Article Search
검색
검색 팝업 닫기

Metrics

Help

  • 1. Aims and Scope

    Gut and Liver is an international journal of gastroenterology, focusing on the gastrointestinal tract, liver, biliary tree, pancreas, motility, and neurogastroenterology. Gut atnd Liver delivers up-to-date, authoritative papers on both clinical and research-based topics in gastroenterology. The Journal publishes original articles, case reports, brief communications, letters to the editor and invited review articles in the field of gastroenterology. The Journal is operated by internationally renowned editorial boards and designed to provide a global opportunity to promote academic developments in the field of gastroenterology and hepatology. +MORE

  • 2. Editorial Board

    Editor-in-Chief + MORE

    Editor-in-Chief
    Yong Chan Lee Professor of Medicine
    Director, Gastrointestinal Research Laboratory
    Veterans Affairs Medical Center, Univ. California San Francisco
    San Francisco, USA

    Deputy Editor

    Deputy Editor
    Jong Pil Im Seoul National University College of Medicine, Seoul, Korea
    Robert S. Bresalier University of Texas M. D. Anderson Cancer Center, Houston, USA
    Steven H. Itzkowitz Mount Sinai Medical Center, NY, USA
  • 3. Editorial Office
  • 4. Articles
  • 5. Instructions for Authors
  • 6. File Download (PDF version)
  • 7. Ethical Standards
  • 8. Peer Review

    All papers submitted to Gut and Liver are reviewed by the editorial team before being sent out for an external peer review to rule out papers that have low priority, insufficient originality, scientific flaws, or the absence of a message of importance to the readers of the Journal. A decision about these papers will usually be made within two or three weeks.
    The remaining articles are usually sent to two reviewers. It would be very helpful if you could suggest a selection of reviewers and include their contact details. We may not always use the reviewers you recommend, but suggesting reviewers will make our reviewer database much richer; in the end, everyone will benefit. We reserve the right to return manuscripts in which no reviewers are suggested.

    The final responsibility for the decision to accept or reject lies with the editors. In many cases, papers may be rejected despite favorable reviews because of editorial policy or a lack of space. The editor retains the right to determine publication priorities, the style of the paper, and to request, if necessary, that the material submitted be shortened for publication.

Search

Search

Year

to

Article Type

Online first

Split Viewer

Online first

Development of a Nomogram to Predict the Risk for Acute Necrotizing Pancreatitis

Junchao Zhang , Xiaxia Weng

Department of Gastroenterology, Xiamen Hospital of Traditional Chinese Medicine, Xiamen, China

Correspondence to: Xiaxia Weng
ORCID https://orcid.org/0009-0002-6310-0642
E-mail 302561422@qq.com

Received: October 4, 2023; Revised: November 15, 2023; Accepted: November 21, 2023

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Gut Liver.

Published online February 22, 2024

Copyright © Gut and Liver.

Background/Aims: Necrotizing pancreatitis (NP) presents a more severe clinical trajectory and increased mortality compared to edematous pancreatitis. Prompt identification of NP is vital for patient prognosis. A risk prediction model for NP among Chinese patients has been developed and validated to aid in early detection.
Methods: A retrospective analysis was performed on 218 patients with acute pancreatitis (AP) to examine the association of various clinical variables with NP. The least absolute shrinkage and selection operator (LASSO) regression was utilized to refine variables and select predictors. Subsequently, a multivariate logistic regression was employed to construct a predictive nomogram. The model's accuracy was validated using bootstrap resampling (n=500) and its calibration assessed via a calibration curve. The model's clinical utility was evaluated through decision curve analysis.
Results: Of the 28 potential predictors analyzed in 218 AP patients, the incidence of NP was 25.2%. LASSO regression identified 14 variables, with procalcitonin, triglyceride, white blood cell count at 48 hours post-admission, calcium at 48 hours post-admission, and hematocrit at 48 hours post-admission emerging as independent risk factors for NP. The resulting nomogram accurately predicted NP risk with an area under the curve of 0.822, sensitivity of 82.8%, and specificity of 76.4%. The bootstrap-validated area under the curve remained at 0.822 (95% confidence interval, 0.737 to 0.892). This model exhibited excellent calibration and demonstrated greater predictive efficacy and clinical utility for NP than APACHE II, Ranson, and BISAP.
Conclusions: We have developed a prediction nomogram of NP that is of great value in guiding clinical decision.

Keywords: Pancreatitis, Acute necrotizing, Nomograms, Bootstrap, Predict

Acute pancreatitis (AP) is an acute disorder of the exocrine pancreas, which is commonly linked to acinar cell injury and local/systemic inflammatory responses.1 While most cases of AP are mild and self-limiting, approximately 20% of patients progress to necrotizing pancreatitis (NP), a severe form of the disease associated with a mortality risk of up to 15%–20%.2 Approximately one-third of patients with NP may develop pancreatic infection during the natural course of AP, thus the relative risk of death is up 2-fold.3 Therefore, early identification of patients with NP is vital to assess disease severity and determine optimal treatment.

Contrast-enhanced computed tomography (CECT) is the primary diagnostic modality for identifying NP. However, its effectiveness is often limited due to NP typically becoming discernible on imaging only 3 days after the onset of the disease; Moreover, it is not accessible in all cases.4 Therefore, there is a need to develop other methods to supplement acute necrotizing pancreatitis assessment.

Several scoring systems have been developed to predict the severity of pancreatitis.5 However, each system has its unique shortcomings. The Balthazar computer tomography (CT) severity index and bedside index for severity in acute pancreatitis (BISAP) score are restricted by the availability of CT. The application of other scoring systems without CT are limited by their complexity. The Ranson score and acute physiology and chronic health evaluation II (APACHE-II) score are based on numerous factors, which are not convenient for calculating. In a prospective study, the predictive capacity of Ranson, APACHE-II, BISAP, and CT severity index scoring systems for pancreatic necrosis was found to be suboptimal, with area under the curves (AUCs) of 0.70, 0.68, 0.61, and 0.75, respectively.6

Thus, the purpose of our study is to overcome these limitations and build a model that provides an accurate and simple prediction for NP development.

1. Study design and population

This retrospective study involved patients who were diagnosed with AP and admitted to Xiamen Hospital of Traditional Chinese Medicine between January 1, 2021, and August 31, 2022. This study received approval from the Ethics Committee of Xiamen Hospital of Traditional Chinese Medicine (IRB number: 2023-K023-01) and adhered to the ethical standards of the Declaration of Helsinki 2013. Due to the retrospective study design, written informed consent was waived. The final manuscript received unanimous approval from all authors.

For the initial diagnosis of AP in our study, non-contrast CT was utilized. CECT can detect the extent of pancreatic necrosis after 72 hours from symptom onset. It has demonstrated an early detection rate of 90% and near-perfect sensitivity beyond 4 days for pancreatic necrosis.7 Thus, we conducted CECT between 72 and 96 hours post-admission to evaluate patients for study inclusion. Exclusion criteria comprised chronic pancreatitis, pancreatic malignancy, and pregnancy. Additionally, to mitigate selection bias, we excluded patients presenting over 72 hours post-symptom onset, those with incomplete data, or those lacking CECT scans.

Patients diagnosed with AP upon admission who exhibited necrosis on CECT within the initial 72 to 96 hours were classified into the necrotizing group, while those without necrosis formed the edematous group.

2. Calculation of sample size

The sample size for the clinical prediction model adhered to the TRIPOD recommendation of 10 events per variable.8 We based our model on the smaller subset of binary classification outcomes multiplied by 10. The model in our study included five independent variables, thus the required positive event was at least 50 (=10×5). The incidence of NP in our study was 25.2% (=55/218×100%), so the required sample size was at least 198 (=50/25.2%). Thus, it is in accordance with the requirements.

3. Definition

Diagnosis of AP required fulfillment of at least two out of three criteria: (1) abdominal pain indicative of AP; (2) serum lipase or amylase levels exceeding three times the upper normal limit; (3) characteristic radiological findings of AP in imaging examinations.9

Hypertriglyceridemia AP was identified when serum triglyceride (TG) levels upon admission were ≥1,000 mg/dL or between 500–1,000 mg/dL with lactescent serum, in the absence of other etiologies.7

NP was characterized by area of non-enhancement in pancreatic tissue on CECT, with necrosis categorized as <30%, 30%–50%, or >50% (massive).10 A systemic inflammatory response syndrome was indicated by two or more of the following criteria: (1) heart rate >90 beats/min; (2) core temperature <36°C or >38°C; (3) white blood count <4,000 or >12,000/mm3; or (4) respirations >20/min or partial pressure of carbon dioxide <32 mm Hg. Organ failure was defined by a score of ≥2 on the modified Marshall scoring system.9

4. Clinical management protocols

Upon admission, patients received personalized conservative management, including intensive resuscitation, monitoring of fluid and electrolyte, nutritional support, and supportive care.

5. Clinical variables

We collected demographic and laboratory data from electronic medical records at the time of admission, including age, gender, etiology, alcohol consumption, smoking status, and the presence of diabetes or hypertension. Laboratory parameters encompassed white blood counts (WBC), hematocrit, C-reactive protein (CRP), procalcitonin (PCT), serum electrolyte levels, TG, total bilirubin, alanine aminotransferase, aspartate aminotransferase, lactic dehydrogenase, creatine, blood urea nitrogen (BUN), and glucose. Additionally, measurements of WBC, hematocrit, CRP, PCT, serum calcium, creatine, and BUN were repeated at 48 hours post-admission. Laboratory tests and radiological exams were performed within 24 hours of admission to calculate APACHE-II and BISAP scores, while the Ranson score was derived from tests conducted within 48 hours of admission. All data were collected and corroborated by two independent researchers.

6. Statistical analysis

Categorical data were presented as frequencies and percentages, and analyzed using the chi-square or Fisher exact tests. Continuous variables following a normal distribution were expressed as mean±standard deviation and analyzed with the two-sided Student t-test. Non-normally distributed variables were represented as medians with interquartile ranges and analyzed using the Mann-Whitney U test. The LASSO regression method was applied for variable selection and data dimensionality reduction. Predictive models were constructed via multiple logistic regression analysis. Nomograms were derived from this multivariate logistic regression analysis. Model performance was evaluated by receiver operating characteristic (ROC) curve, calibration plot, and decision curve analysis (DCA). Internal validation used 500 bootstrap resamples. Statistical computations were performed with R version 4.0.3, and significance was set at a two-sided p-value of <0.05. The nomogram and calibration curve were plotted using rms package and DCA was plotted with the rmda package. ROC was plotted with rms package.

1. Patients’ characteristics

From January 1, 2021, to August 31, 2022, 304 patients with AP were admitted to our hospitals. Application of exclusion criteria led to the removal of 86 patients from the study. The remaining cohort of 218 patients underwent further evaluation. The incidence of NP among these was 25.2%. The patient selection process is depicted in Fig. 1.

Figure 1.Flowchart of the process of patient enrollment. CECT, contrast-enhanced computed tomography; NP, necrotizing pancreatitis.

Detailed demographic and clinical information is described in Table 1. Of the 218 patients, 163 were diagnosed with non-NP, and 55 were diagnosed with NP. The median age did not differ between the two groups and the proportion of male patients was equally high between the two groups (p=0.203). The most common etiology of AP was hyperlipidemia (117/218, 53.7%), followed by others (52/218, 23.9%), gallstone (33/218, 15.1%), and alcohol (16/218, 7.3%).

Table 1. Characteristics of Patients in the Non-NP and NP Groups

Patient characteristicsAll (n=218)Non-NP (n=163)NP (n=55)p-value
Demographics and comorbidities
Male sex171 (78.4)124 (76.1)47 (85.5)0.203
Age, yr40.0 (32.0–48.0)40.0 (32.0–49.0)39.0 (31.5–47.0)0.531
Etiology0.003
Hyperlipidemia117 (53.7)76 (46.6)41 (74.5)<0.001
Gallstone33 (15.1)30 (18.4)3 (5.45)0.020
Alcohol16 (7.34)14 (8.59)2 (3.64)0.223
Others52 (23.9)43 (26.4)9 (16.4)0.132
Smoker85 (39.0)60 (36.8)25 (45.5)0.329
Drinker80 (36.7)54 (33.1)26 (47.3)0.085
Diabetes86 (39.4)51 (31.3)35 (63.6)<0.001
Hypertension41 (18.8)27 (16.6)14 (25.5)0.208
SIRS89 (40.8)50 (30.7)39 (70.9)<0.001
Organ failure56 (25.7)35 (21.5)21 (38.2)0.023
Laboratory data
WBC, ×109/L13.3 (10.9–16.2)12.5 (10.4–15.6)15.3 (12.4–17.8)0.004
HCT, %44.3 (40.1–46.6)44.1 (40.1–46.6)44.7 (40.8–46.3)0.465
CRP, mg/dL0.50 (0.10–2.70)0.37 (0.06–2.24)1.06 (0.21–3.71)0.025
PCT, ng/ml0.06 (0.05–0.16)0.05 (0.05–0.12)0.14 (0.05–0.60)<0.001
LDH, IU/L204 (176–258)202 (175–248)210 (186–293)0.335
BUN, mmol/L4.00 (3.30–4.90)4.00 (3.30–5.00)4.10 (3.35–4.60)0.708
Creatinine, mmol/L73.8 (65.9–83.5)73.9 (66.3–83.2)73.7 (65.6–87.3)0.544
Ca, mmol/L2.33 (2.23–2.42)2.33 (2.23–2.41)2.33 (2.21–2.46)0.778
Glucose, mmol/L8.00 (6.30–12.10)7.80 (6.10–11.20)10.9 (7.95–14.80)<0.001
Total bilirubin, mmol/L16.7 (13.1–24.4)16.7 (13.2–24.6)16.7 (11.6–23.5)0.792
ALT, IU/L30.0 (19.2–51.0)31.0 (20.0–53.5)28.0 (18.5–44.0)0.288
AST, IU/L25.5 (20.0–43.0)24.0 (19.0–43.0)31.0 (21.5–43.5)0.197
Triglyceride, mmol/l5.76 (1.32–15.50)4.25 (1.22–10.10)14.3 (4.09–34.70)<0.001
WBC 48 hr, ×109/L9.80 (7.70–12.70)9.40 (7.40–12.20)10.80 (8.05–14.40)0.029
HCT 48 hr, %40.7 (37.3–43.7)40.6 (37.1–43.2)41.1 (37.5–44.4)0.223
CRP 48 hr, mg/dL126 (59.4–208)103 (50.7–148)226 (153–303)<0.001
PCT 48 hr, ng/ml0.16 (0.05–0.46)0.10 (0.05–0.23)0.80 (0.38–1.86)<0.001
Ca 48 hr, mmol/L2.13 (2.03–2.22)2.15 (2.07–2.23)2.04 (1.88–2.16)<0.001
BUN 48 hr, mmol/L2.80 (2.10–3.48)2.70 (2.05–3.40)3.00 (2.55–4.00)0.008
Creatinine 48 hr, mmol/L70.3 (62.3–79.3)72.1 (62.5–79.6)68.1 (61.1–77.1)0.299
APACHEII3.00 (2.00–5.00)3.00 (1.00–4.00)4.00 (2.00–6.50)<0.001
Ranson1.00 (0.00–2.00)1.00 (0.00–2.00)2.00 (1.00–3.00)<0.001
BISAP1.00 (1.00–2.00)1.00 (1.00–2.00)2.00 (1.00–2.00)<0.001
Outcomes percutaneous catheter drainage2 (1.0)02 (3.6)0.063
ICU admission13 (6.0)1 (0.61)12 (21.8)<0.001
Hospital stays, day7.00 (5.00–8.75)6.00 (5.00–8.00)8.00 (7.00–13.00)<0.001
Mortality000
Percentage of pancreatic necrosis*--
<30%47 (21.6)47 (85.5)
30%–50%7 (3.2)7 (12.7)
>50%1 (0.5)1 (1.8)

Data are presented as number (%) or median (interquartile range).

NP, necrotizing pancreatitis; SIRS, systemic inflammatory response syndrome; WBC, white blood cell; HCT, hematocrit; CRP, C-reactive protein; PCT, procalcitonin; LDH, lactate dehydrogenase; BUN, blood urea nitrogen; Ca, serum calcium; ALT, alanine aminotransferase; AST, aspartate aminotransferase; 48 hr, at 48 hours post-admission; APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis; ICU, intensive care unit.

*Evaluated on the third day after admission.



Patients with hyperlipidemia and diabetes were likely to develop pancreatic necrosis.NP had a higher proportion of organ failure, systemic inflammatory response syndrome, longer hospital stays, and intensive care unit admission compared to edematous AP, with significant differences. At the time of admission, patients with NP showed significantly higher serum WBC, CRP, PCT, glucose, TG than those with non-NP. At admission 48 hours, patients with NP had significantly higher serum WBC, CRP, PCT, calcium, and BUN than non-NP. Among the various scoring systems, APACHE-II, BISAP, and Ranson were significantly different between NP and non-NP.

Of the 28 variables collected from patients, 14 features were selected based on non-zero coefficients generated by LASSO logistic regression analysis (Fig. 2). These selected features included etiology, smoking status, alcohol consumption, diabetes, systemic inflammatory response syndrome, CRP, PCT, TG, and values for WBC, HCT, CRP, PCT, calcium, and BUN at 48 hours post-admission. These features were subsequently utilized in multivariate logistic regression analysis.

Figure 2.Features selection using the LASSO binary logistic regression model. (A) Log (lambda) value of the 28 features in the LASSO model: a coefficient profile plot was produced against the log (lambda) sequence. (B) Parameter selection in the LASSO model used tenfold cross-validation via minimum criterion: partial likelihood deviation (binomial deviation) curves and logarithmic (lambda) curves were plotted. Use the minimum standard and 1 se (1-SE standard) of the minimum standard to draw a vertical dashed line at the optimal value. The optimal lambda produced four nonzero coefficients. LASSO, least absolute shrinkage and selection operator; SE, standard error.

2. Risk factors for NP development

To identify clinical indicators that could predict NP, we conducted a multivariate logistic regression analysis with the aforementioned 14 variables identified by the LASSO regression technique. Ultimately, PCT, TG, WBC at 48 hours, HCT at 48 hours, and calcium at 48 hours were found to be independent predictors (Table 2).

Table 2. Association of Necrotizing Pancreatitis in Multivariable Analysis

VariableMultivariable regression
BOR (95% CI)p-value
Intercept0.3071.360 (0.005–435.8)0.916
PCT0.681.973 (1.285–3.826)0.008
TG0.0481.049 (1.025–1.075)<0.001
WBC 48 hr0.1421.153 (1.055–1.267)0.002
HCT 48 hr0.1131.120 (1.034–1.224)0.008
Ca 48 hr–4.0680.017 (0.001–0.161)0.001

OR, odds ratio; CI, confidence interval; PCT, procalcitonin; TG, triglyceride; WBC, white blood cell; HCT, hematocrit; Ca, serum calcium; 48 hr, at 48 hours post-admission.



3. Nomogram model for NP development

The results from the multivariate logistic regression analysis were employed to develop a nomogram model to predict the probability of NP. The model calculation is as follows: 0.307+0.68×(PCT)+0.048×(TG)+0.142×(WBC at 48 hours post-admission)–4.068×(Ca at 48 hours post-admission)+0.113×(HCT at 48 hours post-admission). The probability of NP can be determined using the nomogram, as illustrated in Fig. 3.

Figure 3.Nomogram prediction of pancreatic necrosis. The value of each variable was scored on a point scale from 0 to 100, after which the scores for each variable were added together. That sum is located on the total points axis, which enables us to predict the probability of necrotizing pancreatitis risk. PCT, procalcitonin; TG, triglyceride; WBC, white blood cell; HCT, hematocrit; Ca, serum calcium; 48 hr, at admission 48th hour.

4. Validation of the predictive accuracy and clinical value of nomogram model

We assessed the accuracy of the nomogram model, which was further validated internally using bootstrap sampling. After 500 bootstrap replications to reduce overfitting, the AUC for the cohort was found to be 0.822 (95% confidence interval [CI], 0.737 to 0.892); the calibration curve demonstrated good agreement between predicted and observed probabilities (H-L test: χ2=12.897, p=0.167). Following 500 bootstrap samples, DCA indicated that at a threshold probability for NP of >8%, the nomogram model provided a positive net benefit over both the treat-all or treat-none strategies (Fig. 4).

Figure 4.(A) Receiver operating characteristic curve for the nomogram was measured by bootstrapping for 500 repetitions. (B) Calibration curve using bootstraps sampling 500 for predicted probability of the nomogram. When the gray line (performance nomogram) was closer to the dotted line (ideal model), the prediction accuracy of the nomogram was better. (C) Decision curve analysis using bootstraps sampling 500 for the prediction model. The dotted line is from the prediction model, the line is for all patients with necrotizing pancreatitis (NP), and the solid horizontal line indicates no patients have NP.

5. Predictive performances and clinical value of Ranson, APACHE-II, BISAP, and nomogram model

ROCs of Ranson, APACHE-II, BISAP, and the nomogram model were depicted in Fig. 5. An ROC curve analysis of the nomogram model yielded an AUC of 0.822 (95% CI, 0.748 to 0.897), indicating good predictive performance with a sensitivity of 82.8% and a specificity of 76.4% at the optimal cutoff value of 0.260. The model's AUC was significantly higher than that of BISAP (AUC, 0.686; 95% CI, 0.612 to 0.761; p<0.001), APACHE-II (AUC, 0.664; 95% CI, 0.578 to 0.749; p<0.001), and Ranson (AUC, 0.666; 95% CI, 0.586 to 0.746; p<0.001) (Fig. 5).

Figure 5.(A) Receiver operating characteristic curves for the nomogram and other existing clinical scoring systems. (B) Decision curve analysis for the nomogram and other existing clinical scoring systems. APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis.

To ascertain the superiority in prediction of NP, we employed the IDI and NRI to compare the discriminative abilities of the nomogram model against other clinical scoring systems. Our findings suggest that the nomogram model exhibits superior accuracy in predicting NP, indicating its potential as a valuable clinical tool (Table 3).

Table 3. Predictive Performance of the Integrated Discrimination Improvement between the Model and Other Clinical Scores

VariableIDI (95% CI)p-valueNRI (95% CI)p-value
Model-APACHE-II0.243 (0.165–0.321)<0.0010.989 (0.718–1.260)<0.001
Model-BISAP0.251 (0.173–0.329)<0.0010.890 (0.620–1.160)<0.001
Model-Ranson0.274 (0.200–0.349)<0.0011.061 (0.802–1.320)<0.001

IDI, integrated discrimination improvement; CI, confidence interval; NRI, net reclassification index; APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis.



The DCA curves for Ranson, APACHE-II, BISAP, and nomogram model are presented in Fig. 5. The nomogram model consistently demonstrated greater net benefit than Ranson, APACHE-II, and BISAP in predicting NP with a threshold probability of >8%.

6. Predictive performances of the nomogram model in hyperlipidemic pancreatitis and no-hyperlipidemic pancreatitis

ROC curve analysis of the nomogram model for hyperlipidemic and no-hyperlipidemic pancreatitis yielded AUCs of 0.795 (95% CI, 0.698 to 0.891) and 0.838 (95% CI, 0.706 to 0.971).

Early diagnosis of NP is crucial for the management of AP. We developed a predictive nomogram for early-stage NP based on multivariate analyses, which may mitigate some adverse effects associated with CECT. The model suggests that higher levels of PCT and TG at admission, as well as elevated WBC, calcium, and BUN at 48 hours post-admission, increase the risk of NP development. This nomogram could serve as a practical tool for predicting NP likelihood in AP patients, utilizing readily available clinical parameters. The nomogram exhibited excellent diagnostic performance (AUC 0.822, sensitivity 82.8%, and specificity 76.4%) and was internally validated using bootstrap sampling. Moreover, its clinical utility was confirmed by DCA, indicating its effectiveness in a clinical environment.

Our study found that NP was associated with a higher incidence of organ failure, longer hospital stays, and increased intensive care unit admissions compared to edematous AP, aligning with findings from Ugurlu et al.11,12 These studies also indicated higher mortality rates for necrotizing AP,11,12 a trend not observed in our data. This discrepancy may stem from our exclusion of patients with incomplete data at 48 hours and without CECT, potentially omitting those with severe conditions leading to early discharge or those unable to undergo CECT. Consequently, the NP group in our study had lower APACHE-II scores. A retrospective study from China recognized the APACHE-II score as a significant mortality risk factor in NP.13 Alternatively, this may be due to early and aggressive care that includes vigilant monitoring of inflammation markers and evaluating patients within the first 72 to 96 hours using CECT.

AP is characterized by non-infected inflammation of pancreatic tissue. The development of pancreatic necrosis is closely related to both the extent of inflammation and the disease severity. Consequently, higher WBC levels, a biomarker of systemic inflammation, are anticipated in NP. Acehan et al.12 demonstrated that WBC levels are independent predictors of pancreatic necrosis. Similarly, a study by Unal and Barlas14 found that WBC levels were significantly higher in NP patients than those with edematous AP. Our findings reinforce this relationship, showing that WBC levels at the 48th hour of admission are associated with an increased odds ratio of 1.153 for NP development.

PCT, a propeptide secreted by hepatocytes and the thyroid in response to severe inflammation and sepsis,1,6 serves as a biomarker in distinguishing necrotizing from edematous pancreatitis.15 Elevated PCT levels have been recognized as early indicators of disease severity.6 A recent retrospective study of 582 AP patients in Turkey found PCT to be an early predictor of NP as suggested by ROC analysis.11 Echoing these results, our study identifies PCT at the 48th hour post-admission as an independent predictor of necrosis, increasing the risk of necrosis development by a factor of 1.97.

NP is linked to third-space fluid loss, leading to hypoperfusion, splanchnic vasoconstriction, and reduced microcirculation in the pancreas.16 Earlier studies17,18 have shown that early fluid resuscitation can diminish the risk of NP, attenuate systemic inflammatory response, prevent multi-organ failure, and maintain pancreatic microcirculation. Hence, hemoconcentration resulting from fluid loss is likely associated with the severity of AP. In a multicenter, prospective study conducted in the United States and the Netherlands, HCT was identified as a potential predictor of NP, with an HCT level at admission ≥44% being associated with an odds ratio of 3.11 for NP development.19 Recent findings by Hidalgo et al.20 also confirmed that HCT at admission correlates with NP development. Moreover, clinical studies have demonstrated that HCT levels at the 48th hour are independent risk factors for necrosis in AP.12 In line with these studies, our research indicates that higher HCT levels at the 48th hour of admission are indicative of an increased risk for necrosis development in AP.

Fatty acids were hydrolyzed from excess TG by pancreatic lipase, which can result in inflammation of the pancreas and intracellular calcium influx, subsequently causing pancreatic necrosis.21 Current evidence suggests that patients with elevated TG levels exhibit a more severe form of pancreatitis. Mosztbacher et al.22 demonstrated that TG levels dose-dependently exacerbate the severity and complications of AP. A prospective observational study in Spain, which focused on a single cohort, revealed that TG levels in the early stages of AP are linked to an increased risk of developing NP.20 Cheng et al.23 found that patients with acute biliary pancreatitis and high TG levels tend to develop more extensive necrosis. Similarly, our research corroborated that elevated TG constitutes an independent risk factor for the development of necrosis in AP.

Hypocalcemia in AP is attributed to the release of free fatty acids, which may lead to the formation of calcium salts and impairment of parathyroid function.24 Yu et al.25 found that patients with hypocalcemia in hyperlipidemic pancreatitis face a higher risk of severe pancreatitis. Another study indicated that serum calcium levels elevate the risk of necrosis development by 0.021-fold at values above 2.04 mmol/L.26 Our research suggests that hypocalcemia at the 48-hour mark of admission is an independent risk factor for the development of NP.

As highlighted in the introduction, common AP scoring systems, such as APACHE-II, Ranson, and BISAP, are limited by the requirement for CT imaging or complex calculations. Furthermore, these systems have not significantly predicted necrosis development.27 Consequently, there is a need for an accurate model that utilizes a small number of variables without radiological findings for clinical application. The nomogram described in our study requires only five parameters (PCT, TG, WBC at 48 hours, Ca at 48 hours, and HCT at 48 hours) to predict NP development, which are widely used and easily computed. Thus, our risk assessment model could has the potential for broad acceptance. Moreover, it demonstrated greater predictive efficacy and was more beneficial in clinical decision-making for patients with NP than the other three AP scoring systems. Despite the limited sample size of our study, which may affect the external validity of the model, we performed internal validation using bootstrap sampling to affirm the nomogram's predictive accuracy and clinical utility.

It is crucial to recognize several limitations of this study. Firstly, its single-center, retrospective design may introduce selection bias. Secondly, although the ROC curve analysis yielded AUCs of 0.795 (95% CI, 0.698 to 0.891) for hyperlipidemic pancreatitis and 0.838 (95% CI, 0.706 to 0.971) for non-hyperlipidemic pancreatitis, hyperlipidemia constitutes over half the etiologies in our cohort, as opposed to typical AP cohorts where alcohol and gallstones predominate. Consequently, the generalizability of our nomogram to all AP cases may be limited. Thirdly, with only three factors (WBC, Ca, and HCT at 48 hours), the nomogram cannot predict necrotizing AP prior to 48 hours post-admission. A multicenter, prospective trial is required to confirm the model's accuracy.

In conclusion, our study presents a nomogram model designed to calculate a risk score and identify patients at an increased likelihood of early NP development. The application of this model as a convenient and specific tool may prove advantageous for clinical decision-making.

No potential conflict of interest relevant to this article was reported.

Study concept and design: J.Z. Data acquisition: J.Z. Data analysis and interpretation: J.Z. Drafting of the manuscript: J.Z. Critical revision of the manuscript for important intellectual content: X.W. Statistical analysis: J.Z. Administrative, technical, or material support; study supervision: X.W. Approval of final manuscript: all authors.

  1. Asim Riaz HM, Islam Z, Rasheed L, et al. The evaluation of inflammatory biomarkers in predicting progression of acute pancreatitis to pancreatic necrosis: a diagnostic test accuracy review. Healthcare (Basel) 2022;11:27.
    Pubmed KoreaMed CrossRef
  2. Jha AK, Goenka MK, Kumar R, Suchismita A. Endotherapy for pancreatic necrosis: an update. JGH Open 2019;3:80-88.
    Pubmed KoreaMed CrossRef
  3. Colvin SD, Smith EN, Morgan DE, Porter KK. Acute pancreatitis: an update on the revised Atlanta classification. Abdom Radiol (NY) 2020;45:1222-1231.
    Pubmed CrossRef
  4. Kiss S, Pintér J, Molontay R, et al. Early prediction of acute necrotizing pancreatitis by artificial intelligence: a prospective cohort-analysis of 2387 cases. Sci Rep 2022;12:7827.
    Pubmed KoreaMed CrossRef
  5. Leonard-Murali S, Lezotte J, Kalu R, et al. Necrotizing pancreatitis: a review for the acute care surgeon. Am J Surg 2021;221:927-934.
    Pubmed KoreaMed CrossRef
  6. Khanna AK, Meher S, Prakash S, et al. Comparison of Ranson, Glasgow, MOSS, SIRS, BISAP, APACHE-II, CTSI scores, IL-6, CRP, and procalcitonin in predicting severity, organ failure, pancreatic necrosis, and mortality in acute pancreatitis. HPB Surg 2013;2013:367581.
    Pubmed KoreaMed CrossRef
  7. Leppäniemi A, Tolonen M, Tarasconi A, et al. 2019 WSES guidelines for the management of severe acute pancreatitis. World J Emerg Surg 2019;14:27.
    Pubmed KoreaMed CrossRef
  8. Moons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162:W1-W73.
    Pubmed CrossRef
  9. Banks PA, Bollen TL, Dervenis C, et al. Classification of acute pancreatitis: 2012. Revision of the Atlanta classification and definitions by international consensus. Gut 2013;62:102-111.
    Pubmed CrossRef
  10. Garg PK, Madan K, Pande GK, et al. Association of extent and infection of pancreatic necrosis with organ failure and death in acute necrotizing pancreatitis. Clin Gastroenterol Hepatol 2005;3:159-166.
    Pubmed CrossRef
  11. Ugurlu ET, Yurekli UF. The effect of procalcitonin and immature granulocyte ratio in predicting the development of acute necrotizing pancreatitis: evidence from 582 cases. Eur Rev Med Pharmacol Sci 2022;26:7514-7521.
    Pubmed CrossRef
  12. Acehan F, Comoglu M, Kayserili FM, Hayat B, Ates I. Factors predicting the development of necrosis in patients presenting with edematous acute pancreatitis. Pancreas 2022;51:1300-1307.
    Pubmed CrossRef
  13. Yu L, Xie F, Luo L, et al. Clinical characteristics and risk factors of organ failure and death in necrotizing pancreatitis. BMC Gastroenterol 2023;23:19.
    Pubmed KoreaMed CrossRef
  14. Ünal Y, Barlas AM. Role of increased immature granulocyte percentage in the early prediction of acute necrotizing pancreatitis. Ulus Travma Acil Cerrahi Derg 2019;25:177-182.
    Pubmed CrossRef
  15. Silva-Vaz P, Abrantes AM, Castelo-Branco M, Gouveia A, Botelho MF, Tralhão JG. Multifactorial scores and biomarkers of prognosis of acute pancreatitis: applications to research and practice. Int J Mol Sci 2020;21:338.
    Pubmed KoreaMed CrossRef
  16. Di Martino M, Van Laarhoven S, Ielpo B, et al. Systematic review and meta-analysis of fluid therapy protocols in acute pancreatitis: type, rate and route. HPB (Oxford) 2021;23:1629-1638.
    Pubmed CrossRef
  17. Warndorf MG, Kurtzman JT, Bartel MJ, et al. Early fluid resuscitation reduces morbidity among patients with acute pancreatitis. Clin Gastroenterol Hepatol 2011;9:705-709.
    Pubmed KoreaMed CrossRef
  18. Wu BU, Johannes RS, Sun X, Conwell DL, Banks PA. Early changes in blood urea nitrogen predict mortality in acute pancreatitis. Gastroenterology 2009;137:129-135.
    Pubmed CrossRef
  19. Koutroumpakis E, Wu BU, Bakker OJ, et al. Admission hematocrit and rise in blood urea nitrogen at 24 h outperform other laboratory markers in predicting persistent organ failure and pancreatic necrosis in acute pancreatitis: a post hoc analysis of three large prospective databases. Am J Gastroenterol 2015;110:1707-1716.
    Pubmed CrossRef
  20. Hidalgo NJ, Pando E, Alberti P, et al. The role of high serum triglyceride levels on pancreatic necrosis development and related complications. BMC Gastroenterol 2023;23:51.
    Pubmed KoreaMed CrossRef
  21. Yang AL, McNabb-Baltar J. Hypertriglyceridemia and acute pancreatitis. Pancreatology 2020;20:795-800.
    Pubmed CrossRef
  22. Mosztbacher D, Hanák L, Farkas N, et al. Hypertriglyceridemia-induced acute pancreatitis: a prospective, multicenter, international cohort analysis of 716 acute pancreatitis cases. Pancreatology 2020;20:608-616.
    Pubmed CrossRef
  23. Cheng L, Luo Z, Xiang K, et al. Clinical significance of serum triglyceride elevation at early stage of acute biliary pancreatitis. BMC Gastroenterol 2015;15:19.
    Pubmed KoreaMed CrossRef
  24. Dong X, Pan S, Zhang D, et al. Hyperlipemia pancreatitis onset time affects the association between elevated serum triglyceride levels and disease severity. Lipids Health Dis 2022;21:49.
    Pubmed KoreaMed CrossRef
  25. Yu S, Wu D, Jin K, et al. Low serum ionized calcium, elevated high-sensitivity C-reactive protein, neutrophil-lymphocyte ratio, and body mass index (BMI) are risk factors for severe acute pancreatitis in patients with hypertriglyceridemia pancreatitis. Med Sci Monit 2019;25:6097-6103.
    Pubmed KoreaMed CrossRef
  26. Xue E, Shi Q, Guo S, et al. Preexisting diabetes, serum calcium and D-dimer levels as predictable risk factors for pancreatic necrosis of patients with acute pancreatitis: a retrospective study. Expert Rev Gastroenterol Hepatol 2022;16:913-921.
    Pubmed CrossRef
  27. Hagjer S, Kumar N. Evaluation of the BISAP scoring system in prognostication of acute pancreatitis: a prospective observational study. Int J Surg 2018;54:76-81.
    Pubmed CrossRef

Article

ahead

Gut and Liver

Published online February 22, 2024

Copyright © Gut and Liver.

Development of a Nomogram to Predict the Risk for Acute Necrotizing Pancreatitis

Junchao Zhang , Xiaxia Weng

Department of Gastroenterology, Xiamen Hospital of Traditional Chinese Medicine, Xiamen, China

Correspondence to:Xiaxia Weng
ORCID https://orcid.org/0009-0002-6310-0642
E-mail 302561422@qq.com

Received: October 4, 2023; Revised: November 15, 2023; Accepted: November 21, 2023

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Background/Aims: Necrotizing pancreatitis (NP) presents a more severe clinical trajectory and increased mortality compared to edematous pancreatitis. Prompt identification of NP is vital for patient prognosis. A risk prediction model for NP among Chinese patients has been developed and validated to aid in early detection.
Methods: A retrospective analysis was performed on 218 patients with acute pancreatitis (AP) to examine the association of various clinical variables with NP. The least absolute shrinkage and selection operator (LASSO) regression was utilized to refine variables and select predictors. Subsequently, a multivariate logistic regression was employed to construct a predictive nomogram. The model's accuracy was validated using bootstrap resampling (n=500) and its calibration assessed via a calibration curve. The model's clinical utility was evaluated through decision curve analysis.
Results: Of the 28 potential predictors analyzed in 218 AP patients, the incidence of NP was 25.2%. LASSO regression identified 14 variables, with procalcitonin, triglyceride, white blood cell count at 48 hours post-admission, calcium at 48 hours post-admission, and hematocrit at 48 hours post-admission emerging as independent risk factors for NP. The resulting nomogram accurately predicted NP risk with an area under the curve of 0.822, sensitivity of 82.8%, and specificity of 76.4%. The bootstrap-validated area under the curve remained at 0.822 (95% confidence interval, 0.737 to 0.892). This model exhibited excellent calibration and demonstrated greater predictive efficacy and clinical utility for NP than APACHE II, Ranson, and BISAP.
Conclusions: We have developed a prediction nomogram of NP that is of great value in guiding clinical decision.

Keywords: Pancreatitis, Acute necrotizing, Nomograms, Bootstrap, Predict

INTRODUCTION

Acute pancreatitis (AP) is an acute disorder of the exocrine pancreas, which is commonly linked to acinar cell injury and local/systemic inflammatory responses.1 While most cases of AP are mild and self-limiting, approximately 20% of patients progress to necrotizing pancreatitis (NP), a severe form of the disease associated with a mortality risk of up to 15%–20%.2 Approximately one-third of patients with NP may develop pancreatic infection during the natural course of AP, thus the relative risk of death is up 2-fold.3 Therefore, early identification of patients with NP is vital to assess disease severity and determine optimal treatment.

Contrast-enhanced computed tomography (CECT) is the primary diagnostic modality for identifying NP. However, its effectiveness is often limited due to NP typically becoming discernible on imaging only 3 days after the onset of the disease; Moreover, it is not accessible in all cases.4 Therefore, there is a need to develop other methods to supplement acute necrotizing pancreatitis assessment.

Several scoring systems have been developed to predict the severity of pancreatitis.5 However, each system has its unique shortcomings. The Balthazar computer tomography (CT) severity index and bedside index for severity in acute pancreatitis (BISAP) score are restricted by the availability of CT. The application of other scoring systems without CT are limited by their complexity. The Ranson score and acute physiology and chronic health evaluation II (APACHE-II) score are based on numerous factors, which are not convenient for calculating. In a prospective study, the predictive capacity of Ranson, APACHE-II, BISAP, and CT severity index scoring systems for pancreatic necrosis was found to be suboptimal, with area under the curves (AUCs) of 0.70, 0.68, 0.61, and 0.75, respectively.6

Thus, the purpose of our study is to overcome these limitations and build a model that provides an accurate and simple prediction for NP development.

MATERIALS AND METHODS

1. Study design and population

This retrospective study involved patients who were diagnosed with AP and admitted to Xiamen Hospital of Traditional Chinese Medicine between January 1, 2021, and August 31, 2022. This study received approval from the Ethics Committee of Xiamen Hospital of Traditional Chinese Medicine (IRB number: 2023-K023-01) and adhered to the ethical standards of the Declaration of Helsinki 2013. Due to the retrospective study design, written informed consent was waived. The final manuscript received unanimous approval from all authors.

For the initial diagnosis of AP in our study, non-contrast CT was utilized. CECT can detect the extent of pancreatic necrosis after 72 hours from symptom onset. It has demonstrated an early detection rate of 90% and near-perfect sensitivity beyond 4 days for pancreatic necrosis.7 Thus, we conducted CECT between 72 and 96 hours post-admission to evaluate patients for study inclusion. Exclusion criteria comprised chronic pancreatitis, pancreatic malignancy, and pregnancy. Additionally, to mitigate selection bias, we excluded patients presenting over 72 hours post-symptom onset, those with incomplete data, or those lacking CECT scans.

Patients diagnosed with AP upon admission who exhibited necrosis on CECT within the initial 72 to 96 hours were classified into the necrotizing group, while those without necrosis formed the edematous group.

2. Calculation of sample size

The sample size for the clinical prediction model adhered to the TRIPOD recommendation of 10 events per variable.8 We based our model on the smaller subset of binary classification outcomes multiplied by 10. The model in our study included five independent variables, thus the required positive event was at least 50 (=10×5). The incidence of NP in our study was 25.2% (=55/218×100%), so the required sample size was at least 198 (=50/25.2%). Thus, it is in accordance with the requirements.

3. Definition

Diagnosis of AP required fulfillment of at least two out of three criteria: (1) abdominal pain indicative of AP; (2) serum lipase or amylase levels exceeding three times the upper normal limit; (3) characteristic radiological findings of AP in imaging examinations.9

Hypertriglyceridemia AP was identified when serum triglyceride (TG) levels upon admission were ≥1,000 mg/dL or between 500–1,000 mg/dL with lactescent serum, in the absence of other etiologies.7

NP was characterized by area of non-enhancement in pancreatic tissue on CECT, with necrosis categorized as <30%, 30%–50%, or >50% (massive).10 A systemic inflammatory response syndrome was indicated by two or more of the following criteria: (1) heart rate >90 beats/min; (2) core temperature <36°C or >38°C; (3) white blood count <4,000 or >12,000/mm3; or (4) respirations >20/min or partial pressure of carbon dioxide <32 mm Hg. Organ failure was defined by a score of ≥2 on the modified Marshall scoring system.9

4. Clinical management protocols

Upon admission, patients received personalized conservative management, including intensive resuscitation, monitoring of fluid and electrolyte, nutritional support, and supportive care.

5. Clinical variables

We collected demographic and laboratory data from electronic medical records at the time of admission, including age, gender, etiology, alcohol consumption, smoking status, and the presence of diabetes or hypertension. Laboratory parameters encompassed white blood counts (WBC), hematocrit, C-reactive protein (CRP), procalcitonin (PCT), serum electrolyte levels, TG, total bilirubin, alanine aminotransferase, aspartate aminotransferase, lactic dehydrogenase, creatine, blood urea nitrogen (BUN), and glucose. Additionally, measurements of WBC, hematocrit, CRP, PCT, serum calcium, creatine, and BUN were repeated at 48 hours post-admission. Laboratory tests and radiological exams were performed within 24 hours of admission to calculate APACHE-II and BISAP scores, while the Ranson score was derived from tests conducted within 48 hours of admission. All data were collected and corroborated by two independent researchers.

6. Statistical analysis

Categorical data were presented as frequencies and percentages, and analyzed using the chi-square or Fisher exact tests. Continuous variables following a normal distribution were expressed as mean±standard deviation and analyzed with the two-sided Student t-test. Non-normally distributed variables were represented as medians with interquartile ranges and analyzed using the Mann-Whitney U test. The LASSO regression method was applied for variable selection and data dimensionality reduction. Predictive models were constructed via multiple logistic regression analysis. Nomograms were derived from this multivariate logistic regression analysis. Model performance was evaluated by receiver operating characteristic (ROC) curve, calibration plot, and decision curve analysis (DCA). Internal validation used 500 bootstrap resamples. Statistical computations were performed with R version 4.0.3, and significance was set at a two-sided p-value of <0.05. The nomogram and calibration curve were plotted using rms package and DCA was plotted with the rmda package. ROC was plotted with rms package.

RESULTS

1. Patients’ characteristics

From January 1, 2021, to August 31, 2022, 304 patients with AP were admitted to our hospitals. Application of exclusion criteria led to the removal of 86 patients from the study. The remaining cohort of 218 patients underwent further evaluation. The incidence of NP among these was 25.2%. The patient selection process is depicted in Fig. 1.

Figure 1. Flowchart of the process of patient enrollment. CECT, contrast-enhanced computed tomography; NP, necrotizing pancreatitis.

Detailed demographic and clinical information is described in Table 1. Of the 218 patients, 163 were diagnosed with non-NP, and 55 were diagnosed with NP. The median age did not differ between the two groups and the proportion of male patients was equally high between the two groups (p=0.203). The most common etiology of AP was hyperlipidemia (117/218, 53.7%), followed by others (52/218, 23.9%), gallstone (33/218, 15.1%), and alcohol (16/218, 7.3%).

Table 1 . Characteristics of Patients in the Non-NP and NP Groups.

Patient characteristicsAll (n=218)Non-NP (n=163)NP (n=55)p-value
Demographics and comorbidities
Male sex171 (78.4)124 (76.1)47 (85.5)0.203
Age, yr40.0 (32.0–48.0)40.0 (32.0–49.0)39.0 (31.5–47.0)0.531
Etiology0.003
Hyperlipidemia117 (53.7)76 (46.6)41 (74.5)<0.001
Gallstone33 (15.1)30 (18.4)3 (5.45)0.020
Alcohol16 (7.34)14 (8.59)2 (3.64)0.223
Others52 (23.9)43 (26.4)9 (16.4)0.132
Smoker85 (39.0)60 (36.8)25 (45.5)0.329
Drinker80 (36.7)54 (33.1)26 (47.3)0.085
Diabetes86 (39.4)51 (31.3)35 (63.6)<0.001
Hypertension41 (18.8)27 (16.6)14 (25.5)0.208
SIRS89 (40.8)50 (30.7)39 (70.9)<0.001
Organ failure56 (25.7)35 (21.5)21 (38.2)0.023
Laboratory data
WBC, ×109/L13.3 (10.9–16.2)12.5 (10.4–15.6)15.3 (12.4–17.8)0.004
HCT, %44.3 (40.1–46.6)44.1 (40.1–46.6)44.7 (40.8–46.3)0.465
CRP, mg/dL0.50 (0.10–2.70)0.37 (0.06–2.24)1.06 (0.21–3.71)0.025
PCT, ng/ml0.06 (0.05–0.16)0.05 (0.05–0.12)0.14 (0.05–0.60)<0.001
LDH, IU/L204 (176–258)202 (175–248)210 (186–293)0.335
BUN, mmol/L4.00 (3.30–4.90)4.00 (3.30–5.00)4.10 (3.35–4.60)0.708
Creatinine, mmol/L73.8 (65.9–83.5)73.9 (66.3–83.2)73.7 (65.6–87.3)0.544
Ca, mmol/L2.33 (2.23–2.42)2.33 (2.23–2.41)2.33 (2.21–2.46)0.778
Glucose, mmol/L8.00 (6.30–12.10)7.80 (6.10–11.20)10.9 (7.95–14.80)<0.001
Total bilirubin, mmol/L16.7 (13.1–24.4)16.7 (13.2–24.6)16.7 (11.6–23.5)0.792
ALT, IU/L30.0 (19.2–51.0)31.0 (20.0–53.5)28.0 (18.5–44.0)0.288
AST, IU/L25.5 (20.0–43.0)24.0 (19.0–43.0)31.0 (21.5–43.5)0.197
Triglyceride, mmol/l5.76 (1.32–15.50)4.25 (1.22–10.10)14.3 (4.09–34.70)<0.001
WBC 48 hr, ×109/L9.80 (7.70–12.70)9.40 (7.40–12.20)10.80 (8.05–14.40)0.029
HCT 48 hr, %40.7 (37.3–43.7)40.6 (37.1–43.2)41.1 (37.5–44.4)0.223
CRP 48 hr, mg/dL126 (59.4–208)103 (50.7–148)226 (153–303)<0.001
PCT 48 hr, ng/ml0.16 (0.05–0.46)0.10 (0.05–0.23)0.80 (0.38–1.86)<0.001
Ca 48 hr, mmol/L2.13 (2.03–2.22)2.15 (2.07–2.23)2.04 (1.88–2.16)<0.001
BUN 48 hr, mmol/L2.80 (2.10–3.48)2.70 (2.05–3.40)3.00 (2.55–4.00)0.008
Creatinine 48 hr, mmol/L70.3 (62.3–79.3)72.1 (62.5–79.6)68.1 (61.1–77.1)0.299
APACHEII3.00 (2.00–5.00)3.00 (1.00–4.00)4.00 (2.00–6.50)<0.001
Ranson1.00 (0.00–2.00)1.00 (0.00–2.00)2.00 (1.00–3.00)<0.001
BISAP1.00 (1.00–2.00)1.00 (1.00–2.00)2.00 (1.00–2.00)<0.001
Outcomes percutaneous catheter drainage2 (1.0)02 (3.6)0.063
ICU admission13 (6.0)1 (0.61)12 (21.8)<0.001
Hospital stays, day7.00 (5.00–8.75)6.00 (5.00–8.00)8.00 (7.00–13.00)<0.001
Mortality000
Percentage of pancreatic necrosis*--
<30%47 (21.6)47 (85.5)
30%–50%7 (3.2)7 (12.7)
>50%1 (0.5)1 (1.8)

Data are presented as number (%) or median (interquartile range)..

NP, necrotizing pancreatitis; SIRS, systemic inflammatory response syndrome; WBC, white blood cell; HCT, hematocrit; CRP, C-reactive protein; PCT, procalcitonin; LDH, lactate dehydrogenase; BUN, blood urea nitrogen; Ca, serum calcium; ALT, alanine aminotransferase; AST, aspartate aminotransferase; 48 hr, at 48 hours post-admission; APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis; ICU, intensive care unit..

*Evaluated on the third day after admission..



Patients with hyperlipidemia and diabetes were likely to develop pancreatic necrosis.NP had a higher proportion of organ failure, systemic inflammatory response syndrome, longer hospital stays, and intensive care unit admission compared to edematous AP, with significant differences. At the time of admission, patients with NP showed significantly higher serum WBC, CRP, PCT, glucose, TG than those with non-NP. At admission 48 hours, patients with NP had significantly higher serum WBC, CRP, PCT, calcium, and BUN than non-NP. Among the various scoring systems, APACHE-II, BISAP, and Ranson were significantly different between NP and non-NP.

Of the 28 variables collected from patients, 14 features were selected based on non-zero coefficients generated by LASSO logistic regression analysis (Fig. 2). These selected features included etiology, smoking status, alcohol consumption, diabetes, systemic inflammatory response syndrome, CRP, PCT, TG, and values for WBC, HCT, CRP, PCT, calcium, and BUN at 48 hours post-admission. These features were subsequently utilized in multivariate logistic regression analysis.

Figure 2. Features selection using the LASSO binary logistic regression model. (A) Log (lambda) value of the 28 features in the LASSO model: a coefficient profile plot was produced against the log (lambda) sequence. (B) Parameter selection in the LASSO model used tenfold cross-validation via minimum criterion: partial likelihood deviation (binomial deviation) curves and logarithmic (lambda) curves were plotted. Use the minimum standard and 1 se (1-SE standard) of the minimum standard to draw a vertical dashed line at the optimal value. The optimal lambda produced four nonzero coefficients. LASSO, least absolute shrinkage and selection operator; SE, standard error.

2. Risk factors for NP development

To identify clinical indicators that could predict NP, we conducted a multivariate logistic regression analysis with the aforementioned 14 variables identified by the LASSO regression technique. Ultimately, PCT, TG, WBC at 48 hours, HCT at 48 hours, and calcium at 48 hours were found to be independent predictors (Table 2).

Table 2 . Association of Necrotizing Pancreatitis in Multivariable Analysis.

VariableMultivariable regression
BOR (95% CI)p-value
Intercept0.3071.360 (0.005–435.8)0.916
PCT0.681.973 (1.285–3.826)0.008
TG0.0481.049 (1.025–1.075)<0.001
WBC 48 hr0.1421.153 (1.055–1.267)0.002
HCT 48 hr0.1131.120 (1.034–1.224)0.008
Ca 48 hr–4.0680.017 (0.001–0.161)0.001

OR, odds ratio; CI, confidence interval; PCT, procalcitonin; TG, triglyceride; WBC, white blood cell; HCT, hematocrit; Ca, serum calcium; 48 hr, at 48 hours post-admission..



3. Nomogram model for NP development

The results from the multivariate logistic regression analysis were employed to develop a nomogram model to predict the probability of NP. The model calculation is as follows: 0.307+0.68×(PCT)+0.048×(TG)+0.142×(WBC at 48 hours post-admission)–4.068×(Ca at 48 hours post-admission)+0.113×(HCT at 48 hours post-admission). The probability of NP can be determined using the nomogram, as illustrated in Fig. 3.

Figure 3. Nomogram prediction of pancreatic necrosis. The value of each variable was scored on a point scale from 0 to 100, after which the scores for each variable were added together. That sum is located on the total points axis, which enables us to predict the probability of necrotizing pancreatitis risk. PCT, procalcitonin; TG, triglyceride; WBC, white blood cell; HCT, hematocrit; Ca, serum calcium; 48 hr, at admission 48th hour.

4. Validation of the predictive accuracy and clinical value of nomogram model

We assessed the accuracy of the nomogram model, which was further validated internally using bootstrap sampling. After 500 bootstrap replications to reduce overfitting, the AUC for the cohort was found to be 0.822 (95% confidence interval [CI], 0.737 to 0.892); the calibration curve demonstrated good agreement between predicted and observed probabilities (H-L test: χ2=12.897, p=0.167). Following 500 bootstrap samples, DCA indicated that at a threshold probability for NP of >8%, the nomogram model provided a positive net benefit over both the treat-all or treat-none strategies (Fig. 4).

Figure 4. (A) Receiver operating characteristic curve for the nomogram was measured by bootstrapping for 500 repetitions. (B) Calibration curve using bootstraps sampling 500 for predicted probability of the nomogram. When the gray line (performance nomogram) was closer to the dotted line (ideal model), the prediction accuracy of the nomogram was better. (C) Decision curve analysis using bootstraps sampling 500 for the prediction model. The dotted line is from the prediction model, the line is for all patients with necrotizing pancreatitis (NP), and the solid horizontal line indicates no patients have NP.

5. Predictive performances and clinical value of Ranson, APACHE-II, BISAP, and nomogram model

ROCs of Ranson, APACHE-II, BISAP, and the nomogram model were depicted in Fig. 5. An ROC curve analysis of the nomogram model yielded an AUC of 0.822 (95% CI, 0.748 to 0.897), indicating good predictive performance with a sensitivity of 82.8% and a specificity of 76.4% at the optimal cutoff value of 0.260. The model's AUC was significantly higher than that of BISAP (AUC, 0.686; 95% CI, 0.612 to 0.761; p<0.001), APACHE-II (AUC, 0.664; 95% CI, 0.578 to 0.749; p<0.001), and Ranson (AUC, 0.666; 95% CI, 0.586 to 0.746; p<0.001) (Fig. 5).

Figure 5. (A) Receiver operating characteristic curves for the nomogram and other existing clinical scoring systems. (B) Decision curve analysis for the nomogram and other existing clinical scoring systems. APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis.

To ascertain the superiority in prediction of NP, we employed the IDI and NRI to compare the discriminative abilities of the nomogram model against other clinical scoring systems. Our findings suggest that the nomogram model exhibits superior accuracy in predicting NP, indicating its potential as a valuable clinical tool (Table 3).

Table 3 . Predictive Performance of the Integrated Discrimination Improvement between the Model and Other Clinical Scores.

VariableIDI (95% CI)p-valueNRI (95% CI)p-value
Model-APACHE-II0.243 (0.165–0.321)<0.0010.989 (0.718–1.260)<0.001
Model-BISAP0.251 (0.173–0.329)<0.0010.890 (0.620–1.160)<0.001
Model-Ranson0.274 (0.200–0.349)<0.0011.061 (0.802–1.320)<0.001

IDI, integrated discrimination improvement; CI, confidence interval; NRI, net reclassification index; APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis..



The DCA curves for Ranson, APACHE-II, BISAP, and nomogram model are presented in Fig. 5. The nomogram model consistently demonstrated greater net benefit than Ranson, APACHE-II, and BISAP in predicting NP with a threshold probability of >8%.

6. Predictive performances of the nomogram model in hyperlipidemic pancreatitis and no-hyperlipidemic pancreatitis

ROC curve analysis of the nomogram model for hyperlipidemic and no-hyperlipidemic pancreatitis yielded AUCs of 0.795 (95% CI, 0.698 to 0.891) and 0.838 (95% CI, 0.706 to 0.971).

DISCUSSION

Early diagnosis of NP is crucial for the management of AP. We developed a predictive nomogram for early-stage NP based on multivariate analyses, which may mitigate some adverse effects associated with CECT. The model suggests that higher levels of PCT and TG at admission, as well as elevated WBC, calcium, and BUN at 48 hours post-admission, increase the risk of NP development. This nomogram could serve as a practical tool for predicting NP likelihood in AP patients, utilizing readily available clinical parameters. The nomogram exhibited excellent diagnostic performance (AUC 0.822, sensitivity 82.8%, and specificity 76.4%) and was internally validated using bootstrap sampling. Moreover, its clinical utility was confirmed by DCA, indicating its effectiveness in a clinical environment.

Our study found that NP was associated with a higher incidence of organ failure, longer hospital stays, and increased intensive care unit admissions compared to edematous AP, aligning with findings from Ugurlu et al.11,12 These studies also indicated higher mortality rates for necrotizing AP,11,12 a trend not observed in our data. This discrepancy may stem from our exclusion of patients with incomplete data at 48 hours and without CECT, potentially omitting those with severe conditions leading to early discharge or those unable to undergo CECT. Consequently, the NP group in our study had lower APACHE-II scores. A retrospective study from China recognized the APACHE-II score as a significant mortality risk factor in NP.13 Alternatively, this may be due to early and aggressive care that includes vigilant monitoring of inflammation markers and evaluating patients within the first 72 to 96 hours using CECT.

AP is characterized by non-infected inflammation of pancreatic tissue. The development of pancreatic necrosis is closely related to both the extent of inflammation and the disease severity. Consequently, higher WBC levels, a biomarker of systemic inflammation, are anticipated in NP. Acehan et al.12 demonstrated that WBC levels are independent predictors of pancreatic necrosis. Similarly, a study by Unal and Barlas14 found that WBC levels were significantly higher in NP patients than those with edematous AP. Our findings reinforce this relationship, showing that WBC levels at the 48th hour of admission are associated with an increased odds ratio of 1.153 for NP development.

PCT, a propeptide secreted by hepatocytes and the thyroid in response to severe inflammation and sepsis,1,6 serves as a biomarker in distinguishing necrotizing from edematous pancreatitis.15 Elevated PCT levels have been recognized as early indicators of disease severity.6 A recent retrospective study of 582 AP patients in Turkey found PCT to be an early predictor of NP as suggested by ROC analysis.11 Echoing these results, our study identifies PCT at the 48th hour post-admission as an independent predictor of necrosis, increasing the risk of necrosis development by a factor of 1.97.

NP is linked to third-space fluid loss, leading to hypoperfusion, splanchnic vasoconstriction, and reduced microcirculation in the pancreas.16 Earlier studies17,18 have shown that early fluid resuscitation can diminish the risk of NP, attenuate systemic inflammatory response, prevent multi-organ failure, and maintain pancreatic microcirculation. Hence, hemoconcentration resulting from fluid loss is likely associated with the severity of AP. In a multicenter, prospective study conducted in the United States and the Netherlands, HCT was identified as a potential predictor of NP, with an HCT level at admission ≥44% being associated with an odds ratio of 3.11 for NP development.19 Recent findings by Hidalgo et al.20 also confirmed that HCT at admission correlates with NP development. Moreover, clinical studies have demonstrated that HCT levels at the 48th hour are independent risk factors for necrosis in AP.12 In line with these studies, our research indicates that higher HCT levels at the 48th hour of admission are indicative of an increased risk for necrosis development in AP.

Fatty acids were hydrolyzed from excess TG by pancreatic lipase, which can result in inflammation of the pancreas and intracellular calcium influx, subsequently causing pancreatic necrosis.21 Current evidence suggests that patients with elevated TG levels exhibit a more severe form of pancreatitis. Mosztbacher et al.22 demonstrated that TG levels dose-dependently exacerbate the severity and complications of AP. A prospective observational study in Spain, which focused on a single cohort, revealed that TG levels in the early stages of AP are linked to an increased risk of developing NP.20 Cheng et al.23 found that patients with acute biliary pancreatitis and high TG levels tend to develop more extensive necrosis. Similarly, our research corroborated that elevated TG constitutes an independent risk factor for the development of necrosis in AP.

Hypocalcemia in AP is attributed to the release of free fatty acids, which may lead to the formation of calcium salts and impairment of parathyroid function.24 Yu et al.25 found that patients with hypocalcemia in hyperlipidemic pancreatitis face a higher risk of severe pancreatitis. Another study indicated that serum calcium levels elevate the risk of necrosis development by 0.021-fold at values above 2.04 mmol/L.26 Our research suggests that hypocalcemia at the 48-hour mark of admission is an independent risk factor for the development of NP.

As highlighted in the introduction, common AP scoring systems, such as APACHE-II, Ranson, and BISAP, are limited by the requirement for CT imaging or complex calculations. Furthermore, these systems have not significantly predicted necrosis development.27 Consequently, there is a need for an accurate model that utilizes a small number of variables without radiological findings for clinical application. The nomogram described in our study requires only five parameters (PCT, TG, WBC at 48 hours, Ca at 48 hours, and HCT at 48 hours) to predict NP development, which are widely used and easily computed. Thus, our risk assessment model could has the potential for broad acceptance. Moreover, it demonstrated greater predictive efficacy and was more beneficial in clinical decision-making for patients with NP than the other three AP scoring systems. Despite the limited sample size of our study, which may affect the external validity of the model, we performed internal validation using bootstrap sampling to affirm the nomogram's predictive accuracy and clinical utility.

It is crucial to recognize several limitations of this study. Firstly, its single-center, retrospective design may introduce selection bias. Secondly, although the ROC curve analysis yielded AUCs of 0.795 (95% CI, 0.698 to 0.891) for hyperlipidemic pancreatitis and 0.838 (95% CI, 0.706 to 0.971) for non-hyperlipidemic pancreatitis, hyperlipidemia constitutes over half the etiologies in our cohort, as opposed to typical AP cohorts where alcohol and gallstones predominate. Consequently, the generalizability of our nomogram to all AP cases may be limited. Thirdly, with only three factors (WBC, Ca, and HCT at 48 hours), the nomogram cannot predict necrotizing AP prior to 48 hours post-admission. A multicenter, prospective trial is required to confirm the model's accuracy.

In conclusion, our study presents a nomogram model designed to calculate a risk score and identify patients at an increased likelihood of early NP development. The application of this model as a convenient and specific tool may prove advantageous for clinical decision-making.

CONFLICTS OF INTEREST

No potential conflict of interest relevant to this article was reported.

AUTHOR CONTRIBUTIONS

Study concept and design: J.Z. Data acquisition: J.Z. Data analysis and interpretation: J.Z. Drafting of the manuscript: J.Z. Critical revision of the manuscript for important intellectual content: X.W. Statistical analysis: J.Z. Administrative, technical, or material support; study supervision: X.W. Approval of final manuscript: all authors.

Fig 1.

Figure 1.Flowchart of the process of patient enrollment. CECT, contrast-enhanced computed tomography; NP, necrotizing pancreatitis.
Gut and Liver 2024; :

Fig 2.

Figure 2.Features selection using the LASSO binary logistic regression model. (A) Log (lambda) value of the 28 features in the LASSO model: a coefficient profile plot was produced against the log (lambda) sequence. (B) Parameter selection in the LASSO model used tenfold cross-validation via minimum criterion: partial likelihood deviation (binomial deviation) curves and logarithmic (lambda) curves were plotted. Use the minimum standard and 1 se (1-SE standard) of the minimum standard to draw a vertical dashed line at the optimal value. The optimal lambda produced four nonzero coefficients. LASSO, least absolute shrinkage and selection operator; SE, standard error.
Gut and Liver 2024; :

Fig 3.

Figure 3.Nomogram prediction of pancreatic necrosis. The value of each variable was scored on a point scale from 0 to 100, after which the scores for each variable were added together. That sum is located on the total points axis, which enables us to predict the probability of necrotizing pancreatitis risk. PCT, procalcitonin; TG, triglyceride; WBC, white blood cell; HCT, hematocrit; Ca, serum calcium; 48 hr, at admission 48th hour.
Gut and Liver 2024; :

Fig 4.

Figure 4.(A) Receiver operating characteristic curve for the nomogram was measured by bootstrapping for 500 repetitions. (B) Calibration curve using bootstraps sampling 500 for predicted probability of the nomogram. When the gray line (performance nomogram) was closer to the dotted line (ideal model), the prediction accuracy of the nomogram was better. (C) Decision curve analysis using bootstraps sampling 500 for the prediction model. The dotted line is from the prediction model, the line is for all patients with necrotizing pancreatitis (NP), and the solid horizontal line indicates no patients have NP.
Gut and Liver 2024; :

Fig 5.

Figure 5.(A) Receiver operating characteristic curves for the nomogram and other existing clinical scoring systems. (B) Decision curve analysis for the nomogram and other existing clinical scoring systems. APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis.
Gut and Liver 2024; :

Table 1 Characteristics of Patients in the Non-NP and NP Groups

Patient characteristicsAll (n=218)Non-NP (n=163)NP (n=55)p-value
Demographics and comorbidities
Male sex171 (78.4)124 (76.1)47 (85.5)0.203
Age, yr40.0 (32.0–48.0)40.0 (32.0–49.0)39.0 (31.5–47.0)0.531
Etiology0.003
Hyperlipidemia117 (53.7)76 (46.6)41 (74.5)<0.001
Gallstone33 (15.1)30 (18.4)3 (5.45)0.020
Alcohol16 (7.34)14 (8.59)2 (3.64)0.223
Others52 (23.9)43 (26.4)9 (16.4)0.132
Smoker85 (39.0)60 (36.8)25 (45.5)0.329
Drinker80 (36.7)54 (33.1)26 (47.3)0.085
Diabetes86 (39.4)51 (31.3)35 (63.6)<0.001
Hypertension41 (18.8)27 (16.6)14 (25.5)0.208
SIRS89 (40.8)50 (30.7)39 (70.9)<0.001
Organ failure56 (25.7)35 (21.5)21 (38.2)0.023
Laboratory data
WBC, ×109/L13.3 (10.9–16.2)12.5 (10.4–15.6)15.3 (12.4–17.8)0.004
HCT, %44.3 (40.1–46.6)44.1 (40.1–46.6)44.7 (40.8–46.3)0.465
CRP, mg/dL0.50 (0.10–2.70)0.37 (0.06–2.24)1.06 (0.21–3.71)0.025
PCT, ng/ml0.06 (0.05–0.16)0.05 (0.05–0.12)0.14 (0.05–0.60)<0.001
LDH, IU/L204 (176–258)202 (175–248)210 (186–293)0.335
BUN, mmol/L4.00 (3.30–4.90)4.00 (3.30–5.00)4.10 (3.35–4.60)0.708
Creatinine, mmol/L73.8 (65.9–83.5)73.9 (66.3–83.2)73.7 (65.6–87.3)0.544
Ca, mmol/L2.33 (2.23–2.42)2.33 (2.23–2.41)2.33 (2.21–2.46)0.778
Glucose, mmol/L8.00 (6.30–12.10)7.80 (6.10–11.20)10.9 (7.95–14.80)<0.001
Total bilirubin, mmol/L16.7 (13.1–24.4)16.7 (13.2–24.6)16.7 (11.6–23.5)0.792
ALT, IU/L30.0 (19.2–51.0)31.0 (20.0–53.5)28.0 (18.5–44.0)0.288
AST, IU/L25.5 (20.0–43.0)24.0 (19.0–43.0)31.0 (21.5–43.5)0.197
Triglyceride, mmol/l5.76 (1.32–15.50)4.25 (1.22–10.10)14.3 (4.09–34.70)<0.001
WBC 48 hr, ×109/L9.80 (7.70–12.70)9.40 (7.40–12.20)10.80 (8.05–14.40)0.029
HCT 48 hr, %40.7 (37.3–43.7)40.6 (37.1–43.2)41.1 (37.5–44.4)0.223
CRP 48 hr, mg/dL126 (59.4–208)103 (50.7–148)226 (153–303)<0.001
PCT 48 hr, ng/ml0.16 (0.05–0.46)0.10 (0.05–0.23)0.80 (0.38–1.86)<0.001
Ca 48 hr, mmol/L2.13 (2.03–2.22)2.15 (2.07–2.23)2.04 (1.88–2.16)<0.001
BUN 48 hr, mmol/L2.80 (2.10–3.48)2.70 (2.05–3.40)3.00 (2.55–4.00)0.008
Creatinine 48 hr, mmol/L70.3 (62.3–79.3)72.1 (62.5–79.6)68.1 (61.1–77.1)0.299
APACHEII3.00 (2.00–5.00)3.00 (1.00–4.00)4.00 (2.00–6.50)<0.001
Ranson1.00 (0.00–2.00)1.00 (0.00–2.00)2.00 (1.00–3.00)<0.001
BISAP1.00 (1.00–2.00)1.00 (1.00–2.00)2.00 (1.00–2.00)<0.001
Outcomes percutaneous catheter drainage2 (1.0)02 (3.6)0.063
ICU admission13 (6.0)1 (0.61)12 (21.8)<0.001
Hospital stays, day7.00 (5.00–8.75)6.00 (5.00–8.00)8.00 (7.00–13.00)<0.001
Mortality000
Percentage of pancreatic necrosis*--
<30%47 (21.6)47 (85.5)
30%–50%7 (3.2)7 (12.7)
>50%1 (0.5)1 (1.8)

Data are presented as number (%) or median (interquartile range).

NP, necrotizing pancreatitis; SIRS, systemic inflammatory response syndrome; WBC, white blood cell; HCT, hematocrit; CRP, C-reactive protein; PCT, procalcitonin; LDH, lactate dehydrogenase; BUN, blood urea nitrogen; Ca, serum calcium; ALT, alanine aminotransferase; AST, aspartate aminotransferase; 48 hr, at 48 hours post-admission; APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis; ICU, intensive care unit.

*Evaluated on the third day after admission.


Table 2 Association of Necrotizing Pancreatitis in Multivariable Analysis

VariableMultivariable regression
BOR (95% CI)p-value
Intercept0.3071.360 (0.005–435.8)0.916
PCT0.681.973 (1.285–3.826)0.008
TG0.0481.049 (1.025–1.075)<0.001
WBC 48 hr0.1421.153 (1.055–1.267)0.002
HCT 48 hr0.1131.120 (1.034–1.224)0.008
Ca 48 hr–4.0680.017 (0.001–0.161)0.001

OR, odds ratio; CI, confidence interval; PCT, procalcitonin; TG, triglyceride; WBC, white blood cell; HCT, hematocrit; Ca, serum calcium; 48 hr, at 48 hours post-admission.


Table 3 Predictive Performance of the Integrated Discrimination Improvement between the Model and Other Clinical Scores

VariableIDI (95% CI)p-valueNRI (95% CI)p-value
Model-APACHE-II0.243 (0.165–0.321)<0.0010.989 (0.718–1.260)<0.001
Model-BISAP0.251 (0.173–0.329)<0.0010.890 (0.620–1.160)<0.001
Model-Ranson0.274 (0.200–0.349)<0.0011.061 (0.802–1.320)<0.001

IDI, integrated discrimination improvement; CI, confidence interval; NRI, net reclassification index; APACHE-II, acute physiology and chronic health evaluation II; BISAP, bedside index for severity in acute pancreatitis.


References

  1. Asim Riaz HM, Islam Z, Rasheed L, et al. The evaluation of inflammatory biomarkers in predicting progression of acute pancreatitis to pancreatic necrosis: a diagnostic test accuracy review. Healthcare (Basel) 2022;11:27.
    Pubmed KoreaMed CrossRef
  2. Jha AK, Goenka MK, Kumar R, Suchismita A. Endotherapy for pancreatic necrosis: an update. JGH Open 2019;3:80-88.
    Pubmed KoreaMed CrossRef
  3. Colvin SD, Smith EN, Morgan DE, Porter KK. Acute pancreatitis: an update on the revised Atlanta classification. Abdom Radiol (NY) 2020;45:1222-1231.
    Pubmed CrossRef
  4. Kiss S, Pintér J, Molontay R, et al. Early prediction of acute necrotizing pancreatitis by artificial intelligence: a prospective cohort-analysis of 2387 cases. Sci Rep 2022;12:7827.
    Pubmed KoreaMed CrossRef
  5. Leonard-Murali S, Lezotte J, Kalu R, et al. Necrotizing pancreatitis: a review for the acute care surgeon. Am J Surg 2021;221:927-934.
    Pubmed KoreaMed CrossRef
  6. Khanna AK, Meher S, Prakash S, et al. Comparison of Ranson, Glasgow, MOSS, SIRS, BISAP, APACHE-II, CTSI scores, IL-6, CRP, and procalcitonin in predicting severity, organ failure, pancreatic necrosis, and mortality in acute pancreatitis. HPB Surg 2013;2013:367581.
    Pubmed KoreaMed CrossRef
  7. Leppäniemi A, Tolonen M, Tarasconi A, et al. 2019 WSES guidelines for the management of severe acute pancreatitis. World J Emerg Surg 2019;14:27.
    Pubmed KoreaMed CrossRef
  8. Moons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162:W1-W73.
    Pubmed CrossRef
  9. Banks PA, Bollen TL, Dervenis C, et al. Classification of acute pancreatitis: 2012. Revision of the Atlanta classification and definitions by international consensus. Gut 2013;62:102-111.
    Pubmed CrossRef
  10. Garg PK, Madan K, Pande GK, et al. Association of extent and infection of pancreatic necrosis with organ failure and death in acute necrotizing pancreatitis. Clin Gastroenterol Hepatol 2005;3:159-166.
    Pubmed CrossRef
  11. Ugurlu ET, Yurekli UF. The effect of procalcitonin and immature granulocyte ratio in predicting the development of acute necrotizing pancreatitis: evidence from 582 cases. Eur Rev Med Pharmacol Sci 2022;26:7514-7521.
    Pubmed CrossRef
  12. Acehan F, Comoglu M, Kayserili FM, Hayat B, Ates I. Factors predicting the development of necrosis in patients presenting with edematous acute pancreatitis. Pancreas 2022;51:1300-1307.
    Pubmed CrossRef
  13. Yu L, Xie F, Luo L, et al. Clinical characteristics and risk factors of organ failure and death in necrotizing pancreatitis. BMC Gastroenterol 2023;23:19.
    Pubmed KoreaMed CrossRef
  14. Ünal Y, Barlas AM. Role of increased immature granulocyte percentage in the early prediction of acute necrotizing pancreatitis. Ulus Travma Acil Cerrahi Derg 2019;25:177-182.
    Pubmed CrossRef
  15. Silva-Vaz P, Abrantes AM, Castelo-Branco M, Gouveia A, Botelho MF, Tralhão JG. Multifactorial scores and biomarkers of prognosis of acute pancreatitis: applications to research and practice. Int J Mol Sci 2020;21:338.
    Pubmed KoreaMed CrossRef
  16. Di Martino M, Van Laarhoven S, Ielpo B, et al. Systematic review and meta-analysis of fluid therapy protocols in acute pancreatitis: type, rate and route. HPB (Oxford) 2021;23:1629-1638.
    Pubmed CrossRef
  17. Warndorf MG, Kurtzman JT, Bartel MJ, et al. Early fluid resuscitation reduces morbidity among patients with acute pancreatitis. Clin Gastroenterol Hepatol 2011;9:705-709.
    Pubmed KoreaMed CrossRef
  18. Wu BU, Johannes RS, Sun X, Conwell DL, Banks PA. Early changes in blood urea nitrogen predict mortality in acute pancreatitis. Gastroenterology 2009;137:129-135.
    Pubmed CrossRef
  19. Koutroumpakis E, Wu BU, Bakker OJ, et al. Admission hematocrit and rise in blood urea nitrogen at 24 h outperform other laboratory markers in predicting persistent organ failure and pancreatic necrosis in acute pancreatitis: a post hoc analysis of three large prospective databases. Am J Gastroenterol 2015;110:1707-1716.
    Pubmed CrossRef
  20. Hidalgo NJ, Pando E, Alberti P, et al. The role of high serum triglyceride levels on pancreatic necrosis development and related complications. BMC Gastroenterol 2023;23:51.
    Pubmed KoreaMed CrossRef
  21. Yang AL, McNabb-Baltar J. Hypertriglyceridemia and acute pancreatitis. Pancreatology 2020;20:795-800.
    Pubmed CrossRef
  22. Mosztbacher D, Hanák L, Farkas N, et al. Hypertriglyceridemia-induced acute pancreatitis: a prospective, multicenter, international cohort analysis of 716 acute pancreatitis cases. Pancreatology 2020;20:608-616.
    Pubmed CrossRef
  23. Cheng L, Luo Z, Xiang K, et al. Clinical significance of serum triglyceride elevation at early stage of acute biliary pancreatitis. BMC Gastroenterol 2015;15:19.
    Pubmed KoreaMed CrossRef
  24. Dong X, Pan S, Zhang D, et al. Hyperlipemia pancreatitis onset time affects the association between elevated serum triglyceride levels and disease severity. Lipids Health Dis 2022;21:49.
    Pubmed KoreaMed CrossRef
  25. Yu S, Wu D, Jin K, et al. Low serum ionized calcium, elevated high-sensitivity C-reactive protein, neutrophil-lymphocyte ratio, and body mass index (BMI) are risk factors for severe acute pancreatitis in patients with hypertriglyceridemia pancreatitis. Med Sci Monit 2019;25:6097-6103.
    Pubmed KoreaMed CrossRef
  26. Xue E, Shi Q, Guo S, et al. Preexisting diabetes, serum calcium and D-dimer levels as predictable risk factors for pancreatic necrosis of patients with acute pancreatitis: a retrospective study. Expert Rev Gastroenterol Hepatol 2022;16:913-921.
    Pubmed CrossRef
  27. Hagjer S, Kumar N. Evaluation of the BISAP scoring system in prognostication of acute pancreatitis: a prospective observational study. Int J Surg 2018;54:76-81.
    Pubmed CrossRef
Gut and Liver

Vol.18 No.2
March, 2024

pISSN 1976-2283
eISSN 2005-1212

qrcode
qrcode

Share this article on :

  • line

Popular Keywords

Gut and LiverQR code Download
qr-code

Editorial Office