Patient characteristics
The estimated smallest sample size was 94. After screening, 282 patients with ACLF who fulfilled the inclusion and exclusion criteria were recruited into the study (Fig. 1). The comparison of baseline characteristics of these patients with or without MOF is shown in Table 1. suPAR and baseline characteristics was measured in all participants. Plasma suPAR at admission was significantly higher in patients with MOF than those without MOF (11.9 (9.1–15.5) vs. 16.4 (11.5–24.0) ng/mL; p < 0.001, Table 1). Significant differences between the two groups were also found for the presence of some clinical events such as HRS and hepatic encephalopathy (HE), laboratory data such as white blood cell count (WBC), international normalized ratio (INR) and total bilirubin (Tbil), and prognostic scoring systems such as CTP, MELD and SOFA scores. All patients were followed at the end of the point.
Table 1
Baseline characteristics of ACLF patients.
|
(A)ACLF (n = 216)
|
(B)ACLF-MOF (n = 66)
|
(C)CHB
(n = 14)
|
(D)HC
(n = 14)
|
p- value
A vs B
|
Age (years), Mean (± SD)
|
47.1 ± 12.6
|
46.4 ± 13.5
|
43.1 ± 14.3
|
42 ± 10.1
|
0.69
|
Male (%)
|
185 (85.6)
|
57(86.4)
|
12(85.7%)
|
12(85.7%)
|
0.88
|
Etiology
|
|
|
|
|
0.07
|
HBV (%)
|
191 (88.4)
|
64 (97)
|
14(100)
|
-
|
|
Others (%)
|
25 (11.6)
|
2 (3)
|
-
|
-
|
|
Clinical feature
|
|
|
|
|
|
Ascites (%)
|
170 (78.7)
|
50 (75.8)
|
-
|
-
|
0.61
|
Cirrhosis (%)
|
110 (50.9)
|
31 (47.0)
|
-
|
-
|
0.57
|
UGIB (%)
|
18 (8.3)
|
8 (12.1)
|
-
|
-
|
0.35
|
HRS (%)
|
5 (2.3)
|
9 (13.6)
|
-
|
-
|
< 0.001
|
HE (%)
|
7 (3.2)
|
24 (36.4)
|
-
|
-
|
< 0.001
|
SBP (%)
|
9 (4.2)
|
6 (9.1)
|
-
|
-
|
0.12
|
Bacterial or fungal infection (%)
|
23 (10.6)
|
12 (18.2)
|
-
|
-
|
0.10
|
Sepsis (%)
|
1 (0.5)
|
2 (3.0)
|
-
|
-
|
0.27
|
Laboratory data
|
|
|
|
|
|
suPAR (ng/mL), Median (IQR)
|
11.9 (9.06–15.5)
|
16.4 (11.4–23.6)
|
2.8 (2.2-4.0)
|
2.3(2.0-2.9)
|
< 0.001
|
WBC (× 109/L), Median (IQR)
|
6.2 (4.8–8.7)
|
8.1 (6.1–12.6)
|
5.1(4.3–6.1)
|
5.4(4.2–6.8)
|
< 0.001
|
Platelets (× 109/L), Median (IQR)
|
101(72–140)
|
122(81–177)
|
150 (116–200)
|
160 (110–231)
|
0.05
|
ALT (IU/L), Median (IQR)
|
190(67–492)
|
267(175–910)
|
45 (16–101)
|
18 (10–24)
|
0.001
|
Albumin (g/L), Mean (± SD)
|
31.1 ± 4.3
|
31.7 ± 4.0
|
40.5 ± 3.5
|
42.7 ± 3.0
|
0.39
|
Bilirubin (mg/dL), Median (IQR)
|
18.1(12.7–24.3)
|
21.0(15.2–26.6)
|
0.8(0.5-1.0)
|
0.7(0.4–0.9)
|
0.039
|
INR, Median (IQR)
|
1.94 (1.73–2.30)
|
3.14 (2.7–3.7)
|
-
|
-
|
< 0.001
|
Creatinine (mg/dL), Median (IQR)
|
0.7 (0.6–0.9)
|
0.8 (0.7–1.1)
|
0.6 (0.5–0.7)
|
0.6(0.5–0.7)
|
0.04
|
Sodium (mmol/L), Mean (± SD)
|
137.2 ± 3.9
|
137.6 ± 4.8
|
140 ± 3.1
|
142 ± 2.0
|
0.56
|
Scores
|
|
|
|
|
|
CTP, Mean (± SD)
|
10.8 ± 1.3
|
11.4 ± 1.7
|
-
|
-
|
0.007
|
MELD, Mean (± SD)
|
21.9 ± 4.6
|
30.4 ± 6.5
|
-
|
-
|
< 0.001
|
SOFA, Mean (± SD)
|
8.6 ± 1.76
|
12.0 ± 1.4
|
-
|
-
|
< 0.001
|
ACLF acute-on-chronic liver failure, ACLF-MOF ACLF complicated with multi-organ failure, CHB chronic hepatitis B, HC healthy controls, UGIB upper gastrointestinal bleeding, HRS hepatorenal syndrome, HE hepatic encephalopathy, SBP spontaneous bacterial peritonitis, WBC white blood cell count., INR international normalized ratio, CTP Child—Turcotte-Pugh, MELD Model for End-stage Liver Disease, SOFA sequential organ failure assessment. |
Baseline plasma suPAR levels and association with ACLF disease progression
There was no difference in plasma suPAR levels between HBV-related and non-HBV-related ACLF (p > 0.05, Appendices, Fig. A. 1). We then determined plasma suPAR among HC, CHB and ACLF and found out suPAR levels in patients with ACLF were markedly higher than those with HC and CHB. (Fig. 2a, 12.16 (7.61–17.57) vs. 2.3 (2.00-2.89) vs 2.7 (2.16-4.00) ng/L; p < 0.001). However, no difference was shown between HC and CHB. We then further compared plasma suPAR levels in ACLF patients with or without complications at admission or during the follow-up period. At admission, patients with HE, HRS, UGBI or infection had higher suPAR levels than those without these complications. Differences in plasma suPAR levels were most pronounced in patients with HRS (30.15 (13.57–36.35) vs. 12.30 (9.33–16.46) ng/L; p < 0.001). (Fig. 2b). There was no statistically significant difference in suPAR levels between patients with or without SBP, ascites, cirrhosis. (Appendices, Fig. A. 1). Patients exhibiting circulatory failure during follow-up also showed significantly higher suPAR levels than patients without circulatory failure (Fig. 2b).
Aside from clinical features, significant correlations with plasma suPAR were also found for clinical laboratory data and prognostic scores (Table 2). All three prognostic scores were correlated with suPAR levels; the strongest correlation was found with the MELD (r = 0.421, p < 0.001) (Table 2). Among the laboratory data, various infection-immunity related data showed a positive association with suPAR levels, including the incidence of bacterial or fungal infection, WBC and PCT. Interestingly, in white blood cells, suPAR was positively correlated with the percentage of neutrophils but did not correlate with the percentage of monocytes. At the same time, suPAR was also negatively correlated with HBcAb and the percentage of lymphocytes, indicating that suPAR was positively correlated with innate immunity, but negatively correlated with adaptive immunity in patients with HBV-related ACLF. As systemic inflammation seemed to be the driver for the development of organ failure in ACLF 2, we also investigated whether the concentration of plasma suPAR was associated with plasma inflammatory cytokines in patients with ACLF. After identifying 27 cytokines in 40 patients, a strong correlation was found between two chemokines, MIP1beta and IL-8, and plasma suPAR levels (r = 0.453, p = 0.007; r = 0.448, p = 0.003, respectively) (Table 2). The rest of the cytokines were either not significantly associated or undetectable (Supplementary, Fig. A.1). Significant correlations were also found between suPAR and liver-related data, kidney-related data, and thyroid-related data, but not heart-related data (Table 2).
Table 2
Association of clinical parameters and prognostic scoring systems with serum suPAR concentrations
Variable
|
Correlation coefficient with serum suPAR (r)
|
p - value
|
Number of patients
|
Age
|
0.203**
|
0.001
|
282
|
Infection-immunity data
|
|
|
|
Bacterial or fungal infection
|
0.118*
|
0.04
|
282
|
HBcAb
|
-0.182**
|
0.003
|
261
|
WBC
|
0.306***
|
< 0.001
|
282
|
Neutrophil (%)
|
0.250***
|
< 0.001
|
271
|
Monocyte (%)
|
0.065
|
0.28
|
271
|
Lymphocyte (%)
|
-0.327***
|
< 0.001
|
271
|
PCT
|
0.205**
|
0.008
|
166
|
CPR
|
0.055
|
0.39
|
215
|
MIP1beta
|
0.453**
|
0.007
|
40
|
IL8
|
0.448**
|
0.003
|
40
|
Liver-related data
|
|
|
|
ALT
|
0.036
|
0.55
|
282
|
Albumin
|
-0.098
|
0.101
|
282
|
Bilirubin
|
0.287***
|
< 0.001
|
282
|
INR
|
0.281***
|
< 0.001
|
282
|
Heart-related data
|
|
|
|
MAP
|
-0.015
|
0.80
|
282
|
hsTnI
|
0.161
|
0.080
|
119
|
Kidney-related data
|
|
|
|
GFR
|
-0.290***
|
< 0.001
|
270
|
Creatinine
|
0.219***
|
< 0.001
|
282
|
Sodium
|
-0.221***
|
< 0.001
|
282
|
Thyroid-related data
|
|
|
|
T3
|
-0.356***
|
< 0.001
|
254
|
FT3
|
-0.270***
|
< 0.001
|
254
|
Scores
|
|
|
|
CTP
|
0.183**
|
0.002
|
282
|
MELD
|
0.425***
|
< 0.001
|
282
|
SOFA
|
0.356***
|
< 0.001
|
282
|
HBcAb antibody against HBV core, PCT procalcitonin, CRP C-reactive protein, MIP1beta macrophage inflammatory protein 1-beta, INR international normalized ratio, MAP mean arterial pressure, hsTnI hypersensitive troponin I, GFR glomerular filtration rate, T3 Total triiodothyronine, FT3 free triiodothyronine. CTP Child—Turcotte-Pugh, SOFA sequential organ failure assessment, MELD Model for End-stage Liver Disease, UGIB upper gastrointestinal bleeding, WBC white blood cell count. *p < 0.05, **p < 0.01,***p < 0.001 |
Survival analysis
During the 30-day follow-up, sixty-two (22.0%) patients died and thirty-four (12.1%) received a liver transplant. During the 90-day follow-up, eighty-two (29.1%) patients died and forty-one (14.5%) underwent liver transplantation.
Baseline plasma suPAR increased in patients who died or underwent transplant (n = 123) during the 90-day follow-up compared to those who survived without liver transplantation (n = 159) (16.03 (11.65–23.70) vs. 11.14 (8.41–14.14) ng/L, p < 0.001; Fig. 2b).
The optimal cut-off point for plasma suPAR in predicting 90-day mortality was 14.7 ng/mL, as calculated by the Youden Index. Based on this optimal cut-off point, Kaplan-Meier curves significantly indicated 30- and 90-day mortality for patients with ACLF. Intriguingly, the effect of suPAR levels on the mortality was significantly larger in patients without cirrhosis or with HE (Fig. 3). The 30-day mortality of patients with both ACLF and cirrhosis was not impacted as much as the 90-day mortality by high suPAR levels (Fig. 3). During 90-day follow-up, patients with high suPAR (suPAR > = 14.7 ng/mL) and with HE had the highest mortality while ACLF patients with low suPAR and without HE had the lowest mortality (Fig. 3).
Correlations between clinical features/laboratory data and short-term mortality were analyzed by univariate Cox regression (Appendices, Table A. 1). Results showed that suPAR was significantly associated with both 90-day and 30-day mortality (p < 0.001). Variables with statistically significant (p < 0.05) in univariate regression analyses and age were included in multivariate models. To explore whether serum suPAR was correlated with the short-term mortality independently of the prognostic scores, suPAR was separately evaluated with MELD and SOFA scores in multivariate analysis. Variables included in two prognostic scores would be ruled out from multivariate models in order avoid collinearity.
Cox’s regression multivariate analysis using the forward step-wise selection method identified suPAR > = 14.7 ng/mL and WBC > = 6.6*109, together with MELD > = 23.1 SOFA > = 9.5, as the independent predictors of both day 90 and day 30 mortality (Table 3). These models were validated by bootstrapping.
Table 3
Multivariate Cox regression models for short-term mortality in ACLF patients
|
HR (95% CI)
|
p - value
|
Bootstrapping
p - value
|
Mortality at 30 days
|
|
|
|
Model 1: MELD score
|
|
|
|
MELD > = 23.1
|
3.62 (1.93–6.76)
|
< 0.001
|
0.001
|
suPAR > = 14.7
|
3.72 (2.15–6.41)
|
< 0.001
|
0.001
|
WBC > = 6.6
|
1.96 (1.10–3.51)
|
0.02
|
0.016
|
Model 2: SOFA score
|
|
|
|
SOFA > = 9.5
|
3.18 (1.67–6.10)
|
< 0.001
|
0.001
|
suPAR > = 14.7
|
3.52 (2.02–6.13)
|
< 0.001
|
0.001
|
WBC > = 6.6
|
2.03 (1.13–3.67)
|
0.02
|
0.017
|
Model 3 Other Clinical Data
|
|
|
|
suPAR > = 14.7
|
4.77 (2.78–8.19)
|
< 0.001
|
0.006
|
Sepsis
|
7.48(2.25–24.85)
|
0.001
|
0.009
|
Cirrhosis
|
-
|
0.25
|
0.30
|
UGIB
|
-
|
0.40
|
0.78
|
HRS
|
-
|
0.09
|
0.15
|
Mortality at 90 days
|
|
|
|
Model 1: MELD score
|
|
|
|
MELD > = 23.1
|
3.19 (1.88–5.41)
|
< 0.001
|
0.001
|
suPAR > = 14.7
|
3.02 (1.90–4.81)
|
< 0.001
|
0.001
|
WBC > = 6.6
|
1.97 (1.21–3.21)
|
0.007
|
0.008
|
Age > = 46.5
|
2.51 (1.54–4.08)
|
< 0.001
|
0.003
|
Model 2: SOFA score
|
|
|
|
SOFA > = 9.5
|
2.54 (1.51–4.29)
|
< 0.001
|
0.001
|
suPAR > = 14.7
|
2.89 (1.78–4.69)
|
< 0.001
|
0.001
|
WBC > = 6.6
|
2.00 (1.22–3.27)
|
0.006
|
0.007
|
Age > = 46.5
|
2.73 (1.68–4.45)
|
< 0.001
|
0.001
|
Model 3 Other Clinical Data
|
|
|
|
suPAR > = 14.7
|
4.52 (2.87–7.14)
|
< 0.001
|
0.003
|
Sepsis
|
7.33 (2.21–24.32)
|
0.001
|
0.02
|
Cirrhosis
|
-
|
0.10
|
0.15
|
UGIB
|
-
|
0.24
|
0.58
|
HRS
|
-
|
0.12
|
0.24
|
UGIB upper gastrointestinal bleeding, HRS hepatorenal syndrome, WBC white blood cell count., MELD Model for End-stage Liver Disease, SOFA sequential organ failure assessment. |
Moreover, analysis of the area under the receiver operating characteristic curve (ROC-AUC) revealed that suPAR may be a useful predictor for both 30- and 90-day mortality in ACLF patients (0.751 and 0.742, respectively) (Table 4).
Table 4
ROC area of suPAR predicting mortality in ACLF patients
|
ROC area (95% CI)
|
Mortality at 30 days
|
|
suPAR
|
0.751 (0.684–0.817)
|
MELD
|
0.732 (0.658–0.807)
|
SOFA
|
0.763 (0.698–0.828)
|
Mortality at 90 days
|
|
suPAR
|
0.742 (0.680–0.805)
|
MELD
|
0.729 (0.663–0.795)
|
SOFA
|
0.726 (0.662–0.789)
|
MELD Model for End-stage Liver Disease, SOFA sequential organ failure assessment. |
suPAR enhanced neutrophil ROS production.
ROS plays a key role in ACLF pathogenesis26. To better understand the role of suPAR in ACLF, we applied suPAR to circulating neutrophils from patients with HBV related-ACLF under stimulation of E. coli. There was enhanced ROS production in neutrophils after suPAR addition, suggesting elevated serum suPAR levels promote disease progress in HBV related-ACLF (Fig. 4, p < 0.05).