Characterization of experimental diabetic neuropathy using multicontrast magnetic resonance neurography at ultra high field strength

Abstract

In light of the limited treatment options of diabetic polyneuropathy (DPN) available, suitable animal models are essential to investigate pathophysiological mechanisms and to identify potential therapeutic targets. In vivo evaluation with current techniques, however, often provides only restricted information about disease evolution. In the study of patients with DPN, magnetic resonance neurography (MRN) has been introduced as an innovative diagnostic tool detecting characteristic lesions within peripheral nerves. We developed a novel multicontrast ultra high field MRN strategy to examine major peripheral nerve segments in diabetic mice non-invasively. It was first validated in a cross-platform approach on human nerve tissue and then applied to the popular streptozotocin(STZ)-induced mouse model of DPN. In the absence of gross morphologic alterations, a distinct MR-signature within the sciatic nerve was observed mirroring subtle changes of the nerves' fibre composition and ultrastructure, potentially indicating early re-arrangements of DPN. Interestingly, these signal alterations differed from previously reported typical nerve lesions of patients with DPN. The capacity of our approach to non-invasively assess sciatic nerve tissue structure and function within a given mouse model provides a powerful tool for direct translational comparison to human disease hallmarks not only in diabetes but also in other peripheral neuropathic conditions.

Figure 1

Behavioural phenotyping of STZ-induced diabetic mice. (a–c) Thermal nociception was tested measuring the behavioural latency [s] using the three different assays ‘Hotplate’, ‘Tailflick’ and ‘Hargreaves’. Significant differences between STZ-induced mice and controls were noted in the ‘Hotplate’ and ‘Hargreaves’. Group data provided as group mean ± s.d. *=p < 0.05, **=p < 0.01, n.s. = nonsignificant, unpaired two-sided t-test. N = 10 vs. 13 mice for ‘Hotplate’ and ‘Tailflick’ assays, n = 20 vs. 26 limbs for ‘Hargreaves’ assay. d-f) Significant correlation was noted for all of the behavioural assays and HbA_1c values (Hotplate: r = 0.59, p = 0.003; Tailflick: r = 0.59, p = 0.003; Hargreaves: r = 0.43, p = 0.04). Linear regression graphs shown (n = 23 mice).

Figure 2

Cross-platform validation of multicontrast MRN at 3 and 9.4 T. (a–e) Illustration and corresponding quantitative findings of the five MR contrasts ‘T2w_norm‘, ‘T2-Time’, ‘proton density (PD)’, ‘fractional anisotropy (FA)’ and ‘apparent diffusion coefficient (ADC)’. Each quadruplet of images shows a representative example MR-image of the control (Ctrl) and the type 1 diabetic (T1D) nerves at 3 and 9.4 T, respectively. A comparison of the respective quantitative fascicle group data is provided underneath, given as as group mean ± s.d. N = 10 (3 T) / 25 (9.4 T) vs. 11 (3 T) / 17 (9.4 T) fascicles, ****=p < 0.0001, unpaired two-sided t-test.

Figure 3

Imaging rationale of ultra high field MRN in experimental DPN. Schematic of the lumbosacral plexus with indication of imaging regions. The upper two images show normal morphological T2-w images of both sciatic nerves (white arrows) within the multicontrast imaging region of the lumbosacral plexus and at the pelvic exit in an axial orientation (a). Lower image depicting a normal axial T2-w image plane of the sciatic nerve along the proximal thigh (b), also indicated by a white arrow. For better orientation, the spinal canal in the image center, surrounded by the hypointense lumbar vertebral column, is indicated by red asterisks.

Figure 4

Multicontrast MRN in STZ-diabetic mice at ultra high field strength. (a–f) Illustration and corresponding quantitative findings of the six MR contrasts ‘T2w_norm‘ at proximal (i.e. pelvic level) and distal (i.e. thigh level) positions, ‘T2-time’, ‘proton density (PD)’, ‘fractional anisotropy (FA)’ and ‘apparent diffusion coefficient (ADC)’. Each pair of images shows a representative image example of the control (Ctrl) and STZ group, respectively. A comparison of the respective quantitative group data is provided underneath, given as group mean ± s.d. N = 20 vs. 26 nerves. *p < 0.05, ****p < 0.0001, n.s. = nonsignificant, unpaired two-sided t-test. White arrows indicating sciatic nerves.

Figure 5

Correlation analysis of MRN parameters with clinical and behavioural variables. Significant correlation was noted for the MRN parameters (a) ‘T2-time’ (r = −0.62, p = 0.002) and (b) ‘T2w_norm Distal’ (r = −0.51, p = 0.014) with the HbA_1c value and for T2-time with the behavioural assays (c) ‘Hotplate’ (r = −0.49, p = 0.018) and (d) ‘Hargreaves’ (r = −0.49, p = 0.017). Corresponding linear regression graphs shown (n = 23 mice).

Figure 6

Light microscopic analysis of the sciatic nerves from STZ-diabetic mice. (a) Gross anatomical findings of Toluidine blue stained semithin sciatic nerve sections from control and STZ animals using light microscopy (LM). Inset shows a representative region in x2-magnification. (b–d) Quantitative findings of the control (Ctrl) and the STZ groups (n = 3 each) comparing the three parameters b) ‘axon density’, (c) ‘myelin density’ (i.e. the myelin fraction within the nerve cross-sectional area (CSA)) and (d) ‘average myelination’ (i.e. myelin area per axon); n.s. = nonsignificant, unpaired two-sided t-test. (e) Histogram plot showing the frequency distribution of the equivalent axon radius within the population of all myelinated axons inside the nerve CSA. Each bar indicates group mean ± s.d. (n = 3 each). *p < 0.05, **p < 0.01, ***p < 0.001, unpaired two-sided t-test.

Figure 7

Ultrastructural analysis of the sciatic nerves from STZ-diabetic mice. (a) Representative ultrastructural findings from control and STZ animals using electron microscopy (EM). Red asterisks indicate myelin infoldings. (b–d) Quantitative findings of the control (Ctrl) and the STZ groups comparing the three parameters (b) Unmyelinated ‘axon density’ (per image frame, n = 16 vs. 17 frames), (c) ‘Unmyelinated fibre area’ (i.e. the unmyelinated fibre fraction per image frame, n = 16 vs. 17 frames) and d) ‘Unmyelinated axon size’ (i.e. all unmyelinated fibres within the image frames considered, median and quartiles indicated, n = 1646 (Ctrl) vs. 878 (STZ)); *=p < 0.05, n.s. = nonsignificant, unpaired two-sided t-test.