Comparative Review of Seizure and Cognitive Outcomes in Resective, Ablative, and Neuromodulatory Temporal Lobe Epilepsy Surgery

Abstract

Resective surgery for drug-resistant temporal lobe epilepsy remains underutilized in the United States. While anteromesial temporal lobectomy consistently achieves the highest rates of long-term seizure freedom, it comes with greater risks for memory and language decline. Magnetic resonance imaging-guided laser interstitial thermal therapy and neuromodulation have gained popularity due to perceived lower surgical risk and faster recovery, although they yield lower rates of sustained seizure freedom. Neuromodulation with vagus nerve, deep brain, or responsive neurostimulation provides an option for patients ineligible for resection or ablation, but overall seizure outcomes remain modest. Balancing improved seizure control with open resection against the potential cognitive advantages of less invasive treatments is complex, requiring careful patient selection. Future research must refine these approaches to optimize results. Thoughtful, individualized decision-making, guided by each patient's clinical scenario and goals, is paramount for achieving the best balance between seizure freedom, cognitive preservation, and overall patient outcome.

Introduction

Despite randomized trials demonstrating the safety and efficacy of anteromesial temporal lobectomy (ATL),^1–3 years of multidisciplinary practice parameters recommending referral for surgical evaluation,^4,5 and refinements of microsurgical approaches to selective amygdalohippocampectomy (SAH), ⁶ resective epilepsy surgery remains woefully underutilized in the United States.^7–9 MRI-guided laser interstitial thermal therapy (LITT) and neuromodulation have grown in popularity as less invasive alternatives touting fewer complications and faster postoperative recovery. ⁸ This review presents the surgical risks and benefits of each approach, focusing on seizure and cognitive outcomes. We also identify knowledge gaps in need of further research.

Surgical Fundamentals

Technical Overview

ATL is a resective approach pioneered by Dr. Wilder Penfield in the 1950s ¹⁰ as he reported pathological findings extending beyond the mesial temporal structures to the temporal pole and superior temporal gyrus. Since then, ATL has come to refer to any open resection of the mesial temporal structures and pole with variable resection of the parahippocampus and additional neocortical regions. Understanding the anatomy of the hippocampus, amygdala, and piriform cortex is critical for safe and effective resection. Open resection provides unparalleled access to the temporal lobe offering: (1) direct visualization of the surgical field theoretically facilitating hemostasis; (2) a contoured and tailored 3-dimensional resection, as compared to the linear restrictions and finite stereotactic freedom in LITT; (3) tactile feedback of tissue stiffness during microsurgical resection to enhance or extend the resection; and (4) access to tissue for pathological review (and when appropriate, for research purposes). Conversely, LITT is a minimally invasive stereotactic procedure. A less than 2 mm diameter catheter is inserted through a less than 1 cm incision. Deep targets can be reached with minimal disruption of the overlying brain. Time-dependent tissue necrosis and near real-time temperature monitoring with MR thermography minimize the risk of off-target injury. “Pull backs” along the insertion trajectory allow for larger ablations; and cerebrospinal fluid heat sinks contoured the ablation to curved structures such as the hippocampus.

Surgical Complications

Recovery and adverse events with LITT or selective laser amygdalohippocampotomy (SLAH), compare favorably to open resection. Patients typically spend 1 night in the hospital and report minimal pain. An 11-center LITT study of 268 patients, ¹¹ the largest to date, reported 0.4% persistent homonymous hemianopsia, 1.1% persistent extraocular movement dysfunction, 0.4% symptomatic hemorrhage, and 0 infections or 30-day perioperative deaths. A meta-analysis estimated the rate of major complications with LITT at 3.8% compared to 10.9% with ATL and 7.4% with SAH. ¹² The favorable safety outcomes are notable given that they reflect the earliest experience with LITT compared to the matured historical experience with open resection, suggesting a more favorable learning curve.

Seizure Outcomes

Early reports of LITT suggest seizure-freedom rates below open resection, particularly when long-term results are considered. One meta-analysis estimated 57% Engel I outcome with LITT compared to 69% for ATL and 66% for SAH historical controls. ¹² Another estimated 64% at 1 year, 47% at 2 years, and 42% at 3 years, ¹³ with no long-term outcomes out to 10 years. This is a concerning trend when compared to seizure-freedom rates of 70-80% at 1 year, 66% at 5 years, and 50% at 10 or more years after ATL.^14–17 There are, however, 2 caveats to consider.

First, LITT outcomes reported to date represent the earliest experience of an evolving surgical technique. For example, current practice favors the use of 2 laser probes to optimize the ablation as opposed to the early single-probe approach. Second, experience to date includes a more heterogenous cohort of patients with variability in the reported LITT outcomes: the largest multicenter observational study to date reports seizure freedom of 55.8% at 1-year and 49.3% at last follow-up in (median 47, range 12-95 months) ¹¹ ; and 1 of the largest single-center experiences (n = 48) reports 60.4% Engel 1 at mean follow-up 50 months, ¹⁸ suggesting that more durable outcomes may be achievable with LITT in a well-selected patient subgroup. The challenge is defining this subgroup. While inconsistent, some retrospective LITT studies have reported better seizure outcomes with mesial temporal sclerosis (MTS)^18,19 and concordant Phase I workup (eg PET hypometabolism, ¹¹ unitemporal interictal discharges ¹⁹ ); and worse outcomes with focal to bilateral tonic-clonic seizures^11,18 and nonlateralizing or discordant Phase I data.^11,18–20 Stereoelectroencephalography (sEEG) seizure onset pattern, in particular hippocampal low-voltage fast activity and low-frequency repetitive spiking, has also been associated with seizure freedom. ²¹ Patients without MTS may also do well if seizure onsets are localized with sEEG. ²² These outcome predictors are nonspecific to LITT and simply mirror those of ATL, offering little guidance in identifying patients who can achieve the same long-term success seen with ATL.

Cognitive Outcomes

Several lines of evidence suggest more favorable cognitive outcomes with LITT as compared to ATL.

1. Comparison of LITT to published experience with open resection: LITT has been shown to preserve language networks when the epileptogenic zone does not involve the lateral and basal temporal lobe language networks.^23–25 LITT also results in less severe and less frequent memory decline when using classic declarative memory measures.^24–27 Mood outcomes are also stable, and there appears to be less likelihood of “double losers” (eg, persons who are not seizure-free following a procedure and experience cognitive decline).^28,29

2. Direct comparisons of SLAH to open resection: Data in this area is limited to a handful of studies with no randomized case-controlled studies completed. Nevertheless, all findings thus far have been favorable to LITT for language, object/face recognition, and declarative memory as compared to larger open resections (with data including ATL and some of SAH).^30–32

3. High-risk patients: Epilepsy patients considered to be at high risk for cognitive decline (eg, in the context of a failed WADA) have successfully undergone LITT while similar patients undergoing open resection have been more likely to decline. ³³

4. Patients undergoing repeat or both procedures: Neuropsychological sequelae were more prevalent with ATL than LITT when consecutively performed in the same patients in a recent study, including decline in visual naming and functional status. ³⁴ Many gaps still need to be addressed.

First, the most favorable cognitive LITT outcomes were generated alongside the less favorable seizure outcomes reported to date. It is possible that as changing LITT practices with larger or differently located ablations potentially alters seizure outcomes for the better, cognitive outcomes could potentially worsen. Second, cognitive outcomes vary among individuals. Many patients show no cognitive decline post-ATL; and some improve, even in the dominant hemisphere, especially if seizures are controlled and medication is reduced. While there is a subset of patients in whom open resection may hasten cognitive decline in the short term, particularly if seizures persist, memory decline may be stopped and even reversed in the long term, if seizures are fully controlled. ³⁵ In parallel, LITT is not without cognitive risk. A recent meta-analysis found that up to 30%, depending on the domain and surgical side, demonstrate postoperative declines in memory or naming following LITT for TLE. ³⁶ Thoughtful and individualized decision-making is essential. Third, patient perspectives, values, and goals must be considered. Over 70% of patients who undergo ATL are satisfied with their outcome, ³⁷ and >85% say they would repeat surgery if they had to do it over. ³⁸ Satisfaction is most strongly predicted by postoperative seizure freedom, medication reduction, and quality of life improvements.^37,38 Those whose presurgical expectations and goals are met are more likely to report satisfaction with their surgical outcome. ³⁹ It is therefore important to consider patient goals in the decision-making process. For example, a truck driver with left MTS and poor preoperative memory and language scores, seeking to resume driving as soon as possible to regain his livelihood might favor open resection over LITT given his low risk of further cognitive decline with resection and higher risk of ongoing seizures with LITT. A college professor with normal brain imaging and a superior cognitive baseline will likely have different priorities.

Survival and Mortality

Immediate perioperative mortality is extremely low with both LITT and ATL, but long-term survival data favor open resection. Compared with medical management, ATL increased survival by 5.0 years (95% CI, 2.1-9.2) and quality-adjusted life expectancy by 7.5 quality-adjusted life-years (95%, CI, −0.8 to 17.4) in 1 decision curve analysis, primarily due to increased years spent without disabling seizures, thereby reducing seizure-related excess mortality and improving quality of life. ⁴⁰ In another study of 1110 patients with DRE (1006 surgically and 104 non-surgically treated; total follow-up of 8126.62 person-years), open resection reduced the mortality rate around 3 times from 25.3 deaths per 1000 person-years to 8.6 per 1000 person-years (P < .001), with seizure-free patients faring 5× better (5.2 per 1000 person-years). ⁴¹ In contrast, in an observational study of 135 patients treated with LITT with 501.3 person-years at risk, ⁴² SUDEP incidence was higher after LITT than in cohorts treated with resective surgery, and similar to nonsurgical controls from pooled historical data. Based on these findings, the paper concludes that LITT is not effective in reducing SUDEP incidence in patients with DRE, likely due to its lower rates of long-term seizure freedom when compared to open resection, and emphasizes the importance of targeting complete seizure freedom when planning surgical therapy, including early consideration for resection when LITT fails.

Neuromodulation When Neither Resection nor Ablation Are Options

Vagus nerve stimulation (VNS), responsive neurostimulation (RNS), and deep brain stimulation of the anterior nucleus of the thalamus (ANT-DBS) are the only neuromodulation devices approved in the US for the treatment of refractory focal epilepsy that has been assessed in appropriately powered double-blind randomized controlled trials. ⁴³ VNS uses scheduled (open-loop) stimulation of the left vagus nerve. Recent models make use of closed-loop stimulation triggered by predetermined changes in heart rate that may indicate ongoing seizure activity. RNS relies on a cranially-seated neurostimulator with 2 depth and/or cortical strip leads, each with 4 electrodes, that are implanted in the identified seizure focus or foci. The RNS device continually interprets electrocorticographic (ECoG) activity and is programmed to deliver closed-loop stimulation in response to specific pre-ictal and ictal ECoG patterns. ⁴⁴ ANT-DBS makes use of open-loop stimulation and is thought to modulate the synchronization of brain activity and epileptic networks through ANT connections with the Papez circuit.^44,45 Approximately half of patients implanted with these devices have a 50% or greater reduction in seizures, and all demonstrate increased efficacy over time. ⁴⁶ A recently published systematic review showed that long-term cognitive outcomes are at least stable following VNS, DBS, and RNS implantation. ⁴⁷ Despite the limitations presented by the published data given heterogeneity in neuropsychological testing, methodologies, and study sizes, there is evidence that neuromodulation is not associated with the adverse cognitive impact that may be associated with resective or ablative epilepsy surgery. This appears to be true with either open-loop or closed-loop neurostimulation techniques, and with stimulation of the vagus nerve, direct stimulation of the seizure focus, or with stimulation of other nodes within the seizure network.

Neuromodulation is therefore an attractive treatment option when neither resection nor ablation are possible. It should not be considered as a first-line therapy given its significantly inferior seizure outcomes (Table 1).

Table 1.

Summary of Seizure Outcomes at 9 Postoperative Years for Available Surgical Therapies of Drug-Resistant Temporal Lobe Epilepsy.

	6-month remission (%)	1-year remission (%)	% seizure-free (defined as 1-year remission) at last follow-up	Duration of seizure freedom at last follow-up	% reduction in seizure-frequency	% with >50% reduction in seizure frequency
ATL 48	93.2%	92.4%	75.2%	11.9 yrs	88.1% overall. 76.2% in “failed” subgroup with ongoing postoperative seizures.	96.4% overall. 90.4% in “failed” subgroup with ongoing postoperative seizures.
LITT 11	65.9%	55.8%	49.3%	3.9 yrs	NA	81.4% overall. 63.3% in “failed” subgroup with ongoing postoperative seizures.
RNS 44	28%	18.4%	62%	3.2 yrs	75%	73%
DBS a	9%	NA	NA	NA	78%	68%
VNS 43	8%	NA	NA	NA	NA	63%

^a Results for DBS available at 7 years.

Open-Ended Questions for Future Research

Significant and accelerated progress has been made in enriching the surgical landscape for drug-resistant TLE, but many questions still need to be answered. We highlight here the few with the highest long-term significance or most immediate urgency.

• Optimizing surgical targeting with LITT and characterizing its long-term outcomes: Defining LITT ablation strategies (what needs to be ablated and how) with better seizure outcomes that more closely mirror those achieved with the gold standard of ATL will have a critical impact. Several retrospective imaging analyses found associations between seizure freedom and ablation of the amygdala, hippocampal head, and portions of the parahippocampal gyrus, rhinal cortices, and piriform cortex while extending the ablation posteriorly has diminishing returns.^49–51 Such research can drive improvements in LITT planning. For example, a study demonstrated that resecting over 50% of the piriform cortex is associated with a 16-fold improvement in seizure-freedom rate after ATL. ⁵² Yet, in a recent cohort of 39 TLE patients undergoing LITT, only 12.8% of the piriform cortex was treated in the ILAE class 1 patients (compared to 2.7% in the remaining ILAE class 2-6 group), ⁵³ highlighting the challenge of sufficiently ablating the piriform cortex while also targeting the hippocampus using a restricted linear approach. Many centers have since adopted a 2-fiber technique, an approach not yet reflected in most of the published outcomes data. As modifying what is ablated to improve seizure outcomes occurs, research will need to demonstrate that cognitive outcomes will not worsen in parallel.

• Comparative effectiveness research: The treatment landscape is now enriched with various surgical interventions including open resection, LITT, RNS, DBS, and VNS, providing the benefit of multiple options for physicians and patients. However, without robust data on efficacy and safety from head-to-head comparisons, the delivery of surgical care moves further from effective care (what works best and is safe) to supply-driven (what doctors offer based on local availability and hospital investments in services) and preference-driven care (what patients prefer among equal options- for example between VNS, RNS, or DBS). This well-studied shift is known to introduce unwarranted variations in healthcare delivery with consistently demonstrated harms to patient outcomes and public health. ⁵⁴ Recruiting patients for the gold standard of randomized clinical trials comparing surgical interventions head-to-head will be challenging, but well-designed observational research applying statistical tools designed for causal inferencing is possible, as was done to compare sEEG and subdural electrode implantations for intracranial recordings. ⁵⁵

• Defining ideal treatment pathways: Beyond comparing individual surgical therapies, research will also be needed to define the best sequential treatment algorithms. For example, performing the less invasive approach of LITT first with open resection planned in case LITT fails may conceivably lead to different scenarios. In scenario A, more patients are encouraged to get surgical therapy sooner. So, even with lower odds of seizure freedom with LITT for any given patient, there may be a net benefit of more “seizure-free years” to the population at large. Conversely, in scenario B, allowing more patients to spend more time in an interval between an unsuccessful LITT and a subsequent successful open resection may have the net loss of increasing SUDEP. Which scenario is more likely to unfold with such a strategy is now unknown. This is compounded by the impact of laser ablation on subsequent ATL. Significant scarring, adhesive, and gliotic changes following LITT could result in distorted anatomy reducing the success of subsequent open resection. ⁵⁶ On the other hand, 65% (13/20) of those going on to ATL after LITT treatment failure in 1 study achieved seizure freedom with complications similar to those with upfront ATL. ¹¹ Importantly, centers (and neurosurgeons) specializing in epilepsy must be familiar not only with minimally invasive approaches but remain capable of offering open resective treatments, in order to be able to offer surgical intervention without bias. Notably, this has implications for emerging neurosurgical trainees seeking to specialize in epilepsy surgery, in order to preserve effective traditional methods while concomitantly innovating and adapting to novel technologies as they emerge on the horizon.

• Individualizing risk versus benefit assessment: Beyond the macroscale of surgical approach or algorithm comparative effectiveness at the population level, tailoring treatment to the micro-scale of individual patient goals and risk tolerance is key. A patient eager to achieve seizure freedom (eg, disabled by seizures or already retired or at risk of losing employment from unpredictable albeit infrequent seizures) may be comfortable with a higher risk of cognitive decline for more guaranteed and quicker seizure freedom. Another patient (eg, busy professional who relies on high-level cognitive tasks for their livelihood) may be comfortable with risking lower odds of seizure freedom to preserve memory. Blanket judgments and assumptions on outcomes based on population studies are overly simplistic as outcomes vary significantly among patients, regardless of surgical approach. Publicly available prediction models and nomograms now assist clinicians in identifying patients at high risk for poor postoperative outcomes and support decision-making and patient counseling^57–61 when open resection is considered. One nomogram predicts cognitive and seizure outcomes depending on whether a hippocampus that is structurally intact on MRI is resected or preserved, providing some ability—albeit rudimentary—to inform a choice among different strategies. These need to be improved to account for the rich variability of possible anatomical resections. No such tools have been developed yet for LITT. Additionally, there is growing evidence and awareness that neuropsychological assessment is not adequately assessing the complexity of brain function (eg, few tools are available to assess functions such as consolidation of memory/rapid long-term forgetting, multimodal learning, the contribution of emotional valence to memory, naturalistic language), and this needs to be improved for the evaluation of all of these procedures.^62–64

• Impact of minimally invasive procedures on the drug-resistant epilepsy (DRE) surgical treatment gap: A major promise of minimally invasive procedures is their potential to enlarge the surgical patient pool as they would attract those uncomfortable with the risks and morbidity of open resection, thus remedying the surgical DRE treatment gap. The data, however, are not straightforward. From 2010 to 2020, the proportion of patients with DRE receiving open resection dropped from 12% to 6%, while the proportion receiving LITT grew from 0% to slightly below 1% with stable volumes since 2014, and that receiving RNS grew from 0% to 2%, albeit with a trend of increasing use. ⁹ The net result is a 25% shrinkage—rather than growth—in the total surgical pool. Robust health services research is necessary to understand the drivers of this surgical landscape shift, and to define its public health implications.

Conclusion

Clinicians should personalize temporal lobe epilepsy surgery, considering seizure likelihood, cognitive risks, surgical complications, and patient goals. Continued seizure risks must also be weighed. If seizure freedom is likely and cognitive risk is low, open resection should be considered. If cognitive risk is high, consider less invasive methods. Always discuss and consider patient goals in surgical decisions. Centers offering LITT should be prepared to perform subsequent open resections and counsel patients about the possible need for additional surgery. Future research is essential to refine and understand a rich surgical landscape.