doi:10.1016/j.ijrobp.2006.06.003 CLINICAL INVESTIGATION Brain RPA CLASSIFICATION HAS PROGNOSTIC SIGNIFICANCE FOR SURGICALLY RESECTED SINGLE BRAIN METASTASIS RAHUL D. TENDULKAR, M.D.,* STEPHANIE W. LIU, B.A., † GENE H. BARNETT, M.D., ‡ MICHAEL A. VOGELBAUM, M.D., PH.D., ‡ STEVEN A. TOMS, M.D., M.P.H., ‡ TAO JIN, M.S., § AND JOHN H. SUH, M.D.* Departments of *Radiation Oncology, ‡ Neurosurgery, and § Biostatistics, Brain Tumor Institute, Cleveland Clinic, Cleveland, OH and † Northwestern University Feinberg School of Medicine, Chicago, IL Purpose: To retrospectively evaluate prognostic factors that correlate with overall survival among patients with a surgically resected single brain metastasis. Methods and Materials: An Institutional Review Board–approved database of the Cleveland Clinic Brain Tumor Institute was queried for patients with a single brain metastasis treated by surgical resection between February 1984 and January 2004. The primary endpoint was overall survival from the date of surgery by the Kaplan-Meier method. Results: A total of 271 patients were included. Statistically significant variables for improved survival on multivariate analysis included age <65 years, lack of extracranial metastases, control of primary tumor, histology (non–small-cell lung carcinoma), and use of stereotactic radiosurgery. The median survival for all patients was 10.2 months. Survival of patients in recursive partitioning analysis (RPA) class 1 was better (21.4 months) than those in RPA class 2 (9.0 months, p < 0.001), RPA class 3 (8.9 months, p  0.15), or the combined group of RPA classes 2 and 3 (9.0 months, p < 0.001). Patients had a median survival of 10.6 months after documented gross total resection and 8.7 months after subtotal resection, which approached statistical signifi- cance (p  0.07). Those who were treated with stereotactic radiosurgery had a median survival of 17.1 months, which was greater than patients who were not treated with stereotactic radiosurgery (8.9 months, p  0.006). Conclusions: This analysis supports the prognostic significance of the RPA classification in patients with a single brain metastasis who undergo surgical resection and adjuvant therapy. RPA class 1 patients have a very favorable prognosis with a median survival of 21.4 months. © 2006 Elsevier Inc. Brain metastasis, Surgery, RPA classification. INTRODUCTION Brain metastases are the most common intracranial tumor, with approximately 170,000 cases in the United States an- nually (1, 2). Approximately 20 –50% of patients present with a single lesion in the brain (3–6). Treatment typically includes some combination of steroids, whole-brain radia- tion therapy (WBRT), stereotactic radiosurgery (SRS), and surgical resection. The optimal management for an individ- ual patient is often controversial (7, 8). The Radiation Therapy Oncology Group (RTOG) devel- oped a prognostic classification system for patients with brain metastases, based on 1200 patients from three ran- domized trials treated with whole-brain radiation (9 –12). The RTOG recursive partitioning analysis (RPA) identified four prognostic factors that significantly correlated with survival: Karnofsky performance status (KPS), control of primary tumor, age, and status of extracranial disease. From these four factors, three prognostic classes were developed. RPA class 1 consists of patients with age 65 years, KPS 70, controlled primary tumor, and no ex- tracranial metastases (median survival 7.1 months). RPA class 3 includes all patients with KPS 70 (median survival 2.3 months). RPA class 2 comprises those remaining pa- tients not included in either RPA class 1 or 3 (median survival 4.2 months). This classification scheme was prospectively validated in a study of patients with unresectable brain metastases (13, 14). Others have retrospectively applied the RPA strat- ification to patients treated with conventional surgery (15) and with radiosurgery (16 –18), and its prognostic value appears to be maintained with these therapies. Agboola et al. demonstrated a median survival of 14.8 months, 9.9 months, and 6.0 months for the three RPA classes, respectively, among 125 patients with one or more brain metastasis treated with a surgical resection followed by WBRT (15). A prospective randomized trial evaluated WBRT with or without SRS for Reprint requests to: John Suh, M.D., Department of Radiation Oncology, Cleveland Clinic, Desk T-28, 9500 Euclid Avenue, Cleveland, OH 44195. Tel: (216) 444-5574; Fax: (216) 445-1068 Received May 18, 2006. Accepted for publication June 5, 2006. Int. J. Radiation Oncology Biol. Phys., Vol. 66, No. 3, pp. 810 – 817, 2006 Copyright © 2006 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/06/$–see front matter 810