HCV infection on-treatment viral kinetics: Do they still have a role?†
Potential conflict of interest: Nothing to report.
Watch the interview with the authors
Answer questions and earn CME
As soon as it was recognized that interferon-based therapy could lead to long-term viral eradication or a sustained virological response (SVR), the search was on for predictors of responses to therapy. With low response rates and the need for prolonged, costly, and difficult therapy, the primary goal was to identify patients who were unlikely to respond to treatment so that fruitless therapy could be avoided. A number of pretreatment patient and viral factors were identified that were associated with treatment outcomes.1 However, none had adequate predictive ability for withholding therapy or guaranteeing a response. A careful analysis of the pivotal registration trials of peginterferon showed that early treatment responses were very predictive of ultimate treatment outcomes.2,3 Patients who failed to achieve a 2-log decline in hepatitis C virus (HCV) RNA levels with 12 weeks of therapy or an early virological response (EVR) had effectively no chance of responding to therapy, and this led to the adoption of early stopping rules.2
Since the identification of EVR as a useful stopping rule, the kinetics of viral decline during therapy have been carefully evaluated and have proven useful in a number of different areas of HCV treatment. Beyond the prediction of treatment failure, early viral kinetics have been useful for predicting favorable responses,4 for developing mathematical models to shed light on the mechanisms of action of antiviral agents,5 and now, in the era of direct-acting antivirals (DAAs), for determining the treatment duration.6,7
Predictive Ability of EVR
On-treatment responses have been proven to be much better predictors of ultimate treatment outcomes than any pretreatment factors.1 On-treatment kinetics predict a response because they are in fact components of the response itself; it is perhaps not surprising that an early response predicts an ultimate response. The on-treatment response is influenced by all of the pretreatment factors, but it also takes into account on-treatment factors such as compliance and dose reductions; this makes it the best global assessment of treatment responsiveness. Hence, it is not surprising that patients who have undetectable HCV RNA levels after 4 weeks of therapy or a rapid virological response (RVR) are very likely to achieve SVR.8,9 By achieving RVR in the first place, these patients have clearly defined themselves as treatment-responsive. The pretreatment factors still matter, but RVR, EVR, or any other on-treatment parameter will likely trump all pretreatment predictors.
Patients with unfavorable pretreatment characteristics are much less likely to achieve RVR, but those who are able to suppress the virus rapidly despite their pretreatment profile are more likely to achieve SVR than patients with favorable baseline characteristics who suppress the virus slowly.10
The importance of RVR is perhaps best illustrated when it is compared with the interleukin-28B (IL-28B) genotype. The presence of single-nucleotide polymorphisms near the IL-28B gene is the strongest pretreatment predictor of a response.11 However, in comparison with the IL-28B genotype, RVR is more predictive of SVR. Mangia et al.10 recently reported that although patients with the unfavorable IL-28B genotype are much less likely to achieve SVR, after 4 weeks of therapy, patients with an unfavorable IL-28B genotype who have undetectable HCV RNA levels are more likely to achieve SVR than patients with a favorable IL-28B genotype who fail to achieve RVR. The achievement of RVR is largely dictated by the IL-28B genotype, but RVR also takes into the account all other pretreatment and on-treatment factors that affect the response, such as ethnicity, body mass index, and compliance. Therefore, RVR should not be included with pretreatment factors in predictive models of treatment outcomes because RVR is itself part of the response and thus always emerges as the strongest predictor of SVR. The question to be asked with pretreatment factors is why some patients achieve RVR or EVR while others do not.
EVR has proved to be a useful predictor of nonresponse with a high negative predictive value, and this makes it useful for defining stopping rules.2 On the other hand, RVR is a strong predictor of ultimately achieving SVR and has a very high positive predictive value but a poor negative predictive value.8 RVR has, therefore, been used to tailor the treatment duration. Patients with a genotype 1 infection who achieve RVR are very likely to achieve SVR. Although there still may be a small benefit from the addition of a DAA, it is hard to improve on >80% rates of SVR.9 European guidelines recommend shortening the duration of therapy to only 6 months for patients who achieve RVR, so the benefit of adding a protease inhibitor becomes even less clear if this approach is followed.12 Whether such patients could truncate their therapy even further with the addition of a DAA will undoubtedly be evaluated in the future.
For patients with genotype 2 or 3, the achievement of RVR is a very useful prognostic factor. Those who achieve RVR have high rates of SVR even if the duration of their therapy is shortened to 12 to 16 weeks.4,13 Although a full 24-week course is superior to a shortened course of therapy even in patients who achieve RVR,4 the benefit is relatively small, and the risk of relapse may be partially mitigated by the use of weight-based ribavirin dosing if a shortened course is to be considered. For patients who do not achieve RVR (particularly those with a genotype 3 infection), there may be merit in continuing therapy beyond 24 weeks. Mangia et al.14 found a marginal but nonsignificant benefit from extending therapy to 36 weeks for non-RVR patients with genotype 3. It is hoped that ongoing studies randomizing non-RVR patients to 24 or 48 weeks of therapy will clarify this issue.
Beyond its prognostic utility, EVR has been very useful for understanding the mechanisms of action of antiviral agents. By carefully evaluating the decline in HCV RNA levels during interferon therapy, Neumann et al.5 showed that interferon led to a biphasic reduction in viremia. Fitting mathematical models to experimental data, they were able to conclude that the first phase of the viral load decline was likely due to the clearance of free virions and the inhibition of viral production, whereas the slower second phase was related to the clearance of infected cells. These models have been extended to include ribavirin.15 Despite its use for more than 10 years, the way in which ribavirin improves treatment outcomes for HCV patients remains a mystery.16 Careful modeling studies have been able to largely exclude some of the proposed hypotheses for ribavirin's mechanism of action. The data support a role for ribavirin in affecting the infectivity of the virus. It may function as a mutagen and lead to a less infectious virus15; alternatively, ribavirin may stimulate a weak antiviral response in cells that makes them less infectable.17 The insights gleaned from these models have been instrumental to our understanding of how interferon and ribavirin act.
More recently, modeling approaches have been applied to novel DAAs with some interesting results. DAAs were developed to target specific steps in the HCV lifecycle, so it would seem that their mechanisms of action should be fairly well understood. However, the inhibition of a single HCV enzyme may affect more than one stage in the lifecycle, and this may mean that the benefits of one class of DAAs are different from the benefits of another. For example, recent work by Dahari et al.18 shows that a nonstructural protein 5A inhibitor likely affects both viral replication and viral assembly. Models incorporating a dual mechanism of action fit the data most accurately.
Modeling is only as good as the experimental data to which the models are applied. Only with frequent sampling of relatively large numbers of patients can models be robustly tested. As DAA therapy continues to become more potent, the frequency of early sampling will become more important. With interferon, weekly sampling is more than adequate because most patients do not suppress the virus before 4 weeks at the earliest. However, with potent DAA combination therapy, many patients will have undetectable levels of HCV RNA with 2 weeks of treatment or even earlier.
Modeling not only is important for determining the mechanism of action of drugs but also may guide the appropriate duration of therapy.6,7 Ultimately, the goal of treatment is to reduce viral levels to less than 1 virion per body. Unfortunately, the available assays are limited to the level of detection of virus in the blood. The extrapolation of the slopes of the decline of HCV RNA levels in the blood can be used to estimate the necessary duration of therapy.19 However, teasing apart differences in the slopes of very potent agents may be difficult, but it is potentially still very important. Slight changes in the initial decline may indicate the need for a longer duration of therapy; however, these types of inferences are possible only with very frequent sampling during the initial therapy. Ideally, just as EVR and RVR are appropriate on-treatment evaluations for determining the likelihood of response and the necessary duration of interferon therapy, responses at earlier time points may very well guide DAA therapy.
Response-Guided Therapy (RGT) With DAAs
The assessment of on-treatment viral responses was initially relatively crude with only 12- and 24-week evaluations. The approval of telaprevir and boceprevir has formalized the concept of RGT.7,20 With both agents, for patients who clear viremia rapidly, the total duration of therapy can be shortened with no reduction in SVR. The RGT approach will become the standard for all future DAAs until a combination therapy is developed that is potent enough to allow for a very short period of therapy for all patients.
The result of RGT is the need for vigilance about the timing and interpretation of HCV RNA testing. Delaying testing for a few days or eventually even a few hours may affect the interpretation of the results. Because of the potency of DAAs, testing 24 hours later than planned may be the difference between detectable and undetectable HCV RNA levels, and this may alter decisions about the treatment duration. With both approved protease inhibitors, shortening the duration of therapy with RGT requires that HCV RNA be undetectable at week 4 of protease inhibitor therapy and remain so to the end of the treatment with the DAA.7,20 A percentage of patients treated with both agents suppressed the virus initially but did not qualify for shortened therapy with subsequent testing. Because of the risk of the emergence of DAA resistance, it is critical to follow stopping rules appropriately. Even very low levels of detectable virus may herald the emergence of resistance, particularly because the viral level may be rising from a previously missed nadir. Whether the timing of testing and the appropriate interpretation of results will be accurately performed in clinical practice with these agents remains to be seen. It is hoped that when DAAs are used in combination, the risk of breakthrough resistance will become much lower, and hence the need for testing to ensure the maintenance of undetectability will be less critical. Ideally, frequent testing may be performed early to determine the length of therapy, but once patients have suppressed the virus, testing will no longer be needed except to confirm SVR.
On-treatment viral responses have defined the prognosis and duration of peginterferon and ribavirin therapy and have shed important light on the ways in which these agents lead to viral clearance. In the era of DAAs, on-treatment viral responses will become more important in the short term to define RGT. Ultimately, treatment will continue to improve, and it is hoped that it will become so effective that RVR, EVR, and the growing compendium of acronyms in HCV therapy will be relegated to history books.