AUTHOR=Fu Chengwei , Zhang Yue , Ye Yongsong , Hou Xiaoyan , Wen Zeying , Yan Zhaoxian , Luo Wenting , Feng Menghan , Liu Bo 

TITLE=Predicting response to tVNS in patients with migraine using functional MRI: A voxels-based machine learning analysis

JOURNAL=Frontiers in Neuroscience

VOLUME=Volume 16 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2022.937453

DOI=10.3389/fnins.2022.937453

ISSN=1662-453X

ABSTRACT=Background: Migraine is a common disorder, affecting many patients. However, for one thing, lacking objective biomarkers, misdiagnosis and missed diagnosis happen occasionally. For another, though transcutaneous vagus nerve stimulation (tVNS) could alleviate migraine symptoms, individual difference of tVNS efficacy in migraineurs hamper clinical application of tVNS. Therefore, it’s necessary to identify biomarkers to discriminate migraineurs and select patients suitable for tVNS treatment.
Methods: Seventy patients diagnosed with migraine without aura (MWoA) and 70 matched healthy controls were recruited to complete fMRI scanning. In study 1, fractional amplitude of low-frequency fluctuation (fALFF) was calculated, and the differences between healthy controls and MWoA were compared. Meaningful voxels were extracted as features to construct discriminating model by support vector machine. The performance of discriminating model was assessed by accuracy, sensitivity and specificity. In study 2, 33 of the 70 patients in study 1 receiving real tVNS were included to construct predicting model. Discriminative features of discriminating model in study 1 were used to predict the reduction of attack frequency after a 4-week tVNS treatment by support vector regression. A correlation coefficient between predicting value and actual value of the reduction of migraine attack frequency was conducted in 33 patients to assess the performance of predicting model after tVNS treatment.
Results: A biomarker containing 3650 features was identified with an accuracy of 79.3%, sensitivity of 78.6%, and specificity of 80.0%(p<0.001). The discriminative features were found in the trigeminal cervical complex/rostral ventromedial medulla(TCC/RVM), thalamus, medial prefrontal cortex(mPFC), and superior temporal gyrus/middle temporal gyrus(TSG/TMG). Then, 70 of 3650 discriminative features were identified to predict the reduction of attack frequency after tVNS with a correlation coefficient of 0.36(p=0.03). The 70 predictive features were involved in TCC/RVM, mPFC, TSG/TMG, middle cingulate cortex (MCC), and insula. The reduction of migraine attack frequency had a positive correlation with right TCC/RVM, left MCC, bilateral mPFC and negative with left insula, and right TSG/TMG, respectively. 
Conclusions:  By machine learning, we proposed two potential biomarkers that could discriminate MWoA and predict the efficacy of tVNS in reducing migraine attack frequency. The pivotal features were located in the TCC/RVM, thalamus, mPFC, and TSG/TMG.