Diagnostic accuracy of nerve ultrasound in hereditary and sporadic non-entrapment neuropathies.

Abstract

The objective of this study is to compare the diagnostic accuracy of nerve ultrasound (US) and nerve conduction studies (NCS) for acquired non-entrapment peripheral neuropathies (PNP) and hereditary motor and sensory neuropathies (HMSN) in a routine clinical setting. The methods are based on a single-center, prospective, examiner-blinded cross-sectional study on three subject groups of healthy controls, PNP (both enrolled by a consecutive recruitment strategy), and HMSN patients (convenience sample). A clinical reference standard based on the neuropathy impairment (NIS) and neuropathy symptoms scores (NSS) was used for PNP as the external validation criterion. Diagnostic accuracy was assessed by receiver-operating curve (ROC) analyses of single-nerve measurements and logit models. Of a total of 676 consecutively screened subjects, 107 (15.8%) were recruited, of which 36 (33.6%) had a PNP. HMSN group consisted of 53 subjects (30 subjects (56.6%) with genetic confirmation). AUCs of best diagnostic logit models to distinguish between controls and PNP patients were 0.86 for US and 0.97 for NCS corresponding to an equivalent specificity [US 93% (95% CI: 83-98%), NCS 89% (95% CI: 78-95%)], but inferior sensitivity of US [US 56% (95% CI: 35-74%), NCS 97% (95% CI: 84-100%)]. For differentiation between PNP and HMSN, both methods had equivalent AUCs of 0.95 corresponding to similar sensitivities/specificities. Simpler diagnostic models based on measurement protocols feasible for clinical routine revealed similar diagnostic accuracies. US has an inferior sensitivity than NCS for acquired PNP, but comparable specificity. For identification of HMSN in a PNP population, US and NCS show comparable performance.