01 Mar Lung Cancer: Screening Model Including Low-Dose CT Can Improve Risk Prediction
MedicalResearch.com Interview with:
Martin C. Tammemägi PhD
Cancer Care Ontario | Prevention & Cancer Control
Lung Cancer Screening Pilot for People at High Risk
Professor (Epidemiology) | Brock University Department of Health Sciences
MedicalResearch.com: What is the background for this study?
Response: Some prediction models can accurately predict lung cancer risk (probability of developing lung cancer during a specified time). Good model predictors include sociodemographic, medical and exposure variables. In recent years, low dose computed tomography (LDCT) lung cancer screening has become widespread in trials, pilots, demonstration studies, and public health practice.
It appears that screening results provides added valuable, independent predictive information regarding future lung cancer risk, aside from the lung cancers directly detected from the diagnostic investigations resulting from positive screens.
MedicalResearch.com: What are the main findings?
Response: In our study, we combined the risk scores from the validated PLCOm2012 lung cancer risk prediction model with LDCT screening results to produce a new lung cancer risk prediction model, the PLCO2012results model. For model development, the screening results data included the three screens in the National Lung Screening Trial (NLST) classified by the Lung-RADS system, with Lung-RADS 1 and 2 treated as normal and Lung-RADS 3 and 4 treated as abnormal. The series of three rounds of screening results were grouped into four roughly similar groups with regard to lung cancer risk. The prediction outcome was incidence lung cancers detected 1 to 4 years after the last screen. The PLCO2012results model was developed in the LSS component of the NLST and was validated in the ACRIN component of the NLST. The predictive accuracy of the PLCO2012results model was good and was significantly better than for the PLCOm2012, which excludes screening results. In the validation data, PLCO2012results had good discrimination with area under the curve (AUC) of 0.76 and good calibration with the 50th and 90th percentile of absolute error between observed and predicted risks being 0.0018 and 0.003 respectively.
Having three consecutive negative screens leads to lowering of the PLCOm2012 estimated risk, while having positive screens increases the PLCOm2012 estimated risk, and the more positive screens one has the greater the increase in PLCOm2012 risk. These risks refer to lung cancers detected starting one year after screening ended and exclude screen-detected lung cancers.
MedicalResearch.com: What should readers take away from your report?
Response: Information provided by the PLCOm2012 and PLCO2012results models can help inform decision-making for lung cancer screening eligibility and screening interval. As a generalization, individuals with positive screens not leading to an immediate lung cancer diagnosis should continue to receive annual as opposed to biennial screening. A sizeable proportion of NLST participants had baseline risk scores that were so high that even three consecutive negative screens did not lower their risk below thresholds used to enrol individuals into screening; i.e., they should continue with annual screening.
In contrast, there are some individuals in the NLST who have PLCOm2012 or PLCO2012results risk estimates that are so low as to indicate that they do not warrant annual screening. For example, of those individuals who were NLST criteria positive and had PLCOm2012 risks <2.6%, only 3 in 1000 (0.3%) had lung cancers diagnosed 1 to 4 years after screening ended. The report presents some implications of the study findings regarding offering the next annual screen based on PLCOm2012 and PLCO2012results estimates. The PLCO2012results model can also be used to identify individuals at relatively high risk of developing lung cancer (e.g., >10% three-year risk) for sampling into research trials.
MedicalResearch.com: What recommendations do you have for future research as a result of this work?
Response: The following questions and points need to be addressed:
- What aspects of a positive screen increase future risk of lung cancer? What are the mechanisms involved? What components of the Lung-RADS classification drive the increase risk of future lung cancer?
- Do lung cancers from individuals with high PLCO2012results risks, in which high-risk is driven by past positive screens, have similar biology’s and stage distributions as lung cancers identified through routine screening?
- The PLCOm2012 and PLCO2012results models identify some individuals who are positive by NLST, USPSTF or CMS criteria for enrolment into screening programs who are at such low risk that they are unlikely to benefit from screening and they should not be offered annual screening. Retrospective analyses and interim prospective analysis from the International Lung Screen Trial confirm this conclusion for the PLCOm2012. Further confirmation is required for the PLCO2012results model.
- When using the PLCO2012results to help guide decisions regarding the next screening, what risk thresholds should be considered in different populations?
- What is the cost-effectiveness of applying the PLCO2012results to guide lung cancer screening program designs?
MedicalResearch.com: Is there anything else you would like to add?
A spreadsheet calculator which combining the PLCOm2012 and PLCO2012results and provides individual’s risk estimation is available, free of charge for non-commercial users, for download at these links:
I am the developer of the PLCOm2012 lung cancer risk prediction model which is the basis for the PLCO2012results model. The PLCOm2012 and PLCO2012results models and calculators are available free of charge for non-commercial users. Brock University holds the rights to licence the PLCOm2012 for commercial use.
Tammemägi MC, ten Haaf K, Toumazis I, et al. Development and Validation of a Multivariable Lung Cancer Risk Prediction Model That Includes Low-Dose Computed Tomography Screening Results: A Secondary Analysis of Data From the National Lung Screening Trial. JAMA Netw Open. 2019;2(3):e190204. doi:10.1001/jamanetworkopen.2019.0204
The information on MedicalResearch.com is provided for educational purposes only, and is in no way intended to diagnose, cure, or treat any medical or other condition. Always seek the advice of your physician or other qualified health and ask your doctor any questions you may have regarding a medical condition. In addition to all other limitations and disclaimers in this agreement, service provider and its third party providers disclaim any liability or loss in connection with the content provided on this website.