We selected 92 CSF and 99 sera from patients with AE and NSAbs (20 samples with each of these Abs: AMPAR, GABA_AR, GABA_BR, IgLON5, LGI1, NMDAR, Caspr2; 19 mGluR5, 17 DPPX, 15 mGluR1), and 50 CSF and 50 sera from individuals without NSAbs to assess the performance of a commercial TBA (Euroimmun). NSAbs were confirmed using highly sensitive and specific in-house TBA (Dalmau et al., Physiol Rev 2017) and CBA. Slides were assessed as “positive” or “negative” by two experienced investigators (FG,JD) and, if discordant, an interrater discussion was performed.

The two raters were concordant in classifying 94% (133/142) of CSF and 88% (131/149) of sera. Among CSF samples, 75% (106/142) were correctly identified, while 19% (27/142) were misclassified: 13 false positive (FP), 14 false negative (FN). As for sera, 66% (98/149) were correctly identified and 22% (33/149) misclassified: 11 FP, 22 FN. The worst performance was observed for NMDAR, GABA_AR, and mGluR5 Abs (not identified in 5/10, 6/10, 5/9 sera and in 4/10, 5/10, 5/10 CSF, respectively). The overall sensitivity of the commercial TBA was 83,7% for CSF and 75,8% for serum, while the specificity was 72,3% for CSF and 72,5% for serum.

The diagnostic performance of a TBA (Euroimmun) for NSAbs is suboptimal. NMDAR antibodies, one of the most frequent NSAbs, can be missed in 50% of patients. Our findings suggest that this TBA should not be used as initial screening for NSAbs.