A population-based cohort study of chest x-ray screening in smokers: lung cancer detection findings and follow-up.

Lorenzo Dominioni,Vincenzo D'Ambrosio,Antonio Paddeu,Fausto Sessa,William Mantovani,Nicola Rotolo,Albino Poli,Massimo Paolucci,Salvatore Pisani,Andrea Imperatori,Valentina Conti

doi:10.1186/1471-2407-12-18

Abstract

BackgroundCase-control studies of mass screening for lung cancer (LC) by chest x-rays (CXR) performed in the 1990s in scarcely defined Japanese target populations indicated significant mortality reductions, but these results are yet to be confirmed in western countries. To ascertain whether CXR screening decreases LC mortality at community level, we studied a clearly defined population-based cohort of smokers invited to screening. We present here the LC detection results and the 10-year survival rates.MethodsThe cohort of all smokers of > 10 pack-years resident in 50 communities of Varese, screening-eligible (n = 5,815), in July 1997 was invited to nonrandomized CXR screening. Self-selected participants (21% of cohort) underwent screening in addition to usual care; nonparticipants received usual care. The cohort was followed-up until December 2010. Kaplan-Meier LC-specific survival was estimated in participants, in nonparticipants, in the whole cohort, and in an uninvited, unscreened population (control group).ResultsOver the initial 9.5 years of study, 67 LCs were diagnosed in screening participants (51% were screen-detected) and 178 in nonparticipants. The rates of stage I LC, resectability and 5-year survival were nearly twice as high in participants (32% stage I; 48% resected; 30.5% 5-year survival) as in nonparticipants (17% stage I; 27% resected; 13.5% 5-year survival). There were no bronchioloalveolar carcinomas among screen-detected cancers, and median volume doubling time of incidence screen-detected LCs was 80 days (range, 44-318), suggesting that screening overdiagnosis was minimal. The 10-year LC-specific survival was greater in screening participants than in nonparticipants (log-rank, p = 0.005), and greater in the whole cohort invited to screening than in the control group (log-rank, p = 0.001). This favourable long-term effect was independently related to CXR screening exposure.ConclusionIn the setting of CXR screening offered to a population-based cohort of smokers, screening participants who were diagnosed with LC had more frequently early-stage resectable disease and significantly enhanced long-term LC survival. These results translated into enhanced 10-year LC survival, independently related to CXR screening exposure, in the entire population-based cohort. Whether increased long-term LC-specific survival in the cohort corresponds to mortality reduction remains to be evaluated.Trial registration numberISRCTN90639073

Highlights

Case-control studies of mass screening for lung cancer (LC) by chest x-rays (CXR) performed in the 1990s in scarcely defined Japanese target populations indicated significant mortality reductions, but these results are yet to be confirmed in western countries
At the end of the active screening period, we evaluated the rate of positive CXR screens, the LC yield, and the invasive procedures performed to follow-up radiographic abnormalities
Improved LC survival persisting over 10 years in the population-based cohort of this study suggests mortality reduction, no conclusion can be drawn about the effectiveness of CXR screening, because survival may be biased by lead-time, selection, length-time and overdiagnosis

Summary

Introduction

Case-control studies of mass screening for lung cancer (LC) by chest x-rays (CXR) performed in the 1990s in scarcely defined Japanese target populations indicated significant mortality reductions, but these results are yet to be confirmed in western countries. Four case-control studies performed in the 1990s in Japan on a population level suggested CXR screening effectiveness, as indicated by LC mortality reductions of about 40% [7,8]. The results of these casecontrol studies were obtained in scarcely defined target populations, were likely biased by self-selection and are yet to be confirmed in western countries. The recently published PLCO results showed that four annual chest radiographs did not reduce LC-specific mortality in volunteers [16], but it remains uncertain whether this finding may be generalized on a population level

Objectives

Methods

Results

Discussion

Conclusion