Patients with brain metastases: hope for recursive partitioning analysis (RPA) class 3

Purpose: The objectives of the present study were (a) to validate the prognostic classification derived from recursive partitioning analysis (RPA) of the Radiation Therapy Oncology Group (RTOG); (b) to identify prognostic factors in class 3; (c) to examine the impact of treatment related variables on the prognosis in class 3. Patients and methods: Nine hundred and sixteen patients with brain metastases had resection and whole brain radiotherapy (WBRT, n = 257) or WBRT alone (n = 659) at our institution from 1985 to 2000. Patients were grouped into RPA classes 1, 2, and 3 (n = 67, 441, and 408, respectively). Results: Median survival of the whole group was 3.4 months. Median survival in classes 1, 2, and 3 was 8.2, 4.9, and 1.8 months, respectively. In class 3, age (<65 years vs. ≥65 years, relative risk (RR) 0.75), status of the primary tumor (controlled vs. uncontrolled, RR 0.86), and the number of brain metastases (single vs. multiple, RR 0.76) were independent prognostic variables. We defined three prognostic subgroups: class 3a (n = 51): age <65 years, controlled primary tumor, single brain metastasis; class 3c (n = 44): age greater than or equal to65 years, uncontrolled primary tumor, multiple brain metastases; class 3b (n = 313): all other patients. Median survival in classes 3a, 3b, and 3c was 3.2, 1.9, and 1.2 months, respectively (P < 0.0001). Intra-class comparisons showed that resection followed by WBRT yielded significantly better survival compared with WBRT alone. Conclusion: Our results validate the RTOG RPA classification for patients with brain metastases. The variables age, status of the primary, and number of brain metastases allow the division of class 3 into prognostic subgroups. Even class 3 patients may benefit from more aggressive treatment strategies. (C) 2002 Elsevier Science Ireland Ltd. All rights reserved.