| |
Interferon responses and spontaneous HCV clearance: Is it all a matter of fat? Editorial
|
| |
| |
Download the PDF here
Download the PDF here
Journal of Hepatology July 2012
Jordan Feld
Toronto Western Hospital Liver Centre, University Health Network, University of Toronto, Toronto, Canada
These data offer some potential explanations to some puzzling aspects of the complicated but slowly emerging picture of HCV, lipids and interferon responses. The possibility that interferon non-response is related to effects on lipids is certainly intriguing and merits further study and exploration of the potential relevance for spontaneous HCV clearance, may be even more important.
see article editorial refers to below
Shortly after the discovery of the hepatitis C virus (HCV), it was recognized that the virus interacts intimately with the lipid machinery of the cell. Lipids are involved in almost all stages of the HCV lifecycle beginning with viral entry and including RNA replication, virion assembly and secretion of mature infectious viral particles [1]. The HCV core protein associates directly with cellular lipid droplets leading to recruitment of the non-structural (NS) proteins NS5A and NS3 to form the replicase complex, which serves as a scaffold for the RNA-dependent-RNA polymerase (NS5B) to begin replication of the viral genome [2]. Viral assembly then takes place and seems to parallel the processing of endogenous very low density lipoproteins (VLDL) with transfer of apolipoprotein B (apoB) and then apoE to form infectious low-density viral particles termed lipoviral particles (LVP) [3]. Impaired lipidation of viral particles results in production of non-infectious high-density virions, whereas increased abundance of LVP is associated with increased HCV infectivity [4]. The presence of increased apoE on the surface of LVP is associated with increased infectivity, possibly by facilitating interaction with lipoprotein receptors on hepatocytes [5]. ApoE may be important for more than infectivity. ApoE expression influences immune responses targeting lipid antigens and therefore its presence on HCV viral particles may be important for virus-host interactions [6].
HCV infection triggers innate immune pathways leading to the production of interferons and the establishment of an intracellular antiviral state [7]. Successful viral clearance however requires the subsequent emergence of a broad and potent HCV-specific adaptive immune response [8]. If infection is not cleared in the months following infection, the adaptive immune response wanes and the infection progresses to a chronic state with stable viral titers but progressive liver disease, after which viral clearance does not occur without antiviral therapy [8]. Responses to interferon-based treatment are largely dictated by the baseline expression of intra-hepatic interferon-stimulated genes (ISGs) [9]. Patients with low baseline ISG expression respond well to interferon with high rates of viral clearance while those with preactivated ISGs show no further gene induction with therapy, leading to interferon non-response and treatment failure [10], [11]. What controls the degree of ISG preactivation remains poorly understood.
The identification of single-nucleotide polymorphisms (SNPs) near the IL28B gene that are strongly associated with spontaneous [12] and treatment-induced viral clearance [13], [14], [15] raised the possibility that levels of interferon lambda (IL28B) were driving baseline ISG expression. However, despite the long and growing list of associations between the IL28B genotype and many aspects of HCV infection, the mechanisms underlying these relationships remain remarkably elusive. Hepatic ISG expression is associated with the IL28B genotype, however most reports have not found differing levels of IL28B itself across the different genotypes [16], [17]. Furthermore, there are some seemingly paradoxical associations. One might have expected that a robust interferon response at baseline would lead to higher rates of spontaneous viral clearance. However, patients with the CC genotype have lower baseline ISG expression [16], [17], [18], perhaps suggesting reduced interferon production, yet have higher rates of spontaneous HCV clearance. Despite their favourable response to therapy, CC patients also tend to have higher baseline HCV RNA titres, a factor consistently shown to be associated with interferon non-response [13]. Finally, patients with the CC genotype have also been shown to have higher cholesterol but less hepatic steatosis [19].
In this issue of the Journal, Sheridan and colleagues explore possible links between interferon responses and HCV-lipid interactions [20]. They hypothesize that HCV may take advantage of the effects of interferon on lipid homeostasis. In a cohort of 72 patients with genotype 1 HCV infection, they evaluated the correlations between levels of the infectious low-density HCV fraction (LVP), lipid parameters (apoE, apoB and cholesterol) and the response to interferon-based therapy. They also examined the role of the IL28B genotype and measured serum levels of interferon-gamma-inducible protein 10 (IP10) as a surrogate for ISG expression. They first show that, as might be expected, apoE levels correlate well with the amount of lipid-rich LVP HCV but not with the higher density, lipid-poor HCV fractions. Lower apoE levels were found in patients with a good response to therapy and in those with the treatment-favourable IL28B genotype (CC). Finally, they found a modest correlation with levels of apoE and IP10, but this was not replicated in a small validation cohort. Based on these findings, the authors wonder if higher levels of IP10, which they interpret to indicate greater interferon and ISG activation, lead to more apoE, which in turn favours the more infectious LVP form of the virus resulting in greater HCV infectivity and reduced treatment response.
An evaluation of the HCV fractions by IL28B genotype yielded interesting results. The CC genotype was associated with higher total HCV RNA levels, as has been shown before [13], but interestingly the correlation was entirely driven by an association with the less infectious non-LVP form of the virus. This may at least partially explain the paradoxical association between the treatment-responsive CC genotype and higher viral loads. The higher viral loads seen in CC patients are composed of the less infectious non-LVP virus and therefore do not negatively affect the response to treatment. Presumably the association of higher viral levels with poor treatment response relates to levels of the more infectious LVP fraction and would likely disappear after controlling for the IL28B genotype. The cohort in this study was too small for such an analysis.
This study raises some interesting possibilities. The authors suppose that endogenous interferon stimulates higher levels of apoE, which ultimately promotes infectious LVP HCV and thus reduces responses to interferon-based therapy. Although this is an interesting hypothesis, it is premature to draw causal links. The association between apoE levels and IP10 was marginal at best and IP10 is only a modest surrogate for ISG activation [21]. Notably, although there was a trend, IP10 levels did not correlate with treatment outcome in this cohort and did not differ by IL28B genotype. ApoE and LVP titre also did not differ by treatment response. However, the differences in levels of apoE, apoB and LDL by IL28B genotype were significant. Rather than an explanation for treatment non-response, the authors might have identified a factor relevant to spontaneous viral clearance. The association between the CC genotype and lower apoE but higher apoB and LDL levels was quite robust. If the CC genotype is associated with lower apoE levels and therefore less infectious virus, this may explain why patients with this genotype are more likely to clear infection. The infectiousness of the virus is likely much more important at the time of acute exposure when viral spread is rapidly progressing than during interferon treatment. It may well be true that the LVP virus is less interferon-responsive, in which case the hypothesis about treatment may also be correct, but this has yet to be shown. The data supporting LVP-related infectivity from this group [22] and others [23] are strong and may be very relevant to spontaneous clearance. In fact, an explanation for spontaneous clearance with the CC genotype that does not require more interferon production would potentially explain why these patients go on to have lower levels of ISG preactivation, a paradox that has until now been unexplained. A theory allowing for lower interferon production in CC patients, would also potentially explain the higher viral titres seen with the CC genotype, whether LVP or non-LVP HCV.
To clarify whether the findings truly relate to interferon will require some additional work. It would be helpful to know if apoE is induced by exogenous (and ideally endogenous) interferon directly and if so what mechanisms are involved. To support the concept that higher apoE leads to interferon non-response, it would be helpful to determine if there is a difference in interferon sensitivity between the LVP and non-LVP viral fractions. It would also be useful to delve further into the issue of spontaneous clearance and the association with the CC genotype. Do all individuals with the CC genotype have lower apoE but higher apoB and LDL levels or is this a phenomenon that requires the presence of HCV, or possibly, as the authors suppose, endogenous interferon? If the IL28B genotype affects apoE levels independent of interferon, the question of the underlying mechanism arises once again. Do patients who clear acute HCV have lower levels of apoE and LVP? If so, would modulation of lipid balance at the time of acute infection be a potential therapeutic strategy?
As with many provocative studies, the work by Sheridan and colleagues may have raised more questions than it answers but it is certainly a valuable contribution. These data offer some potential explanations to some puzzling aspects of the complicated but slowly emerging picture of HCV, lipids and interferon responses. The possibility that interferon non-response is related to effects on lipids is certainly intriguing and merits further study and exploration of the potential relevance for spontaneous HCV clearance, may be even more important.
----------------------
Apolipoprotein-E and hepatitis C lipoviral particles in genotype 1 infection: Evidence for an association with interferon sensitivity
Background & Aims
Hepatitis C virus (HCV) interacts with apolipoproteins B (apoB) and E (apoE) to form infectious lipoviral particles (LVP). Response to peginterferon is influenced by interferon-stimulated genes (ISGs) and IL28B genotype. LDL cholesterol (LDL-C) also predicts interferon response, therefore we hypothesised that LVP may also be associated with interferon sensitivity.
Methods
LVP (HCV RNA density ≤1.07g/ml) and 'non-LVP' (d >1.07g/ml) were measured in 72 fasted HCV-G1 patients by iodixanol density gradient ultracentrifugation and the LVP ratio (LVP/LVP+non-LVP) was calculated. Fasting lipid profiles and apolipoproteins B and E were measured. Interferon-gamma-inducible protein 10kDa (IP10), a marker of ISGs, was measured by ELISA.
Results
Complete early virological response (EVR) was associated with lower apoE (23.9±7.7 vs. 36.1±15.3mg/L, p=0.013), higher LDL-C (p=0.039) and lower LVP ratios (p=0.022) compared to null responders. In multivariate linear regression analysis, apoE was independently associated with LVP (R2 19.5%, p=0.003) and LVP ratio (p=0.042), and negatively with LDL-C (p<0.001). IP10 was significantly associated with ApoB (p=0.001) and liver stiffness (p=0.032). IL28B rs12979860 CC was associated with complete EVR (p=0.044), low apoE (CC 28±11 vs. CT/TT 35±13mg/L, p=0.048) and higher non-LVP (p=0.008). Logistic regression analysis indicated that patients with high LVP ratios were less likely to have EVR (odds ratio 0.01, p=0.018).
Conclusions
In HCV-G1, interferon sensitivity is characterised by low LVP ratios and low apoE levels in addition to higher LDL-C and IL28B rs12979860 CC. Null-response is associated with increased LVP ratio. The association of apoE and LVP with peginterferon treatment response suggests that lipid modulation is a potential target to modify interferon sensitivity.
Introduction
Hepatitis C virus (HCV) interacts with lipids at multiple stages in its lifecycle, including replication on lipid droplets [1], assembly [2], release and viral entry [3], [4]. A variable fraction of HCV in serum is associated with triglyceride-rich lipoproteins (TRLs), such as very-low-density-lipoprotein (VLDL), that contain apolipoprotein B (apoB) and apolipoprotein E (apoE). These particles have been termed lipoviral particles (LVP) [5] and exhibit increased infectivity in vitro [1] and in animal models [6]. We have developed a novel assay to measure LVP in patients using iodixanol density gradient ultracentrifugation and demonstrated an association of LVP (density ≤1.07g/ml) in the fasting state with insulin resistance, triglycerides and lack of early virological response in HCV genotype-1 (HCV-G1) infection [7]. Furthermore, after a high fat meal, the quantity of HCV RNA detected in a very-low-density fraction (d <1.025g/ml) increases on average 26-fold. These very low density HCV particles are derived from both de novo production and intravascular transfer, and are then rapidly cleared [8], [9].
Clearance of TRLs (e.g. VLDL and chylomicrons) from the plasma is facilitated by apoE [10]. ApoE exists in three common isoforms (E2, E3, and E4) that affect LDL receptor binding and modulate innate and adaptive immune responses to lipid antigens [11]. ApoE is now recognised as a central component of the HCV-host lipid interaction through involvement at several stages in the viral lifecycle, including intracellular viral assembly [2], [12], [13], [14] and as a surface constituent of secreted LVP in vivo and in vitro [15], [16]. Electron microscopy indicates that infectious LVP are enriched in apoE on the surface [17], implicating apoE in mediating HCV infectivity via lipoprotein receptors [18], [19].
The discovery that interleukin 28B (IL28B) single nucleotide polymorphisms (SNPs) are the most important genetic determinants of chronicity [20] and response to anti-viral therapy in HCV-G1 infection [21], [22], [23], [24], [25] suggests that previous associations of apoE genotype and clinical outcomes in HCV-G1 need to be re-evaluated [19], [20], [21]. The innate immune response is a critical determinant of clinical outcomes that is mediated through expression of interferon stimulated genes (ISGs). It is a paradox that pegylated interferon-α (PegIFNα) is the mainstay of treatment but activation of ISGs and high pre-treatment endogenous interferons are associated with non-response to anti-viral therapy [26], [27], [28]. Restoration of interferon sensitivity remains a target to minimise resistance to direct acting anti-viral (DAA) drugs in combination therapy [29]. The IL28B non-response alleles have been associated with increased ISG expression [28], [30], [31]. Interferon-γ-inducible-protein-10kDa (IP10/CXCL10) is a circulating ISG product and a quantifiable serum marker for overall ISG activation. IP10 levels are associated with treatment outcome [30] and may improve the predictive value of IL28B genotype for individualised treatment [31]. We and others have previously reported that low apoB-associated cholesterol (i.e. LDL/non-HDL cholesterol) is an independent determinant of sustained virological response (SVR) in HCV-G1 infection [32], [33], [34], and recent data indicates that this remains important in treatment experienced patients treated with DAAs [35], suggesting that lipid pathways may also be affected by ISGs [36].
We hypothesised that interferon sensitivity was also associated with LVP. We therefore evaluated LVP in HCV-G1 patients with evidence of interferon sensitivity who achieved early virological response (EVR) compared to null-responders and those with the favourable IL28B rs12979860 CC genotype compared to the unfavourable CT/TT genotypes. ApoE was measured because of its central role in the HCV-host interaction, and IP10 was measured because of its association with interferon sensitivity and a marker of endogenous ISGs.
Materials and methods
Patients
Patients aged >18years with HCV-G1 infection, including both treatment naïve and previous non-responders, attended for a morning fasting blood test. Individuals that were alcohol dependent, on concurrent lipid lowering or anti-viral therapy or co-infected with hepatitis B or human-immunodeficiency virus, were excluded. Seventy-two patients provided written informed consent to participate. Patients (27/72) returned for a second fasting blood test within 8weeks, with no direct intervention between visits. This group was used to independently confirm any associations identified from the primary analysis. Patients were recruited jointly from two centres (Newcastle upon Tyne Hospitals NHS Foundation Trust and Imperial College London Healthcare Trust). The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by both institutions ethics committees (REC 07/H0902/45; REC 07/S0501/21).
Clinical and laboratory assessment
Total cholesterol, triglyceride, and HDL-cholesterol (HDL-C) were measured by standard automated enzymatic methods using an Olympus AU640 analyser (Olympus, Watford, UK). LDL cholesterol (LDL-C) was calculated using the Friedewald equation [37]. Apolipoprotein A-I, apoB, and apoE were measured on each sample by automated rate nephelometric methods (BNII, Dade Behring Marburg GmbH, Marburg, Germany). Patients had assessment of liver stiffness using transient elastography (FibroScan®) within 8weeks from the time of the fasting blood sample.
Measurement of HCV-LVP
Fasting plasma was fractionated by density using 12.5% self-forming iodixanol density gradient ultracentrifugation as described previously [7]. The LVP fraction was defined as HCV RNA detected in the top 3.5ml at density ≤1.07g/ml. The non-LVP fraction was defined as HCV RNA detected at density >1.07g/ml in the bottom 6.5-ml fraction. The LVP ratio was calculated as: LVP/(LVP+non-LVP).
Measurement of IP10
Serum IP10 levels were measured using an ELISA kit (R&D Systems, Abingdon, Oxon, UK) according to the manufacturer's instructions. Samples were run in duplicate and absorbencies were read at OD450nm with wavelength correction at OD570nm. Serum IP10 concentration was calculated against a standard curve using the regression equation. A coefficient of variation (CV) value was also calculated for each duplicate set of readings from the mean/standard deviation.
ApoE and IL28B genotyping
ApoE genotype was determined using an automated dual fluorescent melting curve technique on genomic DNA isolated from whole blood using a commercial kit (Nucleon DNA extraction kit, Amersham Life Sciences, UK) in 71 patients with 1 genotyping failure in the remaining patient.
Patients (55/72) provided consent for stored DNA to be used for IL28B genotyping. Genomic DNA was isolated from 2ml of whole blood using QIAamp midi kits (QIAGEN, Crawley, UK). The SNP rs12979860 in the IL28B gene was genotyped independently by KBiosciences (Hoddesdon, Hertfordshire, UK) using the KASPar chemistry SNP genotyping system using FRET quencher cassette oligonucleotides (http://www.kbioscience.co.uk). Genotyping for rs12979860 failed in four samples.
Statistical analysis
Continuous data was assessed by normality tests and summarised by the mean±standard deviation if normally distributed. Two sample t-tests were used for comparison of normal variables between IL28B genotypes, and one-way ANOVA for comparison on variables associated with week 12 treatment response and the three apoE isoforms. Metabolic and viral variables that showed non-parametric distributions were either log10-transformed to parametric distributions before hypothesis tests were applied or non-parametric tests (Kruskal-Wallis) were used and summarised as median (Q1-Q3). Univariate analysis of association was performed using Pearson correlation to test associations between parametric variables and Spearman's Rho for association between non-parametric variables. The regression (r) coefficients and a probability (p) from testing against the null hypothesis of no association were determined. p-Values <0.05 were considered to be significant. Variables with significant univariate correlations were analysed for evidence of independent association by multivariate linear regression analysis. Variables associated with week 12 treatment outcome were analysed using a binary logistic regression model. All statistical analyses were performed using Minitab16 (Minitab Inc., State College, USA).
Results
Patient characteristics
Seventy-two predominantly Caucasian patients were recruited to the study. Twenty-seven patients were treatment naïve at the time of the study, of whom 16 have subsequently been treated and 11 remained untreated. Of those treated, 9/16 (56%) achieved a complete EVR with subsequent SVR, 5/16 (31%) were relapsers after a partial EVR and 2/16 (13%) were null responders. Forty-five patients were treatment experienced, of whom 27/45 (60%) were prior null responders and 18/45 (40%) were prior relapsers.
Table 1 indicates variation in clinical and metabolic characteristics in the overall cohort and between complete EVR, partial EVR and null responders. Null responders had a clinical phenotype of older age, higher BMI, higher liver stiffness measurements (LSM), lower LDL cholesterol and apoB, and higher apoE levels than complete EVRs. The overall frequency of the major C allele of rs12979860 was 0.61 (CC=35.3%, CT=51%, TT=13.7%), and the CC genotype was more prevalent in complete EVRs than null responders (63% vs. 18%). There was no significant variation in lipid profiles between visits in those 27 patients who attended for a second study visit (Supplementary Table 1).
IP10 correlates with apoB and liver stiffness
IP10 was measured as marker of ISG expression. The average intra-assay CV for the IP10 ELISA was 3.7% and inter-assay CV was 6.1%. IP10 levels were wide ranging (median 279; min 65, max 1642pg/ml) and often high, consistent with significant induction of ISGs (Table 1). IP10 levels tended to be higher in prior null responders (Table 1). IP10 levels did not correlate directly with any specific viral parameter (vs. total viral load r=-0.021, p=0.864; vs. LVP r=0.179, p=0.150; vs. non-LVP r=0.035, p=0.782; vs. LVP ratio r=0.167, p=0.166).
A relationship between ISGs and lipids was evaluated by correlation analysis of serum IP10 with lipid profiles. There was a significant positive correlation between IP10 and apoE (r=0.369, p=0.004). There was also a significant negative correlation between IP10 and total cholesterol (r=-0.314, p=0.009). This relationship was restricted to LDL-C (r=-0.481, p<0.001) rather than HDL-C (r=0.060, p=0.657) and was likewise reflected by a negative correlation of IP10 with apoB (r=-0.446, p<0.001) (Table 2). The negative correlation between IP10 and LDL-C and apoB was confirmed independently in 27 patients attending for a second visit (IP10 vs. LDL-C, r=-0.498, p=0.026; IP10 vs. apoB, r=-0.464, p=0.034) although the positive correlation between IP10 and apoE was no longer significant in these smaller numbers (r=0.351, p=0.109). IP10 also significantly correlated with liver stiffness (r=0.390, p=0.002). In multivariate linear regression analysis, liver stiffness (p=0.032) and the negative correlation with apoB (p=0.001) were the only independent predictors of IP10 levels (R2=27.7%).
ApoE correlates with HCV-LVP
The mean apoE level was 33±14mg/L and ranged from 12 to 71mg/L for HCV-G1 patients (Table 1). ApoE levels correlated with fasting triglycerides (r=0.467, p=0.001) and trended towards variation corresponding to apoE genotype similarly to previous reports [38] with increased apoE levels in those with E2/E3 vs. E3/E3 or E3/E4 (Supplementary Table 2). Serum apoE levels tended to be lower in those who achieved complete EVR (p=0.068) (Table 1). In univariate analysis, serum levels of apoE correlated significantly with LVP load (r=0.283, p=0.034) and LVP ratio (r=0.313, p=0.014) (Fig. 1 and Table 2). The correlation of apoE with LVP and LVP ratio was confirmed independently in the second fasting sample provided by 27 individuals (apoE vs. LVP-load r=0.553, p=0.040; apoE vs. LVP ratio r=0.555, p=0.041). There was no significant correlation between apoE and total viral load (r=-0.075, p=0.567) or apoE and non-LVP (r=0.011, p=0.937). This was confirmed in the second sample (apoE vs. total viral load r=0.147, p=0.560; apoE vs. non-LVP r=-0.177, p=0.512). ApoE also correlated with IP10 (r=0.369, p=0.004) and triglycerides (r=0.467, p=0.001), and negatively with LDL-C (r=-0.452, p=0.001). In multivariate linear regression analysis, apoE remained significantly associated independently with LDL-C (p<0.001), triglycerides (p=0.002), and LVP (p=0.017) (R2=50.6%, Table 2).
LVP ratio is associated with treatment outcome
LVP ratio was higher in null responders than complete or partial EVRs (p=0.022) (Table 1). EVR was also associated with higher LDL-C (p=0.039), lower BMI (p=0.036), lower liver stiffness (p=0.016) and IL28B CC genotype (p=0.044). Binary logistic regression analysis of these factors associated with EVR (partial or complete) vs. null response revealed that LVP ratio was the most significant independent predictor of EVR (odds ratio 0.01, 95% CI 0.00-0.042, p=0.018) (Table 3), indicating that those with high LVP ratios were less likely to achieve EVR.
Variation of lipid profiles with IL28B genotype
Serum apoE levels were significantly lower in the response-favourable rs12979860 CC genotype compared to the unfavourable CT/TT genotype (28mg/L±11 vs. 35mg/L±13, p=0.048) (Table 4). CT/TT genotype was associated with significantly lower LDL-C (2.46±0.82mmol/L vs. 3.19±0.75mmol/L, p=0.006) and lower apoB (0.8±0.2 vs. 1.0±0.2g/L, p=0.005). Triglycerides, HDL-C and apoA-I were not significantly different between IL28B genotypes.
Variation of viral parameters with IL28B and apoE genotype
We sought to evaluate whether rs12979860 was associated with variation in LVP. Total plasma HCV RNA, LVP-load, non-LVP load and LVP ratio were compared between CC and CT/TT genotypes. HCV RNA total viral load was significantly higher in those with response-favourable CC genotype (6.11±0.63log10IU/ml vs. 5.72±0.49, p=0.022). This difference in total viral load was predominantly accounted for by a higher non-LVP fraction in CC genotype (6.09±0.58log10IU/ml vs. 5.59±0.62log10IU/ml, p=0.008) and reflected by a trend towards lower LVP ratio (CC median 0.15 (0.11-0.26 Q1-Q3) vs. CT/TT median 0.26 (0.13-0.36 Q1-Q3), p=0.190). There was no significant difference in LVP-load between the IL28B genotypes.
We evaluated variation in LVP with apoE genotype because of correlation of LVP and LVP ratio with apoE. Frequencies of the apoE isoforms were E2=0.08, E3=0.73, and E4=0.19. No significant differences in these viral parameters were identified between apoE genotypes (Supplementary Table 2).
Discussion
The present study evaluated the hypothesis that levels of infectious LVP in vivo are associated with interferon sensitivity. The clinical phenotype of null response to peginterferon-α and RBV in our population was characterised by increased LVP ratios. In contrast, EVR was associated with low LVP ratio, low apoE and increased apoB and LDL-C. Low apoE levels were also associated with the favourable IL28B CC genotype. ApoE correlated with LVP and LVP ratio and was the strongest predictor of LVP in multivariate linear regression analysis.
It is established that LVP are low-density HCV particles that have high infectivity [1], [6]. LVP may mask neutralising epitopes, thus evading recognition by neutralising antibodies [39], [40]. We have previously reported that LVP correlate with insulin resistance and triglycerides in HCV-G1 infection [7]. High apoE levels are likely to be advantageous to the HCV lifecycle. ApoE is enriched on the surface of LVP [15], [16], [17] and is important for cell entry [18]. Viral non-structural proteins including NS5A interact with several host proteins involved in regulation of lipid metabolism at the sites of viral replication [41], [42]. An interaction between NS5A and apoE is required for formation of infectious virus [2], [12], [14], [43]. High levels of apoE may therefore be advantageous for the HCV-G1 lifecycle and favour the formation of infectious LVP in vivo.
We were unable to determine whether LVP reduces interferon sensitivity directly, or whether the effects of the interferon response on lipids may impact indirectly on levels of LVP.
The data in this study suggest a direct effect of ISGs on lipid profiles. The notion that interferons can alter lipids was recognised by reports of increased serum triglycerides in the early studies of IFNα therapy [44]. The endogenous interferon response may also alter lipid metabolism. IP10 is an immune-modulatory chemokine and has been considered a candidate serum marker for ISG expression [45]. High ISGs are associated with non-response to peginterferon-α treatment and with the unfavourable IL28B genotype [46], [47]. Serum IP10 levels have been reported to be predictive of viral kinetics [31], [45], [48], [49] and non-response to PegIFNα and RBV [30], [50], [51]. However, IP10 as a single serum marker is unlikely to represent a complete measure of hepatic ISG activation. Notably, unlike ISG expression in the liver, IP10 levels did not significantly differ by IL28B genotype, although we observed increased IP10 levels in null responders. Univariate analysis demonstrated a positive correlation between IP10 and apoE, and a negative correlation with LDL-C and apoB. Multivariate linear regression demonstrated that high IP10 levels were associated with increased liver stiffness and low apoB, supporting the concept that endogenous ISG activation alters VLDL/LDL metabolism and may therefore indirectly favour formation of LVP.
ISGs are also regulated by IL28B genotype. Our new data indicate that the unfavourable CT/TT genotype of rs12979860 was associated with increased levels of apoE but decreased levels of LDL-C and apoB. These data are consistent with a previous report of lipid profiles in HCV-G1 that indicated higher LDL-C and apoB levels in those with the favourable CC genotypes than those with CT/TT [36]. The influence of IL28B genotype on lipid profiles in the general population has not been reported. We hypothesise that a state of high endogenous interferon activation also causes altered lipid profiles, characterised by low apoB and LDL-C but increased apoE, and insensitivity to treatment with peginterferon-α. Low apoB and LDL-C may therefore be surrogate markers that reflect ISG activation and may help explain the clinical associations of low LDL-C with non-response to anti-viral therapy [32], [33], [34].
An alternative hypothesis is that HCV infection per se alters lipid levels and subsequently modifies ISGs expression and interferon sensitivity. Experimental evidence indicates that LVP from chronically infected HCV patients can interfere with maturation of monocyte-derived dendritic cells induced by TLR3 or TLR4 ligands, supporting the hypothesis that LVP may interfere with interferon sensitivity [52].
It remains a paradox that the favourable IL28B genotype is associated with higher total viral load, given the association of low viral load with sustained virological response, but this has been a consistent and hitherto unexplained association with IL28B genotype [21]. Our data indicate that this difference in viral load between IL28B genotypes is manifested largely in the high density 'non-LVP' fraction that has lower infectivity in vitro. Further studies in larger cohorts are required to evaluate whether the association of LVP and non-response to anti-viral treatment [7] are independent of IL28B genotype. However, these preliminary data suggest some variation may be explained by the relative increase of higher density (d >1.07g/ml)/lower infectivity 'non-LVP' in those with the favourable IL28B genotype. In contrast, elevated apoE levels in those with unfavourable IL28B genotype may promote production of more infectious LVPs in vivo.
In summary, our data indicate that apoE is a key mediator in a complex interaction between interferon sensitivity, lipoprotein metabolism and infectious virus in chronic HCV infection. High apoE levels were associated with increased levels of LVP, and null response to peginterferon-α. ApoE was significantly higher in those with the unfavourable IL28B genotype. In contrast, low apoE correlated with low LVP, increased LDL-C, the favourable CC IL28B genotype and EVR. Further understanding of the effects of innate antiviral responses on lipid pathways in HCV may therefore provide novel therapeutic targets to enhance or restore interferon sensitivity, which may be a critical factor in preventing resistance to DAAs [29].
|
|
| |
| |
|
|
|
