Cognitive outcomes following laser interstitial therapy for mesiotemporal epilepsies

Abstract

Objective

To provide a review of cognitive outcomes across a full neuropsychological profile in patients who underwent laser interstitial thermal therapy (LiTT) for mesiotemporal epilepsy (mTLE).

Methods

We examined cognitive outcomes following LiTT for mTLE by reviewing a consecutive series of 26 patients who underwent dominant or nondominant hemisphere procedures. Each patient's pre- and postsurgical performance was examined for clinically significant change (>1SD improvement or decline on standardized scores), with a neuropsychologic battery that included measures of language, memory, executive functioning, and processing speed.

Results

Presurgical performance was largely consistent with previous research, where patients suffering from dominant hemisphere epilepsies demonstrated deficits in verbal learning and memory, whereas patients with nondominant hemisphere scored lower on visually mediated tests. Case-by-case review comparing presurgical to postsurgical scores revealed clinically significant improvement in both dominant and nondominant patients in learning and memory and other aspects of cognition such as processing speed and executive functioning. Of the few patients who did experience clinically significant decline following LiTT, a greater proportion had undergone dominant hemisphere procedures.

Conclusions

Compared with the outcome literature of dominant open anterior temporal lobectomies (ATLs), where postsurgical decline has been documented in up to 40%–60% of cases, our LiTT case series exhibited a much lower incidence of postoperative language or verbal memory decline. Moreover, promising rates of postoperative improvements were also observed across multiple cognitive domains. Future studies exploring cognitive outcomes following LiTT should include comprehensive neuropsychological findings, rather than only select domains, as clinically significant change can occur in areas other than those typically associated with mesiotemporal structures.

Laser interstitial thermal therapy (LiTT) has recently gained traction as a viable surgical approach for mesiotemporal epilepsy (mTLE), particularly in patients with mesial temporal sclerosis (MTS).^1–7 Unlike traditional surgical approaches for epilepsy, LiTT does not require a craniotomy, resulting in lower perioperative morbidity and shorter hospital stays.^1,5,7 As this is a relatively new procedure, long-term seizure outcomes are still pending, but several case series reported comparable outcomes to open surgery in the initial period (6–12 months) following LiTT.^3,5,7 According to one of the largest case series published to date (n = 58) with a longer follow-up period (>12 months), seizure-free rates for patients with mTLE who underwent LiTT were lower than but comparable to those seen in open surgery reports.⁴ Two-year follow-up of the patients included in the present case series demonstrated similar results, with 61.5% of patients classified as “free from disabling seizures” and 26.9% experiencing “only rare disabling seizures.”⁸

Any kind of epilepsy surgery for mTLE carries the possibility of cognitive morbidity. Postsurgical deficits in naming, verbal fluency, and memory are well documented in patients who undergo dominant hemisphere anterior temporal lobectomy (ATL), the surgical gold standard for mTLE.^9–13 This risk of cognitive decline has led to increased interest in alternative, less invasive surgical approaches such as LiTT. It has been theorized that by selectively coagulating amygdalohippocampal cortex, surrounding perihippocampal and neocortical structures are spared, presumably resulting in fewer cognitive deficits.³

However, whether LiTT is safer for cognitive function is not clear. To date, multiple case series have documented both decline and improvement in memory following LiTT.^5,7 For example, a recent review of preliminary findings following LiTT suggests potential for episodic verbal memory decline in patients who undergo dominant hemisphere surgery, whereas nondominant hemisphere patients may experience improvement.⁵ Perhaps this is not surprising given that, while a less invasive procedure, LiTT still requires destruction of potentially functional tissues in the hippocampus and amygdala, areas typically associated with memory.

The present study evaluates cognitive outcomes following LiTT in a clinical sample of patients, exploring both language and memory performance, but also other aspects of cognition such as processing speed, fine motor dexterity, and aspects of executive functioning. We examined both individual cognitive profiles and compared outcomes of patients who underwent dominant vs nondominant hemisphere surgery.

Methods

Standard protocol approvals, registrations, and patient consents

Retrospective review of clinical and neuropsychological data was approved by the University of Miami (UM), Institutional Review Board (IRB). Written informed consent from patients or guardians was not required, as this was a retrospective analysis of deidentified data from a medical record repository designed specifically for use by the Comprehensive Epilepsy Center at the UM. As part of the user agreement, investigators accepted the condition to receive encrypted and deidentified data for use exclusively by UM team members, which could not be shared with outside institutions or investigators to protect patient privacy.

Patient selection

A total of 29 patients underwent LiTT for medically refractory focal epilepsy of mesial temporal origin at the UM Comprehensive Epilepsy Center , a level 4 epilepsy-accredited facility. Participants were seen consecutively between 2013 and 2016. All participants received a comprehensive presurgical evaluation from a multidisciplinary team before determining surgical eligibility including extended video-EEG monitoring, 3-tesla high-resolution thin-cut MRI scans using epilepsy surgery protocols, interictal PET, and comprehensive neuropsychological evaluation. Three patients required intracranial monitoring due to evidence of dual pathology. Two had electrographic evidence of independent bilateral seizure foci, and the third patient an extratemporal cavernoma. These subjects were excluded from the current review, as they may represent clinically different pictures than patients who traditionally undergo surgery for temporal lobe seizures. Of the remaining 26 patients, 14 had radiographic evidence of MTS. Participants were divided into dominant vs nondominant hemisphere groups based on language lateralization established on an fMRI or Wada test. One patient included in the nondominant group had atypical language lateralization (right hemisphere) based on Wada testing and underwent left-sided surgery.

Operative procedures

All LiTT procedures were performed by a single surgeon (J.R.J.) at the UM Hospital. After receiving general anesthesia, patients had a stereotactic frame placed (CRW, Plainsboro, NJ) and underwent thin-cut CT, which was then merged with the preoperative seizure protocol MRI using Medtronic (Minneapolis, MN) Stealth navigation to obtain frame-based coordinates for amygdala and hippocampus. After an occipital craniostomy and insertion of the laser in the operating room, patients were transported to the MRI suite for thermal imaging, which involved 3–5 serial ablations along the occipital trajectory. A more detailed description of the procedure is available in the initial paper describing this case series.⁵

Neurocognitive testing

All patients underwent a comprehensive neuropsychological evaluation during their presurgical workup and 6–12 months following surgery. A team of bilingual neuropsychologists with experience in multicultural and multilingual assessment completed the evaluations in English or Spanish, depending on the patients' level of language proficiency, acculturation, and educational background. Participants' scores were converted to standardized values using age- and education-matched controls from the Halstead-Reitan Neuropsychological Test Battery.¹⁴ Norms used for Spanish-speaking patients were from a large multinational normative project.^15–17 For all other measures, test-specific normative data were used.

Neuropsychological testing encompassed a broad spectrum of domains including general cognitive ability, memory, confrontation naming, verbal fluency, processing speed, fine motor dexterity, and executive functioning. Test selection followed the National Institute of Neurological Disorders and Stroke recommendations and their availability in English and Spanish.¹⁸ General cognitive ability was assessed in English with the Wechsler Adult Intelligence Scale–Fourth Edition (WAIS-IV)¹⁹ or Spanish with the Escala de Inteligencia Wechsler²⁰ para Adultos (EIWA-III). Verbal and visual memory were assessed with the Wechsler Verbal Memory Scale, Fourth Edition Logical Memory subtests,²¹ the Brief Visual Memory Test–Revised,²² and the Rey Complex Figure Test.²³ Patients also received a language-specific verbal learning and memory measure Rey Auditory Verbal Learning Test ²⁴ or the Miami Assessment of Memory Instrument .²⁵ Confrontation naming was assessed with the 60-item version of Boston Naming Test in English²⁶ or the Multilingual Aphasia Examination (MAE-S) in Spanish.²⁷ Verbal fluency measures included semantic and phonemic subtests of the Controlled Oral Word Association Test.^14,17 Processing speed measures included Trails A from the Trail Making Test (TMT)^14,15 and the WAIS-IV¹⁹ or EIWA-II²⁰ Coding subtest. Fine motor dexterity was assessed with the Grooved Pegboard Test.¹⁴ Measures of executive functioning included Trails B from the TMT^14,15 and the WAIS-IV¹⁹ or EIWA-II²⁰ Digit Span subtest. In instances where comparable or robust normative data were not available for Spanish-speaking patients, a percent accuracy score was used instead.

Quantitative analyses

To ensure that dominant and nondominant hemisphere patients did not vary in demographic and clinical variables that could affect interpretation of outcomes (i.e., age, sex, level of education, and duration of seizure disorder), independent sample t tests and chi-square tests with an alpha level of p < 0.05 were conducted before examining neurocognitive performance.

Presurgical cognitive performance across all measures was also considered an important variable when interpreting postsurgical outcomes. Patients' presurgical performance was classified as “impaired” if their standardized scores were fell below the 5th percentile compared with age-, sex-, education-, and language-matched normative groups. Presurgical impairments, both in select domains typically associated with lateralized temporal lobe functions (i.e., verbal and visual learning and memory) and in other cognitive measures of executive functioning, processing speed, and graphomotor performance, are noted in the figure.

Figure. Clinical case-by-case analysis of neurocognitive outcomes.

++ = participant with right hemisphere language dominance (WADA); -- = scores within normal limits; * = below 5th percentile at baseline; ** = below 5th percentile at follow-up; 1SD+ = improvement of at least 1 SD; 1SD− = decline of at least 1 SD; ^^ = scores improved from previously impaired to within normal limits; !! = scores declined from normal limits to impaired; CS = continued seizures; FDS = free of disabling seizures; u = repeated measure unavailable.

Change in cognitive performance following LiTT was assessed from a clinical perspective by reviewing each participant's profile individually to identify areas where a ±1SD change occurred and whether this change (>1SD) resulted in reclassification of performance from within normal limits to a clinically impaired range or vice versa. We refrained from using statistical tests such as repeated-measures analyses of variance because although the entire case series includes 26 patients, when patients are stratified according to demographic and clinical variables of interest (i.e., side of surgery and language of testing), cell sizes did not have adequate power to meet necessary assumptions to use these statistical approaches. Nevertheless, review of each patients' postoperative change revealed important information on overall cognitive outcome following LiTT.

Data availability

Group-level data (mean T-scores and SD) for each neuropsychological measure included in this study are summarized in table 3. Additional information on procedures and statistical analyses can be shared with qualified investigators upon request. Although the study authors support data transparency and are moving toward practices that will allow for sharing in the future, individual deidentified participant data included in the present study cannot be shared, as they were approved by the UM IRB under the condition that all data “would be considered confidential and not disclosed to any third party.” Neuropsychological protocols are also not available for sharing because of legal and ethical reasons described in Standards for Educational and Psychological Testing published by the American Educational Research Association , American Psychological Association, and the National Council on Measurement in Education.

Table 3.

Presurgical and postsurgical performance by side of surgery and language of testing

Results

Demographics and seizure outcomes

The dominant and nondominant LiTT groups did not differ in demographic, surgical, or clinical variables. Of the sample, 58% were male, 85% were white, and 78% self-identified as Hispanic/Latino. Mean years of education was 11.7 ± 2.9 years. Age at onset of seizure disorder ranged from 1.0 to 59.0 years, with a mean of 15.03 ± 13.61. Duration of seizure disorder before undergoing surgery was 27.3 ± 12.9 years. Patients were aged 42.3 ± 12.1 years at the time of surgery. Surgical variables did not differ by group. Number of ablations ranged from 3 to 6, with an average level of energy delivered of 7,709 ± 471.3 (in joules), resulting in a mean ablation size of 3,176.00 ± 161.75 mm. Additional demographic, clinical, and surgical information organized by group is presented in table 1.

Table 1.

Demographic, clinical, and surgical variables by side of surgery

Post-LiTT seizure frequency was estimated for all patients based on information gathered from neurologic follow-up documentation at the time of their postsurgical neuropsychological evaluation, performed between 6 and 12 months after their ablation with a mean of 8.4 ± 3.3 months. At the time of testing, 21 (81%) patients were free from disabling seizures, whereas the remaining 5 (19%) experienced a reduction but continued to experience recurrent events. Seizure outcomes at 2-year follow-up were recently published, with 61.5% of patients classified as free from disabling seizures and 26.9% experiencing only rare disabling seizures.⁸

During postsurgical evaluations, all but 1 patient remained on the same antiepileptic drug (AED) medications they were taking presurgically. One individual had independently discontinued all AED medications, but had remained seizure-free since surgery. There was no statistical difference between groups in the proportion of seizure control or the number of AED prescriptions at follow-up neuropsychological testing. Additional clinical details organized by group are presented in table 2.

Table 2.

Seizure control and postsurgical AED by group

Presurgical neurocognitive performance

We assessed for any presurgical differences in baseline neurocognitive function between patients with dominant vs nondominant hemisphere epilepsies that could explain differential outcomes following surgery. Groups did not vary on individual measure performance. For complete mean scores of each measure organized by group, please see table 3.

Although no considerable differences were found when comparing group means, the overall representation of patients classified as impaired (<5th %ile) was consistent with findings of previous studies examining presurgical neurocognitive performance of dominant vs nondominant surgical candidates.^5,28,29 In our sample, dominant hemisphere patients represented a greater proportion of those with verbal memory, language fluency, and naming deficits. Conversely, patients with nondominant hemisphere epilepsies comprised a greater proportion of those scoring below the 5th percentile on measures of visual learning, graphomotor processing speed, and mental set shifting skills. For a detailed summary of impaired scores organized by cognitive measure, please see the figure.

Cognitive changes after LiTT

Postsurgical changes were derived by looking only at patients who had reliable presurgical and postsurgical scores. Overall, of the 14 patients who underwent dominant hemisphere ablations, 11 had valid repeated measures. Of the 12 nondominant hemisphere patients, 10 had reliable repeat assessment. To address missing data, percentages were calculated by taking the number of patients who exhibited clinically significant change over the number of patients with reliable repeated measures in each domain.

Verbal learning and memory

Of 7 patients who experienced clinically significant change in verbal learning, 6 improved (dominant n = 3 of 11 subjects, 27%; nondominant n = 3 of 9 subjects, 33%). Only 1 patient (9%) from the dominant group experienced >1SD decline with a 54% drop from their presurgical score. For delayed recall, although the same number (n = 6) experienced clinically significant improvement, this occurred with greater frequency in the nondominant hemisphere group (dominant n = 1, 9%; nondominant n = 5, 56%). Decline was only observed in 2 (18%) patients from the dominant hemisphere group who experienced a 25%–33% drop from their presurgical scores. Contextual verbal memory of a short story was less sensitive to postsurgical change with 1 of 6 subjects in the nondominant hemisphere group experiencing improvement (17%) and 2 patients (dominant n = 1 of 10 subjects, 10%; nondominant n = 1 of 6 subjects, 17%) demonstrating decline with a 25%–33% drop from their presurgical scores.

Fewer patients experienced clinically significant change in visual leaning (n = 3, 19%). Two saw improvement (dominant n = 1 of 8 subjects, 13%; nondominant n = 1 of 8 subjects, 13%), and 1 (13%) nondominant patient experienced a 33% decline from their presurgical score. On the other hand, 3 (19%) patients experienced improvement in delayed visual recall (dominant n = 1, 13%; nondominant n = 2, 25%), whereas 3 (38%) dominant hemisphere patients experienced >1SD decline (20%–32% drop from their presurgical scores). On a more complex measure of incidental visual memory, 4 patients saw improvement (dominant n = 3 of 10, 30%; nondominant n = 1 of 6, 17%). Only 1 (10%) patient who underwent dominant hemisphere ablation experienced clinically significant decline (49% drop from their presurgical score).

Confrontation naming

Of a total of 20 patients, 30% experienced clinically significant improvement (dominant n = 2 of 10 subjects, 20%; nondominant n = 4 of 10 subjects, 40%), and 10% declined (dominant n = 1, 10%; nondominant n = 1, 10%). All patients with deterioration in their score (drop of 20%–38% from baseline) were Spanish speakers and were assessed with the MAE.

Verbal fluency

Twenty patients had repeat assessment of phonemic verbal fluency. Thirty-five percent (n = 7) experienced improvement (dominant n = 3 of 11 subjects, 27%; nondominant n = 4 of 9 subjects, 44%). There were no patients who experienced clinically significant decline. Of 17 patients with repeat assessment of sematic verbal fluency, only 1 experienced clinically significant decline with a 24% drop in the score. This patient had undergone a dominant hemisphere procedure and was free of disabling seizures.

Processing speed and executive functioning

The greatest percentage change in postsurgical performance occurred in nonmemory or language domains. Improvement in speeded visual scanning was seen in 60% of patients (dominant n = 4 of 9 subjects, 44%; nondominant n = 5 of 6 subjects, 83%). Only 1 (11%) patient who underwent dominant hemisphere LiTT experienced a clinically significant decline with a 52% drop in the score. However, this patient also experienced a surgically related visual field cut, which has been discussed in a previous article,³⁰ that likely affected performance on this task. Clinically significant improvement in graphomotor processing speed occurred in 29% of patients (dominant n = 3 of 10 subjects, 30%; nondominant n = 2 of 7 subjects, 29%). Fine motor dexterity and speed improved in 22% of patients (dominant n = 2 of 11 subjects, 18%; nondominant n = 2 of 7 subjects, 29%).

Eighty-three percent of patients experienced clinically significant improvement in a mental set shifting task, all from the nondominant hemisphere group (5 of 6 subjects). Similarly, 25% of patients had a greater than 1SD improvement on a working memory measure (dominant n = 3 of 11 subjects, 27%; nondominant n = 2 of 9 subjects, 22%). Only 1 dominant hemisphere patient experienced clinically significant decline in this measure.

Discussion

Although several recent reports have evaluated the efficacy of LiTT at achieving seizure freedom for patients with mTLE, investigations assessing the influence of this procedure on cognitive outcomes have been limited to studies that focus exclusively on select measures of language and memory. Although small sample size, single-center investigation, and lack of alternative surgical approach as a comparison group remain concerns in our study, we try to present a broader review of cognitive changes across multiple domains and further explore how side of surgery may influence overall neuropsychological outcomes.

Overall, improvement in at least 1 measure occurred in 90% of subjects with similar frequency in patients who underwent dominant or nondominant hemisphere ablations. Most improvements occurred in nonmemory domains including processing speed, mental set shifting, and fine motor dexterity. Although a select number of patients in our series who underwent dominant hemisphere procedures experienced a decline in language and memory performance, overall outcomes were favorable when compared to estimated rates of impairment following ATL, and similar to other studies, suggest that LiTT may offer greater preservation of neurocognitive function following surgery relative to open resections.^2,4

Although most patients in this series achieved full remission of disabling seizures, degree of seizure control was also a consideration. However, rates of clinically significant change did not appear to differ between patients who were considered “seizure-free” vs those who still had recurrent seizure activity but no longer experienced “disabling seizures.” Nevertheless, prospective and longitudinal data on likelihood of continued seizure freedom and effects on cognition LiTT are still needed. Further analyses of clinical and surgical variables (i.e., age at the time of surgery and total ablation temperature) are also needed so that we can understand what variables may be predictive of positive surgical outcomes or areas that may be at risk of impairment. Prospective data are needed to fully understand how these variables translate to individual patient outcomes. Future research should integrate volumetric imaging information that will allow us to contrast individual patients' comprehensive neuropsychological profiles with presurgical volumes of ipsilateral and contralateral regions furthering our understanding of how ablation of specific regions in mesial temporal structures can contribute to postsurgical cognitive change.

Last, this study also provides the unique opportunity to evaluate cognitive outcomes in a cohort that has been largely underrepresented in the literature, Spanish-speaking Hispanics in the United States. Although we acknowledge that limited normative data on select measures are available for use with this cohort presents a challenge to generalizability, this is true of all cognitive research conducted with this segment of the population and highlight the need for normative samples that accurately represent changing trend in the population of the United States.

This study highlights the importance of considering a broader perspective on cognitive outcomes that moves beyond selectively examining language or memory domains in isolation. Improvement occurred in a substantial portion of patients (90%). However, this information would have been missed should the focus remain exclusively on language and memory measures, as is the case in most published research discussing epilepsy surgery outcomes.

Although decline was noted in a select number of patients in our series, the incidence remains lower than what previously published literature suggest can occur in more invasive procedures such as ATL, where rates range from 40% to 60%. Overall, these findings suggest that LiTT may offer greater preservation of cognitive function compared with other surgical approaches and that patients may experience improvement in domains not typically associated with mesial temporal structures such as processing speed, aspect of executive functioning, and fine dexterity. We speculate that improvement in these areas is related to significant reduction of seizure burden.

Understanding the overall scope of cognitive change in LiTT may have important implications in educating patients' regarding the potential risk vs benefit equation of undergoing surgery. However, these areas must be investigated further to gather a more comprehensive understanding of other variables that may mediate postsurgical outcomes with this novel technique. Our results suggest that considering presurgical cognitive performance may be of importance in identifying patients at an increased risk of cognitive decline.

Although limitations to this study included selection bias for severely intractable cases of epilepsy and variability in the presence of MTS, the sample is generally consistent with the population of persons seeking surgical treatment for epilepsy. Overall, cognitive outcomes following LiTT in a culturally and linguistically diverse sample appear promising. As we continue to follow this cohort of patients, we will be able to gather longitudinal information that may prove vital in developing tailored cognitive rehabilitative programs for epilepsy surgery patients.

TAKE-HOME POINTS

• → Cognitive changes following LiTT for the treatment of intractable MTS can occur in both memory and nonmemory domains.

• → Cognitive performance improved in patients who underwent both dominant and nondominant hemisphere procedures, but with greater frequency in the nondominant group.

• → In comparison to the outcome literature of dominant open ATL resections, our LiTT case series exhibited a much lower incidence of postoperative language or verbal memory decline.

• → Normal presurgical cognitive performance represented a risk factor for clinically significant decline following LiTT.

Appendix. Authors

Study funding

No targeted funding reported.

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.