1700-P: Association between Genetic Variation and Age of Onset of HNF1A MODY from the Expression Genetic Standpoint

HNF1A MODY is characterized by wide variability of the age of clinical manifestation. The aim of this study was to identify genes which expression would correlate with age of HNF1A-MODY onset in the skeletal muscle, adipose tissue, liver, pancreas, pituitary gland, colon and small intestine. 837 patients were used for GWAS on age of HNF1A-MODY onset. Data were imputed on MIS. All modeling and statistics was done in R and python. The 'PrediXcan' and related datasets were downloaded from the GitHub repository. Co-expression matrices, Spearman's correlation matrices, were estimated via the function 'cor' in R. We have imputed the genetic component of expression using the PrediXcan software. Two genes were associated with age of onset - RP11-535M15.2 in visceral adipose and C2CD4C in colon. Variants in RP11-535M15.2 were associated in the Roslin Gene Atlas PheWAS with metabolic traits and in C2CD4C with T2DM and the development of the pancreas in multiple studies. Using the inferred expression values we also build co-expression matrices (Spearman's correlation) and inferred that in early onset MODY3 (as compared to late onset) HNF1A targets are more tightly co-expressed. Examples of most significantly deregulated pairs of genes are: SUOX (sulfite oxidase) and MNAT1 (CDK-activating kinase assembly factor 1) in small intestine, HIVEP1 (transcription factor) and KCNH6 (potassium channel) in pituitary gland. All four genes have been previously reported to be associated with diabetes (either in genetic or expression studies).

In conclusion, the use of expression genetics may enhance the agnostic analysis of GWAS result and add biological likelihood to the results of association studies.