1. Clinical characteristics
A total of 16 subjects were identified, who were enrolled in this study. 9 patients diagnosed with neurosyphilis. The mean age was 53.78±14.02 years (range: 27-71 years). All patients were non-penicillin-allergic patients who received benzathine penicillin and achieved serological cure or serofast after 6-12 months treatment. In the control group, 7 patients with NIND were recruited from the same hospital. Of them, 2 were diagnosed with migraine, 2 with epilepsy, and 3 with mild cognitive impairment. The mean age was 50.29±12.87 years (range: 27-63 years). TRUST and TPPA tests showed negative. Other infectious diseases (eg. HIV) and autoimmune diseases were excluded from above patients by other tests. Further details are displayed in Table 1.
2. Cytokines antibody array
CSF and serum samples were used for the screening of 80 distinct human chemokine and inflammation cytokines using a pre-fabricated, commercially available capture antibody array, including 18 samples from neurosyphilis before and after treatment (Pre- and Post-), 7 samples from NIND. Detailed information on cytokines is provided in Supplementary Table S1.
The original data obtained from array screen was processed by Raybiotech software for background removal and normalization, was used to generate heatmaps, comparing NIND and neurosyphilis (Pre- and Post-), using OriginPro (Fig. 1A). The heatmap showed that there were significant differences of some cytokines among the three groups at CSF levels, but not at serum level (data not shown).
To better visualize these differences, the volcano plots were generated in GraphPad, plotting fold change (FC) in Pre-therapy vs. NIND, Pre- vs. Post-therapy, Pre-therapy vs. NIND respectively, against the corresponding p-value , the significance threshold of p<0.05 and FC>2 (Fig. 1B). In the CSF level, 11 cytokines (CXCL13, CCL22, G-CSF, CXCL6, CCL3, CXCL10, CCL19, IL-12p40, IL-1a, CCL1, CXCL5) were 5fold change higher and 16 cytokines (ICAM-1, CXCL9, Axl, TARC, TNF RII, IL-7, CCL11, CCL4, CCL8, IL-17, TNFβ, IFN-γ, CXCL12, GM-CSF, CCL24, IL-10) were 2fold change higher in neurosyphilis compared to NIND. After treatment, 5 cytokines (CXCL13, CCL22, G-CSF, CXCL6, CXCL9) were 5fold change decreased and 12 cytokines (IL-12p40, CCL19, CCL27, CCL3, CCL1, ICAM-1, CXCL5, TNF RII, CCL4, IFN-γ, IL-10, TNFβ) were 2fold change decreased in post-therapy compared to Pre-therapy. Moreover, 10 cytokines (CXCL13, CCL22, CXCL10, CCL16, LIGHT, CXCL6, CCL3, CXCL5, CCL11, TARC) were higher and 1 cytokine (CCL27) lower in post-therapy than those in NIND. In the serum level, 3 cytokines (TSLP, CCL20, IL-31) were 2fold change lower and 1 cytokine (OPN) higher in neurosyphilis compared to NIND, and 1 cytokine (CCL20) were 2fold change higher in post-therapy compared to Pre-therapy. After treatment, 2 cytokines (TSLP, IL-31) were lower and 1 cytokine (OPN) higher than those in NIND. Supporting data are detailed in Supplementary Table S2.
Concentrations of all CSF cytokines were analyzed by comparing values between the different groups, using the Mann-Whitney U test, or T test. Multiple cytokines (26/27) were overexpressed in neurosyphilis compared to NIND (Fig. 2A), The expression level of IL-1a increased but not significantly (p=0.3577). After antibiotic therapy, the expression of cytokines (15/17) significantly decreased (Fig. 2B), the expression level of 2 cytokines decreased but not significantly (CXCL9, p=0.0977; IFN-γ, p=0.0547). Regrettably, no significantly changes were observed in the serum samples (data not shown). Supporting data are detailed in Supplementary Table S3.
3. Correlation analyses between cytokines
We investigated correlations among CSF cytokine concentrations themselves. Among 10 out of 80 cytokines were closely correlated with other cytokines (≥40 were correlated), most of them were positively correlated. The correlations were limited to 15 cytokines ( ≤10 were correlated), which CXCL9ht be regulated more independently. CCL25 was negatively correlated with 8 cytokines, among which the expression of CXCL9 and CCL24 was significantly increased in neurosyphilis patients, suggesting a downregulation of CCL25 during neurosyphilis inflammation (Fig. 3). Supporting data are detailed in Supplementary Table S4.
We also examined cytokine correlations between CSF and serum concentrations in order to discriminate to what extent a passive transfer from the periphery into the CSF, or vice versa, might occur. Only 6/80 cytokines showed a significant correlation between CSF and serum values, CCL16, CCL23 and IL-13 concentrations between CSF and serum were positively correlated, NAP-2, IL-6R and IL-12p40 concentrations between CSF and serum were negatively correlated (Fig S1). Supporting data are detailed in Supplementary Table S5.
4. Correlation analyses between cytokine concentrations and CSF parameters
We performed correlations among CSF cytokine concentrations and CSF parameters (Fig. 4). Concerning standard CSF parameters, CSF WBCs significantly correlated with 42 cytokines, Qalb significantly correlated with 37 cytokines and CSF-protein significantly correlated with 36 cytokines. The negative correlations were observed between 22 cytokines and Glucose levels. To evaluate distinct effects between cytokines and immune cell subsets, we analyzed the percentage distribution of immune cell subsets. Granulocyte showed a significantly negative correlation with IL-1ra, IL-2, IL-11 and IL-12p40. Lymphocyte showed a significant positive correlation with CXCL16 and TIMP-2. Monocyte showed a significant positive correlation with 11cytokines (CXCL5, CXCL6, NAP-2, PF4, CCL24, G-CSF, ICAM-1, IL-6, IL-12p4, CXCL9 and CCL5) and negatively correlated with CXCL16 and Lymphotactin. Supporting data are detailed in Supplementary Table S6.
5. Diagnostic and therapeutic effect values of cytokines for neurosyphilis.
Given the marked elevation of CSF G-CSF, ICAM-1, CCL4, CCL19, CCL1, IL-12p40 and TNF RII in neurosyphilis patients, significant reduction after treatment, and significantly correlated with CSF parameters, we further evaluated these cytokines as biomarkers in neurosyphilis diagnosis by the ROC curve analysis. The AUCs of G-CSF, ICAM-1, CCL4, CCL19 and IL-12p40 in CSF were all greater than 0.9, which were 0.9615, 0.9583, 0.9167, 0.9236 and 0.9127, respectively (Fig. 5). And the optimal cut-off values were defined by the sum of maximum sensitivity and specificity, which were 146.6 pg/ml, 2974 pg/ml, 3.283 pg/ml, 1961 pg/ml and 2.428 pg/ml, respectively. Supporting data are detailed in Supplementary Table S7.