Comparison of N-methyl-D-aspartate receptor antibody assays using live or fixed substrates

The diagnostic criteria for N-methyl-D-aspartate receptor antibody (NMDAR-Ab) encephalitis require the presence of CSF antibodies against the NMDAR, whereas serum antibodies are considered specific only if accompanied by CSF antibodies. Current assays include in-house immunochemistry (IHC), or cell-based assays (CBA) which use live (L-CBA) or fixed cells (F-CBA), and commercially available fixed-cells CBA (C-CBA), but these have not been compared in parallel. We compared the L-CBA with F-CBA, C-CBA, and IHC using sera and CSFs archived from > 30,000 received for testing and previously positive by L-CBA. Referring neurologists, if identified, provided "definite" or "unlikely" diagnoses of NMDAR-Ab encephalitis for 31 paired serum-CSF samples and 53 unpaired sera. There was good concordance between paired sera and CSFs, with 13/16 "definite" pairs positive, and 7/8 "unlikely" pairs negative in all in-house assays. In unpaired "definite" sera, L-CBA was most sensitive. However, 19/24 serum samples from "unlikely" patients were positive by L-CBA, with only 5/24 and 1/24 positive by F-CBA and IHC, respectively. In available samples, C-CBA demonstrated high sensitivity for CSF, but surprisingly low sensitivity for serum. Overall, regardless of the technique, CSF results were accurate and easy to interpret, but if CSF is unavailable, negative serum C-CBA results in cases with suspected NMDAR-Ab encephalitis could be repeated by a more sensitive in-house assay. Although these assays are sensitive, particularly for CSF, referral of sera with low pre-test probability should be avoided to reduce clinically-irrelevant "false positive" results.