Development and Validation of a Nomogram for Predicting the Severity of the First Episode of Hyperlipidemic Acute Pancreatitis.

Acute pancreatitis (AP) refers to a common inflammatory disease of the pancreas caused by aberrant activation of pancreatic enzymes characterized by local inflammation of the pancreas and subsequent systemic inflammatory response. It is a prominent digestive emergency that leads to hospitalization. 1 , 2 The common causes of AP encompass biliary diseases, hyperlipidemia, and alcohol consumption. AP caused by serum triglyceride (TG) levels ≥11.30 mmol/L or between 5.65 and 11.30 mmol/L with lipid turbidity is called hyperlipidemic acute pancreatitis (HLAP). With the improvement of living standards and changes in dietary structure, the incidence of HLAP is increasing daily, and HLAP is becoming more severe and affects people at a younger age, which requires vigilance. 3 HLAP has a high incidence of complications and poor prognosis. 4 The incidence of local complications, acute renal failure, and exacerbation in patients with HLAP is higher than that in patients with non-HLAP. 3 Therefore, it is imperative to stratify patients with HLAP based on their risk level in order to facilitate clinical decision-making and optimize treatment administration. Currently, the revised Atlanta Classification in 2012 is employed for grading the severity of AP, 5 but there exists a delay in its implementation. Additionally, there are several scoring systems for early assessment of AP severity, including the Ranson score, Acute Physiology and Chronic Health Evaluation (APACHE) II score, modified computed tomography severity index (MCTSI), and Bedside Index for Severity in Acute Pancreatitis (BISAP) score. 6–8 However, these scoring systems also exhibit specific drawbacks and limitations. Considering the aforementioned predictive models’ limitations and HLAP’s unique pathogenesis and clinical characteristics, there are currently few reports on scoring systems specifically designed for early prediction of HLAP severity. Hence, it is crucial to develop an innovative and straightforward scoring system for the early assessment of HLAP severity. The nomogram is a simple, intuitive, and practical graphical model that provides personalized risk estimation and can assess prognostic outcomes associated with various diseases. 9 , 10 This study aimed to develop and validate a nomogram that can assess the clinical severity of patients with HLAP at an early stage to provide reference values for guiding clinical treatment.

As shown in Figure 1 , 318 patients with first-episode HLAP were included in this study. The training cohort consisted of 225 patients with first-episode HLAP, including 150/225 males (66.7%) and 75/225 females (33.3%). The validation cohort comprised 93 patients with first-episode HLAP, including 57/93 males (61.3%) and 36/93 females (38.7%). There were no significant differences in the demographic characteristics, comorbidities, vital signs at admission, hospital stay, hospitalization expenses, APACHE II score, BISAP score, MCTSI, or laboratory indicators within 24 hours after admission between the training and validation cohorts (P >0.05; Table 1 ). Table 1 shows that the two cohorts were similar, justifying their use as training and validation cohorts. Table 1 General Data and Laboratory Indicators of Patients in the Training and Validation Cohorts Training Cohort (n=225) Validation Cohort (n=93) χ 2 /Z/T P value Age, years 37.52±9.84 38.91±9.90 −1.144 0.253 Sex, male/female 150/75 57/36 0.837 0.360 Pregnancy, n(%) 12(5.3) 1(1.1) 2.054 0.152 Comorbidities Hypertension, n(%) 45(20.0) 13(14.0) 1.600 0.206 Diabetes, n(%) 113(50.2) 37(39.8) 2.877 0.090 Fatty liver, n(%) 182(80.9) 73(78.5) 0.237 0.626 Vital signs Temperature (°C) 36.8(36.5,37.2) 36.7(36.5,37.2) −0.042 0.967 Pulse rate (times/minute) 98.0(84.0,113.50) 101.00(87.0,114.0) −0.817 0.414 Respiratory rate (times/minute) 20.0(19.0,21.0) 20.0(19.0,21.0) −0.232 0.816 Systolic blood pressure (mmHg) 130.0(120.0,140.0) 130.0(118.0,143.0) −0.078 0.937 Diastolic blood pressure (mmHg) 80.0(71.0,88.0) 83.0(73.0,90.5) −1.822 0.068 Mean arterial pressure (mmHg) 97.0(89.0,105.0) 98.0(89.0,108.0) −1.341 0.180 Score at admission APACHE II 5.0(2.0,8.0) 5.0(3.0,6.5) −0.406 0.685 BISAP 1.0(0,2.0) 1.0(1.0,2.0) −1.677 0.094 MCTSI 4.0(2.0,6.0) 4.0(3.0,6.0) −1.303 0.193 MAP/MSAP+SAP 91/134 37/56 0.012 0.913 Hospital stay (days) 12.0(10.0,16.0) 11.0(11.0,14.0) −1.627 0.104 Total expenses (RMB) 27,183.36 (13,724.39,46,304.45) 28,688.94 (21,459.35,46,121.00) −1.536 0.125 Inspection expenses (RMB) 5463.00 (3367.50,8351.50) 6342.00 (3807.00,8807.00) −1.613 0.107 Drug expenses(RMB) 12,325.10 (5818.67,23,826.68) 13,096.33 (8906.29,18,732.28) −0.562 0.574 Clinical index WBC(10 9 /L) 11.76(9.43,14.57) 12.41(9.84,16.05) −1.447 0.148 HB(g/L) 143.11±25.50 147.53±21.15 −1.474 0.141 PLT(10 9 /L) 220.00(175.00,279.50) 218.00(186.50,269.50) −0.246 0.806 HCT(%) 40.48±6.49 41.79±5.18 −1.732 0.084 RDW(%) 13.20(12.80,14.00) 13.30(12.75,13.95) −0.488 0.626 Blood AMY(U/L) 287.00 (158.00,521.50) 313.00 (151.50,554.00) −0.520 0.603 BG(mM) 11.32 (7.80,14.92) 10.30 (7.76,14.58) −0.658 0.510 TG(mM) 16.99(12.13,32.09) 16.28(11.93,22.02) −1.757 0.079 TBIL(μmol/L) 15.80(11.85,22.90) 13.90(10.45,20.20) −1.710 0.087 AST(U/L) 29.00 (19.00,39.25) 27.00 (19.50,37.00) −0.557 0.578 LDH(U/L) 311.00(215.75,436.25) 368.00(244.00,530.50) −1.925 0.054 ALB(g/L) 35.18±6.03 35.58±5.34 −0.559 0.576 BUN(mM) 4.20(3.20,5.65) 3.90(2.55,6.15) −1.739 0.082 Scr(μM) 68.00(58.50,78.00) 64.00(47.50,77.50) −1.772 0.076 HCO 3 − (mM) 20.90(17.35,22.95) 20.00(16.48,23.20) −0.725 0.468 Ca 2+ (mM) 2.17(2.02,2.25) 2.15(1.99,2.22) −1.687 0.092 Na + (mM) 136.07±5.65 135.05±4.89 1.524 0.129 K + (mM) 4.05±0.51 4.15±0.59 −1.556 0.121 Blood phosphorus (mM) 0.74±0.32 0.74±0.28 0.075 0.940 DDi (mg/L) 1.89(0.93,3.56) 2.25(0.93,4.28) −0.746 0.445 Abbreviations : APACHE II, Acute Physiology and Chronic Health Evaluation II; BISAP, Bedside Index for Severity in Acute Pancreatitis; MCTSI, modified computed tomography severity index; WBC, white blood cell count; HB, hemoglobin; PLT, platelet count; HCT, hematocrit; RDW, red blood cell volume distribution width; AMY, amylase; BG, blood glucose; TG, triglycerides; TBIL, total bilirubin; AST, aspartate aminotransferase; LDH, lactate dehydrogenase; ALB, albumin; BUN, blood urea nitrogen; Scr, serum creatinine; HCO 3 − , bicarbonate; Ca 2+ , serum calcium; K + , blood potassium; Na + , blood sodium; DDi, D-Dimer. Figure 1 A flow diagram of the patient recruitment procedure ( A ) training cohort; ( B ) validation cohort). Abbreviations : HLMAP, hyperlipidemic mild acute pancreatitis; HLMSAP, hyperlipidemic moderate-severe acute pancreatitis; HLSAP, hyperlipidemic severe acute pancreatitis. General Data and Laboratory Indicators of Patients in the Training and Validation Cohorts Abbreviations : APACHE II, Acute Physiology and Chronic Health Evaluation II; BISAP, Bedside Index for Severity in Acute Pancreatitis; MCTSI, modified computed tomography severity index; WBC, white blood cell count; HB, hemoglobin; PLT, platelet count; HCT, hematocrit; RDW, red blood cell volume distribution width; AMY, amylase; BG, blood glucose; TG, triglycerides; TBIL, total bilirubin; AST, aspartate aminotransferase; LDH, lactate dehydrogenase; ALB, albumin; BUN, blood urea nitrogen; Scr, serum creatinine; HCO 3 − , bicarbonate; Ca 2+ , serum calcium; K + , blood potassium; Na + , blood sodium; DDi, D-Dimer. A flow diagram of the patient recruitment procedure ( A ) training cohort; ( B ) validation cohort). Among the 225 patients in the training cohort, 91 cases (40.4%) were hyperlipidemic mild acute pancreatitis (HLMAP), 134 cases (59.6%) were hyperlipidemic moderate-severe acute pancreatitis and hyperlipidemic severe acute pancreatitis (HLMSAP +HLSAP). There were no statistically significant differences in the demographic characteristics and comorbidities between the HLMAP and HLMSAP+HLSAP groups (P >0.05; Table 2 ). However, there were statistically significant differences in the vital signs on admission (temperature, pulse rate, respiratory rate, systolic blood pressure, diastolic blood pressure, and mean arterial pressure), APACHE II score, BISAP score, MCTSI, hospital stay, and hospitalization expenses between the two groups (P <0.05; Table 2 ). In the training cohort, the comparison of laboratory indicators between the HLMAP and HLMSAP+HLSAP groups showed that there were no statistically significant differences in HB, PLT, HCT, AMY, TBIL, AST, BUN, K + , and Na + (P >0.05; Table 3 ). There were statistically significant differences in WBC, RDW, BG, TG, LDH, ALB, Scr, HCO 3 − , Ca 2+ , blood phosphorus, and DDi between the two patient groups (P <0.05; Table 3 ). Table 2 Comparison of General Data Between the HLMAP and HLMSAP+HLSAP Groups HLMAP Group (n=91) HLMASP+HLSAP Group (n=134) χ 2 /Z/T P value Age, years 38.11±9.78 37.13±9.89 0.735 0.463 Sex, male/female 64/27 86/48 0.923 0.337 Pregnancy, n(%) 2(2.1) 10(7.5) 2.024 0.155 Comorbidities Hypertension, n(%) 14(15.4) 31(23.1) 2.034 0.154 Diabetes, n(%) 43(47.3) 70(52.2) 0.539 0.463 Fatty liver, n(%) 73(80.2) 109(81.3) 0.044 0.833 Vital signs Temperature (°C) 36.6(36.5,37.0) 36.9(36.5,37.4) −2.744 0.006 Pulse rate (times/minute) 88.42±16.19 107.66±20.26 −7.894 <0.001 Respiratory rate (times/minute) 20.0(19.0,20.0) 20.0(20.0,22.0) −4.998 <0.001 Systolic blood pressure (mmHg) 128.24±15.57 133.13±19.52 −1.998 0.047 Diastolic blood pressure (mmHg) 75.0(70.0,82.0) 82.5(72.0,90.0) −3.110 0.002 Mean arterial pressure (mmHg) 93.0(87.0,100.0) 99.0(89.0,108.0) −3.099 0.002 Score at admission APACHE II 4.0(2.0,6.0) 6.0(3.0,9.0) −4.836 <0.001 BISAP 1.0(0,1.0) 2.0(1.0,2.0) −9.013 <0.001 MCTSI 2.0(2.0,4.0) 4.0(4.0,6.0) −8.711 <0.001 Hospital stay (days) 11.0(8.0,12.0) 14.0(12.0,19.0) −7.474 <0.001 Total expenses (RMB) 12,075.02 (8121.12,21,516.41) 40,363.33 (25,766.59,58,593.22) −10.369 <0.001 Inspection expenses (RMB) 3154.00 (2214.00,4737.00) 6826.25 (4814.75,11,262.50) −8.767 <0.001 Drug expenses (RMB) 5624.43 (2842.27,9744.13) 17,880.41 (11,670.95,31,889.88) −9.451 <0.001 Notes : P value for the comparison between the two groups. The boldfaced and italic P values are statistically different. Abbreviations : APACHE II, Acute Physiology and Chronic Health Evaluation II; BISAP, Bedside Index for Severity in Acute Pancreatitis; MCTSI, modified computed tomography severity index. Table 3 Comparison of Laboratory Indicators Between the HLMAP and HLMSAP+HLSAP Groups Clinical index HLMAP Group (n=91) HLMASP+HLSAP Group (n=134) Z/T P value WBC(10 9 /L) 10.70(9.12,12.83) 12.33(9.86,15.61) −3.113 0.002 HB(g/L) 145.55±22.85 141.45±27.11 1.184 0.238 PLT(10 9 /L) 224.00 (184.00,284.00) 216.50 (168.50,278.50) −0.742 0.458 HCT(%) 40.72±5.79 40.33±6.95 0.440 0.661 RDW(%) 13.00 (12.50,13.40) 13.50 (13.00,14.40) −4.779 <0.001 Blood AMY (U/L) 262.00 (153.00,477.00) 313.00 (164.75,522.00) −1.202 0.229 BG(mM) 9.14 (7.11,14.32) 12.50 (8.56,15.17) −2.505 0.012 TG (mM) 15.74 (9.75,32.28) 18.77 (13.23,31.80) −2.017 0.044 TBIL(μmol/L) 15.50 (11.80,21.60) 16.30 (12.08,24.43) −0.910 0.363 AST(U/L) 27.50 (19.00,35.75) 30.50 (20.75,42.50) −1.309 0.191 LDH(U/L) 218.00 (175.25,276.25) 407.50 (281.75,548.25) −7.814 <0.001 ALB(g/L) 37.90±5.42 33.33±5.74 5.994 <0.001 BUN(mM) 4.20(3.15,5.13) 4.60(3.20,6.03) −1.623 0.105 Scr(μM) 63.5(53.0,74.0) 70.6(62.8,86.5) −4.791 <0.001 HCO 3 − (mM) 21.50(19.30,23.00) 20.00(16.00,23.00) −2.228 0.026 Ca 2+ (mM) 2.21(2.15,2.29) 2.13(1.86,2.23) −5.148 <0.001 Na + (mM) 136.02±4.42 136.11±6.36 −0.120 0.905 K + (mM) 4.00±0.45 4.08±0.55 −1.122 0.263 Blood phosphorus(mM) 0.81±0.27 0.70±0.34 2.386 0.018 DDi (mg/L) 1.11 (0.67,2.07) 2.72 (1.39,3.94) −6.597 <0.001 Notes : P value for the comparison between the two groups. The boldfaced and italic P values are statistically different. Abbreviations : WBC, white blood cell count; HB, hemoglobin; PLT, platelet count; HCT, hematocrit; RDW, red blood cell volume distribution width; AMY, amylase; BG, blood glucose; TG, triglycerides; TBIL, total bilirubin; AST, aspartate aminotransferase; LDH, lactate dehydrogenase; ALB, albumin; BUN, blood urea nitrogen; Scr, serum creatinine; HCO 3 − , bicarbonate; Ca 2+ , serum calcium; K + , blood potassium; Na + , blood sodium; DDi, D-Dimer. Comparison of General Data Between the HLMAP and HLMSAP+HLSAP Groups Notes : P value for the comparison between the two groups. The boldfaced and italic P values are statistically different. Abbreviations : APACHE II, Acute Physiology and Chronic Health Evaluation II; BISAP, Bedside Index for Severity in Acute Pancreatitis; MCTSI, modified computed tomography severity index. Comparison of Laboratory Indicators Between the HLMAP and HLMSAP+HLSAP Groups Notes : P value for the comparison between the two groups. The boldfaced and italic P values are statistically different. Abbreviations : WBC, white blood cell count; HB, hemoglobin; PLT, platelet count; HCT, hematocrit; RDW, red blood cell volume distribution width; AMY, amylase; BG, blood glucose; TG, triglycerides; TBIL, total bilirubin; AST, aspartate aminotransferase; LDH, lactate dehydrogenase; ALB, albumin; BUN, blood urea nitrogen; Scr, serum creatinine; HCO 3 − , bicarbonate; Ca 2+ , serum calcium; K + , blood potassium; Na + , blood sodium; DDi, D-Dimer. Vital signs on admission (temperature, pulse rate, respiratory rate, systolic blood pressure, diastolic blood pressure, and mean arterial pressure), WBC, RDW, BG, TG, LDH, ALB, Scr, HCO 3 − , Ca 2+ , blood phosphorus, and DDi were identified as candidate predictors of MASP or SAP in the univariate analysis (P <0.05). LASSO regression was used to screen the variables with non-zero coefficients further among the results of the univariate logistic regression. The best predictors of MASP and SAP were WBC, RDW, BG, LDH, ALB, Scr, HCO 3 − , Ca 2+ , and DDi ( Figure 2 ). These variables were included in the multivariate logistic regression analysis to construct a regression model. The Hosmer–Lemeshow test yielded a P value of 0.639, indicating that the model was valid. The final results of multivariate analysis showed that the WBC (odds ratio [OR], 1.159; 95% confidence interval [CI], 1.036–1.297; P = 0.010), LDH (OR, 1.008; 95% CI, 1.003–1.012; P <0.001), Scr (OR, 1.029; 95% CI, 1.002–1.056; P=0.034), and DDi (OR, 1.359; 95% CI, 1.005–1.838; P=0.046) were independent risk factors for the onset of MSAP and SAP. ALB (OR, 0.898; 95% CI, 0.825–0.977; P=0.012) and Ca 2+ (OR, 0.067; 95% CI, 0.010–0.794; P=0.031) were independent protective factors for MSAP and SAP attack ( Table 4 ). Table 4 Logistic Regression Analysis of Independent Predictors in HLMSAP and HLSAP β Standard Error Wald P OR 95% CI Lower limit Upper limit WBC 0.148 0.057 6.634 0.010 1.159 1.036 1.297 LDH 0.008 0.002 12.413 <0.001 1.008 1.003 1.012 ALB −0.108 0.043 6.294 0.012 0.898 0.825 0.977 Scr 0.028 0.013 4.483 0.034 1.029 1.002 1.056 Ca 2+ −2.679 1.252 4.573 0.031 0.067 0.010 0.794 DDi 0.307 0.154 3.966 0.046 1.359 1.005 1.838 Abbreviations : WBC, white blood cell count; LDH, lactate dehydrogenase; ALB, albumin; Scr, serum creatinine; Ca 2+ , serum calcium; DDi, D-Dimer; β, regression coefficient; OR, odds ratio; CI, confidence interval. Figure 2 Selection of predictive factors using the least absolute shrinkage and selection operator logistic regression algorithm. ( A ) Least absolute shrinkage and selection operator (LASSO) coefficient profiles of the 17 candidate variables. ( B ) The best value was determined by the two dashed vertical lines drawn according to the minimum mean-square error criterion (left dashed line) and the standard error criterion (right dashed line). In the present study, nine predictors were selected according to the standard error criterion (λ=0.041). Logistic Regression Analysis of Independent Predictors in HLMSAP and HLSAP Abbreviations : WBC, white blood cell count; LDH, lactate dehydrogenase; ALB, albumin; Scr, serum creatinine; Ca 2+ , serum calcium; DDi, D-Dimer; β, regression coefficient; OR, odds ratio; CI, confidence interval. Selection of predictive factors using the least absolute shrinkage and selection operator logistic regression algorithm. ( A ) Least absolute shrinkage and selection operator (LASSO) coefficient profiles of the 17 candidate variables. ( B ) The best value was determined by the two dashed vertical lines drawn according to the minimum mean-square error criterion (left dashed line) and the standard error criterion (right dashed line). In the present study, nine predictors were selected according to the standard error criterion (λ=0.041). According to multivariate regression results, the WBC, LDH, ALB, Scr, Ca 2+ , and DDi were used to construct a MSAP+SAP prediction model; R software was used to visualize the model and obtain a nomogram. A total score reaching approximately 35 points indicates that HLAP could progress to MSAP or SAP. A total score close to 55 points indicates that the risk of progression to MSAP or SAP is as high as 90% ( Figure 3 ). Figure 3 A nomogram for the severity of the first episode of HLAP. Abbreviations : WBC, white blood cell count; LDH, lactate dehydrogenase; ALB, albumin; Scr, serum creatinine; Ca 2+ , serum calcium; DDi, D-Dimer. A nomogram for the severity of the first episode of HLAP. The C-index value was used to evaluate the degree of discrimination, and R software was employed to sample 1000 times through the bootstrap method to obtain a new sample dataset. The C-index value was 0.908 (95% CI, 0.867–0.948) in the training cohort and 0.950 (95% CI, 0.910–0.990) in the validation cohort. The verification results of the two cohorts were similar, with both values exceeding 0.9, indicating that the nomogram derived from this study could effectively discriminate between MAP and MSAP or SAP. The HL goodness-of-fit test (χ 2 =7.015, P=0.535) demonstrated a favorable fit for this model; a calibration curve was used to evaluate the calibration of the nomogram model. As shown in Figure 4 , “apparent” represents the original curve, “ideal” represents the ideal standard curve, and “bias-corrected” represents the calibration curve. The Brier score of the training cohort was 0.128, and that of the validation cohort was 0.109, indicating that the nomogram used in this study could predict MSAP or SAP ( Figure 4 ). Figure 4 Calibration curve for predicting the first episode of HLAP ( A) training cohort; ( B ) validation cohort. Calibration curve for predicting the first episode of HLAP ( A) training cohort; ( B ) validation cohort. The area under the curve (AUC) value was calculated through the R software pROC package to compare the accuracy of the nomogram, MCTSI, APACHE II score, and BISAP score predicting MSAP or SAP. In the training cohort, the AUC of the nomogram, MCTSI, APACHE II, and BISAP were 0.908 (95% CI, 0.867–0.948), 0.821 (95% CI, 0.770–0.873), 0.689 (95% CI, 0.622–0.756), and 0.833 (95% CI, 0.786–0.880), respectively ( Figure 5A ). In the validation cohort, the AUC of the nomogram, MCTSI, APACHE II, and BISAP were 0.950 (95% CI, 0.910–0.990), 0.811 (95% CI, 0.734–0.887), 0.715 (95% CI, 0.612–0.819), and 0.842 (95% CI, 0.768–0.916), respectively ( Figure 5B ). These results indicated that the developed nomogram could accurately predict the HLAP severity and it had a robust performance. Figure 5 Receiver operating characteristic curve for predicting the first episode of HLAP ( A ) training cohort; ( B ) validation cohort. Abbreviations : MCTSI, modified computed tomography severity index; APACHE II, Acute Physiology and Chronic Health Evaluation II; BISAP, Bedside Index for Severity in Acute Pancreatitis. Receiver operating characteristic curve for predicting the first episode of HLAP ( A ) training cohort; ( B ) validation cohort. The “rmda” package in the R software was utilized for generating the decision curve. The four distinct DCA curves correspond to the four distinct clinical diagnostic models. The DCA curve shows that the developed nomogram has a greater net benefit than the other system scores in predicting the severity of disease in patients with HLAP; this demonstrates its utility in clinical decision-making ( Figure 6 ). Figure 6 Decision curve of nomogram and the scoring system for predicting the first episode of HLAP ( A ) training cohort; ( B ) validation cohort). Abbreviations : MCTSI, modified computed tomography severity index; APACHE II, Acute Physiology and Chronic Health Evaluation II; BISAP, Bedside Index for Severity in Acute Pancreatitis. Decision curve of nomogram and the scoring system for predicting the first episode of HLAP ( A ) training cohort; ( B ) validation cohort).

The clinical data of patients with first-episode HLAP, who were admitted to Fujian Medical University Union Hospital between June 2012 and June 2023, as well as the First Affiliated Hospital of Fujian Medical University between October 2015 and October 2022, were retrospectively analyzed. The inclusion criteria were: (1) patients who met the diagnostic criteria for AP with any two of the following three conditions: 5 (i) acute, persistent upper abdominal pain, (ii) serum amylase or lipase level >3 times the upper normal limit, (iii) typical imaging changes of AP; and (2) having blood TG level ≥11.30 mmol/L before treatment or serum blood TG level of 5.65–11.30 mmol/L with lipid turbidity. The exclusion criteria were: (1) AP caused by other factors (gallstones/microlithiasis, alcohol, neoplasia, ischemia, Oddi sphincter dysfunction, drug-induced, and bacterial or viral infections, etc); (2) patients with incomplete clinical data; (3) patients with chronic pancreatitis and recurrent pancreatitis; (4) more than 72 h from onset to admission; and (5) patients referred from other medical institutions. As a result, a total of 225 patients from Fujian Medical University Union Hospital were enrolled in the study as a training cohort, and 93 patients from the First Affiliated Hospital of Fujian Medical University were included in the study as a validation cohort. In each cohort, the patients were divided either into the mild acute pancreatitis (MAP) group or moderate-severe acute pancreatitis and severe acute pancreatitis (MSAP+SAP) group according to the revised Atlanta Classification in 2012. 5 This classification defines MAP as no organ failure and no local or systemic complications, MSAP as transient organ failure (spontaneous recovery within 48 hours) with local or systemic complications, and SAP as persistent organ failure (more than 48 hours). Data were obtained from the electronic medical records of patients with HLAP, including demographic characteristics (sex, age, and pregnancy), comorbidities (hypertension, diabetes, and fatty liver), vital signs on admission (temperature, pulse rate, respiratory rate, systolic blood pressure, diastolic blood pressure, and mean arterial pressure), hospital stay, and hospitalization expenses (total, inspection, and drug expenses). Laboratory indicators including white blood cell count (WBC), hemoglobin (HB), platelet count (PLT), hematocrit (HCT), red blood cell volume distribution width (RDW), blood amylase (AMY), blood glucose (BG), triglycerides (TG), total bilirubin (TBIL), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), albumin (ALB), blood urea nitrogen (BUN), serum creatinine (Scr), bicarbonate (HCO 3 − ), serum calcium (Ca 2+ ), blood potassium (K + ), blood sodium (Na + ), blood phosphorus, and D-Dimer (DDi), were assayed within 24 hours after admission. In addition, the APACHE II, BISAP, and MCTSI scores of patients with HLAP were calculated within 24 hours after admission. The count data expressed as the number of cases or percentages were analyzed using χ 2 test and Fisher’s exact probability method when necessary. The measurement data that follow normal distribution are represented as mean and standard deviation (mean±SD); Student’s t -test represents the comparison between groups. The measurement data that do not follow normal distribution are represented as median and interquartile range (median [P25–P75]); the differences of the groups were compared using the Mann–Whitney U -test. Univariate analysis and least absolute shrinkage and selection operator (LASSO) regression methods were used to screen the best predictors in the training cohort. The screened predictors were included in the multivariate logistic regression analysis. The odds ratios (OR) and 95% confidence intervals (CI) of these factors were analyzed using multivariate logistic regression to determine the predictors of progression to MSAP or SAP in the patients with HLAP. Differences were considered statistically significant at a P-value <0.05. The multivariable logistic regression analysis results in the training cohort were presented in a nomogram. For internal validation, 1000 bootstrap resamples were used to calculate the Harrell consistency index (C-index). Then, the performance of the nomogram was tested in the validation cohort for external verification. The nomogram was constructed based on the multivariate logistic regression analysis, assigning points to each value level of each risk factor and then adding each score to obtain the total score. Ultimately, the predicted probability of the outcome event was calculated by applying a conversion function that establishes a quantitative relationship between the total score and the likelihood of occurrence. We assessed the predictive power of the nomogram using a receiver operating characteristic (ROC) curve analysis. Calibration curves were employed to analyze the difference between the nomogram and the ideal model. Decision curve analysis (DCA) was utilized to appraise the clinical utility of the nomogram. The R software packages “rms”, “pROC”, and “rmda” were used to generate the nomogram, ROC curves, and calibration curves and perform DCA. R software (version 4.3.2) and SPSS software (version 24.0) were used for statistical analysis.

AP is a common inflammatory disease of the pancreas. Hypertriglyceridemia (HTG) has emerged as a prominent etiological factor of AP in young individuals. 11 HLAP is characterized by a unique pathogenesis, rapid progression, and a propensity to worsen in severity, 3 which poses a significant challenge for clinicians in terms of its clinical management. Early diagnosis and assessment of HLAP severity have assumed paramount importance. Clinical scoring systems such as Ranson, BISAP, APACHE II, and MCTSI scores, have been used to predict the severity of the disease in patients with AP. 8 The Ranson score is calculated based on data from admission and within 48 h after admission and cannot be evaluated within 24 h, resulting in a relatively poor timeliness. 12 The BISAP score incorporates five easily-obtainable predictive factors; but its sensitivity and positive predictive values were 70% and 40%, respectively. 13 APACHE II is designed to predict the severity and mortality of patients with AP admitted to the ICU with a large set of mandatory variables. Although the AUC of APACHE II in prediction of SAP is 0.820, it is not specific. 14 Imaging data obtained by computed tomography (CT) is one of the criteria for AP diagnosis. However, CT may underestimate or incorrectly classify AP severity if obtained less than 72 h after the symptoms’ onset. 15 A specific scoring system for HLAP in clinical practice is currently not available. Therefore, there is an urgent need to explore evaluation methods with good sensitivity and specificity that are easily implemented in clinical practice. We selected six independent predictors through multivariate analysis to construct a nomogram. WBC, LDH, ALB, Scr, Ca 2+ , and DDi were independent predictors for the tendency of HLAP to become severe. HTG is the main characteristic feature of HLAP; however, a meta-analysis including 11,965 patients from 16 eligible studies revealed no significant difference in the severity of AP based on the extent of HTG. 16 The WBC count of the patients with HLMSAP and HLSAP was significantly higher than that of the patients with HLMAP. The elevation in WBC count may be attributed to AP-related inflammation or pancreatic infection. Mayer et al concluded that the WBC count provides a good distinction between MAP and SAP. They also confirmed that this difference was evident on the day of admission in the patients with AP. 17 Compared with AP due to other etiological factors, HLAP is more likely to develop into severe systemic inflammatory response syndrome and has a poor prognosis. 18 LDH, as a glycolytic enzyme, is widely distributed in the cytoplasm of tissues, mainly in the myocardium, skeletal muscles, kidneys, and liver. 19 Patients with MASP and SAP are more likely to have cardiac, pulmonary, or renal dysfunction; thus, they have elevated LDH levels. The organ specificity of LDH is poor, but it can indicate the extent of the pancreatitis-caused damage to other organs. 20 At present, LDH has been used by the Ranson, Glasgow, and Japanese Severity Scores to predict the severity of early AP. 8 The level of LDH can serve as a simple and valuable parameter for predicting AP associated with organ failure and pancreatic necrosis. 21 , 22 ALB is a natural plasma protein that is exclusively synthesized by the liver but can be catabolized in most organs. During the development of AP, trypsin and elastase damage vascular endothelial cells, resulting in increased vascular permeability and subsequent penetration of ALB into the tissue space. 23 In addition, the decrease in the ALB levels in patients with MSAP and SAP may be attributed to the liver’s decreased capacity to biosynthesize ALB due to reduced food intake and stimulation of inflammatory factors. 23 During the development of HLAP, pancreatic lipase hydrolyzes high levels of TG in the pancreas and surrounding areas, resulting in a substantial release of free fatty acids that surpasses the binding capacity of ALB. As the ALB levels decrease, further damage to the pancreatic acinar cells and small blood vessels occurs. Thus, serum ALB may act as a serum biomarker for assessing the severity of AP. 24 , 25 The incidences of organ failure and pancreatic necrosis in patients with AP complicated by hypoalbuminemia were significantly higher than those in patients without hypoalbuminemia. 26 , 27 Thus, ALB levels were negatively correlated with the severity of AP. 28 Scr is a critical indicator of kidney function. During AP development, vascular permeability increases with the release of inflammatory mediators, resulting in a significant accumulation of body fluid in the interstitial space. This leads to decrease in renal perfusion and circulating blood volume, ultimately causing an elevation of Scr level and other metabolites. 29 During the development of AP, a plethora of inflammatory factors are released, such as TNF-α, which directly acts on the glomeruli and capillaries, leading to ischemia and tubular necrosis. Cytokines, such as IL-1β, IL-8, and IL-6, exert their effects on endothelial cells, leading to renal ischemia and thrombosis by releasing oxygen-free radicals. 30 Due to the unique pathogenesis pattern of HLAP, the hypercoagulable state induced by HTG gives rise to microcirculation disorders and thrombosis, which in turn leads to renal ischemia and hypoxic damage. 31 Elevated Scr, an indicator of acute kidney, is associated with pancreatic necrosis, organ failure, and mortality in patients with AP. 32 , 33 A large multicenter study conducted in China indicated that the incidence of acute renal failure in patients with HLAP is higher than that in patients with non-HLAP. 3 Early changes in Scr levels, especially in the first 24 hours after admission, can serve as a predictive indicator for the severity of AP, 34 which is consistent with our research findings. Studies have demonstrated that serum Ca 2+ levels can determine the exocrine functions of the pancreas and pathological progression of AP. 35 , 36 AP is usually accompanied by a decrease in serum Ca 2+ levels. Ca 2+ has been proven to be an independent predictor of the development of AP complicated by organ failure. 37 Ca 2+ is negatively correlated with the severity of AP, 38 , 39 which is line with our findings. The hypocalcemia induce by AP may be attributed to the autodigestion of mesenteric fat by pancreatic enzymes, which in turn leads to the release of free fatty acids that subsequently form calcium salts. Furthermore, catecholamines are involved in mediating the translocation of serum Ca 2+ into tissues. 40 Additionally, HLAP’s special pathogenesis involves the hydrolysis of high concentrations of TG in the pancreas and its surroundings by pancreatic lipase, which locally produces large amounts of free fatty acids and forms calcium salts through complexing with Ca 2+ . 41 DDi is a marker of coagulation and fibrinolysis. Elevated levels of DDi suggest a possible hypercoagulable state in the blood. During AP, the HTG-induced hypercoagulable state of blood leads to microcirculation disorders and thrombosis, further leading to ischemia and hypoxia-induced damage. 31 Our previous studies showed that HLAP patients with acute renal failure have significantly higher DDi levels than those without acute renal failure. 42 Previous studies have shown that patients with AP and elevated DDi levels are more likely to develop pancreatic necrosis and organ failure than those with normal DDi levels, indicating a positive correlation between DDi levels and the severity of AP. 43 Changes in the coagulation system are closely associated with AP complications. 44 Therefore, WBC count, LDH, ALB, Scr, Ca 2+ , and DDi are all important indicators that can effectively predict the trend of HLAP towards moderate-severe or severe exacerbation. Nomograms can be readily employed to evaluate the odds of a given clinical outcome in an individual patient. Thus, they are increasingly and frequently used as prognostic tools in clinical decision-making. 9 , 45 We endeavored to develop a nomogram capable of promptly evaluating the progression of patients with HLAP towards moderate-severe or severe exacerbation at an early stage. This nomogram consists of six variables that can be readily measured within 24 h after admission and can predict the risk of progression to MSAP or SAP in patients with HLAP for early intervention and treatment, improving the prognosis of patients with HLAP. The ROC curve of this model was further plotted and revealed an AUC value of 0.908, indicating excellent prediction accuracy and recognition performance. The AUC of this prediction model had obvious advantages compared with that of APACHE II, BISAP, and MCTSI. The calibration curve of this model revealed a strong concordance between the predicted and standard curves. The DCA curve was constructed to evaluate the net benefit of this prediction model, demonstrating its favorable applicability in clinical practice. The nomogram was validated in another provincial tertiary grade A hospital, showing good predictive value and clinical applicability. The overall findings of our study demonstrate that our nomogram exhibits superior predictive reliability, accuracy and optimal net benefit when compared to other clinical scoring systems, such as APACHE II, BISAP, and MCTSI. This study’s nomogram can be further developed into a web calculator or APP in the future. After the user inputs the value of the predictor variable, the risk probability can be calculated based on the prediction model, thereby streamlining the assessment process. There are some limitations in this study. First, this study is a retrospective design, and selection bias is inevitable Therefore, prospective studies are needed, and sample sizes are further expanded to enhance the level of evidence. Second, this nomogram was established based on the clinical data of patients with a first episode of HLAP, and further research is required to determine its applicability to recurrent HLAP. Finally, the lack of validation of this study across diverse populations may limit the extrapolation and generalization of this nomogram to other populations worldwide.

The WBC count, LDH, ALB, Scr, Ca 2+ , and DDi are crucial indicators that can predict the trend of HLAP towards moderate-severe or severe exacerbation. Moreover, the established nomogram prediction model exhibited excellent differentiation, calibration, and clinical applicability, which holds significant implications for early evaluation, timely treatment, personalized management, and prognosis improvement of HLAP.