MA10.03 Balance Between Decreased False Positives and Delayed Diagnosis in Lung Cancer Screening

Abstract

False positives in CT screening for lung cancer can be reduced by raising size thresholds for the definition of a positive result. The higher the threshold, the lower the false positive rate. A higher threshold will, however, delay the diagnosis in those cases which would have been detected earlier using a smaller size threshold. We wanted to determine the balance between these two competing considerations. We reviewed the I-ELCAP database to determine the rate of positive results for screening using increasing size thresholds for the baseline round of screening. We then determined how many cancers would have a delayed diagnosis based on higher size thresholds as well as the histology of these cancers. We then estimated the change in the curability of the cancers due to the delayed diagnoses based on the predicted size one year later. Moving from a 6.0 mm size threshold to initiate work-up to 7.0, 8.0, and 9.0 mm, we found that the rate of positive results decreased from 102.1/1000 (10.2%) to 70.9/1000 (7.1%), 51.0/1000 (5.1%), and 39.6/1000 (4.0%). Assuming that 0.56% of the screening participants would be diagnosed of lung cancer within 1 year of baseline screening and potentially the higher thresholds would result in a delay in diagnosis of a small number (5-6%) of cancers until the next annual repeat screening, the number of cancers that would have delayed diagnosis with these higher thresholds was 0.28, 0.33 and 0.38 per 1000 screening participants. Assuming these cancers were growing with a volume doubling time (VDT) of 200 days, the anticipated absolute decrease in curability of a six-month delay in diagnosis was 1.6%-2.4%, these represent 0.005, 0.006, and 0.007 additional lung cancer deaths per 1000 screening participants for 7.0, 8.0, and 9.0 mm size threshold, respectively. The magnitude of the potential loss of life due to increasing the threshold for positive results, thus delaying the diagnosis, was estimated and ranged between 0.005-0.007 deaths per thousand screening participants. This is particularly meaningful when considering the development of screening programs in countries with limited resources, as the decrease in the estimated cure rate for higher threshold of positive result should be compared with the deaths from lung cancer when not performing any screening at all which is in the range of 100 deaths per thousand screening participants.