Model may help identify patients at greater risk for esophageal cancer
Click Here to Manage Email Alerts
GENEVA — A validated prediction model showed the potential to identify people at greater risk for high-grade intraepithelial neoplasia or esophageal cancer, according to study results presented at World Cancer Congress.
Results suggest the approach could reduce screening demand by more than half, according to researcher Ru Chen, MD, assistant professor at Cancer Hospital Chinese Academy of Medical Sciences,
“This model might be used to guide population-based esophageal cancer screening in high-risk areas,” Chen said during a virtual presentation.
Esophageal cancer is a global health challenge. In 2020, it ranked as the seventh most common type of cancer and the sixth leading cause of cancer death worldwide. More than 50% of cases occur in China.
“Previous studies showed endoscopic screening is effective for reducing incidence and mortality from esophageal cancer, but [it is] limited due to high cost, low detection rate and low compliance,” Chen said. “Risk stratification could be an effective strategy to improve efficiency of esophageal cancer screening.”
However, such prediction tools are limited in certain settings, Chen said.
Her team conducted a prior systematic review that examined 20 esophageal cancer risk prediction models. Investigators observed high risk for bias in all models, which limits their application in practice, Chen said.
Chen and colleagues conducted a multicenter cross-sectional study to develop a risk prediction model in hopes of determining which individuals may benefit most from endoscopy screening.
The analysis included 86,853 people aged 40 to 69 years (median, 53.2 years; range, 47.1-59.7; 58.6% women) who underwent endoscopic screening in areas at high risk for esophageal cancer between 2005 and 2015.
Researchers assigned eligible participants to a derivation cohort (Feicheng; n = 53,579) or validation cohort (Linzhou; n = 33,274) according to regional distribution.
Detection of esophageal cancer or high-grade intraepithelial neoplasia served as the key outcome measure.
Investigators used univariate and multivariate analyses to identify predictors for esophageal cancer and precancerous lesions, and they used logistic regression for model development.
They used area under the receiver operating characteristic curve to estimate discrimination, and they used the Hosmer-Lemeshow test to assess calibration. They performed 10-fold cross-validation to internally validate the model, then used the Linzhou cohort for external validation.
Baseline screening identified 832 people with high-grade intraepithelial neoplasia and 332 people with esophageal cancer.
The final prediction model incorporated six variables: age (multivariate coefficient, 1.12; 95% CI, 1.11-1.14), sex (male vs. female, 1.67; 95% CI, 1.34-2.09), alcohol use (yes vs. no, 1.33; 95% CI, 1.07-1.65), smoking history (yes vs. no, 1.39; 95% CI, 1.11-1.74), BMI ( 24 kg/m2 vs. < 24 kg/m2, 0.68; 95% CI, 0.57-0.81) and family history of upper gastrointestinal cancer (yes vs. no, 1.31; 95% CI, 1.06-1.63).
The model generated an area under the curve of 0.778 (95% CI, 0.76-0.79) in the development set, an area under the curve of 0.775 after internal validation, and an area under the curve of 0.714 (95% CI, 0.69-0.73) in the external validation cohort.
Performing endoscopy on individuals deemed by the model to be at medium risk or high risk detected 88.1% of high-grade intraepithelial neoplasia and esophageal cancer cases in the training set and 81.4% of cases in the validation set. This approach reduced screening demand by 51%, Chen said.
Perspective
Back to TopProfessor David Collingridge, PhD
These are very important findings. The prevalence of esophageal cancer is extremely high, so this is exactly the type of study that should be done in that part of the word. As we have heard throughout the congress, prevention definitely is the best course of action.
Collapse
Chen R, et al. Abstract 1765. Presented at: World Cancer Congress; Oct. 18-20, 2022; Geneva.
Read more about
Click Here to Manage Email Alerts