Longitudinal analysis of genetic susceptibility and body mass index throughout adult life
Song et. al 2017 preprint in Diabetes (Critique by Hannah Lennon)
09 Jan 2018
- SNPs occur once in every 300 nucleotides on average
- Roughly 10 million SNPs in the human genome
- Most commonly, these variations are found in the DNA between genes. Can act as biological markers, helping scientists locate genes that are associated with disease.
- When SNPs occur within a gene or in a regulatory region near a gene, they may play a more direct role in disease by affecting the gene’s function.
Background: Nature paper, Locke et al, 2013 with n=339,224
- 97 BMI-associated SNPs each with an \( \hat \beta>0 \) effect size indicating up-regulation of expression with the allele
Genetic Risk Score (GRS) has values from 0 to 194 and defined by \( GRS=\frac{97 \times \sum_{i=1}^{97} \beta_i x_i }{\sum_{i=1}^{97} \beta_i} \) where \( SNP_i \) is recoded to \( x_i=0,1,2 \) according to number of risk alleles
- Each unit of GRS represents one risk allele
Song study aim: Assess the relationship of BMI over adulthood (ages 18/21 and 45-75) with GRS Question: Is the effect of genetic risk of adult BMI and changes in BMI modified by age? (Test for an interaction Age:GRS)
Data n~10k women, 6.5k men with a genetic data GRS
- with BMI recorded by questionnaire every 2 years and the average used for ages 45, 50, 55, 60,…, 80, 85 years
For person \( i=1, \dots, n, \, \) at time \( t = 18/21, 45, \dots, 85 \),
\( Model \, 1: \quad BMI_{it} = \beta_0 + {\color{red} \beta_1} GRS^*_{i} + \beta_2 age_{it} + \epsilon_{it}, \)
\( Model \, 2: \quad BMI_{it} = \beta_0 + { \color{red} \beta_1 GRS^*_{i}} + \beta_{2} age_{it} + b_0 age_{it} + \epsilon_{it}, \)
\( Model \, 3: \quad BMI_{it} = \beta_0 + { \color{red} \beta_1 GRS^*_{i}} + \beta_{2} age_{it} + \beta_{3} age^2_{it} + b_0 age_{it} + \epsilon_{it}, \)
Conclusion 1: Effect modicication of GRS:Change in BMI in Women only
For each time period 18-45, 45-65 and 65-80
\( Stratified \, Model: \quad \Delta BMI_{i} = \beta_0 + { \color{red} \beta_1 GRS^*_{i}} + \epsilon_{i}, \)
