Multi-metric predictors of radiofrequency-treated trigeminal neuralgias

Evaluation of neurovascular compression-related trigeminal neuralgia (NVC-TN) and its resolution through microvascular decompression are demonstrable by MRI and intraoperatively [Leal et al. (Atrophic changes in the trigeminal nerves of patients with trigeminal neuralgia due to neurovascular compression and their association with the severity of compression and clinical outcomes: Clinical article. J Neurosurg. 2014;120(6):1484-1495)]. Non-NVC-TNs treated by radiofrequency (RF) lack such detectable features. Multimodal integration of pre-surgical diffusion tensor imaging (DTI) and volumetry (VOL) with intraoperative neurophysiology (ION) could improve understanding and performance of RF among non-NVC-TN. We hypothesized that DTI disturbances' localization (central relay versus peripherally) rather than their values bares the most significant predictive value upon outcome and that ION could quantitatively both localize and assist RF of affected branches. The first pre-surgical step evaluated the differences between affected and non-affected sides (by DTI and VOL). Four TN's segments were studied, from peripheral to central relay: Meckel's cave-trigeminal ganglion (MC-TGN), cisternal portion, root entry zone (REZ) and spinal tract [Lin et al. (Flatness of the Meckel cave may cause primary trigeminal neuralgia: A radiomics-based study. J Headache Pain. 2021;22(1):104)]. In the second intraoperative step, we used both ION and patient's testimonies to confirm the localization of the affected branch, evolving hypoesthesia, pain reduction and monitoring of adverse effects [Sindou (Neurophysiological navigation in the trigeminal nerve: Use of masticatory responses and facial motor responses evoked by electrical stimulation of the trigeminal rootlets for RF-thermorhizotomy guidance. Stereotact Funct Neurosurg. 1999;73(1-4):117-121); Sindou and Tatli (Traitement de la névralgie trigéminale par thermorhizotomie. Neurochirurgie. 2009;55(2):203-210)]. Last and postoperatively, each data set's features and correlation with short-term (3 months) and long-term outcomes (23.5 ± 6.7 months) were independently analysed and blind to each other. Finally, we designed a multimodal predictive model. Sixteen non-NVC-TN patients (mean 53.6 ± SD years old) with mean duration of 6.56 ± 4.1 years (75% right TN; 43.8% V3) were included. After 23.5 ± 6.7 months, 14/16 were good responders. Age, gender, TN duration and side/branch did not correlate with outcomes. Affected sides showed significant DTI disturbances in both peripheral (MC-TGNs) and central-relay (REZ) segments. However, worse outcome correlated only with REZ-located DTI disturbances (P = 0.04; r = 0.53). Concerning volumetry, affected MC-TGNs were abnormally flatter: lower volumes and surface area correlated with worse outcomes (both P = 0.033; r = 0.55 and 0.77, respectively). Intraoperatively, ION could not differ the affected from non-affected branch. However, the magnitude of ION's amplitude reduction (ION-Δ-Amplitude) had the most significant correlation with outcomes (r = 0.86; P < 0.00006). It was higher among responders [68.4% (50-82%)], and a <40% reduction characterized non-responders [36.7% (0-40%)]. Multiple regression showed that ION-Δ-Amplitude, centrally located only REZ DTI integrity and MC-TGN flatness explain 82.2% of the variance of post-RF visual analogue score. Integration of pre-surgical DTI-VOL with ION-Δ-Amplitude suggests a multi-metric predictive model of post-RF outcome in non-NVC-TN. In multiple regression, central-relay REZ DTI disturbances and insufficiently reduced excitability (<40%) predicted worse outcome. Quantitative fine-tuned ION tools should be sought for peri-operative evaluation of the affected branches.