Magnetic resonance neurography more often provided perceived additional diagnostic value—defined as lesion detection missed by the other modality or clinically meaningful additional characterization—than high-resolution ultrasound in patients with suspected upper-extremity peripheral neuropathies involving proximal lesions, multiple anatomical regions, or multiple nerves, according to an exploratory subanalysis published in European Radiology.
Researchers conducted a retrospective subanalysis of a previously published prospective observational cohort involving 800 consecutive patients referred to a single tertiary center between November 2015 and February 2022 for both high-resolution ultrasound (HRUS) and magnetic resonance neurography (MRN) because of suspected upper-extremity neuropathy. Among the full cohort, 479 patients (60%) had concordant findings between the 2 modalities, whereas 321 patients (40%) had discordant findings. Following exclusion of 46 non–true-positive cases, 275 patients met criteria for perceived additional diagnostic value from 1 modality.
MRN accounted for 261 of those cases (94.9%), whereas HRUS accounted for 14 cases (5.1%). Among the MRN-favored cases, 136 involved negative HRUS findings with positive MRN findings, while 125 cases showed abnormalities on both modalities but additional clinically relevant characterization by MRN.
To identify predictors associated with differential imaging yield, researchers performed multivariable logistic regression analysis. Lesion topography emerged as the strongest predictor of added value from MRN. Compared with peripheral-only lesions, proximal-only lesions were associated with 5.72 times the odds of added diagnostic value from MRN (95% confidence interval [CI] = 3.42–9.56), while combined proximal and peripheral lesions were associated with 4.59 times the odds (95% CI = 2.31–9.09). Multi-anatomical-region involvement (OR = 1.81, 95% CI = 1.14–2.87) and multi-nerve disease (OR = 1.65, 95% CI = 1.03–2.63) also independently favored MRN.
A classification and regression tree model identified multi-nerve involvement as the first major branching point associated with added value from MRN, followed by lesion distribution within the multi-nerve subgroup. The multivariable model achieved an area under the receiver operating characteristic curve of 0.76, and the classification tree demonstrated a mean cross-validated accuracy of 76.1%.
Clinical symptoms were less predictive than lesion anatomy. Although motor deficits were associated with additional MRN value in univariable analysis, they were not independently associated in the final multivariable model once anatomical distribution and lesion extent were considered. Sensory deficits, pain, prior surgery, and abnormal electrophysiologic findings also were not independently associated with additional MRN value.
HRUS provided additional diagnostic value in fewer than 2% of the overall cohort, primarily when MRN was limited by metal artifacts related to implants or osteosynthetic material. Researchers also noted that HRUS remained useful for superficial focal lesions and for dynamic assessment of periarticular nerve segments during movement, which may be relevant in entrapment neuropathies.
The investigators emphasized that the findings should not be interpreted as evidence of objective diagnostic superiority of MRN over HRUS. The study was conducted at a single tertiary referral center with advanced imaging expertise, and HRUS performance may vary substantially because of operator dependency. In addition, the study population was limited to upper-extremity neuropathies, and the findings may not generalize to lower-extremity or generalized peripheral neuropathies.
The composite reference diagnosis incorporated imaging findings alongside clinical examination, electrodiagnostic testing, and, when available, surgical or histopathologic confirmation. The researchers noted this design may introduce incorporation bias and limit conclusions regarding intrinsic diagnostic accuracy.
“These exploratory results support an individualized imaging approach with MRN to be favored in complex or ambiguous clinical scenarios, and HRUS as a complementary or alternative tool when MRN is unavailable, contraindicated, or metal artifact-prone,” wrote Merle Brunnée, of Heidelberg University Hospital in Heidelberg, Germany, and colleagues.
Disclosures: The researchers reported no conflicts of interest related to the study. Open Access funding was enabled and organized by Projekt DEAL. Some study subjects overlapped with a previously reported cohort.
Source: European Radiology
