Comparison of linear and volumetric criteria for the determination of therapeutic response in dogs with intracranial gliomas

Abstract

Background: Brain tumor therapeutic responses can be quantified from magnetic resonance images (MRI) using 1- (1D) and 2-dimensional (2D) linear and volumetric methods, but few studies in dogs compare these techniques. Hypotheses: Linear methods will be obtained faster, but have less agreement than volumetric measurements. Therapeutic response agreement will be highest with the total T2W tumor volumetric (TTV) method. Therapeutic response at 6-weeks will correlate with overall survival (OS). Animals: Forty-six dogs with intracranial gliomas. Methods: Prospective study. Three raters measured tumors using 1D and 2D linear, TTV, and contrast-enhancing volumetric (CEV) techniques on 143 brain MRI to determine agreement between methods, define therapeutic responses, and assess relations with OS. Results: Raters performed 1D the fastest (2.9 ± 0.57 minutes) and CEV slowest (17.8 ± 6.2 minutes). Inter- and intraobserver agreements were excellent (intraclass correlations ≥.91) across methods. Correlations between linear (1D vs 2D; ρ >.91) and volumetric (TTV vs CEV; ρ >.73) methods were stronger than linear to volumetric comparisons (ρ range,.26-.59). Incorporating clinical and imaging data resulted in fewer discordant therapeutic responses across methods. Dogs having partial tumor responses at 6 weeks had a lower death hazard than dogs with stable or progressive disease when assessed using 2D, CEV, and TTV (hazard ration 2.1; 95% confidence interval, 1.22-3.63; P =.008). Conclusions and Clinical Importance: One-dimensional, 2D, CEV, and TTV are comparable for determining therapeutic response. Given the simplicity, universal applicability, and superior performance of the TTV, we recommend its use to standardize glioma therapeutic response criteria.