Long-Term Multisystem Comorbidity Patterns in Acute Pancreatitis: Insights From Observational and Genetic Analysis.

J.C. takes responsibility for the integrity of the work as a whole, from inception to publication. All authors were responsible for conception and were involved in manuscript writing. T.C., T.F., and Y.Z. contributed to the methodology. J.C. conducted formal analysis and data curation. X.R., Y.S., J.G., L.D., X.W., Y.Z., D.Z., B.J.A., S.Y., X.L., and Z.W. contributed to the supervision. X.W., D.Z., X.Z., J.C., and C.Z. contributed to funding acquisition. All authors have read and approved the final version of the manuscript.

The overall ethical approval for the UK Biobank (REC reference: 21/NW/0157) was provided by the Northwest‐Haydock Research Ethics Committee, U.K. Written informed consent was obtained from the patients to publish this study. This study was conducted under UK Biobank application number 232231.

All participants provided written informed consent at the time of recruitment for their data to be used in health‐related research. Further details on the UK Biobank's ethical framework can be found at http://www.ukbiobank.ac.uk/ethics/ .

Jie Chen is supported by the National Science Foundation of China (82500637), the Natural Science Fund for Excellent Young Scholars of Hunan Province (2025JJ40083), the Natural Science Foundation of Changsha (kq2502174), and the China Postdoctoral Science Foundation (GZC20251322). Chunhua Zhou is supported by the National Natural Science Foundation of China (No.82270667 and No.82570763). Xiaoyan Wang is supported by National Natural Science Foundation of China (No.U23A20492 and No.8217033803), Scientific Research Program of FuRong Laboratory (2023SK2085‐3), and Science Fund for Creative Research Groups of the Natural Science Foundation of Hunan Province (2024J1014). Duowu Zou is supported by National Natural Science Foundation of China (No.82170559). Xiaofeng Zhang was supported by Zhejiang Province's Key R&D Plan Project (Grant No. 2023C03054) and the Construction Fund of Key Medical Disciplines of Hangzhou (No.2025HZGF05).

This study integrated observational and genetic data to explore the multisystem comorbidity patterns following AP. As shown in Figure  1 , we first included participants from the UK Biobank, a large‐scale prospective cohort comprising 502,536 participants aged 40–69 years recruited between 2006 and 2010. The study design incorporated three components: (1) an observational analysis to identify AP‐associated comorbidities; (2) disease trajectory analysis to examine the sequential diagnostic patterns; (3) genetic analyses including PRS PheWAS and LDSC to provide genetic triangulation evidence and explore shared genetic architecture and potential causal relevance of the observed associations. Schematic representation of the study design. AP, acute pancreatitis; FDR, false discovery rate; GWAS, Genome‐wide association study; LDSC, linkage disequilibrium score regression; MR, Mendelian randomization; PheWAS, phenome‐wide association study; PRS, Polygenic risk score; TDI, Townsend deprivation index. From the initial cohort, we excluded participants who withdrew from the study ( n  = 178), followed up less than 6 months ( n  = 382), and those without Townsend deprivation index (TDI) data ( n  = 625), leaving 501,181 individuals included for the current study. Incident AP cases ( n  = 4767) were identified using the first recorded diagnosis of AP from linked hospital inpatient records (International Classification of Diseases (ICD)‐10 codes K85) and primary care data (Read codes mapped to ICD‐9577.0 and ICD‐10 K85.x). The date of the first recorded AP diagnosis was defined as the index date for each case and served as the start of follow‐up. For each incident AP case, we selected up to 10 controls through incidence density sampling, matching on birth year, sex, and TDI. Controls were required to be free of AP at the corresponding case's index date and were assigned the same index date as their matched case (Supporting Information  S1 : Figure S1). To minimize the influence of reverse causality and surveillance bias, follow‐up for all participants started from 6 months after the index date and continued until death, loss to follow‐up, or the end of study (31 October 2022 for England, 31 August 2022 for Scotland, and 31 May 2022 for Wales), whichever occurred first [ 24 ]. All participants have provided informed written consent, and the UK Biobank had ethical approval from the Northwest Multicentre Research Ethics Committee. AP diagnosis was defined by Read version codes mapped to ICD code from the primary care data and ICD code from the inpatient data (Supporting Information  S2 : Table S1) [ 2 ]. Clinical outcomes and death were ascertained from three data sources: (1) primary care records (Read codes mapped to ICD), (2) hospital admissions (ICD‐9/10 codes), and (3) death registries (ICD‐10 codes). Given that individual ICD codes often represent highly specific subtypes, manifestations, or severity stages of related diseases, we used the PheCODE system to aggregate clinically and biologically related ICD codes into standardized phenotypic categories [ 25 , 26 , 27 ]. This approach reduces phenotype fragmentation, improves statistical power, and facilitates large‐scale phenome‐wide analysis. To ensure statistical power while maintaining clinical relevance, we focused on phenotypes occurring in > 1% of AP patients and with ≥ 200 cases in the overall cohort (Supporting Information  S1 : Figure S2) [ 27 , 28 ]. For each phenotype evaluated in the PheWAS analyses, participants with a recorded diagnosis of the corresponding phenotype before the index date were excluded from the risk set for that specific outcome. Consequently, the reported associations represent incident diagnoses after the index date rather than prevalent conditions existing before AP. This outcome‐specific exclusion was intended to reduce misclassification of pre‐existing disease as post‐AP morbidity. The AP‐PRS was constructed using 51 independent (linkage disequilibrium r 2  < 0.01) significant single‐nucleotide polymorphisms (SNPs) associated with AP ( p ‐value < 1 × 10 −5 ) identified from a recent GWAS of 218,595 participants in the Million Veteran Program [ 29 ]. The weighted PRS was established by summing the number of risk‐increasing alleles weighted by effect estimates on AP, which showed significant associations with the risk of AP in our analysis (Supporting Information  S2 : Tables S2–S3). The GWAS summary statistics data of AP and the investigated clinical outcomes were obtained from the FinnGen consortium (R12 release, comprising individuals of European ancestry, including 8446 cases and 437,418 controls) [ 30 ] and the UK Biobank [ 31 ]. In the LDSC and TSMR analyses, a total of 14 independent and significant SNPs with linkage disequilibrium ( r 2  < 0.01) and significance ( p  < 5 × 10 −8 ) were used as instrumental variables for AP (Supporting Information  S2 : Table S4). We first conducted Cox regression to investigate the associations between AP and subsequent comorbidities as well as all‐cause death, adjusting for potential confounders including year of birth, sex, TDI, ethnicity, education, smoking status, alcohol intake, physical activity, BMI, and assessment center (details were shown in Supporting Information  S2 : Table S5) based on prior knowledge [ 27 ]. To account for multiple testing, we applied false discovery rate (FDR) correction using the Benjamini‐Hochberg method, retaining associations with an HR > 1 and FDR‐adjusted p ‐values < 0.05 for further analysis. To characterize and visualize the sequential patterns of AP‐related comorbidities, we performed disease‐trajectory analysis using a systematic three‐stage approach. First, we constructed all possible disease 1 (D1) and disease 2 (D2) pairs and calculated the Phi correlations to identify the co‐occurrence patterns among AP‐associated comorbidities occurring in at least 1% of individuals with AP. Disease pairs with phi > 0 and FDR‐adjusted p ‐values < 0.05 indicating significant co‐occurrence strength were eligible for the next step. Second, we applied binomial tests to ensure temporal sequence, requiring D2 followed D1 in more than 50% of individuals with both diagnoses. Disease pairs with FDR‐adjusted p ‐values < 0.05 were deemed to have a significant direction. Third, we performed nested case‐control analyses matching each D2 case with five controls by birth year and sex, using conditional logistic regression to quantify associations while adjusting for the remaining covariates. Disease pairs with odds ratios (OR) > 1 and FDR‐adjusted p ‐values < 0.05 were considered statistically significant. Additionally, we extended this analytical framework to examine disease‐to‐death trajectories, aiming to characterize the chronological ordering of coded diagnoses and death events after AP. We used genetic approaches to provide complementary evidence for reported associations and dissect the potential genetic basis between diseases. We first examined the associations between PRS of AP and risk of comorbidities through a multivariable logistic regression model (adjusted for age, sex, and the first 10 genetic principal components) within a defined sub‐cohort ( n  = 485,360). The sub‐cohort excluded individuals with missing genetic data, sex chromosome aneuploidy, gender mismatch, or excess relatives (Supporting Information  S1 : Figure S3). Furthermore, LDSC regression [ 32 ] and TSMR analyses were performed to determine the potential shared genetic architecture and causality between AP and comorbidities. LDSC results were presented as genetic correlation (r g ) estimates and standard errors and TSMR analyses were primarily presented by the inverse‐variance weighted (IVW) method with OR and corresponding CIs. For TSMR analyses, we applied Cochran's Q statistics to examine the heterogeneity of SNPs' estimates, MR‐Egger, weighted median and MR‐PRESSO analyses to examine the pleiotropy and consistency of the associations. For the PRS‐PheWAS, LDSC and TSMR analyses, associations with p  < 0.05 were considered statistically significant. We conducted stratified analyses to evaluate potential effect modification across key demographic and clinical variables including sex (male and female), obesity (BMI ≥  25 kg/m 2 , BMI  14 g/day for women and > 28 g/day for men, not heavy drinking: ≤ 14 g/day for women and ≤ 28 g/day for men), and comorbidity burden (Charlson comorbidity index (CCI) = 0 versus CCI≥ 1). In the subgroup analyses, we also separately tested the interaction effects between AP and each of these variables (sex, obesity, smoking status, and alcohol intake). Additionally, we separately analyzed patients with mild and severe AP, where severe cases were defined as those requiring invasive procedures, critical care admission, or developing 30‐day severe complications [ 33 ]. In the sensitivity analyses, we excluded individuals with alcohol abuse disorder and those with pre‐existing chronic pain before the index date to assess potential confounding factors in psychiatric outcomes. We also additionally adjusted gallstone disease and triglyceride levels [ 34 , 35 ]. To further reduce the potential influence of early disease evolution, diagnostic overlap, reverse causality, and enhanced surveillance after AP, we performed lagged sensitivity analyses excluding outcomes occurring within 1, 2, and 3 years after the index date. All statistical tests were two‐tailed and performed using R software version 4.3.3 with the survival, glm, TwoSampleMR, and ldscr packages for statistical modeling, and Cytoscape version 3.10.2 for network visualization. Statistical significance was set at p  < 0.05 for all analyses.

Baseline characteristics of the overall matched screening cohort are shown in Table  1 . The cohort included 4767 individuals with AP and 47,670 matched non‐AP controls. Baseline characteristics of the eligible source cohort before matching are provided in Supporting Information  S2 : Table S6. By design, AP cases and controls were well balanced for birth year, sex, and Townsend deprivation index. Compared with matched controls, individuals with AP were more likely to be of white ethnicity (95.3% vs. 94.0%), non‐college educated (75.2% vs. 67.9%), former or current smokers (52.1% vs. 47.4%), obese (38.4% vs. 25.4%), and physically inactive (45.1% vs. 39.2%). Individuals with AP also had a higher prevalence of baseline diabetes (10.3% vs. 4.8%), cardiovascular disease (48.6% vs. 31.1%), chronic kidney disease (6.6% vs. 3.1%), mental disorders (35.3% vs. 23.3%), and gallstone disease (40.3% vs. 3.6%). Because outcome‐specific exclusions were applied separately in each PheWAS analysis, participants with a prior diagnosis of the corresponding outcome were excluded from the risk set for that specific Cox model. Baseline characteristics of the overall matched screening cohort at the index date. Among the 1429 disease conditions identified by PheCode coding across 17 disease categories, 422 outcomes occurred in > 1% of AP participants, each with more than 200 cases (Supporting Information  S1 : Figure S2). During a mean follow‐up time of 13.7 years, 250 disease outcomes were associated with AP after multiple‐testing correction (Figure  2 and Supporting Information  S2 : Table S7). Specifically, AP exhibited significant associations with digestive comorbidities, including chronic pancreatitis (HR = 99.98, 95% CI: 72.98–136.98), pancreatic cysts and pseudocysts (HR = 22.90, 95% CI: 16.85–31.10), and pancreatic cancer (HR = 2.95, 95% CI: 2.12–4.11). AP individuals also showed increased risks of endocrine and metabolic diseases, such as diabetes mellitus (HR = 3.23, 95% CI: 2.74–3.82) and infectious diseases, septicemia (HR = 2.10, 95% CI: 1.81–2.43). In terms of circulatory system diseases, AP was linked to coronary atherosclerosis (HR = 1.31, 95% CI: 1.16–1.48), myocardial infarction (HR = 1.44, 95% CI: 1.25–1.65), and phlebitis and thrombophlebitis (HR = 1.66, 95% CI: 1.25–2.21). Respiratory diseases including asthma (HR = 1.38, 95% CI: 1.14–1.66), obstructive chronic bronchitis (HR = 1.88, 95% CI: 1.59–2.22), and respiratory failure (HR = 2.21, 95% CI: 1.85–2.65) were also common comorbidities following AP. Additionally, AP was associated with an increased risk of genitourinary diseases, including acute (HR = 1.86, 95% CI: 1.67–2.07) and chronic renal failure (HR = 1.87, 95% CI: 1.61–2.17), and mental disorders like depression (HR = 1.71, 95% CI: 1.24–2.35) and anxiety (HR = 1.50, 95% CI: 1.13–1.99). Furthermore, participants with AP were observed at a higher risk of all‐cause death (HR = 1.62, 95% CI: 1.48–1.77). Significant acute pancreatitis‐related disease outcomes in observational PheWAS in the UK Biobank. The x ‐axis represents distinct phenotypic groups. The y ‐axis shows the Hazard ratio of increased disease risk in patients with acute pancreatitis compared with matched individuals. All risk increases are statistically significant (false discovery rate < 0.05). PheWAS, phenome‐wide association study. Yellow and blue were used in the scatter plot to distinguish diseases from different systems; the colors do not carry any additional meaning. Significant associations were identified in 2441 of the 62,250 disease pairs (D1–D2) and 227 of the 250 disease‐death pairs derived from the 250 AP‐associated comorbidities and all‐cause death (Supporting Information  S1 : Figure S2 and Supporting Information  S2 : Table S8). Figures  3 , 4 , 5 shows temporal diagnostic patterns after AP, mainly including three comorbidity clusters categorized by system and etiology similarities. The first cluster, involved digestive diseases, including chronic pancreatitis, cyst and pseudocyst of pancreas, pancreatic cancer, and duodenal ulcer (Figure  3 ). Subsequently, chronic pancreatitis was temporally followed by increased risk of vitamin D deficiency (OR = 3.91, 95% CI = 1.90–8.03), respiratory failure (OR = 2.39, 95% CI: 1.20–4.73) and acute renal failure (OR = 4.04, 95%CI: 2.65–6.15), and cyst and pseudocyst of pancreas was followed by an increased risk of diabetes mellitus (OR = 6.58, 95% CI: 2.31–18.70), and ultimately, all‐cause death with intervening diagnoses including pneumonia, septicemia, acidosis or sepsis. In addition, pancreatic cancer after AP is strongly associated with all‐cause death (OR = 58.39, 95% CI: 17.58–193.96) (Figure  3 ). Trajectories of the digestive disease cluster in individuals with acute pancreatitis. This plot visualizes the network of acute pancreatitis‐related comorbidities. The color of the arrows connecting the two diseases indicates the strength (hazard ratio or odds ratio) of the associations. Trajectories of the metabolic and endocrine diseases cluster in individuals with acute pancreatitis. This plot visualizes the network of acute pancreatitis‐related comorbidities. The color of the arrows connecting the two diseases indicates the strength (hazard ratio or odds ratio) of the associations. Trajectories of other disease clusters in individuals with acute pancreatitis. This plot visualizes the network of acute pancreatitis‐related comorbidities. The color of the arrows connecting the two diseases indicates the strength (hazard ratio or odds ratio) of the associations. The second cluster primarily involves metabolic and endocrine diseases, beginning with diabetes mellitus, followed by an increased risk of diabetic retinopathy and superficial cellulitis and abscess, with high estimates (OR = 21.69; OR = 16.21). Besides, coronary atherosclerosis (OR = 3.37, 95% CI: 1.86–6.81) and acute renal failure (OR = 3.69, 95% CI: 2.21–6.15) were also observed at higher risks in participants with diabetes mellitus, and ultimately, followed by adverse outcomes mainly included electrolyte imbalance and sepsis, and even all‐cause death (Figure  4 ). The third cluster starts with a sequence of cardiovascular, pulmonary, and mental disorders (Figure  5 ). Specifically, coronary atherosclerosis, phlebitis and thrombophlebitis are frequently followed by complications of cardiac or vascular devices, implants, and grafts (OR = 22.61, 95% CI: 8.14–62.82) and myocardial infarction (OR = 8.49, 95% CI: 1.62–44.60) respectively. Asthma, the common pulmonary disorder after AP, is more likely to be related to obstructive chronic bronchitis (OR = 37.82, 95% CI: 13.61–105.04), sleep disorders (OR = 69.82, 95% CI: 7.52–648.40), and chronic bronchitis (OR = 65.33, 95% CI: 8.26–516.54). Depression and anxiety following AP is mainly associated with esophagitis (OR = 19.13, 95% CI: 3.94–92.77) and major depressive disorder (OR = 5.29, 95% CI: 1.42–19.69), gastritis and duodenitis (OR = 37.64, 95% CI: 4.60–307.78) and psychogenic and somatoform disorders (OR = 11.55, 95% CI: 3.16–42.17). These disorders were subsequently followed by corresponding severe organ damage, such as acute renal failure and respiratory failure. When investigating the associations between PRS of AP and the risk of multiple clinical outcomes, we observed consistent 38 comorbidities with increased risk, including circulatory diseases (such as unstable angina, pulmonary embolism and infarction, myocardial infarction), digestive diseases (such as chronic pancreatitis, cyst and pseudocyst of pancreas, duodenal ulcer), respiratory diseases including pneumococcal pneumonia, neurological diseases including epilepsy, and mental diseases (such as delirium dementia, amnestic and other cognitive disorders) (Supporting Information  S2 : Table S9 and Supporting Information  S1 : Figure S4). As shown in Supporting Information  S2 : Table S10 and Supporting Information  S1 : Figure S4, LDSC analysis revealed significant genetic correlations between AP and several digestive systems, including chronic pancreatitis ( r g  = 1.60), diseases of pancreas ( r g  = 1.45), functional digestive disorders (r g  = 0.59), constipation (r g  = 0.35), and irritable bowel syndrome (r g  = 0.32), as well as psychiatric disorders such as anxiety, phobic and dissociative disorders (r g  = 0.29) and anxiety disorder (r g  = 0.27). These seven findings were further supported by TSMR analysis, which demonstrated potential causal effects of genetically predicted AP on these conditions. Besides, TSMR analysis indicated that genetically predicted AP was associated with an elevated risk of cerebral ischemia and inflammation of the eyelids, with consistent results in MR‐PRESSO sensitivity analysis. MR‐Egger regression detected no possible horizontal pleiotropy. Overall, genetic support was more consistent for digestive and psychiatric outcomes, whereas many other extra‐pancreatic outcomes identified in the observational PheWAS showed weaker, PRS‐only, or no consistent genetic support, indicating incomplete overlap between observational and genetic findings. In the majority of AP‐related comorbidities, the risk across different subgroups defined by sex, BMI, alcohol consumption, smoking status, and comorbidity burden was largely consistent with the primary findings (Supporting Information  S2 : Table S11–15). However, variations were observed in the strength of associations between AP and selected outcomes across different risk factor strata. Specifically, although AP showed a higher risk of cyst and pseudocyst of pancreas in both sexes, generally higher estimated effect sizes (95% CIs without overlap, and p‐interaction < 0.001) were observed in males. Significant associations between AP and phlebitis and thrombophlebitis, duodenal ulcer, depression, anxiety, pancreatic cancer, and asthma were observed in male participants, but did not reach statistical significance in females. Regarding smoking status, the association of AP with anxiety was not significant among non‐smokers, whereas the associations with phlebitis and thrombophlebitis, and asthma were not significant among smokers. Additionally, heavy drinkers showed a relatively higher risk of acute renal failure, while associations with phlebitis and thrombophlebitis, duodenal ulcers, depression, anxiety, and asthma were not significant in this group. In analyses stratified by CCI‐defined baseline comorbidity burden, the major AP‐associated outcomes remained generally consistent across comorbidity strata and were aligned with the primary findings. In analysis stratified by disease severity, patients with severe AP exhibited relatively higher risks of multiple diseases, including chronic pancreatitis, cyst and pseudocyst of pancreas, pancreatic cancer, duodenal ulcer, diabetes mellitus, and coronary atherosclerosis, as well as all‐cause death (Supporting Information  S2 : Table S16). The associations between AP and the majority of comorbidities remained consistent after excluding individuals with alcohol abuse disorder, pre‐existing chronic pain additionally adjusting for gallstone, or additionally adjusting for triglycerides (Supporting Information  S2 : Table S17). Additional lagged sensitivity analyses excluding outcomes occurring within 1, 2, and 3 years after the index date showed that most major associations remained directionally consistent (Supporting Information  S2 : Table S18). Effect estimates for pancreatic outcomes, particularly chronic pancreatitis, were attenuated after longer exclusion windows.

In conclusion, this study establishes AP as a systemic disorder with distinct progression patterns across pancreatic, metabolic, cardiovascular, pulmonary, and psychiatric domains. The integration of observational and genetic evidence provides complementary support for selected AP‐associated comorbidity patterns and may inform more comprehensive approaches to post‐AP care. By shifting focus from acute management to long‐term multisystem surveillance, these findings can inform clinical guidelines and ultimately improve outcomes for the growing population of AP survivors.

This comprehensive study elucidates the multisystem comorbidity burden following AP through integrated observational and genetic analyses of a large prospective cohort. AP was related to 250 comorbidities and all‐cause death, and 38 comorbidities were validated in AP participants with higher PRS. Additionally, LDSC analysis identified shared genetic architecture between AP and 53 comorbidities, with TSMR analysis highlighting potential causal relationships with 9 comorbidities. Our findings reveal three principal temporal comorbidity patterns involving pancreatic, metabolic, and systemic inflammatory pathways, supported by both phenotypic associations and genetic evidence. The results substantially expand the current understanding of AP as not merely an acute abdominal condition affecting the pancreas, but rather a systemic disorder with far‐reaching consequences across multiple organ systems. The most striking finding was the exceptionally high risk of chronic pancreatitis following AP. Although the magnitude of this estimate should be interpreted cautiously, the association remains clinically meaningful because AP is a well‐established risk factor for subsequent CP and represents a key component of the AP–CP disease continuum. This finding is consistent with prior meta‐analytic estimates showing that 10%–36% of patients develop CP depending on recurrence status [ 4 ], and was directionally consistent in lagged sensitivity analyses excluding outcomes occurring within 1, 2, and 3 years after the index date. The progression from AP to CP is also supported by recent long‐term clinical follow‐up data. In a Dutch multicenter cohort, recurrent AP and CP developed in 25% and 6% of patients after a first AP episode, respectively, and male sex, smoking, non‐biliary etiology, and recurrent AP were identified as factors associated with CP progression progression [ 9 ]. The progression from AP to chronic pancreatitis involves multiple potential mechanisms. According to the necrosis‐fibrosis theory, pancreatic injury during AP triggers inflammation, pancreatic duct obstruction, and fibrosis, which are hallmark features of chronic pancreatitis [ 36 ]. Our trajectory analysis extends these observations by demonstrating how chronic pancreatitis subsequently predisposes to nutritional deficiencies and multi‐organ dysfunction. The strong genetic correlations and Mendelian randomization evidence for causality suggest shared biological mechanisms. Previous studies have identified variants in PRSS1 , SPINK1 , and CFTR that predispose to both acute and chronic pancreatic injury [ 37 ]. These findings should be interpreted in the context of AP heterogeneity, as etiology, recurrence, smoking, alcohol exposure, severity, necrosis, ductal injury, and pancreatic interventions may modify the long‐term risk of CP. Together, these findings underscore the importance of early interventions to prevent disease progression, particularly in high‐risk groups such as males, smokers, and patients with severe initial presentations. Metabolic disorders represented another important component of the observed temporal patterns, with diabetes mellitus serving as a key diagnosis after AP. Our findings align with meta‐analytic data showing 15% of AP patients develop diabetes within 1 year [ 38 ], while the progression to microvascular complications highlights the clinical significance of pancreatogenic diabetes. Notably, the temporal sequence from pancreatic structural abnormalities to diabetes in our analysis is consistent with recent prospective follow‐up data from the Goulash‐Plus cohort, which showed that pancreatic morphologic progression and endocrine dysfunction often evolve together after AP, particularly during the first 2 years after the acute episode [ 7 ]. Structural pancreatic damage, including ductal injury, local inflammation, fibrosis, and islet dysfunction, may contribute to post‐pancreatitis diabetes. This has important implications for clinical practice, as current guidelines often fail to address the unique pathophysiology and management needs of post‐pancreatitis diabetes mellitus [ 39 ]. Our findings support the implementation of routine glycemic monitoring in AP survivors, particularly those with pancreatic structural abnormalities. Beyond pancreatic and metabolic sequelae, the cardiovascular and pulmonary complications observed in this study provide novel insights into AP's systemic impact. The associations with myocardial infarction and coronary atherosclerosis persisted after extensive covariate adjustment, suggesting mechanisms beyond shared risk factors. Experimental models propose that pancreatic inflammation triggers systemic release of damage‐associated molecular patterns (DAMPs) that promote endothelial dysfunction and atherogenesis [ 40 , 41 ]. Similarly, the pulmonary associations (obstructive bronchitis and respiratory failure) may reflect both direct inflammatory spillover and secondary effects of pancreatic enzymes on pulmonary surfactant [ 42 , 43 ]. These findings suggest that AP should be considered as an independent risk factor for cardiopulmonary disease. It is necessary for patients with AP to undergo a more aggressive management of cardiopulmonary risk factors. The psychiatric comorbidity patterns revealed in this study address an underrecognized aspect of AP outcomes [ 44 ]. The observed associations with depression and anxiety, supported by genetic correlations, suggest bidirectional brain‐gut‐pancreas interactions that may involve shared inflammatory pathways [ 45 ]. The trajectory from mental health disorders to subsequent gastrointestinal complications (esophagitis, gastritis and duodenitis) raises the possibility that psychological distress may exacerbate digestive symptoms through gut‐brain axis dysregulation. These findings highlight the need for integrated care models that address both physical and mental health in patients with AP. The observed association between AP and all‐cause mortality is consistent with recent follow‐up studies highlighting the vulnerability of the post‐AP period. Czapári et al. reported that mortality after discharge was substantially elevated after AP, with the first 90 days representing a particularly high‐risk window and cardiac failure and sepsis among the leading early causes of death [ 10 ]. A recent population‐based Swedish cohort further showed that long‐term mortality remained increased after AP even after adjustment for comorbidities and censoring for recurrent AP or CP [ 8 ]. These findings, together with our observed trajectories linking AP‐related comorbidities to organ dysfunction, sepsis, and death, support the need to consider mortality risk as part of long‐term post‐AP care rather than focusing only on recurrent pancreatic events. Methodologically, our study integrates phenome‐wide association analysis, trajectory analysis, and genetic approaches to characterize long‐term comorbidity patterns after AP. The trajectory analysis moves beyond static comorbidity assessment by describing temporal diagnostic patterns, while the genetic analyses provide complementary triangulation evidence for selected associations rather than definitive proof of causality. The consistency between observational and genetic findings for key outcomes (e.g., chronic pancreatitis, diabetes) suggests that these associations are unlikely to result solely from confounding. Furthermore, the identification of subgroup variations by sex, disease severity, and lifestyle factors enables more personalized risk stratification. The clinical implications of these findings are substantial. First, the identified trajectories argue for a paradigm shift from reactive to proactive monitoring after AP, with particular attention to high‐risk transitions. Patients with pancreatic structural complications or recurrent/severe AP may benefit from closer CP surveillance, while those with pancreatic cysts, pseudocysts, or metabolic risk factors may warrant periodic glycemic screening. Second, modifiable risk factors should be addressed during follow‐up, particularly smoking cessation, alcohol reduction, and management of biliary disease when appropriate. Third, the multisystem involvement suggests that optimal post‐AP care requires multidisciplinary collaboration between gastroenterologists, endocrinologists, cardiologists, and mental health professionals. Additionally, the genetic findings open new avenues for risk prediction and targeted prevention in individuals with high polygenic risk scores. Several limitations warrant consideration. The observational design cannot fully exclude residual confounding, although genetic analyses help mitigate this concern. Residual confounding by alcohol‐related behaviors and chronic pain may remain, as these factors may be incompletely captured in routine healthcare records. The AP‐PRS was derived from MVP multi‐ancestry GWAS results and tested in the independent UK Biobank cohort, which helped avoid sample overlap and reduce overfitting. However, differences in ancestry composition, age, sex, healthcare setting, and disease ascertainment between MVP and UK Biobank may influence PRS transferability and reproducibility. In addition, PRS, LDSC, and TSMR are subject to limitations including pleiotropy, shared genetic architecture, instrument strength, and cross‐cohort transferability. The UK Biobank's predominantly middle‐aged, European‐ancestry, and healthier volunteer population, together with AP and outcome ascertainment based on routine healthcare records, may limit generalizability to other populations and healthcare settings. Our focus on more common phenotypes (≥ 200 cases) may have missed important rare outcomes. PheCODE aggregation may reduce phenotypic granularity; therefore, findings should be interpreted as broad clinical phenotypes rather than subtype‐specific associations. Detailed AP etiology, recurrence, necrosis, ductal injury, and intervention data were incompletely captured in routine healthcare records, limiting etiology‐specific interpretation of long‐term outcomes. Although individuals with pre‐existing diagnoses of each outcome were excluded from the corresponding analyses, baseline comorbidity differences indicate that some associations may still reflect shared susceptibility, common risk factors, or broader multimorbidity patterns rather than direct consequences of AP. Surveillance bias should also be considered, as individuals with AP may have more frequent healthcare contact and diagnostic evaluation after the index event, increasing the likelihood of detecting subsequent conditions, particularly underdiagnosed outcomes such as depression and anxiety. Although differential surveillance may still partially contribute to some associations, the 6‐month lag period and additional lagged sensitivity analyses excluding outcomes within 1, 2, and 3 years after the index date yielded broadly consistent findings with the primary analyses. Therefore, the identified trajectories should be interpreted as temporal sequences of recorded diagnoses following AP rather than definitive evidence of disease progression or causality. Future research should prioritize several directions: (1) mechanistic studies to elucidate the biological pathways linking AP to extra‐pancreatic diseases; (2) clinical trials testing targeted monitoring and intervention strategies for high‐risk progression pathways; (3) development of integrated risk prediction models incorporating clinical, genetic, and biomarker data; and (4) investigation of health disparities in AP outcomes across diverse populations.

Acute pancreatitis (AP) is an inflammatory disorder of the pancreas that can lead to significant morbidity and mortality [ 1 ]. Over the past 20 years, its global incidence has risen sharply, from 21.4 to 48.2 per 100,000 individuals, making it one of the leading causes of gastrointestinal hospitalization [ 2 ]. Despite improvements in acute management, AP remains a life‐threatening disorder, accounting for over 115,000 deaths annually worldwide [ 3 ]. While the immediate clinical manifestations (e.g., pancreatic necrosis, systemic inflammatory response syndrome (SIRS), and multi‐organ dysfunction) are well characterized, the long‐term systemic consequences of AP remain incompletely understood. Emerging evidence suggests that AP may serve as a precursor to a spectrum of chronic multisystem disorders. Notably, AP, particularly when it recurs, is associated with subsequent chronic pancreatitis, with meta‐analyses estimating that 10% of patients with the first episode of AP and 36% of those with recurrent AP develop this condition [ 4 ]. Recent longitudinal follow‐up studies have further demonstrated substantial risks of recurrent AP, chronic pancreatitis, pancreatic cancer, endocrine dysfunction, and excess mortality after an initial AP episode [ 5 , 6 , 7 , 8 , 9 , 10 ]. Beyond these established pancreatic and survival outcomes, AP has also been implicated in cardiovascular [ 11 , 12 ], pulmonary [ 13 , 14 ], renal [ 15 , 16 , 17 ], neurological [ 18 , 19 , 20 ], and psychiatric disorders [ 21 ]. These associations suggest that AP may trigger or accelerate systemic pathophysiological processes; however, a critical gap is the lack of a comprehensive, data‐driven assessment of AP‐related comorbidity networks. Previous studies have mainly focused on individual disease associations, often in isolation, without examining how these conditions interact or progress over time. Furthermore, while shared genetic risk factors (e.g., variants in PRSS1 , SPINK1 , and CFTR ) have been implicated in both AP and chronic pancreatitis [ 22 , 23 ], the broader genetic architecture linking AP to extra‐pancreatic diseases remains unexplored. In this study, we leveraged large‐scale health records and genetic data from the UK Biobank to systematically assess the multisystem comorbidities of AP. Our objectives were: 1) to identify and quantify the full spectrum of AP‐associated comorbidities using a hypothesis‐free phenome‐wide association study (PheWAS) approach; 2) to characterize temporal diagnostic sequences following AP; 3) to use genetic approaches, including polygenic risk scores (PRS), linkage disequilibrium score regression (LDSC), and two‐sample Mendelian randomization (TSMR), as a triangulation framework to provide complementary evidence for the observed associations and explore shared genetic architecture and potential causal relevance. By integrating the above methods, we aimed to provide a holistic understanding of AP as a systemic disease, uncovering novel pathways for clinical monitoring and mechanistic research.

The authors declare no conflicts of interest.

Supporting Information S1 Supporting Information S2