance approach in GWAS data in PROMIS study, we identified one locus, FLJ33534
rs140133294 that associated with variance of BMI (P-value of 3.1 x 10-8). In subsequent analysis the association
of this locus on BMI was found to be significant modified by smoking (Pinteraction= 0.0005). In paper III, we a
genetic risk score based upon 97 BMI-associated genetic variants was significantly associated with change in BMI
(β=0.014 kg/m2 per allele per 10-year follow-up, SE= 0.006, P=0.015). Three of the BMI- loci (PARK2
rs13191362, C6orf106 rs205262, C9orf93 rs4740619) were individually significantly associated with 10 year
change in BMI. In paper IV, we observed that lifestyle interventions modified the response of MTIF3 rs1885988
genetic variant to weight loss. In the meta-analyzed sample of DPP and Look AHEAD, each copy of the minor Gallele
at MTIF3 rs1885988 was associated with weight loss across all four years of the lifestyle interventions (P=
2.4×10-3), while no association with weight loss was observed in the control arm (P= 0.11). In conclusion, this
thesis work shows that gene-lifestyle interactions influence human weight maintenance on several levels, although
the clinical relevance remains to be determined.