Autoimmune encephalitis: assay choice, pitfalls and predictive value

Abstract Autoimmune encephalitides are rare, antibody-defined and treatable, and commercial assays have moved their diagnosis from specialised research laboratories into routine practice. Over 11 years in our laboratory the number of antibody analyses rose tenfold and the number of analytes per patient rose by 40 %. The proportion of patients with a positive result nevertheless remained near 6 %. The additional analytes identified no additional patients, and the already low pre-test probability per analyte has fallen further. Written for the laboratory, this review aims to provide practical guidance on raising the overall specificity and predictive value of the final report. It sets out what each assay format can and cannot do, which specimen to trust for which antibody, when a titre carries information, how to handle a result that only one technique supports, and which sample belongs in a referral laboratory. It also covers preanalytics, reader competence, external quality assessment and the regulatory framework for in-house assays. In general, the predictive value of a positive result is set not by the specificity of a single assay but by the whole pathway: serum and cerebrospinal fluid (CSF) read together, positives confirmed by a second technique, discordant findings referred rather than reported, and every result matched against the clinical syndrome. We propose graded levels of diagnostic certainty for common result constellations. In the future, reliable neuronal antibody diagnostics will depend less on new antigens than on disciplined reporting: internal training, consensus reading, quality control and dialogue with the requesting clinician.