Purpose: To retrospectively construct a comprehensive multivariate model of cancer recurrence and to design a molecular pathologic substaging system in stage I non-small-cell lung cancer (NSCLC). Methods: All patients with stage I NSCLC resected at Brigham and Women's Hospital (Boston, MA) between 1984 and 1992 with adequate clinical follow-vp were studied. The importance of three demographic characteristics, surgical extent, 11 pathologic features, and seven molecular factors on cancer-free survival was examined. Results: Two hundred forty-four patients were studied, with 25 noncancer deaths and 80 patients with recurrent disease. Significant univariate predictors (P < .05) of cancer recurrence were age older than 60 years, male sex, wedge resection, World Health Organization (WHO) adenocarcinoma subtype solid tumor with mucin, lymphatic invasion, and p53 expression. Multivariate analysis identified nine independent predictors of recurrence: solid tumor with mucin, a wedge resection,;tumor diameter of 4 cm or greater, lymphatic invasion, age older than 60 years, male sex, p53 expression, K-ras codon 12 mutation, and absence of H-ras p21 expression. Multivariate cancer-free survival (CFS) analysis in the 180 patients who underwent lobectomy or pneumonectomy led to the elimination of sex and age, which left six independent factors. Conclusion: Lobectomy or pneumonectomy should be performed in stage I NSCLC. Using the six independent factors for recurrent disease, we propose a pathologic molecular substaging system. Patients with two factors or less are graded la, with a 5-year CFS rate of 87%; those with three factors are graded Ib, with a 5-year CFS rate of 58%; and those with four factors or more are graded Ic, with a 5-year CFS rate of 21%. (C) 1998 by American Society of Clinical Oncology.
