Early infection is an independent risk factor for increased mortality in patients with culture-confirmed infected pancreatic necrosis

The 12-Year Experience of the Hungarian Pancreatic Study Group

The Hungarian Pancreatic Study Group (HPSG) was established with the aim of advancing pancreatology. Our summary outlines the methodologies, key results, and future directions of the HPSG. Methodological elements included, the formation of strategic national and international collaborations, the establishment of patient registries and biobanks, and a strong focus on education and guideline development. Key results encompassed, pioneering research on pancreatic ductal function and the role of cystic fibrosis transmembrane conductance regulator (CFTR) in inflammation, significant advancements in understanding acute and chronic pancreatitis, and the execution of numerous clinical trials to explore new therapeutic approaches. Despite challenges, such as securing funding and translating research into clinical practice, the HPSG's commitment to patient care and scientific innovation has been unwavering. The group aims to deepen research into pancreatic cancer and chronic pancreatitis, conduct more randomized controlled trials (RCTs), and expand its efforts internationally by involving global staff and patients. The authors hope that this summary inspires others to undertake similar initiatives and contribute to the global advancement of medical research and patient care in pancreatology.

Development and validation of an explainable machine learning model for mortality prediction among patients with infected pancreatic necrosis

Background

Infected pancreatic necrosis (IPN) represents a severe complication of acute pancreatitis, commonly linked with mortality rates ranging from 15% to 35%. However, the present mortality prediction tools for IPN are limited and lack sufficient sensitivity and specificity. This study aims to develop and validate an explainable machine learning (ML) model for death prediction among patients with IPN.

Methods

We performed a prospective cohort study of 344 patients with IPN consecutively enrolled from a large Chinese tertiary hospital from January 2011 to January 2023. Ten ML models were developed to predict 90-day mortality in these patients. A benchmarking test, involving nested resampling, automatic hyperparameter tuning and random search techniques, was conducted to select the ML model. Sequential forward selection method was employed to select the optimal feature subset from 31 candidate subsets to simplify the model and maximize predictive performance. The final model was internally validated with the 1000 bootstrap method and externally validated using an independent cohort of 132 patients with IPN retrospectively collected from another Chinese tertiary hospital from January 2018 to January 2023. The SHapley Additive exPlanations (SHAP) method was employed to interpret the model in terms of features importance and features effect. The final model constructed with optimal feature subset was deployed as an interactive web-based Shiny app.

Findings

Random survival forest (RSF) model showed the best predictive performance than other 9 ML models (internal validation, C-index = 0.863 [95% CI: 0.854-0.875]; external validation, C-index = 0.857 [95% CI: 0.850-0.865]). Multiple organ failure, Acute Physiology and Chronic Health Examination II (APACHE II) score ≥20, duration of organ failure ≥21 days, bloodstream infection, time from onset to first intervention <30 days, Bedside Index of Severity in Acute Pancreatitis score ≥3, critical acute pancreatitis, age ≥ 50 years, and hemorrhage were 9 most important features associated with mortality. Furthermore, SHAP algorithm revealed insightful nonlinear interactive associations between important predictors and mortality, identifying 9 features pairs with high interaction SHAP value and clinical significance. Two interactive web-based Shiny apps were developed to enhance clinical practicability: https://rsfmodels.shinyapps.io/IPN_app/ for cases where the APACHE II score was available and https://rsfmodels.shinyapps.io/IPNeasy/ for cases where it was not.

Interpretation

An explainable ML model for death prediction among IPN patients was feasible and effective, suggesting its superior potential in guiding clinical management and improving patient outcomes. Two publicly accessible web tools generated for the optimized model facilitated its utility in clinical settings.

Funding

The Natural Science Foundation of Hunan Province (2023JJ30885), Postdoctoral Fellowship Program of CPSF (GZB20230872), The Youth Science Foundation of Xiangya Hospital (2023Q13), The Project Program of National Clinical Research Center for Geriatric Disorders of Xiangya Hospital (2021LNJJ19).

Infected pancreatic necrosis: outcomes and clinical predictors of mortality. A post hoc analysis of the MANCTRA-1 international study

The identification of high-risk patients in the early stages of infected pancreatic necrosis (IPN) is critical, because it could help the clinicians to adopt more effective management strategies. We conducted a post hoc analysis of the MANCTRA-1 international study to assess the association between clinical risk factors and mortality among adult patients with IPN. Univariable and multivariable logistic regression models were used to identify prognostic factors of mortality. We identified 247 consecutive patients with IPN hospitalised between January 2019 and December 2020. History of uncontrolled arterial hypertension (p = 0.032; 95% CI 1.135-15.882; aOR 4.245), qSOFA (p = 0.005; 95% CI 1.359-5.879; aOR 2.828), renal failure (p = 0.022; 95% CI 1.138-5.442; aOR 2.489), and haemodynamic failure (p = 0.018; 95% CI 1.184-5.978; aOR 2.661), were identified as independent predictors of mortality in IPN patients. Cholangitis (p = 0.003; 95% CI 1.598-9.930; aOR 3.983), abdominal compartment syndrome (p = 0.032; 95% CI 1.090-6.967; aOR 2.735), and gastrointestinal/intra-abdominal bleeding (p = 0.009; 95% CI 1.286-5.712; aOR 2.710) were independently associated with the risk of mortality. Upfront open surgical necrosectomy was strongly associated with the risk of mortality (p < 0.001; 95% CI 1.912-7.442; aOR 3.772), whereas endoscopic drainage of pancreatic necrosis (p = 0.018; 95% CI 0.138-0.834; aOR 0.339) and enteral nutrition (p = 0.003; 95% CI 0.143-0.716; aOR 0.320) were found as protective factors. Organ failure, acute cholangitis, and upfront open surgical necrosectomy were the most significant predictors of mortality. Our study confirmed that, even in a subgroup of particularly ill patients such as those with IPN, upfront open surgery should be avoided as much as possible. Study protocol registered in ClinicalTrials.Gov (I.D. Number NCT04747990).

Escherichia coli infection indicates favorable outcomes in patients with infected pancreatic necrosis

Introduction

Infected pancreatic necrosis (IPN) is a severe complication of acute necrotizing pancreatitis with increasing morbidity. Escherichia coli is the most frequently cultured microorganism in IPN. However, the implications of Escherichia coli infection on the outcomes of patients with IPN remain unclear. Therefore, this study aimed to evaluate the clinical impacts of Escherichia coli infection on IPN.

Methods

A prospective database with consecutive patients with IPN between January 2010 and April 2022 at a tertiary hospital was post-hoc analyzed. The clinical and microbiological characteristics, surgical management, and follow-up data of patients with and without Escherichia coli infection were compared.

Results

A total of 294 IPN patients were enrolled in this cohort. Compared with non-Escherichia coli infection cases (n=80, 27.2%), patients with Escherichia coli infection (n=214, 72.8%) were characterized by more frequent polymicrobial infections (77.5% vs. 65.0%, P=0.04) but a lower occurrence of severe acute pancreatitis (SAP) (42.5% vs. 61.7%, P=0.003). In addition, significantly lower mortality (12.5% vs. 30.4%, p=0.002), fewer step-up surgical interventions (73.8% vs. 85.1%, P=0.025), and a lower rate of multiple organ failure (MOF) (25.0% vs. 40.2%, P=0.016) were also observed in patients with Escherichia coli infection. Multivariate analysis of mortality predictors indicated that MOF (odds ratio [OR], 6.197; 95% confidence interval [CI], 2.373–16.187; P&lt;0.001) and hemorrhage (OR, 3.485; 95% CI, 1.623–7.487; P=0.001) were independent predictors associated with higher mortality in patients with IPN. Escherichia coli infection was significantly associated with a lower mortality (OR, 0.302; 95% CI, 0.121–0.751; P= 0.01).

Conclusion

Escherichia coli infection indicates a favorable prognosis in patients with IPN, although the mechanism needs further investigation.

Characteristics and risk factors for extrapancreatic infection in patients with moderate or severe acute pancreatitis

BACKGROUND

How to detect acute pancreatitis (AP) complicated with infection early and how to arrange the treatment time are still the main problems in the world. There are few reports on the potential relationship between extrapancreatic infections and AP. The purpose of this article was to investigate the characteristics, influencing factors and prognosis of extrapancreatic infection in AP patients with modified Marshall score ≥2 on admission.

MATERIALS AND METHODS

We retrospectively analyzed AP admitted to emergency intensive care unit of Ruijin hospital within 72 h of onset from September 2019 to December 2021. In addition to the patients' baseline data, sites of infection and microorganisms outside the pancreas were collected. Microbial cultures were used to identify infections of the respiratory tract, blood, abdominal cavity, biliary tract, urinary tract and clostridium difficile in feces.

RESULTS

144 patients with AP were included, of which extrapancreatic infection accounted for 40.28%. C-reactive protein, procalcitonin, blood urea nitrogen, serum creatinine, oxygenation index, modified Marshall score, BISAP score and APACHE II score were significantly increased in the extrapancreatic infection group. The risk factors of extrapancreatic infection included blood urea nitrogen, Modified Marshall score and duration of mechanical ventilation. The positive rates of pathogenic bacteria in sputum culture, blood culture, ascites culture and bile culture were significantly higher than those in the 1-3 days after admission. The infection begins to worsen as early as 4-7 days after the onset of symptoms. Extrapancreatic infection is associated with pancreatic necrosis, the rate of laparotomy, length of hospital stay and in-hospital mortality.

CONCLUSION

Our research has confirmed the need to prevent and monitor extrapancreatic infection in the early stage.

Acute Pancreatitis: Diagnosis and Treatment

Acute pancreatitis is a common indication for hospital admission, increasing in incidence, including in children, pregnancy and the elderly. Moderately severe acute pancreatitis with fluid and/or necrotic collections causes substantial morbidity, and severe disease with persistent organ failure causes significant mortality. The diagnosis requires two of upper abdominal pain, amylase/lipase ≥ 3 ×upper limit of normal, and/or cross-sectional imaging findings. Gallstones and ethanol predominate while hypertriglyceridaemia and drugs are notable among many causes. Serum triglycerides, full blood count, renal and liver function tests, glucose, calcium, transabdominal ultrasound, and chest imaging are indicated, with abdominal cross-sectional imaging if there is diagnostic uncertainty. Subsequent imaging is undertaken to detect complications, for example, if C-reactive protein exceeds 150 mg/L, or rarer aetiologies. Pancreatic intracellular calcium overload, mitochondrial impairment, and inflammatory responses are critical in pathogenesis, targeted in current treatment trials, which are crucially important as there is no internationally licenced drug to treat acute pancreatitis and prevent complications. Initial priorities are intravenous fluid resuscitation, analgesia, and enteral nutrition, and when necessary, critical care and organ support, parenteral nutrition, antibiotics, pancreatic exocrine and endocrine replacement therapy; all may have adverse effects. Patients with local complications should be referred to specialist tertiary centres to guide further management, which may include drainage and/or necrosectomy. The impact of acute pancreatitis can be devastating, so prevention or reduction of the risk of recurrence and progression to chronic pancreatitis with an increased risk of pancreas cancer requires proactive management that should be long term for some patients.

EASY-APP: An artificial intelligence model and application for early and easy prediction of severity in acute pancreatitis

BACKGROUND

Acute pancreatitis (AP) is a potentially severe or even fatal inflammation of the pancreas. Early identification of patients at high risk for developing a severe course of the disease is crucial for preventing organ failure and death. Most of the former predictive scores require many parameters or at least 24 h to predict the severity; therefore, the early therapeutic window is often missed.

METHODS

The early achievable severity index (EASY) is a multicentre, multinational, prospective and observational study (ISRCTN10525246). The predictions were made using machine learning models. We used the scikit-learn, xgboost and catboost Python packages for modelling. We evaluated our models using fourfold cross-validation, and the receiver operating characteristic (ROC) curve, the area under the ROC curve (AUC), and accuracy metrics were calculated on the union of the test sets of the cross-validation. The most critical factors and their contribution to the prediction were identified using a modern tool of explainable artificial intelligence called SHapley Additive exPlanations (SHAP).

RESULTS

The prediction model was based on an international cohort of 1184 patients and a validation cohort of 3543 patients. The best performing model was an XGBoost classifier with an average AUC score of 0.81 ± 0.033 and an accuracy of 89.1%, and the model improved with experience. The six most influential features were the respiratory rate, body temperature, abdominal muscular reflex, gender, age and glucose level. Using the XGBoost machine learning algorithm for prediction, the SHAP values for the explanation and the bootstrapping method to estimate confidence, we developed a free and easy-to-use web application in the Streamlit Python-based framework (http://easy-app.org/).

CONCLUSIONS

The EASY prediction score is a practical tool for identifying patients at high risk for severe AP within hours of hospital admission. The web application is available for clinicians and contributes to the improvement of the model.