Neuroimaging Features-Based Prediction Of Headache Attack Frequency In Migraine Patients

Migraine is a chronic paroxysmal neurological disorder characterized by recurrent attacks of moderate or severe throbbing headache in the clinical setting.Patients often accompany with other physiological and emotional maladaptive responses such as nausea,emesis and mood disorder during the migraine attacks,and the headaches can get worse as stresses increase.It is the most concerned problem for patients and their clinicians that how to effectively evaluate and prevent the attack and progression of migraine and relieve neurological and painful conditions caused by migraine.Since attack frequency is a risk factor for progression to chronic migraine,preventive medications should be given when migraine attacks are frequent.However,the migraine attack frequency is often ascertained from patient self-report which may causes the measurement of attack frequency to become unreliable or inaccurate.Thus,the current study aims at identifying migraine attack frequency-related neuroimaging features to objectively evaluate patients' headache frequency.Moreover,the frequency of headache attacks is also an important indicator of the effectiveness of migraine placebo intervention.A systematic review pointed that placebo treatment could effectively reduce migraineur's headache attack frequency,and the responder rate for sham acupuncture was 38%.We designed a single-blinded placebo sham acupuncture treatment trial to evaluate whether the diffusion metrics of white matter tract of interest could predict the 8-week placebo treatment outcomes in patients with migraine without aura.A guideline suggested value of headache frequency was used to divide the migraineurs into the low(MOl)and high(MOh)attack frequency groups.Based on resting-state functional magnetic resonance imaging data,we computed the whole-brain functional connectivity matrix for each participant.Whole-brain functional connectivity was used to build multivariate logistic regression models with model iteration optimization to identify MOl and MOh.We found that the discriminative features were mainly located within the limbic lobe,frontal lobe,and temporal lobe.The best model accurately discriminated MOh from MOl with AUC of 0.91(95%CI [0.86,0.95])in the training/test cohort and 0.79 in the validating cohort.Permutation tests analysis demonstrated that the classification performance of these features was significantly better than chance,but counterpart random features failed to classify MOl and MOh.The current findings suggested that functional connectivity has potential value in assessing the frequency of migraine attacks.After the single-blinded placebo sham acupuncture treatment trial,patients were subdivided into the placebo effective(>50% improvement in migraine attack frequency)and ineffective(<50% improvement in migraine attack frequency)groups.The white matter tract microstructure of the medial prefrontal cortex-amygdala circuit was derived from base diffusion tensor imaging data for each patient.By using the multivariate pattern analysis approach,these features were used to predict the placebo treatment responses.We found that the combined features of diffusion measures from vertices along the pathways of the external capsule and anterior cingulate cortex/ medial prefrontal cortex accurately discriminated effective group from ineffective group with an accuracy of 84.0%.Additionally,we did not find any significant correlation between the attack frequency and the diffusion properties along the fiber pathways of the medial prefrontal cortex-amygdala,which may indicate that there was an indirect relationship between the diffusion properties of the fiber bundles of interest and the migraine attack frequency.