ES12.02 Defining High Risk

Abstract

Globally, lung cancer is the leading cause of cancer death. The National Lung Screening Trial and NELSON trial have shown that LDCT screening can reduce lung cancer mortality by ≥20%. This finding can have major public health impact. However, lung cancer screening is most effective when applied to high-risk individuals. Consequently, defining eligibility to select individuals for screening is important in determining screening programs’ effectiveness. This abstract discusses select topics in defining high risk and determining eligibility for lung cancer screening. Two general approaches to determining eligibility exist. The first uses categories of age, pack-years and smoking quit-years (CATage-PKYR-QY). This method does not quantify risk. Examples include NLST, USPSTF2013 and USPSTF2020, CMS and NELSON criteria. The second method uses mathematical models describing the relationship between predictors and lung cancer. and risk is quantified. Because this approach uses many predictors, models continuous variables, and in some models handles non-linear relationships, they have generally been found to be superior to CATage-PKYR-QY approaches. Examples include the Bach, PLCOm2012 and LCRAT models. 1-3 Comparisons of the USPSTF selection criteria versus model-risk-based approaches have been made by the CISNET Lung Group using micro-simulation modeling.4 Because modeling makes unrealistic assumptions, such as 100% participation and adherence, perhaps the best evidence to date instead comes from the International Lung Screening Trial, which prospectively enrolled >4000 individuals by USPSTF2013 criteria or PLCOm2012 risk.5 At PLCOm2012 ≥1.7%/6y both criteria selected the same number for screening. Both criteria were positive for 3018 individuals. Three cancers were detected in the USPSTF+/PLCOm2012- group, and 23 cancers were detected in the PLCOm2012+/USPSTF- group. The PLCOm2012 cancer detected rate was 18.5% (95%CI 11.5%-26.7%) higher than the USPSTF criteria. Given the average age at enrollment was 63.01 for USPSTF+ and 65.38 for PLCOm2012+ and assuming average death ages for USPSTF+ and PLCOm2012+ were 78 and 76 years (assumes greater comorbidities and competing deaths in PLCOm2012+ group), a crude calculation estimates 41.5 more life-years-gained per 1000 scans by the PLCOm2012 criteria. The USPSTF2013 and USPSTF2020 criteria lead to race and gender disparities, that for Whites versus Blacks is undone using the PLCOm2012 and for men versus women are reduced. The CREST study applied USPSTF criteria and PLCOm2012 to a community-based case series of 883 lung cancer patients in the University of Illinois Hospital and Health Sciences System in Chicago.6 For the USPSTF2020 in Whites and Blacks the criteria sensitivities were 75.4% and 70.6% and for PLCOm2012 ≥1.0%/6y (equivalent to USPSTF2020) were 82.5% and 82.8%, respectively. For the USPSTF2020 in men and women the criteria sensitivities were 71.8% and 64.6% and for PLCOm2012 ≥1.0%/6y were 80.9% and 76.7%. NLST results demonstrate that below a PLCOm2012 risk of 0.64%/6y there is no lung cancer mortality benefit conferred by LDCT screening. Indeed, there were more deaths in the LDCT arm. In the PLCO Trial there were 74,218 smokers. 4777 (6.4%) of them had PKYR ≥20 and <30 and QY≤15, representing the incremental group added by the USPSTF2020 criteria. Of these, 1025 (21.5%) have PLCOm2012 risks <0.65%/6y, and only 2 lung cancers were observed in this group (0.2%/6y). This sizeable group has risks so low that according to NLST results they have no benefit. Expansion of screening needs to be done effectively. Note that these results are based on real people and real data. Should biomarkers be used as the first stage for determining eligibility? To date, we are aware of only one biomarker test that has been evaluated in a RCT for selecting individuals for lung cancer screening: Early Diagnosis of Lung Cancer Scotland trial, which evaluated an autoantibody panel.7 In the intervention arm (N=6088), individuals who tested positive (n=598) received a LDCT screen; test-negative and controls (N=6121) received regular care. 71 lung cancer were detected in the control arm and only 56 were detected in intervention arm. Sensitivity was only 32.1%. The cancer detection rate was 0.9% in 2 years and early stage occurred in 41% of cases. These findings compare poorly against pilots using PLCOm2012 for enrollment, for example, Manchester Lung Health Check had 3.0% detection in 1 year and 80% early stage8; Ontario Health (Cancer Care Ontario) pilot had 1.9% in 1 year and 71% early stage9. Cost-effectiveness analysis found no scenario in which the biomarker was more favourable. Because of greater accuracy and cheaper cost, the PLCOm2012 dominated the biomarker. Thus, in such scenarios, biomarker combined with prediction model may be superior, or biomarker may be useful to identify those at high risk who do not qualifying by other criteria. 1. Bach PB, Kattan MW, Thornquist MD, et al. Variations in lung cancer risk among smokers. Journal of the National Cancer Institute 2003;95:470-8. 2. Tammemagi MC, Katki HA, Hocking WG, et al. Selection criteria for lung-cancer screening. The New England journal of medicine 2013;368:728-36. 3. Katki HA, Kovalchik SA, Berg CD, Cheung LC, Chaturvedi AK. Development and Validation of Risk Models to Select Ever-Smokers for CT Lung Cancer Screening. JAMA 2016. 4. Meza R, Jeon J, Toumazis I, et al. Evaluation of the Benefits and Harms of Lung Cancer Screening With Low-Dose Computed Tomography: A Collaborative Modeling Study for the U.S. Preventive Services Task Force. Rockville, MD.2020. 5. Lam S, Myers R, Ruparel M, et al. PL02.02 - Lung Cancer Screenee Selection by USPSTF Versus PLCOm2012 Criteria – Interim ILST Findings. Journal of Thoracic Oncology 2019;Vol 14, Issue 10, S4. 6. Pasquinelli MM, Tammemagi MC, Kovitz KL, et al. Risk Prediction Model Versus United States Preventive Services Task Force Lung Cancer Screening Eligibility Criteria: Reducing Race Disparities. J Thorac Oncol 2020. 7. Sullivan FM, Mair FS, Anderson W, et al. Earlier diagnosis of lung cancer in a randomised trial of an autoantibody blood test followed by imaging. Eur Respir J 2020. 8. Crosbie PA, Balata H, Evison M, et al. Implementing lung cancer screening: baseline results from a community-based 'Lung Health Check' pilot in deprived areas of Manchester. Thorax 2019;74:405-9. 9. Darling GE, Tammemagi MC, Schmidt H, et al. Organized Lung Cancer Screening Pilot: Informing a Province-wide Program in Ontario, Canada. The Annals of thoracic surgery 2020.