Computed tomography features of recurrent acute pancreatitis based on different etiologies

Abstract

Background

Acute pancreatitis (AP), recurrent acute pancreatitis (RAP), and chronic pancreatitis (CP) are increasingly viewed as stages in a continuous disease spectrum when the underlying etiology remains unresolved. Previous studies have investigated the effect of different etiologies on AP severity, but few have specifically examined the clinical and radiologic characteristics of RAP stratified by etiology. This study aimed to investigate the computed tomography (CT) features of RAP stratified by etiology.

Methods

We retrospectively analyzed the data of 683 RAP patients who underwent contrast-enhanced computed tomography (CECT) at three tertiary hospitals between January 2015 and December 2019. The patients were categorized into five etiologic groups: alcoholic, cholelithiasis-related, hypertriglyceridemia-related, multifactorial, and idiopathic. Clinical and imaging data, including demographic data, 2012 revised Atlanta classification (RAC), Acute Physiology and Chronic Health Evaluation (APACHE) II scores, modified computed tomography severity index (MCTSI) scores, extrapancreatic inflammation on computed tomography (EPIC) scores, the presence and extent of pancreatic necrosis, and local complications, were compared across groups using Kruskal-Wallis and Chi-squared (or Fisher’s exact) tests, with Bonferroni correction for multiple comparisons.

Results

Among the 683 patients {449 male, 234 female; median age: 45 years [interquartile range (IQR), 39–52 years]; median hospital stay: 11 days (IQR, 7–15 days)} included in the study, hypertriglyceridemia was the most common cause of RAP (50.95%), followed by idiopathic causes (18.74%), cholelithiasis (15.37%), alcoholism (9.37%), and multiple causes (5.56%). Patients with hypertriglyceridemic or multifactorial RAP had higher triglyceride [median (IQR): 17.99 (12.12–25.77) vs. 1.33 (0.86–2.00) mmol/L] and blood glucose [median (IQR): 10.55 (7.44–14.33) vs. 7.43 (5.74–10.12) mmol/L] levels, as well as a higher prevalence of pre-existing diabetes (30.46% vs. 9.52%) (all P<0.001). Compared to the patients with biliary RAP, those with hyperlipidemic RAP exhibited milder CT features, including lower rates of necrotizing pancreatitis (NP) (33.05% vs. 48.57%), combined necrosis (CN) (25.57% vs. 40.00%), pancreatic necrosis >50% (3.74% vs. 14.29%), and severe disease (23.85% vs. 40.00%) as defined by the MCTSI (all P<0.05). These patients also had lower EPIC scores [median (IQR): 3 (2–5) vs. 4 (2–5.5)] and shorter hospital stays [median (IQR): 10 (7–15) vs. 13 (9.5–18) days] (both P<0.05). No significant differences in the severity indicators were observed among alcoholic, multifactorial, and idiopathic groups. Alcohol-related RAP occurred predominantly in males (96.9%), while biliary RAP was more frequent in older female patients (both P<0.05).

Conclusions

RAP presents with distinct clinical and CT features depending on its etiology. Hypertriglyceridemic RAP is associated with milder disease severity than biliary RAP. Our findings provide new insights into the etiology-specific manifestations of RAP, and may inform future research on its underlying mechanisms and long-term outcomes.

Introduction

Acute pancreatitis (AP) is a common cause of acute abdominal pain. Recurrent acute pancreatitis (RAP) is defined as two or more well-documented episodes of pancreatitis, separated by more than 3 months of complete clinical and biochemical remission, and no evidence of chronic pancreatitis (CP) on imaging or clinical grounds (1,2). The estimated annual incidence and prevalence of RAP are approximately 8–10 and 110–140 per 100,000 people worldwide, respectively, with a reported mortality rate of 0.9% (2-5).

AP, RAP, and CP are increasingly viewed as stages along a continuous disease spectrum when the underlying etiology remains unresolved (4,6). According to Guda et al. (7), failure to identify and manage causative factors can lead to recurrence or progression to CP. Conversely, Li et al. (8) showed that targeted treatment strategies initiated after an acute episode can significantly reduce the risk of recurrence. Therefore, characterizing the clinical and radiologic features of RAP across different etiologies may improve our understanding of its heterogeneous nature, and support future investigations into disease mechanisms and prognosis.

Studies have investigated how various etiologies affect the severity of AP. For instance, hypertriglyceridemia-induced AP has been associated with more severe disease courses (9-11). Huang et al. (12) reported that alcoholic AP was more severe than biliary AP, while Du et al. (13) found no significant differences in AP severity between etiologic subgroups. Although these studies enrolled large cohorts (n=5,375 and n=1,924, respectively), RAP was largely treated as a subgroup rather than a distinct entity. However, emerging evidence suggests that RAP may differ from initial AP in terms of its underlying mechanisms, natural history, and radiologic manifestations (1,2,14-16). These differences underscore the need for dedicated studies specifically focused on RAP as a distinct clinical entity.

Computed tomography (CT), particularly contrast-enhanced computed tomography (CECT), plays a central role in the diagnosis and severity assessment of AP and its complications, as emphasized in the 2012 revised Atlanta classification (RAC) (17,18). Guda et al. (2) recommends obtaining CT scans 48–72 hours after symptom onset in RAP cases to evaluate pancreatic necrosis in patients with severe clinical presentation. Among the most widely used CT-based scoring tools are the modified computed tomography severity index (MCTSI) and the extrapancreatic inflammation on computed tomography (EPIC) score, both of which provide standardized metrics for assessing the severity of AP (19,20). Despite the well-established use of these imaging criteria in AP, their application in RAP, particularly when stratified by etiology, has not been systematically explored.

To address this gap, we conducted a retrospective study involving three tertiary hospitals to evaluate the CT characteristics of RAP across different etiologies. Our aim was to compare demographic features, types of RAP, patterns of necrosis, local complications, and CT-based severity scores between etiologic groups. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-2260/rc).

Methods

Patients and study design

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Ethical approval was obtained from the Sichuan Provincial People’s Hospital, University of Electronic Science and Technology of China [No. 2024ER(R)133]. All participating hospitals were informed of the study and agreed to the use of shared data. The requirement of informed consent was waived due to the retrospective nature of the study.

This retrospective study was conducted at three tertiary hospitals in Sichuan Province, China (Affiliated Hospital of North Sichuan Medical College, The General Hospital of Western Theater Command, and Suining Central Hospital). The study period spanned from January 2015 to December 2019. All patients were managed according to the guidelines for AP (21-23). Indications for CT scans in patients with RAP were based on consensus guidelines for both RAP and AP (2,24,25). A total of 808 consecutive patients were identified through the International Classification of Diseases codes in the hospital information system. RAP diagnosis required two or more distinctly documented episodes of AP, each followed by complete resolution lasting over 3 months, and an absence of morphological criteria for CP (1,2). The diagnosis of AP required the fulfillment of two of the following three criteria: (I) acute, severe epigastric pain, often radiating to the back; (II) serum lipase (or amylase) activity ≥3 times the upper limit of normal level; and/or (III) characteristic imaging findings consistent with AP (18).

The criteria for determining the etiology of RAP were as follows: (I) alcoholism, characterized by daily alcohol consumption exceeding 60 grams for more than 5 years (26); (II) cholelithiasis, confirmed by imaging evidence of gallstones in the gallbladder and/or bile ducts (27); (III) hypertriglyceridemia, indicated by a triglyceride level exceeding 11.3 mmol/L at hospital admission, or triglyceride levels fluctuating within the range of 5.65 to 11.3 mmol/L, ruling out other causes (28); (IV) multiple etiologies, involving two or more causative factors (2,29); and (V) idiopathic causes, diagnosed when no clear etiology could be established despite comprehensive clinical, laboratory, and imaging evaluations.

According to the World Health Organization, the following age classification criteria were employed: young adults (ages 18–44 years), middle-aged adults (ages 45–64 years), and older adults (≥65 years) (30,31).

Patients were included in the study if they met the following inclusion criteria: (I) were aged ≥18 years; (II) had undergone CECT scans due to RAP symptoms; (III) had undergone laboratory tests and CT scans within a maximum interval of 3 days, either before or after each other; and (IV) had undergone multiple scans during a single RAP episode; only the scan showing the most severe findings was included. Notably, for patients with multiple episodes of RAP, each episode was included as a separate event, provided that all the above criteria (I–IV) were met.

Patients were excluded from the study if they met any of the following exclusion criteria: (I) had acute exacerbation of CP (n=38); (II) had concurrent malignancies or severe chronic wasting disorders (n=25); (III) were pregnant (n=23); and/or (IV) had inadequate CECT imaging quality or incomplete medical records (n=39).

Among 808 patients diagnosed with RAP, 631 patients with 683 episodes who underwent CECT scans were included in the final analysis. Based on etiology, these patients were classified into five groups: alcoholic (n=64), cholelithiasis-related (n=105), hypertriglyceridemia-related (n=348), multifactorial (n=38), and idiopathic (n=128). The cohort comprised 587 patients with a single RAP episode, and 44 patients with multiple RAP episodes. Among the latter, eight experienced three episodes and 36 patients experienced two episodes, resulting in a total of 96 recurrent events. A detailed flowchart of patient selection and grouping is presented in Figure 1.

Figure 1.

Flowchart illustrating patient recruitment in the study. CECT, contrast-enhanced computed tomography; CP, chronic pancreatitis; n, number of patients; RAP, recurrent acute pancreatitis.

CT technology

During hospitalization, all patients underwent at least one abdominal CECT scan and received appropriate treatment for a single RAP episode. Imaging was performed using 1 of 5 multi-detector CT scanners: SOMATOM Definition AS+ 128 (Siemens Healthcare, Forchheim, Germany), LightSpeed VCT 128 (GE Healthcare, Waukesha, WI, USA), Brilliance 64 (Philips Healthcare, Best, Netherlands), Toshiba Aquilion ONE 320 (Toshiba Medical Systems, Otawara, Japan), and SOMATOM Definition Flash (Siemens Healthcare, Forchheim, Germany). Detailed CT acquisition protocols are provided in Appendix 1 and Table S1.

Image analysis

All the CT imaging data were retrieved from the picture archiving and communication system. Two experienced radiologists (the first and second authors, with 6 and 13 years of experience in abdominal radiology, respectively), who were blinded to the clinical data, independently reviewed the CT images.

Under the RAC system (18), each episode of RAP was classified as either interstitial edematous pancreatitis (IEP) or necrotizing pancreatitis (NP). NP was identified based on early heterogeneous pancreatic perfusion or non-uniform enhancement following contrast administration. Areas of impaired perfusion were considered indicative of evolving necrosis. Non-enhancement of pancreatic parenchyma, typically evident one week after symptom onset, was considered diagnostic of necrosis (18). NP was further categorized into three subtypes: extrapancreatic necrosis (EXPN) alone, pancreatic parenchymal necrosis (PN) alone, and combined necrosis (CN). The extent of PN was quantified using Balthazar’s CT grading system as <30%, 30–50%, or >50% (32). Radiologists assessed each CT image to determine the subtype of pancreatitis, necrosis classification, presence of local complications, MCTSI score (19), and EPIC score of (20) for each RAP patient. The scoring criteria for the MCTSI and EPIC are detailed in Tables S2,S3, respectively. Local complications included acute peripancreatic fluid collections (APFCs), acute necrotic collections (ANCs), pancreatic pseudocysts (PPCs), and walled-off necrosis (WON). Under the MCTSI, severity was categorized as mild (a score of 0–2), moderate (a score of 4–6), or severe (a score of 8–10). Odd-numbered scores (such as 3 or 7) do not occur because each component of the scoring system contributes only even-numbered points (19). Any discrepancies between readers were resolved by consensus discussion.

Laboratory and clinical data

The medical records of the 683 enrolled patients were reviewed to collect demographic and clinical information, including age, sex, triglyceride and blood glucose levels, hospital length of stay, pre-existing diabetes (yes/no), and Acute Physiology and Chronic Health Evaluation (APACHE) II scores. The severity of each RAP episode was assessed according to the RAC, which integrates clinical findings and imaging results to categorize cases as mild, moderate, or severe (18).

Interobserver reliability of image analysis

Interobserver agreement between the two radiologists was assessed using intraclass correlation coefficients (ICCs) for the MCTSI and EPIC scores, with an ICC >0.75 indicative of good agreement. Cohen’s kappa coefficients were calculated for the presence of pancreatic necrosis, necrosis type, and local complications. Weighted kappa statistics were applied for the categorical classification of RAP severity according to the RAC. A κ kappa value greater than 0.80 indicated near-perfect agreement.

Statistical analysis

The Kolmogorov-Smirnov (K-S) test was used to assess the distribution of the continuous variables. The non-normally distributed variables, including age, triglyceride level, blood glucose level, hospital stay, APACHE II score, MCTSI score, and EPIC score, are summarized as the median with the interquartile range (IQR), and were compared using the Kruskal-Wallis test. The categorical and ordinal variables [sex, pre-existing diabetes, presence of necrosis (yes/no), local complications, MCTSI grade, and RAC] are presented as the count and percentage. The Chi-squared test or Fisher’s exact test was used for group comparisons as appropriate. The Bonferroni correction was applied to adjust for type I error in multiple pairwise comparisons.

A Spearman’s rank correlation analysis was conducted to examine associations among the MCTSI, EPIC, and APACHE II scores with the RAC. Correlation strength was interpreted as follows: 0.00–0.29 (negligible), 0.30–0.49 (weak), 0.50–0.69 (moderate), 0.70–0.89 (strong), and 0.90–1.00 (very strong) (33).

Multicollinearity among independent variables, including age, sex, etiology, pre-existing diabetes, APACHE II score, MCTSI score, and EPIC score, was assessed using the variance inflation factor (VIF) and tolerance statistics. A VIF <5 and tolerance >0.2 indicated no significant multicollinearity.

Stepwise binary logistic regression using the forward likelihood ratio method was performed to identify independent predictors of severe RAP, as defined by the RAC system. Variables with P values <0.10 in the univariate analysis, along with etiology, were entered into the multivariate model. The odds ratio (OR) and the corresponding 95% confidence interval (CI) were calculated. Model fit was evaluated at each step using the likelihood ratio test.

All the statistical analyses were performed using SPSS 25.0 (IBM Corp., Armonk, NY, USA). A two-sided P value <0.05 was considered statistically significant.

Results

Interobserver agreement

Interobserver agreement was excellent for both MCTSI (ICC =0.956, 95% CI: 0.949–0.962; P<0.001) and EPIC scores (ICC =0.988, 95% CI: 0.986–0.990; P<0.001). Cohen’s kappa coefficients indicated almost perfect agreement for NP (κ=0.923, 95% CI: 0.894–0.952; P<0.001) and local complications (κ=0.879, 95% CI: 0.855–0.903; P<0.001), substantial agreement for necrosis type (κ=0.800, 95% CI: 0.761–0.839; P<0.001), and moderate agreement for the extent of necrosis (κ=0.656, 95% CI: 0.603–0.709; P<0.001). The weighted kappa for the RAC was also high (κ=0.868, 95% CI: 0.840–0.896; P<0.001). Overall, these results demonstrated good-to-excellent interobserver agreement across all the evaluated parameters.

Demographic, clinical, and laboratory characteristics of RAP patients

The demographic and clinical characteristics of all 683 patients, as well as subgroup comparisons based on RAP etiology, are summarized in Table 1. Hypertriglyceridemia was the most common etiology, followed by idiopathic causes, cholelithiasis, alcoholism, and multiple causes. Among the 38 patients classified as having multiple causes, 35 (92.1%) had a combination of biliary and hyperlipidemic etiologies, while the remaining three (7.9%) had both alcoholic and hypertriglyceridemia-related etiologies.

Table 1. Demographic and clinical data of RAP patients based on different etiologies.

Variable	All (n=683)	Different etiologies					P
		Alcoholism (n=64)	Cholelithiasis (n=105)	Hypertriglyceridemia (n=348)	Multiple causes (n=38)	Idiopathic causes (n=128)
Sex							<0.001
Male	449 (65.74)	62 (96.88)	51 (48.57)‡	232 (66.67)‡,§	25 (65.79)‡,§	79 (61.72)‡,§
Female	234 (34.26)	2 (3.13)	54 (51.43)‡	116 (33.33)‡,§	13 (34.21)‡,§	49 (38.28)‡,§
TG (mmol/L)	8.13 [2.10, 19.63]	2.51 [1.49, 3.70]¶,||	1.33 [0.86, 2.00]¶,||	17.99 [12.12, 25.77]	18.39 [8.40, 23.77]	2.01 [1.37, 3.27]¶,||	<0.001
Blood glucose (mmol/L)	8.86 [6.47, 13.16]	7.60 [5.93, 11.07]¶,||	7.43 [5.74, 10.12]¶,||	10.55 [7.44, 14.33]	10.05 [7.49, 16.26]	7.30 [5.97, 9.84]¶,||	<0.001
Age (years)	45 [39, 52]	43 [36, 46.75]§	53 [43.5, 66]	44 [37.25, 50]§	46 [40.50, 51.25]	45 [40, 54]§	<0.001
Age rating							<0.001
Young (18–44 years)	322 (47.14)	38 (59.38)§	28 (26.67)	180 (51.72)§	14 (36.84)	62 (48.44)§
Middle-aged (45–64 years)	311 (45.53)	26 (40.63)	48 (45.71)	165 (47.41)	22 (57.89)	50 (39.06)
Elderly (>65 years)	50 (7.32)	0§,†	29 (27.62)||,†	3 (0.86)§,†	2 (5.26)§	16 (12.50)
Hospital stay (days)	11 [7, 15]	10 [7, 17]	13 [9.5, 18]	10 [7, 15]§	11 [6, 17.25]	10 [8, 15]	0.008
Pre-existing diabetes	160 (23.43)	9 (14.06)¶,||	10 (9.52)¶,||	106 (30.46)	15 (39.47)	20 (15.63)¶,||	<0.001

Data are presented as n (%) or median [interquartile range]. ^‡, compared to alcoholism-related RAP; ^§, compared to cholelithiasis-related RAP; ^¶, compared to hyperlipidemia-related RAP; ^||, compared to multifactorial RAP; ^†, compared to idiopathic RAP. n, number of patients; RAP, recurrent acute pancreatitis; TG, triglyceride.

Cholelithiasis was slightly more common in the female patients, while alcoholism, hypertriglyceridemia, multiple causes, and idiopathic causes accounted for a higher proportion of cases in the male patients. Notably, alcoholic pancreatitis had the highest proportion of male patients, accounting for approximately 96.88% of cases. No statistically significant differences in sex distribution were observed among the hypertriglyceridemia, multiple causes, and idiopathic etiology groups (P<0.001).

Patients with biliary RAP had a higher median age than those with alcoholic, hyperlipidemic, and idiopathic RAP; however, no statistically significant difference in age was observed among the latter three etiologies (P<0.001). Among all etiologies of RAP, elderly patients were most frequently affected by biliary RAP, followed by idiopathic RAP. (P<0.001). Conversely, among all etiologies of RAP, the proportion of young patients was lowest in the biliary and multifactorial groups. While no statistically significant differences were observed among the remaining etiologic groups in terms of age (P<0.001). The median hospital stay was shorter in the patients with hypertriglyceridemia-related RAP than those with cholelithiasis-related RAP, while no statistically significant differences in hospital stay were observed among the other etiologic groups (P=0.008). Patients in the hypertriglyceridemia-related and multifactorial RAP etiologic groups had higher triglyceride and blood glucose levels, as well as a greater prevalence of pre-existing diabetes, compared with the other etiologic groups. No statistically significant differences were found in these parameters between the hypertriglyceridemia-related and multifactorial etiologic groups, or among the remaining three etiologic groups (both P<0.001; Table 1).

CT findings of RAP by etiologies

Table 2 summarizes the distribution of pancreatitis types and the extent of necrosis across different RAP etiologies. The prevalence of NP was lower in patients with hyperlipidemic-related RAP than in those with biliary RAP (P=0.009). No statistically significant differences in the prevalence of NP were observed among the remaining three etiologic groups. Additionally, the prevalence of CN and >50% necrosis was lower in the hyperlipidemia-related group than the alcoholic and biliary groups (P=0.007 and P=0.003, respectively). No statistically significant differences in these parameters were observed among the alcoholic, biliary, multifactorial, and idiopathic groups.

Table 2. Types of RAP based on different etiologies.

Variable	All (n=683)	Different etiologies					P
		Alcoholism (n=64)	Cholelithiasis (n=105)	Hypertriglyceridemia (n=348)	Multiple causes (n=38)	Idiopathic causes (n=128)
IEP	421 (61.64)	32 (50.00)	54 (51.43)¶	233 (66.95)	26 (68.42)	76 (59.38)	0.009
NP	262 (38.36)	32 (50.00)	51 (48.57)¶	115 (33.05)	12 (31.58)	52 (40.63)
NP subgroup							0.007
PN	11 (1.61)	0	4 (3.81)	3 (0.86)	0	4 (3.13)
EXPN	46 (6.73)	3 (4.69)	5 (4.76)	23 (6.61)	4 (10.53)	11 (8.59)
CN	205 (30.01)	29 (45.31)¶	42 (40.00)¶	89 (25.57)	8 (21.05)	37 (28.91)
Degree of necrosis							0.003
<30%	152 (22.25)	19 (29.69)	30 (28.57)	70 (20.11)	6 (15.79)	27 (21.09)
30–50%	14 (2.05)	2 (3.13)	1 (0.95)	9 (2.59)	0	2 (1.56)
>50%	50 (7.32)	8 (12.50)¶	15 (14.29)¶	13 (3.74)	2 (5.26)	12 (9.38)

Data are presented as n (%). ^¶, compared to hyperlipidemia-related RAP. CN, combined necrosis; EXPN, extrapancreatic necrosis; IEP, interstitial edematous pancreatitis; n, number of patients; NP, necrotizing pancreatitis; PN, pancreatic necrosis; RAP, recurrent acute pancreatitis.

Local complications of RAP by etiologies

Figure 2 and Table S4 present the distribution of local complications among all RAP patients across different etiologies. Among the 683 RAP patients, no statistically significant differences were observed in the overall prevalence of complications across the five etiologies (P=0.406). The hyperlipidemic RAP group had the lowest prevalence of WON (P<0.001). Conversely, the occurrence of APFCs, ANCs, and PPCs did not differ significantly among the groups (P=0.003, P=0.249, and P=0.500, respectively).

Figure 2.

Local complications of recurrent acute pancreatitis on CT. ANCs, acute necrotic collections; APFCs, acute peripancreatic fluid collections; CT, computed tomography; n, number of patients; PPCs, pancreatic pseudocysts; WON, walled-off necrosis.

CT severity of RAP by etiologies

Table 3 presents the MCTSI and EPIC scores stratified by RAP etiology. The patients with hyperlipidemic RAP had lower EPIC scores than those with biliary RAP (P=0.038). In addition, the proportion of severe RAP based on the MCTSI grading was lower in the hyperlipidemic group than both the alcoholic and biliary groups (P=0.003); however, no such statistically significant difference was observed between the alcoholic and biliary groups. No statistically significant differences in the EPIC and MCTSI scores or severity grades were observed among the other etiologic groups (Figure 3).

Table 3. The severity of RAP on both CT and clinical scoring systems based on different etiologies.

Severity indicators	All (n=683)	Different etiologies
		Alcoholism (n=64)	Cholelithiasis (n=105)	Hypertriglyceridemia (n=348)	Multiple causes (n=38)	Idiopathic causes (n=128)	P
MCTSI score	6 [2, 8]	6 [2, 8]	6 [4, 8]	6 [2, 6]	6 [4, 6]	6 [2, 8]	0.022
MCTSI							0.003
Mild	175 (25.62)	17 (26.56)	20 (19.05)	97 (27.87)	8 (21.05)	33 (25.78)
Moderately severe	313 (45.83)	19 (29.69)||	43 (40.95)	168 (48.28)	23 (60.53)	60 (46.88)
Severe	195 (28.55)	28 (43.75)¶	42 (40.00)¶	83 (23.85)	7 (18.42)	35 (27.34)
EPIC score	3 [2, 5]	4 [2, 6]	4 [2, 5.5]¶	3 [2, 5]	3.5 [2, 5.25]	3 [2, 5]	0.038
APACHE II score	4 [2, 6]	3.5 [2, 5]§	5 [3, 7]	4 [2, 6]§	4 [3, 6.25]	4 [2, 6.75]§	0.002
RAC							0.334
Mild	211 (30.89)	17 (26.56)	34 (32.38)	110 (31.61)	6 (15.79)	44 (34.38)
Moderately severe	427 (62.52)	40 (62.50)	64 (60.95)	216 (62.07)	31 (81.58)	76 (59.38)
Severe	45 (6.59)	7 (10.94)	7 (6.67)	22 (6.32)	1 (2.63)	8 (6.25)

Data are presented as n (%) or median [interquartile range]. ^§, compared to cholelithiasis-related RAP; ^¶, compared to hyperlipidemia-related RAP; ^||, compared to multifactorial RAP. APACHE II, Acute Physiology and Chronic Health Evaluation II; CT, computed tomography; EPIC, extrapancreatic inflammation on computed tomography; MCTSI, modified computed tomography severity index; n, number of patients; RAC, 2012 revised Atlanta classification; RAP, recurrent acute pancreatitis.

Figure 3.

CT features of RAP of different etiologies. (A) In a 31-year-old man with alcoholic RAP, CT showed IEP with APFCs (as indicated by the thin white arrow). The patient had an APACHE II score of 2, moderately severe RAC, a MCTSI score of 4, and an EPIC score of 1. (B) In a 71-year-old man with biliary RAP, CT showed NP (*) with ANCs (as indicated by the thick white arrow). The patient had an APACHE II score of 7, moderately severe RAC, a MCTSI score of 8, and an EPIC score of 4. (C) In a 40-year-old man with hyperlipidemic RAP, CT showed IEP with APFCs (as indicated by the thin white arrow). The patient had an APACHE II score of 9, moderately severe RAC, a MCTSI score of 6, and an EPIC score of 4. (D) In a 28-year-old woman with RAP due to multiple etiologies, CT showed NP with ANCs (as indicated by the thick white arrow). The patient had an APACHE II score of 3, moderately severe RAC, a MCTSI score of 6, and an EPIC score of 4. (E) In a 51-year-old man with idiopathic RAP, CT showed IEP (as indicated by the thin white arrow). The patient had an APACHE II score of 2, mild RAC, a MCTSI score of 2, and an EPIC score of 0. ANCs, acute necrotic collections; APACHE II, Acute Physiology and Chronic Health Evaluation II; APFCs, acute peripancreatic fluid collections; CT, computed tomography; EPIC, extrapancreatic inflammation on computed tomography; IEP, interstitial edematous pancreatitis; MCTSI, modified CT severity index; NP, necrotizing pancreatitis; RAC, revised Atlanta classification; RAP, recurrent acute pancreatitis.

Clinical severity of RAP by etiologies based on scoring systems

Table 3 presents the APACHE II scores and RAC for all patients, stratified by RAP etiology. The patients with cholelithiasis-related RAP had higher APACHE II scores than those with alcoholic, hypertriglyceridemia-related and idiopathic RAP. However, statistically significant differences in APACHE II scores were only observed between cholelithiasis and these three etiologies (P=0.002). No statistically significant differences in the RAC were found among the different etiologic groups (P=0.334, Figure 3).

Correlation of MCTSI, EPIC, and APACHE II with the RAC by etiologies

In the overall cohort, the MCTSI, EPIC, and APACHE II scores were positively correlated with the RAC. The correlations for both the MCTSI and EPIC scores were moderate (both P<0.001), while the correlation for the APACHE II score was only weak (P<0.001).

In the subgroup analyses by etiology, the correlation between the MCTSI score and RAC was moderate in alcoholic and biliary RAP, strong in hyperlipidemic and idiopathic RAP (all P<0.001), and weak in multifactorial RAP (P=0.011). Similarly, the EPIC scores exhibited moderate correlations across all subgroups, except for the multifactorial etiology group (all P<0.001), where the association was weak (P=0.039). Conversely, the APACHE II scores were weakly correlated with the RAC in alcoholic and hypertriglyceridemic RAP (P=0.001 and P<0.001, respectively), but showed negligible or no correlation in the other subgroups (Table 4).

Table 4. Correlation of MCTSI, EPIC, and APACHE II with the RAC by etiologies.

Correlation pairs	All (n=683)	Different etiologies
		Alcoholism (n=64)	Cholelithiasis (n=105)	Hypertriglyceridemia (n=348)	Multiple causes (n=38)	Idiopathic causes (n=128)
MCTSI/RAC
r	0.674	0.685	0.626	0.702	0.407	0.702
P	0	0	0	0	0.011	0
EPIC/RAC
r	0.532	0.607	0.505	0.541	0.337	0.515
P	0	0	0	0	0.039	0
APACHE II/RAC
r	0.216	0.398	–0.05	0.302	0.076	0.181
P	0	0.001	0.611	0	0.652	0.041

APACHE II, Acute Physiology and Chronic Health Evaluation II; CT, computed tomography; EPIC, extrapancreatic inflammation on computed tomography; MCTSI, modified computed tomography severity index; n, number of patients; RAC, 2012 revised Atlanta classification.

Multicollinearity diagnostics and logistic regression results

The multicollinearity analysis showed that all the VIF values ranged from 1.007 to 1.945, and all the tolerance values exceeded 0.2, indicating no significant collinearity among the predictors (Table S5).

The stepwise logistic regression identified the EPIC score (OR =1.625; 95% CI: 1.340–1.971; P<0.001) and the APACHE II score (OR =1.145; 95% CI: 1.037–1.263; P=0.007) as independent predictors of severe RAP (Table 5). Other variables, including age, sex, etiology, pre-existing diabetes, and MCTSI score, were not retained in the final model due to a lack of statistical significance (Table S6). According to the likelihood ratio test, the addition of the EPIC score significantly improved the model performance (P<0.001), with a further improvement observed upon the inclusion of the APACHE II score (P=0.008; Table S7).

Table 5. Final stepwise logistic regression model for predicting severe RAP (RAC-based).

Variable	B	SE	Wald	df	P value	OR (95% CI) [Exp(B)]
APACHE II	0.135	0.050	7.207	1	0.007	1.145 (1.037–1.263)
EPIC	0.486	0.098	24.342	1	0.000	1.625 (1.340–1.971)
Constant	–5.535	0.569	94.789	1	0.000	0.004 (–)

APACHE II, Acute Physiology and Chronic Health Evaluation II; CI, confidence interval; EPIC, extrapancreatic inflammation on computed tomography; OR, odds ratio; RAC, 2012 revised Atlanta classification; RAP, recurrent acute pancreatitis; SE, standard error.

Discussion

In this study, hypertriglyceridemia emerged as the most common etiology of RAP in our region, and was frequently associated with elevated triglycerides and blood glucose levels, as well as a higher incidence of pre-existing diabetes. These clinical patterns were also observed in patients with multiple etiologies. Compared to the cholelithiasis-induced RAP patients, the hyperlipidemic RAP patients had lower rates of NP, CN, and extensive necrosis (>50%). They also exhibited milder disease severity as indicated by the MCTSI grading, lower EPIC scores, and shorter hospitalization stays. No statistically significant differences were observed among the other etiologic subgroups. Among all etiologies of RAP, biliary RAP was more common in females and older patients, with the highest median age and proportion of elderly individuals. Conversely, the younger patients were the least affected by biliary RAP among all etiologies. Collectively, these findings underscore the distinct demographic and clinical profiles of different RAP etiologies. Hyperlipidemic RAP appears to present with lower clinical severity than biliary RAP, a pattern that contrasts with trends commonly observed in AP. These findings enhance our understanding of how RAP manifestations differ by etiology, and highlight the need to consider these distinctions when evaluating the disease burden and potential treatment strategies.

In our cohort, hypertriglyceridemia was the most common etiology of RAP. Cholelithiasis-related RAP was slightly more prevalent among the female patients, while the other four etiologies (i.e., alcoholism, hypertriglyceridemia, multiple causes, and idiopathic causes) were more frequently observed in the male patients. Among these, alcoholic RAP had the highest proportion of male patients, accounting for 96.88% of cases. The patients with cholelithiasis-related RAP tended to be older than those with alcoholic, hypertriglyceridemia-related, or idiopathic RAP. The patterns of etiology and age/sex distribution in our cohort are consistent with those reported in previous studies of RAP and AP (13,34-40). In the present study, the patients with hypertriglyceridemia-related and multifactorial RAP had elevated triglyceride and blood glucose levels, and a higher prevalence of pre-existing diabetes. Similar clinical characteristics of hyperlipidemic AP have been reported (35,41). These characteristics were also observed in patients with multiple etiologies in our cohort. This may be because hypertriglyceridemia was a contributing factor in most cases of multifactorial RAP.

In this study, NP accounted for approximately 38.36% of RAP. Among these, CN was the most common subtype, followed by EXPN and PN. This distribution pattern is consistent with previous reports on RAP (36,37,39,40). Although the overall prevalence of NP did not differ significantly across etiologies, our study was the first to quantify both the proportion and extent of NP across different RAP etiologies. These findings contrast with previously reported patterns in AP (13).

AP secondary to hypertriglyceridemia is often associated with more severe clinical outcomes (9-11); however, our study revealed a contrasting trend in RAP. Specifically, in our study, the patients with hyperlipidemic RAP had lower rates of NP, CN, pancreatic necrosis >50%, and WON. The patients with hyperlipidemic RAP also had fewer cases of severe RAP based on the MCTSI score, along with lower EPIC scores and shorter hospital stays compared to those with biliary RAP. Our finding in relation to the shorter hospital stays aligns with previous observations for AP (13). This contrast may be due to the early initiation of lipid-lowering therapy following previous episodes of hyperlipidemic AP. Such treatments reduce circulating triglyceride-rich lipoproteins and inhibit their hydrolysis into non-esterified fatty acids (NEFA), which are known mediators of acinar cell injury and systemic inflammation (10,42-44). Additionally, a decrease in chylomicron levels may alleviate plasma hyperviscosity and improve pancreatic microcirculation (45,46). The suppression of oxidative stress and inflammatory cytokine production may also contribute to reducing disease severity (47-49).

Experimental studies have demonstrated that NEFA accumulation and oxidative stress exacerbate pancreatic injury in hyperlipidemic pancreatitis. Interventions such as pancreatic lipase inhibition or dietary fat restriction can mitigate these effects (42,43,50). Lipase inhibitors, including orlistat, have exerted protective effects in animal models by reducing pancreatic necrosis and mortality (51,52). Further, Chang et al. (49) reported that dual-filtration plasma exchange shortened hospital stays in patients with severe hyperlipidemic AP, potentially by removing pro-inflammatory lipoproteins. Collectively, these mechanisms may underlie the milder clinical course observed in patients with hyperlipidemic RAP in our cohort.

Given these clinical and pathophysiological differences, it is essential to explore whether imaging-based severity scores consistently reflect disease burden across different RAP etiologies. Both the MCTSI and EPIC are CT-based scoring systems that reflect the extent of local and systemic disease. The MCTSI incorporates measures of pancreatic necrosis and extrapancreatic complications, offering a morphological assessment of severity (19). Conversely, the EPIC captures inflammation beyond the pancreas and is more indicative of systemic impact (20). In our study, both the MCTSI and EPIC scores showed moderate to strong correlations with the RAC across all RAP etiologies, except for multifactorial RAP. The weaker correlations observed in this subgroup may reflect overlapping pathogenic mechanisms that obscure consistent imaging-severity relationships. Overall, our findings were similar to previous studies in both AP and RAP (13,39,40), supporting the utility of CT-based metrics in evaluating disease burden and distinguishing severity across etiologies.

In contrast to the CT-based scoring systems, the APACHE II score exhibited weak, negligible, or no correlation with the RAC across different etiologies, which suggests that it has limited value in assessing RAP severity. This may be because the APACHE II score primarily reflects systemic inflammatory status and is influenced by non-specific factors such as age. Although the prevalence of NP ranged from 31.58% to 50% among the different etiologic groups, the APACHE II score was not a reliable predictor for NP diagnosis (53,54). Therefore, the use of the APACHE II score alone may be insufficient for evaluating disease severity in RAP.

The stepwise logistic regression in our study identified both the EPIC score and APACHE II score as independent predictors of severe RAP. This result is consistent with our previous findings that the EPIC score (area under the curve ≈ 0.644–0.808) and APACHE II score (area under the curve ≈ 0.652–0.682) exhibited moderate to good discriminative ability in predicting severe RAP (40).

This study had several limitations. First, its retrospective design and the inclusion of patients who underwent CT imaging only may have introduced selection bias; notably, cases of mild RAP may be underrepresented. Second, the prevalence of pre-existing diabetes was primarily based on self-reporting, which could have led to an underestimation due to undiagnosed or asymptomatic cases. Third, the number of patients with multiple etiologies was small, and most involved combinations of biliary and hyperlipidemic causes. This limited our ability to perform stratified analyses based on specific etiologic combinations. Fourth, while prior lipid-lowering therapy may have contributed to the reduced severity observed in patients with hypertriglyceridemia-related RAP, our study lacked detailed patient-level data on treatment type, timing, and adherence, limiting the strength of our conclusions. Finally, as our sample was drawn from a single province in China, the findings may only be applicable to regions with similar patient profiles and may not be generalizable to the entire country or other areas. Taken together, these limitations highlight the need for future large-scale, multi-center studies with comprehensive clinical data to validate and extend our findings.

Conclusions

Hypertriglyceridemia was the most common cause of RAP in our provincial cohort. It was associated with elevated triglyceride and blood glucose levels, as well as a higher prevalence of pre-existing diabetes. Despite these metabolic abnormalities, the clinical course of hyperlipidemic RAP was milder than that of biliary RAP. Biliary RAP was more frequently observed in older adults and female patients, but was less common in younger individuals. No significant differences in severity were observed among alcoholic, multifactorial, and idiopathic etiologies of RAP. While previous studies of AP have occasionally included RAP subgroups, few have focused specifically on CT-based severity differences across etiologies in RAP. By providing comparative imaging and clinical profiles, our study showed that the disease severity of hyperlipidemic RAP is relatively milder than that of biliary RAP. These findings may improve understanding of etiology-specific clinical and imaging features of RAP, and support future research on its underlying mechanisms and long-term outcomes.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Ethical approval was obtained from the Sichuan Provincial People’s Hospital, University of Electronic Science and Technology of China [No. 2024ER(R)133]. All participating hospitals were informed of the study and agreed to the use of shared data. The requirement for informed consent was waived due to the retrospective nature of the study.

Data Sharing Statement

Available at https://qims.amegroups.com/article/view/10.21037/qims-24-2260/dss

qims-15-10-9375-dss.pdf ^{(67.4KB, pdf)}

DOI: 10.21037/qims-24-2260