The demographic characteristics, diagnosis, CSF profiles, results of viral detection, magnetic resonance imaging (MRI) findings, lowest score of the Glasgow coma scale (GCS) during the hospitalization period, and modified Rankin Scale (mRS) at discharge are summarized in Table 1. There was no significant difference in age (P=0.8465) or sex (P=0.8586) among the HSV, VZV, and control groups. Serum samples were collected from 6 patients of the HSV group, 3 patients of the VZV group, and 6 patients from the control group (shown as cases with asterisks in Table 1). The raw data on biomarkers are presented in the Supplementary Table.
Levels of the biomarkers in each group are summarized in Figure 1. There were four significant differences on multiple comparison among the three groups: First, the levels of CSF t-tau were significantly higher in the HSV than control group (P = 0.0016, Figure 1D). Second, CSF p-tau levels were significantly elevated in the VZV group (P=0.0102) and HSV (P=0.0487) groups compared with the control group (Figure 1E). Third, the CSF p-tau/t-tau ratio was significantly higher in the VZV compared not only with the control but also the HSV group (P=0.0389 and P=0.0291, respectively, Figure 1F). Fourth, levels of CSF GFAP were significantly increased in the HSV compared with the control group (P=0.0202, Figure 1J). The HSV and VZV groups trended to have higher CSF sTREM2 and lower CSF Aβ1-42 levels compared with the control. However, these trends did not reach significance (Figure 1I and 1A).
For comparison between herpetic CNS infection and control groups, we combined the data of the HSV and VSV groups, and then compared their biomarker values to those of the control (shown in Figure 2A to 2K). In this comparison, we found the following significant differences: The levels of CSF Aβ1-42, Aβ1-40, and the Aβ1-42/Aβ1-40 ratio were significantly lower in the HSV and VSV combined group (HSV/VSV) compared with the control group (P=0.01836, P=0.0380, and P= 0.0262, respectively) (Figure 2A, 2B, and 2C). CSF t-tau, p-tau, CSF sTREM2, and CSF GFAP levels were significantly elevated in the HSV/VSV group compared with the control group (P= 0.0043, P=0.0007, P=0.0030, and P=0.0139, respectively) (Figure 2D, 2E, 2I, and 2J). The HSV/VZV group tended to have higher CSF p-tau/t-tau, CSF NfL, CSF p-NfH and serum NfL levels compared with the control. However, these trends did not reach significance (Figure 2F, 2G, and 2H).
Results of uni- and multivariate regression analyses between those biomarker values and clinical severity are summarized in Table 2. Scatter plots of those are presented in Figure 3 in cases showing significant correlations on univariate analyses as follows; univariate analyses showed significant negative correlations between the lowest score of GCS during the hospitalization and levels of CSF t-tau (Figure 3A), CSF NfL (Figure 3B), CSF p-NfH (Figure 3C), and serum NfL (Figure 3D) (P< 0.001, P=0.014, P=0.027, and P=0.007 respectively). The correlations between GCS and NfL in CSF and serum were still robust after age adjustment (P=0.014 and P=0.030, respectively), while significance of the relationship of GCS to CSF t-tau and CSF p-NfH disappeared after age adjustment. The levels of CSF Aβ 1-42 (Figure 3E), CSF Aβ 1-42/1-40 ratio (Figure 3F), CSF t-tau (Figure 3G), CSF p-tau/t-tau ratio (Figure 3H), CSF NfL (Figure 3I), CSF p-NfH (Figure 3J), and serum NfL (Figure 3K) were significantly correlated with mRS on discharge (P= 0.037, P=0.047, P=0.012, P= 0.004, P=0.002, P=0.002, and P=0.007, respectively). The correlation coefficient of the association between CSF t-tau and mRS on discharge was positive (i.e., high t-tau levels associated with poor prognosis). Consequently, the correlation coefficient between the CSF p-tau/t-tau ratio (equivalent to reciprocal of CSF t-tau) and mRS on discharge was negative. CSF NfL, CSF pNfH and serum NfL were also positively associated with a poor prognosis. On the other hand, CSF Aβ 1-42 and the CSF Aβ 1-42/1-40 ratio were negatively correlated with the mRS scores (i.e., low Aβ 1-42 levels and Aβ 1-42/1-40 ratio associated with poor prognosis). The significant correlation between CSF NfL and mRS on discharge was only preserved after age adjustment (P=0.018) among those biomarkers, similar to the case in GCS.