摘要目的 利用脑脊液降钙素原(Procalcitonin,PCT)、乳酸及常规生物标志物建立神经外科术后细菌性脑膜炎(Post-neurosurgical bacterial meningitis ,PBM)的预测模型并验证其诊断价值.方法 收集2017年3月至12月于首都医科大学附属北京天坛医院神经外科收治的术后脑膜炎患者213例,根据脑膜炎类别不同分为 PBM 组( n ＝85 )和无菌性脑膜炎( Post-neurosurgical aseptic meningitis,PAM)组(n＝128).所有患者进行14种生物标志物的检测,通过多因素Logistic回归分析选择与脑脊液PCT和乳酸联合诊断的生物标志物,建立预测模型,利用受试者工作特征曲线( ROC)分析预测模型对术后PBM的诊断价值,并对预测模型进行验证.结果 单因素分析显示,脑脊液细胞计数、脑脊液白细胞计数、脑脊液蛋白浓度、脑脊液葡萄糖浓度、脑脊液葡萄糖／血糖比、脑脊液PCT及脑脊液乳酸为预测术后PBM的相关因素(P值均＜0.01).多因素Logistic回归分析显示,脑脊液PCT、脑脊液乳酸、脑脊液蛋白浓度与脑脊液血液葡萄糖比值为预测术后PBM的独立预测因子,建立预测模型为:4.315×脑脊液PCT+0.822×脑脊液乳酸+0.009×脑脊液蛋白浓度-5.480×脑脊液葡萄糖／血糖比-3.074.预测模型的 ROC 曲线下面积最大,为0.947,敏感度和特异度分别为90.60%和85.10%.通过验证,在区分两种脑膜炎上,预测模型的阳性预测值、阴性预测值及诊断正确率分别为84.06%、94.44%与90.40%.结论 采用脑脊液PCT、脑脊液乳酸、脑脊液蛋白浓度与脑脊液血糖比值进行联合检测对神经外科术后PBM具有较好的诊断价值,可缩短患者的诊断时间并提高患者治疗的成功率.

abstractsObjective To investigate the diagnostic value of procalcitonin ( PCT) and lactate in cerebrospinal fluid (CSF) combined with conventional biomarkers for post-neurosurgical bacterial meningitis (PBM).Methods Clinical data of 213 patients with post-neurosurgical meningitis admitted in Beijing Tiantan Hospital, Capital Medical University from March 2017 to December 2017 were retrospectively analyzed, including 85 cases of PBM and 128 cases of post-neurosurgical aseptic meningitis ( PAM).The diagnostic value of CSF procalcitonin , lactate and other 12 conventional biomarkers for PBM was analyzed by multivariate logistic regression.A prediction algorithm was generated and its diagnostic value for PBM was assessed with receiver operating characteristic curve (ROC).Results The univariate analysis showed that CSF cell count, CSF leukocyte count , CSF protein concentration , CSF glucose concentration, CSF glucose／blood glucose ratio, CSF PCT and CSF lactate were significantly associated with PBM.Multivariate logistic regression analysis showed that CSF PCT , CSF lactate, CSF protein concentration and CSF glucose ／blood glucose ratio were independent predictive factors for PBM.The predictive algorithm score ＝4.315 ×CSF PCT+0.822×CSF Lactate+0.009×CSF protein concentration -5.480×CSF glucose／blood glucose ratio-3.074.The predictive algorithm has the largest area under the ROC curve ( AUC ＝0.947), and the sensitivity and specificity of the predictive algorithm score were 90.60% and 85.10%, respectively.The positive predictive value , negative predictive value and the accurate rate of the algorithm in diagnosis of PBM were 84.06%, 94.44% and 90.40%, respectively.Conclusion The predictive algorithm based on the combination of CSF PCT and CSF lactate with CSF protein concentration and CSF glucose ／blood glucose ratio has a good diagnostic value for PBM.It can shorten the diagnosis time of PBM and improve the clinical outcomes.