LITT vs Surgery mTLE Meta-Analysis
(2026)Objective
To compare neuropsychological outcomes—verbal memory, visual memory, and naming—following traditional open surgical approaches (ATL/SAHE) versus MRgLITT in adults with drug-resistant mesial temporal lobe epilepsy, stratified by surgical laterality.
Study Summary
• Left-sided verbal memory decline showed a nonsignificant trend favoring MRgLITT (29%, 95% CI 9%–62%) over open resection (36%, 95% CI 28%–45%; p = .5967)
• Right-sided visual memory decline was comparable between MRgLITT (19%, 95% CI 3%–61%) and open resection (16%, 95% CI 8%–29%; p = .8027)
• Naming improvement occurred in 27% of patients in each group after right-sided surgery; higher seizure freedom rates showed a borderline association with greater naming decline (p = .0508)
Intervention
Systematic review and meta-analysis comparing open resection (anterior temporal lobectomy [ATL] or selective amygdalohippocampectomy [SAHE]) versus MR-guided laser interstitial thermal therapy (MRgLITT) for drug-resistant mesial temporal lobe epilepsy; 34 studies included, searched through June 15, 2025
Inclusion Criteria
Original peer-reviewed English-language studies; adults ≥16 years with drug-resistant mTLE; underwent open resection or MRgLITT; reported proportions with cognitive decline/improvement in ≥1 domain stratified by laterality using validated change methods (RCI, SRB, or ≥1 SD threshold); minimum 6-month postoperative follow-up
Study Design
Arms: Open resection—ATL or SAHE (27 studies) vs MRgLITT (10 studies); 3 studies contributed data to both arms
Patients per Arm: Not reported in available source text
Outcome
• Left-sided verbal memory and right-sided visual memory decline were statistically equivalent across approaches (p = .5967 and p = .8027, respectively)
• A borderline seizure-freedom–naming trade-off was identified (p = .0508), but MRgLITT preserved naming independent of seizure outcomes
Bottom Line
MRgLITT offers a significant and clinically meaningful advantage over open resection for preserving naming after left-sided mTLE surgery, independent of seizure outcomes; verbal and visual memory outcomes are comparable across approaches, and clinicians should weigh the naming benefit against MRgLITT's modestly lower seizure freedom rates when counseling patients.
Major Points
- MRgLITT significantly preserved naming after left-sided surgery compared to open resection (9% vs 43% decline; p < .0001), with the advantage persisting after meta-regression adjustment for seizure freedom rates (β = −1.36, 95% CI −2.20 to −0.51; p = .0016)
- Left-sided verbal memory decline showed a nonsignificant trend favoring MRgLITT (29% vs 36%; p = .5967), consistent with better but not statistically superior preservation
- Right-sided visual memory decline was statistically equivalent between MRgLITT (19%) and open resection (16%; p = .8027)
- Naming improvement after right-sided surgery occurred in 27% of patients in both groups, reflecting substantial cognitive recovery when the dominant hemisphere is preserved regardless of approach
- An exploratory meta-regression identified a borderline association between higher seizure freedom and greater naming decline (p = .0508), suggesting a potential efficacy–language trade-off that warrants individualized counseling
- MRgLITT's naming benefit is independent of seizure outcomes, supporting its use as the preferred approach when language preservation is a priority in left-sided mTLE
Study Design
- Study Type
- Systematic Review and Meta-Analysis
- Randomization
- No
- Sample Size
- 34
- Follow-up
- Minimum 6 months postoperatively; 12-month data prioritized where available
Primary Outcome
Definition: Proportion of patients with cognitive decline or improvement in verbal memory (left-sided surgery), visual memory (right-sided surgery), and naming (left-sided surgery), compared between open resection and MRgLITT
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| Open resection — naming decline (left): 43% (95% CI 27%–61%); verbal memory decline (left): 36% (95% CI 28%–45%); visual memory decline (right): 16% (95% CI 8%–29%) | MRgLITT — naming decline (left): 9% (95% CI 3%–22%); verbal memory decline (left): 29% (95% CI 9%–62%); visual memory decline (right): 19% (95% CI 3%–61%) | - (Naming decline MRgLITT 3%–22% vs open resection 27%–61%; verbal memory decline MRgLITT 9%–62% vs open resection 28%–45%; visual memory decline MRgLITT 3%–61% vs open resection 8%–29%; meta-regression β 95% CI: −2.20 to −0.51) | Naming (left): p < .0001; verbal memory (left): p = .5967; visual memory (right): p = .8027; meta-regression (naming, adjusted): p = .0016 |
Limitations & Criticisms
- Wide confidence intervals for MRgLITT outcomes (visual memory 3%–61%; verbal memory 9%–62%) reflect the small number of MRgLITT studies (n=10) and limited sample sizes, limiting precision
- High between-study heterogeneity likely present given variability in surgical extent, cognitive test batteries, and definitions of 'decline' across included studies
- Retrospective, observational design of all included studies precludes causal inference and introduces potential selection bias
- Patients selected for MRgLITT may systematically differ from open resection candidates in unmeasured ways (e.g., lesion volume, comorbidities), confounding comparative estimates
- Seizure freedom data were not uniformly available across studies, limiting the power of the seizure–cognition meta-regression
- Stereoelectroencephalographic confirmation data were rarely reported, precluding planned subgroup analyses on this variable
- Cognitive assessment instruments varied across studies; pooling proportions across heterogeneous tools may obscure domain-specific differences
Citation
Stavrogianni K, et al. Neuropsychological outcomes comparing traditional surgical approaches and laser interstitial thermal therapy for refractory mesial temporal lobe epilepsy: A systematic review and meta-analysis. Epilepsia. 2026;67(2):588-605. doi:10.1111/epi.18687