Pulsed radiofrequency (PRF) is a well-established neuromodulation technique widely used for managing neuropathic pain. Lumbosacral radicular pain, a common neuropathic condition, is often refractory to conventional treatments. Although dorsal root ganglion (DRG) PRF has emerged as a promising intervention, its therapeutic efficacy is often limited and variable, likely due to an incomplete understanding of its mechanisms. To elucidate the neural mechanisms underlying DRG PRF analgesia, this study characterized treatment-induced alterations in microstate spatiotemporal dynamics and examined their correlation with pain intensity, thereby assessing their potential as neurophysiological markers.
Methods
We recorded high-density electroencephalograms (EEG) in healthy controls and patients before and after DRG PRF treatment. Topographic differences were assessed using topographic analysis of variance (TANOVA). Microstate temporal parameters (duration, occurrence, coverage) and transition probabilities were analyzed. Pearson correlation analysis was performed between transition probabilities and visual analogue scale (VAS) scores.
Results
TANOVA revealed significant differences in microstate topographies among the three groups (p = 0.033), primarily attributed to microstates C and E. Although we found no significant differences in global temporal parameters or transition probabilities, our exploratory analysis revealed a reduction in the transition probability from microstate D to E (Delta TM D to E) in patients before DRG PRF treatment compared to healthy controls (p = 0.016, uncorrected). Notably, this reduction showed a trend toward normalization after treatment. Furthermore, we observed a significant negative correlation between Delta TM D to E and VAS scores (r = −0.459, p = 0.008).
Conclusion
DRG PRF alleviates neuropathic pain by normalizing interactions between large-scale brain networks, as evidenced by topographic reorganizations and trends toward normalized transition dynamics. The sensitivity of microstate metrics to these changes supports their potential as neurophysiological markers for assessing both pain-related brain dysfunction and treatment response.
Trial Registration
The trial was registered on ClinicalTrials.gov with the following number: ChiCTR2500104921.
Dorsal root ganglion pulsed radiofrequency (DRG PRF) is a promising therapy for refractory lumbosacral radicular pain; however, its efficacy is often limited and variable due to an incomplete understanding of mechanisms.
Microstate analysis is an advanced method for studying the spatiotemporal dynamics of large-scale neural networks, which can be used to investigate the central mechanisms of DRG PRF treatment.
This study aimed to determine whether and how DRG PRF treatment exerts its analgesic effect by modulating the spatiotemporal characteristics of aberrant microstates.
What was learned from this study?
DRG PRF treatment significantly altered microstate topographies (primarily microstates C and E) and promoted normalization of the specific transition probability from microstate D to E. This specific change was negatively correlated with visual analogue scale (VAS) scores.
These findings provide evidence that DRG PRF treatment alleviates pain by normalizing interactions between large-scale brain networks, suggesting a central mechanism that extends beyond peripheral effects.
Consequently, microstate metrics show promise as neurophysiological markers for objectively evaluating treatment response in neuropathic pain and warrant further validation in larger cohorts.
Introduction
Pulsed radiofrequency (PRF) is a neuromodulation modality that has been widely utilized in the management of neuropathic pain for decades [1]. Unlike conventional continuous radiofrequency (CRF), which can cause tissue destruction or painful sequelae, PRF delivers short, high-voltage bursts of current to generate an electric field around neural tissues without these adverse effects [2]. Compared to oral analgesics, PRF provides longer and better analgesic effects with minimal complications, making it a valuable option for neuropathic pain management [3].
Anzeige
Lumbosacral radicular pain is a typical neuropathic pain, with an estimated prevalence of 9.9% to 25% [4]. A randomized trial has shown that even surgical interventions may not achieve desired therapeutic outcomes [5]. Radicular pain is defined as “pain [that] radiates in a dermatomal distribution, while in the case of radiculopathy, objective sensory, motor, and/or reflex loss are usually present” [6]. Although the pathophysiology is not fully understood, radicular pain is believed to originate from lesions that either directly compromise the dorsal root ganglion (DRG) mechanically or indirectly compromise the spinal nerve and its roots by inducing axonal ischemia or inflammation [1]. Consequently, the DRG is considered an optimal therapeutic target for radicular pain [7].
Since its introduction over 20 years ago [8], DRG PRF treatment has demonstrated consistent clinical efficacy and an excellent safety profile, leading to growing acceptance in pain management practices [9]. However, several limitations remain. First, although DRG PRF provides longer-lasting analgesia than pharmacological treatments, its clinical benefits often diminish over time [10]. Second, it was found that even when more than 50% pain relief was achieved, a subset of patients continued to report moderate to severe pain [11]. These unresolved issues highlight the critical need to elucidate the underlying mechanisms of DRG PRF.
Previous investigations into the mechanisms of DRG PRF have primarily focused on the spinal cord or relied on animal models [12‐14]. As an application introduced to clinical practice, PRF is believed to alleviate pain by inducing long-term depression (LTD), thereby suppressing pain transmission signals from peripheral nerves to the central nervous system [15]. This mechanism suggests that the effect of DRG PRF is not only restricted to the spinal cord but also related to the supraspinal region. Although studies in animal models have underscored the critical role of supraspinal regions [16, 17], direct human research is still lacking. An electroencephalogram (EEG), which records the scalp’s postsynaptic potentials of neuronal populations with high temporal resolution [18], offers a distinct advantage for investigating brain activity in patients with neuropathic pain. Initial studies have focused on static EEG measures and integrated machine learning to establish them as objective biomarkers for pain assessment, thereby overcoming the clinical limitations of subjective pain assessment [19, 20]. However, the dynamic properties of brain activity are equally crucial. For instance, May et al. demonstrated altered brain dynamics in patients with chronic back pain [21], Dinh et al. found increased frontal connectivity in the theta and gamma bands alongside global network reorganization in patients with chronic pain [22], and Hu et al. reported that laser-induced gamma oscillations recorded from central electrodes both reliably and selectively predicted pain sensitivity across individuals [23]. Collectively, these findings suggest that the temporal dynamics of brain activity may reflect individual pain sensitivity, explain interindividual variability in pain perception, and demonstrate the possible mechanisms of chronic pain.
Microstate analysis provides an advanced framework for characterizing the spatiotemporal dynamics of large-scale neural networks [24]. EEG microstates are defined as “quasi-stable” periods of electrical topography [25] that can be clustered into 4–7 canonical classes based on topographic similarity [26]. Studies employing source imaging and simultaneous EEG–functional magnetic resonance imaging (fMRI) have linked these microstate classes to distinct patterns of large-scale brain network activity [27], underscoring the value of microstate analysis for elucidating the mechanisms of DRG PRF analgesia. Although EEG microstate dynamics are known to be altered in various pain conditions [21, 28‐31], the modulation of these dynamics by DRG PRF treatment remains unexplored.
Anzeige
Therefore, this study aims to investigate the spatiotemporal dynamic changes induced by DRG PRF treatment in patients with lumbosacral radicular pain, assess the correlation between these alterations in microstate dynamics and clinical pain outcomes, and evaluate the potential of microstate metrics to serve as neurophysiological markers for elucidating the neural mechanisms underlying DRG PRF analgesia.
Methods
Participants
The study enrolled patients scheduled for DRG PRF treatment for lumbosacral radicular pain. The inclusion criteria were as follows: age between 18 and 80 years, with no gender restrictions, a preoperative visual analogue scale (VAS) score greater than 3, willingness to participate, and ability to cooperate with the collection of case report information and EEG acquisition. Exclusion criteria included a history of mental illness or cognitive dysfunction, comorbid chronic pain conditions, inability to provide self-reported pain scores, and scalp lesions or infections. All participants provided written informed consent for both the surgical procedure and their participation in the research. Pre-existing necessary medications were maintained throughout the study. EEG recording was performed following the confirmation of a stable pain state. Demographic and clinical data, including age, sex, affected side, use of pain-related medication, pain duration, and pain severity, were collected for subsequent analysis.
Ethical Approval
This study was conducted in accordance with the principles outlined in the Declaration of Helsinki and approved by the Ethics Committee of The Third Xiangya Hospital, Central South University, China (No. R24033). The trial was registered on ClinicalTrials.gov with number ChiCTR2500104921.
Surgical Details
The surgical procedure was performed as described previously [11]. Briefly, patients were placed in the prone position. The target DRGs for PRF treatment were determined by the dermatomal distribution and the results of imaging examinations (unilateral or bilateral). The needle tip position was confirmed under X-ray guidance and through sensory testing. PRF treatment was delivered at 2 Hz with a 20-ms pulse width for three 240-s cycles using a radiofrequency generator (R-2000BA1, Beiqi, Beijing, China). The impedance was maintained below 300 Ω throughout the procedure.
Data Collection
A 128-channel system (Yunshen Technology Limited Company, Beijing, China) was used to acquire resting-state EEG with Cz as the reference. The sampling rate was 1024 Hz. Participants were asked to sit comfortably and remain relaxed. They were then instructed to remain still, minimize eye movements, and keep their eyes closed during the recording. The entire recording process lasted approximately 5 to 6 min. All participants were instructed to remain awake during the recording, and post-recording verbal confirmation was obtained to exclude data from those who reported drowsiness or sleep.
EEG Preprocessing
The EEG data were preprocessed using MATLAB R2020a (The MathWorks Inc., Natick, MA, USA) using the EEGLAB toolbox v2023.1 (https://sccn.ucsd.edu/eeglab/) [32]. The preprocessing pipeline was implemented according to the protocol established by Luo et al. The continuous EEG data were offline bandpass-filtered between 0.5 and 90 Hz, with line noise (48–52 Hz) and its harmonics (98–102 Hz) removed. After interpolating the bad channels with spherical splines, the data were re-referenced to the average reference. Independent component analysis (ICA) was conducted to eliminate ocular, muscular, cardiac, and electrode artifacts. The analyzed signals were then segmented into 2-s epochs, and epochs with amplitudes exceeding 100 μV were rejected. Following the above procedures, datasets with either > 15% bad channels or > 50% rejected epochs were excluded from further analysis [33]. Finally, the analyzed data were bandpass-filtered between 1 and 40 Hz and resampled to 250 Hz, and the mastoid electrodes (M1, M2) were excluded prior to the microstate analysis [24].
Microstate Analysis
Microstate analysis was performed using MATLAB R2022b (The MathWorks Inc., Natick, MA, USA) with the EEGLAB toolbox (v2024.2.1) [32] and MicrostateLab (v2.1) [24]. The entire processing pipeline followed the protocol established by Koenig et al. [26], as illustrated in Fig. 1, and consisted of four steps.
Fig. 1
Microstate analysis pipeline. For each preprocessed electroencephalogram (EEG) recording, global field power (GFP) peaks were identified, and corresponding topographic maps were extracted. These maps were first clustered into five distinct microstate (MS) classes at the individual level. The resulting topographies were then clustered across participants to generate group-level mean microstates, which were labeled A through E based on the MetaMaps template. Finally, each participant's data were back-fitted to these group templates to extract temporal parameters (duration, occurrence, coverage, transition probabilities) for subsequent statistical analysis
First, global field power (GFP) peaks were extracted to compute topographic maps, which were initially clustered to identify five classes of microstate maps per subject. Second, group-level microstate maps were derived through a second clustering step, in which individual maps were aggregated across subjects. In accordance with established guidelines, k-means clustering with 50 iterations was applied to ensure robustness of the results. Group-level maps were computed separately for healthy controls (HC group), the group of patients before DRG PRF treatment (pre-PRF group), and the group of patients after DRG PRF treatment (post-PRF group). Third, the resulting microstate maps were edited and sorted according to the published MetaMaps template [34]. Finally, individual maps were back-fitted to the corresponding group-level templates, and temporal parameters were exported for subsequent analysis (see Supplementary Material for the full dataset).
Anzeige
Four key temporal characteristics were quantified from the microstate time series: duration, occurrence, coverage, and adjusted transition probabilities (denoted as Delta TM X to Y; hereafter referred to simply as “transition probabilities”). Detailed interpretations of these parameters are provided in Table 1 [24].
Table 1
Overview of parameters derived from microstate analysis in this study and their neurophysiological interpretability in relation to brain network activity
Average duration of all microstates belonging to the microstate class X
Stability of neural networks represented by microstate class X
Occurrence of class X
Frequency (per second) that microstates of class X were observed
Frequency of activating neural networks represented by class X
Coverage of class X
Percentage of total time a microstate class is present
Percentage of time that the brain spent in neural networks represented by class X
Adjusted transition probabilities (DeltaTM(X → Y)
Differences in transition probabilities against the transition probabilities as they are expected under randomness, given the relative occurrences of particular microstate classes.
This is calculated as follows:
DeltaTM(X → Y) = \(\frac{OrgTM(X \to Y) - ExpTM(X \to Y)}{{ExpTM\left( {X \to Y} \right)}} \times 100\) where OrgTM((X → Y) is computed according to Nagabhushan Kalburgi et al. [24]. ExpTM(X → Y) is computed according to Lehmann et al. [35]
Relative preference or resistance to entering microstate Y from microstate X, against chance level. Addresses the question of whether there are particular rules (or “syntax”) in the chain of microstates
Statistics
Statistical analyses were conducted using SPSS 27.0 (SPSS Statistics, IBM Corp., Armonk, NY, USA), GraphPad Prism 10.4.2 (GraphPad Software, San Diego, CA, USA), and MATLAB R2022b. Continuous data are presented as mean ± standard deviation, and categorical variables as percentages. The normality of continuous variables was assessed using the Shapiro–Wilk test.
For demographic data, normally distributed continuous variables were compared using one-way analysis of variance (ANOVA), followed by Tukey’s honest significant difference (HSD) test for post hoc multiple pairwise comparisons. Given the absence of pain duration data in the HC group, this variable was compared between the pre- and post-PRF groups using an unpaired t-test. Non-normally distributed continuous variables were analyzed using the Kruskal–Wallis test, followed by Dunn’s test for post hoc comparisons. Categorical variables were evaluated using Fisher’s exact test.
Topographic differences in microstates were assessed using topographic analysis of variance (TANOVA) with a custom script (available at https://github.com/simonruch/topographic_analyses_eeg) and Field Trip (v20240916) [34]. A global TANOVA was first performed to test for any significant topographic differences among the three groups. Following a significant global effect, post hoc pairwise TANOVAs were conducted to identify specific group differences. The effect size for each pairwise comparison was quantified by computing the GFP of the difference between the respective GFP-normalized, group-average maps. To control for multiple comparisons, p-values from post hoc pairwise tests were adjusted using the false discovery rate (FDR) correction.
Anzeige
A mixed-model ANOVA was employed to analyze the temporal parameters (duration, occurrence, coverage), with microstate class as the within-subjects factor and group as the between-subjects factor. Transition probabilities were compared using permutation testing (with a permutation number of 5000). The relationship between the transition probability from D to E (Delta TM D to E) and VAS scores was evaluated using Pearson correlation. A p-value < 0.05 was considered statistically significant.
Results
General Demographics
Following preprocessing, the final dataset included 12 patients before DRG PRF treatment, nine patients after DRG PRF treatment, and 11 healthy controls. All participants were right-handed. Patients in the pre-PRF group presented with moderate to severe pain, whereas those in the post-PRF group reported mild pain or no pain. Demographic and clinical characteristics are summarized in Table 2. A significant difference in VAS scores was found among the three groups (p < 0.001). In contrast, no significant differences were observed in age (p = 0.770), sex (p = 0.453), years of education (p = 0.509), or pain duration (p = 0.520).
Table 2
General demographics and clinical characteristics
HC
Pre-PRF
Post-PRF
p
Number
11
12
9
–
Age (years)
64.82 ± 6.78
65.75 ± 6.36
63.67 ± 6.38
0.770
Sex (female, %)
10 (90.91%)
10 (83.33%)
6 (66.67%)
0.453
Education (years)
8.18 ± 2.09
6.83 ± 2.89
8.00 ± 3.46
0.509
VAS scores***
0.00
5.83 ± 0.83
1.33 ± 1.00
< 0.001
Pain duration
(months)
–
18.83 ± 15.62
23.22 ± 14.54
0.520
Continuous variables are presented as mean ± SD
***p < 0.001
VAS scores visual analogue scale scores; HC healthy control group; pre-PRF the group of patients before pulsed radiofrequency (PRF) treatment; post-PRF the group of patients after PRF treatment
Spatiotemporal Characteristics of Microstates
Microstates were classified into five distinct clusters for each group (Fig. 2). This configuration was selected as it yielded the highest global explained variance (HC: 79.26%; pre-PRF: 75.57%; post-PRF: 81.92%) among models with 4–7 classes, which was consistent with a previous study on chronic back pain [21].
Fig. 2
Group-level microstate topographies. The five canonical microstate classes (A–E) derived for each group are shown. Blue and red colors correspond to negative and positive values, respectively, with intensity representing magnitude. MS: microstate; Patients pre-PRF: patients before pulsed radiofrequency (PRF) treatment; Patients post-PRF: patients after PRF treatment
Microstate-class-window-based TANOVA revealed significant differences (p = 0.033) in topographies among the three groups, primarily driven by microstates C and E (Fig. 3). Post hoc pairwise comparison showed that for microstate C, all group differences remained significant after FDR correction (HC vs. pre-PRF: PFDR = 0.032, GFP diff = 0.342; HC vs. post-PRF: PFDR = 0.047, GFP diff = 0.281; pre-PRF vs. post-PRF: PFDR = 0.032, GFP diff = 0.384). Similarly, for microstate E, all pairwise comparisons were also significant (HC vs. pre-PRF: PFDR = 0.041, GFP diff = 0.326; HC vs. post-PRF: PFDR = 0.019, GFP diff = 0.385; pre-PRF vs. post-PRF: PFDR = 0.047, GFP diff = 0.275).
Fig. 3
Topographic differences in microstates C and E. A topographic analysis of variance (TANOVA) revealed a significant main effect among groups (p = 0.033). Post hoc tests with false discovery rate (FDR) correction showed that these differences were primarily driven by microstates C and E, with all pairwise comparisons (HC vs. pre-PRF, HC vs. post-PRF, pre-PRF vs. post-PRF) remaining significant (PFDR < 0.05). Each row displays the group-average topography for a given microstate class across the three groups. *PFDR < 0.05. HC healthy control group; pre-PRF the group of patients before PRF treatment; post-PRF the group of patients after PRF treatment.
Our results revealed no significant differences in the microstate temporal parameters, including duration, occurrence, and coverage, among the HC, pre-PRF, and post-PRF groups (Fig. 4; Table 3). Global transition probabilities also showed no significant differences among groups.
Fig. 4
Microstate temporal parameters across groups. Group-level comparisons of microstate a duration, b coverage, and c occurrence. A mixed-model ANOVA with microstate class as the within-subjects factor and group as the between-subjects factor revealed no significant differences in any temporal parameters (see Table 2 for details)
The reported p-value represents the significance of the interaction effect between microstate classes and groups
MS microstate; PRF pulsed radiofrequency
Given the limited sample size and the consequent risk of type II error, we performed exploratory pairwise comparisons to identify potential subthreshold alternations in transition dynamics (Table 4). These analyses, which should be interpreted with caution in the absence of an overall significant effect, suggested a specific pattern: Delta TM D to E was reduced in the pre-PRF group compared to the HC group, with a strong trend (p = 0.016, uncorrected for multiple comparisons; Fig. 5a, Table 4). Notably, this measure exhibited a trend toward normalization after DRG PRF treatment, approaching levels observed in the HC group despite most patients reporting residual mild pain.
Table 4
Pair comparisons of transition probabilities
HC
pre-PRF
post-PRF
p
p
p
HC vs. pre-PRF
HC vs. post-PRF
Pre- vs. post-PRF
A to B
−2.76 ± 8.66
−5.35 ± 10.99
−0.29 ± 7.29
0.512
0.564
0.221
A to C
−2.21 ± 6.87
2.33 ± 10.58
−4.12 ± 7.00
0.218
0.636
0.084
A to D
−0.65 ± 8.86
−1.03 ± 9.90
−1.36 ± 10.66
0.920
0.877
0.936
A to E
2.38 ± 9.01
−1.52 ± 9.12
1.08 ± 10.20
0.326
0.765
0.544
B to A
−6.18 ± 10.62
−3.18 ± 10.95
−0.05 ± 6.66
0.478
0.165
0.484
B to C
1.77 ± 8.99
0.18 ± 6.78
1.31 ± 6.02
0.613
0.891
0.729
B to D
4.81 ± 4.77
−1.80 ± 11.84
−2.15 ± 13.41
0.135
0.149
0.938
B to E
−5.10 ± 6.09
−2.54 ± 11.06
−7.01 ± 11.29
0.532
0.664
0.294
C to A
−7.28 ± 9.07
−1.55 ± 10.74
−3.50 ± 7.67
0.146
0.385
0.651
C to B
−0.80 ± 9.79
−3.93 ± 7.73
−3.18 ± 6.06
0.362
0.525
0.833
C to D
5.65 ± 8.56
2.56 ± 9.56
−5.17 ± 15.15
0.529
0.035●
0.135
C to E
−3.16 ± 8.44
−2.94 ± 9.56
4.12 ± 10.01
0.955
0.090
0.098
D to A
−4.55 ± 8.20
−3.48 ± 9.80
2.03 ± 11.90
0.808
0.138
0.218
D to B
−0.60 ± 9.74
−2.67 ± 7.18
−4.85 ± 6.78
0.555
0.241
0.532
D to C
−1.54 ± 12.15
6.94 ± 11.81
−5.36 ± 12.59
0.116
0.514
0.026●
D to E
0.89 ± 7.23
−8.50 ± 11.57
1.85 ± 6.38
0.016●
0.838
0.012●
E to A
−2.09 ± 10.16
−2.25 ± 15.19
−0.82 ± 9.99
0.976
0.809
0.799
E to B
−5.26 ± 9.79
0.08 ± 10.75
−6.44 ± 5.66
0.175
0.791
0.117
E to C
−1.23 ± 7.89
−5.92 ± 7.64
1.88 ± 10.13
0.200
0.432
0.045●
E to D
1.57 ± 11.51
3.97 ± 12.86
−1.90 ± 9.62
0.629
0.500
0.251
Variables are presented as mean ± SD
●Indicates uncorrected p < 0.05
PRF: pulsed radiofrequency
Fig. 5
Altered transition dynamics and clinical correlation. a The transition probability from microstate D to E (Delta TM D to E) was reduced in the patients before the pulsed radiofrequency (PRF) versus the healthy controls, with a strong but nonsignificant trend (p = 0.016, uncorrected). A trend toward normalization was observed after PRF treatment (see Table 4 for complete statistics). Between-group differences in visual analogue scale (VAS) were assessed using the Kruskal–Wallis test, followed by Dunn’s test for post hoc comparisons. b Delta TM D to E showed a significant negative correlation with VAS scores. (r = −0.459, p = 0.008). ● p < 0.05 (uncorrected), ** p < 0.01, **** p < 0.0001
Correlation Between VAS and Transition Probability
We next examined whether alterations in transition probabilities correlated with clinical pain severity. Pearson correlation analysis revealed a significant negative correlation between Delta TM D to E and VAS scores (r = −0.459, p = 0.008; Fig. 5b).
Discussion
This study characterized EEG microstate characteristics in patients with lumbosacral radicular pain before and after DRG PRF treatment, in comparison to healthy controls. To our knowledge, it represents the first EEG investigation to demonstrate DRG PRF treatment-induced alterations in both microstate topography and temporal dynamics. Our findings indicate that while DRG PRF significantly modulated microstate topography in patients with lumbosacral radicular pain, particularly in microstates C and E, it did not affect global temporal parameters (duration, occurrence, coverage). Exploratory analysis further suggested a trend toward normalization in Delta TM D to E following treatment, which was negatively correlated with VAS scores.
We observed significant differences in the topographies of microstates C and E among all groups, suggesting distinct configurations of large-scale brain network activity. Microstate C is primarily associated with the salience network, which is involved in processing personally relevant information, self-referential thinking, and internal awareness [27]. Its alteration may reflect the capture of attentional resources by pain, which acts as a persistent salient stimulus. Microstate E, also linked to the salience network, is implicated in interoceptive and emotional processing, as well as specific cognitive functions [27]. Changes in this microstate may indicate disruptions in emotional and cognitive processes resulting from pain, potentially reflecting a functional overlap with microstate C. Together, these topographic alterations suggest that DRG PRF may alleviate pain salience and restore cognitive control by modulating interactions within these networks.
The transition probability reflects the relative preference for or resistance to transitioning from microstate X to microstate Y, against chance level, thereby indicating whether particular rules (or “syntax”) govern the temporal sequence of microstates [26]. In our study, the pre-PRF group exhibited a reduction in Delta TM D to E compared to both the HC and the post-PRF groups, suggesting that this specific transition occurred less frequently than expected. Although this finding emerged from an exploratory, uncorrected analysis, it demonstrated a strong statistical trend (p = 0.016) and was further supported by a significant negative correlation with VAS scores. These observations suggest that chronic pain may alter the dynamic interplay between attention-executive networks (microstate D) and interoceptive-salience networks (microstate E), and that DRG PRF may help normalize these pathological adaptations.
Microstate analysis originated from Lehmann et al.’s observation that EEG topography remained stable for 80–120 ms before rapidly transitioning to another configuration [25]. Subsequent research utilizing source localization and fMRI-EEG fusion studies has established that distinct microstates correspond to the activity of specific large-scale brain networks. In the present study, we focused on microstates D and E due to the notable reduction in their transition probability in the pre-PRF group. Generally, microstate D is associated with attention and executive functioning, while microstate E is linked to interoceptive-emotional processing and salience detection [27]. The decreased Delta TM D to E suggests a specific disruption in the dynamic switching between these networks in chronic pain states.
Although the duration, occurrence, or coverage of microstate E did not reach statistical significance, all three metrics were consistently lower in the pre-PRF group than in both the HC and post-PRF group. This pattern aligns with previous literature: Tomescu et al. reported a negative correlation between microstate E and somatic awareness [36], which is itself linked to heightened pain sensitivity [37]. In addition, microstate E has been positively correlated with psychological well-being and cognitive planning [38], suggesting that its suppression may reflect impaired affective regulation in pain conditions. Further supporting this interpretation, a study on experimental central sensitization found that the transition probability from E to D correlated positively with mechanical hypersensitivity [39], consistent with our observed negative correlation between Delta TM D to E and VAS scores. Together, these consistencies suggest that the microstate temporal dynamics reflect shared neural pathways across different forms of pain perception.
In contrast to our results, most prior studies report significant alterations in conventional microstate temporal parameters. For instance, one microstate analysis in patients with chronic pain found that microstate D exhibited significantly decreased duration, coverage, and occurrence, along with reduced transition probabilities from microstates A, B, and E to D [21]. Our study did not detect significant changes in these standard temporal parameters for any microstate class. This discrepancy may be attributable to our relatively small sample size. Nevertheless, we observed consistent, directional (though nonsignificant) changes in parameters related to microstates C and E in the pre-PRF group, with a trend to normalization after DRG PRF treatment.
The lack of significant differences in conventional temporal parameters could also reflect active compensatory mechanisms within brain networks. Specifically, the reduced Delta TM D to E may indicate a functional decoupling between interoceptive and executive networks, a potentially adaptive response that limits pain catastrophizing [40] and thereby obscures changes in other temporal metrics.
Divergent microstate findings across studies highlight the context-dependent nature of pain-related network reorganization. For instance, whereas patients with fibromyalgia showed significantly reduced occurrence and coverage of microstate 1 (topographically similar to microstate C) compared to healthy controls [31], individuals with high postsurgical pain after breast cancer surgery exhibited increased occurrence and coverage of the same microstate [29]. These opposing patterns may reflect differences in pain etiology or underlying neuropathic mechanisms. Furthermore, the high spatial similarity between microstates C and E has led some researchers to merge them into a single cluster [41], underscoring the need for clearer functional and neuroanatomical distinctions between microstate classes.
Other pain-related studies report alterations in microstates A and B, which are associated with auditory and visual processing, respectively [27]. Similarly, one study suggested that neuropathic pain and tinnitus share common neural substrates, reflected in specific brain connectivity patterns and microstate profiles [30]. Although pain types and findings vary across studies, a common theme emerges: patients with neuropathic pain consistently exhibit significant alterations in microstate spatiotemporal dynamics. These alterations may represent potential biomarkers for assessing pain severity, highlighting the functional relevance of microstate analysis in pain research.
Limitations
Several limitations of this study should be acknowledged. First, the sample size used to compare microstate alterations before and after DRG PRF treatment was relatively small. Second, the absence of long-term follow-up precludes conclusions regarding the persistence of the observed effects. Despite these limitations, the results revealed meaningful trends that provide a preliminary foundation for future research. Additionally, this study exclusively enrolled patients with lumbosacral radicular pain. Given the demonstrated efficacy of PRF for other neuropathic pain conditions [1], it remains unclear whether PRF interventions at non-DRG sites would elicit similar microstate dynamics. These questions warrant systematic investigation in future studies.
Conclusions
This study demonstrated significant topographical differences in EEG microstates among the HC, pre-, and post-PRF groups, specifically in microstates C and E, suggesting a treatment-induced modulation of large-scale brain network activity. In contrast, global temporal parameters (duration, occurrence, coverage) remained stable, indicating a distinct sensitivity of spatial versus temporal microstate characteristics to DRG PRF treatment. Exploratory analysis further identified a reduction in Delta TM D to E in the pre-PRF group, which trended toward normalization after treatment and was negatively correlated with VAS scores. This further supported a specific alteration in transition dynamics. Collectively, these findings position EEG microstate topography and transition dynamics as sensitive neurophysiological markers of pain-related brain dysfunction and treatment response. The results support the notion that DRG PRF acts not only peripherally but also by normalizing interactions among distributed brain networks.
Acknowledgements
The authors would like to thank all the participants of the study. We fondly remember and profoundly miss our late colleague, Raoxiang Zhang, who passed away on 27 January 2025, prior to the publication of this work. We are forever grateful for her significant contributions and the lasting impact of her work. All surviving authors and an authorized representative have warmly approved her inclusion as a co-author. We also thank Dr. Haocheng Zhou for managing the trial registration and coordinating patient recruitment.
Medical Writing/Editorial Assistance
The authors used automated grammar-checking tools (Grammarly) and an AI-powered writing assistant (DeepSeek) for language polishing and proofreading during manuscript preparation. No financial support was received for the use of these tools.
Declarations
Conflict of interest
Ying Yang, Lanxing Wu, Yan Pan, Xuelian Li, Dong Huang, Raoxiang Zhang, Kuankuan Li, and Yuzhao Huang have nothing to disclose.
Ethical Approval
The study was conducted in accordance with the principles outlined in the Declaration of Helsinki. This study was approved by the Ethics Committee of The Third Xiangya Hospital, Central South University, China (No. R24033). The trial was registered on ClinicalTrials.gov with the following number: ChiCTR2500104921.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/.
Chang MC. Efficacy of pulsed radiofrequency stimulation in patients with peripheral neuropathic pain: a narrative review. Pain Phys. 2018;21(3):E225–34.CrossRef
2.
Byrd D, Mackey S. Pulsed radiofrequency for chronic pain. Curr Pain Headache Rep. 2008;12(1):37–41.CrossRefPubMedPubMedCentral
3.
Vanneste T, Van Lantschoot A, Van Boxem K, Van Zundert J. Pulsed radiofrequency in chronic pain. Curr Opin Anaesthesiol. 2017;30(5):577–82.CrossRefPubMed
4.
Knezevic NN, et al. Low back pain. Lancet. 2021;398(10294):78–92.CrossRefPubMed
5.
Weinstein JN, et al. Surgical vs nonoperative treatment for lumbar disk herniation: the Spine Patient Outcomes Research Trial (SPORT): a randomized trial. JAMA. 2006;296(20):2441–50.CrossRefPubMedPubMedCentral
6.
Peene L, et al. Lumbosacral radicular pain. Pain Pract. 2024;24(3):525–52.CrossRefPubMed
7.
Sapunar D, Kostic S, Banozic A, Puljak L. Dorsal root ganglion - a potential new therapeutic target for neuropathic pain. J Pain Res. 2012;5:31–8.CrossRefPubMedPubMedCentral
8.
Me S. The effects of pulsed radiofrequency field applied to the dorsal root ganglion-a preliminary report. Pain Clinic. 1998;11:109–17.
9.
Vuka I, et al. Efficacy and safety of pulsed radiofrequency as a method of dorsal root ganglia stimulation in patients with neuropathic pain: a systematic review. Pain Med. 2020;21(12):3320–43.CrossRefPubMed
10.
Cohen SP, et al. Pulsed radiofrequency of the dorsal root ganglia is superior to pharmacotherapy or pulsed radiofrequency of the intercostal nerves in the treatment of chronic postsurgical thoracic pain. Pain Phys. 2006;9(3):227–35.
11.
Yang L, et al. Clinical outcome of pulsed-radiofrequency combined with transforaminal epidural steroid injection for lumbosacral radicular pain caused by distinct etiology. Front Neurosci. 2021;15:683298.CrossRefPubMedPubMedCentral
12.
Li X, et al. Pulsed radiofrequency relieves neuropathic pain by repairing the ultrastructural damage of chronically compressed dorsal root ganglion. Neurosci Insights. 2025;20:26331055251339081.CrossRefPubMedPubMedCentral
13.
Lin ML, et al. Pulsed radiofrequency inhibited activation of spinal mitogen-activated protein kinases and ameliorated early neuropathic pain in rats. Eur J Pain. 2014;18(5):659–70.CrossRefPubMed
14.
Perret DM, et al. Application of pulsed radiofrequency currents to rat dorsal root ganglia modulates nerve injury-induced tactile allodynia. Anesth Analg. 2011;113(3):610–6.CrossRefPubMedPubMedCentral
15.
Sluijter ME, van Kleef M. Pulsed radiofrequency. Pain Med. 2007;8(4):388–9.CrossRefPubMed
16.
Fang X, Xu X, Lin X, Liu R. Downregulated spinal IRF8 and BDNF in NAC are involved in neuropathic pain-induced depression relief via pulsed radiofrequency on dorsal root ganglion in rat SNI model. Brain Res Bull. 2019;146:192–200.CrossRefPubMed
17.
Xu X, et al. Pulsed radiofrequency on DRG inhibits hippocampal neuroinflammation by regulating spinal GRK2/p38 expression and enhances spinal autophagy to reduce pain and depression in male rats with spared nerve injury. Int Immunopharmacol. 2024;127:111419.CrossRefPubMed
18.
Hu L. Zhang Z. EEG signal processing and feature extraction: Springer; 2019. p. 2019.
19.
Mussigmann T, Bardel B, Lefaucheur JP. Resting-state electroencephalography (EEG) biomarkers of chronic neuropathic pain. A systematic review. Neuroimage. 2022;258:119351.CrossRefPubMed
20.
Zis P, et al. EEG recordings as biomarkers of pain perception: where do we stand and where to go? Pain Ther. 2022;11(2):369–80.CrossRefPubMedPubMedCentral
21.
May ES, et al. Dynamics of brain function in patients with chronic pain assessed by microstate analysis of resting-state electroencephalography. Pain. 2021;162(12):2894–908.CrossRefPubMedPubMedCentral
22.
Ta Dinh S, et al. Brain dysfunction in chronic pain patients assessed by resting-state electroencephalography. Pain. 2019;160(12):2751–65.CrossRefPubMedPubMedCentral
23.
Hu L, Iannetti GD. Neural indicators of perceptual variability of pain across species. Proc Natl Acad Sci U S A. 2019;116(5):1782–91.CrossRefPubMedPubMedCentral
24.
Nagabhushan Kalburgi S, et al. MICROSTATELAB: the EEGLAB toolbox for resting-state microstate analysis. Brain Topogr. 2024;37(4):621–45.CrossRefPubMed
25.
Lehmann D, Ozaki H, Pal I. EEG alpha map series: brain micro-states by space-oriented adaptive segmentation. Electroencephalogr Clin Neurophysiol. 1987;67(3):271–88.CrossRefPubMed
26.
Michel CM, Koenig T. EEG microstates as a tool for studying the temporal dynamics of whole-brain neuronal networks: A review. Neuroimage. 2018;180(Pt B):577–93.CrossRefPubMed
27.
Tarailis P, Koenig T, Michel CM, Griskova-Bulanova I. The functional aspects of resting EEG microstates: a systematic review. Brain Topogr. 2024;37(2):181–217.CrossRefPubMed
28.
Qiu S, et al. Temporal dynamics of electroencephalographic microstates during sustained pain. Cereb Cortex. 2023;33(13):8594–604.CrossRefPubMed
29.
Li Y, et al. Preoperative resting-state microstate as a marker for chronic pain after breast cancer surgery. Brain Behav. 2023;13(10):e3196.CrossRefPubMedPubMedCentral
30.
Vanneste S, To WT, De Ridder D. Tinnitus and neuropathic pain share a common neural substrate in the form of specific brain connectivity and microstate profiles. Prog Neuropsychopharmacol Biol Psychiatry. 2019;88:388–400.CrossRefPubMed
31.
Gonzalez-Villar AJ, Trinanes Y, Gomez-Perretta C, Carrillo-de-la-Pena MT. Patients with fibromyalgia show increased beta connectivity across distant networks and microstates alterations in resting-state electroencephalogram. Neuroimage. 2020;223:117266.CrossRefPubMed
32.
Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134(1):9–21.CrossRefPubMed
33.
Luo H, et al. An electroencephalography profile of paroxysmal kinesigenic dyskinesia. Adv Sci (Weinh). 2024;11(12):e2306321.CrossRefPubMed
34.
Koenig T, et al. EEG-Meta-Microstates: Towards a more objective use of resting-state EEG microstate findings across studies. Brain Topogr. 2024;37(2):218–31.CrossRefPubMed
35.
Lehmann D, Faber PL, Galderisi S, Herrmann WM, Kinoshita T, Koukkou M, et al. EEG microstate duration and syntax in acute, medication-naive, first-episode schizophrenia: a multicenter study. Psychiatry Res. 2005;138(2):141–56.CrossRefPubMed
36.
Tomescu MI, et al. Spontaneous thought and microstate activity modulation by social imitation. Neuroimage. 2022;249:118878.CrossRefPubMed
37.
Schrepf A, et al. Somatic awareness and tender points in a community sample. J Pain. 2016;17(12):1281–90.CrossRefPubMedPubMedCentral
38.
Si X, et al. The temporal dynamics of EEG microstate reveals the neuromodulation effect of acupuncture with Deqi. Front Neurosci. 2021;15:715512.CrossRefPubMedPubMedCentral
Galambos A, et al. A systematic review of structural and functional MRI studies on pain catastrophizing. J Pain Res. 2019;12:1155–78.CrossRefPubMedPubMedCentral
41.
Custo A, et al. Electroencephalographic resting-state networks: source localization of microstates. Brain Connect. 2017;7(10):671–82.CrossRefPubMedPubMedCentral
Mit Hilfe der PSMA-PET/CT lassen sich Patienten mit Prostatakarzinomen identifizieren, bei denen eine lokale Therapie möglich ist und die z.B. von einer stereotaktischen Radiotherapie profitieren könnten. Die aktuelle Evidenz dazu wurde auf dem DGHO-Kongress vorgestellt.
IDH (Isocitrat-Dehydrogenase)-Mutationen sind bei Gliomen im Erwachsenenalter ein wichtiger prognostischer und prädiktiver Faktor und zugleich ein ideales Target für eine zielgerichtete Therapie. Mittlerweile gibt es mit Vorasidenib eine zugelassene Behandlungsoption.
Die Prognose von Patientinnen und Patienten mit nicht-kleinzelligem Lungenkarzinom (NSCLC) hat sich durch die Einführung von Immuntherapien enorm verbessert. Dennoch erreicht nur ein Teil der Betroffenen eine langfristige und stabile Krankheitsremission. Das hat verschiedene Gründe.
Chronische Schmerzen und therapieresistent: Botulinumtoxin A könnte dafür eine Lösung sein. Wie der Wirkstoff in der Orthopädie eingesetzt wird, welche Evidenz dafür spricht und wie es um die Kostenübernahme steht, erklärte Dr. Stephan Grüner auf dem Kongress für Orthopädie und Unfallchirurgie.
