系统性炎症等级与急性胰腺炎严重程度的相关性及其预测价值分析

Association of systemic inflammation grade with the severity of acute pancreatitis and its predictive value

摘要:

目的 分析系统性炎症等级（SIG）与急性胰腺炎（AP）严重程度的相关性，并评估其预测价值。方法 回顾性分析2022年1月—2024年12月于广西壮族自治区人民医院住院治疗的247例AP患者的临床资料，依据2012年修订的亚特兰大分类标准，将患者分为轻症组（141例）和中重症组（106例）。收集患者入院时的血液实验室指标，计算中性粒细胞与淋巴细胞比值（NLR）和改良格拉斯哥预后评分（mGPS），并将两者联合计算出SIG。采用多因素一般Logistic回归模型分析影响AP严重程度的相关因素，并构建联合预测模型；采用受试者工作特征（ROC）曲线评估SIG评分、联合预测模型对AP严重程度的预测价值。结果 中重症组糖尿病史率、中性粒细胞计数水平、红细胞分布宽度标准差（RDW-SD）水平、C-反应蛋白水平、尿素氮水平、乳酸脱氢酶（LDH）水平、NLR评分、急性胰腺炎床旁严重程度评分≥3分占比、mGPS和SIG评分均高于轻症组，人血白蛋白水平和血清钙水平均低于轻症组。多因素一般Logistic回归分析结果显示：有糖尿病史［O^R=1.958（95% CI：1.020，3.758）］、RDW-SD升高［O^R=1.062（95% CI：1.013，1.113）］及LDH升高［O^R=1.007（95% CI：1.003，1.010）］是中重症急性胰腺炎的危险因素（P <0.05），SIG评分为1分［O^R=0.166（95% CI：0.044，0.621）］是中重症急性胰腺炎的保护因素（P <0.05）。ROC曲线分析显示，SIG、RDW-SD、LDH、急性胰腺炎床旁严重程度评分（BISAP评分）联合预测的曲线下面积（AUC）最大为0.824（95% CI：0.770，0.873），敏感性最高为84.0%（95% CI：0.758，0.897），BISAP评分的特异性最高为89.4%（95% CI：0.832，0.935）。联合预测模型的校准分析显示，Hosmer-Lemeshow检验差异无统计学意义（P >0.05），提示模型预测概率与实际发生率具有较好的一致性。联合预测模型AUC为0.789，Brier score为0.187，校准截距为-0.001，校准斜率为0.686，提示模型具有较好的内部稳定性。结论 SIG与AP严重程度独立相关，单独应用预测效能中等；结合糖尿病史、RDW-SD和LDH构建的联合预测模型可进一步提高对中重症AP的识别能力。

Abstract:

Objective To analyze the association between systemic inflammation grade (SIG) and the severity of acute pancreatitis (AP) and to evaluate its predictive value.Methods The clinical data of 247 patients with AP hospitalized at the People's Hospital of Guangxi Zhuang Autonomous Region between January 2022 and December 2024 were retrospectively analyzed. According to the 2012 revised Atlanta classification, the patients were divided into a mild AP group (n = 141) and a moderately severe/severe AP group (n = 106). Laboratory parameters on admission were collected. The neutrophil-to-lymphocyte ratio (NLR) and modified Glasgow Prognostic Score (mGPS) were calculated and combined to determine SIG. Multivariable logistic regression was used to analyze factors associated with AP severity and construct a composite prediction equation. Receiver operating characteristic (ROC) curves were used to evaluate the predictive value of SIG and the combined model for AP severity.Results The proportions of patients with a history of diabetes and the levels or scores of neutrophil count, red blood cell distribution width-standard deviation (RDW-SD), C-reactive protein, blood urea nitrogen, lactate dehydrogenase (LDH), NLR, mGPS, and SIG were higher in the moderately severe/severe AP group than in the mild AP group, whereas serum albumin and calcium levels were lower. Multivariable logistic regression showed that a history of diabetes [O^R = 1.958 (95% CI: 1.020, 3.758) ], elevated RDW-SD [O^R = 1.062 (95% CI: 1.013, 1.113) ], and elevated LDH [O^R = 1.007 (95% CI: 1.003, 1.010) ] were risk factors for moderately severe/severe AP (P < 0.05), whereas an SIG score of 1 [O^R = 0.166 (95% CI: 0.044, 0.621) ] was a protective factor (P < 0.05). The combination of SIG, RDW-SD, LDH, and the Bedside Index for Severity in Acute Pancreatitis (BISAP) had the largest area under the curve (AUC), at 0.824 (95% CI: 0.770, 0.873), and the highest sensitivity, at 83.96% (95% CI: 0.758, 0.897); BISAP had the highest specificity, at 89.36% (95% CI: 0.832, 0.935). The Hosmer-Lemeshow test for the composite prediction equation showed no significant difference (P > 0.05), indicating good agreement between predicted probabilities and observed incidence. The combined prediction model had an AUC of 0.789, a Brier score of 0.187, a calibration intercept of -0.001, and a calibration slope of 0.686, indicating good internal stability.Conclusion SIG is independently associated with AP severity and has moderate predictive performance when used alone. A composite prediction equation combining SIG with a history of diabetes, RDW-SD, and LDH may further improve the identification of moderately severe and severe AP.