MULTILOCUS ANALYSIS FOR THE IDENTIFICATION OF EPISTATIS IN CARDIOVASCULAR DISEASE

E-mail: antonella.sangalli@univr.it Human genetic studies of several disorders suggest that interactions between genotypes at multiple genes play an im-portant role in modifying the risk of the disease. However the detection of interacting genes is a challenging task. We deter-mined the genotype for 35 candidate genes (63 polymorphisms) in a sample of 728 individuals with angiographically docu-mented coronary artery disease (CAD+, cases), and 311 individuals with angiographically documented normal coronary arter-ies (CAD-, controls). Genes have been selected for their possible involvement in lipid, homocysteine metabolism, blood co-agulation, blood pressure regulation, or inflammation. We conducted an exploratory analysis based on classification tree-based recursive partitioning to discover gene-gene patterns (GGP) associated with an increase risk of CAD. Five suggestive GGPs resulted to be associated with CAD: APOC3 -641 AA / APOC3 1100 CC (OR-corrected: 3.92 [CI: 1.54-9.95] p=0.0041), APOC3 -455 TT / APOC3 1100 CC (OR-corrected: 3.83 [CI: 1.50-9.76] p=0.00488), APOC3 -641 AA / NPR12238 TC/TT (OR-corrected : 3.53 [CI: 1.15-10.79] p=0.0271), APOC3 -455 TT / NPR1 2238 TC/TT (OR-corrected : 3.43 [CI: 1.13-10.66] p=0.0295), SELE128 ser-ser, ser-arg / LIPH480 TT(OR-corrected: 0.14 [CI: 0.03-0.66 ] p= 0.013). Classification tree-based recursive partitioning represents helpful tools to discover multigene patterns that can test with standard methods. The genetic patterns of risk identified by this study indicate that genes involved in lipid metabolism play an important role in CAD risk and they are combined with genetic factors involved in blood pressure or in cell adhesion.