Anatomic Variation

All axons with the minimum axon diameter visible within the field of view of each image were analyzed. Image size was selected to permit accurate classification of myelin appearance (compact or decompacted laminae) and as a result, larger diameter axons may have been disproportionately excluded from analysis due to their higher likelihood of falling partially outside of the field of view. Axons were classified based on the profile of their myelin sheath; specifically, the extent to which the ensheathing myelin was held in compact layers, or whether separation of the myelin laminae was visible. Classifications were adapted from previous studies and defined as follows: axons with compact myelin were characterized by thick, dark myelin appearance with <10% axon circumference affected by decompaction (Figure 2a); axons with moderately decompacted myelin had 10–35% of the axon circumference affected by decompaction (Figure 2b); axons with severely decompacted myelin had 36–99% of the axon circumference with decompaction (Figure 2c); axons with completely decompacted myelin were fully ensheathed in myelin with the appearance of multiple diffuse layers, encompassing 100% of the axon diameter (Figure 2d); and unmyelinated axons with a regular round or oval appearance without surrounding myelin (Figure 2e) but distinct from glial cell processes, which are elongated in shape. Additional categories were included to describe other observed anatomical variations: paranodal profiles defined as axons with a large axolemmal space; lightly myelinated axons surrounded by thick, compact, low electron density (light-appearing) myelin with <10% axon circumference with observed decompaction; and aberrant profiles with constricted axoplasm showing empty membrane folds and aberrant extracellular compartments [33 (link),34 (link),63 (link),67 (link)].
Axons characterized with ‘compact myelin’ were further analyzed to determine axon diameter, myelin thickness, fiber diameter, and G ratio (axon diameter/fiber diameter). The minimum axon diameter was used, and myelin thickness was measured at that same location (Figure 2f). The reported myelin thickness was the average of the two measurements on each side of the minimum axon diameter. Fiber diameter included the minimum axon diameter plus the myelin sheath and axolemmal space (Figure 2f). These measurements were not collected from axons with decompacted myelin because decompaction distorts the apparent thickness of myelin and fiber diameter and prevents consistent measurement. All image analyses were conducted using FIJI Image J (version 1.53) software with investigators blinded to animal identity and treatment status.

In a first step, centrally trained researchers assessed the quality of the original Digital Imaging and Communications in Medicine (DICOM) images at level T12 and L3. In a second step, the research assistants evaluated the CT scans and selected a single slice at level T12. For this purpose, we used SliceOmatic Software version 5.0 (TomoVision, Montreal, Quebec, QC, Canada). We excluded all images with incomplete depiction of T12 and in case that the muscle tissue was out of range and/or if contrast did not allow discrimination. There were no anatomical variations that led to exclusion.
Muscle and visceral tissue were distinguished from subcutaneous adipose tissue using tissue-specific Hounsfield Unit (HU) ranges and anatomical knowledge. We followed the Alberta protocol [23 (link)] and set Hounsfield ranges to −29 to 150 HU for skeletal muscle, −190 to −30 HU for subcutaneous and intramuscular adipose tissue, and −150 to −50 HU for visceral adipose tissue. Muscles included in the cross-sectional measurements at T12 with different muscle groups, including the erector spinae, latissimus dorsi, external and internal oblique, rectus abdominis and external and internal intercostal muscles. Every slice was evaluated twice to improve the interrater reliability with the aim to achieve >99%. Researchers improved the slicing and measuring criteria after the first round.