Application of Lung-Screening Reporting and Data System Versus Pan-Canadian Early Detection of Lung Cancer Nodule Risk Calculation in the Alberta Lung Cancer Screening Study

Abstract

BackgroundFalse-positive scans and resultant needless early recalls can increase harms and reduce cost-effectiveness of low-dose CT (LDCT) lung cancer screening. How LDCT scans are interpreted and classified may impact these metrics. MethodsThe Pan-Canadian Early Detection of Lung Cancer risk calculator was used to determine nodule risk of malignancy on baseline screening LDCTs in the Alberta Lung Cancer Screening Study, which were then classified according to Nodule Risk Classification (NRC) categories and ACR Lung Screening Reporting and Data System (Lung-RADS). Test performance characteristics and early recall rates were compared for each approach. ResultsIn all, 775 baseline screens were analyzed. After a mean of 763 days (±203) of follow-up, lung cancer was detected in 22 participants (2.8%). No statistically significant differences in sensitivity, specificity, or area under the receiver operator characteristic curve occurred between the NRC and Lung-RADS nodule management approaches. Early recall rates were 9.2% and 9.3% for NRC and Lung-RADS, with the NRC unnecessarily recalling some ground glass nodules, and the Lung-RADS recalling many smaller solid nodules with low risk of malignancy. ConclusionPerformances of both the NRC and Lung-RADS in this cohort were very good with a trend to higher sensitivity for the NRC. Early recall rates were less than 10% with each approach, significantly lower than rates using the National Lung Screening Trial cutoffs. Further reductions in early recall rates without compromising sensitivity could be achieved by increasing the NRC threshold to 20% for ground glass nodules or by applying the nodule risk calculator with a 5% threshold to 6- to 10-mm solid nodules under Lung-RADS.