2019 年第 11 期 第 14 卷

重型颅脑损伤患者中性粒细胞与淋巴细胞比值动态变化及预后模型的构建

Dynamic changes of neutrophil to lymphocyte ratio in patients with severe traumatic brain injury and the construction of prognostic model

作者：刘健羽王永谦王维平

英文作者：

单位：200030上海中医药大学附属龙华医院神经外科

英文单位：

关键词：重型颅脑损伤；中性粒细胞与淋巴细胞比值；预后

英文关键词：

• 摘要：

• 【摘要】目的    探讨重型颅脑损伤（sTBI）患者入院后7 d内的中性粒细胞与淋巴细胞比值（NLR）水平变化特点并根据Logistic回归结果构建预测6个月预后的模型。方法    回顾性选取2014年1月至2018年1月在上海中医药大学附属龙华医院治疗的261例sTBI患者，收集sTBI患者入院后7 d内的NLR。根据出院后6个月的格拉斯哥预后量表（GOS）评分将所有患者分为预后良好组(GOS评分≥4分)及预后较差组（GOS评分＜4分）。比较2组患者入院后NLR变化特点及水平，分析出院后6个月预后的危险因素。构建预后预测模型，根据多因素Logistic回归结果联合受试者工作特征曲线评价模型准确性。结果    261例sTBI患者均在出院后6个月获得随访，随访率100%，其中预后较差组59例，预后良好组202例。在入院后的6 d内，预后较差组患者NLR水平均高于预后良好组患者，差异均有统计学意义（均P＜0.01）。2组患者NLR峰值均出现在入院后第3天。多因素Logistic回归结果显示，年龄（越大）、入院时格拉斯哥昏迷量表（GCS）评分（越低）、第1天NLR和第3天NLR（越高）是sTBI患者出院后6个月预后较差的独立危险因素（均P＜0.01），年龄、入院时GCS评分、第1天NLR和第3天NLR预测sTBI患者出院后6个月预后的曲线下面积（AUC）分别为0.720（95%置信区间：0.665～0.772）、0.796（95%置信区间：0.745～0.841）、0.760（95%置信区间：0.708～0.809）、0.780（95%置信区间：0.727～0.829）。根据本研究所得的4个独立危险因素进行预后预测模型的构建，模型1：年龄+入院时GCS评分；模型2：年龄+入院时GCS评分+第1天NLR；模型3：年龄+入院时GCS评分+第3天NLR；模型4：年龄+入院时GCS评分+第1天NLR+第3天NLR。模型1、2、3、4的AUC分别为0.824、0.853、0.932、0.978，其中模型4的AUC最大，表明其预测预后的准确度最高。结论    入院时NLR和入院后第3天出现峰值的NLR是sTBI患者预后较差的独立危险因素。NLR结合年龄及入院时GCS评分共同预测sTBI患者预后的准确性显著提高。

• 【Abstract】Objective    To investigate the changes of neutrophil to lymphocyte ratio(NLR) in patients with severe traumatic brain injury(sTBI) within 7 days after admission; to construct a prediction model for 6-month prognosis based on logistic regression results. Methods    From January 2014 to January 2018, 261 patients with sTBI in Longhua Hospital, Shanghai University of Traditional Chinese Medicine were retrospectively analyzed. NLR values in 7 days after admission were collected. According to Glasgow Outcome Scale(GOS) score at 6 months after discharge, all patients were divided into good prognosis group (GOS score≥4) and poor prognosis(GOS score＜4) group. Dynamic characteristics of NLR after admission and risk factors of 6-month prognosis were analyzed. A prediction model for 6-month prognosis was established. The accuracy of the model was evaluated by multivariate logistic regression and receiver operating characteristic curve. Results    The 6-month follow-up rate was 100%. Among them, 59 patients had poor prognosis and 202 patients had good prognosis. NLR values in 6 days after admission in poor prognosis group were significantly higher than those in good prognosis group(P＜0.01). The peak value of NLR appeared at the 3rd day after admission in both groups. Multivariate logistic regression showed that hith age, low GCS score at admission(low), high NLR at the 1st and 3rd days after admission（NLR-1 d, NLR-3 d） were independent risk factors of poor prognosis(all P＜0.01). Areas under the curve(AUC) of age, GCS score at admission, NLR-1 d and NLR-3 d were 0.720(95% confidence interval: 0.665-0.772), 0.796(95% confidence interval: 0.745-0.841), 0.760(95% confidence interval: 0.708-0.809) and 0.780(95% confidence interval: 0.727-0.829). The prognostic prediction model was constructed based on the 4 risk factors. Model 1: age+GCS score at admission; Model 2: Age+GCS score at admission+NLR-1 d; Model 3: age+GCS score at admission+NLR-3 d; Model 4: age+GCS score at admission+NLR-1 d+NLR-3 d. Model 4 showed the best accuracy(AUC: 0.978 for Model 4＞0.932 for Model 3＞0.853 for Model 2＞0.824 for Model 1). Conclusions    NLR values at the 1st and 3rd days after admission are independent risk factors of poor prognosis in patients with sTBI. NLR combined with age and GCS score shows high accuracy in predicting the prognosis.