The prevalence of chronic neuropathic pain (CNP) in the general population is estimated to be 7%–10% and is expected to rise with shift in demographic trend towards ageing population, increased prevalence of diabetes and increased survival from cancer following chemotherapy [1-4]. It can result in spectrum of complications from psychological to somatic disorders [5].

Neuronal maladaptive mechanisms triggered by neuroinflammatory processes following nerve injuries play a pivotal role in the onset of CNP. These aberrant neural activities extend beyond the pain pathway, encompassing the central autonomic network [6]. Disruption of brain-heart interaction, essential for allostasis, is a characteristic feature of chronic pain.

Spectral analysis of electroencephalography (EEG) in CNP patients has revealed increased power within the 2–25 Hz frequency range and a slowing of the dominant frequency. These EEG spectral changes have been attributed to the phenomenon of thalamo-cortical dysrhythmia, characterized by excessive and widespread slow rhythms in the awake brain, commonly observed in CNP [7-9]. Heart rate variability (HRV) refers to the variation in time intervals between successive heartbeats. It is a measure of the dynamic interplay between the sympathetic and parasympathetic branches of the autonomic nervous system (ANS), reflecting the adaptability and regulation of the cardiovascular system. HRV analysis has emerged as a valuable electrophysiological tool for assessing neuropathic pain [10-13]. Vagal dysfunction, a hallmark of CNP, was further underscored in a study demonstrating enhanced vagal tone, as indicated by time domain parameters NN50 and pNN50, alongside pain relief following central neuromodulation techniques [6].

Considering that pharmacological treatments alone yield favorable outcomes in only 40%–60% of cases due to the condition's multifaceted pathophysiology [14], there is a growing need for non-pharmaceutical management strategies. Complementary and alternative medicine (CAM) modalities, including chiropractic care, acupuncture, hypnosis, yoga, biofeedback, and transcranial direct current stimulation, have shown promise in CNP treatment, despite a lack of comprehensive scientific validation regarding their mechanisms of action [14-16]. To advance towards a multimodal approach in CNP management, it is imperative to gain a deeper understanding of its complex underlying mechanisms.

Quantitative electroencephalography (qEEG) is a non-invasive technique that quantifies the electrical activity of the brain by analyzing the frequency and amplitude characteristics of electroencephalogram signals. Exploring the intricate relationship between HRV and qEEG in CNP patients can offer insights into associated comorbidities and facilitate personalized treatment planning and appropriate follow-up strategies. Despite the proposition that maladaptive neuronal mechanisms in CNP affect both the pain network and the central autonomic network, there remains a dearth of research exploring the correlation between these networks. Therefore, the primary objective of this study was to investigate potential correlations between HRV parameters and spectral EEG analysis in CNP patients and subsequently compare these findings with a healthy control group.

Participant population

The study commenced after securing approvals from the Post Graduate Research Monitoring Committee (PGRMC/ 26.04.2017/14) and the Institute Ethics Committee (JIP/IEC/2017/0169) at Jawaharlal Institute of Postgraduate Medical Education and Research (JIPMER), Puducherry. The sample size for each group in this study was determined to be 26, considering the exploratory nature of this investigation as a pilot study. The calculation was based on [17] an alpha error of 0.05, a beta error of 80% (power of 0.80), and an expected correlation coefficient of 0.5. Given the specialized analysis involving HRV and qEEG in CNP patients, this sample size was deemed appropriate for initial observations and insights. It was a cross-sectional comparative study comparing CNP patients with age and sex matched normal controls. Patients diagnosed as neuropathic pain of non-malignant etiology for more than 3 months and aged between 30–50 years were recruited for the study after obtaining written informed consent. Patients or normal controls with history of 1) organic diseases of brain, 2) neurosurgery, 3) treatment for psychiatric disorders, 4) substance use disorder with narcotics, 5) treatment with old generation anti-epileptics, 6) taking more than two analgesics, 7) hypertension or coronary artery disease or cardiac failure or arrhythmias were excluded from the study. Neuropathic pain patients who can withhold the drugs with proper consultation from the physician for one day prior to the recording were recruited from the pain clinic.

Study procedure

Study participants were advised to abstain from nicotine and alcohol, 24 h prior to recording and were also instructed not to consume tea or coffee on the day of recording. Participants were instructed to shampoo their hair in the morning prior to recording. All recordings were done in the forenoon (between 9 and 11AM) to avoid the influence of diurnal variation. EEG recording was taken after 15 min of rest with the participant in semi-reclined posture doing spontaneous breathing using 19-channel pre-wired EEG caps following the 10–20 international system of electrode placement [18] according to the recommendations of International Federation of Societies of EEG and Clinical Neurophysiology [19,20]. EEG signals were acquired using the Galileo NT- BE light machine manufactured by EB Neuro, Florence, Italy. The acquisition settings for EEG were 7 mm/µV of sensitivity, high frequency (HF) filter at 70 Hz, time constant of 0.3 sec, sampling rate at 512 Hz and notch filter enabled for eliminating 50 Hz electrical line noise. Simultaneous Lead II ECG was recorded using the designated ECG channel of Galileo NT- BE light acquisition device. ECG recordings were taken with sensitivity at 100 mm/µV, low pass filter at 70 Hz, high pass filter at 0.3 Hz, sampling rate at 512 Hz and notch filter enabled for eliminating 50 Hz electrical line noise. The recording lasted for 10 min, in which the participants were asked to relax and keep their eyes open focusing at a distant point for the initial 2.5 min. Subsequently, the participants were asked to close their eyes and remain awake for the next 2.5 min. The same cycle was repeated for the second half of the recording and this was done to ensure that the participants didn’t sleep through the recording.

Data analysis

Spectral analysis of the EEG was conducted offline using Brainstorm version 2.0, a software that operates on the MATLAB Runtime Compiler version 2.0 platform. Once the EEG data were imported into Brainstorm, a bandpass filter (1–40 Hz) was applied to remove muscle artifacts that could interfere with the analysis. From the cleaned data, 10 epochs of 4-sec duration from the eyes-closed state were selected. To avoid selection bias, 5 epochs were chosen from the initial 2.5 min of the eyes-closed state in the first half of the recording, and the remaining 5 epochs were selected from the latter 2.5 min of the eyes-closed state in the second half of the recording. Furthermore, to minimize bias, the epochs were selected at a consistent 10-sec interval. During the selection process, if an epoch was found to have artifacts, it was excluded, and the next epoch was selected after a 10-sec interval. The power spectral analysis of the 10 selected epochs was performed using the Welch transformation (window length of 2 s with 50% overlap), and the frequencies in the output were categorized as delta (1–4 Hz), theta (5–7 Hz), alpha (8–12 Hz), and beta (13–29 Hz).

Following a manual inspection of the ECG signal in LabChart Pro software (AD Instruments), segments of 330 sec that were free from artifacts and ectopic beats were chosen for analysis of R-R interval, which represents the interval between consecutive R waves in the electrocardiogram. The measured R-R intervals were subsequently analyzed using Kubios HRV version 2.0 software (University of Eastern Finland) to estimate the fast Fourier transform (FFT) spectrum spectrum using Welch’s periodogram method with 300 s and 50% overlap, to compute the frequency domain parameters of HRV. The time domain parameters, SDNN, RMSSD, NN50, and the frequency domain parameters total power (TP) (ms²), low frequency (LF) power (ms², nu), HF power (ms², nu), and LF/HF ratio, were obtained. The frequency stratification used was as follows: LF - 0.04–0.15 Hz, HF - 0.15–0.40 Hz. LF and HF in normalized units were calculated using the formulas: LFnu (low frequency normalized units) = LF / (TP − very low frequency [VLF]) × 100 and HFnu (high frequency normalized units) = HF / (TP − VLF) × 100.

Statistical analysis

Age, in years, was presented as the mean with standard deviation, and an independent t-test was performed to verify age matching between the two groups. Sex distribution was expressed as percentages, and the Chi-square test was conducted to assess matching between the two groups. The Shapiro–Wilk test was utilized to determine the data distribution type. HRV parameters were presented as median with interquartile range, and a Mann–Whitney U-test was used to compare these parameters between the study groups. The association between HRV parameters and power spectral data of EEG was examined using partial correlation after controlling for age, sex, and body mass index (BMI). A threshold of p < 0.05 was considered statistically significant. All statistical analyses were performed at a significance level adjusted for multiple comparisons using Bonferroni correction. JASP (Version 0.17.3), R (Version 4.0.3), and RStudio (Version 1.3.1073) were employed for statistical analysis and data visualization.

Participant demographics

A comprehensive summary of the primary demographic characteristics and pain perception intensity of our study participants is presented in Table 1. Specifically, the study groups comprising CNP patients and healthy controls exhibited remarkable comparability in key demographic factors, including age (years), sex distribution, and BMI (kg/m²). This meticulous alignment underscores the methodological rigor of our study, mitigating the influence of potential confounding variables and bolstering the reliability of subsequent comparative analyses.

Table 1 . Comparison of age, sex distribution, level of pain intensity and emotional distress between chronic neuropathic pain patients and healthy controls.

Variable	Patients (n = 26)	Healthy (n = 26)	p-value
Age (yr)	48.5 ± 7.3	45.5 ± 7.5	0.151
Male (%)	9 (34.6)	12 (46.2)	0.397
Female (%)	17 (65.4)	14 (53.8)
BMI (kg/m2)	23.51 ± 2.2	22.41 ± 2.7	0.117
Pain intensity (Wong–Baker Faces Scale score)	6.2 (6)	0 (0)	0.000*
Emotional distress level General Health Questionnaire (GHQ-12)	20.5 (21)	10 (15)	0.000*

Age and body mass index (BMI) are expressed in mean ± SD. Pain intensity, emotional distress level are given in median (range). Unpaired t-test or Mann–Whitney U-test was performed based on type of data distribution. Sex is expressed in percentage and compared with fisher’s exact test. *p < 0.05 is considered statistically significant.

Comparison of HRV parameters

Comparison of HRV between CNP patients and healthy controls (Table 2) shows reduced SDNN, RMSSD, NN50 and pNN50 in time domain analysis. Similarly, the frequency domain analysis reflected the reduced HRV in LF, HF and TP in the patients. Although it also showed a decrease in HFnu and an increase in LFnu and LF/HF ratio in CNP patients, these differences did not reach statistical significance.

Table 2 . Comparison of heart rate variability between chronic neuropathic pain patients and healthy controls.

Variable	Patients (n = 26)	Healthy (n = 26)	Mann–Whitney U	p-value*
SDNN (ms)	20.1 (10.78)	48.2 (16.0)	36.5	0.000*
RMSSD (ms)	15.15 (12.08)	41.55 (17.7)	40	0.000*
NN50 (count)	1.5 (4)	72.5 (60.5)	22.5	0.000*
pNN50 (%)	0.3 (1.03)	22.5 (21.45)	25	0.000*
TP (ms2)	419.5 (502.75)	2,095.5 (1,653)	44	0.000*
HF (ms2)	58 (148)	532 (607)	43	0.000*
HFnu	46.33 (39.16)	62.24 (35.32)	248	0.1
LF (ms2)	82 (138)	328.5 (638.5)	86.5	0.000*
LFnu	53.68 (39.16)	37.77 (35.32)	248	0.1
LF/HF	1.17 (1.86)	0.61 (1.07)	245.5	0.09

All values are expressed as median (interquartile range). SDNN, standard deviation of NN intervals (ms); RMSSD, square root of mean squared differences of successive NN intervals (ms); NN50, number of pairs of adjacent NN intervals differing by more than 50 ms (count); TP, total power (ms²); LF, low frequency (ms²); HF, high frequency (ms²); VLF, very low frequency. LF power in normalized units [LF / (TP − VLF) × 100]; HF power in normalized units [HF / (TP − VLF) × 100]. Comparison between the groups was done by Mann–Whitney U-test. *p < 0.05 is considered statistically significant.

Power spectral density (PSD) comparison of qEEG in the four frequency bands

Table 3 displays a comparison of the group averages of PSD in the delta frequency band between the chronic pain group and healthy controls. Delta frequency power was significantly higher in the chronic pain group across all montages except for the Fp1 electrode position.

Table 3 . Comparison of power spectral density in delta frequency between chronic neuropathic pain patients and healthy controls.

Variable	Pain group PSD (median)	Healthy group PSD (median)	Mann–Whitney U	p-value
delta_Fp1	2.15 (2.09)	1.318 (1.87)	236	0.062
delta_Fp2	2.195 (3.16)	1.176 (0.86)	178	0.003*
delta_F7	4.143 (2.22)	3.207 (2.46)	220	0.031*
delta_F3	1.289 (0.8)	0.789 (0.61)	221	0.032*
delta_F4	1.374 (1.33)	0.773 (0.61)	178	0.003*
delta_F8	3.566 (2.26)	2.037 (1.82)	150	0.001*
delta_T3	4.852 (3.8)	2.764 (1.86)	146	0.000*
delta_C3	3.305 (2.39)	1.763 (1.2)	166	0.002*
delta_C4	3.52 (2.73)	1.911 (0.76)	173	0.003*
delta_T4	4.55 (2.99)	2.665 (1.62)	135	0.000*
delta_T5	6.923 (3.82)	4.133 (2.83)	166	0.002*
delta_P3	5.628 (3.95)	2.95 (1.49)	154	0.001*
delta_P4	5.316 (3.4)	3.063 (2)	128	0.000*
delta_T6	6.649 (3.32)	3.966 (2.59)	138	0.000*
delta_O1	7.886 (6.72)	4.732 (2.97)	137	0.000*
delta_O2	7.592 (6.27)	5.387 (4.17)	196	0.009*
delta_Fz	1.055 (0.92)	0.578 (0.64)	220	0.031*
delta_Cz	3.773 (2.49)	2.044 (1.35)	180	0.004*
delta_Pz	5.893 (4.16)	3.026 (1.78)	136	0.000*

All values are expressed in median and interquartile range. Power spectral density (PSD) of delta frequency is expressed in µV²/Hz over 19 montages of 10–20 system. Comparison between the groups was done by Mann–Whitney U-test. *p < 0.05 is considered statistically significant.

The comparison of PSD in the theta frequency band between the chronic pain group and healthy controls is presented in Table 4. A statistically significant increase in theta power was observed in the patient group across all montages except for the Fp1 and F3 electrode positions in the international 10–20 system.

Table 4 . Comparison of power spectral density of in theta frequency between chronic neuropathic pain patients and healthy controls.

Variable	Pain group PSD (median)	Healthy group PSD (median)	Mann–Whitney U	p-value
theta_Fp1	0.55 (0.25)	0.451 (0.44)	245	0.089
theta_Fp2	0.675 (0.42)	0.36 (0.31)	173	0.003*
theta_F7	1.64 (1.1)	1.045 (0.95)	198	0.010*
theta_F3	0.506 (0.32)	0.444 (0.3)	248	0.100
theta_F4	0.576 (0.39)	0.438 (0.33)	220	0.031*
theta_F8	1.506 (0.91)	0.89 (1.05)	178	0.003*
theta_T3	2.608 (1.42)	1.22 (1.56)	174	0.003*
theta_C3	1.525 (0.87)	0.847 (0.98)	196	0.009*
theta_C4	1.554 (1.36)	0.929 (0.93)	188	0.006*
theta_T4	2.219 (1.24)	1.319 (1.65)	188	0.006*
theta_T5	3.499 (3.29)	2.045 (2.07)	184	0.005*
theta_P3	2.632 (1.6)	1.338 (2.02)	194	0.008*
theta_P4	2.639 (1.6)	1.402 (1.91)	184	0.005*
theta_T6	3.702 (2.69)	2.189 (2.76)	193	0.008*
theta_O1	4.074 (2.5)	2.251 (2.39)	192	0.008*
theta_O2	3.842 (3.38)	2.498 (2.5)	208	0.017*
theta_Fz	0.381 (0.28)	0.233 (0.17)	206	0.016*
theta_Cz	1.62 (1.21)	0.952 (1)	187	0.006*
theta_Pz	2.654 (1.4)	1.302 (1.56)	171	0.002*

All values are expressed in median and interquartile range. Power spectral density (PSD) of theta frequency is expressed in µV²/Hz over 19 montages of 10–20 system. Comparison between the groups was done by Mann–Whitney U-test. *p < 0.05 is considered statistically significant.

Table 5 presents a comparison of the group averages of PSD in the alpha frequency band between the chronic pain group and healthy controls. Although the PSD of the alpha frequency was relatively higher in the pain group, the values did not reach statistical significance when compared to healthy controls.

Table 5 . Comparison of power spectral density in alpha frequency between chronic neuropathic pain patients and healthy controls.

Variable	Pain group PSD (median)	Healthy group PSD (median)	Mann–Whitney U	p-value
alpha_Fp1	0.371 (0.26)	0.329 (0.31)	320	0.742
alpha_Fp2	0.37 (0.29)	0.275 (0.26)	241	0.076
alpha_F7	1.218 (1.05)	1.227 (1.32)	310	0.608
alpha_F3	0.295 (0.26)	0.247 (0.29)	283	0.314
alpha_F4	0.319 (0.33)	0.274 (0.22)	282	0.305
alpha_F8	0.879 (1.27)	0.944 (0.84)	279	0.280
alpha_T3	2.269 (4.57)	1.962 (3.95)	298	0.464
alpha_C3	1.316 (3.24)	1.352 (1.96)	269	0.207
alpha_C4	1.55 (5.2)	1.35 (2.08)	268	0.200
alpha_T4	2.509 (5.86)	2.441 (3.72)	284	0.323
alpha_T5	5.904 (16.42)	5.407 (9.96)	300	0.487
alpha_P3	4.866 (12.65)	3.955 (6.31)	274	0.241
alpha_P4	4.654 (15.85)	3.958 (7.39)	270	0.213
alpha_T6	8.891 (22.28)	5.636 (17.01)	287	0.351
alpha_O1	7.411 (19.13)	6.265 (13.9)	300	0.487
alpha_O2	7.139 (18.54)	6.762 (15.79)	305	0.546
alpha_Fz	0.176 (0.15)	0.168 (0.17)	290	0.380
alpha_Cz	1.162 (3)	0.944 (0.94)	259	0.148
alpha_Pz	4.2 (12.18)	3.79 (5.74)	271	0.220

All values are expressed in median and interquartile range. Power spectral density (PSD) of alpha frequency is expressed in µV²/Hz over 19 montages of 10–20 system. Comparison between the groups was done by Mann–Whitney U-test. p < 0.05 is considered statistically significant.

The comparison of PSD in the beta frequency band between the chronic pain group and healthy controls is shown in Table 6. Beta frequency was significantly higher in patients with chronic pain across electrode positions, except for the frontal electrodes (Fp1, Fp2, F3, F4, Fz) and central electrodes (C3, C4, Cz).

Table 6 . Comparison of power spectral density of in beta frequency between chronic neuropathic pain patients and healthy controls.

Variable	Pain group PSD (median)	Healthy group PSD (median)	Mann–Whitney U	p-value
beta_Fp1	0.198 (0.58)	0.171 (0.18)	310	0.608
beta_Fp2	0.159 (0.49)	0.114 (0.21)	281	0.297
beta_F7	0.294 (0.3)	0.165 (0.13)	178	0.003*
beta_F3	0.139 (0.16)	0.115 (0.08)	236	0.062
beta_F4	0.142 (0.12)	0.096 (0.09)	248	0.100
beta_F8	0.3 (0.18)	0.171 (0.16)	218	0.028*
beta_T3	0.433 (0.36)	0.292 (0.17)	197	0.010*
beta_C3	0.36 (0.23)	0.208 (0.24)	242	0.079
beta_C4	0.365 (0.31)	0.256 (0.31)	251	0.111
beta_T4	0.443 (0.3)	0.271 (0.2)	211	0.020*
beta_T5	0.626 (0.48)	0.425 (0.34)	211	0.020*
beta_P3	0.535 (0.44)	0.337 (0.32)	220	0.031*
beta_P4	0.511 (0.6)	0.35 (0.31)	225	0.039*
beta_T6	0.659 (0.55)	0.386 (0.44)	200	0.012*
beta_O1	0.697 (0.4)	0.421 (0.52)	213	0.022*
beta_O2	0.636 (0.46)	0.406 (0.63)	229	0.046*
beta_Fz	0.065 (0.05)	0.058 (0.04)	268	0.200
beta_Cz	0.298 (0.25)	0.189 (0.28)	246	0.092
beta_Pz	0.495 (0.53)	0.296 (0.3)	223	0.035*

All values are expressed in median and interquartile range. Power spectral density of beta frequency is expressed in µV²/Hz over 19 montages of 10–20 system. Comparison between the groups was done by Mann–Whitney U-test. *p < 0.05 is considered statistically significant.

Correlation of HRV and qEEG in patients with chronic pain and healthy controls

Fig. 1 illustrates the correlations between HRV parameters and PSD in the delta frequency band, presented separately for patients (Fig. 1 - left) and healthy controls (Fig. 1 - right). In the control group, a significant positive correlation was found between the HF band (HF in absolute power, ms²) of HRV and PSD in the delta frequency band of qEEG. These correlations reached statistical significance (p-value < 0.05) in montages F7, F4, and C3. Notably, such significant correlations were absent in patients with chronic pain.

Figure 1. The correlogram presented in this figure illustrates the partial correlation between heart rate variability (HRV) parameters and power spectral densities in the delta frequency band of quantitative electroencephalogram (qEEG) data for both the pain group (left) and the healthy control group (right). These correlations have been adjusted for age, sex, and body mass index. The intensity of color within the correlogram corresponds to the strength of the correlation, with red indicating positive correlations and blue indicating negative correlations across 19 specified EEG electrodes in 10–20 system. Only correlation coefficient values with statistical significance (p < 0.05) are displayed, while non-significant correlations are represented as blank spaces.

Fig. 2 shows the correlations between HRV parameters and PSD in the theta frequency band. In the control group, positive correlations were observed between RMSSD (in FP1 electrodes), NN50, pNN50 (in F3, F4, Fz electrodes), HF in absolute power (ms²) (in FP2, F3, F4, Fz electrodes), and HFnu (in Cz electrode) with the theta band of the qEEG power spectrum. Additionally, LFnu (in Cz electrode) displayed a negative correlation with the theta band. Conversely, among individuals with chronic pain, a positive correlation was evident between LF/HF ratio and the theta band of the qEEG (FP1, F4, T5 electrodes).

Figure 2. The correlogram presented in this figure illustrates the partial correlation between heart rate variability (HRV) parameters and power spectral densities in the theta frequency band of quantitative electroencephalogram (qEEG) data for both the pain group (left) and the healthy control group (right). These correlations have been adjusted for age, sex, and body mass index. The intensity of color within the correlogram corresponds to the strength of the correlation, with red indicating positive correlations and blue indicating negative correlations across 19 specified EEG electrodes in 10–20 system. Only correlation coefficient values with statistical significance (p < 0.05) are displayed, while non-significant correlations are represented as blank spaces.

Fig. 3 illustrates the relationships between HRV parameters and the alpha frequency bands. Among the chronic pain group (Fig. 3 - left), positive correlations between LF/HF ratio and the alpha band of qEEG were observed in all electrodes except P4, Pz, O2, Fz. Conversely, no such correlations were evident in the control group (Fig. 3 - right).

Figure 3. The correlogram presented in this figure illustrates the partial correlation between heart rate variability (HRV) parameters and power spectral densities in the alpha frequency band of quantitative electroencephalogram (qEEG) data for both the pain group (left) and the healthy control group (right). These correlations have been adjusted for age, sex, and body mass index. The intensity of color within the correlogram corresponds to the strength of the correlation, with red indicating positive correlations and blue indicating negative correlations across 19 specified EEG electrodes in 10–20 system. Only correlation coefficient values with statistical significance (p < 0.05) are displayed, while non-significant correlations are represented as blank spaces.

Fig. 4 highlights the correlations between HRV parameters and the beta frequency band. In the control group (Fig. 4 - right), negative correlations were identified between SDNN (in Cz electrode), RMSSD (in F7, T3, C4, T4, T5, P3, P4, T6, O1, O2, Cz, Pz electrodes), NN50, pNN50 (in T5, T6, P3, P4, O1, O2, Pz electrodes), HF in absolute power (ms²) (in C4, T5, T6, P3, P4, O1, O2, Pz electrodes), and the beta band of PSD. Conversely, in the chronic pain group (Fig. 4 - left), positive correlations were observed between the LF/HF ratio and the beta band of the qEEG in T4, T6, O1, O2 electrodes.

Figure 4. The correlogram presented in this figure illustrates the partial correlation between heart rate variability (HRV) parameters and power spectral densities in the beta frequency band of quantitative electroencephalogram (qEEG) data for both the pain group (left) and the healthy control group (right). These correlations have been adjusted for age, sex, and body mass index. The intensity of color within the correlogram corresponds to the strength of the correlation, with red indicating positive correlations and blue indicating negative correlations across 19 specified EEG electrodes in 10–20 system. Only correlation coefficient values with statistical significance (p < 0.05) are displayed, while non-significant correlations are represented as blank spaces.

This study aims to comprehensively examine HRV and qEEG parameters in individuals grappling with chronic pain, drawing comparisons with a demographically matched healthy control. Notably, age, sex, and BMI distribution between the two groups were comparable, as demonstrated in Table 1. It enhances the validity of the ensuing comparative analyses, minimizing potential confounding factors for understanding of the observed differences in HRV and qEEG profiles.

Comparison of HRV

In the realm of time domain and frequency domain analyses (Table 2), a comprehensive evaluation of HRV parameters revealed significant reductions in SDNN, RMSSD, NN50, pNN50, TP (ms²), HF in absolute power (ms²), and LF in absolute power (ms²) among patients grappling with chronic pain. It suggests a reduction in the resting tone of both the parasympathetic and sympathetic branches of the ANS. Patients also exhibited trends towards reduced HF in normalized units and increased LF in normalized units, along with a higher LF/HF ratio, which denote the shifting of the sympatho-vagal balance towards the sympathetic dominance. However, these trends did not reach statistical significance.

While some studies report increased levels of SDNN, NN50, HF, and LF in absolute power among patients enduring chronic pain [21-24], these differences often do not attain statistical significance. Conversely, a broader body of research consistently demonstrates statistically significant reductions in critical HRV parameters, encompassing SDNN, RMSSD, NN50, TP, HF (absolute power), and LF (absolute power), within the chronic pain population due to various etiopathogenesis [6,13,24-31]. They have also reported decreased HFnu with increased LFnu and LF/HF ratio among chronic pain patients.

The interpretation of our resting HRV analysis in individuals with chronic pain leans towards highlighting autonomic dysfunction impacting both branches of the ANS, with notable vagal dysregulation of the heart, rather than emphasizing increased sympathetic activity within the chronic pain conditions [32]. This perspective is substantiated by changes observed in LFnu and LF/HF ratio, suggesting a relative response to pronounced vagal dysfunctions.

The calculation of normalized units involves deriving values from the absolute powers of various bands in HRV. Specifically, LFnu is computed as LF / (TP − VLF) × 100, and HFnu as HF / (TP – VLF) × 100 [33]. Additionally, LFnu can be expressed as LF / (LF + HF), and HFnu as HF / (LF + HF) [34]. In cases where there is a reduction in both LF and HF bands in absolute power, the ratio may remain unaltered if the change is similar in both bands (LF, HF) of the power spectrum. However, if the reduction in the HF component surpasses that of the LF component, it results in an increase in LFnu and a decrease in HFnu. This dynamic reflects the relative change between the two components, indicative of compromised functionality in both limbs of the ANS within the patient group.

Therefore, it can be interpreted that CNP patients exhibit a reduction in overall HRV, affecting both the sympathetic and parasympathetic branches of the ANS. Furthermore, there appears to be a more pronounced dysfunction in the modulation of vagal activity within the heart in these individuals.

Comparison of qEEG

In quantitative EEG analysis, there were statistically significant increase in the PSD observed among the patients with chronic pain in the delta, theta and beta frequencies (Tables 3, 4, and 6). Though the PSD in the alpha band was higher in the patient group with chronic pain, the difference was not statistically significant (Table 5).

LF bands of qEEG (delta and theta) in chronic pain

Delta waves, the slowest recorded brain waves in humans, are typically associated with deep relaxation and restorative sleep. In infants and young children, they play a vital role in promoting rejuvenating sleep and supporting immune function. In wakeful states, the presence of heightened delta wave activity can be indicative of underlying pathology, denoting disruptions in neural processing or cognitive dysfunction [35]. In our study, excluding the FP1 electrode position, qEEG analysis in all other regions in the 10–20 system showed the significant increase in the PSD within delta band range. This heightened state of delta activity may serve as a compensatory mechanism, facilitating neural repair processes and promoting recovery.

Theta waves, known for fostering intuition, creativity, and a natural sense, also support restorative sleep and emotional connection in optimal states. However, excessive theta activity is associated with conditions such as attention deficit hyperactivity disorder and depression, contributing to symptoms like impulsivity and cognitive dysfunction [35]. The higher PSD within the theta band observed predominantly in frontal electrode positions, excluding FP1 and F3 regions in our study, raises the possibility of its association with underlying physiological or pathological processes in chronic pain.

Except few studies [36,37] conducted in pain research involving pain-inducing conditions and chronic psychological pain disorders, our findings are consistent with many studies that have reported increased power spectral densities in the low-frequency bands (delta and theta) [7,9,38-45].

HF bands of qEEG (alpha and beta) in chronic pain

Alpha waves, characterized by their frequency range between beta and theta, facilitate relaxation and calmness as needed. They are prominently observed during states of daydreaming, relaxation, and an inability to focus [35]. Consistent with prior findings [9,36,39,46,47], the comparison of high-frequency bands in spectral analysis of EEG between neuropathic pain patients and normal controls revealed an increased level of PSD in the alpha band. However, this difference lacked statistical significance across electrode positions.

Beta waves, prevalent during wakefulness, facilitate conscious thought and logical reasoning, often inducing stimulation. Excessive prominence of beta activity correlates with heightened anxiety and arousal, hindering relaxation. In line with previous investigations [9,36,39,41,48,49], the comparison of high-frequency bands in spectral analysis of EEG between neuropathic pain patients and normal controls unveiled an elevated PSD in the beta band. This difference reached statistical significance in numerous electrodes within the beta band range, except for frontal (FP1, FP2, F3, F4, Fz) and central (C3, C4, Cz) positions.

Sarnthein et al. [39] provided conclusive evidence for the presence of higher spectral power in the 2–25 Hz frequency range and slowing of the dominant rhythm in neuropathic pain patients. They reaffirmed their findings by following up a subgroup of patients who underwent central lateral thalamotomy for pain relief, and in them, they were able to demonstrate a decrease in power in the theta band along with pain relief. They referred to these excess continuous and widespread slow rhythms occurring in the awake brain of neuropathic pain patients as thalamo-cortical dysrhythmia. The presence of thalamo-cortical dysrhythmia and its resolution accompanied by pain relief following central lateral thalamotomy have been stated by Jeanmonod et al. [50] as well, in a study involving neuropathic pain patients.

Correlation between HRV and qEEG

In the current study we have also made an attempt at finding how the neuronal maladaptiveness in the central autonomic network of CNP patients (manifested as reduced overall HRV and decreased vagal modulation of heart rate) was related with their EEG power spectral findings. In this regard, partial correlation of various HRV parameters with the spectral power of delta, theta, alpha and beta bands of EEG (Fig. 1-4) were done in both groups after controlling for age, sex and BMI parameters.

Vagal (parasympathetic) correlation patterns

In the control group, we observed positive correlations between vagal indices derived from HRV analysis and qEEG parameters primarily in the delta and theta frequency bands. These correlations suggested that lower frequency qEEG activity, particularly in delta and theta bands (Figs. 1 and 2), positively correlated to vagal tone, as evidenced by significant correlation with HF (ms²), NN50, pNN50. This pattern was notable in frontal electrode positions, indicating a potential role of prefrontal cortical activity in vagal modulation.

Conversely, negative correlations between vagal indices and beta wave activity were observed, indicating that higher frequency beta waves, associated with active engagement and focused attention, were negatively associated (Fig. 4) with vagal activity (HF (ms²), NN50, pNN50) at rest. This pattern was particularly evident in posterior EEG leads, suggesting a modulation of vagal tone by posterior cortical activity during restful states.

Loss of vagal association in chronic pain group

However, in the chronic pain group, these correlations between vagal indices and qEEG bands were absent. This indicates a disruption in the association between vagal activity and qEEG parameters in chronic pain individuals, suggesting a loss of the normal modulation of vagal tone by brain activity.

Sympathetic correlation patterns

In healthy control group, except Cz position, where LFnu an index of sympathetic activity is negatively associated with theta band (HF in normalized units, the vagal index is positively correlated to theta band at same Cz position) (Fig. 2), no sympathetic indices were correlated to any of the four bands of qEEG (Fig. 1-4). It is suggesting that cortical activity measured through qEEG at rest are predominantly related to vagal modulation compared to sympathetic regulation.

In contrast to the finding from parasympathetic system, sympathetic activity, as mainly denoted by increased LF/HF ratio (sympatho-vagal balance, more towards sympathetic component), showed positive correlations with theta, alpha, and beta bands (Fig. 2-4) in the chronic pain group. This suggests that theta, alpha, and beta bands, are associated with sympathetic activation in chronic pain conditions.

Overall, our findings suggest a disruption in the association between vagal activity and qEEG parameters in chronic pain individuals, while sympathetic activity is positively correlated with alpha qEEG bands (Fig. 3). This highlights the potential dysregulation of autonomic balance and the loss of normal brain-heart interaction thorough parasympathetic system in chronic pain conditions, underscoring the importance of understanding these physiological mechanisms for targeted therapeutic interventions.

This may prove useful in targeting the treatment of neuropathic pain through CAM like biofeedback, transcranial direct current stimulation chiropractic, acupuncture, hypnosis, and yoga.

Although HRV has been recognized as a potential screening tool for identifying neuropathic pain, identification of autonomic dysfunction related to neuropathic pain from HRV becomes obscured in patients known to have heart diseases or in those on drugs that affect the heart. Therefore, in the current clinical settings where cardiac diseases are not uncommon, the level of correlation between HRV and qEEG may become an equally effective screening tool in identifying neuronal maladaptiveness and disturbance of heart brain interaction at early stages in CNP and may help in deciding the adjunct and non-pharmaceutical management for improving their quality of life.

In conclusion, this study investigates the intricate interplay of HRV and qEEG parameters in individuals coping with chronic pain, comparing them with demographically matched healthy controls. The comparison reveals significant alterations in both HRV and qEEG profiles among chronic pain patients, highlighting the multifaceted nature of their condition. Notably, the HRV analysis indicates a noteworthy reduction in the resting tone of both the parasympathetic and sympathetic branches of the ANS in chronic pain patients. Moreover, qEEG analysis demonstrates significant alterations in PSD, particularly in delta, theta, and beta frequencies. However, the alpha band, though elevated, did not reach statistical significance. The correlation analyses between HRV and qEEG shed light on intriguing differences between healthy controls and chronic pain patients, revealing potential disruptions in brain-heart interaction. Despite these compelling findings, this study has limitations, primarily the relatively small sample size. Future research should delve into diverse pain conditions, consider the impacts of medication, and adopt a more extensive EEG channel configuration to enhance our understanding of these critical physiological parameters in the context of chronic pain.

Limitations of the study

While this study sheds light on the intricate relationship between HRV, qEEG, and persistent pain, it is not without limitations. The relatively small sample size could limit the generalizability of our findings to a broader population. Additionally, the cross-sectional nature of this study provides a snapshot of HRV and qEEG profiles at a specific point in time, necessitating longitudinal studies to better understand the dynamics of these parameters over time. The study primarily focuses on CNP, providing opportunities for future research to explore HRV and qEEG in other types of persistent pain conditions. Moreover, the study does not delve into the impact of various medications on HRV and qEEG, which could be a critical area for investigation. Future studies should also consider incorporating a more diverse range of pain assessment tools and explore how psychological factors might interact with HRV and qEEG. Furthermore, EEG recordings with more channels could have provided greater spatial resolution, suggesting a potential direction for improving future investigations. Overall, further research with larger and diverse cohorts, coupled with a multifaceted approach, is needed to unravel the complexity of autonomic and central nervous system involvement in persistent pain.

We extend our sincere gratitude to the study participants for their valuable contributions.

This study was supported by the intra mural research grant (JIP/Res/Intramural/phs1/2017-18) from Jawaharlal Institute Post-graduate Medical Education Research, Puducherry.

The authors declare no conflicts of interest.