Diametrical Effects of Glucose Levels on Microvascular Permeability of Peripheral Nerves in Patients With Type 2 Diabetes With and Without Diabetic Neuropathy

Although extensively investigated in both clinical and preclinical studies, the pathophysiology of distal symmetric diabetic neuropathy (DN), one of the most frequent and most disabling complications of both type 1 and type 2 diabetes (T2D), remains poorly understood (1–4). As a consequence, therapeutic approaches remain limited to the adjustment of blood glucose levels and the modification of individual risk factors, such as dyslipidemia (5).

While the optimization of blood glucose levels in type 1 diabetes has been shown to prevent the occurrence and progression of DN, studies on intensified glucose control in patients with T2D yielded controversial results (6–8). A meta-analysis conducted by Callaghan et al. (9) found no significant effect of intensive glucose control on the clinical outcome in patients with T2D with regard to DN progression. Apart from that, a phenomenon known as treatment-induced neuropathy has been described in patients with diabetes who undergo fast normalization of HbA_1c levels (10).

Besides hyperglycemia, clinical studies have found microangiopathy to be an important contributor to structural nerve damage in DN. It remains uncertain, however, to what extent hyperglycemia contributes to microangiopathy in patients with T2D since it is not possible to directly assess the effect of blood glucose and HbA_1c levels on affected nerves’ microcirculation. Dynamic contrast–enhanced magnetic resonance neurography (DCE-MRN) at 3 Tesla allows assessment of parameters of microvascular perfusion in vivo. The constant of the examined nerve’s capillary permeability (K^trans), volume fraction of the plasma space (v_p), and volume fraction of the extracapillary extracellular space (v_e) can be calculated by using the extended Tofts model (11,12). Recent studies on DCE-MRN in patients with T2D found that patients with DN showed lower microvascular permeability compared with patients without DN and that a reduction of the sciatic nerve’s microvascular permeability is associated with a decline in electrophysiologic parameters (13,14). In contrast, several studies on animal models found that hyperglycemia (15–18) as well as the deposition of advanced glycation end products result in increased endothelial permeability (19).

The aim of this study was to investigate potential associations of blood glucose and HbA_1c levels with parameters of nerve perfusion obtained from DCE-MRN in patients with T2D with and without DN.

This study was approved by the ethics committee of Heidelberg University Hospital (Heidelberg Study on Diabetes and Complications [HEIST-DiC]; ClinicalTrials.gov identifier NCT03022721, local ethics number S-383/2016). All participants gave written informed consent. Participants were screened and recruited at the outpatient clinic of the Department of Endocrinology, where all clinical, serological, and electrophysiological examinations were performed. MRN and image processing were performed by the Department of Neuroradiology. The sample size of this study was based on that of previous studies on correlations of DCE-MRN parameters with clinical and serological data in patients with and without DN (13,14).

The presented cohort of patients is part of the HEIST-DiC cohort (20,21). A total of 58 patients with T2D (21 women and 37 men) were enrolled in this cross-sectional single-center study between June 2018 and March 2020. Patients were divided into a group of patients without DN and a group with DN based on electrophysiological criteria following a recommendation for controlled clinical trials by the Diabetic Neuropathy Study Group of the European Association for the Study of Diabetes (NEURODIAB) and the Toronto consensus panel (4). DN was diagnosed if parameters of nerve conduction were abnormal in at least two out of three examined nerves and if a symptom or symptoms or a sign or signs of sensory motor polyneuropathy were present. All parameters of nerve conduction were normalized for age and height. For the definition of abnormal parameters of nerve conduction, reference values for electrophysiologic examinations issued by the Normative Data Task Force were used (22). Both groups were matched for age, BMI, fasting state glucose, HbA_1c, duration of diabetes, and glomerular filtration rate (GFR) in order to minimize potential confounding factors.

Overall exclusion criteria were as follows: <18 years of age, pregnancy, an estimated GFR (eGFR) <60 mL/min, any contraindications for MRI or administration of MRI contrast agents, any history of myocardial infarction, coronary heart disease or heart surgery, spine surgery, or lumbar disc extrusion. Furthermore, any risk factors for sarcopenia or neuropathy other than diabetes (e.g., malignant diseases, alcoholism, and hypovitaminosis), any previous or ongoing exposure to neurotoxic agents, and any chronic neurological diseases, such as Parkinson disease, restless leg syndrome, or multiple sclerosis, led to exclusion.

Patients with an eGFR of <60 mL/min were excluded from this study to exclude severe renal insufficiency as a confounding factor.

A detailed medical history was taken for every patient. Blood sampling was performed in the fasting state directly prior to MRN. Samples were directly analyzed at the central laboratory of Heidelberg University Hospital. Cystatin C values were analyzed in each patient to calculate the eGFR.

All electrophysiological studies were performed on the patients’ right leg by two specially trained medical technical assistants with >6 years of experience in electrophysiological assessments on patients with diabetes. Skin temperature was maintained at 32°C throughout the examination. The electrophysiological examination included the assessment of nerve conduction velocities (NCVs) of the tibial, peroneal, and sural nerve, distal motor latencies (DMLs) of the tibial and peroneal nerve, compound muscle action potentials (CMAPs) of the tibial and peroneal nerve, and sensory nerve action potentials (SNAPs) of the sural nerve, as previously published (13,23,24).

An examination of neuropathic symptoms was performed comprising the Neuropathy Disability Score (NDS) and the Neuropathy Severity Scale (NSS) in accordance with the guidelines issued by the German Diabetes Association (25).

High-resolution MRN of the right thigh in a 3.0 Tesla MR-scanner (Magnetom Tim TRIO; Siemens Healthineers, Erlangen, Germany) was performed for every patient using a 15-channel transmit–receive extremity coil. Sequences were centered to the sciatic nerve bifurcation at distal thigh level using the following parameters (13):

1. Axial high-resolution T2-weighted turbo spin echo two-dimensional sequence with spectral fat suppression. Repetition time (TR), 5,970 ms; echo time (TE), 55 ms; field of view (FOV), 160 × 160 mm²; matrix size, 512 × 512; slice thickness, 4 mm; no interslice gap; voxel size, 0.3 × 0.3 × 4.0 mm³; 24 slices; 24 acquired images; and total acquisition time, 4:42 min.

2. Axial T1-weighted volume interpolated breathhold examination (VIBE) sequence for determination of the precontrast T1 time. TR, 3.3 ms; TE, 1.11 ms; FOV, 160 × 160 mm²; matrix size, 128 × 128; slice thickness, 4 mm; interslice gap, 0.8 mm; voxel size, 1.3 × 1.3 × 4.0 mm³; single acquisition at a flip-angle of 5°, 8°, 11°, 14°, and 17°; 24 slices; total of 144 acquired images; and total acquisition time, 30 s.

3. Axial T1-weighted VIBE sequence. TR, 3.3 ms; TE, 1.11 ms; flip angle, 15°; FOV, 160 × 160 mm²; matrix size, 128 × 128; slice thickness, 4 mm; interslice gap, 0.8 mm; voxel size, 1.3 × 1.3 × 4.0 mm³; 50 repetitions; 24 slices; total of 1,200 acquired images; contrast agent administration (Dotarem [Guerbet, Villepinte, France]; 0.1 mmol/kg body weight, flow rate 3.5 mL/s) after completion of the sixth repetition; and total acquisition time, 4:09 min.

MATLAB 7.14.0.0739 (R2012a) and GraphPad Prism 7 were used for all statistical analyses. To test for Gaussian normal distribution, the D’Agostino-Pearson omnibus normality test was applied. The t tests were used for comparisons of two groups, and Pearson correlation coefficients were used for correlation analysis if a Gaussian normal distribution was given. In case of non-Gaussian distributed data, the Mann-Whitney test was used for comparisons of two groups. Likewise, nonparametric Spearman correlation was used for correlation analysis if data did not follow Gaussian distribution. If needed, partial correlation analysis was performed to exclude confounding variables. The level of significance was defined at P < 0.05 for all tests.

The data sets generated during and/or analyzed during the current study are not publicly available due to reasons of patient data protection but are available from the corresponding author upon reasonable request.

A total of 58 patients with T2D (21 women and 37 men) took part in this study. Twenty participants were diagnosed with DN in accordance with the electrophysiologic criteria explained above, while 38 patients did not suffer from DN.

Between the two groups of participants with and without DN, no differences could be found for age (DN, 66.2 ± 8.5 years; no DN, 65.4 ± 8.4 years; P = 0.728), BMI (DN, 28.3 ± 4.3 kg/m²; no DN, 29.2 ± 4.6 kg/m²; P = 0.482), fasting state glucose (DN, 139.3 ± 39.1 mg/dL; no DN, 126.9 ± 28.8 mg/dL; P = 0.309), HbA_1c (DN, 50.7 ± 8.1 mmol/mol and 6.8 ± 0.7%; no DN, 47.5 ± 8.3 mmol/mol and 6.6 ± 1.1%; P = 0.177), duration of diabetes (DN, 8.7 ± 9.7 years; no DN, 8.0 ± 5.9 years; P = 0.66), and GFR calculated from cystatin C levels (DN, 90.0 ± 20.2 mL/min; no DN, 82.8 ± 13.2 mL/min; P = 0.211).

In patients with DN, NSS (DN, 6.2 ± 2.6; no DN, 3.6 ± 3.4; P = 0.008) and NDS (DN, 4.5 ± 2.7; no DN, 2.7 ± 2.6; P = 0.024) were higher compared with patients without DN.

HbA_1c was elevated >47 mmol/mol or 6.5% in 12 of 20 patients in the DN group and in 20 of 38 patients in the group without DN.

K^trans was lower in patients with DN compared with patients without DN (DN, 0.034 ± 0.006 min⁻¹; no DN, 0.041 ± 0.014 min⁻¹; P = 0.036). No such differences could be found for v_p (DN, 0.046 ± 0.0054%; no DN, 0.047 ± 0.011%; P = 0.658) or v_e (DN, 0.014 ± 0.011%; no DN, 0.036 ± 0.047%; P = 0.131). Over all of the participants, no differences were found between woman and men for K^trans (0.0412 ± 0.013 min⁻¹ vs. 0.0372 ± 0.012 min⁻¹; P = 0.364), v_e (0.037 ± 0.047 vs. 0.023 ± 0.035; P = 0.402), or v_p (0.047 ± 0.008 vs. 0.047 ± 0.012; P = 0.365).

In the group with DN, nerve conduction studies of the sural nerve were incomplete in 10 participants with severe DN. All other acquired parameters of nerve conduction were complete in all participants. Tibial and peroneal NCV were lower in patients with DN compared with patients without DN (DN, 36.2 ± 6.2 m/s and no DN, 42.0 ± 3.9, P < 0.001; and DN, 35.4 ± 5.2 m/s and no DN 42.6 ± 3.7 m/s, P < 0.001, respectively). Peroneal (DN, 8.4 ± 3.7 ms; no DN, 4.3 ms; P < 0.001) and tibial DML (DN, 9.0 ± 5.4 ms; no DN, 4.3 ± 1.8 ms; P < 0.001) were longer in patients with DN.

In patients without DN, peroneal NCV correlated negatively with HbA_1c (r = −0.50; P = 0.004). No such correlation could be found for patients with DN.

A detailed overview of all patient characteristics is provided in Table 1.

Over all of the participants, positive correlations were found between K^trans and BMI (r = 0.42; P = 0.001), tibial NCV (r = 0.35; P = 0.011), peroneal NCV (r = 0.47; P < 0.001), and peroneal CMAP (r = 0.28; P = 0.041) as well as sural SNAP (r = 0.31; P = 0.038). A positive correlation with sural NCV failed to reach statistical significance (r = 0.34; P = 0.051). Negative correlations were found for K^trans with peroneal DML (r = −0.41; P = 0.002) and NDS (r = −0.40; P = 0.002).

In patients with T2D without DN, K^trans correlated positively with BMI (r = 0.45; P = 0.004) and negatively with age (r = −0.52; P < 0.001). Moreover, in this group, negative correlations could be found between K^trans and HbA_1c (r = −0.43; P = 0.012) (Fig. 2A) and K^trans and NDS (r = −0.372; P = 0.023). In a partial correlation analysis controlled for BMI, we found a negative correlation between K^trans and HbA_1c (r = −0.42; P = 0.016) as well, whereas partial correlation analysis controlled for BMI and age (r = −0.33; P = 0.071) and age alone (r = −0.30; P = 0.094) showed a similar trend without reaching statistical significance.

Positive correlations of K^trans could be found for peroneal (r = 0.47; P = 0.004) and sural NCV (r = 0.48; P = 0.015) and sural SNAP (r = 0.31; P = 0.035). Another positive correlation was found with v_e (r = 0.83; P < 0.0001). A detailed overview of correlations between MRN imaging parameters and clinical data of participants without DN is provided in Table 2.

In patients with T2D with DN, K^trans showed positive correlations with fasting state glucose (r = 0.46; P = 0.04), HbA_1c (r = 0.55; P = 0.015) (Fig. 2B), tibial NCV (r = 0.45; P = 0.047), and v_e (r = 0.848; P < 0.001). No significant correlations for age, BMI, or GFR could be found in this group. A detailed overview of correlations between MRN imaging parameters and clinical data of participants with DN is provided in Table 3.

In patients with T2D without DN, positive correlations for v_e were found with BMI (r = 0.37; P = 0.023), sural SNAP (r = 0.40; P = 0.033), peroneal NCV (r = 0.54; P < 0.001), and peroneal CMAP (r = 0.38; P = 0.023), while negative correlations were found with age (r = −0.46; P = 0.004), NDS (r = −0.54; P < 0.001), and peroneal DML (r = −0.37; P = 0.031).

For patients with DN, no correlations were found between v_e and clinical, electrophysiological, or serological data.

In patients without DN, v_p showed a positive correlation with peroneal CMAP (r = 0.36; P = 0.035) and a negative correlation with age (r = −0.33; P = 0.041).

In patients with DN, v_p showed a negative correlation with peroneal DML (r = −0.46; P = 0.042) and a similar trend for peroneal NCV (r = −0.42; P = 0.069).

For patients with DN, no correlations were found between v_p and clinical, electrophysiological, or serological data, except for a negative correlation with peroneal DML (r = −0.46; P = 0.042).

This study used 3-Tesla DCE-MRN to investigate potential associations of glucose and HbA_1c levels with peripheral nerve perfusion in patients with T2D. The main findings were: 1) in patients with T2D without DN, K^trans was higher compared with patients with DN, and 2) in patients with T2D with DN, K^trans was positively correlated with fasting state glucose and HbA_1c levels, whereas in patients with T2D without DN, K^trans was negatively correlated with HbA_1c levels.

The finding of lower K^trans values in patients with T2D with DN compared with patients with T2D without DN is in line with results of a reference study on DCE-MRN in T2D (13), in which K^trans was higher in patients with T2D without DN compared with patients with T2D with DN and compared with healthy control subjects, while no such difference could be found between patients with DN and control subjects (13). These findings indicate that T2D is accompanied by a disruption of the blood–nerve barrier that results in an increase of capillary permeability and that, for reasons yet to be determined, is apparently reverted in patients with T2D with DN. The finding that tibial or peroneal NCV were found to be correlated positively with K^trans in the respective patient groups suggests that a reduction of abnormally increased capillary permeability of peripheral nerves contributes to demyelination in T2D (13). In this context, the finding that, in this study, K^trans and peroneal NCV were negatively correlated with HbA_1c levels in patients without DN supports the hypothesis that glucose control in patients with T2D without DN may prevent hyperglycemia-related changes of capillary blood vessels, such as hypertrophy and hyperplasia of endothelial cells, thickening of the basement membrane, and capillary thrombosis, all of which have been described in histological studies of human nerve and muscle biopsies and linked to a decrease of capillary permeability in the later stages of the disease (29,30). In this context, the finding of an apparently normal capillary permeability in patients with T2D with DN may be the result of a larger barrier of diffusion as a result of the disease-related changes to capillary blood vessels mentioned above (29,31,32). Since there were no nerve biopsies obtained in this study to identify histological correlates for MRN findings, this assumption remains hypothetical.

The positive correlations of K^trans with fasting blood glucose and HbA_1c in patients with DN indicate that capillary permeability in patients with T2D with DN is increased by higher blood glucose levels. This is in line with results from preclinical studies that found capillary permeability to be increased due to complex pathways involving an increased leakage of tight junctions of both endothelial and perineurial cells (32), complement activation, and protein kinase C activation (16,17) that is also related to infections and inflammatory reactions (33,34). Thus, a potential explanation for diametric correlations of K^trans and HbA_1c levels in patients with T2D with and without DN might be that in all patients with T2D, increased glucose levels contribute to an increase of capillary permeability, but that in patients without DN, higher glucose levels also contribute to capillary changes, such as basement membrane thickening and damage to endothelial cells that predominantly decrease capillary permeability and, ultimately, contribute to DN, whereas in patients with T2D with DN, the aforementioned changes at capillary level have already taken place (35). Consequently, K^trans is lower in patients with DN compared with patients without DN, and the effect of glucose on capillary permeability predominates in DN patients. This may explain the controversial effects of glucose control on the progression of T2D DN in previous clinical studies (7–9). While our results indicate that glucose control in patients with T2D without DN may prevent microvascular damage, comparable effects cannot be shown in patients with T2D with DN. It remains to be determined, however, whether a decrease in blood glucose levels is consequently accompanied by a decrease in capillary permeability in patients with DN and whether this is of functional significance in the context of nutritional supply to peripheral nerves since glucose and other monosaccharides are transported across the capillary endothelium via both active transport and diffusion (35). Since DCE-MRN only allows us to assess the diffusion of gadolinium-based contrast agents but not the active transport of different molecules from the intracapillary to the extracapillary space, this study does not allow us to draw definite conclusions on the effect of capillary permeability on peripheral nerve nutrition or oxygenation (12).

Another explanation for the results of this study might be disturbances of microvascular flow patterns caused by the microangiopathic alterations outlined above (36,37). Østergaard et al. (37) suggested that altered flow patterns of the (endo-)neurial microcirculation led to an insufficient oxygen and glucose extraction and thereby contributed to DN. As K^trans represents the extraction of gadolinium contrast agent to the extracapillary space, it may indirectly represent a reduction of the extraction of oxygen and glucose to the extracapillary space due to changes of capillary flow. Consequently, the finding of lower K^trans values in patients with DN compared with patients without DN and the negative correlation between K^trans and HbA_1c would indicate that very strict glucose control may result in a critical reduction of glucose supply to peripheral nerves that is potentially harmful in patients with T2D with DN. This assumption is supported by preclinical studies that found that peripheral nerves compensate hypoxic states by increased glycolysis (38) and that low glucose levels in neurons are associated with cytochrome c–induced apoptosis, which suggests a need for a certain level of glucose supply for peripheral nerves and Schwann cells. Further evidence that strict glucose control may not be beneficial for all patients with T2D was provided by the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial, which could show that in patients with cardiovascular risk factors, intensive glucose control may lead to an increased mortality for reasons yet to be determined (39). Still, the assumption that intensive glucose control causes structural nerve damage in patients with T2D with DN should be treated with caution and cannot be proven by this study due to both sample size and the cross-sectional nature of the data. Also, the size of the presented collective does not allow us to exclude all possible confounding factors, although groups were matched for age, BMI, GFR, fasting state glucose, HbA_1c, and duration of diabetes to reduce possible confounders. This study is further limited by the fact that there is an imbalance of women and men between the two patient groups. Although no significant differences were found for parameters of nerve perfusion between women and men, we cannot fully exclude an impact of sex on parameters of nerve perfusion.

In summary, this study found diametrical correlations between HbA_1c levels and parameters of peripheral nerve capillary permeability in patients with T2D with and without DN. The results indicate that glucose control may have different effects on peripheral nerves’ capillary permeability in patients with T2D with and without DN. While the results support the hypothesis that glucose control is a beneficial preventive measure in patients with T2D without DN, the pathophysiological mechanisms underlying the positive correlation of HbA_1c levels and capillary permeability in the DN group and their clinical implications remain to be determined. As a consequence, future longitudinal studies assessing the effect of glucose control on the development of DN in patients with T2D should distinguish between patients with and without DN.

Acknowledgments. The authors thank Anita Pflästerer and Ulrike Bauer (Department of Endocrinology, Diabetology and Clinical Chemistry [Internal Medicine 1], Heidelberg University Hospital, Heidelberg, Germany) for the ongoing support and excellent technical performance of all electrophysiological studies. The authors also thank Dorothea Willich (Department of Neuroradiology, Heidelberg University Hospital) for ongoing support and excellent technical performance of all MRN examinations.

Funding. Z.K. and J.S. received grants from the Deutsches Zentrum für Diabetesforschung. J.M.E.J. received grants from the International Foundation for Research in Paraplegia (P179) and the Else Kröner-Fresenius-Stiftung (2021_EKES29), which provided financial support for personnel expenditures and MRI costs. This study was further supported by Deutsche Forschungsgemeinschaft SFB 1158 to P.N. and M.B. which provided financial support for personnel expenditures, MRI costs, and costs for the technical equipment required for electrophysiological and serological analysis. S.H. and M.B. received support from the Dietmar Hopp Foundation. F.T.K. was supported by the German Research Foundation (KU 3555/1-1) and the Hoffmann-Klose Foundation of Heidelberg University Hospital.

The International Foundation for Research in Paraplegia, Else Kröner-Fresenius-Stiftung, and Deutsche Forschungsgemeinschaft had no influence on the study design, collection, and analysis of data or on the writing of the article.

Duality of Interest. P.N. received grants from Novo Nordisk. M.B. received grants and personal fees from Codman, Guerbet, Bayer, and Novartis; personal fees from Roche, Teva Pharmaceuticals, Springer, and Boehringer; and grants from Siemens. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. C.M.M. contributed to the organization of the participants, collection of MR data, image segmentation, data analysis and interpretation, literature search, writing the manuscript, and arrangement of the figures. L.S. contributed to the organization of the participants and collection and evaluation of clinical, electrophysiological, and serological data. Z.K. and J.S. contributed to the collection and evaluation of clinical, electrophysiological, and serological data. S.H. contributed to the development of the MR sequence protocol and writing of the manuscript. P.N. contributed to the study design and the development of the clinical, electrophysiological, and serological study protocol. M.B. contributed to study design and coordination, development of the MR sequence protocol, and writing of the manuscript. S.K. contributed to development of the clinical and electrophysiological study protocol and to the collection and evaluation of clinical, electrophysiological, and serological data. F.T.K. contributed to image segmentation, programming of image analysis tools, data analysis and interpretation, literature search, writing the manuscript, and arrangement of figures. J.M.E.J. contributed to conception of the study, collection of MRI data, image segmentation, data analysis and interpretation, literature search, writing of the manuscript, and arrangement of figures. C.M.M. and J.M.E.J. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.