Poster Presentation The 13th International Congress of the Immunology of Diabetes Society 2013

Variable associations of non-HLA risk loci with different phases of the autoimmune process in type 1 diabetes (#89)

Johanna Lempainen 1 2 , Antti-Pekka Laine 1 , Anna Hammais 1 , Riitta Veijola 3 4 , Olli Simell 1 , Mikael Knip 5 6 7 , Jorma Ilonen 1 8
  1. University of Turku, Turku, Finland
  2. Turku University Hospital, Turku, Finland
  3. University of Oulu, Oulu, Finland
  4. Oulu University Hospital, Oulu, Finland
  5. University of Helsinki, Helsinki, Finland
  6. Tampere University Hospital, Tampere, Finland
  7. Folkhälsan Research Center, Helsinki, Finland
  8. University of Eastern Finland, Kuopio, Finland

In addition to the HLA region multiple loci around the genome affect the susceptibility to type 1 diabetes (T1D). We set out to characterize which phase of the disease process is affected by individual non-HLA genes by using a panel of 42 known T1D risk-associated SNP markers genotyped by the Sequenom® platform. We performed a survival analysis using data from 1510 children with HLA-DR/DQ associated disease susceptibility participating in the Finnish Diabetes Prediction and Prevention (DIPP) study.  521 children have seroconverted to positivity for at least two of the four autoantibodies (ICA, IAA, GADA, IA-2A) analyzed or developed T1D after appearance of a single autoantibody. Altogether 269 have developed T1D, whereas 989 children have remained negative for all autoantibodies during the follow-up time. Primary survival analysis using Cox regression to compare three genotypes in each SNP showed, that INS rs689 and IFIH1 rs1990760 affected the development of autoantibodies (P<0.0001, and P=0.022, respectively) while PTPN2 rs45450798, CCR5 rs11711054, IL2RA rs2104286 and CD226 rs763361 were associated with the rate of progression from beta-cell autoimmunity to clinical T1D (P=0.001, P=0.025, P=0.036, P=0.042, P=0.045, respectively). PTPN22 rs2476601 was strongly associated with both phases (P<0.0001 for autoantibody development and P=0.025 for progression to T1D) as seen earlier in a subcohort of these children. When either a dominant or recessive model was applied the significance of the findings was strengthened for markers with less strong association in the primary analysis and several borderline effects (P<0.05) on both autoantibody production and progression to T1D could be seen. The results suggest that many of the disease susceptibility genes are affecting distinct phases of the disease process and their role become more apparent when these phases are analyzed separately. These findings may also facilitate mechanistic studies to focus on processes relevant in each phase of the disease process.