Abstract A14: Small Tumor Diameter at Cure Threshold (< 20mm) Among Aggressive Lung Cancers Predicts both Chest X-ray and CT Screening Outcomes in a Novel Simulation Framework

Abstract

Abstract Background: The population distribution of lung cancer growth and tumor diameter at cure threshold is an important component of a model of the natural course of lung cancer. Methods: We simulate outcomes in the Mayo CT lung screening study using an existing simulation model of lung cancer progression calibrated to the Mayo Lung Project (males only) and evaluate the goodness of fit. We perform a likelihood-based analysis to determine whether the conditional probability of five-year lung cancer survival given tumor diameter at detection depends significantly on detection modality. A novel probabilistic model characterizes growth and tumor progression after transformation of a single malignant cell, governed by two parameters f and μ, the branching fraction and cellular mutation rate, respectively. We identify distinct sets of parameter pairs (f,μ), each of which satisfies a method of moments condition given a survival function governing detection in the absence of screening. We examine their goodness of fit with respect to simulated tumor size and five-year lung cancer survival among detected incident lung cancers in the MLP and Mayo CT studies. Results: An existing model of lung cancer progression under-predicts the number of advanced-stage incident lung cancers among males in the Mayo CT study (p-value = 0.004). The probability of five-year lung cancer survival given size at detection depends significantly on detection modality (p-value = 0.0312). Selected models having a median tumor diameter at cure threshold ranging from 16.7–27.0 mm predict both MLP and Mayo CT outcomes. Conclusions: The median lung tumor diameter at cure threshold among aggressive lung cancers may be small (< 20 mm).