摘要

Objectives: Unselected screening for oesophageal adenocarcinoma (OAC) is not justified due to the low absolute risk in the general population. This study aimed to evaluate a risk prediction model in identifying high-risk individuals who might be considered for targeted screening.
Methods: A population-based cohort of 62,576 participants was recruited in 1995-1997 in Nord-Trondelag County, Norway (HUNT) and followed up until 31 December 2015. A model for predicting individuals' absolute risk of OAC was developed using competing-risk regression. The Lorenz curve was used to assess the concentration of OAC patients in high-risk individuals and the feasibility of targeted screening based on individual risk assessment.
Results: During 1,085,137 person-years of follow-up, 29 incident cases of OAC occurred. The model included risk factors for OAC, in which male sex, older age, gastro-oesophageal reflux symptoms, obesity, and tobacco smoking predicted higher risk of OAC. The area under the receiver operating characteristic curve for 10-year risk of OAC was 0.71 (95% confidence interval 0.57-0.85) and for 15-year risk was 0.84 (95% confidence interval 0.76-0.91) after 10-fold cross-validation, with good agreements between observed and predicted risks. The Lorenz curve indicated that 33% of all OAC cases would have occurred in the 5% of the population with the highest risks within 15 years, and 61% of all cases in the top 10% of the population.
Conclusions: Individual risk assessment based on known risk factors for OAC has the potential to identify a selected high-risk group of individuals who may benefit from screening for early detection.

  • 出版日期2018-6