Diffusion-tensor magnetic resonance imaging captures increased skeletal muscle fibre diameters in Becker muscular dystrophy

Abstract

Background: Becker muscular dystrophy (BMD) is an X-linked disorder characterized by slow, progressive muscle damage and muscle weakness. Hallmarks include fibre-size variation and replacement of skeletal muscle with fibrous and adipose tissues, after repeated cycles of regeneration. Muscle histology can detect these features, but the required biopsies are invasive, are difficult to repeat and capture only small muscle volumes. Diffusion-tensor magnetic resonance imaging (DT-MRI) is a potential non-invasive alternative that can calculate muscle fibre diameters when applied with the novel random permeable barrier model (RPBM). In this study, we assessed muscle fibre diameters using DT-MRI in BMD patients and healthy controls and compared these with histology.

Methods: We included 13 BMD patients and 9 age-matched controls, who underwent water-fat MRI and DT-MRI at multiple diffusion times, allowing RPBM parameter estimation in the lower leg muscles. Tibialis anterior muscle biopsies were taken from the contralateral leg in 6 BMD patients who underwent DT-MRI and from an additional 32 BMD patients and 15 healthy controls. Laminin and Sirius-red stainings were performed to evaluate muscle fibre morphology and fibrosis. Twelve ambulant patients from the MRI cohort underwent the North Star ambulatory assessment, and 6-min walk, rise-from-floor and 10-m run/walk functional tests.

Results: RPBM fibre diameter was significantly larger in BMD patients (P = 0.015): mean (SD) = 68.0 (25.3) μm versus 59.4 (19.2) μm in controls. Inter-muscle differences were also observed (P ≤ 0.002). Both inter- and intra-individual RPBM fibre diameter variability were similar between groups. Laminin staining agreed with the RPBM, showing larger median fibre diameters in patients than in controls: 72.5 (7.9) versus 63.2 (6.9) μm, P = 0.006. However, despite showing similar inter-individual variation, patients showed more intra-individual fibre diameter variability than controls-mean variance (SD) = 34.2 (7.9) versus 21.4 (6.9) μm, P < 0.001-and larger fibrosis areas: median (interquartile range) = 21.7 (5.6)% versus 14.9 (3.4)%, P < 0.001. Despite good overall agreement of RPBM and laminin fibre diameters, they were not associated in patients who underwent DT-MRI and muscle biopsy, perhaps due to lack of colocalization of DT-MRI with biopsy samples.

Conclusions: DT-MRI RPBM metrics agree with histology and can quantify changes in muscle fibre size that are associated with regeneration without the need for biopsies. They therefore show promise as imaging biomarkers for muscular dystrophies.

Keywords: Becker muscular dystrophy; diffusion-tensor MRI; histopathology; immunohistochemistry; skeletal muscle.

Figure 1

Schematic showing the diffusion‐tensor magnetic resonance imaging (DT‐MRI) processing pipeline for this study. Representative axial spin‐echo echo planar imaging (SE‐EPI) DT‐MRI and chemical‐shift‐based water‐fat separation, or ‘Dixon’s;, images from the lower leg are shown for a 59‐year‐old Becker muscular dystrophy patient. The top row shows the DT‐MRI raw data, which undergo denoising, Dixon fat‐water decomposition to remove olefinic fat, and distortion correction. The bottom row shows region of interest (ROI) drawing on Dixon images, registration of ROIs to the DT‐MRI data and pixel‐wise thresholding based on low signal‐to‐noise ratio (SNR) and high fat fraction. ROIs, with excluded pixels indicated by a crosshatch pattern, are then used to obtain conventional DT‐MRI and random permeable barrier model metrics per muscle. FA, fractional anisotropy; MD, mean diffusivity.

Figure 2

Representative lower leg ‘Dixon’ chemical‐shift‐based water‐fat separation images and stimulated‐echo diffusion‐tensor magnetic resonance imaging (STE‐DT‐MRI) parameter maps from a 59‐year‐old Becker muscular dystrophy (BMD) patient (A) and a 58‐year‐old male healthy control (B). STE‐DT‐MRI data were acquired with a diffusion time of 330 ms. The BMD patient shows severe fat replacement in the gastrocnemius medialis and lateralis and in the peroneus longus and extensor digitorum longus muscles. This leads to signal voids in these regions in the DT‐MRI data, which were acquired with comprehensive fat suppression.

Figure 3

Boxplots showing fractional anisotropies (FA, top row) and mean diffusivities (MD, bottom row) derived from spin‐echo and stimulated‐echo diffusion‐tensor imaging (SE‐DTI and STE‐DTI, respectively) at three diffusion times, Δ, increasing from left to right. Median region‐of‐interest (ROI) measures are shown for Becker muscular dystrophy (BMD) patients and healthy controls over seven muscles of the lower leg: soleus (SOL), lateral gastrocnemius (GCL), medial gastrocnemius (GCM), peroneus longus (PER), tibialis posterior (TP), tibialis anterior (TA) and extensor digitorum longus (EDL). The boxplots represent median values by thick lines, with hinges corresponding to the 25th and 75th percentiles and dots approximating the raw data points. Boxplots are also ordered by median FA, highlighting a pattern of FA differences across the lower leg where the SOL has the lowest FA and the EDL has the highest, at all diffusion times. Dashed lines represent averages of all ROIs across all participants per diffusion time, showing that FA increases with diffusion time, whereas MD decreases. No statistically significant differences were observed between BMD patients and controls.

Figure 4

Characteristic fibre diameter and membrane permeability results obtained from the random permeable barrier model (RPBM) in Becker muscular dystrophy (BMD) patients and healthy controls. (A) Scatter plots showing the time dependence of axial and radial diffusivities in the soleus muscles of a 58‐year‐old control and a 59‐year‐old BMD patient. Squares and circles denote axial and radial diffusivities, respectively, per diffusion time, and dashed lines indicate fits of the RPBM to the radial diffusivity data per participant. (B) Boxplots showing a and κ for each group over the medial gastrocnemius (GCM), tibialis anterior (TA), tibialis posterior (TP), peroneus longus (PER), extensor digitorum longus (EDL), soleus (SOL) and lateral gastrocnemius (GCL) muscles. Median values are represented by thick lines, with hinges corresponding to the 25th and 75th percentiles, and boxplots are ordered by median fibre diameter. Two‐way analysis of variance showed globally larger fibre diameters in BMD patients versus controls, but no per‐muscle differences between groups. (C) Parameter maps, overlaid on matching ‘Dixon’ chemical‐shift‐based water‐fat separation water images, showing a and κ at the mid‐calf level for the aforementioned BMD patient and control. (D) Histograms showing pixel‐wise distributions of a and κ over the TA muscles of all participants, where BMD patients tend to show larger fibre diameters.

Figure 5

Representative histology images from this study, with zoomed‐in region inset, and examples of intra‐ and inter‐individual distributions. (A) From left to right, example haematoxylin and eosin, laminin and Sirius‐red images, respectively, from a tibialis anterior muscle sample taken from a 48‐year‐old healthy control; and (B) from a 37‐year‐old Becker muscular dystrophy (BMD) patient. (C) An example of semi‐automatic segmentation of the BMD laminin image shown in (B). (D) Histograms showing the distributions of muscle fibre diameters obtained by segmentation of laminin images in (A) and (B); it can be seen that the BMD laminin image contains a broader distribution of fibre sizes, with a trend towards larger fibres, as compared with the control. (E) Boxplots showing the inter‐individual distributions of median laminin‐derived fibre size (left) and the relative area of fibrosis (right), per sample. BMD patients show significantly larger fibre sizes, along with a greater proportion of fibrosis, as compared with controls; however, the inter‐individual variance in median fibre diameter was similar between patients and controls. *Statistically significant group difference at the P = 0.05 level.