Predicting Posttraumatic Epilepsy with MRI: Prospective Longitudinal Morphologic Study in Adults


Address correspondence and reprint requests to Dr. U. Salvolini at Department of Neuroradiology, University Hospital Umberto I, Via Conca, Torrette, 60020 Ancona, Italy. E-mail:


Summary: Purpose: Evaluation of morphologic risk factors for posttraumatic epilepsy (PTE) by using brain magnetic resonance imaging (MRI) in serial assessments ≤2 years after traumatic brain injury (TBI).

Methods: Brain MRI hyperintense (gliosis) or hypointense (hemosiderin) areas or both were assessed in the images of 135 adult TBI inpatients who completed a 2-year clinical, EEG, and MRI study protocol. Overall clinical follow-up for the development of PTE was 5–10 years (median, 102 months). Morphologic risk factors for PTE were evaluated by using Kaplan–Meier curves and Cox regression analysis.

Results: In 20 patients, PTE developed. Kaplan–Meier curves showed that gliomesenchymal sequelae of focal brain lesions (subdural hematomas/contusions) that required surgical treatment (sSDH-C) were a PTE risk factor (p < 0.001), as were sequelae of nonsurgical hemorrhagic contusions with gliosis wall incompletely surrounding hemosiderin dregs (IW) (p = 0.039) and mainly those with time-related changes from incomplete to complete gliosis wall around hemosiderin (I/CW) (p = 0.005); those with early hemosiderin completely surrounded by gliosis (CW) were not (p = 0.821). Cox regression analysis showed that for patients with sequelae of sSDH-C, the PTE risk was 4.38 (p = 0.023) times higher than for those who did not require surgical treatment or underwent surgery because of purely extradural hematoma; for those with IW and I/CW lesions, considered pooled, it was 6.61 times higher (p = 0.014) than for those with CW lesions.

Conclusions: MRI follow-up examination in the early chronic stage can differentiate among low-, intermediate-, and high-risk sequelae of TBI. These findings yield new evidence for, but do not resolve, the debate on posttraumatic epileptogenesis.