33 A tale of two studies: diagnostic algorithms and clinical practice guidelines minimize overdiagnosis and overtreatment and maximize survival in lung cancer screening

Abstract

At the beginning of the new millennium two major lung cancer screening (LCS) studies used different research approaches to study computerized tomographic (CT) LCS. The National Lung Cancer Screen study (NLST) used a randomized control design to compare with chest roentgenogram (CR). The study did not utilize a diagnostic algorithm. The International Early Lung Cancer Action Program (IELCAP) was a prospective, multi-institutional, single-arm registry study incorporating diagnostic algorithms for baseline and annual-repeat screens http://www.ielcap.org/protocols. Although the NLST demonstrated substantial decrease in lung cancer (>20%) and all-cause (>7%) mortality, study subjects had lower 5 year survival (60%) and experienced higher levels of false positive test results and invasive diagnostic tests and surgical operations in patients with benign nodules than IELCAP. The IELCAP study used a form of ‘learning healthcare system’, analyzing accumulating registry data on a semi-annual basis with step-wise modification of the algorithms based upon analysis of results and consensus of assembled principal investigators. The updated IELCAP algorithms have been incorporated into the clinical practice guideline of the National Comprehensive Cancer Network and other professional organizations. Although the IELCAP design does not allow precise estimation of mortality reduction, it has established that application of the algorithms results in substantially higher 10 year actuarial lung cancer specific survival (>80%), lower (13%) rate of false positives and fewer invasive tests and operations on benign nodules. Modifications in the algorithm to date primarily relate to raising the threshold for a positive test result to solid nodules with an average diameter of 6 mm. or greater and changes in the management of non-solid nodules, regardless of size. The algorithms reflect current uncertainty as to how often and how rapidly in-situ adenocarcinomas will progress into invasive adenocarcinomas. Extant data suggest that careful observation using annual CT scans allows identification of progression to invasive adenocarcinoma with demonstration of transition from non-solid to part solid nodules containing a growing solid component, without allowing neoplasms to progress in stage. Based upon retrospective analysis of IELCAP data suggesting that minimally invasive, sub-lobar resectons provide equivalant survival, with lower morbidity, an offshoot, prospective registry study, ‘Initiative or Early Lung Cancer Research on Treatment’ (IELCART), is currently accruing participants to determine whether small screen-detected LC can be treated safely and effectively using minimally invasive, sub-lobar pulmonary resection or radiation therapy modalities. Objectives The objective of this presentation is to review current efforts to reduce potential overdiagnosis and (more important) to avoid overtreatment of screen-detected lung cancer, and their effectiveness and safety. Method Review and analysis of NLST, IELCAP and NCCN diagnostic algorithms, clinical practice gudelines and published results on lung cancer screening. Results Application of diagnostic algorithms, updated after analysis of accumulating research data, yield improvements in survival and reduction in false-positive test results, further non-invasive and invasive testing and overtreatment in the form of resection of benign pulmonary nodules or potentially-overdiagnosed in-situ adenocarcinomas. Conclusions Application of ‘learning healthcare system’ principles to prospective registry research models in lung cancer screening offer potential to improve the safety and effectiveness of lung cancer screening while simultaneously avoiding over-treatment of potentially-overdiagnosed lung cancer. A substantial number of lung cancer deaths can be prevented by lung cancer screening.