Quantitative assessment of diabetic amyotrophy using magnetic resonance neurography—a case-control analysis

Objectives: To quantitatively characterize diabetic amyotrophy (DA), or diabetic lumbosacral radiculoplexopathy, and compare with controls using magnetic resonance neurography (MRN). Methods: Forty controls and 23 DA cases were analyzed qualitatively and quantitatively. Cross-sectional areas (CSAs) of bilateral L3 through S2 lumbosacral nerve roots, femoral nerves, and sciatic nerves (proximal and distal measurements) were measured. A linear model was used to assess the nerve location and case/control effect on angle-corrected CSAs. Intra- and inter-reader analysis was performed using intraclass correlation (ICC). Results: In DA cases, abnormalities of the lumbosacral nerve roots, sciatic, femoral, and obturator nerves were seen in 21/23, 16/23, 21/23, and 9/23, respectively. Denervation abnormalities of multiple abdominopelvic muscles were seen. Quantitatively, the CSA of all measured LS plexus nerve roots and bilateral femoral nerves were significantly larger in DA cases vs. controls by 45% (95% CI, (30%, 49%); p < 0.001). The ICC was moderate for inter-rater analysis = 0.547 (95% CI, 0.456–0.626) and excellent for intra-rater analysis = 0.90 (95% CI, 0.89–92). Conclusions: Multifocal neuromuscular lesions related to diabetic amyotrophy were qualitatively and quantitatively detected on MRN. Qualitative abnormalities distinguished cases from controls, and nerve CSAs of cases were significantly larger than those of controls. Therefore, MRN may be employed as a non-invasive diagnostic tool for the evaluation of diabetic amyotrophy. Key Points: • Qualitative abnormalities of lumbosacral nerve roots, their peripheral branches, and muscles are seen in DA. • The lumbosacral nerve roots and their peripheral branches in diabetic amyotrophy cases are significantly larger in cross-sectional area than non-diabetic subjects by 45% (95 CI, 30%, 49%; p < 0.001). • The ICC was moderate for inter-rater analysis = 0.547 (95% CI, 0.456–0.626) and excellent for intra-rater analysis = 0.90 (95% CI, 0.89–92).