Hepatitis C Direct Acting Antivirals and Ribavirin Modify Lipid but not Glucose Parameters

Chronic hepatitis C (HCV) infection perturbs lipid and glucose metabolism. The influence of direct acting antiviral (DAA) treatment and ribavirin on these measures was evaluated. Furthermore, the effect of HCV cure on these parameters was assessed. Participants were allocated to one of three 12-week treatment groups: non-cirrhotic genotype 1a-paritaprevir/ritonavir/ombitasvir/dasabuvir (PrOD) plus ribavirin; non-cirrhotic 1b-PrOD; compensated cirrhotic 1a or 1b-PrOD plus ribavirin. Fasting insulin, glucose, lipid and apolipoprotein measures were assessed at baseline, Treatment Weeks 4 and 12, and 12 and 24 weeks post-dosing. Twenty-three of 24 participants achieved SVR (PP= 23/24, 96% SVR). Overall, total cholesterol, low-density lipoprotein cholesterol (LDL-C), and triglyceride levels all increased in treatment and post-dosing. However, LDL-C levels decreased during treatment in ribavirin recipients. Fasting glucose, insulin, and HOMA-IR were unchanged during treatment and 12 weeks post-treatment. By 12 weeks post-treatment, controlled attenuation parameter (CAP) scores, a measure of steatosis, increased from baseline (mean 30.3 ± 63.5, p = 0.05). This regimen was safe and highly effective and did not influence glucose metabolism. Ribavirin exposure may mitigate some on-treatment lipid changes. Further mechanistic studies are needed to understand how ribavirin impacts lipid pathways, as there could be therapeutic implications. The metabolic pathophysiology of increased CAP score with HCV treatment requires explanation.


Introduction
Chronic hepatitis C affects 2-3% of the world population and is a leading cause of cirrhosis, hepatocellular carcinoma, and liver transplant [1]. More than 350,000 deaths per year are attributable to hepatitis C (HCV)-related complications [1]. There is evidence that HCV perturbs the metabolic milieu, which in turn influences the risk for complications and response to HCV antiviral treatment. Insulin resistance and type 2 diabetes are associated with an increased risk of hepatocellular carcinoma (HCC) [2], higher transplant complication rates [3], accelerated liver fibrosis [4], and possibly increased morbidity from cardiovascular and metabolic complications [5]. Impaired insulin sensitivity is associated with diminished antiviral treatment outcomes with interferon-based regimens [6,7].
Blood samples were collected for HCV RNA analysis at baseline, Weeks 4 and 12, and 12 weeks and 24 weeks post-treatment.
Measures of fasting insulin, glucose, total cholesterol, HDL-C, LDL-C, TG, HbA1c, apoA1, apoA2, apoB, apoC2, apoC3, and apoE were performed at baseline, Weeks 4 and 12, and 12 and 24 weeks post-treatment. Patients were advised to fast prior to blood draws. HOMA-IR score was calculated as per the following: (glucose × insulin)/22.5. IR was defined as having an HOMA-IR >2. A cut-off of 2 was selected, as this is a recognized standard [27]. Furthermore, our investigator group believed that evaluating a study population with elevated HOMA-IR would be optimal to identify an HCV treatment effect, if present.
Steatosis and liver fibrosis can result as a consequence of metabolic dysfunction. Liver stiffness was determined by transient elastography (FibroScan ® , Echosens SA, Paris, France) [28], performed at baseline, Week 12, and 12 weeks post-dosing. A recording over 12.5 kPa defined cirrhosis. Controlled attenuation parameter (CAP) was utilized as a measure of steatosis and/or inflammation within the parenchyma of the liver.

Statistical Analysis
A danoprevir mono-therapy study identified a 1.6 +/− 1.1 decline in HOMA-IR over 14 days [23]. In HCV patients treated with interferon and ribavirin with or without metformin, a 21% decrease in HOMA-IR in metformin recipients versus 10% in the control group was noted in the first 12 weeks (absolute 1.0 mean change at 24 weeks post-HCV antiviral treatment) [29]. Based on these data, a crude sample size of 8 per group provided 80% power to detect a mean difference in HOMA-IR scores of 1.0 assuming a two-sided comparison, a variance of 0.5, and an alpha of 0.05. This did not account for analysis adjustments for potential confounders, as this was a pilot study.
Patient demographics, glucose, lipids, and apolipoprotein measures, grouped according to treatment arm and differences between baseline characteristics, were evaluated using chi-square and ANOVA at the p < 0.05 level. Fisher's exact test was used for outcomes with fewer than 5 cases. Changes from baseline in HOMA-IR, lipid, and apolipoprotein parameters at Weeks 4 and 12 as well as 12 weeks post-treatment and 24 weeks post-treatment were assessed using linear mixed models or generalized linear mixed models with a logit link. Models consisted of fixed effects for visit, treatment group, and their interactions with time. Models of glucose measures were adjusted for baseline BMI, and models of apolipoproteins were adjusted for baseline viral load. Visit was treated as a categorical variable with baseline as the reference. Models included a random intercept for participant or a repeated statement to control for the clustering effects of time. The estimated mean at each study time point was also determined by the mixed models. Pairwise comparisons of the estimated means within each group with respect to baseline were performed with least significant difference tests in the mixed models. As the primary outcome, changes in HOMA-IR from baseline to 12 weeks post therapy were compared within treatment groups.

Demographics and Baseline Characteristics
Twenty-four participants were enrolled in this study. Of 38 screen failures, 31 had HOMA-IR results <2. Baseline characteristics are outlined in Table 1. The participants were predominantly male with a mean age of 54 years (SD 11.6) and a mean BMI of 30.0 kg/m 2 (SD 4.6). The most common mode of HCV exposure was former injection drug use (48%) followed by blood transfusion (17%). Nine non-cirrhotic genotype 1a-infected participants received PrOD plus ribavirin therapy. Eight non-cirrhotic genotype 1b-infected participants were dosed with PrOD without ribavirin. Seven compensated cirrhotic genotype 1a or 1b-infected participants received PrOD plus ribavirin therapy.

HCV Treatment Response
All participants cleared HCV RNA by Week 4 of treatment. Twenty-three of 24 participants achieved SVR12. One participant with detectable virus at the end of treatment was lost to follow-up prior to the visit occurring 12 weeks post-treatment. The SVR12 was 100% (8/8) with ribavirin-free regimens and 15/16 (94%, PP = 15/16, 94%) in ribavirin-containing treatments (p = 0.47). SVR12 was 7/7 (100%; PP = 7/7, 100%) in cirrhotic participants and 16/17 (94%, PP = 16/17, 94%) in those without cirrhosis (p = 0.51). Ribavirin dosage was reduced to 600 mg in three patients following treatment initiation. One patient had a serious adverse event unrelated to the study drug and discontinued therapy at nine weeks to avoid a drug-drug interaction with colchicine for a severe gout flare resulting in hospitalization. No other serious adverse events were reported during the study.

Metabolic Measures and HCV Treatment-All Patients
The means of the metabolic measures across the study period are presented in Table 2 and changes from baseline at each time point are reported in Figures 1-4. Changes were observed for HbA1c and lipid measures across the study period ( Table 2 and described below). Overall, glucose (p = 0.11), insulin (p = 0.42), and HOMA-IR (p = 0.32) did not change over time (Table 2, Figure 2A-C). Individual HOMA-IR trajectories over time are presented in Figure 1.

HCV Treatment Response
All participants cleared HCV RNA by Week 4 of treatment. Twenty-three of 24 participants achieved SVR12. One participant with detectable virus at the end of treatment was lost to follow-up prior to the visit occurring 12 weeks post-treatment. The SVR12 was 100% (8/8) with ribavirin-free regimens and 15/16 (94%, PP = 15/16, 94%) in ribavirin-containing treatments (p = 0.47). SVR12 was 7/7 (100%; PP = 7/7, 100%) in cirrhotic participants and 16/17 (94%, PP = 16/17, 94%) in those without cirrhosis (p = 0.51). Ribavirin dosage was reduced to 600 mg in three patients following treatment initiation. One patient had a serious adverse event unrelated to the study drug and discontinued therapy at nine weeks to avoid a drug-drug interaction with colchicine for a severe gout flare resulting in hospitalization. No other serious adverse events were reported during the study.

Metabolic Measures and HCV Treatment-All Patients
The means of the metabolic measures across the study period are presented in Table 2 and changes from baseline at each time point are reported in Figures 1-4. Changes were observed for HbA1c and lipid measures across the study period ( Table 2 and          Missing data for all apolipoproteins apoA1-E at W12 (n = 1); W12PT (n = 2); W24PT (n = 4).  * The estimated mean of the metabolic parameters at each study visit is adjusted for treatment group, baseline BMI, and baseline viral load by the linear or generalized linear mixed model. The p-value reflects the overall effect of time on the outcome across all study visits and study groups as estimated by the linear and generalized mixed models. ** HbA1c was also adjusted for the effect of hemoglobin.

Ribavirin and Metabolic Measures
Ribavirin-containing treatment was associated with a decrease in LDL-C at Weeks 4 and 12 ( Figure 3C). This effect did not persist post-treatment. HbA1c decreased during treatment in the ribavirin-exposed group (Week 4: −0.50, 95% CI −0.91-−0.08; Week 12: −1.1, 95% CI −1.5-−0.68), but this change was not sustained post-treatment. No differences in HOMA-IR at each time point were observed in ribavirin-exposed participants compared to those who did not receive ribavirin. Compared to baseline, on-treatment decreases were observed at Week 4 in apoA2 and apoB or at Weeks 4 and 12 in apoE in ribavirin recipients ( Figure 4A,B,F).

Cirrhosis and Metabolic Measures
Participants with cirrhosis had higher baseline insulin and HOMA-IR than non-cirrhotic patients in the RBV-sparing group (HOMA-IR mean difference 2.2, 95% CI 0.58-3.8, p = 0.01; insulin mean difference 52.5, 95% CI 8.5-96.4, p = 0.02). These measures remained unchanged from baseline while on treatment and 12 weeks following dosing in the overall study population as well as in cirrhotic and non-cirrhotic participants. Twenty-four weeks post-treatment, an increase in insulin (mean change 70.3, 95% CI 24.0-116.6) and HOMA-IR (mean change 2.9, 95% CI 1.2-4.7) was noted in cirrhotic participants, which was not observed in non-cirrhotic participants. When the results of a single outlier were removed, this finding was no longer present ( Figure 1). Participants with cirrhosis had similar trajectories of HbA1c and lipid measures compared to those in the non-cirrhotic ribavirin-containing treatment group (Figures 2D and 3A-D). apoA1, apoA2, and apoE were lower at baseline in the cirrhotic participants compared with non-cirrhotic participants ( Figure 4A,B,F). There were no observed changes from baseline in mean glucose and apolipoprotein measures among cirrhotic patients over the study period.

Liver Assessment
Twelve weeks post-treatment, a mean increase from baseline in the controlled attenuation parameter (CAP) was observed (30.3 ± 63.5, p = 0.05) ( Figure 5). No differences in CAP score were observed in the treatment group (p = 0.94). Overall mean liver stiffness scores did not change from baseline at 12 weeks post-treatment (−0.61 ± 5.9 kPA, p = 0.64). No differences were observed in liver stiffness over time by treatment group (p = 0.96) (data not shown). (Figures 4A, B and F).

Cirrhosis and Metabolic Measures
Participants with cirrhosis had higher baseline insulin and HOMA-IR than non-cirrhotic patients in the RBV-sparing group (HOMA-IR mean difference 2.2, 95% CI 0.58-3.8, p = 0.01; insulin mean difference 52.5, 95% CI 8.5-96.4, p = 0.02). These measures remained unchanged from baseline while on treatment and 12 weeks following dosing in the overall study population as well as in cirrhotic and non-cirrhotic participants. Twenty-four weeks post-treatment, an increase in insulin (mean change 70.3, 95% CI 24.0-116.6) and HOMA-IR (mean change 2.9, 95% CI 1.2-4.7) was noted in cirrhotic participants, which was not observed in non-cirrhotic participants. When the results of a single outlier were removed, this finding was no longer present ( Figure 1). Participants with cirrhosis had similar trajectories of HbA1c and lipid measures compared to those in the non-cirrhotic ribavirincontaining treatment group (Figures 2D and 3A-D). apoA1, apoA2, and apoE were lower at baseline in the cirrhotic participants compared with noncirrhotic participants (Figures 4A, B and F). There were no observed changes from baseline in mean glucose and apolipoprotein measures among cirrhotic patients over the study period.

Liver Assessment
Twelve weeks post-treatment, a mean increase from baseline in the controlled attenuation parameter (CAP) was observed (30.3 ± 63.5, p = 0.05) ( Figure 5). No differences in CAP score were observed in the treatment group (p = 0.94). Overall mean liver stiffness scores did not change from baseline at 12 weeks post-treatment (−0.61 ± 5.9 kPA, p = 0.64). No differences were observed in liver stiffness over time by treatment group (p = 0.96) (data not shown).

Discussion
An association between chronic HCV infection and impaired glucose and lipid metabolism has been extensively described in the literature [14][15][16][30][31][32][33]. Although the exact mechanisms by which HCV interferes with these metabolic pathways has not been fully established, there is evidence to suggest that viral clearance achieved with interferon-based HCV treatment may perturb these metabolic measures [8][9][10]18,19,25]. In contrast, there has been minimal evaluation of the metabolic effects of interferon-free DAA treatment or the specific effects of ribavirin while in therapy or after SVR. We present a prospective comprehensive analysis of metabolic findings of individuals undergoing interferon-free, DAA treatment (with and without ribavirin) as well as the post-SVR outcomes of HCV genotype 1 infection.
We observed no consistent on-treatment or post-SVR change in measures of glucose homeostasis. HOMA-IR did not improve from baseline during the treatment phase or post-DAA dosing in the overall study population or any subgroup ( Table 2, Figures 1 and 2A). The latter is consistent with results from a Messiner et al. study in which no difference in HOMA-IR between baseline, on-treatment and the post-treatment phase were observed in patients treated with an IFN-free regimen [24]. This challenges the hypothesis that HCV clearance may reduce insulin resistance and/or reduce the risk of developing diabetes [8][9][10]. As suggested in Figure 1, it is possible that the complex interactions between liver fibrosis status, HCV RNA clearance, and glucose-insulin homeostasis may result in heterogeneous short-and long-term outcomes regarding diabetes risk.
Our evaluation suggests a correlation between the degree of liver fibrosis and insulin resistance (see Figure 2 baseline results). Insulin resistance is a critical factor in promoting the progression of hepatic fibrosis. This is thought to be mediated via the direct effect of insulin on hepatic stellate cells [27]. Insulin resistance is a common and well-known feature of cirrhosis irrespective of etiology and an important contributor to hepatogenous diabetes [34,35].
There was a difference in HbA1c measures based on ribavirin exposure. A decrease in HbA1c levels was observed in ribavirin recipients during the treatment phase. As there were no differences in fasting glucose or insulin levels at these time points, HbA1c may not provide a true reflection of glycemic control in this group. Ribavirin reduces erythrocyte lifespan and in turn may lead to falsely low HbA1c levels [36,37]. It is unclear if other studies have considered the ribavirin effects on hemoglobin when describing improvements in HbA1c levels while on therapy [10,24,38,39]. One recent evaluation specifically indicated that HbA1c was not measured in ribavirin-exposed participants for a minimum of 3 months post-treatment due to the known hemolytic effects of this medication [25]. This group reported that HCV eradication with a sofosbuvir-based regimen resulted in a decrease in HbA1c levels (pre-treatment 6.66 +/− 0.95 versus post-treatment 6.14 +/− 0.65, p < 0.005).
In addition to evaluating measures of glucose homeostasis, we also assessed the lipid profile. As described previously, viral clearance was associated with increased on-treatment and post-treatment total cholesterol, LDL-C, and TG levels (Table 2, Figure 3A,C,D) [18,19,24]. The presence of cirrhosis did not affect lipid levels during or after treatment. Differences in the direction of lipid changes during treatment were observed based on ribavirin exposure, suggesting a direct effect of this medication on lipid homeostasis. Specifically, reduced total cholesterol, LDL-C, and decreased apoB levels were noted in non-cirrhotic RBV recipients at Weeks 4 and 12 ( Figure 3A,C and Figure 4C). Post-treatment, this RBV-specific lipid effect was no longer observed. While the observed overall on-treatment and post-SVR increase in total and LDL-C, as well as TG levels, are consistent with other studies, the marked on-treatment effect of ribavirin on the lipid profile is novel. This warrants further evaluation given the possible role in the management of dyslipidemia.
We conducted an intensive evaluation of apolipoproteins. The formation of HCV lipoviral particles (LVP) occurs within the endoplasmic reticulum at interfaces between lipid droplets and is closely intertwined with lipid and lipoprotein metabolism [40]. As such, apolipoproteins are essential regulators of lipid metabolism and play an important role in the HCV life cycle [41]. apoB forms the protein backbone for the formation of TG lipoproteins including chylomicrons and VLDL as well as post-hydrolysis LDL. In contrast to apoB, other apolipoproteins including apoA1, apoC2, apoC3, and apoE are easily disassociated and can be exchanged between different classes of lipoproteins and the surface of HCV LVPs [41]. apoE association with HCV lipoviral particles enhances infectivity [13,[42][43][44][45]. We observed a decline in apoE during the 12-week treatment phase in ribavirin-exposed participants. This was similarly described by Younossi et al. This further suggests a ribavirin-specific effect on lipid homeostasis.
The downregulation of apoA1 has been associated with decreases in HCV RNA levels and as such has been implicated in viral replication [46]. Our study demonstrated decreased apoA1 levels in non-cirrhotic, RBV unexposed treatment recipients (Table 2, Figure 4A). This was in contrast to Younossi et al. where HCV treatment did not influence apoA1 levels. Although associated with low apoB levels, the role of apoB in HCV assembly is unclear [31,43,47,48]. apoB levels increased from baseline to 24 weeks after treatment in patients treated with RBV irrespective of the degree of liver fibrosis (Table 2, Figure 4C). This is in contrast to the study by Younossi et al. where treatment did not influence apoB levels. apoC2, an activator, and apoC3, an inhibitor of lipoprotein lipase activity, have been implicated in HCV infection via their modulation of the LVP catabolism. Lipoprotein lipase activity has been shown to inversely correlate with HCV RNA levels [49][50][51], and low apoC2 levels correlate with increased HCV infection and more advanced liver disease [50]. Consistent with other studies, we demonstrated an increase in apoC2/C3 levels post-treatment in non-cirrhotic patients irrespective of ribavirin exposure and provide further evidence that more advanced liver disease is associated with low levels of these apolipoproteins pre-and post-treatment (Table 2, Figure 4D,E).
CAP score, a measure of liver steatosis, was noted to increase from baseline at the end of treatment and again 12 weeks after treatment when SVR was achieved ( Figure 5). This finding was consistent irrespective of the presence or absence of cirrhosis or RBV exposure. The explanation of this finding is unclear. The relationship between lipid profile perturbation and liver steatosis resulting from HCV RNA clearance requires further evaluation. This increase in CAP score did not correlate with the fibrosis score, which remained unchanged from baseline-at least during the short period of post-treatment follow-up.
There are limitations requiring consideration in this exploratory study. As all but one participant achieved a SVR12, it was not possible to compare metabolic outcomes between those cured and treatment failures. This was a non-randomized, open-label study and as such is subject to selection bias. Given the small sample size, it was not possible to adjust for all potential confounders, and the potential for Type I and II statistical error is acknowledged. Our study was powered to detect differences in HOMA-IR of 1.0. We were limited in detecting smaller changes in this primary outcome. Some evaluations have suggested that the most marked improvements in insulin sensitivity and glucose metabolism are achieved in patients with higher baseline HOMA-IR levels and established diabetes, respectively [25,38,39]. Our study population was characterized by relatively low HOMA-IR levels at baseline compared to these studies.

Conclusions
Exposure to DAA and ribavirin may influence lipid and apolipoprotein, but not glucose, parameters during HCV treatment. Furthermore, an HCV cure with DAA treatment results in increased lipid levels. Our study suggests that ribavirin exposure may play a role in mitigating some of the on-treatment lipid changes observed as HCV is cleared while on HCV treatment. Hepatic steatosis may also be affected by the clearance and cure of HCV.
Author Contributions: C.C. was the lead investigator on the project and responsible for conception design and protocol development. M.-A.D. and A.C. contributed to study design and protocol development. The data analysis was completed by C.G. and reviewed by M.-A.D., A.C., E.M., and C.C. M.-A.D. was responsible for writing the initial draft of the manuscript. All authors were responsible for editing and approving the final manuscript.
Funding: This research was funded by Abbvie who is acknowledged for providing unrestricted funding and medication for the conduct of this study. CLC has served as an advisor and speaker as well as received program and research support from Abbvie, Gilead, and Merck. MAD has served as an advisor and speaker for Gilead.

Conflicts of Interest:
The authors declare no conflict of interest.