Pretreatment factors predict overall survival for patients with low-grade glioma: A recursive partitioning analysis

Purpose: Three databases were pooled and analyzed to determine which groupi ngs of prognostic factors best predicted overall survival for patients with low-grade gliomas treated with surgery and immediate or delayed radiothera py. Methods and Materials: Databases of patients with low-grade gliomas compile d at the London Regional Cancer Centre (LRCC), the Norwegian Radium Hospita l (NRH), and the University of California, San Francisco (UCSF) were merged . Inclusion criteria for the pooled analysis included: age greater than or equal to 18 years and histologically confirmed low-grade (World Health Orga nization Grade II) supratentorial fibrillary astrocytoma, oligodendroglioma or mixed oligoastrocytoma. Factors analyzed for prognostic significance in cluded: age at diagnosis, gender, seizures at presentation, presence of enh ancement on computed tomography (CT) or magnetic resonance imaging (MRI), K arnofsky Performance Status (KPS) at diagnosis, histology, extent of surgic al resection, timing of radiotherapy, and treating institution. Univariate and multivariate analysis of overall survival for these factors was perform ed. Recursive partitioning was performed to generate prognostic groups usin g these factors. Results: From the combined databases, 401 patients were eligible for analys is. Median survival for the entire group,vas 95 months/7.9 years. On univar iate analysis age 18-40, presence of seizures at presentation, KPS greater than or equal to 70, treating institution, and absence of contrast enhancem ent were associated with improved overall survival. On multivariate analysi s, these factors remained independent predictors of improved overall surviv al. Recursive partitioning analysis yielded four prognostic groups with sta tistically different median survivals (MS): Group I (n = 41: KPS < 70, age > 40) MS 12 months; Group II (n = 34: KPS greater than or equal to 70, age > 40, enhancement present) MS 46 months; Group III (n = 138: KPS < 70, age 18-40 or KPS greater than or equal to 70 age > 40, no enhancement) MS 87 mo nths; Group IV (n = 188: KPS greater than or equal to 70, age 18-40) MS 128 months. Conclusion: Clusters of pretreatment prognostic factors described subgroups of low-grade glioma patients with divergent overall survivals. Considerati on of these prognostic subgroups may be important when considering timing o f interventions for these patients and in the stratification of patients fo r clinical trials. (C) 1999 Elsevier Science Inc.