Early Hepatitis C viral (HCV) kinetics was previously shown to predict sustained virological response (SVR). The current consensus stopping rule is defined at 12 weeks of treatment with peginterferon-alfa and ribavirin. Earlier prediction is needed in order to improve the cost/effect ratio. However, the predictive value of earlier viral kinetics for the current standard of care has not yet been established.
 

The DITTO-HCV European funded project is for now the largest study of frequent early viral kinetics during therapy with peginterferon-alfa and ribavirin. The researchers aim to optimize viral kinetics prediction criteria in order to enable stopping or tailoring treatment as early as possible.

Primary objectives: 1) negative predictive value (NPV) as close as possible to 100% (thus minimalizing stopped patients with missed SVR), 2) high specificity (thus increasing the fraction of patients that can be safely stopped). 3) a high positive predictive value (PPV) using the same criteria.
 

134 chronically HCV infected naïve patients received peginterferon-alfa-2a (40KD) (Pegasys ®) (180 μg qw) plus ribavirin (Copegus ®) (1000-1200 mg qd) for 48 weeks. Of those, 67%, 28%, 4% and 1% had HCV genotype 1, 2-3, 4 or 5 respectively. Viral kinetics were measured according to centralized serum HCV RNA quantifications on days 0, 1, 4, 7, 8, 15, 22 and 29 (Cobas Amplicor^® HCV Monitor v2, Roche). SVR was defined as undetectable serum HCV RNA (<50 IU/ml) after 24 weeks of follow-up post treatment. The early viral response prediction criteria (EVR) and thresholds tested here (see Table 1 below) were prospectively defined.

Study Results

The overall SVR rate was 64%. The EVR studied, together with their NPV, PPV, specificity (SP), statistical significance and number of HCV-RNA quantifications needed (NSMP), are given in Table 1 below.

TABLE 1

The EVR-w12 criteria has good NPV (100%) but its specificity is low (21%), thus relatively few non-responders can be stopped. None of the simple 2nd-slope or Week-1 criteria shows NPV higher than 90%, thus not adequate for a stopping rule.

 

A new composite criteria (DITTO-2nd-slp) gives NPV=100% and specificity=34% already after 4 weeks of treatment. Furthermore, another new composite criteria (DITTO-1st-week) also gives NPV=100% with specificity=28% already after 1 week.

Although composite, these new criteria are simple to calculate and necessitate only 1 or 2 more viral quantifications compared to the current Week-12 criteria. Thus allowing to lower the mean cost by 2410 USD per patient. Note that SVR (56% for gen 1 and 87% for gen 2-3) is predicted independently of genotype.

 

Moreover, by selecting more strict thresholds for the DITTO-1st-week (e.g. 4.5 log at day 4 and 2 log decline at day 7) and DITTO-2nd-slope (e.g, 0.6 log/week) criteria one can obtain higher PPV up to 90%.

 

Conclusions:

Sustained viral response to treatment with peginterferon-alfa-2a (40KD) and ribavirin can be predicted, independent of genotype, as early as 1 or 4 weeks of treatment by new composite, but still simple and practical, criteria of viral kinetics. Similar NPV and PPV and better specificity was obtained here by the DITTO-1st-week or DITTO-2nd-slope prediction criteria, giving rise to significantly lower cost/effect as compared to the current stopping rule at 12 weeks. These results warrant future prospective testing in larger trials, since they can give rise to considerable saving in cost and quality of life related to over-treatment or alternatively allow early tailoring of treatment.

Discussion

Can the new predictive criteria move the current 12- week stopping rule to under a month?

The DITTO-HCV study retrospectively identified two new criteria to classify rapid viral responders and predict the likely treatment outcome. For the DITTO-1^st week criterion viral levels are measured at baseline and two times in the first week of treatment; while the DITTO-2^nd slope criterion uses three measurements of viral levels between the second and fourth week of treatment. 

“Both of these new criteria predicted who was and who was not likely to respond to treatment more accurately than the existing 12-week stopping rule. We had obtained 100 per cent negative predictive value (NPV) and 90 per cent positive predictive value (PPV) with these criteria” said Professor Avidan Neumann, from the Bar-Ilan University, Israel, and the DITTO-HCV study coordinator who presented the study results at an oral session at AASLD.

Using these criteria for stopping treatment in patients predicted early on not to respond will make treatment more cost-effective, even considering the added cost of the extra measurements of viral load.

“This is good news for patients and physicians.  Using these new criteria, we will be able to very early, identify more of those patients who will ultimately not achieve an SVR so we can advise them to stop taking their medication, thus improving their quality of life,” said Professor Neumann.

“We now know that tailoring treatment according to viral kinetics cannot improve SVR rates with the treatment choices we have at the moment.  However, by measuring viral levels earlier in treatment, a test physicians are already familiar with, we can likely move the stopping rule currently defined at week 12 down to week 4 or even week 1 and avoid having patients who will not achieve an SVR being treated unnecessarily. These novel prediction algorithms will hopefully be confirmed soon by other clinical trials and their feasibility in clinical practice will be assessed, thus allowing us to optimize the treatment of hepatitis C patients," concluded Prof. Neumann. 

10/27/03

Reference
AU Neumann and others. EARLY VIRAL KINETICS PREDICTION OF SUSTAINED VIROLOGICAL RESPONSE AFTER 1 OR 4 WEEKS OF PEG-INTERFERON-ALFA-2A AND RIBAVIRIN THERAPY (DITTO-HCV PROJECT).

A Comparison of Standard Treatment Versus Dynamically Individualized Treatment in Patients with Chronic Hepatitis C
 

Combination therapy with peginterferon alfa and ribavirin for 48 weeks achieves sustained virologic response rates (SVR) of 54-61% in patients with chronic hepatitis C. However, the SVR rates are highly variable according to baseline (HCV genotype) and on-treatment parameters (initial viral decline). Thus, it must be anticipated that current standard therapy recommendations lead to under-treatment in some and over-treatment in other individual patients.
 

Comparison between a dynamically individualized treatment schedule according to the early virologic response versus the standard of care combination therapy with peginterferon alfa-2a (40KD) (Pegasys) (PEG-IFN) (180 μg qw) plus ribavirin (Copegus) (RBV) (1000-1200 mg qd) for 48 weeks.

The primary aim of the study was to increase SVR, while optimizing the available drugs, treatment duration, quality of life and the socio-economical burden of therapy. The secondary aim of the study was to enable a comprehensive analysis of viral/host/immune correlates of response to treatment (ongoing).
 

All patients (n=273) were initially treated with PEG-IFN/RBV for 6 weeks and initial viral kinetics were defined according to centralized serum HCV RNA quantifications (Cobas Amplicor HCV Monitor v2, Roche Molecular Systems) on baseline and days 0, 1, 4, 7, 8, 15, 22 and 29.

After classification into viral response categories at 6 weeks, patients were randomized (n=270) to individualized therapy (arms A1, A2, B1, B2, C, D) or continuation of standard therapy (STD arm). Treatment tailoring included: in rapid viral responders (RVR) - discontinuation of RBV (A1) or shortening of treatment duration to 24 weeks (A2); in slow partial responders (SPR) - addition of histamine (B1) or prolongation of treatment to 72 weeks (B2); in flat partial responders (FPR) - addition of histamine (C); in null responders (NUR) - retreatment with high-dose of PEG-IFN (360 μg qw) plus RBV (D). SVR was defined as undetectable serum HCV RNA (< 50 IU/ml) at the end of 24 weeks of untreated follow-up.
 

Study Results

Demographic and baseline virologic parameters were similar in the standard and the individualized treatment groups. According to the initial viral decline patients were categorized as RVR (51% for genotype 1 and 94% for genotype 2-3), SPR (35% and 5% accordingly), FPR (5% and 0% accordingly), NUR (10% and 1% accordingly) or unclassified (1%). The overall SVR rates for genotype 1 patients were 49% for the individualized and 56% for the standard treatment arm (NS), and for genotype 2-3 patients 90% and 87% respectively (NS). SVR rates (by ITT analysis) according to HCV genotype and within each initial viral response category and arm are given in Table 1 below.

Table 1

Conclusions

The overall sustained virologic response rates of 53% in genotype 1 and 88% in genotype 2-3 patients with chronic hepatitis C treated with Pegasys (peginterferon alfa-2a/40KD) in combination with ribavirin support previously presented data.

Individualized treatment according to initial viral kinetics appears to be clinically feasible, but did not improve the sustained virologic response rate with the drugs and dosages usable at the time of this study.

Nevertheless, the possibility that in rapid viral responders discontinuation of ribavirin does not decrease the sustained virologic response rate warrants future prospective trials.

Discussion

Rapid Viral Responders Have Most Promising Results

Rapid viral responders were prospectively defined as patients whose HCV RNA declined by at least 99 per cent during the first month of treatment. The study found that in this sub-group, even the most difficult-to-treat genotype 1 patients could achieve 83% SVR.

A relatively high SVR rate (71%) was obtained even for the rapid responding patients treated with ribavirin for only the first 6 weeks instead of the standard regimen of ribavirin during the whole peginterferon treatment of 48 weeks. These SVR rates are similar to that achieved by genotype 2/3 patients (88%), who traditionally have been easier to treat.

“It is therefore critical for these rapid responding patients to be identified” said Professor Zeuzem.  

Supported by the European Community (QLK2-2000-00836), Hoffmann La-Roche and Maxim Pharmaceuticals.

10/24/03

References

S Zeuzem and others (for the The DITTO-HCV study--Dynamically Individualized Treatment of Hepatitis C Infection and Correlates of Viral/Host Dynamics study).

INTERNATIONAL, MULTICENTER, RANDOMIZED, CONTROLLED STUDY COMPARING STANDARD VERSUS DYNAMICALLY INDIVIDUALIZED TREATMENT IN PATIENTS WITH CHRONIC HEPATITIS C (DITTO-HCV PROJECT). Abstract 317 (poster). 54^th AASLD. October 24-28, 2003. Boston, MA.

Multivariate Models Predict Outcome of Therapy for HCV at Baseline and at Weeks 4 and 8

Given the current burden of hepatitis C virus (HCV) therapy, physicians have been seeking a reliable early stopping rule for patients undergoing pegylated combination therapy who will be non-responders.

French researchers used data from 174 chronically infected HCV patients to develop multivariate models (MM) to predict non-response (NR), sustained response (SR) and relapse (RR) from information available in the first 8 weeks of therapy. They compared information from univariate models (UMs), based solely on quantitative or qualitative HCV RNA information, with that derived from MMs.

174 chronically infected HCV patients (94 treatment naïve and 80 previously failed) were treated with PEG-Intron (pegylated interferon alfa-2b) 1.5 mg/kg/week plus ribavirin 800-1200 mg/day. Naïve genotype 1 or 4 patients and previously failed patients were treated for 48 weeks; patients with HCV genotype 2, 3, or 5, for 24 weeks.

SR was defined as undetectable HCV RNA by the VERSANT^R HCV RNA Qualitative Assay (TMA) (Bayer Diagnostics) at end-of-treatment (EOT) and follow-up at week 24 (FU 24). Serum HCV RNA was quantified by the VERSANT^R HCV RNA 3.0 Assay (bDNA) (Bayer Diagnostics); at baseline, weeks 4, 8, EOT, and FU24. If the HCV RNA concentration was below 615 IU/mL, specimens were then tested by HCV Qual (TMA) (limit of detection ≤ 9.6 IU/mL.)

 

Univariate models: Baseline UM utilized a prediction threshold of 6.14 log₁₀ copies/mL to predict outcome. Week 4 and 8 UM used a 2 log₁₀ drop rule ( ≥ or < 2 log drop in viral load from baseline; UM2 log) or clearance of virus or lack thereof by the HCV Qual (TMA) (UM TMA).

Multivariate models: MM employ ordinal regression with similarity least squares technology to assign a given outcome to the highest probability of response. Different numbers of "critical patients" (n= 33-57) were used to derive the models using design of experiment and functional techniques for each time point; the models were then tested on the remaining patients (n >110). Model variables included: HCV RNA concentrations at baseline and weeks 4 and 8, gender, age, baseline ALT, inflammatory and fibrosis scores, genotype and treatment status (naïve or previous therapy).

 

Study Results

The comparative sensitivity specificity, positive and negative predictive values (PPV and NPV) at for the UMs and MMs are shown in the table below. Baseline and week 4 models combined NRs and RRs into non-responder group, whereas week 8 model correctly separated SRs, RRs, and NRs.

Prediction of SR or NR At Baseline and Weeks 4 and 8 of Therapy

1. percent of SR identified by model; 2 percent of non-responders identified by model; 3 percent of SR predicted by model who are true SR; 4 percent of NR predicted by model who are true NR; 5 ≥ or <6.14 log₁₀ c/mL; 6 ≥ or < 2 log₁₀ change in c/mL; 7 HCV RNA detected or not detected by TMA.

Conclusions: At each time point, the multivariate models demonstrates greater sensitivity, specificity, PPV and NPV than do the univariate models. Multivariate models can identify SR and non SR (NR and RR as one group) at baseline and week 4. By week 8 multivariate models can correctly identify SR, RR, and NR.

The 100% prediction of non-response by the multivariate model at week 8 with the 97% NPV of < a 2 log drop suggests a week 8 stopping rule could be employed.

11/07/03

Reference
M Martinot-Peignoux and others. MULTIVARIATE MODELS PREDICT OUTCOME TO HEPATITIS C VIRUS THERAPY AT BASELINE AND AT WEEKS 4 AND 8. Abstract 1215 (poster). 54^th Annual Meeting of the American Association for the Study of Liver Diseases. October 24-28, 2003. Boston, MA.

HIV-HCV Coinfection Status Does Not Affect HCV Viral Kinetic Response to Interferon

Viral kinetic modeling of HCV response to interferon-based therapy provides important insights into factors associated with treatment outcomes. HIV-HCV coinfected patients appear to have lower overall response rates than that observed in monoinfected subjects, but the reason for this is not clear. This research group speculated that kinetic responses in key parameters would be decreased in coinfected subjects.
 

HIV-HCV coinfected patients and HCV monoinfected patients prospectively matched for treatment, genotype, age (+ 5 yrs), gender, race, and histology were evaluated.

Coinfected patients were randomized within the context of a large U.S. multicenter trial (ACTG 5071) to receive Pegasys (pegylated interferon alfa-2a) + ribavirin vs. Roferon (interferon alfa-2a) + ribavirin. Quantitative HCV RNA (Roche COBAS AMPLICOR) kinetic testing was performed at 0, 6, 12, 24, 48, and 72 hours and at days 7, 14. Non-linear regression and linear models were evaluated in an effort to best predict response and identify prognostic factors.
 

Study Results

Twenty-seven subjects underwent viral kinetic sampling and evaluation. These included twelve HIV-HCV case subjects (10 men, 2 women) and 15 matched HCV controls (some patients double-matched).

The mean age of coinfected subjects was 46.1 years (range, 38-60). Among HIV+ subjects the mean CD4+ count was 325 cells/mm³ (range, 175-855). 11/12 had baseline HIV RNA <400 copies/ml. Mean HCV viral load was 6.6 log IU/ml among coinfected vs. 6.2 log IU/ml in controls.

75% of coinfected subjects had HCV genotype 1. The remainder were genotype 2. Twenty-five (25%) of coinfected patients had no detectable virus 24 weeks after completion of 48 weeks of therapy (SVR).

Overall SVR in control subjects was 46.6%. Efficiency (ε) of Phase 1 response (λ) slope at 72 hours, lambda₂ (slope of Phase 2 decline) were calculated (see table below).

Efficiency of clearance (ε) at 72 hours was highly associated with actual viral clearance across all groups (p=0.001) but λ₂ was not. Regression analysis failed to demonstrate a relationship between ε and baseline CD4+ count, HCV viral load or genotype. In contrast, #955;₂ was significantly associated with genotype.

Viral kinetic parameters were not predictive of SVR.
 

Conclusions

• Viral kinetic modeling demonstrated that the efficiency of clearance at 72 hours was significantly associated with viral clearance during treatment;
• Coinfection status did not affect key kinetic parameters;
• PEG-IFN was superior to standard interferon in terms of viral clearance efficiency, particularly in the coinfected group; and
•  Diminished SVR in coinfected patients may be related to immune factors that are operative after reduction of viral loads to undetectable levels.

11/10/03

Reference
KE Sherman and others. MODELING HCV VIRAL KINETIC RESPONSE TO PEG-IFN/RIBAVIRIN IN HCV-HIV COINFECTED PATIENTS. Abstract 315 (poster). 54^th Annual Meeting of the American Association for the Study of Liver Diseases. October 24-28, 2003. Boston, MA.