Lung Cancer Screening in Clinical Practice: A Five-year Review of Frequency and Predictors of Lung Cancer in the Screened Population

Evaluate (1) frequency, type, and lung cancer stage in a clinical lung cancer screening (LCS) population, and (2) association between patient characteristics and Lung-RADS with lung cancer diagnosis.This retrospective study enrolled individuals undergoing LCS between 1/1/2015 - 6/30/20. Individuals' sociodemographic characteristics, Lung-RADS scores, pathology-proven lung cancers, and tumor characteristics were determined via electronic health record and our health system's tumor registry. Associations between the outcome of lung cancer diagnosis within one year after LCS and covariates of sociodemographic characteristics and Lung-RADS score were determined using logistic regression.Of 3,326 individuals receiving 5,150 LCS exams, 102 (3.1%) were diagnosed with lung cancer within one year of LCS; most of these cancers were screen-detected (97/102, 95.1%). Over the study period, there were 118 total LCS-detected cancers in 113 (3.4%) individuals. Most LCS-detected cancers were adenocarcinomas (62/118, 52%), 55.9% (65/118) were Stage I, and 16.1% (19/118) were Stage IV. The sensitivity, specificity, positive predictive value, and negative predictive value of Lung-RADS in diagnosing lung cancer within one year of LCS was 93.1%, 83.8%, 10.6%, and 99.8%, respectively. At multivariable analysis controlling for sociodemographic characteristics, only the Lung-RADS score was associated with lung cancer (OR for one unit increase in Lung-RADS, 95% CI: 4.68, 3.87-5.78).The frequency of LCS-detected lung cancer and Stage IV cancers was higher than reported in the National Lung Screening Trial. Though Lung-RADS was a significant predictor of lung cancer, the PPV of Lung-RADS is relatively low, implying opportunity for improved nodule classification.