Two Spinal-Tap Markers Could Change How Doctors Diagnose a Hidden Brain Disease

In a German multicenter study of 92 patients followed for about three years, two spinal-tap markers gave neurologists something they had not had before in autoimmune encephalitis: a paired read on whether the disease is present and how severe it is likely to run.

Autoimmune encephalitis is a rare, severe disorder in which the immune system attacks the brain. It often strikes young adults, mimics psychiatric illness or viral infection, and is still misdiagnosed or diagnosed late. The new study, published in Neurology: Neuroimmunology & Neuroinflammation and drawn from the German Network for Research on Autoimmune Encephalitis and the Hannover Medical School cerebrospinal fluid (CSF) biobank, screened 2,330 patients and enrolled 92 adults with definite disease across the three most common antibody-defined forms: 53 with anti-NMDAR (antibodies against the N-methyl-D-aspartate receptor), 20 with anti-LGI1 (leucine-rich glioma-inactivated 1), and 19 with anti-CASPR2 (contactin-associated protein-like 2) encephalitis. Median follow-up was 38 months.

The first marker is kappa free light chain (KFLC), a protein fragment produced inside the brain when B cells are activated locally. Intrathecal KFLC synthesis, meaning the brain is making the protein rather than receiving it from blood, appeared in 94% of anti-NMDAR cases, 50% of anti-LGI1 cases, and 53% of anti-CASPR2 cases. Those rates beat two older CSF tests: oligoclonal bands, which look for restricted antibody patterns in the spinal fluid, and pleocytosis, a simple cell count. KFLC caught more cases of NMDAR encephalitis than either.

The second marker is neurofilament light chain (NfL), a structural protein that leaks from damaged neurons. It is the same protein Siemens Healthineers markets in a CE-marked blood-based assay currently framed around multiple sclerosis. In the German cohort, each 1-standard-deviation rise in CSF NfL z-score, a normalized measure of how far a patient's value sits above the reference mean, tracked to roughly a 10-point higher score on the CASE (Clinical Assessment Scale in Autoimmune Encephalitis) severity scale at baseline, independent of age, sex, antibody subtype, and other clinical factors (β = 0.61). NfL was also followed longitudinally, so the same patient can be tracked over time rather than scored once.

Pairing the two gives clinicians a workflow rather than a single test. KFLC answers "is there intrathecal inflammation consistent with autoimmune encephalitis right now, even before antibody results return?" NfL answers "given that the disease is here, how much neuronal damage has already happened, and is it getting worse?" For patients and families, that means faster treatment decisions and more honest conversations about likely course. For clinicians, it is an add-on to antibody testing, not a replacement.

Pooled across all control groups, which included relapsing multiple sclerosis, varicella zoster encephalitis, noninflammatory neurologic disease, and antibody-positive non-AE cases, KFLC's diagnostic specificity was 44%. That is modest. The 90% specificity figure that circulates in marketing copy applies only to the contrast with noninflammatory neurologic controls, which is the easier comparison. Against multiple sclerosis or viral encephalitis, KFLC adds signal but does not separate the diseases on its own. The clinically honest read: KFLC is good at flagging inflammation inside the central nervous system and is especially strong in NMDAR disease; it is not a stand-alone rule-in test.

The access picture is mixed. Mayo Clinic Labs offers KFLC as KCSFP, a nephelometry test (a light-scattering method for measuring protein concentration), but only as part of the MSP3 multiple sclerosis cascade, which is a real ordering constraint for clinicians who want the test alone. NfL is more widely available, though the Siemens assay is currently positioned for MS workflows rather than AE. A clinician trying to use either test for autoimmune encephalitis today is essentially off-label in U.S. commercial lab catalogs.

The cohort is registry-based and German, so generalizability to U.S. emergency departments and to antibody subtypes outside the three studied is an open question. The authors also point to longitudinal NfL trajectories in the abstract; the full text will need to confirm that the β = 0.61 effect holds across follow-up rather than only at baseline. Adjacent work, including a meta-analysis of NfL in neuronal surface antibody AE and a long-term outcomes study in LGI1 encephalitis, supports the broad direction without changing the headline: a CSF-only dual-marker read, paired with antibody testing, is the most concrete diagnostic and prognostic gain autoimmune encephalitis has had in years.

The next step is a prospective U.S. or multi-country cohort that re-runs the same pairing in real-world emergency settings. Until then, the practical message for clinicians is narrower than the press release: order KFLC when you suspect NMDAR encephalitis and the antibody panel will take days, and follow NfL as a longitudinal track once disease is confirmed.