Second-Phase Hepatitis C Plasma Viral Kinetics Directly Reflects Reduced Intrahepatic Burden of Hepatitis C Virus

Abstract

Background

Mathematical models explain how antivirals control viral infections. Hepatitis C virus (HCV) treatment results in at least 2 phases of decline in viremia. The first phase reflects clearance of rapidly produced virions. The second phase is hypothesized to derive from loss of infected cells but has been challenging to prove.

Methods

Using single-cell methods, we quantified the number of hepatitis C virus (HCV)-infected hepatocytes in liver biopsies taken before and within 7 days of initiating direct-acting antivirals (DAAs) in a double-blinded randomized controlled trial testing 2 (sofosbuvir-velpatasvir) versus 3 (sofosbuvir-velpatasvir-voxilaprevir) DAAs.

Results

We employed thousands of intrahepatic measurements in 10 persons with chronic genotype 1a HCV infection: median proportion of infected hepatocytes declined from 11.3% (range, 1.3%–59%) to 0.6% (range, <0.3%–5.8%), a loss of 75%–95% infected hepatocytes. Plasma viremia correlated with numbers of HCV-infected hepatocytes (r = 0.77; P < .0001). Second-phase plasma dynamics and changes in infected hepatocytes were indistinct (P = .16), demonstrating that second-phase viral dynamics derive from loss of infected cells. DAAs led to a decline in intracellular HCV RNA and interferon-stimulated gene expression (P < .05 for both).

Conclusions

We proved that second-phase viral dynamics reflect decay of intrahepatic burden of HCV, partly due to clearance of HCV RNA from hepatocytes.

Clinical Trials Registration

NCT02938013.

Hepatitis C virus (HCV) infects 70 million people worldwide [1]. After infection, up to 85% of people develop persistent viremia. HCV is a positive-sense, single-stranded RNA virus without a DNA intermediate, making it unusual among viruses as a prevalent cause of chronic infection. A hallmark of chronic infection is that plasma HCV viremia is remarkably stable despite decades of infection, underscoring the equilibrium between replication and control that results in a virological set point. We previously found a link between plasma HCV RNA levels and the number of HCV-infected hepatocytes [2–6]. Indeed, because the virus principally infects hepatocytes in the liver and does not have a latent phase, HCV is a model infection for understanding the quantitative relationship between intracellular and extracellular phases of the viral lifecycle. In the present investigation, we extended this link by studying both phases during antiviral therapy.

Direct acting antivirals (DAAs) have revolutionized HCV treatment, resulting in high rates of sustained virologic response (SVR). Treatment with DAAs leads to at least 2 distinct phases of decline in plasma viremia: a rapid first phase occurs over the first 24 hours, while a slower second phase continues until viremia is no longer clinically detectable. The first phase results from rapid clearance of virions from circulation [7]. The second phase has been speculated to derive from the slower loss of infected hepatocytes, but this has not been proven because the intrahepatic burden of HCV has not been easily estimated until recently [3–6]. The rate of loss of infected hepatocytes is the key determinant of the required duration of DAA treatment to achieve SVR. That duration is necessarily longer than the time when plasma HCV RNA levels become undetectable. We designed an intensive clinical study of liver and plasma viral dynamics in persons with chronic HCV genotype 1a infection who were treatment naive and who initiated DAAs. We analyzed plasma and paired liver biopsies collected from participants before and during the first week of treatment with 2 (sofosbuvir/velpatasvir [SV]) or 3 (sofosbuvir/velpatasvir + voxilaprevir [SVV]) DAAs in a randomized controlled clinical trial (clinicaltrials.gov identifier NCT02938013). The principal objectives were to (1) analyze the relationship between the second-phase plasma viral decline and of loss of infected hepatocytes, and (2) test whether the addition of a third DAA would accelerate early plasma or liver viral dynamics.

METHODS

Enrollment and Study Observations

Ten adults with chronic genotype subtype 1a HCV infection (≥ 6 months) who were treatment naive and who had METAVIR ≤ F2 (transient elastography liver stiffness <12.5 kPa) were enrolled after confirmation that they were had negative serological testing for coinfection with human immunodeficiency virus (HIV) or hepatitis B virus (HBV) [8]. Inclusion and exclusion criteria and schedule of events are stated in Supplementary Methods. Briefly, participants had phlebotomy and a liver biopsy performed on day 0, just prior to administration of study drug; phlebotomy continued at structured intervals for 14 days and a second liver biopsy was performed between days 4 and 7 (Supplementary Figure 1). Participants were admitted to the clinical research unit on day 0 and randomly assigned to receive treatment with sofosbuvir 400 mg/velpatasvir 100 mg (Epclusa: Gilead Sciences) or sofosbuvir 400 mg/velpatasvir 100 mg/voxilaprevir 100 mg (Vosevi; Gilead Sciences); regimens were administered as 1 tablet daily for 7 days, after which all participants received sofosbuvir 400 mg/velpatasvir 100 mg for an additional 11 weeks. HCV RNA testing and deep sequencing was performed on plasma samples and single-cell HCV analysis and innate immune characterization was performed on liver tissues (Supplementary Methods). All participants gave written informed consent to the protocol and for the use of human specimens as approved by the Johns Hopkins Institutional Review Board.

Statistical Analysis

We used Student t tests or nonparametric versions (Wilcoxon tests) and Fisher exact tests to compare continuous and discrete variables across treatments, as indicated in the text. To compare variables, such as slopes, within individuals, we used paired versions of those tests. We used linear mixed models to estimate the slopes of decay of HCV RNA in plasma in each participant and to compare differences between treatment groups.

RESULTS

We enrolled 10 persons with chronic genotype 1a HCV infection into a clinical trial of SV versus SVV for 7 days; all participants received SV for an additional 11 weeks (total 12 weeks of treatment; Supplementary Figure 1). Median age was 52 years (range, 28–60 years), 5/10 were women, and 7/10 were black (Table 1). Median liver stiffness was 7 kPa (range, 4–11 kPa), and baseline plasma HCV RNA levels were 5.74 log₁₀ IU/mL (range, 5.09–6.94 log₁₀ IU/mL). There were no differences in randomization between the SV and SVV groups in any demographic characteristic (Table 1). All participants achieved SVR at 12 weeks (SVR12).

Baseline plasma HCV RNA levels were not significantly different between SV and SVV groups. Plasma viral kinetics showed rapid first-phase declines by day 1 followed by slower second-phase declines over the subsequent 6 days (Figure 1 and Supplementary Figure 2). There were no differences in plasma HCV RNA decay between SV and SVV groups over the first 24 hours. By day 7, there was a 0.9-log difference in plasma HCV RNA fold-change that did not reach significance: the median change for SV was −3.96 log₁₀ IU/mL (range, −2.85 to −4.34 log₁₀ IU/mL), compared to −4.86 log₁₀ IU/mL (range, −3.99 to −5.20 log₁₀ IU/mL) for SVV (P = .056): 2/5 persons in the SVV group had undetectable plasma HCV RNA levels by day 7, while all persons in the SV group still had detectable plasma HCV RNA levels by day 7 (Supplementary Table 1). We fit a simple biphasic (biexponential) model to the viral load over the first 7 days and found a significant difference in second-phase slopes between the 2 groups (P = .02) without differences in first-phase slopes.

We used single-cell laser capture microdissection (scLCM) to isolate 300 to 400 hepatocytes per liver biopsy per time point, as we have done previously [3–6]. We improved on our original method by depositing single hepatocytes into individual microfuge tubes and extracted their RNA separately, rather than combining hepatocytes for analysis. We quantified 7SL, an abundant intracellular RNA, in each hepatocyte by quantitative polymerase chain reaction (qPCR) to confirm single cells and to exclude cell fragments. Out of 7680 total captures, we excluded 1141 (15%) as fragments (Supplementary Figure 3A), also confirming that HCV RNA amounts were not correlated with the amount of material captured (Supplementary Figure 3B).

The median proportion of HCV-infected hepatocytes at baseline was 11.3% (range, 1.3%–59%) and did not differ between SV and SVV (P = .31; Table 2). In both groups, the proportion of HCV-infected hepatocytes declined at second biopsy to a median of 0.6% (range, <0.3%–5.8%), reflecting loss of 75% to > 95% HCV-infected hepatocytes within 7 days. There were no significant differences in proportions of infected hepatocytes in second biopsies between SV and SVV groups.

We used a mixed effect approach to compare slopes of decay in numbers of infected hepatocytes between the SV versus SVV, extrapolating these numbers from the proportions. Although there were only 2 datapoints per participant, this approach uses all data simultaneously and allows estimation of population means for slopes of decay. There were no differences in slopes of decay between the 2 treatment groups. We extrapolated the time to presumed intrahepatic eradication of HCV, assuming 100% adherence and log-linear loss of HCV-infected hepatocytes at the same rate as observed in the biopsies. To this end, we estimated the burden of infection in the whole liver for each person at baseline and calculated the time until only 1 HCV-infected hepatocyte remained. For the participant who did not have detectable intrahepatic HCV at second biopsy, a conservative value of the limit of detection was assumed as if a single infected hepatocyte was found. We calculated the predicted time to eradication for each person, estimating a median of 53 days, which is consistent with clinical guidelines for the duration of DAA treatment that is required to achieve SVR in >95% of people. Separating by group, SVV had a median predicted time to eradication of 49 days (range, 26–70 days) compared to SV, which had a median predicted time of 58 days (range, 28–116 days) (Figure 1). One individual who received SV had a predicted time to eradication of 116 days, which was inconsistent with their observed SVR after 12 weeks of treatment. In this individual, however, 13/234 (5.6%) hepatocytes were infected in the first biopsy and 5/345 (1.45%) hepatocytes were infected at the second biopsy; thus, if we consider the lower limit of the (binomial) 95% confidence interval as also a reasonable estimate of the fraction infected cells (0.47%), then the time for eradication in this person would be 63 days.

There was a close correlation between plasma HCV RNA and the contemporaneous estimated total hepatic burden of HCV-infected hepatocytes (r = 0.77; P < .0001; Figure 2). We overlaid each participant's plasma HCV RNA levels corresponding to their second-phase decline with their change in intrahepatic burdens over the first 7 days (Figure 3). Strikingly, a comparison of the slopes of decay of virus in plasma with the slopes of decay of infected hepatocytes revealed that these were not statistically different (P = .16, Wilcoxon paired test). Taken together with the predicted times to eradication, these data strongly indicate that the number of infected cells is the key determinant of plasma viremia, and that the decay in infected hepatocytes is a major contributor to second-phase plasma kinetics and, consequently, time to eradication.

As noted earlier, HCV is unique among common chronic viral infections in lacking a DNA intermediate. Hypothetically, then, cells can be individually cured by DAAs that interrupt production and accumulation of nascent RNA genomes after residual HCV RNAs are packaged into virions and exit from the cell. To test this hypothesis, we quantified HCV RNA in individual infected hepatocytes before and after DAA initiation. HCV-infected hepatocytes showed substantially decreased levels of HCV RNA in second compared to first biopsies (P = .0012; Figure 4). However, despite similar HCV RNA levels between SV and SVV before treatment (P = .65), we found no evidence that 3 DAAs diminished intracellular HCV RNA levels more rapidly than 2 DAAs (P = .15) (Figure 4).

To test the role of preexisting resistance to DAAs in viral dynamics, we performed deep sequencing on plasma samples at baseline of distinct regions of the HCV genome corresponding to major resistance-associated substitutions (RAS) in the NS3, NS5A, and NS5B genes that we first PCR amplified (Supplementary Tables 2 and 3). All individuals had NS3 RAS present at baseline, excluding 1 participant in the SV group without detectable NS3 reads. Among the evaluable participants, 4/4 SV participants had NS3 RAS that were >5% of all reads and 2/4 had NS3 RAS that were >50% of all reads (Q80K and Q80L), while at least 3/5 SVV participants had NS3 RAS >5% of all reads at that location and 2/5 had NS3 RAS that were >50% of all reads (both Q80K). NS5A RAS were present in all participants but composed <1% of all reads, except for 1 SVV individual for whom 19.4% of all reads were L31M. No NS5B RAS were detected in any individual. We saw no evidence of emergent NS5A RAS in plasma during DAA treatment by day 4 (Supplementary Table 3). The NS3 amplicon was not detectable in any individuals after DAA initiation. Collectively, we found no evidence of differences in preexisting or emergent RAS between SV and SVV.

There is a strong association between HCV infection and intrahepatic expression of interferon-stimulated genes (ISGs) [9]. The viral nonstructural gene NS3 has been shown to cleave the intracellular innate immune signaling adapters mitochondrial antiviral signaling protein (MAVS) and TIR domain-containing adapter inducing interferon-β (TRIF) [10–12]. These actions of NS3 are thought to contribute to HCV persistence by attenuating the intracellular antiviral ISG response. Conversely, pharmacologic NS3 inhibition has been proposed to restore those antiviral responses. Because voxilaprevir is a potent NS3 inhibitor, we quantified ISG expression in hepatocytes from SVV participants compared to SV participants by targeting interferon induced transmembrane protein-3 (IFITM3), an abundant ISG that is a hallmark of the innate immune response to chronic HCV. At baseline, 64% of hepatocytes in SV participants expressed IFITM3 mRNA compared to 72% of hepatocytes in SVV participants (P = .0117). At second biopsy, 50% percent of hepatocytes showed ISG expression in SV participants compared to 65% in SVV participants (P < .0001).

We also quantified intracellular amounts of IFITM3 at baseline and after DAAs: although similar at baseline (P = .199), IFITM3 levels decreased at the time of the second biopsy (P = .0477 for SV and P < .00001 for SVV; Figure 5). Importantly, IFITM3 levels in the SVV group appeared to decrease more than in the SV group (decreases of 24.8 copies vs 5.7 copies, respectively, P < .00001), suggesting that protease inhibitor intensification may modulate type 1 interferon signaling, although not in a predictable manner.

DISCUSSION

In the present study, we have provided empirical evidence that links the second-phase decline in viral levels to changes in the number of HCV-infected hepatocytes within the first week of initiating DAAs. The single-cell quantification of HCV RNA revealed that intracellular HCV RNA levels were significantly diminished with DAAs, confirming in people a previously speculated mechanism for the action of DAAs: cure of already infected cells. Our data informed the estimated times until viral eradication, which are consistent with HCV treatment guidelines.

The second phase of viral decline during combination DAA treatment had been attributed to the loss of HCV-infected cells [13], but this has been difficult to prove. While plasma HCV RNA levels do reflect an aspect of the intrahepatic burden of infection, these levels rapidly become undetectable during DAA treatment [14]. The second-phase viral decline, if indeed tied to the loss of HCV-infected cells, is a direct measure of the time to viral eradication with therapy. Our calculated times to eradication were consistent with treatment durations prescribed by the HCV treatment guidelines. Nevertheless, although presently the guidelines advocate treatment durations of 8–12 weeks that achieve SVR in 95% of treatment-naive persons, there are still people who do not achieve SVR with those durations and who might benefit from prolonged treatment. Alternatively, there may be people who require significantly shorter durations of treatment. Patient-targeted therapy, therefore, might be best directed by understanding the intrahepatic burden of infection and the rate of loss of HCV-infected hepatocytes over time. Indeed, our intrahepatic measurements strongly support that plasma HCV RNA levels may be a reasonable surrogate for intrahepatic burden. A natural extension of this concept is in targeting patients with low pretreatment plasma HCV RNA levels for shortened therapy with treatment intensification, which has variable support from previous trials [15–21]. Notably, we did not observe a significant difference in the intrahepatic HCV burden between SV and SVV, which is consistent with the high rate of non-SVR with 8 weeks of SV and with the inadequate response among people with genotype 1a who are treated with 8 weeks of SVV.

Innate immune modification by NS3 protease inhibitors has been speculated. We did not find clear evidence that the addition of voxilaprevir enhanced intracellular ISG expression in hepatocytes, instead finding that intracellular IFITM3 expression decreased in both the SV and SVV groups. The decrease in intrahepatic ISG expression within 7 days of DAA initiation may have other clinical implications: there is a risk of HBV reactivation in people who are HCV coinfected that has been tied to derepression of HBV after normalization of type 1 interferon signaling [22–24]. The rapidity of hepatocellular ISGs decline after treatment initiation that we found strengthens this hypothesis.

We were met with several challenges in the present study. Our sample size is small, due to the nature of intensive study of each individual, including 2 biopsies a short time apart. Another related challenge is that whereas we examined 300–400 hepatocytes per liver biopsy and obtained consistent results, our measurements are from a small fraction of the full liver. These limitations are mitigated by the strong correlation between the estimates of intrahepatic HCV burden and contemporaneous plasma HCV RNA levels (Figure 2), implying that the HCV virologic set point is governed by the fraction of infected hepatocytes. Another challenge is the limit of reporting of our assay of 1 IU/cell, previously calculated as approximately 2 copies/cell [6]. Although we routinely observe values lower than this limit, they are outside the log-linear range and upon repeat testing are not always reproduced. Moreover, our extrapolated time to viral eradication from the liver is remarkably consistent with what is observed clinically, underscoring the accuracy of our estimates of HCV-infected hepatocytes.

In closing, we investigated liver and plasma viral kinetics in people with chronic genotype 1a HCV infection who were treatment naive, embedded in a clinical trial testing treatment intensification with the HCV protease inhibitor voxilaprevir in addition to sofosbuvir and velpatasvir. We demonstrated that second-phase plasma viral kinetics derive directly from the loss of HCV-infected hepatocytes after initiating treatment and were not influenced by preexisting RAS. The estimated time to viral eradication was consistent with real-world data. In addition, intracellular HCV RNA levels decreased within 7 days, supporting cure of individual cells as a novel mechanism of DAA action. Treatment intensification with voxilaprevir enhanced second-phase viral dynamics and may have modulated intracellular innate immune signaling. The intracellular phase of viral infections is a key, but overlooked, component of human viral infections, and its quantitative study is critical for establishing the number of infected cells, a major determinant of the virological set point and time to eradication.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We express gratitude to the study participants and coordinators for their support, efforts, and contributions.

Author contributions. J. S. contributed data acquisition, data analysis, and drafted the manuscript. A. B., R. M. R., and M. S. S. conceived the study, performed data analysis, and drafted the manuscript. J. L.-L. and K. B. performed data acquisition. J. Q. performed data acquisition and data analysis. K. W. performed study enrollment.

Financial support. This work was supported by National Institutes of Health (grant numbers R01 DA016065 to M. S. S. and R01 AI116868 to A. B. and R. M. R.). Study medications were provided by Gilead Sciences.

References

Author notes