中文摘要：

目的探讨过氧化物酶体增殖物激活受体（PPAR）α／δ／γ单核苷酸多态性（SNP）基因-基因交互作用对原发性高血压（EH）的影响。方法研究对象均来自于“江苏省多代谢异常和代谢综合征综合防治研究（PMMJS）”队列人群。采用单纯随机抽样方法抽取820名研究对象的基线血标本进行PPAR α／δ／γ（PPARα：rs135539、rs1800206和rs4253778；PPARδ：rs2016520和rs9794；PPARγ：rs10865710、rs1805192、rs4684847、rs709158和rs3856806）多态性检测，运用广义多因子降维法（GMDR）模型检测10个SNP的基因-基因交互作用与EH的关联。结果调整性别、年龄、体重指数、空腹血糖、甘油三酯、高密度脂蛋白胆固醇、高脂饮食、低纤饮食和体力活动后，以EH、SBP和DBP质量性状为结局的GMDR最优模型分别为七、九维度模型（EH：交叉一致性为9／10和10／10，平均检验准确度为0．5862和0．5885）、五、九维度模型（SBP：交叉一致性为10／10和8／10，平均检验准确度为0．6055和0．6011）和八、九维度模型（DBP：交叉一致性均为10／10，平均检验准确度为0．5926和0．5972）。SBP和DBP数量性状的最优模型分别为四、五维度模型（SBP：交叉一致性为10／10和8／10，平均检验准确度为0．6111和0．6072）和五维度模型（DBP：交叉一致性为9／10，平均检验准确度为0．5753）。结论PPAR的10个SNP中多个SNP之间存在交互作用，且对血压水平具有显著影响。

英文摘要：

Objective To explore the impact of the gene-gene interaction among the single nucleotide polymorphisms （SNPs） of peroxisomc proliferator-activated receptor α/δ/γ on essential hypertension （EH）. Methods Participants were recruited based on the previous work of the PMMJS （Prevention of Multiple Metabolic Disorders and Metabolic Syndrome in Jiangsu Province） cohort study in Jiangsu province of China. A total number of 820 subjects were randomly selected from the cohort and received gene polymorphism detection covered ten SNPs： PPAR α/δ/γ （PPARα ： rs135539, rs1800206 and rs4253778 ; PPARδ ： rs2016520 and rs9794; PPARγ： rs10865710, rs1805192, rs4684847, rs709158 and rs3856806）. Generalized Multifactor Dimensionality Reduction （GMDR） model was used to evaluate the association between gene-gene interaction among the ten SNPs and EH. Results After adjusting factors as gender, age, BMI, FPG, TG, HDL-C, high fat diet, low fiber diet and physical activity, results from the GMDR analysis showed that the best qualitative trait models were 7/9-dimensional model （EH： cross-validation consistency were 9/10 and 10/10,prediction accuracy were 0.5862 and 0.5885） , 5/9-dimensional model （SBP： cross-validation consistency were 10/10 and 8/10, prediction accuracy were 0.6055 and 0.6011 ）, and 8/9-dimensional model （DBP： cross-validation consistency both were 10/10, prediction accuracy were 0.5926 and 0.5972）, while the best quantitative trait models were 4/5-dimensional model （SBP： cross-validation consistency were 10/10 and 8/10, prediction accuracy were 0.6111 and 0.6072）, and 5-dimensional model （DBP： cross-validation consistency were 9/10, prediction accuracy were 0.5753）. Conclusion Interactions among ten SNPs of PPARs seemed to have existed and with significant impact on the levels of blood pressure.