OBJECTIVETo investigate the association of ten SNP at peroxisome proliferator-activated receptors (PPARα, δ, γ) with hypertriglyceridemia and the gene-gene interaction.

METHODSParticipants were recruited from the Prevention of MetS and Multi-metabolic Disorders in Jiangsu province of China Study (PMMJS). A total of 820 subjects were selected from the 4083 participants who had received follow-up examination, by using simple random sampling. Participants in baseline and follow-up study surveys were both collected blood samples 11 ml in the morning after at least 8 hours of fasting. Blood samples which collected at the baseline were subjected to PPARα, PPARδ and PPARγ genotype analyses. Blood samples which collected at the follow-up were used to measure serum triglyceride levels. The logistic regression model was used to analyze the association between different SNP and hypertriglyceridemia, and the generalized multifactor dimensionality reduction (GMDR) was applied to explore the gene-gene interaction.

RESULTSThe samples included 474 in the non-hypertriglyceridemia group and 346 in the hypertriglyceridemia group. The genotype frequencies of rs1800206 in the hypertriglyceridemia group were 211 (61.0%) for LL, 132 (38.2%) for LV and 3 (0.9%) for VV, and in the non-hypertriglyceridemia group were 411 (86.7%) for LL, 59 (12.4%) for LV and 4(0.8%) for VV (χ(2) = 74.18, P < 0.01). V allele frequencies of rs1800206 in the hypertriglyceridemia group was 138(19.9%), and in the non-hypertriglyceridemia group was 67 (7.1%) (χ(2) = 60.62, P < 0.01). The genotype frequencies of rs2016520 in the hypertriglyceridemia group were 177 (51.2%) for TT, 154 (44.5%) for TC and 15 (4.3%) for CC, and in the non-hypertriglyceridemia group were 211 (44.5%) for TT, 212 (44.7%) for TC and 51 (10.8%) for CC(χ(2) = 15.93, P < 0.01). C allele frequencies of rs2016520 in the hypertriglyceridemia group was 184(26.6%), and in the non-hypertriglyceridemia group was 314 (33.1%) (χ(2) = 8.07, P < 0.01). The genotype frequencies of rs3856806 in the hypertriglyceridemia group were 149 (43.1%) for CC, 156 (45.1%) for CT and 41 (11.8%) for TT, and in the non-hypertriglyceridemia group were 269 (56.8%) for CC, 170 (35.9%) for CT and 35 (7.4%) for TT (χ(2) = 15.93, P < 0.01). T allele frequencies of rs3856806 in the hypertriglyceridemia group was 238(34.4%), and was 240 (25.3%) in the non-hypertriglyceridemia group (χ(2) = 15.96, P < 0.01). The genotype frequencies of rs1805192 in the hypertriglyceridemia group were 145 (41.9%) for PP, 158(45.7%) for PA and 43(12.4%) for AA, and in the non-hypertriglyceridemia group were 314 (66.2%) for PP, 137(28.9%) for PA and 23(4.9%) for AA (χ(2) = 50.92, P < 0.01). A allele frequencies of rs1805192 in the hypertriglyceridemia group was 244(35.2%), and was 183 (19.3%) in the non-hypertriglyceridemia group(χ(2) = 52.89, P < 0.01). After adjusting age, gender, smoking, alcohol consumption, high-fat diet, low -fiber diet and occupational physical activity factors, rs1800206, rs2016520, rs3856806 and rs1805192 were significantly associated with hypertriglyceride, while the OR (95%CI) was 3.88 (2.69 - 5.60), 0.71 (0.52 - 0.96), 1.40 (1.03 - 1.90) and 2.56 (1.88 - 3.49), respectively (P < 0.05). GMDR model analysis showed that the second-order model (rs1800206 and rs1805192) was the best model when quality traits of triglyceride was chosen as outcome (P < 0.01); while third-order model (rs1800206, rs1805192 and rs2016520) was the best model when quantitative traits of triglyceride was chosen as outcome (P < 0.01).

CONCLUSIONThe rs1800206, rs2016520, rs3856806 and rs1805192 were significantly associated with hypertriglyceridemia. There was a gene-gene interaction between multiple SNP.