Spinal Fluid Antibody Cutoff of 1 nmol/L Separates True GAD65 Neurological Disease from Incidental Positivity

A single number measured in spinal fluid may settle one of the most stubborn diagnostic puzzles in autoimmune neurology. In a retrospective cohort study at Mayo Clinic, researchers report that a cerebrospinal fluid GAD65-antibody level of 1 nmol/L or higher, measured by radioimmunoassay, best separated patients with true GAD65-antibody neurological disease from those whose antibody positivity was clinically irrelevant. The finding, published in Annals of Clinical and Translational Neurology, arrives amid rising concern that indiscriminate autoantibody testing is generating misdiagnoses and harmful treatments in patients who never had the disease at all.

Antibodies against glutamic acid decarboxylase 65, an enzyme involved in synthesizing the neurotransmitter GABA, have long occupied an awkward position in clinical practice. True GAD65-antibody neurological disease shows discrete recognized manifestations, notably stiff person spectrum disorder, cerebellar ataxia, encephalitis, and temporal lobe epilepsy. Yet many patients carry the antibody without any of these syndromes, and the antibody can appear incidentally in people with entirely different diagnoses. The study’s authors note that indiscriminate testing has made clinically irrelevant antibody detections the commonest precipitant of misdiagnosis and iatrogenic harm in this field, with GAD65 among the most problematic targets.

Current practice leans heavily on serum antibody titers. A serum GAD65-antibody level exceeding the neurological cutoff of more than 20 nmol/L by radioimmunoassay, or more than 10,000 IU/mL by enzyme-linked immunosorbent assay, is considered more accurate for identifying true disease. But the new data expose how unreliable that threshold can be. Among 50 patients whose GAD65 positivity turned out to be incidental, 24, or 48 percent, had serum titers above the 20 nmol/L threshold. The serum measurement achieved a sensitivity of 0.983, meaning it almost never missed a true case, but a specificity of only 0.520, meaning it flagged roughly as many irrelevant positives as relevant ones at that cutoff.

The study included 167 adults seen at Mayo Clinic who had positive serum and cerebrospinal fluid GAD65-antibodies, tested in samples collected less than a week apart. Most were female, 119 of 167 or 71.3 percent, and White, 117 of 167 or 70.1 percent, with symptom onset at a median age of 50 years and a range of 18 to 80. True GAD65-antibody disease accounted for 70 percent of the cohort, 117 of 167 patients, manifesting as stiff person spectrum disorder in 42 cases, cerebellar ataxia in 25, temporal lobe epilepsy in 24, limbic encephalitis in 8, and combinations of these syndromes in 18. The remaining 50 patients had an alternative confirmed diagnosis explaining their illness, defining the incidental positivity group.

To build a diagnostic tool, the investigators split patients into a discovery cohort from January 2021 to December 2025 and a validation cohort from January 2016 to December 2020. They classified true disease using established phenotypes, existing diagnostic criteria for stiff person spectrum disorder and limbic encephalitis, or very high serum titers above 20 nmol/L confirmed by two independent autoimmune neurologists. Incidental positivity required an alternative confirmed diagnosis. The researchers then compared serum and spinal fluid antibody titers, along with calculated ratios and indices intended to reflect intrathecal synthesis, the production of antibodies within the nervous system itself.

The spinal fluid measurement outperformed every alternative. Cerebrospinal fluid GAD65-antibody titers were markedly higher in true disease, with a median of 8.1 nmol/L in the discovery cohort and 4.3 nmol/L in the validation cohort, compared with medians of 0.2 and 0.16 nmol/L in the incidental groups. A cutoff of 1 nmol/L or greater achieved a combined sensitivity of 0.897 and specificity of 0.940 across both cohorts. In receiver operating characteristic analysis pooling discovery and validation data, the spinal fluid model produced an area under the curve of 0.950, with a 95 percent confidence interval of 0.915 to 0.996 and an odds ratio of 14.1. The serum model performed respectably at 0.858 but was statistically inferior to its spinal fluid counterpart.

The derived intrathecal synthesis metrics told a different story. The GAD65-antibody ratio, the GAD65-antibody index, and the GAD65-antibody to IgG ratio, all calculated from paired serum and spinal fluid values, showed less robust or no statistically significant differences between true and incidental disease. Among routine spinal fluid constituents, only oligoclonal bands consistently differed between the groups, elevated in 36 of 84 true disease patients, or 42.9 percent, versus 4 of 37 incidental patients, or 10.8 percent, yielding a sensitivity of 0.429 and specificity of 0.892. Correcting the models for oligoclonal band status, prior or current immunotherapy exposure, or coexisting autoimmunity changed neither the cutoffs nor their performance.

The cutoff also rescued cases the serum threshold would have missed. In 12 of the 117 true disease patients, or 10.3 percent, serum titers fell below the 20 nmol/L threshold, yet 11 of those 12 had spinal fluid levels above 1 nmol/L, ranging from 11.2 to 19.8 nmol/L. Across the combined cohort, applying the spinal fluid cutoff agreed with the clinical diagnosis in 152 of 167 cases, or 91 percent, with a positive predictive value of 97.2 percent and a negative predictive value of 79.7 percent. The authors found no clinical pattern among the discordant cases. Notably, the only true disease patient with titers below both the serum and spinal fluid thresholds had received four immunotherapies before paired antibody testing was performed.

The researchers propose the 1 nmol/L spinal fluid radioimmunoassay cutoff as a diagnostic branchpoint when evaluating a patient for true GAD65-antibody disease, adding to prior work that supported a discriminatory value for spinal fluid antibodies. They emphasize that spinal fluid testing does not replace established serum thresholds or clinical judgment but provides an adjunctive method to avoid misdiagnosis, unnecessary workups, and to rationally consider immunotherapy. The cutoff carries utility in two scenarios: confirming phenotypically incompatible cases whose serum levels exceed the established threshold, and, less frequently, diagnosing patients with compatible phenotypes such as stiff person spectrum disorder whose serum antibodies sit in the 10 to 20 nmol/L gray zone.

The study carries important limitations. Clinical charts contained missing values, the population came from a specialty-referral center, and results may not generalize to other detection methods such as ELISA. A potential incorporation bias looms largest: because serum GAD65-antibody titers helped define true disease in the first place, the serum test’s poor showing may be partly an artifact of the study design, and the spinal fluid test’s strong performance may be correspondingly inflated. Even with independent expert review of diagnoses, patients with atypical presentations of true disease may have been labeled as incidental in the absence of other biomarkers. Within these constraints, the authors conclude that a spinal fluid GAD65-antibody cutoff of 1 nmol/L or higher performed very well in diagnosing true neurological disease and may offer diagnostic clarity in clinical practice, potentially sparing patients whose antibody is an innocent bystander from aggressive and unnecessary immune treatment.

Subject of Research: Diagnostic performance of cerebrospinal fluid GAD65-antibody cutoffs for identifying true GAD65-antibody neurological disease

Article Title: Cerebrospinal Fluid GAD65‐Antibody Levels Accurately Identify Relevant Neurological Diseases

Article References: Herman, M. C., Marzello, M., Tharwani, Z. H., Kotaru, S., Piura, Y. D., Newman, E., Day, G. S., McKeon, A., Lopez‐Chiriboga, A. S., & Irani, S. R. (2026). Cerebrospinal Fluid GAD65 ‐Antibody Levels Accurately Identify Relevant Neurological Diseases. Annals of Clinical and Translational Neurology, Article acn3.70578. https://doi.org/10.1002/acn3.70578

Image Credits: AI Generated

DOI: 10.1002/acn3.70578

Keywords: GAD65 antibodies, cerebrospinal fluid, autoimmune neurology, stiff person spectrum disorder, diagnostic cutoff, radioimmunoassay, misdiagnosis, intrathecal synthesis, oligoclonal bands, temporal lobe epilepsy, limbic encephalitis, Mayo Clinic