Elevated serum levels of soluble CD14 in HBeAg‐positive chronic HBV patients upon Peginterferon treatment are associated with treatment response

Abstract Pegylated IFNα (PEG‐IFN) is one of the treatment options for chronic HBV (CHB) patients. However, the high patient treatment burden and limited response rate together clearly ask for biomarkers to predict PEG‐IFN response. Soluble CD14 (sCD14) is considered a marker for immune activation and has been shown to predict clinical outcome of HIV infection. However, studies on sCD14 in CHB infection are inconclusive, and its relationship with clinical outcome is largely unknown. Here, we measured sCD14 levels in CHB patients and investigated whether changes in sCD14 level related to PEG‐IFN response. Serum sCD14 levels were determined in 15 healthy controls, 15 acute self‐limited HBV, 60 CHB patients in different disease phases and 94 HBeAg+ CHB patients at week 0 and week 12 of a 52‐week PEG‐IFN treatment. Response to PEG‐IFN treatment was defined as HBeAg seroconversion or HBeAg loss at 26 weeks post‐treatment. The mean sCD14 level in acute HBV patients (3.0 µg/mL) was significantly higher than in CHB patients (2.4 µg/mL) and healthy controls (2.4 µg/mL). In CHB patients receiving PEG‐IFN, a significant increase in sCD14 was found after 12‐week treatment (median week 0:2.1 µg/mL; week 12:3.7 µg/mL). After 12‐week treatment, the fold change (FC = w12/w0) in sCD14 was significantly higher in responders compared to nonresponders (HBeAg seroconversion: median FCresponder = 2.1 vs FCnonresponder = 1.6; HBeAg loss: median FCresponder = 2.2 vs FCnonresponder = 1.5). Receiver operating characteristic curves demonstrated that FC‐sCD14wk12/wk0 levels can be of significant value as a stopping rule to select patients at week 12 who are not likely to benefit from further PEG‐IFN treatment.


| INTRODUC TI ON
About 1/3 of the world's population has been exposed to hepatitis B virus (HBV) during their lifetime, and over 250 million individuals worldwide suffer from chronic HBV (CHB) infection. 1 These patients are at increased risk of developing liver fibrosis, cirrhosis and hepatocellular carcinoma (HCC). 2 Currently, there are two main treatment options for CHB patients: nucleos(t)ide analogues (NA) and pegylated IFNα (PEG-IFN). NA interfere with viral replication but do not eliminate the virus have to be given lifelong and may lead to viral escape variants. The other treatment, PEG-IFN, aims to induce long-term immunological control over the virus after finite treatment duration, typically 1 year. PEG-IFN, however, has a moderate antiviral effect resulting in a sustained response in only ~25% of patients 3 and a serious risk of adverse events. 4 Considering the limited response rate, high costs and severe side effects, indicators to predict which individuals are most likely to respond to PEG-IFN, are highly needed. Especially, early on-treatment predictors are of interest to discontinue PEG-IFN and minimize treatment burden for the nonresponding patients.
CD14 is a protein that plays an important role in immunity.
CD14 acts as a co-receptor with Toll-like receptor 4 for the detection of bacterial lipopolysaccharides (LPS). 5 It also recognizes other pathogen-associated molecular patterns such as bacterial lipoteichoic acid. 6 It exists in two forms, a membrane-attached form (mCD14) and a soluble form (sCD14). mCD14 is a glycosylphosphatidylinositol-anchored glycoprotein expressed on the surface of various cells including myeloid cells 7,8 and nonhematopoietic liver cells. 7 sCD14 is produced by activated monocytes, macrophages and primary human hepatocytes. 9,10 In serum, it may serve as a biological marker for inflammatory disease activity as increased serum levels of sCD14 have been detected in patients with bacterial infections, viral infections, 11,12 rheumatic diseases 13,14 and multiple sclerosis. 15 sCD14 has been widely studied as a biomarker in HIV infection and liver disease. Elevated sCD14 levels were found in HIV-infected patients compared with seronegative controls, 12,16,17 and sCD14 levels could independently predict disease progression and mortality for both HIV-1 and HIV-2 infections. 16,18,19 In viral hepatitis, high levels of baseline plasma sCD14 were associated with worse clinical outcome in a mixed cohort of CHB and chronic HCV patients. 20 Furthermore, elevated serum sCD14 levels were associated with HBV-related HCC in CHB. 21 Finally in HCV, the rise in sCD14 after 12 weeks of PEG-IFN treatment was a negative predictor of treatment response. 22 For HBV, however, neither the relation between serum sCD14 levels and the natural history of disease nor clinical outcome upon PEG-IFN treatment has been clearly established.
Here, we measured and analysed sCD14 levels in acute HBV infection, treatment-naive CHB patients from various disease stages and HBeAg+ CHB patients receiving PEG-IFN. For the latter, we characterized the relationship of sCD14 serum levels to PEG-IFN treatment response and assessed the value of sCD14 as an early on-treatment indicator to stop treatment of patients who are unlikely to respond.

| Patient selection
Sixty CHB patients who never received treatment attending the outpatient hepatology clinic of the ErasmusMC and 15 age-matched, acute self-limited HBV patients and healthy controls were included in our study (Table 1). Serum samples were prospectively collected and stored at −80°C. Exclusion criteria for selection in the study are described in detail elsewhere. 23 A total of 94 HBeAg+ CHB patients, who were previously enrolled in an investigator-initiated multicentre randomized trial (99-01 study) comparing PEG-IFN monotherapy or combination therapy with lamivudine (LAM), were also included. 24 The inclusion and exclusion criteria are similar to the original study. 24 Additional inclusion criteria for the current analysis were completion of the 26-week follow-up phase of the original study and availability of a baseline and 12-week serum sample for quantification of sCD14 levels (n = 94). All patients gave written informed consent according to the standards of the local ethics committee.

| Study design
A total of 94 HBeAg-positive CHB patients who are from 99-01 study, as described before, 24 in brief, received combination therapy with weekly doses of 100 μg PEG-IFN (PegIntron, Schering-Plough) and a daily dose of 100 mg LAM (Zeffix, GlaxoSmithKline) or monotherapy with 100 μg/wk PEG-IFN and placebo. The total duration of treatment was 52 weeks. All the patients were followed up for 26 weeks post-treatment. Patient serum samples that used for sCD14 measurement were collected at week 12 during PEG-IFN treatment and stored at −80°C. To reduce the likelihood of early patient dropout, the dose of PEG-IFN was lowered to 50 μg/wk in both treatment groups after 32 weeks.

| Acute but not chronic HBV infection or its disease phases associated with elevated sCD14 levels compared to healthy individuals
In order to characterize serum sCD14 levels in different HBV disease phases, sCD14 levels were first assessed in an age-matched cohort of acute HBV patients, treatment-naive CHB patients and healthy controls. Serum sCD14 levels were significantly elevated in acute selflimiting HBV patients (mean of 3.0 µg/mL) compared to CHB patients (2.4 µg/mL) or healthy control individuals (2.4 µg/mL) ( Figure 1A).
The sCD14 levels in acute HBV patients tended to positively correlate to serum ALT levels (Spearman r = 0.53, P = 0.05; data not shown).
F I G U R E 1 Serum sCD14 levels in healthy controls, acute HBV patients and chronic HBV patients. A, Box and whisker plots of serum sCD14 levels in healthy controls (n = 15), acute HBV patients (n = 15) and chronic HBV patients (n = 60). The top and bottom of the whiskers represent the measured maximum and minimum values, respectively. The bottom and top of the box represent the first and third quartiles, respectively. Mean sCD14 is depicted as horizontal lines within the boxes and were 3.0 µg/mL for acute HBV patients, 2.4 µg/mL for healthy controls and 2.4 µg/mL for chronic HBV (CHB) patients. *P < 0.05, ***P < 0.001 by one-way ANOVA and Tukey's multiple comparison test. B, Box and whisker plots of serum sCD14 concentration of CHB patients as in A but stratified for the four phases of CHB infection. Means were 2.1 µg/mL for IT (Immune tolerant; n = 15), 2.6 µg/mL for IA (immune active; n = 15), 2.4 µg/mL for IC (inactive carrier; n = 15), and 2.3 µg/mL for E-neg (HBeAg-negative; n = 15). ns, not significant by one-way ANOVA Next, we compared sCD14 levels among the four clinical CHB phases, categorized on virological, biochemical and serological parameters, but did not find any significant differences among these groups or between the groups and healthy controls ( Figure 1B). Furthermore, for these CHB patients, we did not find any correlation between sCD14 and liver damage (by serum ALT) or fibrosis stage (data not shown).

| sCD14 levels rise upon treatment with PEG-IFN
The underlying mechanism of action of PEG-IFN treatment is the activation of the immune system. To investigate the effect of this treatment on serum sCD14 levels and to assess association of sCD14 with clinical outcome, sCD14 levels were determined in 94 CHB patients from a retrospective cohort. 24  as reported. 24 Previously, it was determined that treatment of PEG-IFN with LAM did not influence response rates compared with PEG-IFN only, 24 and we here did not detect a difference of sCD14 levels between CHB patients receiving PEG-IFN with or without LAM (Table 2). Therefore, sCD14 data from these groups were pooled for further analysis (hereafter referred to as PEG-IFN unless specified).
Baseline sCD14 levels did not significantly differ between sexes, or across HBV genotypes A-D. Correlation between baseline sCD14 and baseline HBsAg level was observed (Spearman r = 0.22, P = 0.03; data not shown). Serum sCD14 levels were significantly elevated after 12 weeks of PEG-IFN treatment in paired patient samples ( Figure 2A). Median baseline level of sCD14 was 2.1 μg/mL (IQR 1.8-3.0 μg/mL), and the median level at week 12 was increased to 3.7 μg/mL (IQR 2.8-5.1 μg/mL) ( Figure 2B). No correlation was found between sCD14 levels (week 0 and 12) and ALT, fibrosis (Ishak score) or necroinflammation. sCD14 level at week 12 was associated with genotype (   (Table 2). Interestingly, we found that the mean FC-sCD14 (wk12/wk0) was significantly higher in responders compared to nonresponders  Figure 3B,D). Importantly, FC-sCD14 (wk12/wk0) was not associated with sex, HBV genotypes and treatment, however, was influenced by ethnicity as was also observed for the sCD14 levels at baseline and week 12 ( Table 2). Because ethnicity is a cofounder for FC-sCD14 (wk12/wk0) levels, we also performed this operation without the Asian patients ( Figure S2). When excluding Asian patients, the AUC was even higher ( Figure S2).
Knowing FC-sCD14 wk12/wk0 is associated with HBeAg seroconversion and HBeAg loss, we next studied the correlation of    Table 3).

| Predictive value of the change of sCD14 at week 12 upon PEG-IFN treatment
Combining of both markers revealed an even higher sensitivity (84% for FC-sCD14 wk12/wk0 = 2.0 and 61% for FC-sCD14 wk12/ wk0 = 1.5 in combination with HBsAg) when compared with using either HBsAg or for FC-sCD14 wk12/wk0 alone (Table S1 and Table 3).
Thus, it may be possible to exclude even more nonresponders after 12 weeks of treatment when using both serological markers together.
Taken all together, like HBsAg decline, the change in sCD14 wk12/ wk0 may thus also serve as an effective stopping criterion and may even be a superior negative predictive value.

| D ISCUSS I ON
This is the first study to describe the association of sCD14 levels with PEG-IFN treatment response in CHB. We found that a high rise in serum sCD14 in CHB patients at the first 12 weeks of PEG-IFN treatment is associated with an increased probability of clinical response defined as either HBeAg seroconversion or HBeAg loss.
Besides this major finding, we also found serum sCD14 levels ele- phase with a similar age group. We observed that the mean sCD14 level in the immune active phase of CHB was slightly higher as compared to the mean sCD14 level found in healthy controls which is in line with the previously observed relation between hepatic inflammation and sCD14 levels. 20 Possibly, previous studies reporting increased sCD14 levels in CHB included a subset of CHB patients with more severe liver disease, 20,21 while in our cohort, ~86% treatmentnaive CHB patients have a relative low fibrosis level (<F2 by fibroscan). Furthermore, we here demonstrate that ethnicity influences sCD14 levels which was reported before in rheumatoid arthritis patients 30 and we previously reported the same for age, 25 indicating that also these parameters may be a cause of discrepancy.
Why and how sCD14 is secreted remains elusive. sCD14 is thought to be produced by monocytes, macrophage, Kupffer cells and primary human hepatocytes, which can be increased upon activation by, for example, pathogen-associated patterns and/or pro-inflammatory cytokines. The pro-inflammatory cytokine PEG-IFN is a very potent activator of the immune system 31,32 that could directly or indirectly enhance release of sCD14 by hepatocytes and/or myeloid cells. 7,8,33 In our study, we found that the elevation in sCD14 in responders to PEG-IFN was significantly higher compared with those in nonresponders, which could therefore mean that inflammation/immune activation may be more effective and/or different in responders. The PEG-IFN induced increase in sCD14 fits well with the notion that sCD14 is a marker of immune activation in a number of acute and chronic inflammatory conditions, including liver inflammation. 30,34 Since microbial translocation is known to contribute to inflammation in viral hepatitis patients, 20 the observed increased sCD14 levels in HBV patients might also be partly due to release of CD14 from Kupffer cells upon stimulation by microbial products such as LPS. 5 We previously demonstrated that HBsAg can activate Kupffer cells. 35 In addition, we showed that HBsAg-sCD14 complexes exist in serum of CHB patients and that HBsAg induces dendritic cell activation in a CD14-dependent manner. 25 26 Our study shows that the response to PEG-IFN therapy in HBeAg+ patients can also be deduced from the change of serum sCD14 at week 12 with respect to week 0. Selecting nonresponders based on the change in sCD14 levels was even slightly more sensitive than the stopping rules defined in the 2017 EASL guideline. Importantly, the association of sCD14 change with response to treatment was independent of HBV genotype which makes it unique compared to other reported predictors so far. The value of using combinational markers of sCD14 change and HBsAg level for treatment response should be validated by others in future. These results demonstrate that sCD14 could serve as a valuable additional early immune marker for PEG-IFN treatment response to stop ineffective treatment early on.
Although this study is the first to describe that the change of sCD14 might be used as an stopping rule in PEG-IFN treatment at week 12, there are some limitations. First, a relatively small number of patients were included in our study that limited us to draw firm conclusions on the use of sCD14 as a biomarker in CHB treatment.
Second, although the original study contained 2 treatment arms, we decided to pool the patients since LAM did not add to overall response rate, 24 and no differences in sCD14 levels were found between CHB patients receiving PEG-IFN with/without LAM (Table 2). Third, we were not able to perform external validation on using the change in sCD14 as an stopping reference for PEG-IFN treatment at week 12, because an additional independent data set is not available to us. Thus, follow-up studies on other patient cohort are needed to validate the use of sCD14 to stop PEG-IFN treatment for those who will not benefit. Fourth, ethnicity is a confounder in our study. Our data suggest that predictive value of sCD14 may be lower in patients with Asian ethnicity that had lower sCD14 concentrations across all time points. Since only a limited number of patients with Asian ethnicity were represented in our cohort, validation of our findings in an independent cohort with more Asian patients is needed.
In conclusion, serum sCD14 has the potential as a biomarker to select patients who are unlikely to respond to PEG-IFN treatment and thus suffer unnecessarily from its severe side effects. Our data indicate that a high rise in serum sCD14 level upon treatment is associated with a successful HBV-directed immune response. This finding contributes to our understanding of the immunological requirements for viral eradication and calls for further research.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest regarding the publication of this article.