Abstract:

Objective　To construct a model to predict the risk of severe hypertriglyceridemic acute pancreatitis (sHTGAP) by the LASSO regression model. Methods　The clinical data of 315 patients with hypertriglyceridemic acute pancreatitis (HTGAP) treated in Hanzhong Central Hospital between January 2019 and March 2022 were retrospectively analyzed. The patients were divided into an sHTGAP group (94 cases), including 59 males and 35 females who were (37.51±7.08) years old, and a non-sHTGAP group (221 cases), including 135 males and 86 females who were (41.87±6.28) years old, according to the disease severity. The clinical data and laboratory indicators were compared between the two groups. t test, rank sum test, and χ² test were applied. The LASSO regression model was used to screen the characteristic factors and construct the risk score model. Results　There were no statistical differences in gender, diabetes history, hypertension, smoking history, drinking history, and diet between the two groups (all P>0.05). The scores of Acute Physiology and Chronic Health Evaluation Ⅱ (APACHE Ⅱ) and Modified CT Severity Index (MCTSI), white blood cell count (WBC), C-reactive protein (CRP), glucose (Glu), and triglycerides (TG) in the sHTGAP group were higher than those in the non-sHTGAP group [9.00 (7.00, 11.00) vs. 4.00 (3.00, 6.00), 6.00 (5.00, 7.00) vs. 4.00 (3.00, 6.00), 12.22 (9.89, 14.34)×10⁹/L vs. 10.44 (8.58, 12.39)×10⁹/L, 164.25 (128.32, 217.58) mg/L vs. 56.15 (37.13, 76.95) mg/L, (12.16±2.99) mmol vs. (6.27±2.42) mmol, and (14.05±5.10) mmol/L vs. (11.46±4.78) mmol/L]; the hospitalization time in the sHTGAP group was shorter than that in the non-sHTGAP group [13.00 (8.00, 16.00) d vs. 9.00 (5.00, 13.00) d]; the calcium ion (Ca²⁺) in the sHTGAP group were lower than that in the non-sHTGAP group [(2.12±0.09) mmol/L vs. (2.17±0.09) mmol/L]; there were statistical differences (all P<0.05). The area under the curve (AUC) of the risk score model constructed by the LASSO regression was 0.998; it was better than any single indicator. Conclusions　APACHE Ⅱ score, MCTSI score, hospitalization time, WBC, and levels of CRP, Glu, Ca²⁺, and TG are closely related to the occurrence and severity of sHTGAP. Comprehensive evaluation of these indicators is of great significance to improve the prognosis of patients with sHTGAP.

Key words:

Acute pancreatitis, Hypertriglyceridemia, Severity, LASSO regression, Risk score

摘要：

目的　通过LASSO回归模型构建预测重症高甘油三酯血症性急性胰腺炎（sHTGAP）风险模型。方法　回顾性分析，选取2019年1月至2022年3月在汉中市中心医院治疗的315例高甘油三酯血症性急性胰腺炎（HTGAP）患者作为研究对象，根据疾病严重程度分为sHTGAP组（94例）和Non-sHTGAP组（221例）。sHTGAP组男59例、女35例，年龄（37.51±7.08）岁；Non-sHTGAP组男135例、女86例，年龄（41.87±6.28）岁。对比两组患者临床资料和实验室指标，统计学方法采用t检验、秩和检验、χ²检验。采用LASSO回归模型筛选特征因素并构建风险评分模型。结果　两组患者性别、糖尿病史、高血压、吸烟史、酗酒史、饮食习惯比较，差异均无统计学意义（均P>0.05）。sHTGAP组急性生理学与慢性健康评估Ⅱ（APACHE Ⅱ）评分、多层螺旋CT严重指数（MCTSI）评分、白细胞计数（WBC）、C反应蛋白（CRP）、血糖（Glu）、甘油三酯（TG）均高于Non-sHTGAP组［9.00（7.00，11.00）分比4.00（3.00，6.00）分、6.00（5.00，7.00）分比4.00（3.00，6.00）分、12.22（9.89，14.34）10⁹/个比10.44（8.58，12.39）10⁹/个、164.25（128.32，217.58）mg/L比56.15（37.13，76.95）mg/L、（12.16±2.99）mmol比（6.27±2.42）mmol、（14.05±5.10）mmol/L比（11.46±4.78）mmol/L］，住院时间长于Non-sHTGAP组［13.00（8.00，16.00）d比9.00（5.00，13.00）d］，钙离子（Ca²⁺）低于Non-sHTGAP组［（2.12±0.09）mmol/L比（2.17±0.09）mmol/L］，差异均有统计学意义（均P<0.05）。LASSO风险评分模型的曲线下面积（AUC）为0.998，优于单一指标。结论　APACHE Ⅱ评分、MCTSI评分、住院时间、WBC、CRP、Glu、Ca²⁺和TG水平与sHTGAP的发生和严重程度密切相关，综合评估这些指标对于提高sHTGAP患者预后意义重大。

关键词:

急性胰腺炎, 高甘油三酯血症, 严重程度, LASSO回归, 风险评分