Magnetic Resonance-guided Laser Interstitial Thermal Therapy for Drug-Resistant Epilepsy: A Systematic Review and Individual Participant Data Meta-Analysis.

Magnetic resonance-guided laser interstitial thermal therapy (MRgLITT) has emerged as a popular minimally invasive alternative to open resective surgery for drug-resistant epilepsy (DRE). We sought to perform a systematic review and individual participant data meta-analysis (IPDMA) to identify independent predictors of seizure outcome and complications following MRgLITT for DRE. Eleven databases were searched from January 1 st , 2010 to February 6 th , 2021 using the terms: "MR-guided Ablation Therapy" and "epilepsy". Multivariable mixed-effects Cox and logistic regression identified predictors of time-to-seizure recurrence, seizure freedom, operative complications, and postoperative neurological deficits. From 8,705 citations, 46 studies reporting on 450 MRgLITT DRE patients (mean age 29.5±18.1 years, 49.6% female) were included. Median postoperative seizure freedom and follow-up duration were 15.5 and 19.0 months, respectively. Overall, 240 (57.8%) of 415 patients (excluding palliative corpus callosotomy) were seizure free at last follow-up. Generalized seizure semiology (HR=1.78, p=0.020) and non-lesional MRI findings (HR=1.50, p=0.032) independently predicted faster time-to-seizure recurrence. Cerebral cavernous malformation (CCM) (OR=7.97, p<0.001) and mesial temporal sclerosis/atrophy (MTS/A) (OR=2.21, p=0.011) were independently associated with greater odds of seizure freedom at last follow-up. Operative complications occurred in 28 (8.5%) of 330 patients and were independently associated with extratemporal ablations (OR=5.40, p=0.012) and non-lesional MRI studies (OR=3.25, p=0.017). Postoperative neurological deficits were observed in 53 (15.1%) of 352 patients and were independently predicted by hypothalamic hamartoma (HH) (OR=5.92, p=0.006) etiology and invasive EEG monitoring (OR=4.83, p=0.003). Overall, MRgLITT is particularly effective in treating patients with well-circumscribed lesional DRE, such as CCM and MTS/A, but less effective in non-lesional cases or lesional cases with a more diffuse epileptogenic network associated with generalized seizures. This study identifies independent predictors of seizure freedom and complications following MRgLITT that may help further guide patient selection for MRgLITT.