Sensitivity and specificity of electrodiagnostic criteria for CIDP using ROC curves: comparison to patients with diabetic and MGUS associated neuropathies

Abstract

Introduction: Electrodiagnostic testing plays a key role in the characterization of neuropathies. To this end, sets of electrodiagnostic criteria have been proposed to define chronic inflammatory demyelinating polyneuropathy (CIDP). These criteria sets differ because of the number of data points within the sets, the number of required abnormal measures within a criterion, and the cutoff value of each measure.

Objective: To evaluate the sensitivity and specificity of the published criteria for defining CIDP in comparison to diabetic polyneuropathy (DMPN) and monoclonal gammopathy of undetermined significance associated neuropathies (MGUS-PN).

Design/methods: Electrodiagnostic studies of 21 patients with biopsy proven CIDP, 35 patients with MGUS-PN, and 82 patients with DMPN were analyzed. Data were compared against 4 different published criteria sets. Receiver operator characteristic (ROC) curves were used to determine the ideal threshold values for individual electrodiagnostic parameters.

Results/conclusion: None of the currently published criteria sets could adequately separate CIDP from DMPN and MGUS-PN. Analysis of our data using ROC curves shows that the best discrimination was achieved using the following criteria: (1) FWL>145%ULN in one nerve or 110%ULN in two nerves in separate nerve roots. (2) Motor CV<90%LLN in four nerves with at least one nerve <70%LLN. Furthermore, proximal-to-distal amplitude ratio alone is not adequate to define conduction block. Although electrodiagnostic studies are important for evaluating CIDP, these studies by themselves cannot be used to define this neuropathy.