A Resting-State FMRI Study On Drug Mechanisms Of Herpes Zoster Neuralgia

Due to a reactivation of varicella zoster virus and its damage towards peripheral nerves,herpes zoster neuralgia(HZN)is characterized by persistent pain,hyperalgesia and allodynia,thus seriously influencing patients' quality of life.Pharmacotherapy with antidepressants and anticonvulsants is the most common pain management protocol for HZN.Most HZN patients are the elders with poor immune function and often suffer from multiple side effects of drugs.While pharmacotherapy fails,risks of getting intractable postherpetic neuralgia(PHN)increase dramatically.With functional magnetic resonance imaging(fMRI),previous studies have demonstrated that HZN/PHN is associated with abnormal activity in central pain modulation system.However,the pathogenesis and analgesic mechanisms in HZN/PHN still remain largely unknown.The present work aimed at exploring brain functional signatures for HZN treatment prognosis and the relevant mechanisms underlying drug treatment,in terms of functional segregation and integration.Support vector machine(SVM),a classifier of machine learning method,was employed in fMRI data analysis to help identify these functional neurologic signatures.Twenty-night right-handed HZN patients were acquired resting-state fMRI scans before treatment,including 14 with medication-resistant pain(aged 59.6±6.6)and 15 with medication-sensitive pain(aged 62.3±7.3).In study 1,functional segregation analysis with regional homogeneity(ReHo)and fractional amplitude of low frequency fluctuation(fALFF)measurements was conducted.ReHo and fALFF features were then input into two SVM classifiers to predict treatment outcome.The brain regions with predictive ReHo/fALFF features were considered significant regional neurologic signatures for prognosis.In study 2,functional integration analysis with seed-based functional connectivity(SB-FC),a hypothesis-driven method,was performed.We firstly derived seeds form significant predictive brain regions in study 1.SB-FC features were then obtained and used to construct SVM for HZN classification.Those features contributing to prediction were deemed SB-FC signatures of HZN.In study 3,with independent component analysis(ICA),another method of functional integration analysis,we extracted 5 resting-state networks(RSNs)to calculate RSNs-FC features and trained an SVM for prediction.Also,those significant features represented RSNs-FC signatures of HZN.The current analysis found:(1)ReHo and fALFF features for SVM yielded classification accuracies of 96.5% and 93.1%,respectively.Regional neurologic signatures located in pain descending inhibitory system(eg,cerebellum,parahippocampus,thalamus,postcentral gyrus)and advanced cognitive areas(eg,frontal lobe).(2)Using SB-FC features from each seed respectively,SVM could successfully classify patients with poor response from those with good response,with an accuracy between 75.9~96.6%.The SB-FC signatures were mainly involved in medial pain system with hippocampus connecting to cerebellum,orbito-frontal cortex,limbic lobe,and in lateral pain system displaying robust connectivity between primary somatosensory cortex and cerebellum,frontal lobe,cingulate gyrus and supplementary motor areas.(3)Predictive RSNs-FC signatures were those FC patterns between default mode network,left executive network,motor network,basal ganglia network,as well as between motor network and salience network.The anatomic basis of RSNs-FC signatures was largely consistent with that of SB-FC signatures.Additionally,default mode network may be an important RSN in pain modulation of HZN.Taken together,functional neurologic signatures regarding drug treatment prognosis of HZN comprised brain regions in pain descending inhibitory system and cognitive areas.Common central acting agents for HZN may restore abnormal function of networks centering in hippocampus or frontal lobe,or regulate RSNs-FC between default mode network and other networks to produce antinociceptive and analgesia effect.These findings may provide important neuroimaging information to promote the efficiency of HZN diagnosis and treatment.