Study objectives: The International Staging System for Lung Cancer (ISSLC) was revised in 1997. Validation studies are numerous but include only selected surgical patients. This study aims to verify the following: (1) the reliability of the ISSLC in an unselected lung cancer population; (2) the likely improvement in prognostic capability of the new classification; and (3) the possibilities for further improvements.
Design: Analysis of a single institution database over a 16-year period from 1983 to 1998.
Setting: Community-based hospital and second referral level institution for a province of 500,000 people.
Patients: The study included 1,296 consecutive patients (1,117 men), with pathologically documented lung cancer (46% with squamous cell cancer), staged both clinically (77%) and pathologically (23%), and treated, for the most part, with chemotherapy (52%).
Interventions: Anthropometric, clinical, and laboratory data were recorded prospectively. Survival analysis was performed by the Kaplan-Meier method and Cox multivariate regression analysis.
Measurement and results: The 1997 revised ISSLC classification fit well with the cohort studied. Each stage and substage significantly differed from each other, except for stage IIA. In this stratum, there were only 13 patients. Comparing the 1986 and the 1997 classifications, no substantial differences were observed (log-rank statistics, 295 vs 293, respectively; p < 0.0001). Independent of the classification used, the Cox models were always highly predictive of the outcome. The only way to increase their efficiency was to replace the variable stage with the original TNM descriptors.
Conclusions: Since grouping different TNM subsets into one stage is not really helpful, we might choose to use TNM descriptors in clinical practice and in research.