Online Program

Return to main conference page

Friday, January 12
Fri, Jan 12, 10:30 AM - 12:15 PM
Crystal Ballroom A
Statistical Methods for Cancer Screening Using Risk Prediction Models, Polygenic Risk Scores, and Simulations

Risk-based screening for second primary lung cancer (303957)

Iona Cheng, Cancer Prevention Institute of California 
Scarlett Gomez, Cancer Prevention Institute of California 
*Summer S Han, Stanford University 
Sylvia Plevritis, Stanford University 
Gabriel Rivera, Kaiser Permanente Fresno Medical Center 
Martin Tammema¨gi, Brock University 
Heather Wakelee, Stanford University 

Keywords: risk-stratification, risk prediction model, lung caner, screening, CT screening, second primary lung cancer, competing risk analysis, cancer screening

Lung cancer (LC) causes significant mortality worldwide. With the widespread adoption of computed tomography (CT), the number of long-term LC survivors is expected to increase. How to guide screening for LC survivors with potential risks of second primary lung cancer (SPLC) is a critical public health concern. In this study, we develop a prediction model for SPLC using various competing risk methods and examine the risk-stratification ability of the prediction model. Decision curve analysis is used evaluate the clinical utility of the model for selecting eligibility criteria for screening, by calculating the model’s net benefit under varied risk thresholds. The stratification by deciles of estimated risk using the prediction model shows that the observed incidence of SPLC is significantly higher in the 10th decile (12.5%) versus the 1st decile group(2.9%). The decision curve analysis yields a range of risk thresholds (1%-11.5%) at which the clinical net benefit of the risk model is larger than those under hypothetical “all- screening” or “no-screening” scenarios. The risk-stratification approach for SPLC can be potentially useful in identifying LC survivors to be screened by CT.