Preventing Verbal Memory Decline: Anatomic Clues to Predict Risk of Laser Ablation Versus Resection?

Abstract

Impact of White Matter Networks on Risk for Memory Decline Following Resection Versus Ablation in Temporal Lobe Epilepsy

Kaestner E, Stasenko A, Schadler A, Roth R, Hewitt K, Reyes A, Qiu D, Bonilha L, Voets N, Hu R, Willie J, Pedersen N, Shih J, Ben-Haim S, Gross R, Drane D, McDonald C R. J Neurol Neurosurg Psychiatry. 2024 Jan;11: jnnp-2023-332682. doi: 10.1136/jnnp-2023-332682. Online ahead of print.

Background: With expanding neurosurgical options in epilepsy, it is important to characterize each option’s risk for postoperative cognitive decline. Here, we characterize how patients’ preoperative white matter (WM) networks relate to postoperative memory changes following different epilepsy surgeries. Methods: Eighty-nine patients with temporal lobe epilepsy with T1-weighted and diffusion-weighted imaging as well as preoperative and postoperative verbal memory scores (prose recall) underwent either anterior temporal lobectomy (ATL; n = 38) or stereotactic laser amygdalohippocampotomy (SLAH; n = 51). We computed laterality indices (ie, asymmetry) for volume of the hippocampus and fractional anisotropy (FA) of 2 deep WM tracts (uncinate fasciculus [UF]) and inferior longitudinal fasciculus [ILF]). Results: Preoperatively, left-lateralized FA of the ILF was associated with higher prose recall (P < .01). This pattern was not observed for the UF or hippocampus (Ps > 0.05). Postoperatively, right-lateralized FA of the UF was associated with less decline following left ATL (P < .05) but not left SLAH (P > .05), while right-lateralized hippocampal asymmetry was associated with less decline following both left ATL and SLAH (Ps < 0.05). After accounting for preoperative memory score, age of onset, and hippocampal asymmetry, the association between UF and memory decline in left ATL remained significant (P < .01). Conclusions: Asymmetry of the hippocampus is an important predictor of risk for memory decline following both surgeries. However, asymmetry of UF integrity, which is only severed during ATL, is an important predictor of memory decline after ATL only. As surgical procedures and pre-surgical mapping evolve, understanding the role of frontal-temporal WM in memory networks could help to guide more targeted surgical approaches to mitigate cognitive decline.

Commentary

For decades, epilepsy surgery centers have been striving to utilize and refine tools to predict and potentially minimize risk of memory decline after temporal lobe epilepsy surgery, especially verbal memory decline after dominant hemisphere surgeries. While it is well established that individuals undergoing dominant hemisphere anterior temporal lobectomy (ATL) with older age of epilepsy onset, normal magnetic resonance imaging (MRI), and strong verbal memory function on neuropsychological testing are at risk for verbal memory decline, ¹ centers often conduct additional testing including Wada and/or functional MRI to further estimate risk. There remains significant debate regarding the added risk and potential benefit of these procedures, and while functional MRI has demonstrated significant limitations in assessing memory function (vs language), many now favor this non-invasive tool over the invasive Wada. ¹ This debate highlights the need for additional, improved, and ideally noninvasive tools to support presurgical counseling on cognitive outcomes of temporal lobe epilepsy surgeries and to tailor surgery selection based on a more refined balance of cognitive risk and seizure outcome prediction. This is especially important in light of current trends in epilepsy surgery, with increased minimally invasive options available, which have clear trade-offs compared to traditional resective surgery and typically sacrifice some potential seizure benefit to reduce cognitive risk.

A particularly crucial gap in recent research literature has been data to help clinicians accurately weigh the trade-offs of dominant temporal lobe resection versus mesial temporal laser ablation (stereotactic laser amygdalohippocampotomy [SLAH]). This technique was rapidly adopted in many centers starting more than a decade ago and quickly transformed the epilepsy surgery landscape, at least in part due to SLAH's more rapid recovery and reduced hospital length of stay compared to ATL, and due to initial hope that seizure outcomes would be comparable to resection (but with lower cognitive risk). Initial literature on SLAH outcomes was limited in rigor, being single-center case series, and with minimal or no cognitive outcome data. However, in the subsequent years, literature has mounted to suggest that (1) seizure freedom after SLAH is not as high as after ATL ² and (2) there appears to be reduced cognitive decline for many individuals following SLAH when compared to historical data on temporal lobectomy.^3–6

While much further work is needed to fully characterize cognitive outcomes of SLAH compared to other surgical approaches and better predict individual patient outcomes, the multicenter analysis by Kaestner et al ⁷ makes significant strides forward in this area of research. Kaestner et al provide new data characterizing verbal memory outcomes of SLAH versus temporal lobectomy, generating hypotheses regarding anatomic mechanisms, and providing preliminary information about potential predictors of verbal memory decline. The study was conducted at 3 centers with varied clinical practice and included 89 patients who underwent either ATL (38) or SLAH (51) and had pre- and postoperative verbal memory evaluated using the Wechsler Memory Scale Logical Memory Delayed Recall test. Postoperative testing was conducted a median of 8 months after surgery. Pre-operative diffusion-weighted MRI fractional anisotropy (FA) values were calculated for the hippocampi and 2 key white matter (WM) tracts disrupted in ATL but not SLAH: uncinate fasciculus (UF) and inferior longitudinal fasciculus (ILF). These were evaluated for association with postoperative verbal memory decline. The main analysis was done among the subset of individuals with left temporal lobe epilepsy (N = 41); right temporal epilepsy patients were controls. Postoperative verbal memory decline was defined as a 1 standard deviation (SD) decline of scaled scores based on epilepsy-specific norms, and sensitivity analyses also examined reliable change indices (with 95% confidence interval). Based on the 1 SD decline of scaled scores, 27% of left ATL patients and 17% of left SLAH patients had a significant verbal memory decline (vs 10% and 0% for right ATL and SLAH, respectively).

The main study findings suggest a relationship between hippocampal MRI and verbal memory decline after both surgery types, and with UF only after ATL. Correlation analyses evaluated the association between laterality index of MRI FA for each anatomic area of interest with occurrence of verbal memory decline. The findings indicate an association of hippocampal leftward laterality with verbal memory decline after both surgery types, consistent with prior work indicating a structurally intact hippocampus predicted greater functional decline after surgery. ⁸ Among the WM tracts examined, Kaestner et al ⁷ found an association of uncinate fasciculus left laterality with verbal memory decline only after ATL, and no association of ILF with decline after either surgery. This may suggest individuals with preoperative leftward UF FA could represent a specific subgroup of left temporal epilepsy patients for whom verbal memory decline risk would be lower with laser ablation than ATL. The finding also supports the hypothesis that severing the UF during ATL is a mechanism for verbal memory decline in some individuals.

While the study by Kaestner et al has various limitations, it is a potentially important step toward future accessible, noninvasive evaluation tools for presurgical cognitive outcome prediction. Limitations include the small sample size, which limited clinically important analyses among individuals with dominant hemisphere surgery and postoperative seizure freedom. Nonetheless, this sample appears to be the largest SLAH cognitive outcomes sample to date, and a key strength is the direct comparison with ATL. Other limitations include use of only a single verbal memory measure; also, there may be debate as to whether the measure evaluated is the ideal verbal memory measure. Future analyses with larger samples and additional verbal memory measures are needed to demonstrate reproducibility. Despite these limitations, since MRI diffusion-weighted imaging (DWI) is a widely available technique collected during routine epilepsy workups, theoretically a future widely available clinical tool for additional noninvasive surgery risk prediction could emerge from this line of work. If future studies demonstrate reproducibility and individual patient-level predictive value of preoperative hippocampal and UF laterality, these measures could potentially be incorporated in standard epilepsy surgery workups, enhancing surgical technique selection and presurgical risk/benefit counseling.

In the current epilepsy surgery landscape where multiple options exist with different seizure outcome and cognitive risk profiles, tools to help clinicians better understand the individualized risk/benefit for patients specifically among the current surgical techniques are needed. While much additional work is required as outlined above, Kaestner et al's findings raise the possibility that DWI of the UF might help identify patients most likely to benefit cognitively in selecting a potentially less efficacious laser ablation over a resective surgery. Likewise, hippocampal DWI might prove useful to identify patients to counsel about risk for verbal memory decline from either surgery, perhaps further supplanting the role of the more invasive Wada procedure.