Premium
Acute thalamic damage as a prognostic biomarker for post‐traumatic epileptogenesis
Author(s) -
Manninen Eppu,
Chary Karthik,
Lapinlampi Niina,
Andrade Pedro,
Paananen Tomi,
Sierra Alejandra,
Tohka Jussi,
Gröhn Olli,
Pitkänen Asla
Publication year - 2021
Publication title -
epilepsia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.687
H-Index - 191
eISSN - 1528-1167
pISSN - 0013-9580
DOI - 10.1111/epi.16986
Subject(s) - fractional anisotropy , diffusion mri , epilepsy , medicine , magnetic resonance imaging , logistic regression , concordance , nuclear medicine , epilepsy surgery , population , confidence interval , anesthesia , radiology , psychiatry , environmental health
Objective To identify magnetic resonance imaging (MRI) biomarkers for post‐traumatic epilepsy. Methods The EPITARGET (Targets and biomarkers for antiepileptogenesis, epitarget.eu) animal cohort completing T 2 relaxation and diffusion tensor MRI follow‐up and 1‐month‐long video‐electroencephalography monitoring included 98 male Sprague‐Dawley rats with traumatic brain injury and 18 controls. T 2 imaging was performed on day (D) 2, D7, and D21 and diffusion tensor imaging (DTI) on D7 and D21 using a 7‐Tesla Bruker PharmaScan MRI scanner. The mean and standard deviation (SD) of the T 2 relaxation rate, multiple diffusivity measures, and diffusion anisotropy at each time‐point within the ventroposterolateral and ventroposteromedial thalamus were used as predictor variables in multi‐variable logistic regression models to distinguish rats with and without epilepsy. Results Twenty‐nine percent (28/98) of the rats with traumatic brain injury (TBI) developed epilepsy. The best‐performing logistic regression model utilized the D2 and D7 T 2 relaxation time as well as the D7 diffusion tensor data. The model distinguished rats with and without epilepsy (Bonferroni‐corrected p ‐value < .001) with a cross‐validated concordance statistic of 0.74 (95% confidence interval [CI] 0.60–0.84). In a cross‐validated classification test, the model exhibited 54% sensitivity and 91% specificity, enriching the epilepsy rate within the study population from the expected 29% to 71%. A model using the D2 T 2 data only resulted in a 73% enriched epilepsy rate (regression p ‐value .007, cross‐validated concordance 0.70, 95% CI 0.56–0.80, sensitivity 29%, specificity 96%). Significance An MRI parameter set reporting on acute and subacute neuropathologic changes common to experimental and human TBI presents a diagnostic biomarker for post‐traumatic epileptogenesis. Significant enrichment of the study population was achieved even when using a single time‐point measurement, producing an expected epilepsy rate of 73%.