RANO seizure working group-Tumor-Related Epilepsy Assessment Tool (RANO-TREAT) to assess seizure control for glioma treatment trials and clinical practice

Background No standardized method exists for seizure assessment in glioma clinical trials. We describe the development and evaluation of RANO-TREAT (Tumor Related Epilepsy Assessment Tool) for seizure assessment and its association with changes on brain MRI. Methods Patients with glioma/glioneuronal tumors and ≥ 1 prior seizure along with clinicians completed RANO-TREAT in conjunction with brain MRIs, yielding multiple RANO-TREAT scores at clinic visits over time. Unweighted (primary) and weighted (post-hoc) scores were correlated with disease progression via MRI in all patients and patients with IDHmt tumors, separately. Cohorts were randomly split by patient into cohort-specific training and validation sets. Weights for RANO-TREAT items were defined by multivariable generalized estimating equation models in cohort-specific training sets and validated in cohort-specific validation sets. A nomogram was developed using overall cohort training and validation sets. Results Four hundred and ninety patients (310 IDHmt tumors) had ≥ 1 visits and 285 patients (168 IDHmt tumors) had ≥ 2 visits. Unweighted RANO-TREAT scores (OR:1.01; 95%CI:0.998–1.02; P = .13) and score changes (OR:1.00; 95%CI:0.99–1.02; P = .63) were not associated with progressive disease on MRI. Post-hoc analysis using training and validation sets demonstrated weighted RANO-TREAT scores were correlated with progressive disease in both overall cohort validation set (OR:2.51; 95%CI:1.80–3.52; P < .0001) and IDHmt cohort validation set (OR:4.53; 95%CI:2.11–9.75; P = .0001). Weighted analyses for patients with ≥ 2 visits showed similar associations in validation sets. Conclusions This prospective study suggests an association of seizure control evaluated by a new standardized tool with disease progression in glioma. This tool requires further systematic evaluation in glioma clinical trials alongside more traditional endpoints.