Recursive partitioning analysis of 1999 radiation therapy oncology group (RTOG) patients with locally-advanced non–small-cell lung cancer (LA-NSCLC): Identification of five groups with different survival

Purpose: Survival of patients with locally-advanced non–small-cell lung cancer (LA-NSCLC) is predicted by the stage of the disease and other characteristics. This analysis was undertaken to identify these characteristics in a large cooperative group patient population, as well as to define subgroups of the population with differing outcomes. Patients and Methods: Analysis included 1,999 patients treated in 9 RTOG trials between 1983 and 1994 with thoracic irradiation (RT) with ( n = 355) or without chemotherapy (CT). Results: In univariate analysis, the following characteristics were significantly associated with an improved survival: use of CT, CT delivered without major deviation, abnormal pulmonary function tests, normal hemoglobin, protein, LDH and BUN, presence of dyspnea, hemoptysis, cough or hoarseness, uninvolved lymph nodes, T1 or T2 stage, no malignant pleural effusion (PE), weight loss of < 8%, Karnofsky performance status (KPS) of at least 90, adenocarcinoma histology, female gender, and age less than 70 years. Recursive partitioning analysis (RPA) was subsequently applied to identify 5 patient subgroups with significantly different median survival times (MST): Group I, KPS of ≥ 90, who received chemotherapy (MST 16.2 months); Group II, KPS of ≥ 90, who received no CT, but had no PE (MST 11.9 months); Group III, KPS < 90, younger than 70 years, with non-large cell histology (MST 9.6 months); Group IV, KPS ≥ 90, but with PE, or KPS < 90, younger than 70 years, and with large cell histology, or older than 70 years, but without PE (MST 5.6–6.4 months); Group V, older than 70, with PE (MST 2.9 months). Conclusion: Cisplatinum-based CT improves survival, for excellent prognosis of LA-NSCLC patients, over RT alone. The presence of a malignant pleural effusion is a major negative prognostic factor for survival. The identification of RPA prognostic groups among patients with LA-NSCLC provides prognostic information and may serve as a basis of stratification in future trials.