Extent of resection and survival in supratentorial infiltrative low-grade gliomas: analysis of and adjustment for treatment bias

Abstract

Any correlation between the extent of resection and the prognosis of patients with supratentorial infiltrative low-grade gliomas may well be related to biased treatment allocation. Patients with an intrinsically better prognosis may undergo more aggressive resections, and better survival may then be falsely attributed to the surgery rather than the biology of the disease. The present study investigates the potential impact of this type of treatment bias on survival in a series of patients with low-grade gliomas treated at the authors' institution. We conducted a retrospective study of 148 patients with low-grade gliomas undergoing primary treatment at our institution from 1996-2011. Potential prognostic factors were studied in order to identify treatment bias and to adjust survival analyses accordingly. Eloquence of tumor location proved the most powerful predictor of the extent of resection, i.e., the principal source of treatment bias. Univariate as well as multivariate Cox regression analyses identified the extent of resection and the presence of a preoperative neurodeficit as the most important predictors of overall survival, tumor recurrence and malignant progression. After stratification for eloquence of tumor location in order to correct for treatment bias, Kaplan-Meier estimates showed a consistent association between the degree of resection and improved survival. Treatment bias was not responsible for the correlation between extent of resection and survival observed in the present series. Our data seem to provide further support for a strategy of maximum safe resections for low-grade gliomas.