Hippocampal sclerosis and temporal lobe epilepsy following febrile status epilepticus: The FEBSTAT study

Abstract

Objective: This study was undertaken to determine whether hippocampal T2 hyperintensity predicts sequelae of febrile status epilepticus, including hippocampal atrophy, sclerosis, and mesial temporal lobe epilepsy.

Methods: Acute magnetic resonance imaging (MRI) was obtained within a mean of 4.4 (SD = 5.5, median = 2.0) days after febrile status on >200 infants with follow-up MRI at approximately 1, 5, and 10 years. Hippocampal size, morphology, and T2 signal intensity were scored visually by neuroradiologists blinded to clinical details. Hippocampal volumetry provided quantitative measurement. Upon the occurrence of two or more unprovoked seizures, subjects were reassessed for epilepsy. Hippocampal volumes were normalized using total brain volumes.

Results: Fourteen of 22 subjects with acute hippocampal T2 hyperintensity returned for follow-up MRI, and 10 developed definite hippocampal sclerosis, which persisted through the 10-year follow-up. Hippocampi appearing normal initially remained normal on visual inspection. However, in subjects with normal-appearing hippocampi, volumetrics indicated that male, but not female, hippocampi were smaller than controls, but increasing hippocampal asymmetry was not seen following febrile status. Forty-four subjects developed epilepsy; six developed mesial temporal lobe epilepsy and, of the six, two had definite, two had equivocal, and two had no hippocampal sclerosis. Only one subject developed mesial temporal epilepsy without initial hyperintensity, and that subject had hippocampal malrotation. Ten-year cumulative incidence of all types of epilepsy, including mesial temporal epilepsy, was highest in subjects with initial T2 hyperintensity and lowest in those with normal signal and no other brain abnormalities.

Significance: Hippocampal T2 hyperintensity following febrile status epilepticus predicted hippocampal sclerosis and significant likelihood of mesial temporal lobe epilepsy. Normal hippocampal appearance in the acute postictal MRI was followed by maintained normal appearance, symmetric growth, and lower risk of epilepsy. Volumetric measurement detected mildly decreased hippocampal volume in males with febrile status.

Keywords: MRI volumetry; T2 signal intensity; febrile seizures; hippocampal malrotation.

Figure 1:

Effect of T2 Hyperintensity on Hippocampal Growth: Model comparing T2-Norm and T2-Hyper HVs at mean ages for follow-up (FU) intervals. A) HVs of the right sided hyperintense hippocampi (T2-Hyper Right) are plotted along with HVs of their left contralateral homologues (Contralateral to T2-Hyper Right) and right sided T2-Norm HVs (T2-Norm Right). Estimated HV of T2-Hyper Right was significantly smaller (p≤0.001) than that of right T2-Norm and contralateral HVs at all FU intervals after Acute. B) HVs of the left sided hyperintense hippocampi (T2-Hyper Left ) are plotted with HVs of their right contralateral homologues (Contralateral to T2-Hyper Left) and left sided T2-Norm HVs (T2-Norm Left). Estimated HV for T2-Hyper Left was significantly smaller than the contralateral side(p≤0.001) and T2-Norm Left HVs (p=0.014) at the 10-year follow-up. T2-Hyper subject numbers at the acute, 1-, 5- and 10-year follow-up intervals in the T2 group were 15, 6, 4 and 6 subjects with right T2 hyperintensity and 7,5,5 and 5 subjects with left T2 hyperintensity respectively. In the T2-Norm group the respective numbers were 144, 113, 78 and 76 subjects. Error bars indicate 95%CI.

Figure 2

Evolution of T2Scores During Follow-up of T2-Hyper subjects: The difference between right and left was significant at five years (p=0.009) and approached significance at ten years (p=0.07). Error bars indicate 95% confidence intervals (95%CI).

Figure 3:

Cumulative Incidence of Epilepsy by Diagnostic Group: Forty-four subjects developed epilepsy. Overall 10 year cumulative epilepsy incidence was 0.302 (95%CI: 0.22 – 0.375). Ten year cumulative incidences by diagnostic group were: T2-Hyper 0.71 (0.32–0.90), Ext Hpc Abnl (Extrahippocampal Abnormality) 0.48 (0.14 – 0.76), Dysmorphic 0.32 (0.12 – 0.53) and T2-Norm 0.23 (0.15 – 0.32). Gray’s Test indicated significant differences between groups (p=0.0011). Vertical slash marks indicate censoring of subjects lost to follow-up at those time points. The number of subjects remaining at risk for epilepsy at each of the 5 year time points are shown under the graph. Eliminating the Duke pilot and Columbia cohorts, ten year cumulative incidence of epilepsy in the FEBSTAT cohort alone was 0.291 (0.202 – 0.369) and by diagnostic groups the cumulative incidences were: T2-Hyper 0.64 (0.01–0.87), Extrahippocampal Abnormality 0.49 (0.05 – 0.72), Dysmorphic 0.25 (0.003 – 0.43) and T2-Norm 0.25 (0.15 – 0.34), (p=0.038).

Figure 4:

HVs of Control group vs T2-Norm group. Timepoints used in the plot are mean ages of the T2-Norm subjects at the acute, one-, five- and ten-year MRIs. A) HVs of Control vs. T2-Norm group. There was significant difference between Control and T2-Norm at age 6.97 (p=0.0087) and 12.85 years (p= 0.0027). B) Male Control vs. male T2-Norm: Male T2-Norm HVs were almost significantly smaller than male Controls at acute MRI (mean age 1.58 yrs) ( p= 0.08), and became significantly smaller at the one-year (mean age 2.6yrs) ( p= 0.02), five-year (mean age 6.97 years) ( p= 0.014) and ten-year (mean age 12.85 years) ( p= 0.047) follow-ups. C) Female Control vs. female T2-Norm HVs were not significantly different at any of the four follow-up timepoints. Error bars are 95%CI.

FIGURE 5:

Asymmetry Index by Diagnostic Group vs. Mean Age (Timepoints are mean ages at the follow-up timepoints of the T2-Norm group as in Figure 3) There were no significant differences between estimated AIs of Control and T2-Norm groups nor between Control and the Extrahippocampal abnormalities group. Dysmorphic AIs were significantly different from Control and T2-Norm at all points (p<0.01) and from Extrahippocampal abnormalities at 2.6 (p = 0.046) and 6.97 (p=0.011) years. T2-Hyper group AIs were significantly different from Control and T2-Norm at all points (p<0.0001). Dysmorphic and T2-Hyper groups were significantly different at 6.97 and 12.85 years (p<0.001). Error bars are 95%CI. Error bars are not shown on Control, T2-Norm and Extrahippocampal Abnormality groups because all bars were overlapping.