Disrupted default mode network dynamics in recuperative patients of herpes zoster pain

Abstract Introduction Previous studies of herpes zoster (HZ) have focused on acute patient manifestations and the most common sequela, postherpetic neuralgia (PHN), both serving to disrupt brain dynamics. Although the majority of such patients gradually recover, without lingering severe pain, little is known about life situations of those who recuperate or the brain dynamics. Our goal was to determine whether default mode network (DMN) dynamics of the recuperative population normalize to the level of healthy individuals. Methods For this purpose, we conducted resting‐state functional magnetic resonance imaging (fMRI) studies in 30 patients recuperating from HZ (RHZ group) and 30 healthy controls (HC group). Independent component analysis (ICA) was initially undertaken in both groups to extract DMN components. DMN spatial maps and within‐DMN functional connectivity were then compared by group and then correlated with clinical variables. Results Relative to controls, DMN spatial maps of recuperating patients showed higher connectivity in middle frontal gyrus (MFG), right/left medial temporal regions of cortex (RMTC/LMTC), right parietal lobe, and parahippocampal gyrus. The RHZ (vs HC) group also demonstrated significant augmentation of within‐DMN connectivity, including that of LMTC‐MFG and LMTC‐posterior cingulate cortex (PCC). Furthermore, the intensity of LMTC‐MFG connectivity correlated significantly with scoring of pain‐induced emotions and life quality. Conclusion Findings of this preliminary study indicate that a disrupted dissociative pattern of DMN persists in patients recuperating from HZ, relative to healthy controls. We have thus provisionally established the brain mechanisms accounting for major outcomes of HZ, offering heuristic cues for future research on HZ transition states.


| INTRODUC TI ON
Herpes zoster (HZ), commonly known as shingles, represents a recrudescence of latent varicella-zoster virus (VZV) that inflames and injures spinal or cranial sensory ganglia. Typically, a unilateral erythematous rash develops in affected skin, accompanied by various pain sensations, such as burning, stabbing, soreness, bloating, or allodynia. 1,2 One epidemiologic study has determined that ~12.5% of older patients (≥50 years) with HZ may experience postherpetic neuralgia (PHN) 3 months after onset of zoster, and the proportion rises with advancing age. 3 As the most common and intractable sequela of HZ, PHN has received the most attention to date, given its effects on patient quality of life and the considerable social burden inflicted. 4,5 To understand its mechanism for precision treatment, a number of neuroimaging changes have been identified in patients with acute HZ or PHN, compared with healthy individuals, particularly exhibiting abnormal function in classic sensory-discriminative and emotional-linked areas (ie, thalamus, insula, frontal lobe, brainstem, temporal lobe, and limbic lobe). [6][7][8][9][10] However, after comparable episodes of acute shingles, the majority of patients gradually recover, and the pain attenuates. Lack of prescribed treatments or clinical follow-up in this recuperative population has hampered investigations, creating a scarcity of data on their life situations and brain dynamics. Procuring such data is indispensable in deciphering HZ transition mechanism. The present preliminary study was conducted to ascertain whether brain dynamics in patients recuperating from HZ normalize to the level of healthy individuals.
In previous efforts, widely distributed abnormalities of brain activity in acute and chronic pain have reflected multidimensional dysfunction of pain sensation, pain-related emotion, and cognition. 11 An isolated, predefined brain area is simply incapable of such complexity.
A better understanding of pain-related neurobiology would thus require investigations of brain networks. Resting-state networks (RSNs) have proven important in high-order brain functions and neuropsychiatric disorders, [12][13][14] incorporating brain regions with coherent neuronal activity of low-frequency blood oxygenation level-dependent (BOLD) signal fluctuations. 15 Among the various known RSNs, the default mode network (DMN) is seemingly the most widely studied and well characterized, supporting internal mentation (such as memory, prospection) and acting as a sentinel to monitor the external environment. [16][17][18][19][20] Interest has been considerable in terms of correlating the DMN with pain intensity, negative mood, or pain rumination. 21 29 In contrast to the PHN, our recuperating patients were diagnosed by a clinician, using the following criteria: (a) self-appraised pain intensity score <4/10 by visual analog scale (VAS: 0 [no pain] to 10 [worst pain imaginable]); (b) ≥3-month duration after onset of acute shingles; and (c) no medical treatment rendered for HZ. Exclusion criteria for both RHZ and HC groups were the following: (a) special HZ (of ear, eye, or viscera or asymptomatic); (b) history of ongoing acute or chronic pain attributable to headaches, toothaches, arthritis, or cervical/lumbar spondylopathy; (c) psychiatric or neurological disorders (eg, epilepsy or head injury); (d) any severe major health condition, such as cardiovascular disease or renal insufficiency; and (e) contraindications to MRI.

| Assessment of pain and emotional parameters
All questionnaires were completed 1 hour prior to brain scans. Each

| Acquisition of fMRI data
All MRI scans were performed using a 3.0-Tesla MRI scanner (GE Discovery 750; GE Healthcare) equipped with an 8-channel head coil. During these procedures, each patient was positioned supine, the head firmly restrained by foam pads; and earplugs were provided to reduce noise during scanning. They were asked to remain still as long as possible, keeping eyes closed but staying awake.
High-resolution structural T1-weighted images were acquired using a fast spoiled gradient recalled echo pulse sequence as follows: rep-

| Resting-state fMRI (rs-fMRI) preprocessing
In preprocessing of fMRI time series volume data, the Resting-State fMRI Data Analysis Toolkit (REST, V1.8; http://www.restf mri.net) and Statistical Parametric Mapping (SPM12; http://www.fil.ion.ucl.ac.uk/ spm) were accessed, using MATLAB platform (MathWorks). The first 10 volumes of each functional time series were discarded to avoid transient signal changes before magnetic field steady states were reached, allowing subjects to acclimate in this scanning environment.
We then corrected the other images for timing differences (slice 37 used as reference) and head motion, determining translation (mm) and rotation (degrees) by six parameters (three each, translation and rotation). There were no group-wise exclusions due to head motion beyond 2 mm of displacement or 2° of rotation. Subsequently, we spatially normalized images to the Montreal Neurological Institute (MNI) space using EPI templates with resampling voxel sizes of 3 × 3 × 3 mm. All images generated were spatially smoothed using a Gaussian kernel of 6 × 6 × 6 mm, full width at half maximum (FWHM).

| Identification of DMN via Independent Component Analysis (ICA)
Independent component analysis is a powerful tool based on datadriven blind source separation that captures essential components of multivariate rs-fMRI signals, extracting independent sources from mixed sources. 30,31 Concatenated preprocessed rs-fMRI data of each group were subjected to ICA, utilizing Group ICA of fMRI Toolbox (GIFT) software (v4.0; TReNDS, https://trend scent er.org/ softw are/). Independent components (ICs) were estimated at 37 (RHZ group) and 41 (HC group) by minimum description length (MDL) criteria, identified via Infomax algorithm. [32][33][34] For each IC, the time courses mirrored waveforms of specific coherent brain activity patterns, and pattern intensities across voxels were expressed in corresponding spatial maps. To display voxels appropriate for individual ICs, intensities of each map were converted to z-values. 35 The DMN component of each group was extracted successfully, adopting a higher signal-to-noise ratio than that traditionally implemented. [36][37][38] Subject-specific spatial maps and time courses were then configured by temporospatial multiple regression back-reconstruction approach in GIFT.

| Second-level analysis of the DMN
One-sample t test was invoked for DMN spatial maps to evaluate within-group data integrity (P < .05, false discovery rate [FDR] criterion corrected), powered by Statistical Parametric Mapping (SPM12).
To further restrict DMN comparisons between groups, the DMN spatial maps of both groups were combined, creating a DMN mask.
Two-sample t test was engaged to determine between-group differences in DMN spatial maps, with significance set at P < .05 (corrected by FDR and cluster size >43).

| Seed-based within-DMN functional connectivity (FC) analysis
To examine the within-DMN functional connectome, regions of interests (ROI) were extracted from significantly discrepant DMN spatial maps of RHZ and HC groups, based on two-sample testing of above-mentioned areas (including MFG, right/left MTC, right parietal lobe, right/left parahippocampal gyrus, and inferior frontal cortex [IFC]). We then identified one of these seven regions as ROI, performing voxel-wise within-DMN FC analysis by using the combined DMN spatial map as a mask. Correlation coefficients were ultimately converted (Fisher's z-transformation) to a normal distribution. A total of seven voxel-wise within-DMN connectivity analyses were completed.

| Statistical analysis
Demographic and clinical variables of RHZ and HC groups were assessed, expressing continuous variables as mean ± SD and testing for normality by Shapiro-Wilk test. Intergroup differences in variables with no evidence against normality were subjected to independent two-sample t test, applying chi-squared or Fisher's exact test to categorical variables. All computations were driven by standard software (SPSS v24.0; IBM Corp), setting significance at P < .05.
Intergroup FC differences were assessed by two-sided unpaired t test. To exclude possible confounding effects, age, sex, and education level were included as covariates in two-sample testing.

| Demographics and clinical characteristics
A total of 60 participants (RHZ group, 30; HC group, 30) were selected for this study. As shown in Table 1, there were no statistically significant differences between groups in terms of gender

| Group differences in DMN
In comparing DMN spatial maps by group, stronger connectivity within the DMN was shown for RHZ (vs HC) group, including MFG, right/left MTC, right parietal lobe, right/left parahippocampus, and IFC areas (P < .05, FDR corrected) (Figure 2A, Table 2).
Representative multislices of discrepant DMN spatial map between two groups are provided in Figure 2B.

| D ISCUSS I ON
In this preliminary study, the following were major findings: (a) con- It is important to note that our study of the DMN was prompted by the existing body of knowledge on its role in a wide range of sensory and cognitive processing functions, including pain-related rumination, attention, and memory. 16,18,39,40 Disruption of the DMN has been identified in many mental and psychological conditions, including Alzheimer's disease, traumatic brain injury, epilepsy, autism, and major depressive disorders. 41 The three aforementioned DMN regions (LMTC, MFG, and PCC) marked by abnormal connectivity are key nodes in pain regulation.
Temporal cortex is one area of the brain involved in pain perception and modulation, as confirmed in human neuropathic states and in animal models. 46  On the other hand, the chronification of pain may be promoted through corticostriatal projections, perhaps relying on dopamine receptor activation (or lack thereof) in the ventral tegmental area-nucleus accumbens (VTA-NAc) reward pathway. [54][55][56] There is also neuroimaging evidence that patients with chronic pain display enhanced connectivity between mPFC and other DMN regions, such as PCC/PCu and retrosplenial cortex, linked to pain catastrophizing and rumination (ie, a form of thought characterized by repetitive attention to discomforting pain stimuli and negative emotions). 57,58 Those afflicted with chronic pain also bear abnormal interactions between DMN and the descending modulatory system, viewed as an underlying mechanism of rumination on chronic pain. 22  along with rumination-related within-DMN connectivity alterations, and the changes were not short term in nature ( Figure 5). We speculated these connectivity abnormalities were remnants of acute HZ.
A recent study has further shown that patients with chronic pain due to spondyloarthritis exhibit complex relations between pain, resilience, and mPFC seed-based within-DMN connectivity. Resilience is a positive psychological factor, enabling rebound in response to adverse events, like pain. 65,66 In the setting of chronic pain, resilience is linked to pain catastrophizing, adjustment, and acceptance through As a pivotal subregion of the mPFC, ventral medial prefrontal cortex (vmPFC) is often considered a sensory-visceromotor aggregator associated with mood regulation, motivation, and social behavior. 69 Neuroimaging studies indicate that individual emotional states directly affect vmPFC activity. 19 The diminished vmPFC activity negatively correlates with anxiety self-ratings, reflecting a dynamic balance between attention and certain anxiety states. 70,71 In our patients, there was a trend toward positive correlation between mPFC-related within-DMN connectivity and anxiety scores in the RHZ group, falling short of statistical significance. Nevertheless, this raises the possibility that increased mPFC-related connectivity in patients recuperating from HZ may signal disruptions in anxiety states and individual brain homeostasis ( Figure 5).
Posterior cingulate cortex, another key node of DMN, is activated during attention regulation and is extensively connected to medial temporal lobe memory systems. 72 with chronic back pain, complex regional pain syndrome, osteoarthritis, and tonic pain stimuli; and the intensity of PCC connectivity seems to increase in those with temporomandibular disorders. 57,74 Thus, the increased LMTC-PCC connectivity observed in our RHZ group members may underlie an altered ability for some particular tasks, such as attention and memory ( Figure 5). Whether they also may develop cognitive impairment (as in patients with PHN) is the subject of further investigation. 75 Finally, recent studies have contended that anterior (mPFC) and posterior (PCC) DMN subnetworks achieve a dynamic balance vital for maintenance of cognitive function. 76 In our study, the aberrant anterior and posterior within-DMN connectivity shown by the RHZ group seems to link such zonal disequilibrium with pain-induced rumination and emotional regulation after extinction of acute protracted pain. This finding enhances our understanding of reciprocity between subnetworks responsible for sensory, emotional, and cognitive dimensions of pain ( Figure 5).
The current preliminary study has limitations that may be reme- also be considered and are presently being explored. Finally, we did not measure behavioral changes through cognitive or attention-demanding tasks or questionnaires to prove a correlation between intensity of PCC-related connectivity and cognitive performance.
In conclusion, the present efforts are the first to characterize disruption of DMN patterns and within-DMN connectivity in patients recuperating from HZ, despite resolution of major symptoms (ie, rash and pain). We have also determined a relation between within-DMN connectivity and pain-induced negative emotional states, attributing residual pain rumination, emotional alterations, and resilience to F I G U R E 5 Schematic diagram of regions with aberrant FC within-DMN and its related function in RHZ patients. The MTC may be related to pain perception, modulation, and HZ-PHN chronification. The alternative LMTC FC within DMN in RHZ patients can be accounted for by the alteration of brain activity from acute herpes zoster to the recuperative period. The mPFC may be related to pain rumination and pain-induced negative emotion and its dynamic alteration within-DMN may not disappear in the short term after acute pain period. The PCC may be related to cognitive impairment, but whether RHZ patients show cognitive impairment, as seen in PHN patients, warrants further investigation. In summary, RHZ patients showed disequilibrium in anterior and posterior subnetworks after acute herpes zoster pain. HCs, healthy controls; MFG, middle frontal gyrus; MTC, medial temporal cortex; PCC, posterior cingulate cortex; RHZ, recuperative patients of herpes zoster DMN divergence. Our preliminary results may help clarify the brain dynamics during recovery that likely persist well beyond the acute period, providing heuristic cues for additional research on neuromechanisms of HZ transition.

CO N FLI C T S O F I NTE R E S T S
The authors declare no conflict of interest.