Chest X-ray screening improves outcome in lung cancer. A reappraisal of randomized trials on lung cancer screening.

It is believed that population-based screening for cancer should be advocated only when screening reduces disease-specific mortality. Four randomized controlled studies on lung cancer screening have been conducted in male cigarette smokers, and none has demonstrated reduced mortality. Accordingly, no organization that formulates screening policy advocates any specific early detection strategies for lung cancer. Yet, despite this public policy against screening, there is considerable evidence that chest x-ray screening is associated with earlier detection and improved survival. Two randomized trials, the Memorial Sloan-Kettering and Johns Hopkins Lung Projects, were specifically designed to evaluate the effectiveness of sputum cytologic study. Both evaluated the efficacy of the addition of sputum cytologic studies to annual chest radiographs, and both demonstrated that cytologic study did not favorably influence outcome. All individuals in experimental and control groups in both studies had annual chest radiographs. Because survival rates observed in both studies were about three times higher than predicted, based either on the National Cancer Institute's Surveillance Epidemiology and End Results database or based on the American Cancer Society's annual Cancer Statistics, raises the possibility that the periodic chest radiographs performed in all patients in both studies contributed to an improved outcome. In the Mayo Lung Project and in the Czechoslovak study on lung cancer screening, the experimental groups underwent a program of relatively intensive and regular rescreening with chest radiographs and sputum cytologic study, while the control groups underwent either less-frequent rescreening or no rescreening. In both studies, the screened groups achieved meaningful improvements in stage distribution, resectability, and survival. However, increases in cumulative incidence of lung cancer in the experimental group in both studies (which in the Mayo Lung Project reached statistical significance) prevented significant improvements in survival from translating into corresponding reductions in mortality. The possibility that screening may be associated with lung cancer "overdiagnosis" has been widely postulated to account for higher survival and incidence rates and equivalent mortality rates. However, analysis of autopsy information and of disease outcome in individuals with screen-detected early stage lung cancer who do not undergo surgical resection strongly supports the conclusion that screening does not lead to overdiagnosis of lung cancer. Similarly, lead-time and length bias do not adequately account for the differences in cumulative incidence observed in the Mayo and Czech studies.(ABSTRACT TRUNCATED AT 400 WORDS)