The predictive value of CXCL13 in suspected Lyme neuroborreliosis: a retrospective cross-sectional study.

The role of CXCL13 as a marker of Lyme neuroborreliosis (LNB) is under investigation, and CXCL13 is not part of routine diagnostics in suspicion of LNB. Our aim was to find the optimal cut-off value of CXCL13 for LNB in a Danish population and to investigate the role of CXCL13 both in early LNB and as a discriminatory marker between LNB and other neuroinflammatory disorders. We conducted a retrospective cross-sectional study including all patients with a cerebrospinal CXCL13 test performed at the Department of Clinical Immunology, Odense University Hospital, Denmark, between 1 January 2015 and 31 December 2018. We included 619 patients, of which 51 had definite LNB, 14 patients had possible LNB with neurological symptoms suggestive of LNB and pleocytosis but no intrathecal Borrelia antibodies, eight patients had prior LNB and 546 had no LNB. With an optimal CXCL13 cut-off of 49 ng/L we found a sensitivity of 100% and specificity of 94% (AUC 0.988, 95% CI 0.980-0.996) when patients treated with antibiotics prior to lumbar puncture were excluded (n = 130). All patients with possible LNB had a CXCL13 value above the cut-off value; 18/546 patients (3.3%) without LNB had a CXCL13 value ≥ 50 ng/L. While CXCL13 cannot be used as a stand-alone test, it can be used as a reliable additional marker in treatment-naive patients suspected of LNB. CXCL13 can be used to monitor treatment response in LNB patients.