Potential conflict of interest: Dr. Corsa owns stock in Gilead. Dr. Snow-Lampart owns stock in Gilead. Dr. Flaherty owns stock in Gilead. Dr. Kitrinos owns stock in Gilead. Dr. Miller owns stock in Gilead. Dr. Liu owns stock in Gilead.
One major challenge in the treatment of chronic hepatitis B is to maintain long-term viral suppression without promoting the selection of drug-resistant mutations. We analyzed data from 347 hepatitis B e antigen-negative and 238 hepatitis B e antigen-positive patients receiving tenofovir disoproxil fumarate (TDF) in an open-label, long-term extension of two phase 3 studies. To date, resistance analyses have been completed for patients receiving up to 288 weeks (6 years) of TDF. Population sequencing of hepatitis B virus (HBV) polymerase/reverse transcriptase (pol/RT) was attempted for all patients at baseline, and any patient who remained viremic (HBV DNA ≥400 copies/mL [≥69 IU/mL]) at week 288 or at the end of treatment with TDF (n = 52) or emtricitabine (FTC)/TDF (n = 7). Phenotypic analyses were performed in HepG2 cells using recombinant HBV containing patient pol/RT sequences. Approximately half of the patients on open-label treatment who qualified for genotyping had pol/RT sequence changes compared to baseline (23/52 [44%] on TDF, 4/7 [57%] on FTC/TDF). Most changes were at polymorphic sites and none were associated with TDF resistance. Virologic breakthrough occurred infrequently and was associated with nonadherence to study medication in the majority of cases (12/16, 75%). Per protocol, 57 patients (10%) were eligible to switch to FTC/TDF; the majority had HBV DNA <400 copies/mL at their last study visit regardless of whether they switched to FTC/TDF (n = 34) or maintained TDF monotherapy (n = 17). No patient exhibited persistent viremia (HBV DNA never <400 copies/mL) after week 240. Conclusion: TDF monotherapy maintains effective suppression of HBV DNA through 288 weeks of treatment with no evidence of TDF resistance. (Hepatology 2014;59:434–442)
The treatment goals for chronic hepatitis B (CHB)-infected patients are to improve quality of life and decrease the risk of life-threatening complications by potent and durable suppression of hepatitis B virus (HBV) replication.[1-5] As most patients with CHB will require long-term treatment with nucleos(t)ide analogs, one of the greatest challenges is to maintain effective viral suppression without the selection of resistance mutations. Long-term treatment with first-generation nucleos(t)ide analogs leads to relatively high cumulative resistance rates, 71% after up to 4 years of treatment with lamivudine (LAM), and 29% in hepatitis B e antigen (HBeAg)-negative patients and 20% in HBeAg-positive patients after up to 5 years of treatment with adefovir (ADV).[7, 8] Furthermore, 25% of HBeAg-positive patients and 11% of HBeAg-negative patients developed resistance after up to 2 years of treatment with telbivudine. Entecavir (ETV) has a markedly lower rate of resistance: 1.2% among treatment-naïve patients treated for up to 5 years. However, ETV resistance develops rapidly in patients with prior resistance to LAM: the 5-year cumulative probability of developing ETV resistance is 51% in patients harboring LAM resistance.
Tenofovir disoproxil fumarate (TDF), an oral prodrug of tenofovir, is a nucleotide analog with potent anti-HBV activity that was approved for the treatment of CHB in 2008. In vitro, TDF has been shown to be equally effective against multiple HBV genotypes (A-H) as well as against LAM-resistant isolates. To date, there have been no confirmed reports of resistance selection during treatment with TDF for CHB.[14-16] Two multicenter, 48-week, randomized controlled trials of 641 predominantly treatment-naïve patients with CHB demonstrated that TDF was safe and significantly more effective than ADV, with 76% of HBeAg-positive and 93% of HBeAg-negative patients achieving complete HBV DNA suppression (HBV DNA <400 copies/mL). After week 48, patients could enroll in long-term open-label studies of TDF for an additional 7 years. This study describes the genotypic and phenotypic results from patients treated with TDF in these studies for up to 288 weeks (6 years) in order to evaluate the rate of resistance selection in these long-term trials with TDF.
Patients and Methods
The protocol, study design, patient enrollment criteria, and analysis methods for these clinical trials (GS-US-174-0102 [study 102] and GS-US-174-0103 [study 103]) have been previously described.[8, 15-17] All patients who completed the 48-week double-blind phase and consented to a liver biopsy at week 48 were eligible to enter open-label treatment with TDF for up to an additional 7 years. At or after week 72, patients with HBV DNA ≥400 copies/mL (≥69 IU/mL) (confirmed at two consecutive visits) could add emtricitabine (FTC) as a fixed-dose FTC/TDF combination tablet at the discretion of the investigator.
Population sequencing of the HBV polymerase/reverse transcriptase (pol/RT) gene was attempted on serum samples for all patients at baseline, annually in viremic patients with or without virologic breakthrough (HBV DNA ≥400 copies/mL), and in any viremic patient at the time of early discontinuation or switch to FTC/TDF. All patients were documented by pill counts to have been on study drug at the time of testing. Any patients who were viremic at the end of a treatment year and subsequently experienced HBV DNA suppression without treatment alteration were genotyped during the study but not included in the final analysis. All genotypic analysis methods have been reported previously. Based on an alignment of pol/RT sequences from all patients at baseline, conserved sites were defined as positions with only one amino acid present or at which two amino acids were present but the prevalence of the minority amino acid was <1%. All other positions within the pol/RT sequence were considered polymorphic. Overall, 64% of amino acid positions in the pol/RT sequence were classified as conserved and 36% as polymorphic.
In vitro phenotypic analyses of tenofovir susceptibility were attempted for patients who developed conserved-site changes, patients who developed polymorphic-site changes (if observed in more than one patient), and patients who experienced protocol-defined virologic breakthrough. Virologic breakthrough was defined as two consecutive HBV DNA values ≥400 copies/mL after previously achieving HBV DNA <400 copies/mL or a confirmed increase of ≥1 log10 copies/mL from the HBV DNA nadir while on study drug. Phenotypic analyses were conducted as previously described using HepG2 cells transiently transfected with recombinant HBV-containing patient serum pol/RT quasispecies.[16, 18] If a recombinant virus containing the sequence change of interest could not be generated from patient serum, the mutation was created by site-directed mutagenesis (QuikChange site-directed mutagenesis kit, Stratagene) with either a genotype A (pHY92) or genotype D (pCMVHBV) laboratory isolate. The interassay variability for tenofovir susceptibility is ≤2 fold of the mean EC50 values.
For patients on TDF shown to harbor the rtN236T mutation by population sequencing, the rtN236T mutant percentage was determined with MultiCode RTx allele-specific PCR modified for HBV rtN236T detection as described previously.
Adherence to Study Medication
For patients qualifying for genotypic analysis, adherence was assessed by evaluation of plasma tenofovir levels (liquid chromatography/mass spectroscopy) and by analysis of drug accountability records associated with case report forms or included in clinical deviation logs. An undetectable tenofovir level in the plasma indicated the patient had not been adherent to medication for 24 hours or longer. Nonadherence was defined as having undetectable tenofovir levels in the plasma and/or >30% returned pills at a given visit.
Overall, 641 patients enrolled across both studies (375 HBeAg-negative patients in study GS-US-174-0102 and 266 HBeAg-positive patients in study GS-US-174-0103). Of these, 585 patients (389 randomized to TDF and 196 to ADV) enrolled in the open-label phase, and 466 patients completed week 288 on therapy (298 randomized to TDF and 168 randomized to ADV); 119 patients discontinued during the open-label phase (Fig. 1). The majority of patients discontinued the study with HBV DNA <400 copies/mL (99/119, 83%). Overall, 18/119 patients withdrew due to a safety, tolerability, or efficacy reason; however, 15/18 (83%) of these patients had HBV DNA <400 copies/mL at the time of discontinuation.
A cumulative analysis of the pol/RT genotypic data from all patients who entered the open-label treatment phase and were viremic (HBV DNA ≥400 copies/mL) with or without virologic breakthrough at their last timepoint on treatment (week 288 or at early discontinuation) or switched to FTC/TDF combination therapy was conducted. The analysis includes all patients who initiated open-label treatment with genotypic data from their last visit on TDF monotherapy as well as their last visit on FTC/TDF, if applicable. Overall, the majority of patients who entered open-label treatment (533/585, 91%) had HBV DNA <400 copies/mL at their last timepoint on TDF. Fifty-two patients (9%) qualified for genotypic analysis at their last visit on TDF monotherapy (Fig. 2A; Supporting Table 1A). Few patients (39/585, 6.7%) switched to FTC/TDF combination therapy during the first 5 years of open-label treatment; seven also qualified for genotypic analysis at their last visit on FTC/TDF (Fig. 2B; Supporting Table 1B).
Patients With HBV DNA ≥400 Copies/mL
The majority of patients who qualified for genotypic analysis at their last timepoint on TDF or FTC/TDF had HBV DNA ≥400 copies/mL in the absence of protocol-defined virologic breakthrough (TDF, 39/52 [75%]; FTC/TDF, 4/7 [57%]) (Fig. 2; Supporting Table 1). While patients with HBV DNA ≥400 copies/mL qualified for genotypic analysis throughout the open-label period, the majority qualified during year 2 (TDF, 30/39 [77%]; FTC/TDF 1/4 [25%]) during which time HBV DNA was still decreasing from baseline. Most of the patients who qualified for genotypic testing with HBV DNA ≥400 copies/mL in the absence of protocol-defined virologic breakthrough were adherent to study medication (TDF, 33/39 [85%]; FTC/TDF, 4/4 [100%]).
Genotypic analysis of the 39 patients on TDF monotherapy who had viremia in the absence of virologic breakthrough revealed that 21/39 patients (54%) had no changes in HBV pol/RT compared to baseline, consistent with what was observed at year 3 (Fig. 3A). Six of 39 patients (15%) were unable to be genotyped. Of the patients with HBV pol/RT changes, the majority (9/12, 75%) had unique polymorphic-site changes. Three patients had conserved-site changes: rtV173L+rtL180M+rtM204V (n = 1), rtL101L/F (n = 1), and rtA307A/T (n = 1). In all cases, the conserved-site changes were observed in the absence an increase in alanine aminotransferase (ALT). For the four patients on FTC/TDF, two (50%) had no changes in the HBV pol/RT compared to baseline, one had a polymorphic-site change (rtL91I/L) that was observed in another patient with virologic breakthrough (see below), and one patient had the conserved-site changes rtL180M±rtA181T±rtM204V in the absence of an increase in ALT (Fig. 3A). No new conserved-site changes were detected beyond year 3.
Patients With Virologic Breakthrough
Across open-label treatment, virologic breakthrough episodes occurred at the last visit on TDF (n = 13) or FTC/TDF (n = 3) (Fig. 2; Supporting Table 1). Approximately one-third of the virologic breakthrough episodes (6/16, 38%) were associated with ALT elevations (data not shown). The majority of patients with protocol-defined virologic breakthrough were found to be nonadherent to study medication at the time of testing, as determined by undetectable tenofovir levels in the plasma (TDF, 12/13[92%]; FTC/TDF, 0/3 [0%]). For the three patients with virologic breakthrough on FTC/TDF, HBV DNA at the time of breakthrough was >3 log10 copies/mL lower compared to baseline and continued to decline after breakthrough in 2/3 patients. ALT levels were transiently elevated in one patient and remained stable in the other two patients.
Genotypic analysis of the 13 patients on TDF monotherapy with virologic breakthrough revealed that eight had no sequence changes in pol/RT, three had unique polymorphic-site changes, and two could not be genotyped. For the three patients on FTC/TDF, one had no sequence changes in pol/RT, one had a unique polymorphic-site change, and one had a polymorphic-site change (rtL91I/L) that was observed in another patient with HBV DNA ≥400 copies/mL (see above) (Fig. 3B).
Phenotypic analysis of baseline and postbaseline isolates was attempted for the four patients with conserved-site changes detected at their last visit on TDF or FTC/TDF, as well as for one of the two patients harboring the polymorphic-site change rtL91I/L (Table 1). Two patients, one with LAM resistance mutations (rtL180M+rtM240V) and one with a LAM/ADV resistance mutation (rtA181T) detected postbaseline, were treatment-naïve at study entry; however, clonal analysis of baseline samples from each revealed the presence of all mutations at low levels prior to study entry (<10%; Supporting Fig. 1). Overall, there were no new conserved-site changes detected after the year 3 analysis; the rtL91I/L polymorphic-site change developed in one patient on FTC/TDF during year 2 and a second patient on FTC/TDF during year 6. This mutant from one patient, along with all other postbaseline pools and clones from patients with conserved-site changes, remained sensitive to tenofovir, with fold change values <2 compared to baseline.
Table 1. Phenotypic Results for Patients Harboring Conserved-Site Changes or Common Polymorphic Site Changes
Phenotypic analysis was also conducted with patient isolates from nine episodes of confirmed virologic breakthrough. All patient isolates remained sensitive to tenofovir, with fold change values <2.
Effect of the Switch to FTC/TDF on Outcomes
Overall, 528/585 (90%) patients had HBV DNA <400 copies/mL by week 72 and did not meet the criterion for receiving combination therapy. The remaining 10% of patients (n = 57) did meet the eligibility criterion for receiving combination FTC/TDF at or after week 72 (confirmed HBV DNA ≥400 copies/mL at two consecutive visits). Of these, 18/57 (32%) remained on TDF monotherapy; 39/57 (68%) switched to FTC/TDF. Most patients had HBV DNA <400 copies/mL at week 288 or the last visit on study regardless of whether they remained on TDF monotherapy (15/18, 83%) or switched to FTC/TDF combination therapy (30/39, 77%). For the 12 patients who remained viremic (11 of whom discontinued), the majority (10/12, 83%) were adherent to study medication with a median HBV DNA of 3.8 log10 copies/mL at the last visit in the study. Overall, 72% of patients eligible to receive FTC/TDF (41/57) remained in the study at week 288. For the 16 patients who discontinued, 11 (69%) discontinued between weeks 72 and 144, and the median HBV DNA at discontinuation was 3.5 log10 copies/mL. The elevated level of HBV DNA likely reflects that patients had a median of 40 weeks FTC/TDF or eligibility for FTC/TDF prior to discontinuing the study.
Further HBV DNA analyses were conducted to better understand why eligible patients did or did not switch to FTC/TDF combination therapy. The median HBV DNA values at baseline for patients who switched to FTC/TDF were >1 log10 copies/mL higher than those for patients who qualified for combination therapy but did not switch to FTC/TDF (9.67 versus 8.09 log10 copies/mL respectively; Fig. 4). This difference in HBV DNA was maintained through week 72, when the median HBV DNA for patients who did switch to FTC/TDF remained >1 log10 copies/mL higher (3.71 log10 copies/mL) compared with patients who remained on TDF monotherapy (2.31 log10 copies/mL). Higher HBV DNA levels may explain the investigator's decision to switch patients to FTC/TDF versus maintain TDF monotherapy.
Patients With Persistent Viremia
Because prolonged viral replication in the presence of antiviral therapy can lead to the selection of resistance, patients with persistent viremia (never achieved HBV DNA <400 copies/mL) were further analyzed. The number of patients with persistent viremia decreased steadily over time to none by week 240 (Fig. 5). In order to determine if low-level pol/RT resistance not detected by population sequencing was contributing to the persistence of HBV DNA in these patients, clonal analysis (∼20 clones per timepoint) was attempted on a subset of persistently viremic patients at week 144 (n = 4) and week 192 (n = 3). Two of the week 192 patients had viral loads too low to be genotyped, precluding clonal analysis. For the remaining four patients (one patient was tested at both week 144 and week 192), a total of 21 distinct conserved-site changes were observed in on-treatment clones compared to baseline clones. Twenty of the conserved-site changes were detected in one to two clones from one patient; however, one conserved-site change was observed in more than one patient (rtF183L, one clone each in two patients). Phenotypic analysis of a site-directed mutant containing rtF183L did not exhibit reduced susceptibility to tenofovir (fold change value <2).
Patients With Resistance Mutations for Other Nucleos(t)ide Analogs
Population-based sequencing detected lamivudine and/or adefovir resistance mutations (LAM-R/ADV-R) in 12 patients at the time they initiated TDF. The majority of these patients (9/12, 75%) were previously treated with LAM and/or ADV. Ten of these 12 patients remained in the study at week 288 and all had HBV DNA <400 copies/mL. Among the two patients who discontinued, one had LAM-R and discontinued prior to week 24 and one had LAM/ADV-R and was lost to follow-up by week 120. The patient who discontinued at week 120 maintained rtA181T throughout the study. Both patients achieved >4 log10 copies/mL decline in HBV DNA on TDF treatment, with the patient harboring rtA181T achieving complete suppression of HBV DNA prior to discontinuation.
Five additional patients had LAM-R and/or ADV-R detected while on TDF or FTC/TDF (Supporting Fig. 1). Although population sequencing did not detect LAM-R and/or ADV-R prior to TDF treatment, four patients were subjected to clonal analysis or allele-specific PCR, and in all cases LAM-R and/or ADV-R mutations were present at a low frequency (<10%) prior to TDF treatment.
A challenge in the long-term management of CHB is to maintain effective viral suppression without promoting the selection of treatment-resistant HBV mutations. In the present study, TDF monotherapy provided durable antiviral efficacy following up to 288 weeks of treatment, and comprehensive genotypic and phenotypic analyses detected no evidence of TDF resistance. Virologic breakthrough was rare, transient, and typically associated with nonadherence. Moreover, no patient remaining in the study at week 288 was persistently viremic (HBV DNA never <400 copies/mL), and patients who maintained persistent viremia through weeks 144 and 192 did not have detectable low level resistance.
The genotypic findings in the present report are similar to what was reported after 144 weeks of TDF treatment. There was no increase in the number of patients qualifying for genotypic analysis each year and the breakdown of sequence changes observed in pol/RT was similar despite an additional 144 weeks of TDF treatment. Overall, 91% of patients had complete suppression of HBV DNA at their last visit on TDF or FTC/TDF. Among patients qualifying for resistance analysis at their last visit on TDF, the majority (56%) had no genotypic changes and 80% of the patients with sequence changes had unique polymorphic-site changes attributable to natural sequence variation. Phenotypic analysis revealed that none of the pol/RT amino acid substitutions detected were associated with resistance to TDF. Additionally, there was no increase in the incidence of virologic breakthrough with longer treatment duration. The majority of patients with virologic breakthrough were found to be nonadherent to study medication. This result is consistent with a recent study of virologic breakthrough in CHB patients receiving treatment in clinical practice that showed nonadherence is often associated with the observation of virologic breakthrough and that breakthrough is also seen in the absence of detectable resistance mutations. In our study, the episodes of virologic breakthrough were often transient upon resumption of therapy.
All patients who remained in the study at week 288 had achieved HBV DNA <400 copies/mL during the open-label treatment period; the number of persistently viremic patients decreased over time. It is important to note that the decline in persistent viremia was not driven by the early discontinuation of TDF nonresponders: 83% of patients who discontinued had HBV DNA <400 copies/mL at their last visit in the study. This suggests that the longer a patient remains on TDF therapy, the greater the likelihood of achieving complete viral suppression. A separate subanalysis focused on patients with high baseline viral load (≥9 log10 copies/mL) observed similar results. Importantly, only a minority of patients (9/585, 2%) were persistently viremic at year 3 and beyond, and none of these patients who were adherent to study medication experienced virologic breakthrough or had detectable resistance to TDF through 288 weeks of treatment.
Overall, the results of this long-term study of TDF, demonstrating no detectable TDF resistance through 288 weeks of treatment, compare very favorably with the results for other treatments for CHB. The selection of virologic resistance during treatment with ADV and/or LAM is well documented.[6-8] ETV is the only CHB therapy with a rate of resistance (1.2%) comparable to that of TDF in treatment-naïve patients. Cumulative resistance to ETV remains low among patients with genotypic resistance to ADV (4% in patients treated for a median of 19 months), but is high in patients with LAM-R (51% after 5 years) and both ADV-R and LAM-R (73% after 3 years).[11, 26] In contrast, the majority of patients in our study with LAM-R or ADV-R mutations maintained HBV DNA <400 copies/mL through 288 weeks of TDF treatment. Several studies have demonstrated the long-term efficacy of TDF in LAM- and/or ADV-R populations[27, 28] (Berg et al., submitted), although other studies have observed an attenuated response in ADV-R populations.
One of the strengths of this study is that it followed a large cohort of well-characterized patients undergoing up to 288 weeks of continuous therapy with TDF who were subjected to rigorous genotypic and phenotypic analyses. One limitation is that patients with HBV DNA ≥400 copies/mL at or after week 72 could add FTC to their treatment regimen (switch to fixed-combination FTC/TDF); however, only 39 patients (out of 466 who completed week 288) added FTC. There was no notable difference in the virologic response or genotypic findings between patients on FTC/TDF combination therapy and those on TDF monotherapy, and adding FTC had no discernible effect on the viral decay kinetics or genotypic profiles. Moreover, the virologic response among patients who qualified to add FTC but who remained on TDF monotherapy was comparable to those who did switch to FTC/TDF, although subjects who remained on TDF monotherapy had >1 log10 copies/mL lower HBV DNA levels at week 72 compared to those who added FTC. This suggests that the option to switch to FTC/TDF in these studies was unlikely to impact the overall long-term results and lack of TDF resistance.
In conclusion, the present results demonstrate that TDF monotherapy maintained effective viral suppression over up to 288 weeks of continuous therapy without the selection of TDF resistance. This study is ongoing and resistance surveillance will continue through 384 weeks of continuous treatment.
This study was sponsored by Gilead Sciences, Inc. The study was conducted in compliance with all regulatory obligations and the Institutional Review Board and informed consent regulations at each investigational site. In collaboration with the investigators, Gilead Sciences, Inc. designed the study, conducted statistical analyses, and interpreted the data. The authors had unrestricted access to the primary data and were not limited by Gilead Sciences, Inc. in the writing of this article. The corresponding author takes full responsibility for the authenticity of the data and statistical analysis. The authors thank Brandi Chappell, Maria Curtis, Florence Myrick, James Schawalder, and Yuao Zhu for excellent technical assistance. The authors thank Amy Lindsay, PhD, and Mariana Ovnic, PhD, of Percolation Communications LLC for providing editorial assistance during article development. Gilead Sciences, Inc. provided financial support for article development.