Diagnostic usefulness of plexus MRI in chronic inflammatory demyelinating polyradiculopathy without electrodiagnostic criteria of demyelination.

Background and purpose The usefulness of plexus magnetic resonance imaging (MRI) in the diagnosis of chronic inflammatory demyelinating polyradiculopathy (CIDP) without definite European Federation of Neurological Societies/Peripheral Nerve Society (EFNS/PNS) electrodiagnostic criteria is currently unclear. Methods Data from consecutive patients with clinical manifestations suggesting CIDP, with or without (CIDP-D and CIDP-ND, respectively) definite EFNS/PNS electrodiagnostic criteria, and referred for plexus MRI in our imaging centre were retrospectively analysed. An expert committee of neurologists compared the level of suspicion of CIDP in CIDP-ND patients to the blinded/unblinded MRI findings. Plexus MRI was reviewed by a neuroradiologist blinded to the final diagnosis. Results In all, 38 patients were assessed with suspected CIDP-ND [7/38 (18%) probable; 13/38 (34%) possible; 18/38 (47%), no EFNS/PNS electrodiagnostic criteria], plus 10 with CIDP-D. Thirty-six of the 38 (95%) fulfilled clinical criteria of CIDP variants, including pure sensory neuropathy in 22/36 (61%). Plexus MRI showed abnormalities in 22/38 (58%) patients including increased nerve signal intensity on T2-weighted images in 22/22 (100%), nerve enlargement in 20/22 (91%) and contrast enhancement in 8/22 (36%). Plexus MRI enabled the expert committee's final diagnosis to be adjusted in 7/38 (18%) patients, and in conjunction with nerve conduction studies was a supportive criterion to classify 7/24 (29%) patients as definite CIDP. MRI abnormalities were more asymmetrical (P = 0.03) and less diffuse (P = 0.1) in CIDP-ND than in CIDP-D. Conclusions Our observations suggest that plexus MRI makes a valuable contribution to the diagnosis of CIDP-ND patients. Further studies are needed to investigate inter-rater reliability of clinical and imaging criteria of CIDP in these patients, and the impact on outcomes.