| Contributors: |
Cagol, A.; Cortese, R.; Barakovic, M.; Schaedelin, S.; Ruberte, E.; Absinta, M.; Barkhof, F.; Calabrese, M.; Castellaro, M.; Ciccarelli, O.; Cocozza, S.; De Stefano, N.; Enzinger, C.; Filippi, M.; Jurynczyk, M.; Maggi, P.; Mahmoudi, N.; Messina, S.; Montalban, X.; Palace, J.; Pontillo, G.; Pröbstel, A. K.; Rocca, M. A.; Ropele, S.; Rovira, À.; Schoonheim, M. M.; Sowa, P.; Strijbis, E.; Wattjes, M. P.; Sormani, M. P.; Kappos, L.; Granziera, C. |
| Description: |
IMPORTANCE Multiple sclerosis (MS) misdiagnosis remains an important issue in clinical practice.OBJECTIVE To quantify the performance of cortical lesions (CLs) and central vein sign (CVS) in distinguishing MS from other conditions showing brain lesions on magnetic resonance imaging (MRI).DESIGN, SETTING, AND PARTICIPANTS This was a retrospective, cross-sectional multicenter study, with clinical and MRI data acquired between January 2010 and May 2020. Centralized MRI analysis was conducted between July 2020 and December 2022 by 2 raters blinded to participants' diagnosis. Participants were recruited from 14 European centers and from a multicenter pan-European cohort. Eligible participants had a diagnosis of MS, clinically isolated syndrome (CIS), or non-MS conditions; availability of a brain 3-T MRI scan with at least 1 sequence suitable for CL and CVS assessment; presence of T2-hyperintense white matter lesions (WMLs). A total of 1051 individuals were included with either MS/CIS (n = 599; 386 [64.4%] female; mean [SD] age, 41.5 [12.3] years) or non-MS conditions (including other neuroinflammatory disorders, cerebrovascular disease, migraine, and incidental WMLs in healthy control individuals; n = 452; 302 [66.8%] female; mean [SD] age, 49.2 [14.5] years). Five individuals were excluded due to missing clinical or demographic information (n = 3) or unclear diagnosis (n = 2).EXPOSURES MS/CIS vs non-MS conditions.MAIN OUTCOMES AND MEASURES Area under the receiver operating characteristic curves (AUCs) were used to explore the diagnostic performance of CLs and the CVS in isolation and in combination; sensitivity, specificity, and accuracy were calculated for various cutoffs. The diagnostic importance of CLs and CVS compared to conventional MRI features (ie, presence of infratentorial, periventricular, and juxtacortical WMLs) was ranked with a random forest model.RESULTS The presence of CLs and the previously proposed 40% CVS rule had a sensitivity, specificity, and accuracy for MS of 59.0% (95% CI, 55.1-62.8), 93.6% ... |