Creation of a Pediatric Choledocholithiasis Prediction Model

[Establishment and Validation of a Predictive Model for Gallstone Disease in the General Population: A Multicenter Study]

Objective

Gallstone disease (GSD) is one of the common digestive tract diseases with a high worldwide prevalence. The effects of GSD on patients include but are not limited to the symptoms of nausea, vomiting, and biliary colic directly caused by GSD. In addition, there is mounting evidence from cohort studies connecting GSD to other conditions, such as cardiovascular diseases, biliary tract cancer, and colorectal cancer. Early identification of patients at a high risk of GSD may help improve the prevention and control of the disease. A series of studies have attempted to establish prediction models for GSD, but these models could not be fully applied in the general population due to incomplete prediction factors, small sample sizes, and limitations in external validation. It is crucial to design a universally applicable GSD risk prediction model for the general population and to take individualized intervention measures to prevent the occurrence of GSD. This study aims to conduct a multicenter investigation involving more than 90000 people to construct and validate a complete and simplified GSD risk prediction model.

Methods

A total of 123634 participants were included in the study between January 2015 and December 2020, of whom 43929 were from the First Affiliated Hospital of Chongqing Medical University (Chongqing, China), 11907 were from the First People's Hospital of Jining City (Shandong, China), 1538 were from the Tianjin Medical University Cancer Institute and Hospital (Tianjin, China), and 66260 were from the People's Hospital of Kaizhou District (Chongqing, China). After excluding patients with incomplete clinical medical data, 35976 patients from the First Affiliated Hospital of Chongqing Medical University were divided into a training data set (n=28781, 80%) and a validation data set (n=7195, 20%). Logistic regression analyses were performed to investigate the relevant risk factors of GSD, and a complete risk prediction model was constructed. Factors with high scores, mainly according to the nomograms of the complete model, were retained to simplify the model. In the validation data set, the diagnostic accuracy and clinical performance of these models were validated using the calibration curve, area under the curve (AUC) of the receiver operating characteristic curve, and decision curve analysis (DCA). Moreover, the diagnostic accuracy of these two models was validated in three other hospitals. Finally, we established an online website for using the prediction model (The complete model is accessible at https://wenqianyu.shinyapps.io/Completemodel/, while the simplified model is accessible at https://wenqianyu.shinyapps.io/Simplified/).

Results

After excluding patients with incomplete clinical medical data, a total of 96426 participants were finally included in this study (35876 from the First Affiliated Hospital of the Chongqing Medical University, 9289 from the First People's Hospital of Jining City, 1522 from the Tianjin Medical University Cancer Institute, and 49639 from the People's Hospital of Kaizhou District). Female sex, advanced age, higher body mass index, fasting plasma glucose, uric acid, total bilirubin, gamma-glutamyl transpeptidase, and fatty liver disease were positively associated with risks for GSD. Furthermore, gallbladder polyps, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and aspartate aminotransferase were negatively correlated to risks for GSD. According to the nomograms of the complete model, a simplified model including sex, age, body mass index, gallbladder polyps, and fatty liver disease was constructed. All the calibration curves exhibited good consistency between the predicted and observed probabilities. In addition, DCA indicated that both the complete model and the simplified model showed better net benefits than treat-all and treat-none. Based on the calibration plots, DCA, and AUCs of the complete model (AUC in the internal validation data set=74.1% [95% CI: 72.9%-75.3%], AUC in Shandong=71.7% [95% CI: 70.6%-72.8%], AUC in Tianjin=75.3% [95% CI: 72.7%-77.9%], and AUC in Kaizhou=72.9% [95% CI: 72.5%-73.3%]) and the simplified model (AUC in the internal validation data set=73.7% [95% CI: 72.5%-75.0%], AUC in Shandong=71.5% [95% CI: 70.4%-72.5%], AUC in Tianjin=75.4% [95% CI: 72.9%-78.0%], and AUC in Kaizhou=72.4% [95% CI: 72.0%-72.8%]), we concluded that the complete and simplified risk prediction models for GSD exhibited excellent performance. Moreover, we detected no significant differences between the performance of the two models (P>0.05). We also established two online websites based on the results of this study for GSD risk prediction.

Conclusions

This study innovatively used the data from 96426 patients from four hospitals to establish a GSD risk prediction model and to perform risk prediction analyses of internal and external validation data sets in four cohorts. A simplified model of GSD risk prediction, which included the variables of sex, age, body mass index, gallbladder polyps, and fatty liver disease, also exhibited good discrimination and clinical performance. Nonetheless, further studies are needed to explore the role of low-density lipoprotein cholesterol and aspartate aminotransferase in gallstone formation. Although the validation results of the complete model were better than those of the simplified model to a certain extent, the difference was not significant even in large samples. Compared with the complete model, the simplified model uses fewer variables and yields similar prediction and clinical impact. Hence, we recommend the application of the simplified model to improve the efficiency of screening high-risk groups in practice. The use of the simplified model is conducive to enhancing the self-awareness of prevention and control in the general population and early intervention for GSD.

Using artificial intelligence to predict choledocholithiasis: can machine learning models abate the use of MRCP in patients with biliary dysfunction?

BACKGROUND

Prompt diagnosis of choledocholithiasis is crucial for reducing disease severity, preventing complications and minimizing length of stay. Magnetic resonance cholangiopancreatography (MRCP) is commonly used to evaluate patients with suspected choledocholithiasis but is expensive and may delay definitive intervention. To optimize patient care and resource utilization, we have developed five machine learning models that predict a patients' risk of choledocholithiasis based on clinical presentation and pre-MRCP investigation results.

METHODS

Inpatients admitted to the Royal Hobart Hospital from 2018 to 2023 with a suspicion of choledocholithiasis were included. Exclusion criteria included prior hepatobiliary surgery, known hepatobiliary disease, or incomplete records. Variables related to clinical presentation, laboratory testing, and sonographic or CT imaging were collected. Four machine learning techniques were employed: logistic regression, XGBoost, random forest, and K-nearest neighbours. The three best performing models were combined to create an ensemble model. Model performance was compared against the American Society for Gastrointestinal Endoscopy (ASGE) choledocholithiasis risk stratification guidelines.

RESULTS

Of the 222 patients included, 113 (50.9%) had choledocholithiasis. The most successful models were the random forest (accuracy: 0.79, AUROC: 0.83) and ensemble (accuracy and AUROC: 0.81). Every model outperformed the ASGE guidelines. Key variables influencing the models' predictions included common bile duct diameter, lipase, imaging evidence of cholelithiasis, and liver function tests.

CONCLUSION

Machine learning models can accurately assess a patient's risk of choledocholithiasis and could assist in identifying patients who could forgo an MRCP and proceed directly to intervention. Ongoing validation on prospective data is necessary to refine their accuracy and clinical utility.

Machine learning to predict pediatric choledocholithiasis: A Western Pediatric Surgery Research Consortium retrospective study

BACKGROUND

The purpose of this study was to accurately predict pediatric choledocholithiasis with clinical data using a computational machine learning algorithm.

METHODS

A multicenter retrospective cohort study was performed on children <18 years of age who underwent cholecystectomy between 2016 to 2019 at 10 pediatric institutions. Demographic data, clinical findings, laboratory, and ultrasound results were evaluated by bivariate analyses. An Extra-Trees machine learning algorithm using k-fold cross-validation was used to determine predictive factors for choledocholithiasis. Model performance was assessed using the area under the receiver operating characteristic curve on a validation dataset.

RESULTS

A cohort of 1,597 patients was included, with an average age of 13.9 ± 3.2 years. Choledocholithiasis was confirmed in 301 patients (18.8%). Obesity was the most common comorbidity in all patients. Choledocholithiasis was associated with the finding of a common bile duct stone on ultrasound, increased common bile duct diameter, and higher serum concentrations of aspartate aminotransferase, alanine transaminase, lipase, and direct and peak total bilirubin. Nine features (age, body mass index, common bile duct stone on ultrasound, common bile duct diameter, aspartate aminotransferase, alanine transaminase, lipase, direct bilirubin, and peak total bilirubin) were clinically important and included in the machine learning algorithm. Our 9-feature model deployed on new patients was found to be highly predictive for choledocholithiasis, with an area under the receiver operating characteristic score of 0.935.

CONCLUSION

This multicenter study uses machine learning for pediatric choledocholithiasis. Nine clinical factors were highly predictive of choledocholithiasis, and a machine learning model trained using medical and laboratory data was able to identify children at the highest risk for choledocholithiasis.

Magnetic resonance imaging following the demonstration of a normal common bile duct on ultrasound in children with suspected choledocholithiasis: what is the benefit?

BACKGROUND

The role of MRI in evaluating children with an in situ gallbladder and suspected choledocholithiasis following a negative or inconclusive US is unclear.

OBJECTIVE

To determine whether MRI benefits children with suspected choledocholithiasis and a normal common bile duct (CBD) without stones on US.

MATERIALS AND METHODS

We conducted a retrospective 10-year review of paired US and MRI (within 10 days) in children 18 years or younger with suspected choledocholithiasis. With MRI as a reference standard, two reviewers independently evaluated the images for CBD diameter, choledocholithiasis, cholelithiasis and pancreatic edema. Serum lipase was recorded. We calculated exact binomial confidence limits for test positive predictive values (PPVs) and negative predictive values (NPVs) using R library epiR.

RESULTS

Of 87 patients (46 female, 41 male; mean age 14 years, standard deviation [SD] 4.6 years; mean interval between US and MRI 1.6 days, SD 1.8 days), 55% (48/87) had true-negative US, without CBD dilation/stones confirmed on MRI; 5% (4/87) had false-positive US showing CBD dilatation without stones, not confirmed on MRI; 33% (29/87) had true-positive US, with MRI confirming CBD dilatation; and 7% (6/87) had false-negative US, where MRI revealed CBD stones without dilatation (2 patients) and CBD dilatation with or without stones (4 patients). Patients with false-negative US had persistent or worsening symptoms, pancreatitis or SCD. The overall US false-negative rate was 17% (6/35). Normal-caliber CBD on US without stones had an NPV of 89% (48/54, 95% confidence interval: 0.77-0.96).

CONCLUSION

MRI adds little information in children with a sonographically normal CBD except in the setting of pancreatitis or worsening clinical symptoms. Further evaluation is warranted in children with elevated risk of stone disease.

Active learning-based multistage sequential decision-making model with application on common bile duct stone evaluation

Multistage sequential decision-making occurs in many real-world applications such as healthcare diagnosis and treatment. One concrete example is when the doctors need to decide to collect which kind of information from subjects so as to make the good medical decision cost-effectively. In this paper, an active learning-based method is developed to model the doctors' decision-making process that actively collects necessary information from each subject in a sequential manner. The effectiveness of the proposed model, especially its two-stage version, is validated on both simulation studies and a case study of common bile duct stone evaluation for pediatric patients.

Acute Gallstone Pancreatitis: If a Picture Is Worth a Thousand Words, How Many Images Do We Need?

Introduction Accurate diagnosis and prompt definitive management of choledocholithiasis are vital in acute gallstone pancreatitis. The sensitivity of detection of choledocholithiasis varies across imaging modalities. Magnetic resonance cholangiopancreatography (MRCP) is the most sensitive but may not be necessary, resulting in both delayed definitive management and increased costs. We aimed to evaluate the range of radiological investigations patients with acute gallstone pancreatitis underwent and the clinical appropriateness of MRCP when performed. Methods This was an observational study of patients diagnosed with acute gallstone pancreatitis between January 1, 2019 and November 30, 2021 in a district general hospital in London, UK. A detailed review of patient records, laboratory and radiological results, and endoscopic and/or operative intervention was undertaken. Results One hundred consecutive patients diagnosed with acute gallstone pancreatitis (median age 57 years) were included. Seventy-nine had a transabdominal ultrasound (USS), 46 had CT, and 40 patients had MRCP. The median waiting time for these investigations was 1, 0, and 4 days, respectively. Choledocholithiasis was identified in 21 patients (4 on USS, 5 on CT, and 12 on MRCP). As definitive management, 37% underwent endoscopic retrograde cholangiopancreatography, and 57% underwent laparoscopic cholecystectomy. A total of 19% of patients were readmitted with pancreatitis prior to definitive management. Conclusions First-line imaging investigations such as USS and CT can detect some cases of choledocholithiasis in patients with acute gallstone pancreatitis, but not all. Despite expenses in terms of cost and length of hospital stay, MRCP remains an essential resource to detect cases of choledocholithiasis not captured by USS or CT. We recommend establishing a guideline to streamline imaging in assessing acute gallstone pancreatitis.

The Laparoscopic Cholecystectomy and Common Bile Duct Exploration: A Single-Step Treatment of Pediatric Cholelithiasis and Choledocholithiasis

BACKGROUND

In recent years, complicated biliary tract diseases are increasingly diagnosed in children. Laparoscopic exploration of the common bile duct (LCBDE) followed by laparoscopic cholecystectomy has gained popularity in children. The aim of this study was to investigate the outcomes of LCBDE in children and compare them with the treatment outcomes of previously used endoscopic retrograde cholangiopancreatography (ERCP).

METHODS

From January 2000 to January 2022, a total of 84 children (78.5% female) underwent laparoscopic cholecystectomy with a median follow-up of 11.4 (IQR 8, 14) years. Of these, 6 children underwent laparoscopic cholecystectomy (LC) + ERCP and 14 children underwent LCBDE for choledochiothiasis. The primary end point of the study was the success of treatment in terms of the incidence of complications, recurrence rate, and rate of reoperation. Secondary endpoints were stone characteristics, presenting symptoms, duration of surgery, and length of hospital stay.

RESULTS

The majority of patients were female in both groups (83.5% vs. 85.7%), mostly overweight with a median BMI of 27.9 kg/m² and 27.4 kg/m², respectively. Obstructive jaundice, colicky pain, acute pancreatitis, and obstruction of the papilla were the most common symptoms in both groups. The majority of patients (68%) had one stone, whereas two or more stones were found in 32% of patients. The median diameter of the common bile duct was 9 mm in both groups. The procedure was successfully completed in all patients in the ERCP group. In the group of patients treated with LCBDE, endoscopic extraction of the stone with a Dormia basket was successfully performed in ten patients (71.4%), while in the remaining four patients (28.6%) the stones were fragmented with a laser because extraction with the Dormia basket was not possible. The median operative time was 79 min in the LCBDE group (IQR 68, 98), while it was slightly longer in the ERCP group, 85 min (IQR 74, 105) (p = 0.125). The length of hospital stay was significantly shorter in the LCBDE group (2 vs. 4 days, p = 0.011). No complications occurred in the LCBDE group, while two (40%) complications occurred in the ERCP group: pancreatitis and cholangitis (p = 0.078). During the follow-up period, no conversions, papillotomies, or recurrences were recorded in either group.

CONCLUSIONS

Exploration of the common bile duct and removal of stones by LCBDE is safe and feasible in pediatric patients for the treatment of choledocholithiasis. Through this procedure, choledocholithiasis and cholelithiasis can be treated in a single procedure without papillotomy or fluoroscopy. Compared with LC + ERCP, LCBDE is associated with a shorter hospital stay. The incidence of complications was rather low but not statistically significant.