Sensitivity of the UT-Hopkins Score in 350 patients with antibody-mediated autoimmune encephalitis

Early recognition of autoimmune encephalitis (AE) is critical but remains challenging because of overlapping clinical features with viral encephalitis (VE) and delays in antibody testing. The UT-Hopkins Score was recently developed to assist in the clinical differentiation of AE from VE at presentation. We aimed to assess the sensitivity of this score in a large national cohort of antibody-mediated AE and across major antibody-defined subgroups.

We conducted a retrospective cohort study including 350 patients with antibody-mediated AE diagnosed according to international consensus criteria at the French Reference Center. Antibodies included anti-N-methyl-D-aspartate receptor (NMDAR), leucine-rich glioma-inactivated 1 (LGI1), contactin-associated protein-like 2 (CASPR2), γ-aminobutyric acid type B receptor (GABA_BR), and glutamic acid decarboxylase 65 (GAD65). The UT-Hopkins Score was calculated using four binary variables. Sensitivity was assessed using a cutoff of ≥ 2 points and analyzed overall, in a reweighted model reflecting national antibody distribution, and across antibody subtypes.

Overall, 90.3% (95% CI, 86.7-93.0) of patients scored ≥ 2. In the reweighted model, sensitivity increased to 92.9% (95% CI, 86.3-96.6). Sensitivity was highest in GAD65 (100%) and NMDAR (99%) subgroups, remained high in LGI1 (94%) and CASPR2 (90%), but was lower in GABA_BR (56%), characterized by higher comorbidity burden, acute onset, fewer psychiatric or memory features, and robust CSF inflammation.

In this large cohort of antibody-mediated AE, the UT-Hopkins Score demonstrated high sensitivity across most antibody subtypes, supporting its use to detect autoimmune causes early in patients presenting with encephalitis.