MRI acquisition

MRI scanning was performed on a 3.0-Tesla Siemens Trio TIM Scanner (Siemens Medical Solutions, Erlangen, Germany) at the Radiology Department, the First Affiliated Hospital of Nanchang University. Each participant underwent a conventional T1-weighted and T2-weighted MRI scan, a resting-state functional magnetic resonance imaging (rs-fMRI) scan and a three-dimensional, high-resolution T1-weighted structural MRI scan. First, conventional T1-weighted imaging (repetition time [TR]=250 ms, TE=2.46 ms, slices = 19, slice thickness = 5 mm, gap = 1.5 mm, and field of view [FOV]=220×220 mm) and T2-weighted imaging (TR=4,000 ms, TE=113 ms, slices = 19, slice thickness = 5 mm, gap = 1.5 mm, and FOV=220×220 mm) were performed. Then, rs-fMRI data were obtained using an echo planar imaging sequence (TR=2,000 ms, TE=30 ms, flip angle = 90°, FOV=220×220 mm, matrix = 64×64, 4 mm slice thickness, 30 interleaved axial slices, 240 acquisitions). The high-resolution anatomic three-dimensional T1 sequence had the following parameters: TR=1900 ms, TE=2.26 ms, flip angle=9°, FOV=215×230 mm, matrix = 240×256, slice thickness = 1.0 mm, 176 sagittal slices. For the resting-state scan, all participants were instructed to keep their eyes closed, clear their minds, and remain awake.

MRI data preprocessing

Conventional T1-weighted imaging and T2-weighted imaging were performed to rule out structural brain lesions before pre-processing. rs-fMRI images and structural images were preprocessed using a toolbox for Data Processing and Analysis of Brain Imaging²¹ (DPABI) (http://rfmri.org/dpabi) running on Matlab 7.14.0 (MathWorks, Natick, MA, USA). Preprocessing comprised the following steps: 1) rejection of the first ten functional volumes for stabilization; 2) slice-timing correction; 3) head motion correction; 4) linear transformation: coregistration of the high-resolution structural images to the mean functional image; 5) spatial normalization to the Montreal Neurological Institute template and resampling to 3×3×3 mm voxels; 6) linear detrending and nuisance signal removal, including white matter, cerebrospinal fluid, Friston 24-parameter, and global signal as covariates. In this study, four participants were excluded according to the head motion criteria, which included a maximum spin (x, y, z) of <3.0° and a maximum cardinal direction displacement (x, y, z) of <3.0 mm; the group differences in head motion between PHN patients (n=18) and HCs (n=18) were evaluated according to framewise displacement (FD) criteria based on the method of van Dijk et al.²²

ALFF analysis

ALFF was calculated using DPABI (http://rfmri.org/dpabi). First, fast Fourier transformation was used to convert the time series to the frequency domain for each voxel. Second, the square root of the power spectrum was calculated. Third, the square root was averaged across a predefined frequency interval, and ALFF was calculated as the averaged square root in a voxel. Finally, the ALFF of each voxel was divided by the global mean value to reduce the global effects of variability across participants.²³

ALFF is an effective method for estimating the absolute strength or intensity of low-frequency fluctuations. To investigate PHN-related brain oscillatory activity, we calculated the ALFF for the typical frequency band (0.01–0.08 Hz), the slow-4 band (0.027–0.073 Hz), and the slow-5 band (0.01–0.027 Hz).

Statistical analysis

First, we used Student’s t-tests to analyze the effects of head motion in ALFF analysis of rs-fMRI, and mean FD was applied as a covariate in the group comparisons of the ALFF.²⁴

Then, two-sample t-tests were performed to examine differences between the groups in the typical frequency band (0.01–0.08 Hz). Next, to analyze the main effects of group, frequency band, and their interaction, a two-way ANOVA with factors of condition (PHN, HCs) and frequency band (slow-4, slow-5) was performed. Finally, post hoc tests were performed to determine the regional group differences in ALFF. Age, gender, and FD values were used as covariates, and the significance threshold correction was based on the Gaussian random field theory with a voxel-level P<0.025 and cluster-level P<0.05.

ALFF analysis of the typical frequency band (0.01–0.08 Hz)

The ALFF results revealed significantly higher ALFF values in the typical frequency band (0.01–0.08 Hz) than the global average in the PHN and HC groups, as shown in Figure 1. Next, we compared these ALFF patterns between the two groups and found a significant decrease in ALFF values in the right prefrontal cortex (PFC, Brodmann area 10/46) in PHN patients. In contrast, ALFF values were significantly increased in the bilateral brain stem/cerebellum anterior lobe (BS/CAL) in the PHN group (Figure 1; Table 2).

Open in a separate window

Figure 1

In the typical frequency band (0.01–0.08 Hz), regions with significant ALFF values in the (A) PHN and (B) HC groups are shown separately (false discovery rate correction, P<0.025), as well as their (C) between-group differences (Gaussian random field correction, voxel-level P<0.025 and cluster-level P<0.05).

Abbreviations: ALFF, amplitude of low-frequency fluctuation; HCs, healthy controls; L, left; LH, left hemicerebrum; PHN, postherpetic neuralgia; R, right; RH, right hemicerebrum.

ALFF changes in different frequency bands

The main effects from the two-way repeated-measures ANOVA are shown in Figure 2A and Table 3. Brain regions with a main effect of group included significantly increased ALFF (PHN > HCs) regions in the BS/CAL, the left PCUN/ superior middle temporal gyrus (sMTG), and the right PCUN and significantly decreased ALFF (PHN < HCs) regions in the right cerebellum posterior lobe (CPL), the right PFC, the left PFC, and the right supplementary motor area/M1 (SMA/M1).

Open in a separate window

Figure 2

Frequency-associated ALFF analyses between the PHN and HC groups.

Notes: (A) Main effects of group, (B) main effects of frequency band, and (C) their interactions, with Gaussian random field theory correction, voxel-level P<0.025 and cluster-level P<0.05.

Abbreviations: ALFF, amplitude of low-frequency fluctuation; HCs, healthy controls; L, left; LH, left hemicerebrum; PHN, postherpetic neuralgia; R, right; RH, right hemicerebrum.

Table 3

Frequency-associated ALFF differences, with Gaussian random field theory correction, voxel-level P<0.025 and cluster-level P<0.05

Brain area	BA	Peak (MNI)	Peak	Voxel size
		x, y, z
Main effects of group
Bilateral BS/CAL		−6, 24,−33	5.744	1,986
Left PCUN/sMTG	39, 21, 19, 22	−15, 60, 27	5.350	1,331
Right PCUN	40, 7	18, 63, 42	4.781	1,081
Right PFC	10, 9	42, 33, 33	−4.840	1,075
Right CPL		3, 93, 9	−4.809	727
Right SMA/M1	6, 4	−27, 9, 69	−4.455	402
Left PFC	10, 9	−27, 57, 15	−5.222	350
Main effects of frequency band
Right MFG	47, 32	24, 21, 18	4.421	321
Bilateral BS/CAL		−3, 39,−12	3.964	206
Left MFG	10,11	−9, 36, 9	3.188	293
Right BS		12, 33, 57	−2.914	130
Bilateral MCC	24	12, 21, 18	−4.117	716
Right Ins/STG	22, 13	45, 3, 3	−3.908	116

Open in a separate window

Abbreviations: ALFF, amplitude of low-frequency fluctuation; BA, Brodmann area; BS, brain stem; CAL, cerebellum anterior lobe; CPL, cerebellum posterior lobe; Ins, insular; MCC, midcingulate cortex; MFG, medial frontal gyrus; MNI, Montreal Neurological Institute; PCUN, precuneus; PFC, prefrontal cortex; SMA, supplementary motor area; sMTG, superior middle temporal gyrus; STG, superior temporal gyrus.

Several brain regions showed a significant main effect for frequency band (Figure 2B; Table 3), including a significantly increased ALFF (slow-5 > slow-4) in the right medial frontal gyrus (MFG), the left MFG, the bilateral BS/CAL, and a significantly decreased ALFF (slow-5 < slow-4) in the right BS, the bilateral midcingulate cortex, and the right insular/superior temporal gyrus. There were no significant interactions between different groups and frequency bands (Figure 2C).

Further post hoc t-tests showed bidirectional changes in ALFF across slow-4 and slow-5 frequencies between the PHN and HC groups. Significantly decreased ALFF values were identified in the bilateral cuneus/lingual gyrus and the right PFC in the slow-4 band, but significantly increased ALFF values were identified in the bilateral BS/CAL and the left parieto-occipital cortex (POC) in the slow-5 band (Figure 3; Tables 4 and ​and55).

Open in a separate window

Figure 3

Difference in the ALFF values in the brain between postherpetic neuralgia patients and healthy controls across slow-5 (A) and slow-4 (B) frequencies, with Gaussian random field theory correction, voxel-level P<0.025 and cluster-level P<0.05.

Abbreviations: ALFF, amplitude of low-frequency fluctuation; HCs, healthy controls; L, left; LH, left hemicerebrum; PHN, postherpetic neuralgia; R, right; RH, right hemicerebrum.

Table 4

In the slow-5 band (0.001–0.027 Hz), ALFF differences between the PHN and HC groups with Gaussian random field theory correction, voxel-level P<0.025 and cluster-level P<0.05

Brain area	BA	Peak (MNI)	Peak	Voxel size
		x, y, z
PHN > HCs
Bilateral BS/CAL		–6, 24,–21	4.552	466
Left parieto-occipital cortex	19, 39, 7	–36, 69, 30	4.245	196

Open in a separate window

Abbreviations: ALFF, amplitude of low-frequency fluctuation; BA, Brodmann area; BS, brain stem; CAL, cerebellum anterior lobe; HCs, healthy controls; MNI, Montreal Neurological Institute; PHN, postherpetic neuralgia.

Differential ALFF values between groups in the typical frequency band (0.01–0.08 Hz)

Compared with HCs, PHN patients showed significantly decreased ALFF values in the right PFC and increased ALFF values in the BS/CAL in the typical band. The PFC is the hub region of the default mode network (DMN);²⁵ the DMN most commonly shows the highest intrinsic activity during rest, and this activity declines during some attention-demanding cognitive tasks.^25,26 The reduction in intrinsic activity in the DMN may be due to the maintenance of attention in PHN pain; pain sensations conflict with the resting state of the brain, which is dominated by the DMN.²⁷ The PFC is also a vital region in the cognitive control network²⁸ and is involved in cognitive decision making, such as pain evaluation and response decisions.²⁹ Therefore, the reduced activity of the PFC may be an underlying reason for the cognitive symptoms of depression in PHN patients. Other previous neuroimaging studies also suggested microstructural abnormalities,³ decreased interhemispheric intrinsic connectivity,⁹ and reduced ALFF values¹⁵ in the PFC in PHN patients.

In addition, we found significantly increased ALFF values in the BS/CAL; this result is consistent with a recent functional neuroimaging study showing increased regional homogeneity (ReHo) and fractional aptitude of low-frequency fluctuation (fALFF) in the brain stem and cerebellum in PHN patients.¹⁶ The cerebellum is part of the pain matrix³⁰ and is involved in nocifensive behavior.³¹ Activation of the cerebellum is often reported during the processing of pain. The CAL is an important region in the cerebellum, and the CAL and BS are closely linked in structure and function. The BS is a major part of pain processing and modulation of nociceptive input; thus, the BS/CAL may be a key region involved in descending pain modulation.

Main effect of the groups and frequency bands

The main effect of the groups in ANOVA showed significant differences in ALFF values between PHN patients and HCs, including increased ALFF values (PHN > HCs) in the bilateral BS/CAL, the left PCUN/sMTG, and the right PCUN and decreased ALFF values (PHN < HCs) in the right CPL, the right PFC, the left PFC, and the right SMA/M1. Changes in ALFF values in the BS/CAL and the PFC were the same as those in the typical frequency band between groups. The PCUN/sMTG is an information processing region. The PCUN continuously gathers information from both external and internal milieu and plays vital roles in monitoring sensory information³² and implementing a wide range of higher-order cognitive functions.³³ The MTG might be involved in the processing of somatosensory stimuli.³⁴ Therefore, the increased ALFF values in the PCUN/sMTG could be explained by the increases in spontaneous neuronal activity in the information processing regions, which show functional compensation or plasticity. The CPL plays an important role in pain regulation.³⁵ Cerebellar areas are activated by painful stimulation in chronic neuropathic pain patients.^35,36 Nociception may induce activation of the CPL in PHN patients. Moreover, the SMA is part of the pain matrix,¹⁵ which is involved in motor responses to pain. In the past, researchers have reported that pain-related regions (SMA/M1,³⁷ cerebellum,³⁸ S1,³⁹ and S2⁴⁰) play a vital role in sensorimotor control, and increased functional connectivity in the SMA has been reported in chronic low back pain patients.⁴¹

In addition, the main effect of frequency band revealed increased ALFF (slow-5 > slow-4) in the MFG, the left MFG, and the bilateral BS/CAL and significantly decreased ALFF (slow-5 < slow-4) in the right BS, the bilateral midcingulate cortex, and the right insular/superior temporal gyrus. Different neurophysiological mechanisms induce different oscillatory bands and have varying physiological functions.⁴² Our results are similar to previous rs-fMRI studies, showing that the lower frequency band has higher power in the prefrontal, occipital, and parietal cortices. The higher frequency band has lower power in subcortical structures, including the thalamus and the basal ganglia.⁴² The ventromedial prefrontal cortices in the slow-5 band are more dominant than those in the slow-4 band.¹⁷

Frequency-dependent alterations in ALFF in PHN patients

Interestingly, we found bidirectional alterations in ALFF values across slow-4 and slow-5 frequency bands between the PHN and HC groups, which represent a novel discovery regarding frequency-dependent alterations in PHN patients. We observed significantly decreased ALFF values in the bilateral cuneus/lingual gyrus and the right PFC in the slow-4 band but significantly increased ALFF values in the bilateral BS/CAL and the left POC in the slow-5 band. Decreased ALFF values in the right PFC and increased ALFF values in the bilateral BS/CAL were consistent with the results of the typical frequency band; combined with the slow-4 and slow-5 bands, these values may be more sensitive for detecting alterations in spontaneous brain activity in PHN patients. However, this idea requires further study. Moreover, increased ALFF values in the slow-5 band in the left POC were positively correlated with VAS scores. The POC may be involved in the integration of visual information and somatosensory signals;⁴³ additionally, the POC is activated in response to the prickle sensation induced by cold stimuli.⁴⁴ Thus, we suggest that increased neuronal activity in the POC in the slow-5 band may indicate that patients were enduring sharp pain, and pain intensity may play a dominating role. In addition, increased ALFF values in the slow-5 band in the bilateral BS/CAL were positively correlated with disease duration. The BS/CAL is a pain processing and modulation region; long-term PHN pain may aggravate pain processing and modulation in the BS/CAL and may induce spontaneous neuronal activity.

This work was supported by grants from the Jiangxi Province Health Department Support Program (grant nos. 20181057 and 2017A026) and the Jiangxi Province Education Department Support Program (grant no. GJJ170075).