Univariate and multivariate statistical analysis of high-grade gliomas: the relationship of radiation dose and other prognostic factors

Abstract

Univariate and multivariate statistical analyses were used to examine the relationships between duration of survival and multiple variables in the presentation and treatment of 82 patients with high-grade gliomas (16 grade 3, 66 grade 4). The median survival time of the eight patients who received less than or equal to 40 Gy to the tumor bed was 16 weeks and was 17 weeks for the three who received between 40 and 50 Gy. Patients who received 50-60 Gy had a median survival time of 62 weeks, compared to 54 weeks in patients who received 60-70 Gy. These differences in median survival time were statistically significant between the extremes (p = 0.0001), as well as between the 40-50 Gy group and the 50-60 Gy group (p = 0.02). However, no significant difference could be detected between the groups receiving 50-60 Gy versus 60-70 Gy. Univariate analysis also identified preoperative performance status, age, histologic grade, extent of surgery, and seizure history as prognostic factors. Cox multivariate analysis was performed to identify variables that were significant in independently predicting duration of survival. Although contemporary studies have shown many variables to be significant in predicting survival, our analysis found that many of them were not independent predictors. The variables which independently predicted improved duration of survival were greater total radiation dose to the tumor bed (p less than 0.0001), superior preoperative performance status (p = 0.003), and grade 3 versus grade 4 (p = 0.04). Younger age at diagnosis was marginally significant (p = 0.07). In the group of 60 patients receiving greater than 50 Gy, a discriminant analysis was also performed. The patients were divided into two groups based on apparent clusters of survival times: greater than or equal to 60 weeks versus less than 60 weeks. The only variable that was found to be predictive of membership in the cluster with longer survival was the presence of seizure activity (p = 0.02). Although univariate and multivariate analyses both showed an apparent statistically significant improvement in survival with increasing total radiation dose to the tumor bed, no additional benefit could be demonstrated for doses greater than 60 Gy.