gms | German Medical Science

Introduction: The identification and enumeration of persons at elevated risk to develop lung cancer is a necessary prerequisite to the investigation and institution of screening procedures, e.g. with low-dose CT. Validated risk prediction models provide the means to combine comprehensive information on a person’s smoking history with personal clinical and familial disposition to lung diseases. The PLCOm2012 model showed good calibration in the prospective German EPIC cohorts [1]. Compared to fixed eligibility criteria risk estimates have been shown to improve sensitivity and specificity in identifying groups at high risk [1], [2]. In a modified version the PLCOall2014 model provides risk estimates for non-smokers [2]. This allows investigating the distribution and the potential for high-risk stratification of lung cancer risk estimates in a general population setting.

Methods: We applied the PLCOall2014 risk prediction model based on age, sex, smoking duration, smoking intensity, COPD, BMI, education, and cancer history to data from the population representative survey of the DEGS-study. With the help of population weights we assessed the distribution of lung cancer risk estimates in the German population in general and in strata of age, gender and smoking exposure.

Results: Risk estimates to develop lung cancer within 6- years were varying between 0 and 16.8%. Risk in women was on average lower (highest quintile D8 =0.2%) than in men (D8= 0.6%). Average age-class specific estimates were 0.02% (age group 25-29) to 2.7% (age 75+) in men, and 0.01% (age group 25-29) to 0.7% (age 75+) in women. Approximately 11% (8%) of all men and 4% (3%) of women had risk estimates beyond 1.5% (2%) respectively, which are risk limits currently discussed as potential inclusion criteria for lung-cancer screening [2], [3]. Persons currently smoking on a daily basis had mean risk 0.5% (D8=1.5%), while persons currently smoking only occasionally had mean risk 0.008% (D8= 0.006%), much lower than former smokers with average 1.0% risk (D8=1.5%).

Discussion: The potential benefit and cost of preventive measures are directly related to the size of high-risk groups and to their estimated risk. Here cost not only relates to the financial burden of screening persons who will never develop the disease, but to the general burden of alarm and harm from further diagnostic measures for persons with false-positive findings. Through the concept of net-benefit analysis well calibrated risk estimates may directly be related to the diagnostic quality of screening methods in terms of true and false positive rates. In this context it is relevant to assess the size of high-risk population groups and their risk, e.g. as highest quintile estimates. According to our analyses only beyond the age of 60 40% (33%) of men and 10% (8%) of women have risk estimates above 1.5% (2%), and they might thus be considered as eligible candidates for lung-cancer screening. At lower ages less than 5% have 6-year risk estimates to develop lung cancer beyond 1%. Higher age is associated with higher rates of comorbidity and competing risks, which need to be considered when eligibility criteria for lung cancer screening are discussed.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.