Comparison of commercial and in-house tissue-based and cell-based assays for the detection of autoantibodies targeting neuronal surface proteins: a prospective cohort study

David Goncalves ¹ Marie Benaiteau ^2,3 Veronique Rogemond ^2,3 Sterenn Closs ^2,3 Anne-Laurie Pinto ^2,3 Maroua Dhairi ^2,3 Marine Villard ^2,3 Géraldine Picard ^2,3 Fabien Nicole ¹ Jerome Honnorat ^2,3*

• ¹ Service d'immunologie biologique, Hôpital Lyon Sud, Hospices Civils de Lyon, Pierre-Bénite, France
• ² Centre de référence des syndromes neurologiques paranéoplasiques et de l'encéphalite auto-immune, Hospices Civils de Lyon, Lyon, Rhône-Alpes, France
• ³ MELIS-UCBL-CNRS UMR5284. INSERM U1314, Université Claude Bernard Lyon 1, Lyon, France

The detection of antibodies targeting neuronal antigens is a keystone for the diagnosis of autoimmune encephalitis (AE) and paraneoplastic neurological syndromes (PNS). This study aimed to compare the performance of a commercial tissue-based immunofluorescence assay (cIFA) to that of an in-house IFA (hIFA) for the screening of autoantibodies targeting neuronal surface proteins in the cerebrospinal fluid (CSF) and to compare the performance of commercial cell-based assays (cCBA) to that of in-house CBA (hCBA) in serum samples. Between March and June 2021, 2135 CSF samples and 524 serum samples from 2283 patients referred to the French Reference Center on PNS and AE were prospectively included. CSF samples were all tested using 3 different assays: cIFA, hIFA, and cCBA. Serum samples were all tested using at least 1 cCBA and 1 hCBA for the detection of the following autoantibodies: CASPR2, GABABR, and LGI1. Among the 2135 CSF tested, 93 (4.4%) were positive using both cIFA and hIFA, 1 (0.05%) was positive using only cIFA, and 6 (0.3%) were positive using only hIFA. Among the double-positive samples, 37 (39.8%) were positive using cCBA for the following autoantibodies: anti-NMDAR (n=16), -LGI1 (n=8), -CASPR2 (n=7), -GABABR (n=5), and –DPPX (n=1) autoantibodies. The remaining 56 (60.2%) double-positive samples were negative using cCBA and additional tests were performed to identify the autoantibodies according to the pattern observed on the IFA. The only sample positive using cIFA but negative using hIFA was positive for anti-LGI1 autoantibodies using cCBA. Among the 6 samples negative using cIFA but positive using hIFA, only one sample was positive with cCBA for anti-NMDAR autoantibodies. These data indicate that, in CSF, cIFA and hIFA performed similarly for the detection of autoantibodies targeting neuronal surface proteins. Regarding serum samples, cCBA and hCBA were both positive in 3 patients for CASPR2, 4 patients for LGI1, and 1 patient for GABABR. A positive cCBA and negative hCBA was observed in 2 patients for LGI1 and 4 patients for GABABR. A lack of specificity of GABABR cCBA is suspected as CSF explorations were negative in 3 of these patients and none presented clinical features highly suggestive of AE.