Low positive predictive value of computed tomography screening for lung cancer irrespective of commonly employed definitions of target population

• Published in: International journal of cancer Vol. 149; no. 1; pp. 58 - 65
• Main Authors: Krilaviciute, Agne, Brenner, Hermann
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01.07.2021; Wiley Subscription Services, Inc
• Subjects: Age; Cancer; Clinical trials; Computed tomography; Epidemiology; Eurobarometer; Eurobarometer 87.1 (ZA6861 v1.2.0); Lung cancer; Medical research; pack-years; preclinical cancer; predictive value; Risk assessment; screenin; screening; Smoking; Tomography; lung cancer; screening; pack-years; predictive value; preclinical cancer
• ISSN: 0020-7136; 1097-0215; 1097-0215
• DOI: 10.1002/ijc.33522

Screening for lung cancer (LC) by low‐dose computed tomography (LDCT) has been demonstrated to reduce LC mortality in randomized clinical trials (RCTs), and its implementation is in preparation in many countries. However, definition of the target population, which was based on various combinations of age ranges and definitions of heavy smoking in the RCTs, is subject to ongoing debate. Using epidemiological data from Germany, we aimed to estimate prevalence of preclinical LC and positive predictive value (PPV) of LDCT in potential target populations defined by age and smoking history. Populations aged 50 to 69, 55 to 69, 50 to 74 and 55 to 79 years were considered in this analysis. Sex‐specific prevalence of preclinical LC was estimated using LC incidence data within those age ranges and annual transition rates from preclinical to clinical LC obtained by meta‐analysis. Prevalence of preclinical LC among heavy smokers (defined by various pack‐year thresholds) within those age ranges was estimated by combining LC prevalence in the general population with proportions of heavy smokers and relative risks for LC among them derived from epidemiological studies. PPVs were calculated by combining these prevalences with sensitivity and specificity estimates of LDCT. Estimated prevalence of LC was 0.3% to 0.5% (men) and 0.2% to 0.3% (women) in the general population and 0.8% to 1.7% in target populations of heavy smokers. Estimates of PPV of LDCT were <20% for all definitions of target populations of heavy smokers. Refined preselection of target populations would be highly desirable to increase PPV and efficiency of LDCT screening and to reduce numbers of false‐positive LDCT findings. What's new? Lung cancer screening using low‐dose computed tomography (LDCT) is nearing implementation in various countries. However, defining the target population, particularly which individuals will benefit most from LDCT screening, remains a critical issue. Here, using data from Germany, the authors investigated different combinations of age range and smoking history and examined how well these combinations estimated preclinical lung cancer prevalence. Analyses show that expected positive predictive values of lung cancer screening by LDCT were below 20 percent for heavy smokers, within all age ranges investigated. The findings highlight the need for improved target population preselection for lung cancer screening by LDCT.