Abstract
Donos C, Breier J, Friedman E, Rollo P, Johnson J, Moss L, Thompson S,
Thomas M, Hope O, Slater J, Tandon N Epilepsia
2018;59:1421–1432. Laser interstitial thermal therapy (LITT) is a minimally invasive
surgical technique for focal epilepsy. A major appeal of LITT is
that it may result in fewer cognitive deficits, especially when
targeting dominant hemisphere mesial temporal lobe (MTL) epilepsy.
To evaluate this, as well as to determine seizure outcomes following
LITT, we evaluated the relationships between ablation volumes and
surgical or cognitive outcomes in 43 consecutive patients undergoing
LITT for MTL epilepsy. All patients underwent unilateral LITT targeting mesial temporal
structures. FreeSurfer software was used to derive cortical and
subcortical segmentation of the brain (especially subregions of the
MTL) using preoperative magnetic resonance imaging (MRI). Ablation
volumes were outlined using a postablation T1-contrasted MRI. The
percentages of the amygdala, hippocampus, and entorhinal cortex
ablated were quantified objectively. The volumetric measures were
regressed against changes in neuropsychological performance before
and after surgery. A median of 73.7% of amygdala, 70.9% of hippocampus, and 28.3% of
entorhinal cortex was ablated. Engel class I surgical outcome was
obtained in 79.5% and 67.4% of the 43 patients at 6 and 20.3 months
of follow-up, respectively. No significant differences in surgical
outcomes were found across patient subgroups (hemispheric dominance,
hippocampal sclerosis, or need for intracranial evaluation).
Furthermore, no significant differences in volumes ablated were
found between patients with Engel class IA vs Engel class II-IV
outcomes. In patients undergoing LITT in the dominant hemisphere, a
decline in verbal and narrative memory, but not in naming function
was noted. Seizure-free outcomes following LITT may be comparable in carefully
selected patients with and without MTS, and these outcomes are
comparable with outcomes following microsurgical resection. Failures
may result from non-mesial components of the epileptogenic network
that are not affected by LITT. Cognitive declines following MTL-LITT
are modest, and principally affect memory processes.
Youngerman BE, Oh JY, Anbarasan D, Billakota S, Casadei CH, Corrigan
EK, Banks GP, Pack AM, Choi H, Bazil CW, Srinivasan S, Bateman LM,
Schevon CA, Feldstein NA, Sheth SA, McKhann GM II for the Columbia
Comprehensive Epilepsy Center Co-Authors. Epilepsia
2018;59:595–606. Selective laser amygdalohippocampotomy (SLAH) using magnetic
resonance-guided laser interstitial thermal therapy (MRgLITT) is
emerging as a treatment option for drug-resistant mesial temporal
lobe epilepsy (MTLE). SLAH is less invasive than open resection, but
there are limited series reporting its safety and efficacy,
particularly in patients without clear evidence of mesial temporal
sclerosis (MTS). We report seizure outcomes and complications in our first 30 patients
who underwent SLAH for drug-resistant MTLE between January 2013 and
December 2016. We compare patients who required
stereoelectroencephalography (SEEG) to confirm mesial temporal onset
with those treated based on imaging evidence of MTS. Twelve patients with SEEG-confirmed, non-MTS MTLE and 18 patients
with MRI-confirmed MTS underwent SLAH. MTS patients were older
(median age 50 vs 30 years) and had longer standing epilepsy (median
40.5 vs 5.5 years) than non-MTS patients. Engel class I seizure
freedom was achieved in 7 of 12 non-MTS patients (58%, 95%
confidence interval [CI] 30%–86%) and 10 of 18 MTS patients
(56%, 95% CI 33%–79%), with no significant difference between
groups (odds ratio [OR] 1.12, 95% CI 0.26–4.91, P = .88).
Length of stay was 1 day for most patients (range 0-3 days).
Procedural complications were rare and without long-term
sequelae. We report similar rates of seizure freedom following SLAH in patients
with MTS and SEEG-confirmed, non-MTS MTLE. Consistent with early
literature, these rates are slightly lower than typically observed
with surgical resection (60%–80%). However, SLAH is less
invasive than open surgery, with shorter hospital stays and
recovery, and severe procedural complications are rare. SLAH may be
a reasonable first-line surgical option for patients with both MTS
and SEEG confirmed, non-MTS MTLE.
Grewal SS, Zimmerman RS, Worrell G, Brinkmann BH, Tatum WO, Crepeau
AZ, Woodrum DA, Gorny KR, Felmlee JP, Watson RE, Hoxworth JM, Gupta
V, Vibhute P, Trenerry MR, Kaufmann TJ, Marsh WR, Wharen RE Jr, Van
Gompel JJ [Epub ahead of print] J Neurosurg doi:
10.3171/2018.2.JNS171873 Although it is still early in its application, laser interstitial
thermal therapy (LiTT) has increasingly been employed as a surgical
option for patients with mesial temporal lobe epilepsy. This study
aimed to describe mesial temporal lobe ablation volumes and seizure
outcomes following LiTT across the Mayo Clinic's 3 epilepsy
surgery centers. This was a multi-site, single-institution, retrospective review of
seizure outcomes and ablation volumes following LiTT for medically
intractable mesial temporal lobe epilepsy between October 2011 and
October 2015. Pre-ablation and post-ablation follow-up volumes of
the hippocampus were measured using FreeSurfer, and the volume of
ablated tissue was also measured on intraoperative MRI using a
supervised spline-based edge detection algorithm. To determine
seizure outcomes, results were compared between those patients who
were seizure free and those who continued to experience
seizures. There were 23 patients who underwent mesial temporal LiTT within the
study period. Fifteen patients (65%) had left-sided procedures. The
median follow-up was 34 months (range 12–70 months). The mean
ablation volume was 6888 mm3. Median hippocampal ablation
was 65%, with a median amygdala ablation of 43%. At last follow-up,
11 (48%) of these patients were seizure free. There was no
correlation between ablation volume and seizure freedom (p = 0.69).
There was also no correlation between percent ablation of the
amygdala (p = 0.28) or hippocampus (p = 0.82) and seizure outcomes.
Twelve patients underwent formal testing with computational visual
fields. Visual field changes were seen in 67% of patients who
underwent testing. Comparing the 5 patients with clinically
noticeable visual field deficits to the rest of the cohort showed no
significant difference in ablation volume between those patients
with visual field deficits and those without (p = 0.94). There were
11 patients with follow-up neuropsychological testing. Within this
group, verbal learning retention was 76% in the patients with
left-sided procedures and 89% in those with right-sided
procedures. In this study, there was no significant correlation between the
ablation volume after LiTT and seizure outcomes. Visual field
deficits were common in formally tested patients, much as in
patients treated with open temporal lobectomy. Further studies are
required to determine the role of amygdalohippocampal ablation.Objectives
Methods
Results
Significance
Objective
Methods
Results
Significance
Objective
Methods
Results
Conclusions
Commentary
MRI-guided laser interstitial therapy (MRg-LITT), also known as stereotactic laser ablation, uses laser energy to ablate neural tissue. When used on medial temporal structures for the treatment of medial temporal lobe epilepsy (MTLE), the operation can be called stereotactic laser amygdalohippocampotomy (SLAH; “-otomy” because those structures are not technically removed as in amygdalohippocampectomy). Since the first reports of its use (1, 2), SLAH has gained very rapid and widespread adoption. Thus, we now have a reasonably large experience to examine, with seven independent groups having published their data at various outcome points (2–7), the most recent example of which was published by Donos et al. To provide a backdrop for comparison, I will provide an aggregation of these reports.
Comparing across different cohorts can be confounded by differences in reporting conventions, even when all use the Engel classification. Some of the following comments on a few issues are relevant:
By convention, the Engel classification is not applied until the 1-year anniversary, an approach endorsed in the International League Against Epilepsy (ILAE) rating system. Outcomes less than 12 months are therefore not typically presented with the Engel classification. Outcomes at 12 or more months are also more readily compared with Wiebe et al.'s (8) benchmark publication on anterior temporal lobectomy (ATL), which tabulated its data at that time point;
Reports documenting outcomes at “last follow-up” can conflate both 1) patients with outcomes less than 12 months that might be seizure free for a time but recur before they reach 12 months, and 2) patients with much longer-term outcomes, who suffer late recurrences just as with open resections;
The Engel 1D classification is for patients that have ‘generalized
seizures with antiepileptic drug (AED) discontinuation’. This bucket is a controversial one, designed for patients whose
seizure(s) is the result of too rapid discontinuation. We have
adopted a conservative approach to this in our work: patients that
“miss a couple of doses” and have their habitual
complex partial seizure are called Engel 2. Some reports, however,
do not stipulate this and so some blurring in outcomes occurs around
this designation.
With these caveats in mind, we move to the tabulation. The reports published prior to Donos et al. comprise 175 unique patients, from series with Ns of 5 (8), 15 (6), 23 (5), 30 (9), 21 (7), 23 (3), and 58 (4). Overall, 101 of 175 (57.7%) were free of disabling seizures at a variety of outcome points. Further audit yielded 101 patients with mesial temporal sclerosis (MTS) with Engel 1 outcomes in 63 (62.4%). Remember that this group contains some patients with less than 12-months follow-up. More conservatively, confining the aggregation to those patients with 12 or more months follow-up yields Engel 1 outcomes in 96 of 168 (57.1%) patients overall, and 58 of 91 (63.7%) with MTS.
How does this compare with the latest series? Donos et al. is an early adopting site that performed their first surgery in June 2012. They performed SLAH on 43 patients with MTLE, predominantly with MTS (n = 34). All patients without MTS had their onset zones validated by stereoelectroencephalography (sEEG). Clinical outcomes were tabulated at 6, 12, and 24 months; however, the length of follow-up for individual patients was not noted. Thirty-nine (personal communication; erratum in progress) of 43 patients had 6 months or more outcome, so at least four patient's outcomes were tabulated somewhere between 0 and 6 months following surgery. One-year outcomes were available for only 18 patients.
Superior Seizure-free Outcome
The authors reported that 67.4% of 43 patients were Engel 1 at a mean of 20.3 months ± 13.8 (SD) follow-up. However, contained within this number are 25 patients with a less than 12-month outcome, four of whom had a less than 6-month follow-up. Thus, this number potentially inflates the effectiveness as compared, say, with the Wiebe et al. (8) ATL benchmark. The authors further report that 79.5% were Engel 1 at 6-months follow-up (N = 39). Again, this number has potential to be inflated as compared with 12-month outcomes. Unfortunately, the only 12-month outcome data are in the form of a Kaplan-Meier analysis (Engel 1A, 78% [CI 63%–92%]; Engel 1, 85% [CI 73%–97%]). While those numbers look remarkable, this is not an ‘intent-to-treat’ type of analysis, and not strictly comparable to other series’ 12-month data.
The results in Donos et al., therefore, look very promising but are not strictly comparable to the series’ results discussed above: 1) With respect to the ‘last follow-up’ tally, indeed their data appears superior to the aggregated data, 67.4% versus 57.7% Engel 1, caveat emptor. 2) With respect to 12 or more month data, we do not have it from Donos et al.'s 18 patients. If we make the leap to compare their 6 or more month data to the aggregate 12 or more month data (in truth, few patients that make it to 6 month as Engel 1 deteriorate by 12 months) (3), Donos et al.'s results again appear substantially superior to the aggregated data, 79.5% versus 57.1% Engel 1. 3) There are no confidence intervals for either Donos et al.'s 6-month data, or for the aggregated data I provided. The closest we have are the data from my group's paper following 58 patients for 12 months, 53.4% (95% CI 40.8–65.7%) (3). The Donos et al. mean is outside our CI, but we do not know their lower bound.
Volume Ablated Matters
Given the likely superior results attained by Donos et al., what are the contributing factors? It is possible that they more aggressively ablated the medial temporal structures (in fact they used thermal parameters as high as 95%, which is quite high and will usually quickly trigger off the laser due to heating near the fiber), but this is difficult to determine with respect to lesion parameters, which vary by power and time. Importantly, if higher parameters were used, they were done so with careful attention to location of the probe, as there were no cranial nerve deficits or visual field deficits, for which the group should be congratulated.
To shed more light on this, the authors analyzed ablation volumes using the FreeSurfer software. In the 39-patient subset for which they quantified the ablation volume, the percentages of ablation were 73.7 ± 13.4% of amygdala, 70.9 ± 12.6% of hippocampus, 30.8 ± 9.9% of parahippocampal gyrus, and 28.3 ± 15.3% of entorhinal cortex. How does this compare with other series? 1) Total volume ablated was not provided by Donos et al. to compare with other series. 2) The percentage ablation values of the amygdala and hippocampus, but not the parahippocampal gyrus or entorhinal cortex, were larger than previously published by Kang et al. (6) (48–50% of amygdala; 52.8–62.5% hippocampus; 27.7–36.8% parahippocampal gyrus; 22.3–25.1% entorhinal cortex). Notably, the latter only achieved Engel 1 outcomes in four of 11 patients overall, and four of 10 with MTS, the lowest Engel 1 outcomes in the aggregated series. Comparing the volumes with Jermakowicz et al. (4), Donos et al. again found slightly greater amygdala ablations (73.7% vs 59–66%), but slightly smaller hippocampal ablations (70.9% vs 75–80%). Donos et al. had much lower percentage of the hippocampal tail ablated, which could indicate it is less important (likely to be the case) or may have been due to methodologic differences. It was harder to compare other structures across these two studies.
These results support that ablation volume may well have contributed in part to the outcomes, most particularly and not expectedly, in the amygdala. Similar to Jermakowicz et al. (4), Donos et al. found no significant difference in ablation percentages comparing Engel 1A to Engel 2 to 4 patients; and they did not replicate the former's observation that hippocampal head remnant prognosticated poor outcome, but it is possible they universally achieved better ablation of the medial uncus. It is important to note that ‘absence of evidence is not evidence of absence’: the failure of these studies to detect an influence of a particular structure's ablation percentages on outcome is at least as likely a manifestation of the variance of the volume analysis as it is true absence of effect, as those variances in each study are very high. In the absence of robust evidence to the contrary, it is important to remember that the technical goal is complete ablation of the hippocampus as close to the tectal plate (posterior body) as possible, and as complete of an ablation of the amygdala as feasible.
MRI Normal MTLE Remains a Legitimate Indication
An excellent outcome was achieved in the small number of patients in this series with non-MTS MTLE (6/9, 67%) (above caveats notwithstanding). Ostensibly, this was the result of using sEEG in seven of them (5/7 Engel 1), although one of two without sEEG became seizure free as well. It certainly makes sense to confirm onsets related to the ablated structures in most/all cases. However, others have used sEEG as well. Tao et al. (7) achieved Engel 1 in only three of 10 (30%) non-MTS patients and Youngerman et al. (9) in seven of 12 (58%). sEEG per se is important, but not the only factor in non-MTS cases; certainly, ablation volume and perhaps specific targets also factor in. Many non-MTS patients benefit well from this procedure, although as with open surgery these patients are prone to a higher rate of failure.
Cognitive Impact
Neuropsychological testing was performed in 31 patients at approximately 6 months and related to the volumetric analysis of the ablation zones for predictive analytics. Similar to the results my group has reported (9), there was no change in naming or category or semantic fluency. Group-wise there was a statistically significant decline in domain-specific memory measures, although no correction was used for multiple tests. Subgroup analysis confirmed that these changes occurred even in the MTS patients, including a decrease in verbal IQ and increase in performance IQ in the dominant hemisphere patients. These memory declines are not unexpected but should be taken in context. As we described (3), verbal memory decline occurred in 8.2% of patients overall, and 15% of dominant-side patients, whereas it improved in 10%, and improved in 24% of nondominant patients. These results are far below the 30 to 60 percent rates of decline reported for patients undergoing open resection. Finally, Donos et al. did a linear regression of postablation cognitive change against preoperative test scores, preoperative hippocampal volume, and the percentage of hippocampus or entorhinal cortex ablated. While there was an association between larger hippocampal and entorhinal ablations with worse memory performance, only the preoperative test performance was a statistically significant predictor of postoperative performance, and even at that, only when not correcting for multiple (39) comparisons.
Conclusion
There is by now a fairly sizable published experience with MRg-LITT SLAH. Although variations in timing of outcome reporting and inconsistencies in Engel classification use compromise comparisons across studies and aggregation of findings, strong statements can nevertheless be made about the effectiveness of SLAH. Although the results reported in Donos et al. appear to be superior, and likely point to the benefit of carefully targeted and maximized ablations while maintaining safe distances from cranial nerves and optic pathways, they fit into a continuum of results that can be summarized. Including their experience, but using their more than 6-month outcome pool (n = 39) rather than the whole cohort (which includes 4 patients whose outcomes were tallied prior to 6 months, an epoch when failures do occur; there are few failures in the literature between 6 and 12 months), by my count there have been published outcomes on 211 SLAHs, of whom 129 (61.1%) have been free of disabling seizures (Engel 1). This includes patients with MTS and those without, and occasional other dual pathologies. Censoring for only MTS patients yields 135 patients, of whom 86 (63.7%) are Engel 1 (this does not imply that the remainder were non-MTS; patients were lost in this analysis due to absence of information as to MTS status as well). When analysis is confined to more than 12-month outcomes, the Engel 1 numbers are 57.1% (96/168) for all patients and 63.7% (58/91) for patients with MTS.
This is a substantial accumulation of consecutive patients from multiple centers where all patients have been accounted for. While not class 1 evidence, it yields strong evidence for the effectiveness (and overall safety) of SLAH. The results, while not controlling for selection or investigator bias, as the case in a randomized controlled trial, such as Wiebe et al. (8), are getting close to their benchmark of 64% of operated patients (85% with MTS) free of disabling seizures following ATL. And neurocognitive results support the conclusion that SLAH protects critical functions, including naming and object recognition, that are at high risk from open resection, and even does better with respect to memory outcomes. While a randomized controlled trial would be nice to compare SLAH with open resection, it will never happen in the United States given availability of MRg-LITT and patients’ strong preferences. We await the results of the ongoing SLATE (Stereotactic Laser Ablation for Temporal Lobe Epilepsy; ClinicalTrials.gov Identifier: NCT02844465), a prospective single-arm industry-sponsored clinical trial that is well underway. However, the data accumulated to date strongly support the use of SLAH to treat MTLE.