OBJECTIVE

In this study we aimed to investigate the functional connectivity of brain regions involved in sensory processing in diabetes with and without painful and painless diabetic peripheral neuropathy (DPN) and the association with peripheral nerve function and pain intensity.

RESEARCH DESIGN AND METHODS

In this cross-sectional study we used resting-state functional MRI (fMRI) to investigate functional brain connectivity of 19 individuals with type 1 diabetes and painful DPN, 19 with type 1 diabetes and painless DPN, 18 with type 1 diabetes without DPN, and 20 healthy control subjects. Seed-based connectivity analyses were performed for thalamus, postcentral gyrus, and insula, and the connectivity z scores were correlated with peripheral nerve function measurements and pain scores.

RESULTS

Overall, compared with those with painful DPN and healthy control subjects, subjects with type 1 diabetes without DPN showed hyperconnectivity between thalamus and motor areas and between postcentral gyrus and motor areas (all P ≤ 0.029). Poorer peripheral nerve functions and higher pain scores were associated with lower connectivity of the thalamus and postcentral gyrus (all P ≤ 0.043). No connectivity differences were found in insula (all P ≥ 0.071).

CONCLUSIONS

Higher functional connectivity of thalamus and postcentral gyrus appeared only in diabetes without neuropathic complications. Thalamic/postcentral gyral connectivity measures demonstrated an association with peripheral nerve functions. Based on thalamic connectivity, it was possible to group the phenotypes of type 1 diabetes with painful/painless DPN and type 1 diabetes without DPN. The results of the current study support that fMRI can be used for phenotyping, and with validation, it may contribute to early detection and prevention of neuropathic complications.

Diabetic peripheral neuropathy (DPN) is a common complication in diabetes and affects up to 50% of all individuals with diabetes. Of individuals with DPN, 15–25% also have painful DPN (14). Unfortunately, early diagnosis and proper treatment of neuropathic complications remain complex and challenging. This is mainly attributed to the limited knowledge of the underlying mechanisms of painful and painless DPN.

Alterations of the central nervous system (CNS) in diabetes, both with and without painful and painless DPN, are strongly evidenced (515). However, the impact of DPN on CNS and the impact of CNS alterations on sensory abnormalities need further exploration. We previously found structural alterations in brain areas central in sensory processing in type 1 diabetes with and without painful and painless DPN (16). The areas include thalamus, somatosensory cortex located in the postcentral gyrus, and insula (17). Brain function at rest can noninvasively be investigated using resting-state functional magnetic resonance imaging (fMRI). The technique allows evaluating functional brain connectivity between brain regions by detecting spontaneous fluctuations in the blood oxygenation level-dependent signal, which indirectly reflect neuronal activity (18).

Structural loss and functional dysfunction have been reported in the thalamus of type 1 diabetes per se (810) and type 1 diabetes with painful and painless DPN (6,13,14). The postcentral gyrus has also demonstrated alterations in DPN (6,13,19). Moreover, structural and functional changes in insula have been reported in the presence of neuropathic pain (20). A better insight into the functional alterations in sensory and pain-related brain regions might improve understanding of the underlying mechanisms of painful and painless DPN in type 1 diabetes.

In this study we aimed to investigate functional connectivity alterations in thalamus, primary somatosensory cortex, and insula in well-phenotyped cohorts of subjects with type 1 diabetes with painful DPN, type 1 diabetes with painless DPN, type 1 diabetes without DPN, and healthy control subjects. Also, we aimed to investigate the association of the connectivity alterations with peripheral nerve functions and pain measurements.

Design and Participants

This cross-sectional, observational, case-control study was a part of the clinical study named MEDON (Methods of Early Detection of diabetic peripheral Neuropathy) (16,21). Between August 2019 and April 2021, 80 participants were recruited through the outpatient clinic at the Department of Endocrinology, Steno Diabetes Center North Denmark, Aalborg University Hospital, into four groups: 20 individuals with type 1 diabetes and painful DPN, 20 individuals with type 1 diabetes screened for probable painless DPN, in accord with the Toronto consensus (1), 20 individuals with type 1 diabetes without DPN, and 20 healthy control subjects. Each participant in one group was age- and sex-matched to one participant in each of the other three groups. Thus, the participants were matched 1:1:1:1 on age (±2 years) and sex across the four groups. We ensured this by initially recruiting the least prevalent participants, those with painful DPN. Then one participant’s sex and age from the painful DPN group were used to recruit one matching participant to the DPN group, one participant to the group without DPN, and one participant to the group of healthy control subjects. While the MEDON study included type 1 diabetes with probable painless DPN based on peripheral vibration perception threshold >25 V, in the current fMRI study we additionally confirmed the presence and absence of DPN using nerve conduction study in accord with the Toronto consensus (1). Hence, the final cohort of the four groups used for the fMRI analyses is 19 individuals with type 1 diabetes and painful DPN, 19 individuals with type 1 diabetes and confirmed painless DPN, 18 individuals with type 1 diabetes without DPN, and 20 healthy control subjects. See Supplementary Fig. 1, phenotyping and clinical parameters, and results for further details.

The subjects in the diabetes groups were clinically diagnosed with type 1 diabetes. All participants were included if they were between 18 and 70 years old and excluded if they met one of the following: abnormalities in the thyroid or parathyroid metabolism, impaired liver or kidney function, history of or current alcohol abuse or drug abuse, cancer or history of chemotherapy, presence of chronic viral infections, severe skin disease, critical ischemia of the lower extremities, pregnancy, and factors that preclude MRI. All participants provided informed written consent before enrollment in the study. The study was conducted according to the Declaration of Helsinki. The North Denmark Region Committee on Health Research Ethics granted the ethics approval (N-20190003), and the study was registered with ClinicalTrials.gov (NCT04078516).

Phenotyping and Clinical Parameters

All participants with type 1 diabetes were prescreened based on their medical records. Painful DPN was clinically confirmed by two independent medical doctors and supported by a Douleur Neuropathique 4 Questions (DN4) questionnaire, a 10-item validated screening tool to identify painful DPN (22,23). Participants with a score ≥4 were classified as having painful DPN. Additionally, current pain intensity and peak and average pain intensity for the last 4 weeks were measured with a numeric rating scale (NRS) ranging from 0 to 10, where 0 represents no pain and 10 represents the worst pain ever possible (24).

Participants with painless DPN were included based on a vibration perception threshold >25 V on their first toe performed using biothesiometry. To confirm painless DPN, the participants were additionally tested for the presence of an abnormal sural nerve with a nerve conduction study conducted on the right leg with standardized skin temperature by the Department of Neurophysiology, Aalborg University Hospital. The results were processed according to the local reference values. The thresholds used for abnormalities of the sural nerve were minimum peak amplitude 6 µV, minimum conduction velocity 41 m/s, and maximum peak latency 4.2 ms. Measurements not detectable due to severe nerve damage were denoted with zero. In the current study we present only the velocity and amplitude of the sural nerve (sensory nerve) and peroneal nerve (motor nerve).

A standardized battery of quantitative sensory testing according to the protocol of the German Research Network on Neuropathic Pain (DFNS) was performed (25,26).

Information on demographic and clinical characteristics was obtained, including sex, age, diabetes duration and onset, retinopathy/nephropathy status, and CNS-acting medications. HbA1c was analyzed in blood samples, and the blood glucose level was measured before the MRI session.

MRI Acquisition

The MRI was obtained with the 3.0-Tesla GE SIGNA Premier Scanner (GE Healthcare, Milwaukee, WI) at the Department of Radiology, Aalborg University Hospital. A 48-channel head coil was used, and the head was fixed with use of foam pads. The participants were instructed to relax, to close their eyes, to avoid falling asleep, and to try not to think of anything specific during the resting-state fMRI acquisition. Gradient echo–planar images with 384 volumes were obtained with the following parameters: repetition time 1,000 ms, echo time 30 ms, flip angle 60°, matrix 64 × 64, field of view 24 cm, voxel size 3.8 × 3.8 × 3.8 mm, hyperband factor 2, and scan time 6 min and 24 s. Before the fMRI scan, a high-resolution T1-weighted three-dimensional structural scan (MPRAGE) was obtained for the anatomical registrations of the fMRI scans. The following parameters were used for this structural scan: repetition time 8.4 ms, echo time 3.6 ms, flip angle 8°, field of view 25 cm, resolution 0.78 × 0.78 mm, and slice thickness 0.80 mm.

fMRI Analyses

fMRI Preprocessing

The functional and structural images were visually inspected for artefacts. The preprocessing was executed in CONN toolbox, version 20b (https://www.nitrc.org/projects/conn) (27), running in MATLAB [version 9.7.0.1586710 (R2019b); The MathWorks, Natick, MA] (https://se.mathworks.com/products/matlab.html). The default preprocessing pipeline in CONN toolbox was used, including the following steps: motion correction; slice timing correction; Artifact Detection Tools (ART)-based outlier detection; segmentation of cerebrospinal fluid, white matter, and gray matter; and normalization to standard Montreal Neurological Institute (MNI) brain template for both structural and functional images, and smoothing with an 8-mm full-width half-maximum Gaussian kernel. The voxel size after normalizing was 2.0 × 2.0 × 2.0 mm. During denoising, the signal from white matter and cerebrospinal fluid and the 12-head motion parameters derived from spatial motion correction were added as confounders (aCompCor strategy implemented in the CONN toolbox to control for physiological and movement confounders) (2830). Data were band-pass filtered to a 0.008–0.09 Hz frequency window.

Seed-Based Functional Connectivity

The functional connectivities of the bilateral thalamus, bilateral postcentral gyrus, and bilateral insula were investigated with use of a seed-based analysis approach in CONN toolbox, also called seed-to-voxel analysis. This analysis characterizes the connectivity pattern between a predefined region of interest (ROI), also called seeds, and the rest of the brain by correlating the average of the time course signal from the voxels in one ROI with the time course of each voxel in the entire brain. As a result of this, functional connectivity maps were obtained (31). In the current study, six predefined ROIs from the CONN atlas, which covered the entire structure, were used and included: bilateral thalamus, bilateral postcentral gyrus, and bilateral insula. SPM12 (Wellcome Trust Centre for Neuroimaging, London, U.K.) (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/) was used for the seed-based group-level analyses.

For comparison of the seed-based connectivity between the groups, one-way ANOVA was used. The primary significance threshold was set to P ≤ 0.001, and cluster level corrected for multiple comparisons was presented (P < 0.05, family-wise error [FWE] corrected) with clusters ≥100 voxels (mm3).

Functional Connectivity Extraction for Correlation Analyses

A Fisher r-to-z transformation was applied between the predefined ROIs to their significantly connected clusters to extract z scores for the correlation analyses. The connected clusters were extracted from MarsBaR 0.44 toolbox (31).

Statistical Analyses

Data are presented as means ± SD or n (%) unless otherwise stated. The assumption of normal distribution was checked with Q-Q plots.

One-way ANOVA for parametric data and Kruskal-Wallis test for non-parametric data were used to determine group differences in the demographic, clinical, and peripheral assessments. χ2 or Fisher exact test was used to test for differences in sex, retinopathy/nephropathy status, and medications. Any group-wise differences were detected with post hoc analyses, and Bonferroni corrections were used to correct multiple comparisons.

Spearman correlation tests were used to detect any associations between functional connectivity and peripheral nerve functions and pain scores.

Analyses were performed in IBM SPSS Statistics (version 26.0; Armonk, NY, IBM). P < 0.05 was considered significant.

The clinical and demographic characteristics of the participants are summarized in Table 1 and also partially published elsewhere (16,21). Due to claustrophobia, one participant in the diabetes group with painful DPN and one participant in the diabetes group without DPN did not complete the MRI session. Additionally, one individual allocated to the group with diabetes and painless DPN did not have confirmed painless DPN and one individual in the group without DPN had abnormal nerve conduction study results and were therefore excluded in the analyses. Hence, the final cohort included in the statistical analyses comprised 19 subjects with type 1 diabetes with painful DPN, 19 with type 1 diabetes with painless DPN, 18 with type 1 diabetes without DPN, and 20 healthy control subjects (Supplementary Fig. 1).

Table 1

Demographic and clinical characteristics

T1DM with painful DPN (n = 19)T1DM with painless DPN (n = 19)T1DM without DPN (n = 18)Healthy control subjects (n = 20)P
Sex (n male/n female) 9/10 10/8 9/9 10/10 0.991 
Age (years) 51.4 ± 9.7 52.6 ± 9.0 50.6 ± 9.1 51.5 ± 9.2 0.923 
BMI (kg/m2) 27.3 (24.5–31.0) 27.7 (24.2–30.3) 28.0 (24.1–30.2) 24.3 (23.0–28.6) 0.318 
HbA1c (%) 8.6 ± 3.2A,B 8.9 ± 3.1A 7.9 ± 3.0B 5.3 ± 2.4C <0.001* 
HbA1c (mmol/mol) 70 ± 11A,B 74 ± 10A 63 ± 9B 34 ± 3C <0.001* 
Diabetes duration (years) 30.1 ± 14.0A,B 33.7 ± 8.1A 23.1 ± 11.0B  0.020* 
Age of diabetes onset (years) 21.3 ± 14.2 18.9 ± 10.9 27.5 ± 17.3  0.183 
Retinopathy, n (%) 18 (94.7)A 17 (89.5)A 11 (61.1)A 0B <0.001* 
Nephropathy, n (%) 5 (26.3) 3 (15.8) 0.011 
Medications for neuropathic pain, n (%)      
 Amitriptyline 3 (15.8) 0.039 
 Duloxetine 7 (36.8)A 0B 0B 0B <0.001* 
 Gabapentin/pregabalin 4 (21.1) 0.008 
 Opiates 4 (21.1) 0.008 
Other CNS-acting medications, n (%) 4 (21.1) 5 (26.3) 0.007 
Peripheral measurements      
 Vibration perception threshold 40.0 (22.0–45.0)A 45.0 (32.0–50.0)A 14.5 (11.5–18.0)B 12.0 (10.0–16.5)B <0.001* 
 Sural nerve amplitude (µV) 0.8 (0.0–3.6)A 0.0 (0.0–3.3)A 5.7 (3.1–8.0)B 10.3 (5.5–13.6)B <0.001* 
 Sural nerve velocity (m/s) 27.0 (0.0–40.0)A 0.0 (0.0–38.0)A 47.5 (44.8–50.3)B 54.5 (47.5–57.8)B <0.001* 
 Peroneal nerve amplitude (mV) 0.0 (0.0–0.0)A 0.0 (0.0–0.4)A 6.3 (3.5–10.7)B 8.5 (6.3–12.3)B <0.001* 
 Peroneal nerve velocity (m/s) 0.0 (0.0–0.0)A 0.0 (0.0–38.0)A 49.0 (44.0–52.3)B 53.0 (51.0–58.8)B <0.001* 
 Cold detection threshold (°C) 20.5 (7.4–26.5)A 12.4 (2.1–20.0)A 28.4 (26.7–30.4)B 30.1 (25.6–30.8)B <0.001* 
 Warm detection threshold (°C) 45.3 (42.2–47.5)A 44.8 (40.5–49.6)A 39.4 (35.9–43.2)B 37.5 (35.3–41.2)B <0.001* 
Neuropathic pain assessments      
 DN4 score 5.0 (4.0–6.0)A 0.0 (0.0–2.0)B 0.0 (0.0–0.0)B 0.0 (0.0–0.0)B <0.001* 
 NRS, peak pain last 4 weeks 8.0 (6.0–9.0)A 0.0 (0.0–1.0)B 0.0 (0.0–0.0)B 0.0 (0.0–0.0)B <0.001* 
 NRS, average pain last 4 weeks 5.0 (4.0–8.0)A 0.0 (0.0)B 0.0 (0.0–0.0)B 0.0 (0.0–0.0)B <0.001* 
T1DM with painful DPN (n = 19)T1DM with painless DPN (n = 19)T1DM without DPN (n = 18)Healthy control subjects (n = 20)P
Sex (n male/n female) 9/10 10/8 9/9 10/10 0.991 
Age (years) 51.4 ± 9.7 52.6 ± 9.0 50.6 ± 9.1 51.5 ± 9.2 0.923 
BMI (kg/m2) 27.3 (24.5–31.0) 27.7 (24.2–30.3) 28.0 (24.1–30.2) 24.3 (23.0–28.6) 0.318 
HbA1c (%) 8.6 ± 3.2A,B 8.9 ± 3.1A 7.9 ± 3.0B 5.3 ± 2.4C <0.001* 
HbA1c (mmol/mol) 70 ± 11A,B 74 ± 10A 63 ± 9B 34 ± 3C <0.001* 
Diabetes duration (years) 30.1 ± 14.0A,B 33.7 ± 8.1A 23.1 ± 11.0B  0.020* 
Age of diabetes onset (years) 21.3 ± 14.2 18.9 ± 10.9 27.5 ± 17.3  0.183 
Retinopathy, n (%) 18 (94.7)A 17 (89.5)A 11 (61.1)A 0B <0.001* 
Nephropathy, n (%) 5 (26.3) 3 (15.8) 0.011 
Medications for neuropathic pain, n (%)      
 Amitriptyline 3 (15.8) 0.039 
 Duloxetine 7 (36.8)A 0B 0B 0B <0.001* 
 Gabapentin/pregabalin 4 (21.1) 0.008 
 Opiates 4 (21.1) 0.008 
Other CNS-acting medications, n (%) 4 (21.1) 5 (26.3) 0.007 
Peripheral measurements      
 Vibration perception threshold 40.0 (22.0–45.0)A 45.0 (32.0–50.0)A 14.5 (11.5–18.0)B 12.0 (10.0–16.5)B <0.001* 
 Sural nerve amplitude (µV) 0.8 (0.0–3.6)A 0.0 (0.0–3.3)A 5.7 (3.1–8.0)B 10.3 (5.5–13.6)B <0.001* 
 Sural nerve velocity (m/s) 27.0 (0.0–40.0)A 0.0 (0.0–38.0)A 47.5 (44.8–50.3)B 54.5 (47.5–57.8)B <0.001* 
 Peroneal nerve amplitude (mV) 0.0 (0.0–0.0)A 0.0 (0.0–0.4)A 6.3 (3.5–10.7)B 8.5 (6.3–12.3)B <0.001* 
 Peroneal nerve velocity (m/s) 0.0 (0.0–0.0)A 0.0 (0.0–38.0)A 49.0 (44.0–52.3)B 53.0 (51.0–58.8)B <0.001* 
 Cold detection threshold (°C) 20.5 (7.4–26.5)A 12.4 (2.1–20.0)A 28.4 (26.7–30.4)B 30.1 (25.6–30.8)B <0.001* 
 Warm detection threshold (°C) 45.3 (42.2–47.5)A 44.8 (40.5–49.6)A 39.4 (35.9–43.2)B 37.5 (35.3–41.2)B <0.001* 
Neuropathic pain assessments      
 DN4 score 5.0 (4.0–6.0)A 0.0 (0.0–2.0)B 0.0 (0.0–0.0)B 0.0 (0.0–0.0)B <0.001* 
 NRS, peak pain last 4 weeks 8.0 (6.0–9.0)A 0.0 (0.0–1.0)B 0.0 (0.0–0.0)B 0.0 (0.0–0.0)B <0.001* 
 NRS, average pain last 4 weeks 5.0 (4.0–8.0)A 0.0 (0.0)B 0.0 (0.0–0.0)B 0.0 (0.0–0.0)B <0.001* 

Data are means ± SD unless otherwise indicated. Overall P values are presented. T1DM, type 1 diabetes mellitus. Statistical differences between groups are denoted with A,B, and C, where groups including the same letters (such as for HbA1c where the groups T1DM with painful DPN and T1DM with painless DPN are denoted with A) indicate no statistical differences. Not having the same letters (such as for HbA1c where T1DM with painless DPN is denoted with A and T1DM without DPN is denoted with B) indicates pairwise statistical differences.

*

Significant difference.

Median (interquartile range).

No pairwise statistical significance.

Overall, the four groups were comparable in sex, age, and BMI, and the three groups with diabetes were also comparable in age at diabetes onset (all P ≥ 0.18). The post hoc analyses revealed significantly higher HbA1c in all groups with diabetes compared with healthy control subjects (post hoc, all P < 0.001). Additionally, higher HbA1c and longer diabetes duration were observed in the group with diabetes and painless DPN compared with the group with diabetes without DPN (post hoc, all P ≤ 0.020). More participants with painful DPN were using duloxetine compared with the other three groups (post hoc, all P ≤ 0.008). No differences in other CNS-acting medications were found between the groups (post hoc, all P > 0.050). All peripheral measurements (nerve conduction measurements and cold/warm detection thresholds) reflected worse nerve function in the group with diabetes with painful DPN and the group with diabetes with painless DPN compared with the diabetes group without DPN and healthy control subjects (post hoc, all P ≤ 0.017). Additional data on the quantitative sensory testing are presented in Supplementary Table 1. Also, as expected, pain scores were higher in the painful DPN group compared with the other three groups (post hoc, all P < 0.001).

Resting-State Functional Connectivity Patterns

A detailed overview of the thalamic and postcentral gyral connectivity to other brain regions compared in the four groups can be found in Table 2, and some of the significant regions are illustrated in Fig. 1.

Figure 1

Resting-state functional connectivity of left thalamus with supplementary motor cortex (SMC) and superior frontal gyrus (SFG) (A), right thalamus with supplementary motor cortex and superior frontal gyrus (B), left postcentral gyrus (PostCG) with precentral gyrus (PreCG) and postcentral gyrus (C), and right postcentral gyrus with precentral gyrus and postcentral gyrus (D). Blue, higher connectivity in subjects with type 1 diabetes without DPN compared with healthy control subjects; red, lower connectivity in subjects with type 1 diabetes with painful DPN compared with subjects with type 1 diabetes without DPN. HC, healthy control subjects; T1DM, type 1 diabetes mellitus; w/o, without.

Figure 1

Resting-state functional connectivity of left thalamus with supplementary motor cortex (SMC) and superior frontal gyrus (SFG) (A), right thalamus with supplementary motor cortex and superior frontal gyrus (B), left postcentral gyrus (PostCG) with precentral gyrus (PreCG) and postcentral gyrus (C), and right postcentral gyrus with precentral gyrus and postcentral gyrus (D). Blue, higher connectivity in subjects with type 1 diabetes without DPN compared with healthy control subjects; red, lower connectivity in subjects with type 1 diabetes with painful DPN compared with subjects with type 1 diabetes without DPN. HC, healthy control subjects; T1DM, type 1 diabetes mellitus; w/o, without.

Close modal
Table 2

Thalamic and postcentral gyral connectivity to other brain areas

Connectivity contrastROIConnected regionsPeak MNI coordinates (x, y, z)Cluster size (mm3)PFWE corrected
T1DM without DPN > HC Thalamus, L Bilateral, supplementary motor cortex −14, −2, 66 466 <0.001 
  Bilateral, superior frontal gyrus    
T1DM with painful DPN < T1DM without DPN Thalamus, L L, supplementary motor cortex −14, 0, 66 329 0.011 
 L, superior frontal gyrus    
T1DM with painful DPN < HC Thalamus, L L, caudate −12, 2, 18 301 0.017 
T1DM without DPN > HC Thalamus, R Bilateral, supplementary motor cortex −10, 0, 66 1,247 <0.001 
  Bilateral, superior frontal gyrus    
T1DM with painful DPN < T1DM without DPN Postcentral gyrus, L L, precentral gyrus −34, 22, 52 326 0.016 
 L, postcentral gyrus    
  R, postcentral gyrus 40, −32, 40 364 0.010 
  R, supramarginal gyrus    
T1DM with painful DPN < T1DM without DPN Postcentral gyrus, R L, precentral gyrus −34, −24, 52 331 0.015 
 L, postcentral gyrus    
  R, precentral gyrus 40, −20, 40 283 0.029 
  R, postcentral gyrus    
T1DM with painful DPN < HC Postcentral gyrus, R Bilateral, postcentral gyrus 12, −40, 70 362 0.010 
Connectivity contrastROIConnected regionsPeak MNI coordinates (x, y, z)Cluster size (mm3)PFWE corrected
T1DM without DPN > HC Thalamus, L Bilateral, supplementary motor cortex −14, −2, 66 466 <0.001 
  Bilateral, superior frontal gyrus    
T1DM with painful DPN < T1DM without DPN Thalamus, L L, supplementary motor cortex −14, 0, 66 329 0.011 
 L, superior frontal gyrus    
T1DM with painful DPN < HC Thalamus, L L, caudate −12, 2, 18 301 0.017 
T1DM without DPN > HC Thalamus, R Bilateral, supplementary motor cortex −10, 0, 66 1,247 <0.001 
  Bilateral, superior frontal gyrus    
T1DM with painful DPN < T1DM without DPN Postcentral gyrus, L L, precentral gyrus −34, 22, 52 326 0.016 
 L, postcentral gyrus    
  R, postcentral gyrus 40, −32, 40 364 0.010 
  R, supramarginal gyrus    
T1DM with painful DPN < T1DM without DPN Postcentral gyrus, R L, precentral gyrus −34, −24, 52 331 0.015 
 L, postcentral gyrus    
  R, precentral gyrus 40, −20, 40 283 0.029 
  R, postcentral gyrus    
T1DM with painful DPN < HC Postcentral gyrus, R Bilateral, postcentral gyrus 12, −40, 70 362 0.010 

> and < represent higher and lower connectivity (i.e., T1DM without DPN > HC means higher connectivity in T1DM without DPN compared with HC). FWE-corrected significant thresholds at P < 0.05 are reported together with cluster size. HC, healthy control subjects; L, left; R, right; T1DM, type 1 diabetes mellitus.

The left and right thalamus ROIs showed higher functional connectivity to bilateral supplementary motor cortex with associated superior frontal gyrus in the diabetes group without DPN compared with healthy control subjects (both P < 0.001). The connectivity between the left thalamus and supplementary motor cortex/superior frontal gyrus was lower in the group with painful DPN compared with the group without DPN (P = 0.011) (see Fig. 1A and B).

Overall, the left and right postcentral gyrus revealed lower connectivity to the precentral gyrus and, furthermore, within and to their respective opposite postcentral gyrus in those with painful DPN compared with the group without DPN (all P ≤ 0.029) (see Fig. 1C and D). Moreover, the connectivity within the right postcentral gyrus and between the left and right postcentral gyrus was lower in the group with painful DPN compared with healthy control subjects (all P ≤ 0.010) (not illustrated).

No significant differences in connectivity patterns of insula were observed between the four groups (all P > 0.071).

Associations Between Brain Connectivity and Peripheral Nerve and Pain Measurements

The correlation analyses included all participants with diabetes (n = 56) to examine the association between the parameters independent of the individual groups. Connectivity z scores between an ROI (thalamus and postcentral gyrus) and the connected brain region (see Table 2) were used for the correlation analyses. Since the thalamus–to–motor area connectivity pattern was similar in comparing those without DPN with healthy control subjects and those without DPN with those with painful DPN, a mean of both connectivity z scores was used. In total, two thalamus–to–motor area connectivity parameters and five postcentral gyrus–to–precentral/postcentral gyrus connectivity parameters were calculated. The connectivity parameters (except one) were significantly correlated to large sensory and motor peripheral nerve fiber functions (amplitude and conduction velocity), small nerve fiber measurements, and pain scores (DN4 and NRS scores) (see Supplementary Table 2).

In general, the following was observed for both thalamic and postcentral gyral connectivities: lower connectivity z scores were associated with lower sural and peroneal nerve amplitudes and conduction velocities and cold detection thresholds (all r > 0.271, P < 0.043). Lower connectivity z scores were associated with a higher warm detection threshold, DN4 score, and peak and average pain over the prior 4 weeks (all r < 0.282, < 0.035) (see Supplementary Table 2).

The correlations were stronger for the thalamic connectivities compared with postcentral gyral connectivities (see Supplementary Table 2 and Fig. 2A–D for examples). Especially, the mean thalamus–to–motor area connectivity was strongly correlated with the peripheral nerve functions and pain score measurements. Overall, the correlations between thalamic connectivity and nerve amplitudes showed the possibility of roughly grouping into healthy control subjects, subjects with type 1 diabetes without DPN, and subjects with type 1 diabetes with painful/painless DPN. The latter is conceptually illustrated in Fig. 2E and F (without further statistical analysis). The healthy control subjects had intact sural nerve amplitude and low connectivity (gray circle), whereas individuals with diabetes without DPN in general revealed both lower nerve amplitudes and increased connectivity (yellow circle). Overall, individuals with painless and painful DPN demonstrated further abnormal nerve amplitude in connection with a decrease in connectivity, with no clear difference between the diabetes group with painless DPN (blue circle) and painful DPN (red circle). The same pattern was observed for the peroneus nerve amplitude-connectivity plot (Fig. 2F) but also for nerve conduction velocity for both sural and peroneus nerves (not illustrated).

Figure 2

AD: Correlation graphs between thalamic/postcentral gyral connectivity and nerve conduction measurements. E and F: Conceptualizing the association between thalamus seed-based connectivity and nerve amplitudes. PreCG, precentral gyrus; PostCG, postcentral gyrus; SFG, superior frontal gyrus; SMC, supplementary motor cortex; T1DM, type 1 diabetes mellitus.

Figure 2

AD: Correlation graphs between thalamic/postcentral gyral connectivity and nerve conduction measurements. E and F: Conceptualizing the association between thalamus seed-based connectivity and nerve amplitudes. PreCG, precentral gyrus; PostCG, postcentral gyrus; SFG, superior frontal gyrus; SMC, supplementary motor cortex; T1DM, type 1 diabetes mellitus.

Close modal

In this study we investigated the functional connectivity patterns of thalamus, postcentral gyrus, and insula in individuals with type 1 diabetes and painful DPN, type 1 diabetes and painless DPN, type 1 diabetes without DPN, and healthy control subjects. Thalamus and the postcentral gyrus showed increased connectivity to motor areas in subjects with type 1 diabetes without DPN compared with both subjects with type 1 diabetes with painful DPN and healthy control subjects. In other words, the painful DPN group had lower connectivity than the diabetes group without DPN. No connectivity differences between the groups were found for the insula. Lower connectivities of thalamus and postcentral gyrus were associated with poorer peripheral nerve functions and higher pain intensity.

Thalamus and postcentral gyrus (the latter constitutes the primary somatosensory cortex) are parts of the spinothalamic tract. Peripheral sensory inputs are transmitted to the thalamus and then to the primary somatosensory cortex (32). The regions generally receive sensory inputs such as touch, pain, and temperature (17,33) and have previously been investigated in type 1 diabetes (6,811,13,14). However, their functional involvement in painful and painless DPN is not fully elucidated. In general, we found lower thalamus–to–motor area connectivity in individuals with painful DPN compared with individuals with type 1 diabetes without DPN but not compared with individuals with painless DPN and healthy control subjects. Previous brain imaging studies mainly suggest higher thalamus activity in individuals with neuropathic pain but also report differences to painless DPN and healthy control subjects (3437). One study investigating the thalamic connectivity to other brain regions found higher connectivity to the insula and primary somatosensory cortex in subjects with type 1 diabetes and irritable painful DPN compared with subjects with type 1 diabetes and nonirritable painful DPN (35). However, since a control group was not included, it was not possible to conclude whether the findings were specific to painful DPN. In another study investigators reported that individuals with painful DPN had elevated thalamic perfusion, which may be related to increased neuronal activity, compared with individuals with painless DPN and diabetes without DPN and healthy control subjects (36). There are several possible explanations for the lower connectivity observed in our study in the painful DPN group and the lack of difference between the painful and painless DPN groups and between painful DPN and healthy control groups. Nearly all participants in the painful DPN group had confirmed and severe DPN. Severely damaged nerves lead to decreased transmission of pain signals to the CNS and probably a lower connectivity of the thalamus. The result is that pain sensation decreases, which may also be reflected in our painful DPN group only experiencing mild pain instead of severe pain (19). Also, the level of neuropathic pain medication used, particularly duloxetine, was higher in the painful DPN group compared with the other three groups. Duloxetine has previously been shown to reduce functional brain responses in depressive disorders and may be an additional reason for lower connectivity in the painful DPN group (38). The current study was not designed to exclude CNS-acting medications, and further studies are needed to consider these. Furthermore, the painful DPN group was heterogeneous in terms of the presence of DPN. Therefore, individuals in the group are likely to possess different connectivity patterns, and further phenotyping of these individuals may be needed (35).

Even though previous studies have investigated similar brain areas, the differences between studies in the phenotyping of painful and painless DPN, other group characteristics and disease complications, and the approach of fMRI methods used make it difficult to compare results directly.

In our study, insula did not show any differences in the connectivity between the four groups. Studies have demonstrated structural alterations of the insula in type 1 diabetes (16,39,40), and other studies have reported no differences in insular activity in stimulating with pain in type 1 diabetes with or without neuropathic complications (6,41). However, to our best knowledge, no studies have investigated the insula’s connectivity at rest in groups of subjects with type 1 diabetes both with and without painful and painless DPN.

Previous studies investigating painful DPN have reported that functional alterations of brain regions involved in pain processing, including thalamus and somatosensory cortex, are associated with severity of DPN and pain perception (6,15,34). Results of other studies suggested that thalamic atrophy and dysfunction were associated with reduced sensory function in type 1 diabetes with DPN (14,42). We found that lower thalamic and postcentral gyral connectivity parameters were associated with both poorer peripheral nerve functions and increased pain intensity ratings. This indicates a strong link between peripheral nerve changes and altered brain functions. The associations were strongest for thalamic connectivity measurements, possibly due to the anatomy of the spinothalamic tract, where the sensory inputs are received to the thalamus first and transmitted through the third-order neurons to the primary somatosensory cortex (32). The strongest associations for thalamic connectivities were found with the nerve amplitude of the peripheral sensory and motor nerves, where lower connectivity was associated with poorer nerve amplitude. The amplitude partly reflects the number of axons in a sensory nerve and the integrated function of the motor axons (43). Lower thalamus–to–motor area connectivity could reflect axonal loss together with a reduced integrated function of the peripheral nerves.

Based on our correlation graphs of the thalamus–to–motor area connectivity in combination with nerve conduction measurements, it was possible to exploratively distinguish among those with type 1 diabetes with painful/painless DPN, those with type 1 diabetes without DPN, and healthy control subjects. The groups with painful and painless DPN were similar regarding the connectivity but also regarding the nerve conduction measurements. The higher connectivity observed in type 1 diabetes without DPN raises speculations of whether the increased thalamic connectivity is a result of neuroplastic mechanisms to compensate and maintain normal sensory processing/inputs despite a possible subclinical dysfunction of nerve function in this group. The presence of clinically detectable DPN later in the disease process may lead to a loss of the increased connectivity in the group without DPN. Compensatory mechanisms to preserve normal ability have previously been observed in the brain of children with type 1 diabetes, where greater brain functional modulation was associated with better cognitive performance (44). Since this is a cross-sectional study, no causal conclusions can be reported. Longitudinal studies are needed to support whether the connectivity findings observed are associated with the developmental process of painful and painless DPN.

Even though our study consisted of well-matched and well-phenotyped groups, it had some limitations. First, the included study participants, especially those with painful DPN, were not free of analgesics and other CNS-acting medications, which may have affected the resting-state fMRI results, and this issue should be considered in future studies. Second, the current seed-based analysis could be expanded to involve other pain regions, including anterior cingulate cortex and ventrolateral periaqueductal grey, in which functional connectivities have been reported to be altered in painful DPN (34). Third, confounding factors such as hyperglycemia and hypoglycemic events may also influence functional connectivity (41,45) and may be obtained in future studies together with information on the duration of pain, pain score at MRI visit, and mood. Fourth, in our study, the participants were highly selected to ensure the group with painless DPN had DPN but no pain and that the group with painful DPN had neuropathic pain. This may cause selection bias and may not reflect the reality of the clinical settings. However, this study design was necessary and adapted for understanding of central mechanisms of painful and painless DPN. For obtaining a deeper understanding of the CNS involvement in painful and painless DPN, several MRI modalities testing the interplay among structural, functional, and metabolic changes may be combined. It is also of interest to combine MRI findings of both the CNS and peripheral nervous system to understand their interaction in the development of painful and painless DPN.

In conclusion, both thalamus and postcentral gyrus revealed higher functional connectivity to motor areas in subjects with type 1 diabetes without DPN compared with both subjects with type 1 diabetes with painful DPN and healthy control subjects. These connectivity patterns were additionally associated with peripheral nerve functions and pain intensity ratings. The increased connectivity was attenuated along with the DPN severity. The correlations with thalamic connectivity measurements were stronger than with postcentral gyral connectivities. Our study contributes to further understanding of the functional alterations of brain regions involved in sensory processing in type 1 diabetes and neuropathic complications. Functional connectivity assessments may be essential to include as a technique to gain information about alterations in the CNS in further investigation of diabetic phenotypes and to evaluate potential biomarkers for early detection and progression of painful and painless DPN. In future resting-state fMRI studies, machine learning approaches will likely be beneficial to improve the phenotyping of painful and painless DPN.

Clinical trial reg. no. NCT04078516, clinicaltrials.gov

This article contains supplementary material online at https://doi.org/10.2337/figshare.21330507.

Acknowledgments. The authors acknowledge Kenneth K. Jensen, Department of Radiology, Aalborg University Hospital, for his assistance in data collection.

Funding. This work was partly supported by Augustinus Fonden, København, Denmark (grant 19-1302).

The funding source did not influence the study.

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

Author Contributions. S.S.C. contributed to study conceptualization, data acquisition, investigation, project administration, and writing of the original draft. J.R. contributed to study conceptualization, data acquisition, investigation, project administration, and review and editing of the original draft. C.D.M. contributed to study conceptualization and review and editing of the original draft. N.E. contributed to study conceptualization, supervision, and review and editing of the original draft. J.B.F. contributed to study conceptualization, supervision, and review and editing of the original draft. T.M.H. contributed to study conceptualization, supervision, and review and editing of the original draft. All authors approved the final version of the article and the decision to submit and publish the manuscript. T.M.H. 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.

1.
Tesfaye
S
,
Boulton
AJM
,
Dyck
PJ
, et al.;
Toronto Diabetic Neuropathy Expert Group
.
Diabetic neuropathies: update on definitions, diagnostic criteria, estimation of severity, and treatments
.
Diabetes Care
2010
;
33
:
2285
2293
2.
Tesfaye
S
,
Vileikyte
L
,
Rayman
G
, et al.;
Toronto Expert Panel on Diabetic Neuropathy
.
Painful diabetic peripheral neuropathy: consensus recommendations on diagnosis, assessment and management
.
Diabetes Metab Res Rev
2011
;
27
:
629
638
3.
Tesfaye
S
,
Selvarajah
D
.
Advances in the epidemiology, pathogenesis and management of diabetic peripheral neuropathy
.
Diabetes Metab Res Rev
2012
;
28
(
Suppl. 1
):
8
14
4.
Abbott
CA
,
Malik
RA
,
van Ross
ERE
,
Kulkarni
J
,
Boulton
AJM
.
Prevalence and characteristics of painful diabetic neuropathy in a large community-based diabetic population in the U.K
.
Diabetes Care
2011
;
34
:
2220
2224
5.
Selvarajah
D
,
Wilkinson
ID
,
Davies
J
,
Gandhi
R
,
Tesfaye
S
.
Central nervous system involvement in diabetic neuropathy
.
Curr Diab Rep
2011
;
11
:
310
322
6.
Selvarajah
D
,
Wilkinson
ID
,
Fang
F
, et al
.
Structural and functional abnormalities of the primary somatosensory cortex in diabetic peripheral neuropathy: a multimodal MRI study
.
Diabetes
2019
;
68
:
796
806
7.
Zhang
Q
,
Zhang
P
,
Yan
R
, et al
.
A single-blinded trial using resting-state functional magnetic resonance imaging of brain activity in patients with type 2 diabetes and painful neuropathy
.
Diabetes Ther
2019
;
10
:
135
147
8.
van Duinkerken
E
,
Schoonheim
MM
,
Steenwijk
MD
, et al
.
Ventral striatum, but not cortical volume loss, is related to cognitive dysfunction in type 1 diabetic patients with and without microangiopathy
.
Diabetes Care
2014
;
37
:
2483
2490
9.
Moulton
CD
,
Costafreda
SG
,
Horton
P
,
Ismail
K
,
Fu
CHY
.
Meta-analyses of structural regional cerebral effects in type 1 and type 2 diabetes
.
Brain Imaging Behav
2015
;
9
:
651
662
10.
Filip
P
,
Canna
A
,
Moheet
A
, et al
.
Structural alterations in deep brain structures in type 1 diabetes
.
Diabetes
2020
;
69
:
2458
2466
11.
Hansen
TM
,
Muthulingam
JA
,
Brock
B
, et al
.
Reduced gray matter brain volume and cortical thickness in adults with type 1 diabetes and neuropathy
.
Neurosci Res
2022
;
17
:
66
72
12.
Selvarajah
D
,
Wilkinson
ID
,
Emery
CJ
, et al
.
Early involvement of the spinal cord in diabetic peripheral neuropathy
.
Diabetes Care
2006
;
29
:
2664
2669
13.
Selvarajah
D
,
Wilkinson
ID
,
Maxwell
M
, et al
.
Magnetic resonance neuroimaging study of brain structural differences in diabetic peripheral neuropathy
.
Diabetes Care
2014
;
37
:
1681
1688
14.
Hansen
TM
,
Frøkjær
JB
,
Selvarajah
D
, et al
.
Reduced thalamic volume and metabolites in type 1 diabetes with polyneuropathy
.
Exp Clin Endocrinol Diabetes
2022
;
130
:
327
334
15.
Tseng
MT
,
Chiang
MC
,
Chao
CC
,
Tseng
WYI
,
Hsieh
ST
.
fMRI evidence of degeneration-induced neuropathic pain in diabetes: enhanced limbic and striatal activations
.
Hum Brain Mapp
2013
;
34
:
2733
2746
16.
Croosu
SS
,
Hansen
TM
,
Røikjer
J
,
Mørch
CD
,
Ejskjaer
N
,
Frøkjær
JB
.
Gray matter brain alterations in type 1 diabetes - findings based on detailed phenotyping of neuropathy status
.
Exp Clin Endocrinol Diabetes
2022
;
130
:
730
739
17.
Raju
H
,
Tadi
P
.
Neuroanatomy, somatosensory cortex
. In
StatPearls. Treasure Island, FL, Stat Pearls Publishing, 2020. Accessed 24 March 2022. Available from https://www.ncbi.nlm.nih.gov/books/NBK555915/
18.
Glover
GH
.
Overview of functional magnetic resonance imaging
.
Neurosurg Clin N Am
2011
;
22
:
133
139
,
vii
19.
Li
J
,
Zhang
W
,
Wang
X
, et al
.
Functional magnetic resonance imaging reveals differences in brain activation in response to thermal stimuli in diabetic patients with and without diabetic peripheral neuropathy
.
PLoS One
2018
;
13
:
e0190699
20.
Wang
N
,
Zhang
YH
,
Wang
JY
,
Luo
F
.
Current understanding of the involvement of the insular cortex in neuropathic pain: A narrative review
.
Int J Mol Sci
2021
;
22
:
2648
21.
Røikjer
J
,
Croosu
SS
,
Hansen
TM
, et al
.
The histamine-induced axon-reflex response in people with type 1 diabetes with and without peripheral neuropathy and pain: a clinical, observational study
.
J Pain
2022
;
23
:
1167
1176
22.
Bouhassira
D
,
Attal
N
,
Alchaar
H
, et al
.
Comparison of pain syndromes associated with nervous or somatic lesions and development of a new neuropathic pain diagnostic questionnaire (DN4)
.
Pain
2005
;
114
:
29
36
23.
Spallone
V
,
Morganti
R
,
D’Amato
C
,
Greco
C
,
Cacciotti
L
,
Marfia
GA
.
Validation of DN4 as a screening tool for neuropathic pain in painful diabetic polyneuropathy
.
Diabet Med
2012
;
29
:
578
585
24.
Haefeli
M
,
Elfering
A
.
Pain assessment
.
Eur Spine J
2006
;
15
(
Suppl. 1
):
S17
S24
25.
Rolke
R
,
Baron
R
,
Maier
C
, et al
.
Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values
.
Pain
2006
;
123
:
231
243
26.
Rolke
R
,
Magerl
W
,
Campbell
KA
, et al
.
Quantitative sensory testing: a comprehensive protocol for clinical trials
.
Eur J Pain
2006
;
10
:
77
88
27.
Whitfield-Gabrieli
S
,
Nieto-Castanon
A
.
Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks
.
Brain Connect
2012
;
2
:
125
141
28.
Patriat
R
,
Molloy
EK
,
Birn
RM
.
Using edge voxel information to improve motion regression for rs-fMRI connectivity studies
.
Brain Connect
2015
;
5
:
582
595
29.
Zhu
X
,
Helpman
L
,
Papini
S
, et al
.
Altered resting state functional connectivity of fear and reward circuitry in comorbid PTSD and major depression
.
Depress Anxiety
2017
;
34
:
641
650
30.
Pelletier-Baldelli
A
,
Andrews-Hanna
JR
,
Mittal
VA
.
Resting state connectivity dynamics in individuals at risk for psychosis
.
J Abnorm Psychol
2018
;
127
:
314
325
31.
Brett
M
,
Anton
J-L
,
Valabregue
R
,
Poline
J-B
.
Region of interest analysis using an SPM toolbox
.
Abstract presented at the 8th International Conference on Functional Mapping of the Human Brain
,
2–6 June 2002
,
Sendai, Japan
.
Available on CD-ROM in Neuroimage
.
2002
;
16
32.
Al-Chalabi
M
,
Gupta
S
.
Neuroanatomy, spinothalamic tract
. In
StatPearls. Treasure Island, FL, Stat Pearls Publishing, 2019. Accessed 25 August 2022. Available from https://www.ncbi.nlm.nih.gov/books/NBK507824/
33.
Torrico
TJ
,
Munakomi
S
.
Neuroanatomy, thalamus. In StatPearls
. In
StatPearls. Treasure Island, FL, Stat Pearls Publishing, 2019. Accessed 24 March 2022. Available from https://www.ncbi.nlm.nih.gov/books/NBK542184/
34.
Segerdahl
AR
,
Themistocleous
AC
,
Fido
D
,
Bennett
DL
,
Tracey
I
.
A brain-based pain facilitation mechanism contributes to painful diabetic polyneuropathy
.
Brain
2018
;
141
:
357
364
35.
Teh
K
,
Wilkinson
ID
,
Heiberg-Gibbons
F
, et al
.
Somatosensory network functional connectivity differentiates clinical pain phenotypes in diabetic neuropathy
.
Diabetologia
2021
;
64
:
1412
1421
36.
Selvarajah
D
,
Wilkinson
ID
,
Gandhi
R
,
Griffiths
PD
,
Tesfaye
S
.
Microvascular perfusion abnormalities of the thalamus in painful but not painless diabetic polyneuropathy: a clue to the pathogenesis of pain in type 1 diabetes
.
Diabetes Care
2011
;
34
:
718
720
37.
Croosu
SS
,
Hansen
TM
,
Brock
B
,
Drewes
AM
,
Brock
C
,
Frøkjær
JB
.
Altered functional connectivity between brain structures in adults with type 1 diabetes and polyneuropathy
.
Brain Res
2022
;
1784
:
147882
38.
López-Solà
M
,
Pujol
J
,
Hernández-Ribas
R
, et al
.
Effects of duloxetine treatment on brain response to painful stimulation in major depressive disorder
.
Neuropsychopharmacology
2010
;
35
:
2305
2317
39.
Northam
EA
,
Rankins
D
,
Lin
A
, et al
.
Central nervous system function in youth with type 1 diabetes 12 years after disease onset
.
Diabetes Care
2009
;
32
:
445
450
40.
Stantonyonge
N
,
Sampedro
F
,
Méndez
J
,
Martínez-Horta
S
,
Chico
A
,
Gómez-Anson
B
.
Structural gray and white matter differences in patients with type 1 diabetes and impaired awareness of hypoglycemia
.
J Clin Endocrinol Metab
2021
;
106
:
450
458
41.
Bolo
NR
,
Musen
G
,
Simonson
DC
, et al
.
Functional connectivity of insula, basal ganglia, and prefrontal executive control networks during hypoglycemia in type 1 diabetes
.
J Neurosci
2015
;
35
:
11012
11023
42.
Selvarajah
D
,
Wilkinson
ID
,
Emery
CJ
, et al
.
Thalamic neuronal dysfunction and chronic sensorimotor distal symmetrical polyneuropathy in patients with type 1 diabetes mellitus
.
Diabetologia
2008
;
51
:
2088
2092
43.
Kane
NM
,
Oware
A
.
Nerve conduction and electromyography studies
.
J Neurol
2012
;
259
:
1502
1508
44.
Foland-Ross
LC
,
Tong
G
,
Mauras
N
, et al.;
Diabetes Research in Children Network (DirecNet)
.
Brain function differences in children with type 1 diabetes: a functional MRI study of working memory
.
Diabetes
2020
;
69
:
1770
1778
45.
van Duinkerken
E
,
Schoonheim
MM
,
IJzerman
RG
, et al
.
Altered eigenvector centrality is related to local resting-state network functional connectivity in patients with longstanding type 1 diabetes mellitus
.
Hum Brain Mapp
2017
;
38
:
3623
3636
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.