Abstract
Working Memory Network Plasticity After Anterior Temporal Lobe Resection: A Longitudinal Functional Magnetic Resonance Imaging Study.
Stretton J, Sidhu MK, Winston GP, Bartlett P, McEvoy AW, Symms MR, Koepp MJ, Thompson PJ, Duncan JS. Brain 2014;137:1439–1453.
Working memory is a crucial cognitive function that is disrupted in temporal lobe epilepsy. It is unclear whether this impairment is a consequence of temporal lobe involvement in working memory processes or due to seizure spread to extratemporal eloquent cortex. Anterior temporal lobe resection controls seizures in 50–80% of patients with drug-resistant temporal lobe epilepsy and the effect of surgery on working memory are poorly understood both at a behavioural and neural level. We investigated the impact of temporal lobe resection on the efficiency and functional anatomy of working memory networks. We studied 33 patients with unilateral medial temporal lobe epilepsy (16 left) before, 3 and 12 months after anterior temporal lobe resection. Fifteen healthy control subjects were also assessed in parallel. All subjects had neuropsychological testing and performed a visuospatial working memory functional magnetic resonance imaging paradigm on these three separate occasions. Changes in activation and deactivation patterns were modelled individually and compared between groups. Changes in task performance were included as regressors of interest to assess the efficiency of changes in the networks. Left and right temporal lobe epilepsy patients were impaired on preoperative measures of working memory compared to controls. Working memory performance did not decline following left or right temporal lobe resection, but improved at 3 and 12 months following left and, to a lesser extent, following right anterior temporal lobe resection. After left anterior temporal lobe resection, improved performance correlated with greater deactivation of the left hippocampal remnant and the contralateral right hippocampus. There was a failure of increased deactivation of the left hippocampal remnant at 3 months after left temporal lobe resection compared to control subjects, which had normalized 12 months after surgery. Following right anterior temporal lobe resection there was a progressive increase of activation in the right superior parietal lobe at 3 and 12 months after surgery. There was greater deactivation of the right hippocampal remnant compared to controls between 3 and 12 months after right anterior temporal lobe resection that was associated with lesser improvement in task performance. Working memory improved after anterior temporal lobe resection, particularly following left-sided resections. Postoperative working memory was reliant on the functional capacity of the hippocampal remnant and, following left resections, the functional reserve of the right hippocampus. These data suggest that working memory following temporal lobe resection is dependent on the engagement of the posterior medial temporal lobes and eloquent cortex.
Commentary
Neuropsychologic evaluation has a long and storied history in the setting of epilepsy surgery, and much has been learned about the relationship between brain structure and cognitive functions in this context (1). However, in some respects, the focus of the majority of cognitive studies in epilepsy has been restricted to a limited range of functions for many years, such as episodic declarative memory and language. Stretton et al. (2014) join in a recent movement to allow broader neuroscience findings to guide research in epilepsy, while making use of the ever-growing armamentarium of neuroimaging techniques and analyses (2–4). These researchers have been building upon fMRI studies examining working memory functions in healthy individuals, extending this research to the realm of temporal lobe epilepsy (TLE) surgery patients (5, 6). Such studies broaden our knowledge of functions that may be abnormal in this patient group or potentially at risk of decline following surgery, and enhance our understanding of the underlying mechanisms supporting these abilities.
The current study extends prior work by Stretton et al. (5), which employed an fMRI task involving visual working memory to explore the neural underpinnings of this function. Working memory involves a short-term use of memory and attention, allowing us to hold information in mind while completing a task based on these skills. It is a cognitive construct more on the continuum of attention, which is important for productively processing information in the moment, and necessary for establishing more enduring, long-term memories. These investigators employed an “n-back” working memory fMRI paradigm involving visual stimuli for their study, which has been a standard procedure used in neuroscience for decades (7). Each subject is presented with a sequence of stimuli, and is required to indicate when the current stimulus matches the one from n steps earlier in the sequence. Task difficulty is adjusted by raising or lowering the position of the item that the subject must recall (e.g., “n = 2” corresponds to two items earlier in the presentation).
Prior fMRI studies with healthy controls have suggested that these working memory tasks require bilateral activation of frontoparietal networks (8). Additionally, Stretton et al. have more recently demonstrated that the bilateral medial temporal lobe regions of healthy controls increasingly deactivate as working memory load is increased (5), suggesting that engaging frontoparietal working memory networks likely requires the inhibition of mesial temporal lobe function for successful task completion. Stretton et al. have shown that both left and right TLE patients exhibit diminished activation of parietal regions as compared to controls when compared during these tasks at the group level, and fail to show deactivation of mesial temporal lobe regions (6). They have suggested that these abnormal patterns are responsible for baseline working memory deficits sometimes observed in TLE patients.
In the current study, Stretton et al. examine working memory performance using their fMRI paradigm and neurocognitive testing both presurgically and 3 and 12 months postsurgically in a series of TLE patients (16 left/17 right) and a parallel group of 15 healthy controls. A real strength of the study is that it employs a repeated measures design with the inclusion of healthy control subjects. This allows for control over within-group changes in fMRI signal over time and behavioral practice effects. Stretton et al. demonstrate that working memory is decreased in TLE patients relative to controls at baseline, but does not worsen at the group level following surgery. Moreover, the left TLE group showed significant gains following surgery, although only improvement trends were observed in the right TLE group.
With regard to functional activation patterns during the visual-spatial working memory task, right and left TLE patients exhibited much weaker activation of parietal regions during the task as compared to controls at baseline. In patients, as demonstrated previously (6), the ipsilateral anterior hippocampus failed to deactivate. In contrast, left TLE patients showed progressively greater deactivation of posterior hippocampus and right (contralateral) hippocampus than did controls at baseline. The investigators point to this baseline difference as a possible reason for the more notable behavioral improvement observed in the left TLE patients. That is, perhaps the left TLE patients are exhibiting a better functioning network at baseline than the right TLE patients, providing them with a greater chance for recovery. If true, it might be possible to predict who would improve most on similar tasks after successful seizure control. In contrast, although the right TLE sample eventually showed the same deactivation of the contralateral posterior hippocampal remnant and greater activation of the ipsilateral superior parietal region by 12-month follow-up, these changes did not correlate with improved behavioral performance. The authors feel these patterns suggest postsurgical functional recovery is not as useful as presurgical reorganization of function.
A few potential confounds in this study involve heterogeneity of the patients in the TLE sample, differences between subjects in postsurgical changes in antiepileptic drug regimens, and variability in the timespan between presurgical assessment and surgery. All of these variables relate to the necessary clinical considerations of an epilepsy surgery cohort, where patient care ultimately trumps methodologic design issues. However, future work with larger cohorts of patients may allow for greater control over such variables, and add more confidence to the conclusions. As the current study used a visual-spatial working memory task only, another explanation could be that greater improvement in left TLE patients reflects that right hemispheric structures are more critical for this task, and that such structures are more intact in the left TLE group. Thus, left TLE patients could potentially improve more once seizure activity is stopped. This option is addressed in the discussion section, although future studies should explore the use of different stimulus parameters. Finally, behavioral results are presented at the group level, leaving one wondering about the pervasiveness of WM limitations in TLE both at baseline and post-operatively. These issues appear to remain unresolved in the research literature, and results can be highly dependent upon the particular task WM employed.
Overall, this study highlights the manner in which fMRI protocols can be used to provide insight into functional recovery over time and shed light on the underlying neural networks supporting specific functions. It also provides evidence that supports the idea that decreased seizure activity may result in improved functioning distal from the epileptogenic foci, yet it highlights a possible temporal lobe mechanism that many would not have expected for this function. It would appear that the mesial temporal lobe structures must be deactivated to allow the most efficient use of the frontoparietal regions for carrying out ofthe working memory task being studied. If these regions are already damaged, improvements in other parts of the network may not be sufficient to produce behavioral gains. Finally, the importance of drawing upon broader neuroscience research as these investigators have done cannot be underestimated, as such a strategy will ultimately help to propel the epilepsy field forward and open up fresh avenues for research.