Preoperative Risk Prediction of Futile Surgery in Gastrointestinal Tract Malignancies

Background

Gastrointestinal tract (GI) malignancies remain some of the most common and lethal cancers worldwide. While surgery remains curative, a cohort of patients will succumb to postoperative complications or early recurrence of disease rendering the surgical care futile. A major challenge is identifying patients who may not benefit from surgery i.e. “futile surgery”. The aim of this study was to develop a tool to predict the risk of futile surgery and identify high-risk cohorts for the most common malignancies in eight GI organs.

Methods

The National Cancer Database was used to identify patients with GI malignancies of the anus, colorectum, esophagus, gallbladder, liver, pancreas, and extrahepatic and intrahepatic bile ducts (2012 to 2020). The most common histological subtype of the malignancies was investigated for each organ. Futile surgery was defined as death within one year of surgery from any cause. Preoperative clinicopathological features including patient-related factors (e.g. age, sex, socioeconomic status, Charlson comorbidity index, facility type) and tumor-related factors (e.g. American Joint Commission on Cancer [AJCC] clinical T- and N-stage, and grade of tumor differentiation), and history of neoadjuvant therapy were considered as potential predictors. For each cancer type the study population was randomly split into a training (2/3rd) and testing (1/3rd) cohort. Multivariable logistic regression was performed on the training cohort to develop predictive models for each cancer type to predict the risk of futile surgery and high-risk class was defined based on predicted risk. The results were validated in the testing cohort.

Results

A total of 111,437 patients were identified and their clinicopathological data were analyzed. The rate of futile surgery ranged from 6.8% to 27.7% across cancer types. Predictors of futile surgery included both patient- and tumor-related factors. Predictive accuracy of the models varied across cancer types with the area under the curve (AUC) ranging from 0.620 to 0.728. High-risk categories were defined and the rate of futile surgery in this cohort ranged from 21.3% to 50.5% across the different types of cancer. The models demonstrated a similar performance in the test cohort.

Conclusions

A cohort of patients undergoing resection for GI malignancies experiences futile surgery. A predictive tool was developed to predict an individual’s risk of futile surgery. Integration of this tool in clinical practice could provide objective data to patients and providers to help in shared decision-making resulting in improved healthcare delivery and efficiency.