Objective: To assess the prognostic value of a series of clinicopathological variables in stage I nonsmall cell lung cancer, for tumors up to 3 cm in diameter.
Patients and method: The study included 271 patients. Survival was analyzed with the Kaplan-Meier method. The Cox model was used for multivariate analysis.
Results: Five- and ten-year survival were 78.63% and 67.59%, respectively. Survival did not significantly depend on sex, age, extent of resection, histology, visceral pleural invasion, level of bronchial invasion or T1 versus T2. The decade in which resection was performed did affect survival (P=.0037). Five-year survival was 58% for operations between 1970 and 1980, 77% for operations between 1981 and 1990, and 84% for operations between 1991 and 2000. Tumor size also affected survival (P=.0046), which was 86% for patients with tumors of less than or equal to 2 cm in diameter and 73% for those with tumors of more than 2 cm in diameter. In the multivariate analysis both variables entered into regression, remaining predictive of survival.
Conclusion: We found evidence for a prognostic stage migration (Will Rogers phenomenon) according to the decade in which resection was performed and that tumor size affected survival in our population. Finally, the current system of TNM staging fails in conforming groups of patients with a homogenous prognosis.