Objective To determine free light chains (FLC) in cerebrospinal fluid (CSF) and serum, calculate the FLC index, and establish a model to evaluate its application value in diagnosing neurosyphilis.

Methods This study included 87 syphilis patients, comprising 48 neurosyphilis patients (NS group) and 39 non-neurosyphilis patients (n-NS group). CSF and serum samples were collected from all patients to measure kappa free light chains (FLC-κ) and lambda free light chains (FLC-λ). The FLC index, FLC-κ index, and FLC-λ index were calculated. Receiver operating characteristic (ROC) curves were used to evaluate these indices. A neurosyphilis diagnostic model based on the FLC index was established and its visualization was verified.

Results Statistically significant differences were observed between the two groups in serological syphilis screening, CSF nucleated cell count, CSF microprotein, CSF IgG, IgG synthesis rate, IgG index, and albumin quotient (P < 0.05). The areas under the ROC curves (AUC) for the FLC-κ index, FLC-λ index, and albumin quotient were 0.998 (95% CI: 0.990-1.000), 0.988 (95% CI: 0.961-1.000), and 0.635 (95% CI: 0.496-0.751), respectively. In the FLC-κ index model, the FLC-κ index (P < 0.05) significantly affected the diagnosis of neurosyphilis, while in the FLC-λ index model, only the FLC-λ index (P < 0.05) had a significant effect. The training ROC curve AUC for the diagnosis model based on the FLC-κ index was greater than 0.950. Its calibration curve showed good predictive performance when the predicted probability was close to 80%, but poor performance when the predicted probability was 40%-60%. The AUC for the diagnosis model based on the FLC-λ index was 0.926, with good predictive performance when the predicted probability was close to 85%, and poor performance when the predicted probability was 40%-70%.

Conclusion FLC-κ and FLC-λ have high diagnostic performance for neurosyphilis, and the FLC index has certain application value in the diagnosis of neurosyphilis.