Reduced Thalamic γ-Aminobutyric Acid (GABA) in Painless but Not Painful Diabetic Peripheral Neuropathy

Diabetic peripheral neuropathy (DPN) is the commonest complication of diabetes mellitus, affecting around 50% of all people with diabetes, and is associated with significant morbidity and mortality (1). DPN is the predominant initiating risk factor for foot ulceration, which frequently leads to lower-leg amputation. People with DPN may suffer with neuropathic pain (painful DPN) or may have no pain (painless DPN). Painful DPN can result in sleep impairment, mood disorders, and a reduced quality of life. Although recent developments have improved our understanding of the pathophysiology painful DPN, the underlying mechanisms remain poorly understood (1), and hence the lack of effective treatments (2).

There is a considerable body of research that has demonstrated central nervous system (CNS) involvement in DPN. These studies have identified structural CNS changes associated with DPN, including a reduction in spinal cord cross sectional area (3); reduced cortical thickness/gray matter volume in regions of the brain associated with somatosensory function (4); and functional changes including higher sensory cortical energy metabolism (5–7). One important region of the brain involved in pain pathways is the thalamus, which receives and modulates ascending sensory signals from the peripheral nervous system before transmitting sensory information to the rest of the brain regions involved in pain processing and somatosensory function. Our group and others have demonstrated altered thalamic structure, function, and blood flow in painful compared with painless DPN (4–6). In painless DPN, there is evidence of neuronal dysfunction, reduced thalamic volume, and impaired blood flow, whereas, in contrast, painful DPN is associated with a preservation of functional and structural parameters, and hyperdynamic vascular supply (5,6,8). Thus, although recent studies have given us an insight into cerebral structural, functional, and microvascular alterations in DPN, the molecular mechanisms underpinning these findings are still lacking.

An important mechanism involved in maintenance of neuropathic pain is central sensitization, mediated by imbalance of neurotransmitters such as the excitatory neurotransmitter glutamate and inhibitory neurotransmitter γ-aminobutyric acid (GABA), which both may be detected using MRS. GABA regulates transmission of nociceptive signals and inhibits release of pronociceptive neurotransmitters, such as glutamate, which mediates synaptic transmission of pain sensation (9). Rodent models of chronic pain demonstrate that reduced GABA and increased glutamate activity contribute to central sensitization in chronic pain. However, human studies suggest that underlying neurometabolite levels may be unique in different pain conditions (10).

In this study, we used MRS to investigate the role of GABA and glutamate neurotransmitter levels within the thalamus, in a cohort that includes both painless and painful DPN. We hypothesize that GABA levels will be reduced and glutamate levels increased in painful DPN, with the converse seen in painless DPN.

Fifty-nine participants who were consecutive patients with type 2 diabetes (T2D) (n = 44) were recruited from the Sheffield Teaching Hospitals NHS Foundation Trust outpatient clinics. Healthy volunteers (HV) (n = 15) were recruited from promotional material within the hospital and served as a control group. Participants with T2D were diagnosed according to World Health Organization criteria (11). Exclusion criteria included nondiabetic neuropathies, history of alcohol consumption of more than 24 units a week, diabetic neuropathies other than DPN, neurological disorders that may confound radiological or clinical assessments, or other factors that would preclude MRI (e.g., pacemaker and claustrophobia), pregnancy, major lower-limb amputation, antidepressant medication use, recurrent severe hypoglycemia, concurrent severe psychological/psychiatric conditions, and moderate-to-severe pain from causes other than DPN. All participants gave written, informed consent before their participation in the study, which had prior ethics approval by the Sheffield Research Ethics Committee (Sheffield, U.K.).

Participants underwent detailed clinical and neurophysiological assessments; full details are available within the Supplementary Material. In brief, this included evaluation of neuropathic pain symptoms and signs (Doleur Neuropathique en 4); assessment of clinical peripheral neurological status (Neuropathy Impairment Score of the Lower Limb [NIS(LL)]; Cardiac Autonomic Function Tests using the O’Brien protocol (12); nerve conduction studies; quantitative sensory testing (13); and skin biopsy with assessment of intraepidermal nerve fiber density (IENFD), using methods as previously described (14). The neuropathy composite score of NIS(LL)+7 was obtained by combining the NIS(LL) plus seven tests of nerve function (15). This is a validated, composite measure of neuropathy severity.

On the basis of clinical and neurophysiological assessments, participants with T2D were divided into three groups according to the Toronto consensus recommendations (16): 1) no DPN—patients with normal clinical and neurophysiological assessments; 2) painless DPN—patients with confirmed DPN and no neuropathic pain; and 3) painful DPN—patients with painful neuropathic symptoms involving the feet and/or legs (Doleur Neuropathique en 4 ≥ 4) in a distal symmetrical pattern with confirmed DPN. Neuropathic pain severity was measured using a 0–10 point score using a visual analog scale.

Participants did not take neuropathic pain medications 48 h before the scan. MR data were acquired at 3.0 Tesla (Ingenia 3.0T; Philips Healthcare, Best, the Netherlands). Anatomical images were acquired using a T1-weighted magnetization-prepared rapid acquisition gradient-echo sequence: repetition time 7.2 ms, echo time 3.2 ms, flip angle 8°, and voxel size 0.9 mm³. MRS data were obtained from a 1-cm³ voxel placed over the left thalamus. GABA-edited Mescher–Garwood point-resolved spectroscopy (17) data were acquired with the following parameters: echo time 68 ms; repetition time 1.8 s; spectral width 2 kHz; frequency selective editing pulses (15 ms) applied at 1.9 ppm and 7.46 ppm. The water signal was minimized using a chemical shift selective radiofrequency pulse. Spectroscopic postacquisition data analysis was performed using Gannet (18). Relative resonance areas from GABA and glutamate relative to unsuppressed water (GABA:H₂O and glutamate:H₂O) and GABA relative to glutamate (glutamate:GABA) were calculated and used for further statistical comparative analysis.

Analyses were performed using SPSS Version 28.0 (SPSS; IBM Corporation, Armonk, NY). Continuous data that are parametric were described as means and SD (±); continuous nonparametric data were described as median and interquartile range (IQR); and categorical variables were described as number and percentage. Group differences were compared using one-way ANOVA (normally distributed), χ2 test (categorical variables), and Kruskal-Wallis rank test (nonnormally distributed) as appropriate. Post hoc analysis was performed using least significant analysis for normally distributed data and Mann-Whitney U test for nonnormally distributed data. ANCOVA was performed to examine differences in mean GABA:H₂O between the study groups using age, BMI, and sex as covariates. Relationships between MRS and clinical variables were assessed using Spearman’s correlation.

The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Table 1 shows demographic, clinical, and metabolic variables in participants. There was no significant difference in age between the groups or in HbA_1c between diabetes groups, although participants with no DPN (P = 0.046, 95% CI −7.96, −0.07) and painful DPN (P = 0.005, 95% CI 1.89, 9.92) had a significantly greater BMI compared with HV. Duration of diabetes was longer in painful DPN compared with no DPN (Mann-Whitney U, P = 0.017). Total cholesterol was lower in painless DPN compared with no DPN (Mann-Whitney U, P < 0.01) and painful DPN (P = 0.017). There was a significant difference in retinopathy status between no DPN and painful DPN (P = 0.04) and painless DPN (P = 0.03). There was no difference in retinopathy status between the two DPN groups (P = 0.438). Urine albumin-to-creatinine ratio was significantly higher in painful DPN (2.1 [IQR 31.5]) compared with no DPN (Mann-Whitney U, P < 0.01). The mean pain score (visual analog scale) was 6.8 (±2.7) in participants with painful DPN.

Table 2 shows neurophysiological, skin biopsy, and MRS results. As expected, DPN groups had significantly higher NIS(LL)+7 scores (Kruskal-Wallis, P < 0.01) and lower IENFD (Kruskal-Wallis, P < 0.01). Other neurophysiological measures also indicated greater neurological impairment in the DPN groups, other than tibial latency and pressure pain threshold. The NISLL+7 was significantly higher in painful compared with the painless DPN group (Mann Whitney U test, P = 0.02), and cold detection threshold was significantly lower (Mann Whitney U, P = 0.02), but both DPN groups had similar IENFD (P = 0.72).

The DPN group (painful and painless DPN) had a significantly lower GABA:H₂O ratio (1.53 ± 0.2, ANOVA, P < 0.01) compared with that of HV (1.8 ± 0.2, P = 0.029, 95% CI 0.02, 0.41) and no DPN groups (1.8 ± 0.4, P < 0.01, 95% CI 0.11, 0.49) (Fig. 1A). When examining each group separately, there was a significant group effect (ANOVA, P < 0.01) (Table 2 and Fig. 2A). Post hoc analysis showed GABA:H₂O in painless DPN was significantly lower compared with HV (P = 0.01, 95% CI −0.50, −0.06) and no DPN (P < 0.01, 95% CI −0.58, −0.14). GABA:H₂O was also significantly lower in painful DPN compared with no DPN (P = 0.047, 95% CI −0.45, −0.003). There was no statistically significant difference in GABA:H₂O between painful and painless DPN (P = 0.22), however. GABA:H₂O group analysis remained significant once adjusted for age, BMI, and sex (ANCOVA, P = 0.047). On adjusted post hoc analysis, the GABA:H₂O for painless DPN remained statistically different from no DPN (P < 0.01, 95% CI −0.89, 0.570) but not HV; also, the difference between painful and no DPN was no longer significant. There were no group differences in other MRS ratios in the combined group (glutamate:H₂O, ANOVA, P = 0.271) (Fig. 1B) (glutamate:GABA, P = 0.257) or separate group analysis (ANOVA, glutamate:H₂O, P = 0.287) (Fig. 2B) (glutamate:GABA, P = 0.149). There were no sex differences in thalamic GABA:H₂O levels (P = 0.304).

There was a significant negative correlation between GABA:H₂O and neuropathy parameters, including NISLL+7 (r −0.373, P = 0.04), IENFD (r 0.366, P = 0.005), vibration detection threshold (r 0.311, P = 0.016), and warm detection threshold (r −0.344, P = 0.008). There was also a correlation between duration of diabetes (r −0.358, P = 0.008).

Using MRS, we examined alterations in the neurotransmitters GABA and glutamate within the thalamus in well-phenotyped patients with DPN. The main findings of the study are that there is a reduction in GABA:H₂O in patients with DPN (painful and painless DPN combined) compared with those without neuropathy (no DPN and HV combined). Moreover, patients with painless DPN had the lowest mean GABA:H₂O compared with no DPN and HV, whereas levels of painful DPN were not different between other groups, after adjustment for age, BMI, and sex. Additionally, there was significant correlation between GABA:H₂O and measures of both large and small fiber nerve function. There was no difference in glutamate ratio levels among the study groups.

The results of our study are novel, as they investigate the levels of GABA and glutamate associated with both neuropathic pain in painful DPN and a reduction in somatosensory function in painless DPN. To our knowledge, this is the first study to examine GABA levels in people with objective impaired neuronal functional, associated with peripheral neuropathy, without concomitant neuropathic pain. However, our results agree with other studies that have shown sensorimotor GABA levels correlating with tactile discrimination/performance in autism and in healthy volunteers (19,20). The finding of reduced GABA in the thalamus in painless DPN could reflect the decrease in the number of neurons and/or reduction in neuronal function. This explanation would be consistent with previous studies demonstrating a reduction in thalamic function (6), gray matter volume (8), and blood flow in painless DPN (5). That a relationship between thalamic GABA levels and peripheral neurologic impairment exists is further strengthened by the correlation between GABA:H₂O and measures of peripheral neuropathy severity.

The study findings demonstrate that GABA:H₂O levels are partially preserved in painful DPN. There are some confounding factors that might account for GABA:H₂O not being clearly different between participants with painful and painless DPN, such as those with painful DPN having a higher BMI than nonneuropathy groups and having more severe neuropathy than painless DPN. The prevailing evidence from animal models suggests impaired GABA inhibitory function and increased glutamate excitatory function in chronic pain (9). However, the cerebral levels of neurotransmitters detected using MRS in response to acute pain and also in chronic pain conditions remains uncertain (21,22). In chronic pain conditions, studies have found both a reduction and an increase in GABA (10), with a trend toward an increased GABA level, although studies investigating thalamic GABA in neuropathic pain (spinal cord injury and trigeminal neuralgia) (23,24) found a reduction in GABA. However, one other small study in DPN also found a trend toward greater thalamic GABA:H₂O in painful DPN compared with healthy volunteers (25). Our results demonstrating partially preserved GABA:H₂O in painful compared with painless DPN could reflect heightened neuronal function/activity (5) or perhaps greater inhibitory activity in an attempt to overcome increased peripheral pain impulses. Glutamate alterations were not seen in DPN, perhaps suggesting that GABA alterations play a more prominent role in the central mechanisms of the condition; indeed, other neuroimaging studies have not found alterations in cerebral glutamate (10).

There are some limitations to our study, in addition to the clinical group differences as previously mentioned. The study also only used one modality of imaging; therefore, we can only hypothesize as to the relationship with other study findings. Moreover, the patients had well-established DPN and longstanding painful DPN, and future studies will need to investigate patients earlier in the disease course. Additionally, the study was cross-sectional, and thus cannot comment on the causality of our findings. Finally, although the sample size was adequate to detect a group difference in GABA:H₂O, it may have been too small to detect differences in glutamate.

To conclude, this study, which contained a well-characterized cohort of participants, demonstrates lower levels of the inhibitory neurotransmitter GABA:H₂O ratio in the thalamus of patients with DPN, with partial preservation in painful DPN. A further prospective understanding of the cerebral neuronal excitatory/inhibitory balance inferred from MRS in other pain-processing regions of the brain may help determine the mechanistic basis of pain perception in DPN and inform future treatment approaches. Furthermore, studies need to be performed to look at the impact of interventions that increase thalamic GABA levels on the intensity of neuropathic pain.

Acknowledgments. I.D.W. is deceased.

Funding. This study was funded by Sheffield Teaching Hospitals NHS Trust, charitable fund.

The funding source had no influence in any aspect pertinent to this study.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. P.S., D.S., I.D.W., and S.T. conceived the study. P.S. and M.G. performed data collection. R.A.E. and I.D.W. designed the MRS protocol and analyzed MRS data. P.A. analyzed skin biopsy data. P.S., G.S., D.S., M.G., R.G., I.D.W., and S.T. interpreted the results, and P.S., G.S., and S.T. prepared the manuscript with input from other co-authors. S.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.