Abstract PR13: External validation of a risk prediction model for lung cancer among smokers

Early detection strategies for lung cancer may be improved by using valid risk prediction models to identify persons at highest risk for the disease. However, external validation of lung cancer risk prediction models has been limited. We sought to externally validate the PLCOM2012 model, which predicts the probability of lung cancer within six years on the basis of age, race, education, body mass index, chronic obstructive pulmonary disease, personal history of cancer, family history of lung cancer, and smoking status, quantity, duration, and quit years, in the Kaiser Permanente Northern California (KPNC) Research Program on Genes, Environment, and Health (RPGEH) cohort. To increase comparability to the populations of smokers used to initially develop and validate the PLCOM2012 model, we restricted our analysis to the 28,757 ever smokers ages 55 to 74 with no history of lung cancer, no history of other non-melanoma skin cancers in the prior five years, and complete data on all model predictors. For each person, the predicted probability of lung cancer risk was estimated with data ascertained from the RPGEH survey on all predictors except quit years, which was ascertained from electronic health records. Using KPNC Cancer Registry data, we identified 672 diagnosed with lung cancer within six years post-survey. Both calibration and discrimination were examined to assess model performance. Calibration was assessed by determining the mean absolute difference in observed and predicted probabilities of lung cancer for each decile of predicted risk. Discrimination was assessed by estimating the area under curve (AUC). The absolute difference in observed and predicted probabilities of lung cancer risk was generally small: Citation Format: Lori C. Sakoda, Laurel A. Habel, Khanh K. Thai, Charles P. Quesenberry, Jr. External validation of a risk prediction model for lung cancer among smokers. [abstract]. In: Proceedings of the AACR Special Conference: Improving Cancer Risk Prediction for Prevention and Early Detection; Nov 16-19, 2016; Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2017;26(5 Suppl):Abstract nr PR13.