超敏C-反应蛋白与缺血性脑卒中发病的前瞻性队列研究

目的 探讨超敏C-反应蛋白（high sensitivity C-reactive protein，hs-CRP）预测缺血性脑卒中发病的适宜切值并评价hs-CRP水平升高与缺血性脑卒中的关联效应。方法 2009年5~9月，采用整群抽样方法在宜兴市官林镇调查35岁以上的居民，共2 238名。排除基线缺血性脑卒中、心血管疾病史者及信息不全者，共2 211人纳入随访（中位随访时间为5.07人年）分析。计算hs-CRP预测缺血性脑卒中发病的生存受试者工作特征（receiver operator characteristic，ROC）曲线下面积，并根据约登指数最大值确定切点。采用Cox比例风险回归模型分析hs-CRP水平升高与缺血性脑卒中发病的关联。结果 在随访期间共发生96例缺血性脑卒中，hs-CRP预测缺血性脑卒中发病的生存ROC曲线下面积为0.647，约登指数最大值为0.270，对应hs-CRP切点为1.17 mg/L。一致性分析结果显示，hs-CRP水平≥1.17 mg/L和≥1.00 mg/L预测缺血性脑卒中发病的一致性极好（Kappa=0.877）。Cox回归分析结果显示，调整协变量后hs-CRP升高（≥1.00 mg/L）与缺血性脑卒中发病关联有统计学意义（HR=2.238，95%CI:1.388~3.608，P=0.001）。结论 hs-CRP≥1.00 mg/L适合作为宜兴地区人群缺血性脑卒中发病预测和高危人群干预的适宜切值。

Association of high sensitive C-reactive protein with stroke: a prospective follow-up study

Objective To investigate an optimal cut-off value of high sensitivity C-reactive protein (hs-CRP) for the prediction of ischemic stroke in Yixing population and to explore the association between hs-CRP level and the risk of ischemic stroke. Methods A total of 2 238 participants over 35 years old were investigated in a rural area (Guanlin, Yixing, Jiangsu; from May to September 2009) using an epidemiological cluster sampling approach. A total of 2 211 participants were followed (median follow-up time was 5.07 person years) after excluding those who had history of ischemic stroke or cardiovascular disease and those whose disease history was incompleted and those who failed to be detected hs-CRP level. Fitted with survival receiver operator characteristic (ROC) curve in which hs-CRP level predicts the incidence of stroke, the optimal cut-off point was determined according to the maximum Youden's index. Cox proportional hazard model was applied to detect the association between hs-CRP level and the incidence of stroke. Results A total of 96 participants developed ischemic stroke in the period of follow-up. The Youden's index reached the maximum (0.270) when the area under ROC curve was 0.647 and the level of hs-CRP was 1.17 mg/L accordingly. The result of symmetric measure showed that predictive incidences of stroke based on the levels of hs-CRP ≥ 1.17 mg/L and ≥ 1.00 mg/L were consistent. Cox regression analyses showed a significant association of high hs-CRP(≥ 1.00 mg/L) with the risk of incident ischemic stroke even after adjusting for the covariates including age, gender, BMI, GLU, HDL-C, LDL-C, TC, TG, smoking and drinking, and hypertension and T2DM (HR=2.238, 95% CI:1.388-3.608,P=0.001). Conclusions Hs-CRP ≥ 1.00 mg/L is appropriate for predicting the incidence of stroke and intervening with the high risk population in Yixing population.