Baseline elevated serum angiopoietin-2 predicts long-term non-regression of liver fibrosis after direct-acting antiviral therapy for hepatitis C

We previously revealed that Angiopoietin-2 (Ang2) predicts non-regression of liver fibrosis based on liver stiffness measurement (LSM) at 24 weeks after anti-hepatitis C virus (HCV) treatment. In this study, we extended the observational period to 96 weeks to investigate the factors associated with non-regression after treatment with direct-acting-antivirals (DAAs). Patients treated with DAAs who underwent transient elastography at baseline and 24 and 96 weeks after DAA therapy were included. Baseline and post-treatment serum Ang2 levels were measured. Liver fibrosis stages were defined based on LSM. Multivariate regression was used to evaluate factors associated with non-regression of liver fibrosis between various time points. In total, 110 patients were included. Of these, 11% showed non-regression of LSM-based fibrosis stage at 96 weeks after DAA therapy. In multivariate analysis, advanced liver fibrosis stage and high baseline Ang2 levels were significantly associated with non-regression at 96 weeks. In patients with advanced liver fibrosis (F3/4), baseline Ang2 levels were associated with non-regression of liver fibrosis stage. Between SVR24 and SVR96, post-treatment Ang2 levels and controlled attenuation parameter values at SVR24 were significantly associated with non-regression of liver fibrosis stage in patients with F3/4. Thus, serum Ang2 levels are an important target for monitoring and therapy.


Definition of liver fibrosis stage and regression after successful HCV eradication.
FibroScan 502 (Echosens, Paris, France) was used for LSM and controlled attenuation parameter (CAP) evaluation with the M-probe and XL-probe. As described previously, patients were placed in the supine position with the right hand at the most abducted position during the FibroScan examination procedure 13,25 . Similar to our previous report, at least 10 valid measurements were obtained, and effective measurements were defined as those with a success rate of > 60% and an interquartile range of < 30% 13,25 .
The fibrosis stage was defined according to transient elastography data (FibroScan; Echosens) as described previously, with cut-off values of 7.1 kPa for F ≥ 2, 9.5 kPa for F ≥ 3, and 12.5 kPa for F4 13,26 . In addition, according to a previous report, regression of liver fibrosis was defined as follows: in patients with liver fibrosis stage F2 to F4, the liver fibrosis stage decreased by more than 1 stage after DAA therapy; in patients with liver fibrosis F0/1, the liver fibrosis stage did not deteriorate 13,26 . Statistical analyses. Continuous variables were analysed by the paired Mann-Whitney U-test, and categorical data were analysed by the chi-squared test. A multivariate logistic regression analysis with stepwise forward selection was performed with variables identified as significant at P ≤ 0.005 in univariate analyses of the factors associated with non-regression of liver fibrosis between baseline and at SVR96 and identified as significant at P ≤ 0.001 in univariate analyses of the factors associated with non-regression of liver fibrosis between SVR24 and SVR96. The cutoff value was based on the receiver operating characteristic (ROC) curve by maximizing the Youden index. All P values were two-tailed, and P < 0.05 was defined as statistically significant. Statistical analyses were performed using SPSS version 24.0 (IBM Japan, Tokyo, Japan).

Results
Baseline patient characteristics. A total of 208 patients with HCV infection who received IFN-free DAA therapy between October 2014 and July 2016 and were regularly followed-up at Hokkaido University Hospital were screened. Of these, 110 patients with all FibroScan examination data at baseline, SVR24, and SVR96, complete clinical information, and preserved serum samples obtained at baseline and after treatment were included in this study ( Figure S1). Table 1 shows the baseline characteristics of these 110 patients and a comparison of LSM-based liver fibrosis stages F0-2 and F3-4. The median age of patients was 66 years (range, 22-87 years), and 69 patients (62.7%) were female. The baseline median aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels were 39 IU/L (range,  and 38 IU/L (range, 6-273), respectively, and the median platelet count was 16.2 × 10 4 /μL (range, 2.6-37.3 × 10 4 /μL).
Rate of non-regression at 96 weeks after successful HCV eradication by DAA therapy and associated factors. Figure 1 summarizes the changes in LSM-based liver fibrosis stage at 96 weeks after successful HCV eradication by DAA therapy. As the baseline liver fibrosis stage increased, the rate of non- We further analysed the baseline predictive factors associated with non-regression of liver fibrosis at 96 weeks after DAA initiation. Table 2 provides a comparison of various factors in patients with or without non-regression of liver fibrosis stage based on LSM at 96 weeks after DAA initiation. A univariate analysis revealed that baseline liver fibrosis stage (F0-2 vs. 3-4, P = 0.002), FIB-4 index (P = 0.005), angiopoietin-2 (P = 0.004), HCV-RNA (P = 0.025), and AST (P = 0.029) were significantly associated with non-regression at 96 weeks after DAA initiation. Subsequently, we included significant factors identified in the univariate analysis (P ≤ 0.005) in a multivariate logistic regression analysis (i.e., the liver fibrosis stage, FIB-4 index, and Ang2 levels). As shown in Table 2, the multivariate logistic regression analysis revealed that baseline fibrosis stage (odds ratio 4.56, 95% confidence interval, 1.13-18.3; P = 0.033) and Ang2 level (odds ratio 1.004, 95% confidence interval, 1.00-1.01; P = 0.039) were significantly associated with non-regression of liver fibrosis stage based on LSM at 96 weeks after DAA initiation. We subsequently conducted ROC analysis to set the cutoff value of the baseline Ang2 level, which  www.nature.com/scientificreports/ predicts non-regression of liver fibrosis at 96 weeks after DAA therapy. As shown in Supplementary Fig. 2, the cutoff value was set at 395 pg/mL (sensitivity, 0.75; specificity, 0.776; ROC-AUC, 0.759). Table 3 summarizes the results of a subgroup analysis of patients with baseline fibrosis stage F3/4. In this group, high Ang2 at baseline was significantly associated with non-regression of liver fibrosis stage based on LSM at 96 weeks after DAA initiation (P = 0.024).

Rate of non-regression of LSM-based liver fibrosis stage between 24 and 96 weeks after successful HCV eradication and associated factors.
Subsequently, we analysed the rate of non-regression between SVR24 and SVR96 and associated factors. As shown in Fig. 2, as the fibrosis stage increased at SVR24, the rate of non-regression at 96 weeks after DAA therapy increased. Among patients with stage F4 at SVR24, 58% (7/12) showed non-regression of liver fibrosis stage based on LSM, compared with 15% (17/110) for all patients.
We subsequently conducted a subgroup analysis of patients with liver fibrosis stage F3/4 at SVR24 (Table 5). A high Ang2 level post-treatment and high CAP value at SVR24 were significantly associated with non-regression of liver fibrosis stage between SVR24 and SVR96.

Discussion
DAAs are revolutionary anti-HCV drugs. Numerous patients have experienced successful HCV eradication in the past several years due to these novel DAAs. However, recent studies have revealed that even after successful HCV eradication by DAAs, the occurrence of HCC and deterioration of liver function are sometimes observed. The deterioration of liver fibrosis after HCV eradication is closely associated with HCC occurrence 11 and causes the deterioration of liver function; thus, the regression of liver fibrosis after successful HCV eradication is a clinically important issue. Previously, we reported that 25% (29/116) of patients show non-regression of LSM-based liver fibrosis stage at 24 weeks after DAA completion 13 . In this study, 11% of patients (12/110) showed non-regression www.nature.com/scientificreports/ of LSM-based fibrosis stage at 96 weeks after the completion of DAAs. Thus, over a longer observation period, more patients with successful HCV eradication by DAAs could experience regression of the LSM-based fibrosis stage. A multivariate regression analysis revealed that baseline fibrosis stage and Ang2 levels were significantly associated with non-regression at 96 weeks after the completion of DAAs. These results are consistent with those of our shorter observational study (spanning 24 weeks) showing that the baseline fibrosis stage and Ang2 levels are significantly determinants of non-regression after DAA therapy 13 . In addition, the observed association between an advanced fibrosis stage (F3/4) at baseline and non-regression of liver fibrosis after anti-HCV therapy is consistent with the results of a previous study in the IFN era 11 . In a subgroup analysis of patients with advanced fibrosis (F3/4) at baseline (Table 3), we further showed that a high Ang2 level at baseline is significantly associated with non-regression of liver fibrosis stage at 96 weeks after DAA Table 3. Factors associated with non-regression of liver fibrosis stage based on LSM between baseline and SVR96 in HCV-infected patients with baseline advanced liver fibrosis, who were treated with DAAs. HCV hepatitis C virus, BMI body mass index, AST aspartate aminotransferase, ALT alanine aminotransferase, γGTP γ-glutamyl transpeptidase, FIB-4 fibrosis 4, CAP Controlled Attenuation Parameter. a Data are shown as median (range) values. *Statistically significant difference, P < 0.05.

Regression
Non-regression P value www.nature.com/scientificreports/ www.nature.com/scientificreports/ completion. Thus, high baseline Ang2 levels might be an important predictive factor for non-regression, even for long time periods (96 weeks) after DAA completion. Because LSM could be affected by liver inflammation due to HCV infection, we analysed the period between 24 and 96 weeks after DAA completion (Table 4). Although a multivariate analysis revealed that advanced liver fibrosis stage (F0-2 vs. F3/4) alone was associated with non-regression, a subgroup analysis of patients with F3/4 at SVR24 revealed that high CAP and Ang2 values post-treatment were significantly associated with non-regression of liver fibrosis stage (Table 5). CAP is a well-established surrogate marker of liver steatosis 22 . Liver steatosis is a risk factor for the development of liver fibrosis 27 . Thus, even after successful HCV eradication, patients with liver steatosis should be carefully followed up and should take measures to improve liver steatosis. Elevated post-treatment Ang2 levels were also significantly associated with non-regression between 24 and 96 weeks after the completion of DAA therapy in patients with F3/4 at SVR24 points, similar to the results obtained for the comparison between baseline and SVR96. Thus, serum Ang2 levels are a potential predictive marker of non-regression of liver fibrosis, even long after the completion of DAAs.
The ANG-TIE2 pathway has a unique effect on vascular stability. Ang1 is mainly expressed in mesenchymal cells and has agonistic effects on Tie2-mediated signalling, resulting in vessel stabilization and endothelial barrier function 15,16 . Ang2 is mainly expressed in endothelial cells and is increased by VEGF, TGF, and hypoxia 28 . Hypoxia-induced Ang2 is hypoxia-inducible factor-1-(HIF1) dependant and is expressed only at sites of vascular remodeling, thereby playing a crucial role in destabilizing vessels for normal or pathological angiogenesis 29 .
In addition, it has been reported that portal hypertension-induced slow blood flow causes increased Ang2 expression 19,20 . Ang2 has an antagonistic effect on Tie2-mediated signalling, resulting in the inhibition of ANG1-TIE2-mediated signalling, causing vascular instability, leakage, and inflammation 28 .
Several studies have shown that elevated serum Ang2 is a candidate biomarker in various liver diseases. Increased Ang2 expression in liver tissues is associated with the occurrence and recurrence of HCC after DAA treatment for hepatitis C 20 . In this previous study, importantly, serum Ang2 levels after DAA therapy were significantly associated with liver tissue Ang2 expression levels. In addition, Lefere et al. reported the possibility that serum Ang2 levels could distinguish patients with NASH from those with simple liver steatosis 30 . Mauro et al. reported that elevated serum Ang2 is associated with mortality and kidney outcomes in patients with decompensated cirrhosis with acute kidney injury 17 . In addition, we have previously shown that baseline elevated Ang2 levels could predict non-regression of liver fibrosis stage based on LSM at 24 weeks after DAA treatment in patients with HCV infection. In this study, we showed that this association persists over a longer period, even at 96 weeks after DAA treatment in patients with HCV infection. Thus, elevated Ang2 could persist for a long duration and might have a pathogenic effect on liver fibrosis. It has been reported that elevated Ang2 expression causes vascular leakage and inflammation and might promote the progression of liver fibrosis 31 .
Recently, Ang2 has been identified as a potential therapeutic target 28 in cancer and ophthalmologic diseases 28 . Furthermore, the effectiveness of the inhibition of both Ang2 and VEGF in diabetic macular oedema has been reported 32 . Similarly, in liver disease, anti-Ang2 therapy is a potential novel therapeutic option. Lefere et al. observed higher serum Ang2 levels in patients with NASH than in patients with simple liver steatosis 30 , and the inhibition of Ang2 restored liver fibrosis in a NASH mouse model. In addition, Pauta et al. reported that in a liver fibrosis rat model induced by CCl4, liver fibrosis and liver inflammation were reduced by the administration of an anti-angiopoietin 2 antibody 31 .
Accordingly, there is substantial evidence that the inhibition of Ang2 could be a novel therapeutic strategy in liver disease. In the present study, patients with elevated serum Ang2 levels showed non-regression of liver fibrosis, suggesting that anti-Ang2 therapy might lead to regression.
This study had several limitations. It was a retrospective study, and the number of patients was relatively small. Thus, a prospective, large-scale, and multicentre study is required to validate the findings. In addition, we evaluated the liver fibrosis stage using LSM. The gold standard for staging is liver biopsy; however, this procedure is invasive and carries a risk of sampling error. Thus, we utilized the LSM-based liver fibrosis stage, and this should be considered when interpreting the study results.
In conclusion, an advanced liver fibrosis stage and baseline high serum Ang2 levels are associated with nonregression of liver fibrosis stage at 96 weeks after the completion of DAA therapy. Thus, careful monitoring is necessary in these patients, even after successful HCV eradication by DAAs.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.