Predictors of Response to Chronic Hepatitis C Treatment

Future Virology

Predictors of Response to Chronic Hepatitis C Treatment

Ezequiel Ridruejo

Future Virology. 2012;7(11):1089-1101.

Abstract
Chronic hepatitis C is a growing health problem worldwide that has attracted increased attention in recent years. Treatment with peginterferon and ribavirin combination had previously been the standard of care. In 2011, a new treatment with protease inhibitors, telaprevir and boceprevir was approved, and a new standard of care was defined. Previous predictors of response have been redefined, and while IL28B, fibrosis stage and hepatitis C virus viral load testing continue to have their value, viral kinetics during treatment defining viral response have emerged as the strongest predictor for achieving sustained virologic response with the new treatment. New therapies are expected in the near future, and current treatment predictors of response may soon change.

Introduction
Chronic hepatitis C is a growing health problem worldwide that has attracted increased attention in recent years. Approximately 3% of the world's population (130–170 million people) is chronically infected with HCV. ^[1,2] HCV is one of the most prevalent blood-borne infections, with a higher prevalence than HIV (~1.1 million infected) and HBV (0.8–1.4 million infected) in western countries. ^[3] Chronic HCV infection progresses asymptomatically, and almost 75% of patients are unaware of the diagnosis when they present with complications of cirrhosis, portal hypertension or develop a hepatocellular carcinoma (HCC) many years later. ^[4] In western countries HCV infection is one of the most frequent causes of death from end-stage liver diseases and HCC. ^[5] In the last decade, HCV-related morbidity doubled, and HCC related to HCV increased almost threefold. ^[6] Its impact on liver-related morbidity and mortality is expected to reach its peak in the next decade. ^[7]

The objective with antiviral therapy is to obtain a sustained virologic response (SVR), defined as undetectable HCV RNA 24 weeks after the end of treatment. In the last decade, treatment with pegylated interferon 2a or 2b (pegIFN) combined with weight-based ribavirin (RBV) for 48 weeks (genotypes 1, 4, 5 and 6) or 24 weeks (genotypes 2 and 3), has been considered the standard of care (SOC) for HCV treatment. With pegIFN/RBV treatment, SVR rates were 40–50% in genotype 1 and 70–80% in genotypes 2/3-infected patients in western countries. ^[8] Some studies showed that prolonging treatment up to 72 weeks in genotype 1 slow responders (HCV RNA detectable at week 12 but undetectable at week 24), can increase SVR rates. ^[9–12] But, prolonged treatment is associated with higher costs, increases in adverse events, and cannot be used in treatment-intolerant patients.
Many viral and host factors affect treatment response, and not achieving a SVR might be related to a combination of them. HCV genotype and IL28B host genotype are the strongest predictors of pegIFN/RBV therapy outcome. ^[13]IL28B designates single-nucleotide polymorphisms (SNPs) in the interferon λ gene region in chromosome 19, and it is related to interferon responsiveness. Other factors such as high viral load, older age, black race and advanced fibrosis or cirrhosis negatively influence SVR rates. ^[14,15]

In 2011 the US FDA and the EMA approved telaprevir (TVR) (Incivek/Incivo™, Vertex) and boceprevir (BOC) (Vitrelis™, Merck) for its use in HCV genotype 1 infected patients. These two NS3/4A serine protease inhibitors (PIs) are the first generation of direct-acting antiviral (DAA) drugs to be approved for its use in clinical practice. Used in combination with pegIFN/RBV, PIs increased SVR rates up to 68–75% in naive patients and to 41–52% in previous nonresponders. ^[16–19] With the new treatment regime of PI-based triple therapy, some of the predictors of response to dual pegIFN/RBV-based therapy are less important. In fact, some host factors are used differently. Nonetheless, baseline host and viral factors and early viral kinetics are still important determinants for patient counseling and management using BOC or TVR combination treatment.

Why Do We Need Predictors of Response to Treatment?
The factors that determine the likelihood of achieving SVR are called predictors of response. They can be classified as viral- or host-related, or as pre- or on-treatment depending on the time point of evaluation.
       
Treatment with pegIFN/RBV dual therapy is costly and is associated with side effects. The triple therapy regimen including BOC or TVR increases the costs and the complexity of treatment further, and more frequent and severe adverse events are expected with these newer therapies. Before initiating treatment, it is useful for patients and physicians to determine the likelihood of achieving a SVR, so that they can decide whether treatment benefits outweigh its costs and risks.
In the pegIFN/RBV dual therapy era, predictors of response helped the patient and the physician to decide whether or not to start treatment. There were no other treatment options, and antivirals were just a promise. Today, in naive patients, it is possible to predict response to dual or triple therapy and choose between them. There are a small percentage of patients, with good predictors of response, who are still candidates for pegIFN/RBV dual therapy. Also, naive patients with poor predictors of response with mild liver disease may be candidates for awaiting the new generation of antivirals. In the future, these regimens might be more effective, less toxic, and perhaps pegIFN-free.
       
Before the approval of DAA, patients who had failed a previous course of pegIFN/RBV had few treatment options. These included consensus interferon or retreatment with pegIFN/RBV. Previous relapsers and null responders can achieve 24–50% and 4–14% SVR rates, respectively. ^[20–23] Even though PIs increase SVR rates to 41–52%, there is still a percentage of nonresponders to the first generation of PIs. ^[18,19] In this subset of patients it is very important to try to predict which patients will benefit with this treatment, and which patients are candidates to wait for treatment with the next generation of DAAs.
       
Many interesting viral and host predictors of response have been described. We will discuss those that are widely available to medical practitioners and suitable for being used in routine clinical practice.

Pretreatment Predictors of Response
Predictors of response to pegIFN/RBV dual therapy have been described, and are shown in Box 1. Some host predictors are fixed (e.g., age) but others can be modified (e.g., body weight). Data about predictors of response to triple therapy are beginning to be demonstrated. Because interferon responsiveness continues to be important with the newer regimes, the same factors may influence treatment response to PI-based therapy. Viral genotype and baseline viral load still influence PI treatment response, but mutations in certain regions of the viral genome appear as new treatment predictors. Due to different rates of resistance, SVR rates to PIs may be different between subtypes a and b in genotype 1 (G1) infected patients. ^[24]

Viral Factors

HCV Genotype.
HCV genotype is the strongest baseline predictor for response for dual therapy: the highest SVR rates have been achieved in G2 (~80%) and the lowest in G1 (~40–50%)-infected patients. ^[8] Other smaller studies suggested that G4 is poorly responsive; ^[25] that G5 is similar to G3; ^[26] that G3 is less responsive than G2; ^[27] and that, similarly to G2, G6 is very responsive. ^[25] The first generation of PIs has been approved for G1 treatment, but these drugs are ineffective in most non-G1-infected patients. A second-generation of NS3/4A PIs, nucleos(t)ide and nonnucleoside inhibitors of the NS5B RNA polymerase and NS5A complex inhibitors, have now reached Phase II and even Phase III clinical stage development for treatment across all HCV genotypes. ^[28] HCV genotype value as a predictor of SVR in this scenario will need to be redefined.
       
HCV G1 Subtype. G1 has two subtypes, 1a and 1b. Subtype 1b has a higher prevalence in Europe and 1a in the USA. Patients infected with G1 subtype b achieved slightly higher SVR rates with triple therapy (~7–10% difference), than those infected with subtype a. Given that subtype 1a has a lower barrier to resistance to PIs, resistance-associated variants (RAVs) are more frequent with this subtype than with viral subtype 1b. ^[29] In cases of treatment failure, 1a-associated RAVs appear earlier, 1b-associated RAVs disappear faster, and the double V36M + R155K mutation needed for RAV appearance occurs almost exclusively in subtype 1a. ^[30]

Baseline HCV Viral Load.
HCV RNA levels before initiating treatment predict the likelihood of obtaining SVR with pegIFN/RBV dual therapy. ^[27,31] This measure has a lower predictive value for the more IFN-sensitive genotypes associated with higher SVR rates, such as G2 and 3 compared with G1. HCV G3-infected patients with <400,000, 400,000–800,000 and >800,000 IU/ml baseline viral loads, achieved SVR rates of 81, 70, and 59%, respectively, when treated with pegIFN alfa-2a and RBV 800 mg per day for 24 weeks. In G2 patients, SVR rates for the same HCV RNA levels were 82, 79, and 73%, respectively. ^[32]

On the contrary, in G1 patients low baseline HCV RNA level (<600,000–800,000 IU/ml or less) was shown to be an independent predictor of achieving SVR. ^[32,33] Based on these results, an abbreviated regimen (24-week treatment with pegIFN/RBV) may be indicated for patients with G1 and a low versus high baseline viral load, and undetectable HCV RNA after 4 weeks of treatment. There is no current agreement on the most discriminatory HCV RNA level, which ranges between 400,000 and 800,000 IU/ml (5.6–5.9 log ₁₀ IU/ml) ( Table 1). ^[34]

Baseline HCV RNA level is less predictive of SVR when using PIs in G1 patients. In the SPRINT-2 study, patients treated with pegIFN/RBV/BOC with <800,000 IU/ml baseline HCV RNA levels achieved 76–85% SVR rates, while for patients with ≥800,000 IU/ml, SVR rates were 61–63%. SVR rates in the SOC control arm were 64 and 33% for ≥800,000 and <800,000 IU/ml, respectively. ^[16] Patients with baseline viral load ≤400,000 (vs >400,000 IU/ml) had a higher SVR rate, with odds ratio (OR): 3.9 (95% CI 2.1–7.1), p < 0.001. ^[16] In the ADVANCE study, patients treated with pegIFN/RBV/TVR achieved high SVR rates irrespective of their baseline viral load: 78% with <800,000 IU/ml and 74% with ≥800,000 IU/ml. The SVR rates in the SOC control arm were 70 and 36%, respectively ( Table 2 & Table 3. ^[7] Both studies showed that PI-treated patients with baseline lower viral loads tend to have a slightly higher likelihood of achieving SVR, but the differences in SVR rates between high and low HCV RNA levels in these patients is less marked than in pegIFN/RBV-treated patients. ^[30]

Host Factors
Age. Univariate and multivariate analyses performed in most of the randomized control trials in patients treated with pegIFN/RBV dual therapy showed that younger age significantly correlated with the likelihood of obtaining SVR. Furthermore, higher SVR rates were obtained in patients younger than 40–45 years old. ^[32] In Phase III clinical trials using PIs, younger patients achieved slightly higher SVR rates. Patients younger than 40 years old treated with BOC achieved higher SVR rate than those treated with SOC: OR: 1.5 (95% CI 1.0–2.1), p = 0.03. ^[16] Patients younger than 45 years old treated with TVR had a SVR rate of 83% versus 70% in those older than 45; in the SOC arm, SVR rates were 52 and 38%, respectively ( Table 2). ^[17]

Sex. Female patients had been shown to achieve higher SVR rates than males in two studies using the old combination of standard IFN and RBV (p < 0.004). However, in both of the pegIFN/RBV registration trials, no statistically significant correlation was found between sex and SVR. ^[32] According to data in the Phase III clinical trials, sex has no impact on SVR response to PIs. ^[16,17] SVR rates were 74% in males and 75% in females in the TVR arm, and 45 and 43%, respectively, in the SOC arm ( Table 2). ^[17]

Race.
African–Americans had lower SVR rates than non-African–Americans patients when treated with pegIFN/RBV therapy. In randomized control trials SVR rates were 19–28% in African–Americans versus 39–52% in non-African–Americans. These lower SVR rates were observed in genotype 1-infected patients treated with pegIFN/RBV for 48 weeks. There was no difference in SVR rates in patients infected with other genotypes. ^[32] Although not as extensively studied, Latinos (Hispanic) patients treated with pegIFN/RBV also showed lower SVR rates than Caucasians. However, another study including Latino patients of European ancestry showed similar SVR rates to those observed in Caucasians. ^[35] Finally, Asian HCV-infected patients tended to achieve higher SVR rates than Caucasians. ^[32] How ethnicity influences HCV treatment outcome is unknown, but race is strongly correlated with differences in favorable IL28B allele frequencies. Host genetics may be the explanation.
       
In black patients, PI-based therapy improves SVR rates when compared with pegIFN/RBV. However, SVR rates in these patients are still lower than in nonblack patients. IFN-responsive black patients achieving early virologic responses showed higher SVR rates. Among nonblack patients, the rate of a SVR was 40% with SOC and was significantly higher with BOC: 67 and 68% (p < 0.001). Among black patients, the SVR rate was 23% in SOC, and 42 and 53% in BOC group (p = 0.004, vs SOC). Black patients treated with BOC showed significantly lower SVR rates than nonblack patients: OR: 0.5 (95% CI 0.3–0.7), p < 0.001. ^[16] Among black patients treated with TVR, SVR rate was 62 and 58%, as compared with 25% in the pegIFN/RBV group. In nonblack patients, TVR-treated patients achieved 75% SVR rates, while in the pegIFN/RBV arm it was 46%. ^[17] Even though Hispanic patients represent a small percentage of the treated population (~10%), they obtained a similar response rate to Caucasians: 74% in the TVR arm and 35% in the pegIFN/RBV arm ( Table 2). ^[17] There is no information on SVR rates in Asians in the Phase III registration trials. In a study from Japan, SVR rates in response to TVR for 12 weeks combined with pegIFN/RBV for 24 weeks were higher when compared with SOC (73.0 vs 49.2%, respectively, p = 0.002), and similar to SVR rates in nonblacks in Phase III trials. ^[36]

Fibrosis.
In patients with chronic HCV infection advanced fibrosis, especially cirrhosis, is associated with a diminished treatment response. Multilogistic regression analyses in pegIFN/RBV randomized controlled trials did not show a significant association between cirrhosis and lower SVR rates. However, lower SVR rates can be demonstrated when they were directly compared between cirrhotic (44% for pegIFN α2b and 43% for pegIFN α2a) and noncirrhotic patients (57% for pegIFN α2b and 58% for pegIFN α2a). ^[32] One possible explanation for the lack of correlation between treatment outcome and cirrhosis in the multivariate analysis is the small percentage of cirrhotic patients in these studies. The mechanisms of reduced IFN responsiveness in cirrhotic patients is unknown, and is possibly multifactorial.
       
Patients with advanced fibrosis or cirrhosis treated with PI triple therapy also showed diminished SVR rates, possibly related to reduced IFN responsiveness in patients with advanced liver disease. However, cirrhotic patients treated with PI-based triple therapy have higher SVR rates than those treated with pegIFN/RBV alone. On the contrary, patients treated with BOC or TVR-based therapy have very high SVR rates if they have mild liver disease (F0–F2 stages of fibrosis).
SVR rates in patients with advanced fibrosis were lower than in those with mild fibrosis, although the numbers of patients with a Metavir fibrosis score of 3 or 4 were small (7–9%). SVR rates according to fibrosis stage were: F0–2: 67% in BOC versus 38% in SOC group (p < 0.001); and F3–4: 41–52% in BOC arm versus 38% in SOC arm (p > 0.05). Absence of cirrhosis was a good predictor for obtaining a SVR in the SPRINT-2 trial: OR: 2.5 (95% CI 1.4–4.6) p = 0.003. ^[16] In the ADVANCE trial the numbers of patients with advance fibrosis were also small: 14–16% F3 and 6–7% F4 in the TVR arm. In this subgroup of patients SVR rates were 62% in the TVR arm and 33% in the SOC arm; in F0–2 patients, SVR rates were 81% in the TVR arm and 46% in the SOC arm ( Table 2 & Table 3. ^[17]

IL28B Genotype.
Genome-wide association studies have identified two single SNPs near the IL28B gene or lambda interferon 3 in chromosome 19. These SNPs are strongly associated with HCV clearance, whether spontaneous or treatment-related. ^[37–39] Given that IL28B genotype strongly influences dual pegIFN/RBV treatment outcome, its predictive value is more relevant in the more difficult-to-treat HCV genotypes: 1 and 4.
       
Caucasian G1 patients with the rs12979860 CC, CT, and TT genotypes treated with pegIFN/RBV dual therapy achieved 69, 33, and 27% SVR rates, respectively; SVR rates for black patients were 48, 15, and 13%, respectively. ^[40] Caucasians and Asian patients with the rs8099917 TT, GT and GG genotypes treated with pegIFN/RBV, showed similar SVR rates. ^[38] Similar SVR rates with both genotypes had been reported in Hispanic patients with European ancestry. ^[41] Based on these results, IL28B CC genotype has been considered the strongest baseline predictor of SVR (OR: 5.2 vs non-CC genotype; p <0.0001) ( Table 1). ^[40] The mechanisms by which IL28B influences pegIFN/RBV treatment outcome, are unknown.
       
However, IL28B predictive value is diminished in patients treated with PIs. In all of the Phase III trials, data on IL28B genotype was retrospectively obtained in a proportion of patients. Naive patients with unfavorable IL28B genotypes treated with TVR or BOC in the ADVANCE and SPRINT-2 trials achieved higher SVR rates than those treated with SOC. ^[42–44] Incremental benefits of BOC or TVR-based therapy were less remarkable in patients with the favorable CC genotype. However, it allowed the shortening of therapy in most patients.
       
A subanalysis from the ADVANCE trial showed that treatment with TVR combined with pegIFN/RBV increased SVR rates across all IL28B genotypes, with the largest increases shown in patients with the CT or TT genotype. SVR rates among patients with the CC genotype were 90% with TVR 12 weeks plus pegIFN/RBV, 84% with TVR 8 weeks plus pegIFN/RBV, and 64% with pegIFN/RBV alone. For the same treatment groups, SVR rates were 71, 57, and 25%, respectively, in CT genotype patients; and 73, 59 and 23%, respectively, for TT genotype patients. In this trial, information about IL28B status was obtained in 42% of the patients, all of whom were white. ^[42]
Data from the SPRINT-2 study, also limited by the relatively small numbers of patients included (62% of the population), showed that the benefit of BOC in patients with the CC genotype seems less marked. In this subgroup, SVR rates were 82% with BOC and 78% in the pegIFN/RBV control arm. As in the TVR trial, patients with unfavorable IL28B genotypes achieved significantly higher SVR rates when treated with BOC-based triple therapy versus pegIFN/RBV alone. SVR rates were 65% in the BOC-based response-guided therapy (RGT) arm, 71% in the BOC for 48 weeks arm, and 28% in the pegIFN/RBV arm for patients with the CT genotype. For the same treatment arms SVR rates were 55, 59, and 27%, respectively in TT genotype patients. ^[43,44] In treatment-naive patients treated with BOC-based therapy, IL28B CC versus non-CC genotype predicted SVR in a multiple stepwise logistic regression model: OR 4.5 (p < 0.001) ( Table 3). ^[43]

In the same subanalysis of the SPRINT-2 trial, the predictive value of the IL28B rs8099917 SNP was evaluated. A crossanalysis of rs12979860 and rs8099917 SNPs revealed that most CC patients at the rs12979860 locus had the corresponding favorable TT at the rs8099917 locus. However, the converse was not true because TT patients at rs8099917 were just as likely to be CC versus non-CC. SVR rates were 78% in TT, 67% in GT and 50% in GG rs8099917 BOC-treated patients. Combining the two SNPs did not offer enhanced predictability of SVR, and although both rs12979860 and rs8099917 have utility in predicting SVR, it appears that the rs12979860 locus is more predictive when considering all treatment groups, including the control groups. ^[44]

SVR rates in the CC genotype patients treated in the pegIFN/RBV control arms differs between the two PI trials. Patients treated in the ADVANCE trial achieved a 64% SVR rate, while in the SPRINT-2 trial this was 78%. SVR rates in the ADVANCE trial are lower than those previously reported, and this may be related to the small sample size. ^[37] It seems that BOC and TVR-based treatment slightly improves SVR rates in the favorable IL28B genotype, but significantly improves it in CT and TT genotypes. One major advantage detected in both studies was that most CC genotype naive patients were able to reduce treatment duration: 78% CC versus 57% of CT versus 45% of TT received shortened treatment with TVR, ^[42] and 89% CC versus 53% of CT versus 45% of TT patients received shortened treatment with BOC. ^[43,44]

Recent data add information on short courses of treatment in favorable IL28B genotypes. A retrospective analysis of the PROVE 2 trial showed that IL28B CC naive patients treated for 12 weeks with TVR/pegIFN/RBV achieved SVR in 100% of the treated patients (14 out of 14). ^[45] In IL28B CC patients treated with TVR for 12 weeks combined with pegIFN/RBV for 24 weeks, and in patients treated with SOC, SVR rates were 94 and 64%, respectively. SVR rates were lower in non-CC genotype patients. ^[45] Shortened therapy is of great benefit, even if SVR rates remain close to those obtained with pegIFN/RBV alone. Currently, there are ongoing studies in IL28B CC patients treated with reduced duration PI-based regimes. ^[101]

The predictive value of IL28B genotype with PI therapy is limited for treatment-experienced patients. In retrospective analyses of the REALIZE ^[46] and RESPOND-2 ^[43,44] trials, patients' previous treatment response appears to have a better predictive value than the IL28B genotype. SVR rates decrease in a stepwise mode from relapsers to partial responders to null responders in the REALIZE trial. ^[46] Also, SVR rates were lower in partial responders than in relapsers in the RESPOND-2 trial (this trial did not include null responders). CC patients receiving TVR-based therapy achieved higher SVR rates (79%) than non-CC patients (60% in CT and 61% in TT). However there was no SVR rate difference according to IL28B genotype according to previous treatment response: in previous relapsers, SVR rates were 88% in CC, and 85% in CT and TT patients; in previous partial responders they were 63% in CC, 58% in CT and 71% in TT; and they were 40% in CC, 29% in CT and 11% in TT previous null responders treated with TVR-based therapy in the REALIZE trial. ^[46]

Response to pegIFN based lead-in phase (LI) was the strongest predictor of treatment response (OR 2.2, p = 0.025) in the RESPOND-2 trial, even stronger than IL28B. Among previous partial responders or relapsers, CC and CT genotype patients treated with RGT or fixed-duration BOC therapy achieved higher SVR rates than those treated with pegIFN/RBV. Although the number of patients included in this subanalysis was small, TT genotype patients treated with fixed-duration BOC achieved higher SVR rates than those treated with either RGT BOC or pegIFN/RBV alone. ^[43] As demonstrated in naive patients, IL28B genotype predicted the possibility of shortening therapy in previous nonresponders: 79% CC versus 46% CT versus 63% TT patients received shortened treatment with BOC. ^[43,44]

Although IL28B genotype has major role in predicting the initial response to pegIFN treatment, there are other factors influencing SVR. Previous response to pegIFN is the strongest predictor of PI-based therapy SVR in treatment-experienced patients. Even though IL28B genotyping has limited value in this subgroup of patients, it may add information about patients' possibilities of achieving SVR and about the possibility of reducing treatment duration. Once the lead-in response is known, reflecting IFN responsiveness, IL28B value as a predictor of response becomes less important.

Previous Response to pegIFN/RBV.
As previously mentioned, nonresponders to pegIFN/RBV are only candidates for PI base therapy, since they increase SVR rates to 41–52%,. ^[18,19] But there are still a percentage of patients nonresponsive to the first generation of PIs, and one of the strongest predictors of SVR is the type of previous response to pegIFN/RBV. Three categories have been defined for these patients. Null responders are patients whose HCV RNA level did not decline by at least 2 log IU/ml at treatment week 12; partial responders are patients whose HCV RNA level dropped by at least 2 log IU/ml at treatment week 12, but in whom HCV RNA was still detectable at treatment week 24; and relapsers are patients whose HCV RNA became undetectable during treatment, but then reappeared after treatment ended. ^[8]

Phase 3 trials using BOC and TVR-based therapy have been performed in HCV G1 treatment-experienced patients. Categories of non-response have been used to predict SVR. As expected, SVR rates were higher in PI-based regimes than in pegIFN/RBV arms. In the two BOC therapy arms in the RESPOND-2 trial, SVR rates were 66 and 59%, and were 21% in the control group. In the BOC-containing arms, SVR rates were higher in prior relapsers (75 and 69%) than in prior partial responders (52 and 40%). In the SOC control arm, SVR rates were 29% for prior relapsers and 7% for prior partial responders; null responders were not included in this study. ^[18] Overall SVR rates in the TVR-containing groups in the REALIZE trial were 64% and 66%. In these groups SVR rates according to previous response were 83 and 88% in relapsers, 59 and 54% in partial responders, and 29 and 33% in null responders. In the control SOC group SVR rates were 24, 15 and 5%, respectively. ^[19] Previous response to pegIFN/RBV treatment influences the outcome of PI-based triple therapy. This issue highlights the importance of adequately documenting previous type of response. The highest SVR rates occurred in prior relapsers, a lower rate was achieved in partial responders, and the lowest rate occurred in null responders (only with TVR). ^[18,19]

Previous response to pegIFN/RBV can be combined with the other predictive factors mentioned previously to increase its predictive value. In the RESPOND-2 trial, F0–2 stage had similar SVR rates in both BOC groups: 68 and 66%, versus 23% in pegIFN/RBV group. However, in patients with F3–4 stages, SVR rates were higher in patients treated with the BOC 48-week duration regime (68%) than in those treated with BOC RGT (44%) or pegIFN/RBV alone (13%). ^[46] Relapsers had similar SVR rates in the TVR arm independent of fibrosis stage: F0–2 86%, F3 85% and F4 84%. But SVR rates were lower according to fibrosis stage in partial responders (F0–2 72%, F3 56% and F4 34%) and even lower in null responders (F0–2 41%, F3 39% and F4 14%). ^[19] The predictive values of the different factors evaluated in the BOC trials are shown in Table 3.
       
Thus, achieving SVR in treatment-experienced patients will mainly depend on the type of prior response to pegIFN/RBV, as well as on the other predictors mentioned. Also other reasons for treatment failure, such as inadequate dosing or side-effect management, have to be taken into account when deciding to retreat a patient.
       
Adherence.
Previous SOC required a moderately complex regimen of twice-daily oral RBV and weekly subcutaneous injections of pegIFN, as well as frequent monitoring of laboratory results and adverse effects. It had been demonstrated that strict adherence to pegIFN/RBV regimen was associated with higher SVR rates. ^[47–50] Triple PI-based therapy increases treatment complexity: it increases the number of pills to be taken with food, not low-fat in the case of TVR; increases the number of laboratory controls; increases the number, and in some cases the severity of side effects, some only related to the PIs (i.e., rash, disgeusia, etc.); and PI misuse increases the risk of RAV emergence. One benefit, is that treatment can be shortened in a large proportion of patients with PI-based therapy. ^[8,16–19] Studies using PIs outside clinical trials will show how this treatment is tolerated in clinical practice and how treatment adherence impacts SVR rates.

On-treatment Predictors of Response
Early viral kinetics are useful in predicting treatment response when treatment has been initiated. With pegIFN/RBV treatment, it was clear that early viral kinetics are the strongest predictor of achieving SVR: obtaining a rapid virologic response (RVR, defined as HCV RNA being undetectable at week 4 of therapy) was the strongest predictor of SVR, no matter what adverse baseline predictors may have been present. ^[51] Patients who have achieved RVR with PI-based triple therapy are also likely to achieve SVR. On the contrary, HCV RNA still being detectable at week 12 of therapy is associated with a low likelihood of obtaining SVR. ^[16,17]

Pretreatment host and viral characteristics affect early viral kinetics. Once treatment has been initiated, outcome depends on how fast HCV RNA becomes undetectable. A retrospective study of the IDEAL trial showed that the IL28B genotype loses its predictive value for SVR after including RVR in the analysis. A patient with CC genotype who did not obtain RVR, has a lower probability of achieving SVR than a patient who has CT or TT genotype but has achieved RVR. ^[40] According to IL28B genotype, patients achieving RVR (only 14% of the treated patients) had 85 (CC), 76 (CT) and 100% (TT) SVR rates (p = 0.25 CC vs non-CC genotype). On the contrary, in patients not achieving RVR, IL28B status predicted SVR rates: 66 (CC), 31 (CT) and 24% (TT) SVR rates (p < 0.001 CC vs non-CC genotype). ^[40] Another study showed that the strongest predictor of SVR is RVR (OR: 5.35; 95% CI: 2.80–10.19; p < 0.0001), while IL28B CC can still predict SVR in the multivariate analysis (OR: 2.66; 95% CI: 1.54–4.61; p < 0.0001). ^[52] It also showed that IL28B predicted HCV RNA undetectability at each of the week 4, week 8, and week 12 testing points during pegIFN/RBV treatment. These results are consistent with IL28B genotype being strongly associated with the viral kinetics of IFN response. On-treatment viral kinetics may represent a final common pathway of response reflecting IL28B genotype. So, once a virological response has been achieved, SVR is set and independent of IL28B. Based on these results, achieving RVR is considered to be the strongest predictor of SVR versus all baseline predictors, including IL28B genotype (OR: 9.1, vs no RVR and non-CC IL28B genotype; p < 0.001) ( Table 1). ^[40] In this setting of a RGT with pegIFN/RBV treatment, IL28B genotype might not add significant clinical information to early viral response.
In the PI-therapy era, early viral kinetics remained as crucial determinants of treatment response. On-treatment virologic response definitions varied between pegIFN and PI-based therapy ( Table 4). The pegIFN/RBV LI used in the BOC-containing regimes gives important data about IFN responsiveness. Naive patients treated with BOC-based triple therapy in the SPRINT-2 trial not achieving a >1 log ₁₀ decline in HCV RNA during the LI had a 28–38% SVR rate following triple therapy, compared with a 79–81% SVR rate in those achieving a >1 log ₁₀ decline. ^[16] In a multivariate analysis of the Phase III trials data, LI response was the strongest predictor for achieving SVR (OR: 9.0; p < 0.001). ^[53] When LI response and IL28B are included in the analysis, the latter loses its predictive value for SVR ( Table 3). ^[44]

In treatment experienced patients, a subanalysis of RESPOND-2 trial showed that BOC regime, type of previous non-response, low baseline viral load, and mild fibrosis were significantly associated with SVR (without including IL28B genotype). ^[54] However, when lead-in response was added into the model it was found to be the strongest predictor of SVR (OR: 5.2; p < 0.001), even stronger than previous type of non-response (OR: 3.0; p < 0.0001). When IL28B genotype was added to the model (available for 66% of patients), its predictive value was lost, and response to LI continued to be a significant predictor for SVR (OR: 1.8; p < 0.0001) ( Table 3). ^[44]

BOC-based therapy requires HCV RNA responses at week 4 (after LI) and at week 8 to implement RGT. IL28B genotype was found to be a strong predictor of both responses in a retrospective analysis performed after SPRINT-2 and RESPOND-2 trials. ^[43,44] It was found that most IL28B CC genotype naive (89%) and previously treated (82%) patients presented as HCV RNA-negative at week 8, and therefore were able to reduce treatment duration with BOC. In CT and TT genotypes, 52% of naive and 51% of previously treated patients were able to reduce treatment duration. Also, the favorable IL28B CC genotype strongly predicted HCV RNA decline ≥1 log ₁₀ after LI in naive patients (OR: 15.8; 95% CI: 6.3–39.8; p < 0.001) and in previous nonresponders (OR: 4.5; 95% CI: 1.5–13.7; p = 0.007) in the same analysis. Nonetheless, the highest SVR rates were achieved by BOC-treated patients, who achieved a ≥1 log ₁₀ decline in HCV RNA levels after LI, regardless of the IL28B genotype ( Table 3). ^[43,44]

Even with TVR, response to LI can be particularly useful to establish whether a prior null responder patient has chances to achieve SVR with a new course of treatment. In patients treated with TVR-based therapy, response type to prior treatment was shown to be a stronger predictor of SVR than on-treatment response to LI in a retrospective analysis of the REALIZE trial. ^[55] In the TVR LI arm 88% of prior relapsers, 54% of prior partial responders and 33% of prior null responders achieved SVR. Moreover, TVR-treated patients achieving >1 log ₁₀ decline in HCV RNA levels after LI achieved SVR in 94% of prior relapsers, and 59% of prior partial responders; in the same arm, patients not achieving this >1 log ₁₀ decline had SVR rates of 62 and 56%, respectively, for the same categories of nonresponse. In prior null responders, achieving ≥1 log ₁₀ versus <1 log ₁₀ decline in HCV RNA levels after LI was associated with a >3-fold higher likelihood of achieving SVR: 54 versus 15%, respectively. ^[55] Lead-in phase response to pegIFN/RBV in previous null responders may help in making the decision whether or not to proceed to PI-based triple therapy in this subgroup of patients.
Early viral responses help in identifying possible patients failing to treatment. Patients failing response after LI at week 4 of treatment, as well as those not achieving low HCV RNA levels after initiating PI-based treatment, are unlikely to obtain SVR. In the TVR arms in the ADVANCE and ILLUMINATE trials, none of those patients with HCV RNA levels >1000 IU/ml at week 4 of therapy obtained SVR. So, this criterion was established as a futility rule for TVR to stop all treatment at this point. ^[17,56,57] Almost none of the BOC-treated patients with HCV RNA levels ≥100 IU/ml at week 12 of treatment achieved SVR. So, this criterion was established as a futility rule for BOC to stop all treatment at this point. ^[58,59]

Patients treated in the BOC-containing arms in RESPOND-2 and SPRINT-2 trials who become HCV RNA-undetectable by week 8 achieved 81–100% SVR rates, regardless of IL28B genotype. ^[44] None of the patients failing to achieve >1 log ₁₀ decline after LI at week and >3 log ₁₀ decline in HCV RNA at week 8 achieved SVR (0 out of 44). ^[44,60] Despite having a <1 log ₁₀ decline at week 4, patients who became HCV RNA-negative by week 8, achieved SVR in 83% of cases. ^[44] In this subgroup of patients, 3 log decline by week 8 may be used as an additional stopping rule.
Following futility rules to stop PI-based treatment is critical, because they strongly predicted the possibility for not achieving SVR ( Table 5). If PI-based therapy is maintained despite its failure, not only is resistance emergence almost certain, but this will increase resistant virus fitness and may compromise future therapies.

Conclusion
The first generation of PI-based therapy, BOC and TVR, combined with pegIFN/RBV, has increased SVR rates across many viral and host factors. Almost all patients are more likely to achieve SVR with PIs than with the older SOC. However, baseline patient characteristics may help in making decisions regarding whether to treat or not, or how to treat a given patient. Also they may help in providing information to the patients about their probabilities of achieving SVR and the probable duration of therapy. In HCV genotype 1-infected patients treated with PI-based triple combination therapy, it is clear that baseline factors help to predict SVR, but on-treatment virologic response appears to be the strongest factor predicting SVR.

Future Perspective
New generations of direct antiviral therapies are in the pipeline for several companies. Even IFN-free regimes could be a possibility in the near future, perhaps in 3–6 years. These new drugs will change the HCV therapeutic scenario, as well as the predictors of response. We will see in the future how these treatments work, and how we can predict which treatment is the best option for which patient. There is no doubt that HCV treatment will become 'personalized', and there is hope for finding suitable predictors of response for each drug for a given patient.

http://www.medscape.com/viewarticle/775545_6