Emergence of resistance-associated variants during sofosbuvir treatment in chronically infected hepatitis E patients

Abstract

Background and aims: Chronic HEV infections remain a serious problem in immunocompromised patients, as specifically approved antiviral drugs are unavailable. In 2020, a 24-week multicenter phase II pilot trial was carried out, evaluating the nucleotide analog sofosbuvir by treating nine chronically HEV-infected patients with sofosbuvir (Trial Number NCT03282474). During the study, antiviral therapy reduced virus RNA levels initially but did not lead to a sustained virologic response. Here, we characterize the changes in HEV intrahost populations during sofosbuvir treatment to identify the emergence of treatment-associated variants.

Approach and results: We performed high-throughput sequencing on RNA-dependent RNA polymerase sequences to characterize viral population dynamics in study participants. Subsequently, we used an HEV-based reporter replicon system to investigate sofosbuvir sensitivity in high-frequency variants. Most patients had heterogenous HEV populations, suggesting high adaptability to treatment-related selection pressures. We identified numerous amino acid alterations emerging during treatment and found that the EC 50 of patient-derived replicon constructs was up to ~12-fold higher than the wild-type control, suggesting that variants associated with lower drug sensitivity were selected during sofosbuvir treatment. In particular, a single amino acid substitution (A1343V) in the finger domain of ORF1 could reduce susceptibility to sofosbuvir significantly in 8 of 9 patients.

Conclusions: In conclusion, viral population dynamics played a critical role during antiviral treatment. High population diversity during sofosbuvir treatment led to the selection of variants (especially A1343V) with lower sensitivity to the drug, uncovering a novel mechanism of resistance-associated variants during sofosbuvir treatment.

Graphical abstract

FIGURE 1

Viral diversity in patients chronically infected with HEV under sofosbuvir treatment. (A) Frequency of nonreference variants within the RdRp of HEV ORF1 compared to the time of treatment initiation (baseline). The cutoff for biological relevance (lower dashed line) was set at 3.287% according to Illumina error rates. The top dashed line (50%) indicates consensus sequence mutations. Each dot represents a variant at a given time point, which is color coded; (B) quantification of variants for each patient and time point that occur above the error cutoff. The size ratio of the dot indicates the percentage of consensus sequence changes compared to the total number of mutations. Labels indicate size normalized variation. Abbreviations: RdRp, RNA-dependent RNA polymerase.

FIGURE 2

Identification of patient-derived consensus substitution. The left panel shows HEV RNA quantity over the treatment period. The dark gray highlighted area refers to the time during which patients were treated with sofosbuvir. Red dots indicate time points at which variants were engineered into the replicon system (Figure 3). The heatmap (right side) shows all changes in the amino acid consensus sequence and their respective frequency in percent per time point.

FIGURE 3

Impact of patient-derived consensus variants on sofosbuvir sensitivity and viral fitness. Amino acid substitutions found in P1–P4 and P6–P9 at specific time points during sofosbuvir treatment (bottom panel) were introduced into a p6 Kernow-C1 replicon harboring a Gaussia luciferase reporter gene and delivered into HepG2 cells by electroporation. For subsequent dose-response experiments, cells were treated with sofosbuvir concentrations ranging from 100 µM to 0.015 µM for 72 hours. (A) EC₅₀ values were determined by nonlinear regression. Numbers at the top indicate fold changes compared to wild-type. Error bars = 95% CI. (B) Replication capacity of variants was determined by plotting the measurement of RLU of vehicle-treated (DMSO) cells in GraphPad Prism (middle panel). Error bars = ± SD. Dashed lines represent wild-type control (A) and replication-deficient control (GAA) (B). n ≥ 3. Abbreviation: HepG2, human hepatoma cells; RLU, relative light units.

FIGURE 4

Emergence of A1343V as a resistance-associated substitution. (A) All positions with at least 1 substitution above 50%. Each box represents a patient with a substitution at that position; see the color code in the legend. The grey shaded area refers to all samples sequenced at that time point. (B) Alpha-fold model of the HEV RdRp (based on Kernow-C1 p6 nucleotide sequence). Residues with consensus sequence changes are highlighted in red. Abbreviation: RdRp, RNA-dependent RNA polymerase.

FIGURE 5

Impact of single amino acid variants on sofosbuvir sensitivity and viral fitness. (A) Shannon entropy of HEV-3 RdRp sequences (sliding window = 20 and step = 1) from publicly available sequences was downloaded from NCBI. Colored areas correspond to polymerase domains: purple, finger; pink, palm; green, thumb. (B) Frequency of amino acids at positions of interest (1343, 1383, 1384, 1479, and 1634; highlighted in red and with label) in the publicly available datasets and the intrahost diversity data. (C) Pearson correlation (r) of SNV data at baseline and during treatment. p values below 0.05 were considered significant. (D) HEV-3 amino acid sequence variability in the proportion of encoded amino acids of selected residues from the public database. (E) In vitro analysis of single amino acid variants on viral sensitivity to sofosbuvir treatment (error bars = 95% CI), and numbers at the top indicate fold changes compared to wild-type, and (F) replication capacity (error bars = ± SD). HepG2 cells were transfected with single amino acid variants harboring a Gaussia luciferase gene and treated with various doses (100 µM–0.015 µM) of sofosbuvir in vitro. Supernatants were sampled after 72 hours, and variant replication was measured through reporter luciferase read-out and normalized to the respective vehicle-treated control (DMSO). EC₅₀ values and replication capacity were determined by the 4-parameter logistic model and measurement of relative light units of vehicle-treated (DMSO) cells, respectively. Dashed lines represent wild-type control (E) and replication-deficient control (GAA) (F). n ≥ 3. Abbreviations: HepG2, human hepatoma cells; RdRp, RNA-dependent RNA polymerase; SNV, single nucleotide variant.