Practically applicable nerve ultrasound models for the diagnosis of axonal and demyelinating hereditary motor and sensory neuropathies (HMSN).

Purpose: To develop specific diagnostic ultrasound (US) models for hereditary motor and sensory neuropathies (HMSN) in patients with primarily demyelinating or axonal polyneuropathies (PNP) according to standard nerve conduction studies (NCS) criteria.

Methods: Single-centre, examiner-blinded cross-sectional study in acquired PNP (consecutive recruitment strategy) and HMSN patients (convenience sample). Allocation into demyelinating or axonal phenotype via easily applicable NCS criteria. Assessment of single measurements by receiver-operating curve (ROC) analysis, development of diagnostic models based on the best measurement values in ROC.

Results: Of 85 enrolled subjects, 53 (62%) had HMSN and 32 (38%) acquired PNPs, and 60 subjects (71%) had demyelinating and 25 (29%) axonal PNP. ROC area under the curve of means of the z-transformed 5 best measurement values was 0.87 for demyelinating and 0.99 for axonal HMSN. Diagnostic models showed high accuracy for both demyelinating (84% sensitivity, 86% specificity) and axonal HMSN (100% sensitivity and specificity). As a measure of variability of morphologic changes, standard deviations of z-transformed measurements were compared for acquired PNP and HMSN. In contrast to previous reports of more homogenous nerve enlargements in HMSN, standard deviations were higher in HMSN than in acquired PNP. Additionally, the performance of previously published models for the diagnosis of HMSN in demyelinating PNP was compared. Previously published models showed lower sensitivities (50-58%), but comparable specificities (91-100%) when applied to NCS-criteria defined demyelinating PNP group.

Conclusion: Diagnostic ultrasound models for HMSN in patients with demyelinating or axonal neuropathies show high accuracy and can contribute to differential diagnosis in clinical routine.