Clinical Features And Establishment Of An Early Prediction Model For Severity Of Acute Pancreatitis

Background and purpose:Acute pancreatitis（AP）is a sudden inflammatory response of the pancreas,which is related to various factors leading to pancreatic self-digestion.Clinically,biliary acute pancreatitis（BAP）,hyperlipidemic acute pancreatitis（HLAP）,and alcoholic acute pancreatitis（AAP）are common and have relatively clear causes.Different etiologies of AP have different pathophysiological characteristics,and understanding the clinical features of AP of different etiologies is of great value for early etiological diagnosis.The 2012 revision of the Atlanta classification criteria classifies AP into mild acute pancreatitis（MAP）,moderate severe acute pancreatitis（MSAP）,and severe acute pancreatitis（SAP）.The overall mortality rate of AP is about 1%,but the mortality rate of SAP patients may be as high as 30～40%.Stratification of severity using certain indicators at or shortly after admission and early diagnosis of etiology can guide the development of optimal treatment which can reduce morbidity and mortality.The purpose of this study is to elucidate the clinical characteristics of different etiologies of AP.Finally,exploring the effectiveness of some routine laboratory indicators as potential biomarkers for early assessment of AP severity.Methods:A retrospective analysis was performed for the data of 596 AP patients hospitalized in the Second Affiliated Hospital of Nanchang University from June 2020to June 2022.1.According to the different etiologies of the study subjects,they are divided into BAP group,HLAP group,AAP group,and other etiological AP group,1)Analyzing the composition ratio of AP etiology,2)The sex,age,diabetes history,fatty liver history,complications（including AKI,ARDS,SIRS）,disease severity,and other indicators were compared among the above groups.2.According to the Atlanta classification criteria for acute pancreatitis,the study subjects were divided into MAP group（n=498）and more severe AP group（including MSAP group and SAP group）（n=98）.Laboratory results from the patient’s earliest venous blood samples are collected,such as WBC,Hb,PLT,N,L,HCT,RDW,TC,TG,HDL,LDL,Alb,Ca²⁺,GLU,BUN,Cr,Cr/Alb,NLR,and TG/HDL were calculated.Firstly,through logistic regression analysis,the predictors independently correlated with the occurrence of more severe AP were obtained,and the correlation between the predictors and the BISAP score was further analyzed.Then,the predictive value of the predictors was evaluated by the AUC value,cut-off value,sensitivity,and specificity obtained by ROC curve analysis.Finally,the predictive model was established through logistic regression analysis and its efficiency was evaluated.Results:1.A total of 596 AP patients were included,including 188 cases of BAP（31.54%）,184 cases of HLAP（30.87%）,38 cases of AAP（6.38%）,and 186 cases of other etiological AP（31.21%）.2.General data between different etiologies of AP were analyzed,and we found that compared with BAP,HLAP,AAP and other etiologic APs were more common in men,with a lower average age of onset,and a higher proportion of fatty liver（P<0.05）.The proportion of HLAP with diabetes was higher than that in BAP,AAP,and other etiological AP（P<0.05）.Among the BAP,HLAP,and AAP,the proportion of SIRS occurring in HLAP was the highest（P<0.05）,but there was no difference in severity and incidence of AKI and ARDS（P>0.05）.3.Through univariate and logistic regression analysis of laboratory indicators,we found that high levels of GLU,BUN,Cr,Cr/Alb,and low levels of Ca²⁺were independent risk factors for more severe AP.The predictive efficacy of GLU,BUN,Cr,Cr/Alb,and Ca²⁺on AP condition was evaluated by using the receiver operating characteristic（ROC）curve,the area under the ROC curve（AUC）was 0.677,0.674,0.642,0.659 and 0.674,respectively.4.Spearman correlation analysis showed that GLU,BUN,Cr,and Cr/Alb levels were positively correlated with BISAP scores（r=0.261,0.301,0.191,0.268,P<0.05）,and Ca²⁺levels were negatively correlated with BISAP scores（r=-0.265,P<0.05）.5.Finally,Ca²⁺,GLU,BUN,and Cr/Alb were selected to establish the prediction model,and the specific formula was as follows:Logit（P）=-1.824*Ca²⁺（mmol/L）+0.124*GLU（mmol/L）+0.141*BUN（mmol/L）+0.412*Cr/Alb.By drawing the ROC curve,we found that the AUC of the prediction model was 0.755,which had a good prediction value,and the HL test showed that the Goodness of fit of the prediction model was good.Conclusion:1.Biliary disease and hyperlipidemia were the top two causes of AP in our hospital.The clinical features of AP vary between different etiologies.HLAP and AAP are more common in young people and men,and most of them have disorders of glucose and lipid metabolism.Among the BAP,HLAP,and AAP groups,the proportion of SIRS in HLAP was the highest,while there was no difference in the severity and incidence of AKI and ARDS between the groups.2.Serum Ca²⁺,GLU,BUN,Cr,and Cr/Alb levels were correlated with BISAP scores,and they were also independent influencing factors for more severe AP.3.The prediction model based on Ca²⁺,GLU,BUN,and Cr/Alb can improve the prediction performance of more severe AP.