A risk model for lung cancer incidence

Area of Science:

• Epidemiology
• Biostatistics
• Oncology

Background:

• Accurate lung cancer risk models are crucial for targeted screening and chemoprevention trials.
• Existing models may not fully capture individual risk, particularly in diverse populations.

Purpose of the Study:

• To develop and validate a novel lung cancer incidence risk model using prospective cohort data.
• To assess the predictive accuracy of smoking history, occupational/environmental exposures, and genetic factors.

Main Methods:

• Utilized survival analysis on a large cohort (169,035 ever-smokers) from the European Prospective Investigation into Cancer and Nutrition (EPIC) study.
• Developed separate models for current and former smokers, incorporating age at smoking initiation and smoking duration.
• Validated the model using an independent test set, measuring accuracy with the area under the receiver operator characteristic curve (AUC).

Main Results:

• The developed model, using smoking information alone, achieved a high AUC of 0.843 (0.810-0.875) in ever-smokers, outperforming the Bach model (AUC 0.775).
• Additional risk factors (occupational/environmental exposures, SNPs) showed negligible impact on predictive accuracy.
• Prediction for never-smokers was poor, indicating the model's primary utility for individuals with smoking history.

Conclusions:

• A straightforward, generalizable lung cancer risk model based on lifetime smoking exposure demonstrates high predictive accuracy in ever-smokers.
• The model's performance highlights the paramount importance of detailed smoking history in lung cancer risk assessment.
• This tool can aid in prioritizing individuals for lung cancer screening and chemoprevention interventions.