Implant volume as a prognostic variable in brachytherapy decision-making for malignant gliomas stratified by the RTOG recursive partitioning analysis

Purpose: When an initial retrospective review of malignant glioma patients (MG) undergoing brachytherapy was carried out using the Radiation Therapy Oncology Group (RTOG) recursive partitioning analysis (RPA) criteria, it revealed that glioblastoma multiforme (GBM) cases benefit the most from implant. In the present study, we focused exclusively on these GBM patients stratified by RPA survival class and looked at the relationship between survival and implanted target volume, to distinguish the prognostic value of volume in general and for a given GBM class. Methods and Materials: Between 1991 and 1998, 75 MG patients were treated with surgery, external beam radiation, and stereotactic iodine-125 (I-125) implant. Of these, 53 patients (70.7%) had GBMs, with 52 (98%) having target volume (TV) data for analysis. Stratification by RPA criteria showed 12, 26, 13, and I patients in classes III to VI, respectively. For analysis purposes, classes V and VI were merged. There were 27 (51.9%) male and 25 (48.1%) female patients. Mean age was 57.5 years (range 14-79). Median Karnofsky performance status (KPS) was 90 (range 50-100). Median follow-up time was 11 months (range 2-79). Results: At analysis, 18 GBM patients (34.6%) were alive and 34 (65.4%) were dead. Two-year and 5-year survivals were 42% and 17.5%, respectively, with a median survival time (MST) of 16 months. Two-year survivals and MSTs for the implanted GBM patients compared to the RTOG database were as follows: 74% vs. 35% and 28 months vs. 17.9 months for class III; 32% vs. 15% and 16 months vs. 11.1 months for class IV; 29% vs. 6% and 11 months vs. 8.9 months for class V/VI. Mean implanted TV was 15.5 cc (range 0.8-78), which corresponds to a spherical implant diameter of 3.1 cm. Plotting survival as a function of 5-cc TV increments suggested a trend toward poorer survival as the implanted volume increases. The impact of incremental changes in TV on survival within a given RPA class of GBMs was compared to the RTOG database. Looking at absolute differences in MSTs: for classes III and IV, there was little effect of different TVs on survival; for class V/VI, a survival benefit to implantation was still seen at the target volume cutoff (TV > 25 cc). Within a given RPA class, no significant differences were found within class III; for class IV, the most significant difference was at 10 cc (p = 0.05); and for class V/VI, at 20 cc (p = 0.06). Conclusion: For all GBM patients, an inverse relationship between implanted TV size and median survival is suggested by this study. However, when GBM patients are stratified using the RTOG's RPA criteria, the prognostic effect of implant volume disappears within each RPA survival class. At the critical volume of 25 cc, which approximates an implant of 5-cm diameter (upper implantation limit of many CNS brachytherapy protocols), the "poorest" prognosis GBM patients stratified by RPA still demonstrate a survival benefit with implant. We suggest that any GBM patient meeting brachytherapy recognized size criteria be considered for I-125 implant. (C) 2001 Elsevier Science Inc.