Patients’ characteristics
The median age of the patients with HBV-ACLF was about 43 years. During the follow-up period, 25 patients with HBV-ACLF survived, while 15 died. Thus, the overall mortality rate was 37.5%. Twenty-nine (72.5%) patients with HBV-ACLF were clinically diagnosed with cirrhosis before enrollment. The baseline characteristics of the participants are shown in Table 1. No significant differences existed among the three groups in age (P=0.170) or gender (P=0.877).
Luminex detected the serum level of forty-eight inflammatory factors in patients with HBV-ACLF
We measured the serum concentrations of 48 cytokines by luminex among three groups. (Figure 1 A). The level of IL-6, IL-10, IL-15, IL-18, M-CSF, IP-10 and CXCL9 were significantly higher in patients with HBV-ACLF than in either patients with CHB or NC subjects. While the level of EGF, PDGF-AA, PDGF-AB/BB, MDC and sCD40L were significantly lower in patients with HBV-ACLF than other groups. (Figure 1 B)
We then determined the correlation between clinical outcome and these 12 cytokines and found four cytokines with predictive value including IL-6, IL-15, CXCL9 and MDC.
IL-6 correlated with disease severity in HBV-ACLF
The serum level of IL-6 was significantly higher in non-surviving patients (median 14.41pg/ml, range 1.07–114.91pg/ml) than in surviving patients (median 6.98 pg/ml, range 1.53–15.10 pg/ml, P<0.01; Figure 2 A). While the serum levels of IL-15, CXCL9 and MDC were statistically different between surviving groups and non-surviving groups. We subsequently analyzed the correlation between the serum levels of these cytokines and PTA levels, INR levels and disease severity parameter like MELD and MELD-Na scores. Interestingly, positive correlations were found between IL-6 serum level and MELD score (r=0.359, P=0.0234), MELD-Na score (r=0.365, P=0.0204; Figure 2A). In addition, MDC serum level had negative correlation with MELD-Na score (r=-0.386, P=0.014; Figure 2C). Besides, CXCL9 serum level seem to have no correlation with MELD and MELD-Na score, while correlated with PTA (r=-0.365, P=0.022) and INR (r=-0.395, P=0.012; Figure 2 D). However, IL-15 have no correlation with these parameters. Collectively, these findings indicated IL-6, MDC, CXCL9 may have more predictive value.
Increased serum level of IL-6 and CXCL9 indicated poor prognosis in patients with HBV-ACLF
The ROC curve analysis was used to evaluate the value of four cytokines mentioned earlier in predicting prognosis (Figure 3A). In addition, the ES index was used to identify the optimal cutoff value, defined as the value that maximized the sensitivity and specificity. The ROC curve analysis results were shown in Table 2.The area under the ROC curve (AUROC) of IL-6 was 0.703 (95% CI: 0.507–0.899, P=0.034), CXCL9 was 0.691 (95% CI: 0.510–0.871, P=0.046). Compared with the MELD score (AUROC=0.765, P=0.005), MELD-Na score (AUROC=0.779, P=0.004), IL-6 and CXCL9 still had a promising prognostic value.
Patients were then divided into two groups by each cut-off value, including IL-6, CXCL9, MELD, MELD-Na. The 90-day survival rate was examined by the Kaplan–Meier analysis between higher group and lower group. The log-rank test revealed significant differences between the higher group and the lower group in 90-day survival rate (IL-6, P=0.032; CXCL9, P=0.012; MELD, P=0.001; MELD-Na, P<0.001; Figure 3B). Collectively, these data suggested that an increased frequency of IL-6 and CXCL9 could be useful predictors of mortality in patients with HBV-ACLF.
Increased serum level of IL-6 could be an independent predictor of mortality
Baseline clinical and laboratory variables were analyzed as possible predictors of mortality. The basic characteristics of surviving and non-surviving patients with HBV-ACLF are summarized in Table 3. Compared with surviving patients, non-surviving patients were had higher levels of WBC, INR, Cr and PCT, had lower levels of PLT and Na (Table 3). Next, the Cox regression analysis was used to identify predictors for HBV-ACLF. In the univariate analysis, the serum level of IL-6, IL-15, CXCL9, WBC, PLT, INR, Na, CR, PCT, MELD score, MELD-Na score were factors associated with a higher risk of mortality (Table 4). Next, we evaluated these significant variables in a multivariate Cox regression analysis by using forward stepwise (likelihood ratio) selection. Only IL-6 (HR=1.026, P=0.025), WBC (HR=1.344, P<0.001), and PLT (HR=0.979, P=0.009) were found to be independent baseline predictors of mortality in patients with HBV-ACLF (Table 4).