来源：风湿病与关节炎,2017,6（9）:15-19,24.

1689例骨关节炎患者血小板参数的变化及关联规则挖据研究

（临床研究）

文建庭¹，刘　健²，忻　凌²，万　磊²，方妍妍¹，董文哲¹

　　【摘　要】目的：基于关联规则的方法挖掘骨关节炎患者血小板参数的变化及与其他实验室指标的相关性。方法：纳入在安徽中医药大学第一附属医院住院的骨关节炎患者的病历资料，采用流式细胞法测定骨关节炎患者全血中血小板相关参数指标，运用SPSS 21.0软件对骨关节炎患者血小板计数（PLT）、血小板压积（PCT）、血小板平均体积（MPV）、血小板宽度（PDW）、大血小板比率（P-LCR），炎症指标及免疫指标等实验室指标进行相关性分析及二元Logistic回归分析，并运用SPSS Clementine 11.1软件Aprior模块分析血小板参数与炎症、代谢及免疫指标等实验室指标的关系。结果：①共纳入骨关节炎患者1689例，与PLT计数正常值比较，上升991例（58.67%），正常517例（30.61%），下降181例（10.72%）。②相关性分析结果，PLT、PCT与红细胞沉降率（ESR）、IgA、补体C3、补体C4、α1-AGP呈正相关，与SOD、高密度脂蛋白胆固醇（HDL-C）呈负相关；P-LCR与ESR、IgA、补体C3、补体C4、α1-酸性糖蛋白（α1-AGP）呈负相关，与SOD、HDL-C呈正相关，与其他指标无相关性；PDW与ESR、IgA、补体C3、α1-AGP呈负相关，与SOD、HDL-C呈正相关，与其他指标无相关性；MPV与ESR、补体C3、补体C4呈负相关，与其他指标无相关性。③Logistic回归示WBC、hs-CRP、IgA、补体C3、补体C4、尿酸（UA）是PLT的危险因素（P < 0.01或P < 0.05），WBC、ESR、补体C3、UA、α1-AGP、血总胆固醇（TC）是PCT的危险因素（P < 0.01或P < 0.05），超敏C-反应蛋白（hs-CRP）、α1-AGP是PDW的危险因素（P < 0.01或P < 0.05），UA、TC是MPV的危险因素（P < 0.01或P < 0.05），补体C4、三酰甘油（TG）是P-LCR的危险因素（P < 0.01或P < 0.05）。④设最小支持度为20%，最小置信度为70%，PLT计数升高与hs-CRP、α1-AGP、补体C3、IgG升高，HDL-C下降相关联；PCT、PDW下降与hs-CRP、ESR升高相关联；MPV下降与hs-CRP、ESR、HDL-C升高相关联。结论：活动期骨关节炎患者PLT参数主要表现为指标的升高，同时与炎症、免疫及代谢指标的升高呈相关性，其表达的失调可能参与了骨关节炎的发病及活动，其具体机制有待进一步探讨。

　　【关键词】　骨关节炎；血小板参数；关联规则；数据挖掘

Study on Changes of Platelet Parameters in 1689 Patients with Osteoarthritis and Association Rules

WEN Jian-ting,LIU Jian,XIN Ling,WAN Lei,FANG Yan-yan,DONG Wen-zhe

　　【ABSTRACT】Objective:To investigate changes of platelet parameters in patients with osteoarthritis and their correlation with other laboratory indexes based on association rules.Methods:Clinical data of patients with osteoarthritis treated in the First Affiliated Hospital of Anhui University of Chinese Medicine were included in the study.Flow cytometry was used to determine the relative parameters of blood platelet in patients with osteoarthritis.SPSS 21.0 was used to make a correlation analysis and a Logistics regression analysis to PLT,PCT,MPV,PDW,P-LCR,inflammatory index and immune index.SPSS Clementine 11.1 software was used to make an Aprior module analysis to the relationship between platelet parameters,inflammation index,metabolism and immune indexes.Results:①Among the 1689 cases,991 cases（58.67%）were above the normal PLT count value,the value of 517 cases（30.61%）kept the same,and 181 cases（10.72%）were below the normal PLT count value.②The results of correlation analysis showed that PLT and PCT were positively correlated with ESR,IgA,complement C3,complement C4,and α1-AGP,and negatively correlated with SOD and HDL-C.P-LCR was negatively correlated with ESR,IgA,complement C3,complement C4 and α1-AGP,and positively correlated with SOD and HDL-C,and had no correlation with other indexes.PDW was negatively correlated with ESR,IgA,complement C3,α1-AGP,and was positively correlated with SOD and HDL-C,and had no correlation with other indexes.MPV was negatively correlated with ESR,complement C3,complement C4,and had no correlation with other indexes.③The Logistic regression showed that WBC,hs-CRP,IgA,complement C3,complement C4,UA were the risk factors of PLT（P < 0.01 or P < 0.05）;WBC,ESR,complement C3,UA,α1-AGP and TC were the risk factors of PCT（P < 0.01 or P < 0.05）;hs-CRP and α1-AGP were the risk factors of PDW（P < 0.01 or P < 0.05）;UA and TC were the risk factors of MPV（P < 0.01 or P < 0.05）;and complement C4,TG were the risk factors of P-LCR （P < 0.01 or P < 0.05）.④Set 20% for the minimum support degree and 70% for the minimum confidence level,the increase of PLT count was associated with the increase of hs-CRP,a1-AGP,complement C3 and IgG and the decrease of HDL-C;the decrease of  PCT and PDW was associated with the increase of hs-CRP and ESR;and the decrease of MPV was associated with the increase of hs-CRP,ESR and HDL-C.Conclusion:PLT parameters of patients with osteoarthritis in the active stage show the increase of index,and are correlated with the increase of inflammation,immune and metabolic index.Its expression disorder may be involved in the pathogenesis and activity of osteoarthritis,and its specific mechanism needs to be further discussed.

　　【Keywords】 osteoarthritis;platelet parameters;association rules;data mining