Dataset Information

Prediction Model of Anastomotic Leakage Among Esophageal Cancer Patients After Receiving an Esophagectomy: Machine Learning Approach.

ABSTRACT:

Background

Anastomotic leakage (AL) is one of the severe postoperative adverse events (5%-30%), and it is related to increased medical costs in cancer patients who undergo esophagectomies. Machine learning (ML) methods show good performance at predicting risk for AL. However, AL risk prediction based on ML models among the Chinese population is unavailable.

Objective

This study uses ML techniques to develop and validate a risk prediction model to screen patients with emerging AL risk factors.

Methods

Analyses were performed using medical records from 710 patients who underwent esophagectomies at the National Clinical Research Center for Cancer between January 2010 and May 2015. We randomly split (9:1) the data set into a training data set of 639 patients and a testing data set of 71 patients using a computer algorithm. We assessed multiple classification tools to create a multivariate risk prediction model. Our ML algorithms contained decision tree, random forest, naive Bayes, and logistic regression with least absolute shrinkage and selection operator. The optimal AL prediction model was selected based on model evaluation metrics.

Results

The final risk panel included 36 independent risk features. Of those, 10 features were significantly identified by the logistic model, including aortic calcification (OR 2.77, 95% CI 1.32-5.81), celiac trunk calcification (OR 2.79, 95% CI 1.20-6.48), forced expiratory volume 1% (OR 0.51, 95% CI 0.30-0.89); TLco (OR 0.56, 95% CI 0.27-1.18), peripheral vascular disease (OR 4.97, 95% CI 1.44-17.07), laparoscope (OR 3.92, 95% CI 1.23-12.51), postoperative length of hospital stay (OR 1.17, 95% CI 1.13-1.21), vascular permeability activity (OR 0.46, 95% CI 0.14-1.48), and fat liquefaction of incisions (OR 4.36, 95% CI 1.86-10.21). Logistic regression with least absolute shrinkage and selection operator offered the highest prediction quality with an area under the receiver operator characteristic of 72% in the training data set. The testing model also achieved similar high performance.

Conclusions

Our model offered a prediction of AL with high accuracy, assisting in AL prevention and treatment. A personalized ML prediction model with a purely data-driven selection of features is feasible and effective in predicting AL in patients who underwent esophagectomy.

SUBMITTER: Zhao Z

PROVIDER: S-EPMC8367102 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundAnastomotic leakage (0-30%) after esophagectomy is a severe complication and is associated with considerable morbidity and mortality. The aim of this study was to determine which treatment for anastomotic leakage after esophagectomy have the best clinical outcome, based on the currently available literature.MethodsA systematic literature search was performed in Medline, Embase, and Web of Science until April 2017. All studies reporting on the specific treatment of cervical or intrathoracic anastomotic leakage following esophagectomy with gastric tube reconstruction for esophageal or cardia cancer were included. The primary outcome parameter was postoperative mortality. Methodological quality was assessed by the Newcastle-Ottawa Quality Assessment Scale.ResultsNineteen retrospective cohort studies including 273 patients were identified. Methodological quality of all studies was poor to moderate. Mortality rates of intrathoracic anastomotic leakages in the treatment groups were as follows: conservative (14%), endoscopic stent (8%), endoscopic drainage (8%), endoscopic vacuum-assisted closure system (0%), and surgery treatment group (50%). Mortality rates of cervical anastomotic leakages in the treatment groups were as follows: conservative (8%), endoscopic stent (29%), and endoscopic dilatation (0%).DiscussionDue to small cohorts, heterogeneity between studies, and lack of data regarding leakage characteristics, no evidence supporting a specific treatment for anastomotic leakage after esophagectomy was found. A severity score based on leakage characteristics instead of treatment given is essential for determining the optimal treatment of anastomotic leakage. In the absence of robust evidence-based treatment guidelines, we suggest customized treatment depending on sequelae of the leak and clinical condition of the patient. PrDepartment of Surgery, Radboudumc, P.O.B. 9101/618 NLactical advices are provided.Trial registrationRegistration number PROSPERO: CRD42016032374.

Project description:BackgroundEarly diagnosis of anastomotic leakage (AL) after esophageal resection is crucial for the successful management of this complication. Inflammatory serological markers are indicators of complications during the postoperative course. The aim of the present study was to evaluate the prognostic value of routine inflammatory markers to predict anastomotic leakage after transthoracic esophageal resection.MethodsData from all consecutive patients undergoing transthoracic esophageal resection between January 2010 and December 2016 were analyzed from a prospective database. Besides clinicodemographic parameters, C-reactive protein, white blood cell count and albumin were analyzed and the Noble/Underwood (NUn) score was calculated to evaluate their predictive value for postoperative anastomotic leakage. Diagnostic accuracy was measured by sensitivity, specificity, and negative and positive predictive values using area under the receiver operator characteristics curve.ResultsOverall, 233 patients with transthoracic esophageal resection were analyzed, 30-day mortality in this group was 3.4%. 57 patients (24.5%) suffered from AL, 176 patients were in the AL negative group. We found significant differences in WBCC, CRP and NUn scores between patients with and without AL, but the analyzed markers did not show an independent relevant prognostic value. For CRP levels below 155 mg/dl from POD3 to POD 7 the negative predictive value for absence of AI was > 80%. Highest diagnostic accuracy was detected for CRP levels on 4th POD with a cut-off value of 145 mg/l reaching negative predictive value of 87%.ConclusionsIn contrast to their prognostic value in other surgical procedures, CRP, WBCC and NUn score cannot be recommended as independent markers for the prediction of anastomotic leakage after transthoracic esophageal resection. CRP is an accurate negative predictive marker and discrimination of AL and no-AL may be helpful for postoperative clinical management. Trial registration The study was approved by the local ethical committee (S635-2013).

Project description:BackgroundAnastomotic leakage (AL) has always been one of the most serious complications of esophagectomy with gastric conduit reconstruction. There are many strong risk factors for AL in clinical practice. Notably, the tension at the esophagogastric anastomosis and the blood supply to the gastric conduit directly affect the integrity of the anastomosis. However, there has been a lack of quantitative research on the tension and blood supply of the gastric conduit. Changes in extracellular matrix collagen reflect tension and blood supply, which affect the quality of the anastomosis. This study aimed to establish a quantitative collagen score to describe changes in the collagen structure in the extracellular matrix and to identify patients at high risk of postoperative AL.MethodsA retrospective study of 213 patients was conducted. Clinical and pathological data were collected at baseline. Optical imaging of the "donut" specimen at the anastomotic gastric end and collagen feature extraction were performed. Least absolute shrinkage and selection operator (LASSO) regression models were used to select the significant collagen features, compute collagen scores, and validate the predictive efficacy of the collagen scores for ALs.ResultsLASSO regression analysis revealed three collagen-related parameters in the gastric donuts: histogram mean, histogram variance, and histogram energy. Based on this analysis, we established a formula to calculate the collagen score. The results of the univariate analysis revealed significant differences in the preoperative low albumin values (P=0.002) and collagen scores between the AL and non-AL groups (P=0.001), while the results of the multivariate analysis revealed significant differences in the collagen scores between the AL and non-AL groups (P=0.002). The areas under the curve (AUCs) of the experimental and validation cohorts were 0.978 [95% confidence interval (CI): 0.931-0.996] and 0.900 (95% CI: 0.824-0.951), respectively.ConclusionsThe collagen score established herein was shown to be related to AL and can be used to predict AL in patients who underwent esophagectomy.