Verbal memory outcomes after temporal lobe resection in pediatric epilepsy

1077 

Abstract Submission 

Freya Prentice^1,2, Maria Eriksson^1,2, Sara Shavel-Jessop^1,2, Patricia Martin-SanFilippo^1,2, Frederique Liegeois^1,2, Torsten Baldeweg^1,2

¹University College London, London, UK, ²Great Ormond Street Hospital, London, UK

Freya Prentice  
University College London|Great Ormond Street Hospital
London, UK|London, UK

Maria Eriksson  
University College London|Great Ormond Street Hospital
London, UK|London, UK

Sara Shavel-Jessop  
University College London|Great Ormond Street Hospital
London, UK|London, UK

Patricia Martin-SanFilippo  
University College London|Great Ormond Street Hospital
London, UK|London, UK

Frederique Liegeois  
University College London|Great Ormond Street Hospital
London, UK|London, UK

Torsten Baldeweg  
University College London|Great Ormond Street Hospital
London, UK|London, UK

Verbal memory deficits are consistently documented in adults after left temporal lobe resection (TLR) for intractable epilepsy (Lee et al., 2002) but not in children. This may reflect less lateralized memory networks in the developing brain as well heterogeneity in the use of neuropsychological measures of verbal memory across pediatric studies. The task specificity model (Saling, 2009) suggests that tests which involve the delayed recall of unrelated word pairs are more sensitive to hippocampal functioning due to the specific role of the hippocampus in binding unrelated information. It would, therefore, be expected that decline on this measure would be greatest after left TLR including hippocampal removal (TLR+H). In this retrospective study, we aimed to replicate previous findings (Law et al. 2017; Danguecan & Smith, 2019) that declines in word pair recall in children are more likely after left TLR with hippocampectomy (TLR+H) compared to without and right TLR. Secondly, we sought to identify additional clinical and neuroimaging predictors of word pair decline.

We included 44 children who underwent TLR for intractable epilepsy at Great Ormond Street Hospital between 2000-2022 and had pre- and postoperative assessments on the Children's Memory Scale (CMS). Change scores were calculated from the difference between pre- and postoperative scores and decline was defined based on published Reliable Change Indices (Busch et al., 2015). The extent of TLR was coded based on surgical reports and confirmed by the visual inspection of the MRI scan by an experienced neuroanatomist. Where patients had a preoperative language fMRI, a laterality index was calculated in a frontal region of interest for a verb generation task.

Rates of RCI-defined declines on the word pairs subtest of the CMS differed between all four groups (see Table 1; X2=9.14, p=.027), with decline only demonstrated in those who underwent left TLR+H. Within the left TLR+H group, greater postoperative decline in word pairs was associated with higher preoperative word pair scores (r=-0.48, p=.039) and an older age at surgery (r=-0.66, p=.002), but not postoperative seizure freedom (t(1.17)=0.14, p=.909; see Figure 1). Preoperative left language lateralization was significantly correlated with an older age at surgery (r=0.61, p=.026, N=13), but not word pair decline (r=-0.52, p=.069, N=13). A final linear regression model including preoperative word pair scores and age at surgery explained 46% of the variance in word pair decline (adjusted r2=0.46, p=.003).

We replicated previous findings that decline on the delayed word pairs subtest was specific to patients with a left TLR+H, with 32% of this group experiencing a clinically meaningful decline. Within the left TLR+H group, all patients who experienced a decline had surgery after 12 years of age and an older age at surgery was associated with decline. This may reflect increasingly lateralized verbal memory functions with age, resulting in greater verbal memory deficits after left TLR+H, similar to those observed in adults. Left language lateralisation was associated with an older age at epilepsy surgery, however, language lateralisation was not predictive of word pair decline. The power of this analysis was likely limited by a small and entirely left-lateralized group. Larger fMRI samples are needed to examine whether the greater word pair declines seen with an increasing age at surgery partially reflect increasing left lateralisation of language and memory networks over development.

Long-Term Memory (Episodic and Semantic) ¹

Neural Plasticity and Recovery of Function

Early life, Adolescence, Aging

Activation (eg. BOLD task-fMRI) ²

Cognition

Development

Epilepsy

FUNCTIONAL MRI

Memory

PEDIATRIC

STRUCTURAL MRI