Lung Cancer: More Efficient CT Screening Model Proposed

dr_Martin_C_Tammemä Interview with:
Dr. Martin C. Tammemägi

Professor (Epidemiology), Brock University
Department of Health Sciences
St. Catharines, Ontario, Canada L2S 3A1

Medical Research: What is the background for this study? What are the main findings?

Dr. Tammemägi: Lung cancer is the leading cause of cancer death in North America and the world. Lung cancer survival following diagnosis is generally poor, in the range of 10% to 15%, and has improved little over the last four decades. The biggest recent breakthrough for reducing lung cancer mortality came with the findings of the National Lung Screening Trial (NLST), a large, well-conducted randomized screening trial, which demonstrated that low dose computed tomography (LDCT) screening versus chest X-ray (CXR) screening can reduce lung cancer mortality by 20%. Currently, most guidelines for selecting screenees for lung screening use the NLST enrolment criteria of 30 or more pack-years smoked, former smokers must have quit smoking within 15 years and ages between 55 and 74, or use a variant of the NLST criteria. The US Preventive Services Task Force (USPSTF) essentially recommends using the NLST criteria but extended the inclusion age to 80 years.

The current study applied the PLCOm2012 lung cancer risk prediction model1 to NLST data and identified that the risk above which lung cancer mortality is consistently lower in the LDCT arm compared to the CXR arm, is ≥1.51% 6-year risk (65th percentile). The USPSTF and the PLCOm2012 risk ≥0.0151 criteria were then applied to the Prostate Lung Colorectal and Ovarian Cancer Screening Trial (PLCO) intervention arm smokers (the PLCOm2012 was developed in PLCO controls) to determine who would be selected for lung cancer screening. Compared to USPSTF criteria, the PLCOm2012 risk ≥0.0151 threshold selected 8.8% fewer individuals, but identified 12.4% more lung cancers (sensitivity 80.1% vs. 71.2%), and had fewer false positives (specificity 66.2% vs. 62.7%). 26% of smokers who were USPSTF criteria positive had risks below the PLCOm2012 risk ≥0.0151 threshold. Of PLCO former smokers who quit more than 15 years ago, 8.5% had PLCOm2012 risk ≥0.0151, suggesting that they might benefit from screening (2.9% of them developed lung cancer in 6 year). None of 65,711 never-smokers in the PLCO had PLCOm2012 risk ≥0.0151, indicating that never-smokers should not be screened. Individuals age ≥65–80 years had significantly higher risks and more lung cancers than those 55-64 years.

Medical Research: What should clinicians, patients and public health officials take away from your report?

Dr. Tammemägi: Selecting ever-smokers for lung cancer screening using the PLCOm2012 and risk threshold ≥0.0151 is more efficient than using the USPSTF criteria. Lung cancer screening programs should be systematic and be undertaken in institutions in which high quality care and follow-up can be provided. Such programs should consider using the PLCOm2012 risks ≥0.0151 criterion for enrolling screenees.

Medical Research: What recommendations do you have for future research as a result of this study?

Dr. Tammemägi: The superior efficiency of selecting individuals for lung cancer screening using PLCOm2012 risks ≥0.0151 over USPSTF and NLST criteria needs to be validated in different populations. One can test this by applying both USPTF or NLST criteria and the PLCOm2012 risk ≥0.0151 criteri a together and enrolling individuals based on positivity by either criteria. Once over 7,000 individuals have been enrolled, statistically significant differences should become apparent in size of samples selected, numbers of lung cancers identified and positive predictive values.


  1. Tammemagi MC, Katki HA, Hocking WG, et al. Selection criteria for lung-cancer screening. The New England journal of medicine 2013;368:728-36.


Tammemägi MC, Church TR, Hocking WG, Silvestri GA, Kvale PA, et al. (2014) Evaluation of the Lung Cancer Risks at Which to Screen Ever- and NeverSmokers: Screening Rules Applied to the PLCO and NLST Cohorts. PLoS Med 11(12): e1001764. doi:10.1371/journal.pmed.1001764