WO2010062676A1 - Modeling evolutionary dynamics of hepatitis-c resistant variants in patients dosed with telaprevir - Google Patents

Abstract

The invention relates to a method of measuring and predicting the evolutionary dynamics of HCV genotype 1 population when dosed with a composition including a protease inhibitor. The method includes using a multivariant viral kinetic model consisting of two-step process of random mutation and selection by competition among various HCV variants for shared limited healthy target cells constituting replication space to measure and predict the evolutionary dynamics of HCV genotype (1) population thereby providing useful information on dosing and responsiveness to compositions including a protease inhibitor. The model parameters are estimated from pre-, during-, and post-dosing plasma HCV RNA and post-dosing clonal sequencing data using a scheme that integrates all dosing periods to impose consistency and to improve estimation results.

Description

Modeling evolutionary dynamics of Hepatitis-C resistant variants in patients dosed with telaprevir

Cross -Reference

[001] The present application claims priority to U.S. Application No. 61/109,578 filed on October 30, 2008, the contents of which are incorporated herein by reference in its entirety.

Technical Field of The Invention

[002] The invention relates to a method of measuring and predicting the evolutionary dynamics of HCV genotype 1 population when dosed with a composition including a protease inhibitor. The method includes using a multivariant viral kinetic model consisting of a two-step process of random mutation and selection by competition among various HCV variants for shared limited healthy target cells constituting replication space to measure and predict the evolutionary dynamics of HCV genotype 1 population thereby providing useful information on dosing and responsiveness to compositions including protease inhibitors.

Background

[003] Hepatitis C virus (HCV) is a chronic disease estimated to infect 170 million people worldwide. Recently, several specifically-targeted antiviral therapies for HCV (STAT-C) have been developed, including inhibitors to HCV NS 3/4 A protease and HCV NS5 polymerase. Development of STAT-C, similar to its counterpart in HIV₅ has been challenged by emergence of resistant variants. These variants are members of the estimated vast number of HCV quasi-species — a consequence of the high replicative rate of HCV and its error-prone RNA polymerase. One of the STAT-C compounds is telaprevir (TVR, also known as VX-950), an HCV NS3/4A protease inhibitor that has demonstrated potent activity in HCV replicon assay and in clinical trials. For patients dosed with TVR the plasma HCV RNA and prevalence of variants post-TVR dosing measured by clonal sequencing have been reported. [004] Models of HCV viral kinetics in patients treated with interferon, pegylated interferon, and ribavirin, have been presented. These models have assumed the HCV population within a subject to be relatively homogeneous with respect to sensitivity to these antiviral agents. However, as a consequence of its high replication rate and error-prone polymerase, HCV exists as a quasispecies. In fact, recent data from clinical trials evaluating HCV protease inhibitors have revealed the presence of viral variants with varying levels of sensitivity to these agents. Viral variants have also been detected at levels around 10^"3 of wild-type NS3^»4A HCV (WT) level prior to dosing in treatment-naϊve subjects. Upon exposure to protease inhibitors, the composition of the HCV quasispecies was altered, as revealed by sequencing of plasma HCV RNA and isolated viral clones obtained from subjects dosed with telaprevir and boceprevir. These variants have also been reported to exhibit reduced fitness and reduced susceptibility to other protease inhibitors in vitro. [005] In these models, the observed biphasic declines in HCV RNA during treatment were explained with a rapid plasma viral clearance rates and a slower infected-cell clearance rates. Recently, higher second-phase declines in wild-type HCV have also been observed in patients dosed with TVR.

[006] Models of population dynamics and emergence of resistance have also been developed for HIV. In these models, variants were assigned different replicative rates, either through their infection or production rates, or both. Typically, these models were parameterized using few during treatment HIV-RNA, CD-4 counts, and fraction of few resistant variants, however, many of them lacked sufficiently sampled data measurements to allow confident estimates of model parameters.

Summary of the Invention

[007] In general, the invention relates to a method of measuring and predicting the evolutionary dynamics of HCV genotype 1 population when dosed with a composition including a protease inhibitor. The method includes using a multivariant viral kinetic model consisting of a two-step process of random mutation and selection by competition among various HCV variants for shared limited healthy target cells constituting replication space to measure and predict the evolutionary dynamics of HCV genotype 1 population thereby providing useful information on dosing and responsiveness to compositions including protease inhibitors. [008] The model can also be used to measure fitness of variants by representing the variants in the absence of a protease inhibitor by their different replication capacities in competing for the limited replication space, and in the presence of a protease inhibitor by how much protease inhibitor blocked their replications.

[009] The model can also be used to investigate the importance of replication space dynamics, mutations during treatment, and pre-existing variants on the overall response.

[010] The model parameters, including the rate constants for wild-type and variant production, plasma virion clearance, infected-cell clearance, effective protease inhibitor concentration, were estimated from 1) pre-, during-, and post-dosing plasma

HCV RNA, 2) postdosing clonal sequencing of variants from the subjects, and 3) sensitivity of variants to the protease inhibitor in replicon cells, using a scheme that integrated all these data to impose consistency.

[011] The results demonstrated at least three advantages of estimating fitness using the modeling approach proposed here when compared to the relative fitness (RF) method:

1. The modeling approach was valid at both increasing and constant viral load while the RF method was valid only for either of these conditions but not both.

2. The modeling approach relied on more data than the two-point measurements used by the RF method and thus, could estimate fitness from more subjects.

3. The modeling approach produced fitness estimates that were more consistent with the data.

[012] The resulting variant fitness estimates from the modeling approach demonstrated reduced replication capacity of variants, with production rates that were < 68% of the wild-type rate. The pre-dosing prevalence of variants was predicted to be near their respective influx mutation rates from wildtype, with the prevalence of single-mutant variants of < 3.7xlO^~4 fraction of the total population, and that of double-mutant variants was < 9x1 CF^S fraction of the total population. [013] In one aspect, the invention features a method of adjusting the dosing level of a composition comprising a protease inhibitor administered to a patient. The method includes measuring plasma HCV RNA levels from a patient; utilizing the measured HCV RNA levels in a multi- variant kinetic model to calculate the responsiveness of the patient to the administered composition comprising a protease inhibitor; comparing the calculated responsiveness to a predetermined responsiveness to compositions comprising a protease inhibitor. The method may further include adjusting the dosing level of the composition comprising a protease inhibitor administered to a patient based upon the comparison of the calculated responsiveness to the predetermined responsiveness. The multi-variant kinetic model may account for one or more HCV genotype 1 resistant variants selected from R155M, T54A, T54S, V36M, R155K, V36A, A156S, R155T, V36M/R155K, A156T, A156V, and V36M/T54S. The method may further utilize the measured HCV RNA levels in a multi-variant kinetic model to calculate the responsiveness of the patient to the administered composition comprising a protease inhibitor. This includes determining the fitness. The plasma HCV RNA levels from a patient may be measured within the first 20 days of administration. The measured HCV RNA levels are utilized in the multi-variant kinetic model to calculate the initial responsiveness of the patient to the administered composition comprising a protease inhibitor. The initial responsiveness is compared to a predetermined responsiveness and based upon that comparison calculating a concentration of a protease inhibitor to be subsequently administered. [014] In some embodiments, the protease inhibitor is a NS3/4A inhibitor. In a preferred embodiment, the NS3/4A inhibitor is telaprevir.

[015] All of the documents cited herein are incorporated herein by reference in their entireties.

Detailed Description of the Figures

[016] Figure 1. a, a superset of HCV genotype Ia variants uncovered in patients dosed with telaprevir and mutations among them. Each node represents a variant of which the amino-acid mutation is printed in bold font. The underlying NS3/4A protease nucleotide sequences at selected positions are listed immediately below. Legends: solid lines, a one-nucleotide change; dotted lines, a two -nucleotide change, b, a schematic of a multi-variant viral kinetic model proposed here. Boxes with solid borders, state variables; boxes with dashed borders, fluxes. Model parameters were either estimated (solid fill) or pre-computed (striped fill), c, d, correspondence between data and best- fit model for Patient 1. Points, data; solid line, best-fit model; dashed lines, predicted variant contribution to the overall plasma HCV RNA; circles, HCV RNA levels of variants (limited to variants with prevalence > 5%). [017] Figure 2. Correspondence of data and best-fit models of alternative cases of replication space 7¹ dynamics applied to Patient 1. Legends: Black lines, HCV RNA; grey lines, T/T_maκ; solid lines, T followed Equation 1 and s was optimally estimated at 0.03 h^"1 ; dashed lines, T followed Equation 1 and s was fixed at 1 h^"1; dotted lines, T followed Equation 5 and γ was fixed at 0.05 h^"1 - a value comparable to s = 1 h^"1 in Equation 1 (dashed lines).

[018] Figure 3. Perturbation analysis of best-fit models of Patient 1. (a), if mutations during treatment were absent. Legends: diamonds, data; solid lines, models without mutations during treatment; dotted lines; models with mutations during treatment, (b), if resistant variants were not present at 0.4 day before dosing. Legends: diamonds, data; lines, models with no variants present at 0.4 day before dosing. [019] Figure 4. Comparison of models with and without mutations among variants for Patient 1. Legends: solid lines, without mutations; dotted lines; with mutations. [020] Figure 5. Comparison of cases with (δ, ≠ 0) and without (δ_L = 0) telaprevir enhanced infected-cell clearance rates. In δi ≠ 0 case, δ[ was estimated from data while δ₀ was fixed at 5.2 x 10^~3 h^~'; in δ_{ ≠ 0 case, δ₀ was estimated from data while δi was fixed at zero, a, objective functions for both cases, b, c, correspondence between plasma HCV RNA (b) and clonal sequencing (c) data and best-fit models for Patient 1. Legends: solid lines, best-fit models with S₁ = 0 case; dotted lines, best-fit models with 8_\ ≠ 0 case; dashed lines, variant HCV RNA predicted by best-fit models with δj = 0 case.

[021] Figure 6. Sensitivity analysis of plasma HCV RNA and variant prevalence to model parameters in an example of a two-variant viral kinetic model. Simulation was generated with base parameters of/, = l ; /₂= .01; p = 0.2; β = 0.05; c = 0.2; ^₁ = 0.04; S₀ = 0.01 ; s = 0.1; s = 0.01 ; m = 10^"4; [tvr] = 13; IC_50,, = 11.35; IC₅₀,₂= 500; h = 2.012; and perturbations of two-times higher /₂, p, c, δ_[t [tvr], S₀, and 10-times higher s and m. Legends: dashed lines, base parameterization; solid lines, perturbed parameterization; black, ensemble HCV RNA; lighter gray, variant 1; darker gray, variant 2. [022] Figure 7. Correspondence between data and model estimation in three additional subjects, a, Subject 2 HCV RNA levels; b, Subject 2 variant prevalence; c, Subject 3 HCV RNA levels; d, Subject 3 variant prevalence level; e, Subject 4 HCV RNA levels; f, Subject 4 prevalence. Diamonds, data; solid line, best-fit model; dashed lines, predicted variant contribution to the overall plasma HCV RNA; circles, HCV RNA levels of variants (limited to variants with prevalence > 5%). [023] Figure 8. The roles of replication space (T) kinetics to model estimates. Two cases of T kinetics were examined: First, T synthesis rate s was estimated from data; Second, s was fixed to 1 h^™1. a, maximum likelihood objective functions; b, distribution of estimated s values; c, estimated absolute replication rates; d, correspondence between / values; e-f, f values of two representative variants. Legends: est_s, first case; fix_s, second case.

[024] Figure 9. Small contribution of mutations during treatment to the emergence of resistant variants, even when the mutation rate is 10-times higher. Legends: points, data; black solid line, plasma HCV RNA of best-fit model with 10-times higher mutation rate than base value; dotted lines, contribution of variants to plasma HCV RNA with 10-times higher mutation rate than base; dashed lines, contribution of variants to plasma HCV RNA if mutations do not occur during treatment. The base mutation rate was 1.2 10^'4 per nucleotide changes per replication cycle. [025] Figure 10. Correspondence between in vivo and in vitro replicative fitness. Panel a, correlation with in vivo fitness estimated by modeling in this manuscript. Panel b, correlation with in vivo fitness estimated by relative fitness. The fitness measured in replicon cells has been previously reported.

[026] Figure 11. Estimation results for different values of mutation rates. Panel a, maximum likelihood objective values for mutation rates m of 1.2e-4/cycle, 1.2e- 3/cycle, and 1.2e-5/cycle. The objective functions are the lowest with m=l .2e-4/cycle, suggesting best correspondence between data and models with this value of mutation rates. Panel b, estimated fitness for m=1.2e-4/cycle. Panel c, estimated fitness for m=1.2e-3/cycle. Panel d, estimated fitness for m=1.2e-5/cycle. Similar ranking of fitness estimates were obtained with these different values of mutation rates. [027] Figure 12. The mutational network for genotype Ib patients. [028] Figure 13. Sensitivity of estimation results to assumption of β values applied to data from Subject 1. Panel A, Maximum likelihood objective values. The objective values were similar despite large variations of assumed /? values. Panel B, estimated production rate constant p Production rate p is related to l/β. Panel C, estimated fitness of V36M. Panel D, estimated fitness of R155K. Estimated fitness converged to similar values despite large variations of assumed β values. The estimation was repeated 2000 times for this subject with different β values (with β/ β_o= io^prandom; βrandom as a random variable with mean=0, std=l). Initial seed and bounds for p were adjusted to maintain constant (pβT_mΑX/(cδ)) values (p//?₀=10^"prandom; po as the estimated/? when/?=/?o).

[029] Figure 14. a, comparison of fitness estimates using the relative fitness (RF) and the modeling methods proposed here. Legends: circles, the RF using the first two clonal sequencing data; squares, the RF using the last two clonal sequencing data; triangles, the production rate ratio from the modeling method, b, the number of variant-subjects estimated by each method. Legends: solid bars, the RF using the first two clonal sequencing data; cross-hatched bars, the RF using the last two clonal sequencing data; hatched bars, the production rate ratio from the modeling method. [030] Figure 15. The information content of the day 14 clonal sequencing data, a, correspondence between the f values obtained from estimations including and excluding the day 14 data; b, correspondence between measured and predicted variant prevalence. Legends: points, predictions when the day 14 data were excluded; circles, predictions when the day 14 data were included; bar indicated 95% confidence interval of the predictions.

Detailed Description of the Invention

[031] In one embodiment, the invention provides a method of modeling treatment of a hepatitis C infected patient with a protease inhibitor, comprising the step of quantifying the patient's response to one or more dosing regimens of said protease inhibitor with a viral dynamic model employing one or more of Equations 1, 2, 3, 4(a), 4(b), 4(c) and 5.

[032] In one aspect, the invention provides a method wherein the viral dynamic model includes parameters for hepatitis C genotype Ia or Ib. [033] In one aspect, the invention provides a method wherein the protease inhibitor is a NS3/4A protease inhibitor. In a preferred embodiment, the NS3/4A protease inhibitor is telaprevir.

[034] In one aspect, the invention provides a method wherein the viral dynamic model simulates rebound kinetics of a resistant HCV variant employing Equations 6-7.

[035] In one aspect, the invention provides a method wherein the viral dynamic model employs normalizing a plasma viron value with a baseline value.

[036] In one aspect, the invention provides a method wherein the viral dynamic model is implemented numerically by employing Equations 9- 13.

[037] In one aspect, the invention provides a method wherein the viral dynamic model calculates a mutation rate by exponentiating the rate of 1.2 x 10^~4 per nucleotide position per replication cycle for a single mutation by the number of nucleotide mutations.

[038] In one aspect, the invention provides a method wherein the viral dynamic model employs a mutation rate (m) in a range of between 1.2 x 10^"5 changes/site/cycle and 1.2 x 10^"3 changes/site/cycle.

[039] In one aspect, the invention provides a method wherein the viral dynamic model is optimized with the maximum likelihood objective function defined with

Equation (16).

[040] In one aspect, the invention provides a method wherein the optimization is solved as a two-step process consisting of integration and optimization steps, wherein the integration is performed by employing the staggered corrector sensitivity analysis method and the optimization is performed by employing the nonlinear programming/NLP approach.

[041] In one aspect, the invention provides a method wherein the dynamic viral model calculates an initial decline of HCV variant levels upon dosing of the protease inhibitor by employing Equation 17.

[042] In one aspect, the invention provides a method wherein the dynamic viral model calculates the fitness of variants relative to the fitness of the most-fit variant employing Equation (18) or (19). [043] In one aspect, the invention provides a method wherein the dynamic viral model calculated the fitness of variants relative to the fitness of the most-fit variant at specific time points using a time-derivative of the HCV RNA levels.

[044] In one embodiment, the invention provides a method of adjusting the dosing level of a composition comprising a protease inhibitor administered to a patient, the method comprising: measuring plasma hepatitis C RNA levels from a patient; utilizing the measured hepatitis C RNA levels in a multi-variant kinetic model to calculate the responsiveness of the patient to the administered composition comprising said protease inhibitor; and comparing the calculated responsiveness to a predetermined responsiveness to compositions comprising said protease inhibitor.

[045] In one aspect, the invention provides a method further comprising adjusting the dosing level of the composition comprising a protease inhibitor administered to a patient based upon the comparison of the calculated responsiveness to the predetermined responsiveness.

[046] In one aspect, the invention provides a method wherein the multi-variant kinetic model accounts for one or more of hepatitis C genotype 1 resistant variants selected from R155M, T54A, T54S, V36M, R155K, V36A, A156S, R155T,

V36M/R155K, A156T, A156V, and V36M/T54S.

[047] In one aspect, the invention provides a method wherein utilizing the measured hepatitis C RNA levels in a multi- variant kinetic model to calculate the responsiveness of the patient to the administered composition comprising a protease inhibitor includes determining the fitness.

[048] In one aspect, the invention provides a method wherein the plasma hepatitis C

RNA levels from a patient are measured within the first 20 days of administration.

[049] In one aspect, the invention provides a method wherein the measured hepatitis

C RNA levels are utilized in the multi- variant kinetic model to calculate the initial responsiveness of the patient to the administered composition comprising a protease inhibitor.

[050] In one aspect, the invention provides a method wherein the initial responsiveness is compared to a predetermined responsiveness and based upon that comparison calculating a concentration of the protease inhibitor to be subsequently administered.

[051] In one aspect, the invention provides a method wherein the protease inhibitor is a NS3/4A protease inhibitor. In a preferred embodiment, the NS3/4A protease inhibitor is telaprevir.

[052] In one aspect, the invention provides a method wherein the viral dynamic model simulates rebound kinetics of a resistant HCV variant employing Equations 6-7.

[053] In one aspect, the invention provides a method wherein the viral dynamic model employs normalizing a plasma viron value with a baseline value.

[054] In one aspect, the invention provides a method wherein the viral dynamic model is implemented numerically by employing Equations 9-13.

[055] In one aspect, the invention provides a method wherein the viral dynamic model calculates a mutation rate by exponentiating the rate of 1.2 x 10^"4 per nucleotide position per replication cycle for a single mutation by the number of nucleotide mutations.

[056] In one aspect, the invention provides a method wherein the viral dynamic model employs a mutation rate (in) in a range of between 1.2 x 10^~5 changes/site/cycle and 1.2 x 10^"3 changes/site/cycle.

[057] In one aspect, the invention provides a method wherein the viral dynamic model is optimized with the maximum likelihood objective function defined with

Equation (16).

[058] In one aspect, the invention provides a method wherein the optimization is solved as a two-step process consisting of integration and optimization steps, wherein the integration is performed by employing the staggered corrector sensitivity analysis method and the optimization is performed by employing the nonlinear programming/NLP approach.

[059] In one aspect, the invention provides a method wherein the dynamic viral model calculates an initial decline of HCV variant levels upon dosing of the protease inhibitor by employing Equation 17. [060] In one aspect, the invention provides a method wherein the dynamic viral model calculates the fitness of variants relative to the fitness of the most-fit variant employing Equation (18) or (19).

[061] In one aspect, the invention provides a method wherein the dynamic viral model calculated the fitness of variants relative to the fitness of the most-fit variant at specific time points using a time-derivative of the HCV RNA levels.

[062] In one embodiment, the invention provides a computer system for modeling treatment of a hepatitis C infected patient with a protease inhibitor, comprising a computer-readable medium storing a computer program for quantifying a patient's response to one or more dosing regimens of said protease inhibitor with a viral dynamic model of Equations l-4a to provide quantified patient's response to the dosing regimens.

[063] In one aspect, the invention provides a computer system wherein the viral dynamic model includes parameters for hepatitis C genotype Ia or Ib.

[064] In one aspect, the invention provides a computer system for modeling treatment of a hepatitis C infected patient with a protease inhibitor, wherein the protease inhibitor is a NS 3/4 A protease inhibitor. In a preferred embodiment, the NS3/4A protease inhibitor is telaprevir.

[065] In one aspect, the invention provides a computer system wherein the viral dynamic model simulates rebound kinetics of a resistant HCV variant employing

Equations 6-7.

[066] In one aspect, the invention provides a computer system wherein the viral dynamic model is implemented employing normalizing a plasma viron values with baseline values.

[067] In one aspect, the invention provides a computer system wherein the viral dynamic model is implemented numerically employing Equations 9-13.

[068] In one aspect, the invention provides a computer system wherein the viral dynamic model calculates a mutation rate by exponentiating the rate of 1.2 x 10^"4 per nucleotide position per replication cycle for a single mutation by the number of nucleotide mutations. [069] In one aspect, the invention provides a computer system wherein the viral dynamic model employs a mutation rate (m) in a range of between 1.2 x 10^"5 changes/site/cycle and 1.2 x 10^"3 changes/site/cycle.

[070] In one aspect, the invention provides a computer system wherein the viral dynamic model is optimized with the maximum likelihood objective function defined with Equation (16).

[071] In one aspect, the invention provides a computer system wherein the optimization is solved as a two-step process consisting of integration and optimization steps, wherein the integration is performed by employing the staggered corrector sensitivity analysis method and the optimization is performed by employing the nonlinear programming/NLP approach.

[072] In one aspect, the invention provides a computer system wherein the dynamic viral model calculates an initial decline of HCV variant levels upon dosing of the protease inhibitor by employing Equation 17.

[073] In one aspect, the invention provides a computer system wherein the dynamic viral model calculates the fitness of variants relative to the fitness of the most-fit variant employing Equation (18) or (19).

[074] In one aspect, the invention provides a computer system wherein the dynamic viral model calculated the fitness of variants relative to the fitness of the most-fit variant at specific time points using a time-derivative of the HCV RNA levels.

[075] The model captures the emergence of resistant variants by increased availability of replication space vacated by wild-type HCV that rapidly disappeared when dosing started. The best-fit model highlighted the importance of modeling the

HCV population as an evolutionary process with competition among variants and mutations among them. The dependency of the estimation results to the replication space dynamics was examined and illustrates that the ratio of production rates of variants to wild-type were estimated more robustly than the absolute replication rates of the wild-type HCV.

[076] The model is useful for examining the dependency of the estimation results to the dynamics of the target cells and the significance of higher second-phase declines in patients dosed with telaprevir using this modeling framework. Dynamic models of evolution in HCV variants

[077] In vivo evolution of the HCV variants was modeled following Equations l^i-a. The model was implemented as deterministic differential equations, which assumes that the system may be represented as an average behavior.

V₄ = ∑ p mjjO - CjWj - cVi Vi ⁽3>

Variant V, represents a virion with characterized amino-acid substitution(s) and different sensitivities to telaprevir when measured in vitro. Variant V infects target cells T to form variant- /-infected cells /,■ at rate βTVi. It is assumed that each infected cell I₁ is infected by only one variant, and each variant competes for the same target cells T. Target cells T also represent limited replication "space" shared by all variants. Each I,- was infected by only one variant, and each variant competes for the same target cells T. Target cells T also represent limited "replication space" shared by all variants. Each I, produces a population of variants at production rate pϊ_h where a m,j - fraction of this production mutates to produce variant j. The m_{l X} were normalized to follow W_t1; + ∑μ i røi_j = 1.

[078] A direct antiviral reagent such as telaprevir reduces production rates of these variants by blockage factors (1 - €,^■). These resistant variants assumed different production rates pfj, but the same infection β and clearance c rates. This assumption is consistent with observed mutations in the NS3/4A protease regions of these variants whose function is to cleave polyprotein as a crucial step in the HCV replication cycle. The infected cell clearance rates δ assumed a linear function of telaprevir antiviral blockage described later, to be consistent with the observed higher second phase declines. The production rate ratio /• quantify variant / replication advantage (or disadvantage) in the absence of telaprevir. In the presence of telaprevir, variant / production rate is further reduced by (1-e,), with e,- a function of telaprevir concentration consistent with measurements in replicon cells. The competition among these variants of differing production and clearance rates constitutes the selection process of the HCV evolutionary dynamics.

[079] Each variant in the viral population represents a virion with specific nucleotide mutation(s) and amino-acid substitutions from wild-type in the NS3/4A protease region and possesses different sensitivity to telaprevir when measured in vitro in both enzymatic assay and replicon cells. The variants identified with high confidence by clonal sequencing that were present in the quasispecies either at > 5% at two measurement points or > 10% at one time point were examined. The number of variants per subject ranged from 2 to 6; the number of variants for each subject is provided in Table 1. These clonal sequencing results identified amino acid differences in HCV NS3^β4A that correlated with changes in telaprevir resistance in vitro. A larger network representation of quasispecies containing an even greater number of variants could have provided a more complete picture, but was not examined here because no in-subject kinetic data were available to estimate their fitness, and/or no in vitro data were available on their resistance to telaprevir. [080] Table 1 : Number of variants with fitness estimates for each subject.

[081] This list of variants corresponds to fit variants one or two nucleotide mutations away from wild-type. A larger network of quasispecies spanning a greater number of mutations can also be implemented in the multi- variant model. The list of variants in each patient was limited to detectable variants of that patient. This selection limits the variant members of a viral population to the most-fit variants closest to wild-type. [082] The mutation rates were pre-computed by assuming a rate of 1.17x1 CT⁴ per nucleotide position per replication cycle, a value obtained from previously studied rate of mutations in NS3/4A protease regions. The specific mutation rates between two variants were computed by exponentiating the value above by the number of nucleotide mutations between these variants. These rates were genotype specific. For example, to produce a mutation V36M, genotype 1 a requires a single mutation, while genotype Ib requires two mutations.

[083] Antiviral activities of telaprevir were implemented by assuming a dual role. Telaprevir blocks the production of HCV by inhibiting the activity of the NS3^»4A protease with blockage factors ε-, calculated using Equation 4a. The blockage factors for all variants within a subject were calculated using a single effective telaprevir concentration [TVR], with its value estimated from the HCV RNA, variant prevalence dynamics within each subject, and in-vitro susceptibility of variants to telaprevir. The susceptibility factor IC^o_j and hill coefficient h_\ were estimated from in-vitro susceptibility of variant / to telaprevir (in logarithmic scale of HCV RNA) and are provided in Table 2. The second role of telaprevir is to enhance infected cell clearance δ, a parameter contributing to the second-phase decline. WT S values were up to 10-times higher in subjects dosed with telaprevir than in subjects treated with Peg-IFN/RBV; only a 0.2-fold increase in the second-phase decline is explained by increased telaprevir blockage alone. Moreover, we examined the case if δ were not enhanced by telaprevir (<5|=0; <5₀. estimated). The results (Figure 5) suggest that enhanced δ was not observed in variants that were the least susceptible to telaprevir. These observations were incorporated into the model by assuming that δ_x- increased proportionally to the blockage factor S₁, given in Equation 4c.

Table 2: Variants susceptibility to telaprevir as measured in replicon cells.

aThe IC50 values reported here were obtained from the logarithmic- scale of replicon HCV RNA; the values may differ from that reported in Table 7 and in that were estimated from the linear-scale of replicon HCV RNA.

[084] Prior to dosing, the differential equations were initialized at steady-state. This steady-state initialization was chosen to be consistent with the fact that many of the patients have been infected by HCV for over 20 years — a timescale much longer than the dosing duration examined here. This steady-state solution was used to obtain the pre-dosing variant prevalence. During dosing, replication rates of variants were reduced by blockage factors constrained by in-vitro results. At post-dosing, these blockages were removed. Consequently, the WT and variants present would compete for available replication space with competitive advantages governed by no-drag fitness of WT and variants.

[085] The data was limited to only allow estimates of variant fitness, or the net difference between the viral replication and clearance rates, while the models require estimates of both production rate ratio f,- and infected-cell clearance rates δy for each variant. To resolve this degeneracy, the model assumes values of δ,- given below and only estimated f_/. The values of Bi assumed a linear function of telaprevir antiviral blockage, given by Equation 4b below,

S₄ = S₀ + eA Vi (4b)

where δ₀ is the no-drug infected-cell clearance rates, δ_t is the maximum additional infected-cell clearance rates when patients are dosed with telaprevir, and e,- is the telaprevir antiviral blockage to variant /. In most runs, δ₀ was assumed at 5 x 10^~3 IT¹. Two alternative cases were also studied: First, δ, followed a Heaviside step function of e,-, and second, δi = 0. [086] Alternatively, we used Equation 4c below,

δj = δ₀ + 6| (- log₁₀ (1 - GJ)) y_?- (_4c)

[087] The majority of the results were reported with representation of replication space T given by Equation 1. We also examined another T representation given by Equation 5. The values of T₀ and /_0,i were fixed prior to estimation to the steady-state values of_r and /_; obtained from models with Equation 1. To obtain a similar rate of T increase, γ values were related to parameters in Equation 1 by γ=s/(7o+∑, /o_j).

τ = 7(T₀ + X; /O,,- - Γ - ^; /O (5) i i

Parameter Estimation

[088] The estimation used both plasma HCV RNA and clonal sequencing data simultaneously. The estimation minimizes the maximum likelihood objective function. This simultaneous estimation allowed fitness estimation in subjects with HCV RNA levels below the clonal sequencing detection limit (100 IU/mL) at the end of telaprevir dosing but with detectable HCV RNA within a week after completion of dosing. Parameters estimated for each patient include c, δ_h p, fi and [tvr]. Parameter bounds are provided in Table 3. Fitness parameter _/j was estimated for each variant; the number of assessed variants for each subject varies between 2 to 6 (Table 1). The remaining parameters were pre-computed prior to estimation runs, which include m,-,^ β=0.05h^'], and J₀=O-OOSh^'1. Target-cell clearance d was constrained by s/d=T_max=l0. Parameters β and s/d ratio were fixed because our data only allowed estimates of the overall basic reproductive ratio R_OiWτ= pβT_max/(cδ). The fitness estimates are robust to the assumed β value (Figure 13). Susceptibility factors IC_50J and h_\, were fixed during each estimation. The model was used in cases when s was estimated simultaneously with other parameters described above, and when it was fixed to 1 rf '. Antiviral production blockage of telaprevir to variant i (1 - e,-) could also be constrained to follow the sensitivity to telaprevir measured in replicon cells.

Table 3: Bounds on the estimated parameters

[089] For the most fit resistant variant / to rebound, it must replicate faster than it is being cleared. In the context of Equation 1-3 when the rebound started, the following assumption that Σ, β7V,- « dT, 67V_/ « δ/,-, and /?(l - e, ]/_//,- » cV_;, was applied and thus, the rebound kinetics may be approximated by Equation 6-7

I ϊ ∞ 0Tr rebound V ^v i₁ (6)

where T_tebQuad is T when HCV RNA start rebounding. For fast T regeneration, r_rcbound = T_max. The overall viral replication rates VjIVj would simplify to p{\ - e,-)fβT_max. The model does not allow dissection of the contributions of each of these parameters to the overall replication rates because the measurements only observed HCV RNA but neither infected cells nor target cells. To resolve this degeneracy, β and T_max are fixed to values that are numerically easy.

Numerical implementation

[090] The simulations were implemented by normalizing the plasma virion value with the baseline values obtained after solving the steady-state initial condition. This normalization allowed better numerical conditioning and more accurate integration; clearance and replication rates, the balance of which is implicit in the baseline viral load, were estimated directly from HCV RNA decline and rebound. The simulation and estimation were implemented using Jacobian Software (R) (Numerica Technology, Inc.)- The dynamic estimation problem utilized the control vector parameterization approach. The Non-Linear Programming (NLP) subproblem was computed using a Successive Quadratic Program method implementing Large-scale Broyden-Fletcher- Goldfarb-Shanno (LBFGS). The gradient of the objective function was computed using the staggered corrector sensitivity analysis method.

[091] The dynamic simulation was implemented with integration relative and absolute tolerances of 10^~6. The optimization was implemented using multistart with random initial seeds, repeated 300 times. To check further the convergence to a global optimum, we required that the optimal parameter values converged with a criterion that depended on the objective function sensitivities to the parameters. [092] The model is Equations 1-3 were implemented numerically as the normalized form to baseline HCV RNA level (V₀), given as:

I = s^f - dt - ^T_j fftVi (H) β tVi - δiit (12)

Where The Vo-normalized model (Equations 11-13) demonstrated that parameters s, β are affected by V₀, but not parameters d, S₁-, p, β, c,-.

Expected increase in the second-phase decline by increased telaprevir blockage

[093] The second phase decline (A₂) is related to the clearance and blockage parameters according to the following equation:

A₂ = 0.5(c + S) - [(c - δ)² + 4(1 - e)cδ)f⁵ (14)

If two drugs exist with different blockage factors ε_A and ε_B but the same clearances c and δ, then the expected difference in X₁ is given by the following equation: λ^ - λf = [(c - δ)² + 4(1 - e_B)cδ)f⁵ - [(c - δf + 4(1 - e_A)cδ)f'^b

(15) Assuming typical clearance rate constants c and S for Peg-IFN and RBV (c=10 d^"!,

10^"0J (estimated blockage for WT with Peg-IFN/RBV therapy) to (l -ε)=10^{"2 12} (estimated blockage for WT with TVR monotherapy) would increase the second phase decline by only 0.04 d^{" 1}, or 0.2- fold. Because the observed second-phase decline in subjects dosed with telaprevir was 10-fold, mechanisms other than increased blockage are needed to explain the observed steeper decline.

Calculation of mutation rates

[094] Previously reported HCV mutation rates range from 1.5 x 10³ nucleotide changes/site/y to 5 x 10³ nucleotide changes/site/year. These values were converted to per nucleotide position per replication cycle by assuming an average length of the HCV replication cycle of 9.5 days calculated as (l/c+l/<S) with typical values for c and δ. These calculations resulted in a mutation rate (m) of 1.2 x 10^"4/site/cycle. [095] The mutation rates were computed prior to each estimation by assuming a rate of 1.2 x 10⁴ per nucleotide position per replication cycle. The specific mutation rates between two variants were computed by exponentiating the mutation rate for a single mutation by the number of nucleotide mutations between these variants. These rates were genotype specific. For example, to produce NS3-4A protease mutation V36M, genotype Ia requires a single nucleotide mutation (from codon GTG to ATG₁ for amino acid position 36), while genotype Ib requires two mutations (from GTT to ATG).

[096] To further examine the sensitivity of the estimation results to the assumed mutation rates, we repeated the estimations with 10-fold higher and 10-fold lower rates. The results are provided in Figure 11 and Table 4. The objective function values were the lowest for the case with m of 1.2 x 10^"4/cycle, suggesting that this rate produced the best correspondence between data and model. The ranking of estimated fitness was qualitatively similar in the three different m values explored, suggesting that the fitness were robust to the assumption of m values in the range of [ 1.2 x 10^"5, 1.2 x 10^"3] changes/site/cycle. The fitness estimates of the double mutant variants (V36M/R155K, V36M/T54S) were affected the most by m; lower m produced higher estimates of production rate ratio / for these variants. This relationship could be explained by the fact that lower m corresponded to lower pre-dosing levels of these double mutants, and to correspond to the levels measured at day 14, the estimation converged to faster replication rates (or higher/ values).

Table 4: Estimated fitness for three different mutation rates: base (1.2 ] 0-4/cycle), 10- fold higher (1.2 10-3/cycle), and 10-fold lower (1.2 10-5/cycle).

Maximum likelihood objective function definition

[097] The estimation problem solved an optimization problem with the maximum likelihood objective function defined below:

Φ ^■ M^ML^L(,p)

(16)

Parameter p is the set of time- invariant parameters to be estimated, N is the total number of measurements of all variables, N_v is the number of measured variables (N_μ=2, HCV RΝA and variant prevalence), N,,,_j is the number of measurements of variable j, ^J ^fc is the j-th observation of variable k and Z_jk(p) is the corresponding value computed from the model, a^ is the standard deviation of measurement ^zi^k . In the analysis, the variables estimated were the log_]0 of plasma HCV RΝA and variant prevalence (normalized to the scale of 0-10 to allow comparable weighting to the logi₀ of plasma HCV RNA).

Dynamic optimization setup

[098] The dynamic estimation problem utilized the control vector parameterization approach (Goh, et al., Automatica 1988;24:3), where the dynamic optimization was solved as a two-step process consisting of integration and optimization steps. The integration was performed by using the staggered corrector sensitivity analysis method (Feehery, et al., Applied Numerical Mathematics 1997;25:41). The optimization was solved using nonlinear programming/NLP approach; implemented using a Successive Quadratic Program method of Large-scale Broyden-Fletcher- Goldfarb-Shanno (LBFGS) (Zhu C, Byrd RH, Lu P, Nocedal J. L-BFGS-B: FORTRAN Subroutines for Large Scale Bound Constrained Optimization: EECS Department, Northwestern University; 1994). The dynamic simulation was implemented with integration relative and absolute tolerances of 10^"6. The estimation was implemented using multi-start with random initial seeds, repeated until best local optimum converged to a global optimum. Determination of global optimality

[099] The optimization step in the control vector parameterization was implemented as a nonlinear programming approach and will converge to local optimal solutions. Unfortunately, a rigorously global, nonlinear dynamic optimization remains difficult to implement. Here, we chose to perform multi-start from random initial seeds. For each subject, estimation runs from 300 random initial seeds were performed. We required that the lowest local optimal objective functions (within 10^"4 relative and absolute tolerance) were obtained for at least 5 times. Furthermore, for these runs with lowest objective values, the optimal parameter values were compared to determine if they converged to similar values. Because of the different objective value sensitivity of these parameters, the convergence criterion for the optimal parameter values were determined by their absolute differences in values, normalized by the sensitivity of the objective function to the parameters at the optimal solution. The multi-start optimization does not guarantee rigorous global optimality; however, the final optimum solution is likely to be the global optimum because of sufficient sampling of parameter space and the good correspondence with experimental data. Details on the normalization have been published previously (Adiwijaya, et al, Molecular BioSystems 2006; 2:650).

Relative contribution of factors leading to an initial decline of variant levels on dosing

[0100] Upon dosing with telaprevir, variant RNA levels were predicted to decline initially because of two major factors: blockage of replication by telaprevir and reduced influx mutations from WT due to rapid WT clearance. The reduction of variants' replication flux by telaprevir blockage can be approximated as -ε-_t fo I_t. The reduction of influx mutation by WT clearance can be approximated as m_WTii /_wτ /_wτ. Thus, assuming that prior to dosing V-_t/V_WT =

the ratio of reduced replication of variant / by blockage to that by mutation can be approximated by the following Equation:

Ratio =

α?)

= £ifiVi/(m_Wτ,ifwτVwτ)

[0101] We assumed a mean effective concentration [TVR] of 4 μM for all variants and calculated ε from Equation 5. For the V36M variant with intermediate susceptibility (IC₅₀=4.73μM, hill coefficient h = 3.5, /_V36M= 0.68), this ratio was 0.97. For the A156T variant with high resistance and low fitness (IC₅₀ = 1000 μM, h=\, f_AI56τ=0.1), the ratio was 4 x 10^"4. For the V36M/R155K variant with high resistance and high fitness (IC₅₀= 142 μM, h = 3.5,

this ratio is 4 x 10^"6. The fact that these ratios are all <1 suggests that the reduction in influx mutations from WT, rather than the increased telaprevir blockage, dominated the initial reduction of the variants' replication rates.

Comparison of fitness estimates from the modeling approach and from relative fitness method. [0102] Relative fitness (RF) were calculated for each pair of the three available clonal sequencing measurements, using the criterion to include variants described above: RF] ₂ ^used th^e first two measurements, and RF_2ι3 used the last two measurements. If a variant was undetectable in any of the measurement, its value was assumed to be at 5%, which is the precision limit of the clonal sequencing method. RF values were calculated using a formula proposed from the literature given in Equation 18.

where V₁ and W_T are the viral loads of variant i and WT, and tj, t₂ are time points when two consecutive measurements were taken. Using this definition, RF' > 0 implied that variant is more fit than wild-type.

[0103] The relative fitness were also calculated from the time derivative of the trajectories in the best-fit models according to the equation below.

[0104] This method is similar to relative fitness approach (Holland, JJ. , et al., J. Virol., 1991. 65: 2960; Goudsmit, J., et al, J. Virol., 1996. 70: 5662), normalized to 0-100 scale. The correspondence between these previously reported in-vivo fitness estimates, the current in-vivo fitness estimates presented herein, and the in-vitro fitness estimates reported by others are provided in Figure 10. Figure 10a shows the correspondence of the current in-vivo estimates to the in-vitro fitness estimates. With the exception of estimates for variant V36A (which appeared to be an outlier in the in- vitro estimates), all these fitness estimates were in good agreement. In contrast, Figure 10b shows correspondence of our previous in-vivo estimates with the in-vitro estimates: the in-vivo fitness estimates of many of the variants were higher than those of in-vitro fitness.

[0105] An example from a subject (Subject 2, Figure 7a and b, Table 5) demonstrated why / estimated from the current model corresponded better with all data than / estimated using the RF method. For this subject, the clonal sequencing data showed that although both WT and V36A comprised <5% of the population at Day 14, both were detectable at Day 21 , with V36A prevalence relative to that of WT increasing between Days 14 and 21. The current model predicted that at Day 14, when telaprevir dosing was discontinued, V36A RNA levels were about 2-log higher than those of WT. Thus, despite the reduced fitness of V36A over WT, V36A would continue to infect the majority of the target cells T between Days 14 and 21. However, by Day 150, WT dominated the quasispecies because of its higher fitness. If we computed RF based on prevalence data at Day 14 and 21 only (RF \_n2) for WT and V36A, assuming the same Day 14 prevalence levels of 5%, then RF_{l2ι V36A VS. WT} =0.337, a value >0 that misleadingly implies that V36A is more fit than WT. However, this conclusion is inconsistent with the RF calculated between Day 21 and Day 150 (RF₂^) of -0.101, a value <0 which implies that WT is more fit than V36A. On the other hand, the current modeling approach calculates /_V36A ⁼ 0.578, and correctly accounts for higher V36A levels than WT levels at Day 14, higher V36A levels at Day 21, and reduced levels of V36A at Day 150.

Table 5: Fitness estimates using relative fitness and production rate ratio (f) in Subject 2.

[0106] Based on the current modeling results, we also can calculate RF values at specific timepoints using a time-derivative of the HCV RNA levels (model-derived RF, Table 5). For the V36A variant, the Day 15 model-derived RF was much higher than the RF at Day 100 (-0.671 vs. 0.023), demonstrating dependency of RF values to the specific timing of sample collection. This example demonstrates the advantages of the modeling approach for estimating variants' fitness. Model predictions

[0107] The prediction of day 14 variant prevalence were obtained by re-estimating the model using data excluding day 14 clonal sequencing and used the optimal parameters to generate predictions. The uncertainty in the prediction was obtained from 100 random simulation runs, using a normal Iy- distributed noise model for each estimated parameter. The standard deviation of each estimated parameter was obtained from each optimal estimate using the Fisher Information Matrix.

[0108] A parameterized multi- variant viral dynamic model was developed to represent the antiviral response of subjects to telaprevir and to estimate the fitness of variants resistant to telaprevir. The dynamics of a population of evolving HCV genotype 1 variants detectable in patients dosed with telaprevir were represented as a two-step process consisting of random point mutation and selection by competition among variants of differing replicative rates for a shared limited "replication space." The superset of these variants and mutations among them are shown in Figure Ia for genotype Ia and in Figure 5 for genotype Ib. A schematic of the model is shown in Figure Ib and described by Equations 1-3 and 4a-c. Replicative fitness of variants was represented by their different production rate constants pf[. [0109] The multi-variant viral kinetic model was parameterized simultaneously from pre-, during-, and post-dosing plasma HCV RNA and post-dosing clonal sequencing data in patients dosed with telaprevir alone. Parameters estimated include rate constants for each variant production rate ratio to wild-type HCV f_h wild-type production p, plasma clearance c, infected-cell clearance in the presence of telaprevir δ_h effective telaprevir concentration [tvr], and in some cases, target-cell synthesis s. Other parameters were precomputed, including infected- cell clearance in the absence of telaprevir δ₀, and mutation rates m. The f,- represents the replicative advantage (or disadvantage) of variants in competition for limited replication space in the absence of telaprevir. In the presence of telaprevir, variant i production rate is further reduced by a factor (1 - e,), of which values were determined by effective telaprevir concentration and were constrained by in vitro measurements of variants sensitivity to telaprevir. The contributions of each model parameter to the observables, including pre-, during-, and post-dosing HCV RNA and post-dosing variant prevalence, are shown in Figure 6. [0110] The estimated fitness obtained from 26 subjects suggests reduced replicative capacity of all telaprevir-resistant variants analyzed compared to WT. Table 6 summarizes estimated production rate ratio (f) for all variants. The variants were sorted based on their resistance to telaprevir as measured in replicon cells. The first 7 variants (R155M to A156S) are low-level resistant variants (defined as variants with IC₅₀ ≤ the mean estimated effective telaprevir concentration in vivo when telaprevir is dosed orally at 750mg q8h). Among all variants, V36M and R155K single mutant variants with low-level telaprevir resistance have the highest/ values of 0.68 and 0.66, respectively; suggesting that all variants detected had average replications ranges from 1% to 68% of WT replication. Among the high-level resistant variants, the double mutant V36M/R155K had the highest / of 0.51. Comparing the list of variants for genotypes Ia and Ib, only V36A, T54A and A156T variants were shared between both subtypes. For all variants detected within a subject, we found that their /values were positively correlated to their catalytic rates k_cal (average per subject Pearson correlation coefficient was 0.746), and were negatively correlated to the logarithmic of their enzymatic inhibition constants /C₅₀ (average correlation coefficient was -0.509).

Table 6: Estimates of variants replication rate ratio to wild-type HCV (f) and corresponding predictions of their pre-dosing prevalence. Four more variants detected only for one subject were not shown.

[0111] After an initial sharp decline in HCV RNA levels during dosing with telaprevir, HCV RNA levels increased in some subjects. The model predicted this increase to be caused by pre-existing variants that had sufficient fitness. This growth was enabled by an increase in available replication space due to the rapid clearance of WT. Mutations during treatment contributed negligibly to the dynamics of HCV RNA increase. The fitness estimates revealed that in the absence of telaprevir, average variant replicative fitness ranges from 1% to 68 % of the wild-type NS3^#4A HCV (WT) fitness. This suggests that in the absence of telaprevir, all telaprevir-resistant variants observed here pre-existed and had reduced fitness compared to WT. [0112] For an HCV population to survive, its most fit variant must replicate faster than its elimination rate (i.e., the most-fit variant / basic reproductive ratio or Ro_1/= (1- ∑i≠itnii+∑i≠itnii f/fd pfiβs/(cδd) must be >1). From the average replication and clearance rate constants calculated with the model presented herein, the Ro_,wτ ^was 33 (with 95% confidence intervals of 10 and 50 and the corresponding f_c was 0.03 (95% confidence interval of [0.02, 0.10])). The estimated fitness of a few resistant variants present in several subjects, including R155M, A156T, and A156V, had R₀ values <1, suggesting that without WT mutations, these variants would not have survived independently even without treatment. The calculated pre-dosing prevalence of variants was near their respective influx mutation rates. Using the / values listed in Table 6, the corresponding pre-dosing prevalences are tabulated in the last column of this table. Because of the impaired replication of variants, the calculated prevalences of all variants were within 3.13-fold of the levels predicted from the respective influx mutation rates alone.

[0113] The corresponding pre-dosing prevalence of variants suggest values near their respective influx mutation rates. Using the / values listed in Table 7, the corresponding pre-dosing prevalence was tabulated in the last column of this Table.

The resulting prevalence suggested that the single mutant variants were near their i respective influx mutational rates from WT (within ^1~Λ -fold, or < 3.13-fold). For the double-mutant variants, their prevalence were up to 4-times higher than the values

/ "ZWT, i \ predicted from WT mutations only ^ W-i ', because of the significant contributions of influx mutations from single-mutant variants to their formation. For example, the average predosing fluxes of V36MR155K formation originated from its self- replication, double nucleotide mutations from WT, and single nucleotide changes from V36M were 51, 16, and 33%, respectively, highlighting the significant contribution of influx mutations from V36M.

Table 7: Summary of the estimation results of multiple model scenarios applied to data of Patient 1. E, estimated from data; P, pre-computed and fixed during estimations.

[0114] Comparing genotypes Ia and Ib, no statistically significant differences were observed in estimated parameter values common for both genotypes. Table 8 summarizes estimated parameters for the two subtypes. Comparing each parameter individually, including replication rate constants, clearance rate constants, and effective telaprevir concentration, no statistically significant differences between the two subtypes were observed. Comparing both subtypes, target cell synthesis rate s showed a borderline significant difference (P=0.03, Wilcoxon two-sided test); however, we were not confident that this difference was real because of the high number of values considered to be outliers. Variant V36A was the only variant for which fitness was estimated from more than two subjects in each of genotypes Ia and Ib. Comparing the two subtypes, /_V36A was either borderline significantly different or not significantly different depending on the assumed s values CP=0.04 when s was fixed, and P=O.11 when s was estimated).

Table 8: Estimated parameters broken down by genotypes.

Fixed target cell synthesis s : Estimated target cell synthesis s

Parameters Genotype Ia Genotype Ib P Genotype Ia Genotype Ib P n mean (std) n mean (std) n mean n mean

(std)^a (std) ^a viral 10 11 0.30 9 9 0.64 production p [d-¹] 4.1(2.9) 2.6(1.2) 4.6(2.4) 4.1(2.4) plasma virion 12 14 0.48 12 14 0.37 clearance c [d^"!] 12.0(4.6) 13.4(4.8) 11.3(4.6) 13.0(4.8) infected cell 12 14 0.80 12 14 0.95 clearance δ

2.4(1.0) 2.4(1.0) 3.1(1.2) 3.4(1.2) effective 12 14 0.41 12 14 0.76 telaprevir concentration [TVR] [μM] 5.1 (2.0) 4.4(2.3) 3.7(0.9) 3.6(1.2) target cell 12 14 n.a. 12 14 0.03* synthesis s 14.4(11.8 [d-¹] 24.0(n.a.) 24.0(n.a.) 4.8(8.9) ) relative 8 0.60 (0.18) 13 0.41 (0.19) 0.04 8 0.58 12 0.43 0.11 fitness fv36A (0.18) (0.21) ^a Mean and standard deviations were inter-subjeci statistic, excluding possible outliers.

[0115] An example estimate of a patient demonstrates that the best-fit model corresponded well to data. Results in Figure Ic and Figure Id show an example of a subject who received 450 mg telaprevir q8h for 2 weeks, and whose plasma HCV RNA levels rebounded on dosing. The correspondence between data and model for additional subjects is provided in Figure 7; for all subjects, the correspondence as reflected by the maximum likelihood objective values, are shown in Figure 8a. Assuming a pre-dosing steady-state, the best-fit model predicted that HCV with WT NS3^Φ4A protease dominated the HCV quasispecies population as the most fit variant in the absence of any drug, and variant prevalences were near the levels predicted from HCV mutation rates. At during-dosing, the plasma HCV RNA of both the data and the model corresponded well, with a biphasic decline kinetic for an HCV population consisting of mostly wild-type HCV (WT) for the first week and the rate of emergence by resistant variants during the second week. At post-dosing, the HCV RNA of both the data and the model returned back to pre-dosing level within two weeks. Moreover, the model agreed with the clonal sequencing data that V36MR155K variant that dominated (80%) the population at the end of dosing were outcompeted by WT and variants V36M and R155K — two mildly-resistant variants with higher /values. The data that V36M persisted for up to 200 days were captured by the model as follows: Immediate after dosing stoppage, V36M initially out- competed WT for available replication space because of its 10⁴-times higher prevalence. Consequently, the HCV population must wait for the much-slower V36M- infected-cell turnover before WT dominated the population again. At pre-dosing, the model predicted a WT prevalence of > 99.9%, consistent with other data not used here that the prevalence was > 95%.

[0116] The resulting fitness estimates demonstrated advantages of modeling approach compared to traditional fitness measures such as the relative fitness (RF) method. Figure Ic summarizes the fitness estimates for both RF and modeling methods. Two kinds of RF values were computed from the first two points of the clonal sequencing data (RF_{1 &} circles), and from the last two points (RF_2i3, squares). Some of RF₁₁₂ values were > 0, which misleading implied that some variants were more fit than WT, and were inconsistent with the RF₂₁₃ values that were all < 0. The modeling method estimated time- invariant production rate ratio / (triangles). The ranking of / values among variants correlated with the ranking of the RFl, 2 values. The f values obtained were all < 1 , suggesting that all variants were less fit than WT and were consistent with the RF2,3 values. Figure Id counted the number of variants estimated by these methods. The modeling method estimated a total of 96 variant-subjects, more than either counts of the RF values (40 for RF_{1 j2}, and xx for RF_2,3). The largest difference in the counts originated from subjects whose HCV RNA at end of dosing were below detectable limits of clonal sequencing (10³ IU/ml) but was still detectable (> 10 IU/ml). [0117] An example application to data of a subject demonstrates the qualitative differences between RF and modeling methods. Figure 7 summarizes data of Subject 1 and their correspondence to the best-fit model. Clonal sequencing data at day 21 for

this subject detected higher V36A than WT, and consequently, the

was 0.337 — a value > 0 that misleadingly implied that V36A was more fit than WT (because their prevalence were assumed the same at day 14, Methods). This conclusion contradicted to the later clonal sequencing data, which showed that WT

had outcompeted V36A to dominate the population (and

= -0.101). The modeling method estimated that V36A was less fit than WT (Zv_36A = 0.578). The model correctly estimated less fit V36A while also captured its higher prevalence at day 21 by predicting that V36A were up to 10³-times more than WT at day 14, allowing the earlier to infect more healthy target cells by day 21 despite its reduced

JlpVZQA fitness. In essence, the model suggested that ^1>2 > 0 was an artifact of the assumptions made for V36A and WT undetectable at day 14 by clonal sequencing, and that the modeling method that used the during-dosing data allowed us to predict more accurately the HCV RNA levels of these variants at day 14. [0118] The best-fit model to Subject 1 also highlighted different RF values when viral load was growing and when it was steady at the pre-dosing level. For example, at

ΓCΓ J_I E J ' day 15 when the viral load was still growing, the obtained from time- derivative of best-fit model trajectories was -0.671 — much less than the later

^, V3GΛ

^Λf d of -0.023 when the viral load had returned back to the pre-dosing level. This difference in RF values could be analyzed through the relationship between the RF and the modeling methods developed elsewhere: RF,- = (f,- ~ϊ)pT -(δ,- -δ_wτ). Comparing the growing viral load condition at day 15 to the constant load condition at day 100, the RF values changed (despite the time- invariant / values) because the

healthy target cells T decreased from its maximum of s/d to its limiting value of $^p . [0119] The during-dosing kinetics of HCV RNA rebound by resistant variants relied on increased "replication space" T as WT disappeared. WT disappeared rapidly at early during-dosing period, causing the mutations from WT to variants — a major component of variants replication fluxes — to be reduced. Resistant variants also declined initially because of reduced influx of new mutations from WT and, in variants with low-level resistance, because of blockage by telaprevir. For variants to emerge, this reduction in replication fluxes must be compensated by increased T. As the total viral load declined, replication space was predicted to increase. Such an increase, along with sufficient on-dosing fitness of variants, is necessary for emergence of variants. Had T remained constant, none of the variants would have rebounded because each variant would have had a reduced replication flux due to the reduced number of WT available to generate the variant. For the subject shown in Figure Ic and Figure Id, variants with a single mutation (V36M or R155K) or a double mutation (V36M/R155K) within their NS3^#4A protease were responsible for the increase in HCV RNA levels detected initially on Day 6. WT levels were predicted to increase again around Day 8 because of back mutations from variants. When telaprevir dosing was stopped, WT, V36M, and R155K variants out-competed the V36M/R155K variant, and WT eventually regained dominance of the HCV quasispecies population to reach a predicted level of >95% of the viral population in 300 days, although V36M persisted for up to 200 days in this subject. The model predicted that immediately after dosing was stopped, V36M initially out-competed WT for available replication space because it was relatively fit and it was 10⁴-times more prevalent. V36M persisted because infected-cell clearance was relatively slow. When T dynamic follows Equation 1, its values increased from a pre-dosing value of δc/^βp to a maximum (T_max) of sld, with a kinetic governed mostly by s. To separate out the kinetic of T from its magnitude, we constrained (-T_max) = sld to a constant. Because T was not directly observable, we examined three cases of T kinetic in the estimates: 1) target cell T followed Equation 1 with synthesis rate s estimated, 2) T followed Equation 1 with s fixed to its upper bound (I h^{" 1}), and 3) T followed Equation 5. For these three cases, the models corresponded well with observed data (Figure 2), suggesting robustness of the models to these assumptions of T dynamics. For T dynamics represented by Equation 1, increasing s implies faster dynamics for target cells to reach their maximum levels (T_maγ), resulting in an earlier occurrence of breakthrough. The objective values for the first two cases (s estimated and s fixed) for all subjects are shown in Figure 8a. Both cases of s produced similar objective values, suggesting robustness of the model fits to the ^ values in the range examined. The estimated s values (Figure 8b) had a logarithmic median of lO^"088 h^'\ a value comparable to the regeneration rate of liver tissues (10^~α3 - 10^~06 h^"'). The optimal estimates of Ro, _WT varied with s; higher s correlated to lower estimates of Rø_,wτ (Figure 8c). However, the estimates of the production rate ratio / were more robust to the s values (Figure 8d-f).

[0120] The contribution of T regeneration rates are demonstrated further by the estimates to data of Patient 1. The estimated parameters for both first (estimated s) and second (fixed s) cases applied to this patient are summarized in the first two columns of Table 7, and the correspondence between models and data are shown in Figure 2. Comparing the first to the second case for this patient, optimal s was slower at 0.030 (vs. 1.0 h^"1), and the objective function improved to 2.3 (vs. 3.0). This slower s was reflected in the slower kinetic for T to reach its maximum value and delayed rebounds of HCV RNA (Figure 2, solid vs. dashed lines). In the second case, lower estimates of p^βT^'max (0.045 vs. 0.15 hf¹) was obtained to compensate for its inability to delay rebound using s. Both cases had comparable fit to clonal sequencing data (Figure 2 solid vs. dashed lines) and their estimated / were comparable within a relative difference of 9.2%.

[0121] Examination of different representations of replication space T dynamics demonstrated robustness of the above estimates. When T followed Equation 5, T increased from a pre-dosing value of T₀ to its maximum (T_raax) of T_o+I_o with a rate given by a first-order rate constant γ. To allow comparable kinetics and magnitude of Tto those of Equation 1, we chose corresponding

parameter values:

^To+Jo . When the model with Equation 5 was used to estimate data of Patient 1, we obtained comparable results as those with Equation 1 (third column of Table 7 and dotted lines in Figure 2). The objective function improved to 1.85, with improved fit to the plasma HCV RNA data, and comparable fit to the clonal sequencing data. However, Equation 5 could not guarantee non-negativity of T and was not explored further.

[0122] Patients dosed with telaprevir have been observed to possess steeper second phase declines. What if, instead of attributing the steeper declines to telaprevir, we ascribed them to patients natural variability? We subsequently compared the results of the first case above, where telaprevir-enhanced infected-cell clearance rates (δ₍) was estimated while constraining the infected-cell clearance rates in the absence of drugs (δ₀ = 5.2 x 10^~3 hf¹), to a case when δ₀ was estimated while constraining δ₍ = 0. Because both cases had the same number of estimated parameters, their objective functions are good surrogates of the quality of fits. The results were summarized in Figure 5. When we compared the (δj = 0) case to the first (δ₍ ≠ 0) case, the objective functions worsened in 20/26 patients (Figure 5a), suggesting that the latter is more consistent with the data. To understand the underlying mechanisms behind Figure 5 the superiority of (δ_! ≠ 0) case, we examined further the best-fit models to data of Patient 1. The estimation results for (S₁ ≠ 0) case are given in the last column of Table 7 and Figure 5b and c. Compared to the (δj ≠ 0) case, models with δ] = 0 obtained a worse objective function in this patient (4.63 vs. 3.01), consistent with the general trends observed from all patients above. For (δj = 0) case, higher δ₀ (9.9 x 10^~2 vs. 5.2 x 10^"3 hf¹) and/ (0.942-0.989 vs. 0.535-0.787) were obtained to satisfy both the steep second phase declines observed by plasma HCV RNA between day 1 and 7 and the persistence of the variants observed up to 200 days post telaprevir dosing. To be consistent with the higher /, pre-dosing prevalence of single mutant variants were predicted to be much higher (10^~2 vs. 10^"4 fraction of WT), and the best-fit model could not capture plasma HCV RNA data as well, especially between day 1 and 10 (Figure 5b, solid vs. dotted black lines). This example further supports a model with faster infected-cell clearance in the presence of telaprevir.

[0123] To examine the importance to include mutations among variants, we compared the simulation results of best-fit model to Patient 1 above to those if we did not include mutations. The results are shown in Figure 4. Without mutations, WT were predicted at much lower level at the end of dosing because of the absence of back- mutations from valiants. Consequently, WT would take much longer to dominate the population again than the measured 50% WT at 212 days. If we re-estimated all parameters using a model without mutations, then the optimal estimates would choose a much slower δi and [tvr] to satisfy the 50% WT prevalence data at day 212 but the model compromized the fit to the first 10 days HCV RNA (data not shown). This example highlights the importance of modeling the data with a population of evolving variants including their cross-mutations.

[0124] To examine the contribution of mutations during treatment to the HCV RNA rebound dynamics, the model was modified by eliminating mutations during treatment among variants while assuming the same pre-dosing prevalence for all variants. The results are shown in Figure 3a. The dynamics of the resistant variants V36M, R155K, and V36M/R155K were similar with and without mutations during treatment, demonstrating that the emergence of resistant variants was attributed primarily to the on-dosing fitness of pre-existing variants and to the increase in replication space, with minimal contribution of mutations during treatment. In the absence of mutations during treatment, WT would continue to decline during telaprevir dosing instead of persisting because of reverse mutations from variants. When the mutation rates were increased by 10-fold and the same estimation and perturbation analysis was repeated, the contribution of mutations during treatment was still negligible with respect to the dynamics of variants (Figure 9). This suggests that when variants pre-exist, variants on-dosing fitness and increase in replication space — and not mutations during treatment — are the dominant factors contributing to HCV RNA rebound by variants. [0125] To examine the likelihood of these resistant variants pre-existing before dosing, the time necessary to generate these variants, if they did not pre-exist, was estimated. The best-fit model for Subject 1 above was reinitialized with an HCV population consisting only of WT up to 0.4 d before dosing started. The results are provided in Figure 3b. Starting with an all- WT population, the variants would rapidly increase to reach a level within 0.5-log of their steady-state prevalence within 0.2 d. Moreover, the predicted HCV RNA rebound on dosing would be delayed compared to the observed rebound. This delay reflects the contribution of variant-infected cells that have not reached steady-state by 0.4 d. The poorer (delayed) fit of this modified simulation compared to the one started with a steady-state level of valiants before dosing further highlights the likelihood that these variants pre-exist at a steady-state level. Because most subjects in this study have been infected with HCV for years, these analyses suggest that these variants exist prior to dosing.

[0126] To examine the contribution of end-of dosing clonal sequencing data to the fitness estimates in the modeling method, we re-estimated the model without these data. The results were provided in Figure 15. Figure 15a demonstrates that re- estimated / values without these data correlated closely with the ones obtained by including these data. Figure 15b compared the predicted day 14 variant prevalence from the re-estimated parameters to the data. The predictions correlated with the measured prevalence, with a correlation coefficient of 0.8518. As a comparison, the predictions obtained when day 14 data were included correlated with the data with a coefficient of 0.9383. These results demonstrated that day 14 clonal sequencing data contributed some to the estimation of fitness, but missing these data were not devastating to the estimates.

[0127] The HCV viral dynamics in subjects dosed with telaprevir were represented by a multi- variant model that included the heterogeneity of variants' fitness, and resistant profiles in the HCV quasispecies. During telaprevir dosing, the overall viral load initially declined as WT was inhibited and replication space available to variants increased, allowing pre-existing variants with sufficient on-dosing fitness to emerge. Unlike during HIV infection, where replication space can be quantified by measuring healthy CD4+ cells replication space in HCV-infected subjects cannot be measured directly. However, the concept of limited replication space is important in HCV infection because HCV RNA levels reach a steady-state value in chronically infected subjects, indicating limited resources for viral replication. Biologically, the replication space in HCV may be limited by the number of healthy hepatocytes, or by other factors necessary for viral replication within these cells (e.g., factors for RNA and/or protein synthesis, or for inhibition of the double- stranded RNA induced signaling pathway).

[0128] The increase in replication space and the on-dosing fitness of variants were the primary determinants of HCV RNA rebound during telaprevir dosing, with negligible contribution from mutations during treatment. The finding of variants prior to dosing and within a week on-treatment in other studies suggested that the variants contributing to virologic rebound in the present study were likely to pre-exist. The pre-existence of valiants is supported by the modeling results; had they not pre-existed, calculations based on the HCV mutation rate, replicative rate, and HCV RNA level at baseline indicated rapid (within hours) generation of these variants. A complementary computational analysis of HIV viral dynamics has also identified the likelihood of HIV variants pre-existing. Moreover, rapid virologic rebound in HIV-infected subjects treated with nonnucleoside reverse transcriptase inhibitors has also been attributed to the pre-existence of variants, similar to the observed viral dynamics in subjects studied here. A caveat here is that the risk of other (novel) resistant variants being generated on-treatment is higher if HCV is not cleared rapidly. Higher HCV RNA levels translate to more replication cycles and thus, larger risk of developing new variants. The higher risk further highlights the importance of treating HCV- infected subjects with potent regimens.

[0129] Based on data observed during the emergence of telaprevir resistant variants, the model described herein estimates a replication space synthesis rate s in subjects chronically infected with HCV that is higher than the s obtained for subjects chronically infected with HIV. This high s is consistent with the high regenerative rates of hepatocytes. Additionally, this high s may reflect the additional influx of healthy cells as a result of HCV RNA elimination in infected-cells (cured cells), consistent with the observed faster second-phase decline in subjects with HCV dosed with telaprevir as compared to the decline in subjects treated with IFN/RBV. [0130] All variants resistant to telaprevir estimated here have reduced replicative fitness in vivo compared to WT in the absence of telaprevir. The phenotypic and structural bases for the resistance and reduced fitness of these variants have been discussed elsewhere. The reduced fitness is consistent with 98% of HCV-infected subjects having quasispecies dominated by WT prior to dosing and with the finding that after telaprevir dosing was stopped, WT regained its dominance in the quasispecies. Resistant HIV variants also have reduced fitness. However, in contrast to treatments in HIV, treatments in HCV may result in a sustained viral response

(clearance of virus).

[0131] The fitness estimates herein included data of on-dosing HCV RNA and 3-7 month clonal sequencing. Current estimates suggest that variant V36M/R155K is less fit than valiant R155K, consistent with data that showed increased prevalence at later times (Day 21-23 vs. Day 14; Month 3-7 vs. Day 21-23) of WT, V36M and R155K as compared to the decreased prevalence of V36M/R155K.

[0132] No significant difference in estimated viral dynamic parameters was observed between genotypes Ia and Ib. In both genotypes, estimated parameters including replication rate, clearance rates, and fitness of the V36A variant were similar. This suggests that any difference in response between genotypes is mainly attributed to their respective genetic barriers, or the number of nucleotide changes needed from

WT to generate these variants.

[0133] The HCV RNA response in subjects dosed with telaprevir monotherapy and the estimate of variants' fitness have been quantified using a multi- variant viral dynamic model. Model perturbations suggest that because of high HCV mutation rates, variants exist prior to dosing.

[0134] The models were parameterized for pre-, during-, and post-dosing periods simultaneously. This simultaneous approach implied constrained relationships among the pre-dosing variants prevalence, the during-dosing HCV RNA declines and rebounds, and the post-dosing prevalence dynamics. Consequently, the simultaneous approach allowed us to use data from patients whose plasma HCV RNA at the end of dosing had fallen below the detection limit of the clonal sequencing method (10³

IU/ml) but were still detectable (10 IU/ml). Moreover, the modeling approach proposed here estimated more accurately variants fitness when compared to other methods.

[0135] The emergence of resistant variants depends on increased availability of

"replication space" T vacanted by wild-type virus upon telaprevir dosing. We found that slower T regeneration rates would delay the development of resistance. The requirement of increased T also suggests that we would observe emergence of resistant variants only when patients were dosed with a reagent that has high anti- viral blockage to eliminate WT rapidly. Moreover, the kinetics of T regeneration had some effects on the estimates of the absolute replication rate of WT, but much less effects on the estimates of the relative ratio of variant replication rates to WT. We also found that two different representations of T dynamics were interchangeable as long as they had comparable kinetics and magnitude.

[0136] The model highlights the significance of clonal sequencing data to the estimates. Upon completion of dosing, the plasma HCV RNA representing the ensemble of variants returned back to its pre-dosing level within two weeks, but the variant composition returned to pre-dosing distribution more slowly. Moreover, examination of sensitivities of clonal sequencing data to model parameters showed that the clonal sequencing data is most sensitive to relative replication rates /, further suggesting that without clonal sequencing data we would not obtain as good / estimates.

[0137] Previously, significantly higher HCV RNA second-phase declines have been observed in patients dosed with telaprevir. Here, we provided an additional support for this observation. Our computational experiments showed that, compared to the model proposed here, a competing model that did not include higher infected-cell clearance rates in the presence of telaprevir would be inferior in capturing observed data. Our data cannot distinguish alternative hypothesis of whether these higher declines are specific to protease inhibitor, or whether they are generally applicable to any reagents with high anti-viral blockage. Although we mostly examined a model where the increased in infected- cell clearance is a linear function of antiviral blockage, we would have obtained similar fits by using a Heaviside step function; Both functions would have similar maximum rate enhancement affecting WT and no effect to highly- resistant variants. The difference would apply only to mildly-resistant variants during- dosing, which contributed to the observables only in a limited time window after WT disappeared and before highly-resistant variants dominated.

[0138] Compared to models of resistance development in HIV, the model proposed here has many similarities in its structure and some subtle differences. Both models produced similar results of the pre-dosing prevalence calculation to be near the mutation rates of each variant (for sufficiently non-fit variants) and multiple phases as telaprevir dosing starts: an initial decrease in the valiant level because of reduced mutations from wild-type, followed by their eventual emergence as replication space increases until the limit of replication space. However, our data and model applied to HCV did not observe the damped oscillations as observed in HIV. We avoided the oscillations by using T regeneration rates higher than those in HIV, to be consistent with the known high regenerative capacity of hepatocytes. Moreover, we assigned different values of infected cell clearance rates in the presence and absence of telaprevir.

[0139] WT δ values were up to 10-times higher in subjects dosed with telaprevir than in subjects treated with Peg-IFN/RBV; much more rapid second-phase decline than the 0.2-fold increase in the decline predicted by increased telaprevir blockage alone. [0140] On the other hand, high-level resistant variants appeared not to observe this increased δ. If we fitted the data to a model without enhancement of δ (<S]= 0), then the model fit to the data was inferior (Figure 5a showed that objective values from cSι=O case is higher than objective values obtained from δy≠O case). Both of these observations provided limits on the assumed δ for variants with intermediate resistance. Given the data scarcity to estimate δ for variants with intermediate resistance, we chose to assume a simple linear function provided in Equation 5. [0141] Biologically, we hypothesized that the phenomenon may be related to the interference of HCV in the innate immune response. Elimination of WT HCV by telaprevir may allow previously WT-dominant infected cells to restore their normal innate immune response whilst variant levels are low, resulting in a more rapid second-phase decline. The reduction in telaprevir blockage for variants with reduced susceptibility to telaprevir may reduce the magnitude of the second-phase decline. [0142] The multi-variant viral kinetic modeling proposed here represents multiple aspects of the data coherently. First, the higher estimated / values of genotype Ia variants when compared to those of genotype Ib appears to explain the higher fraction of genotype Ia subjects rebounding when treated with telaprevir alone and the longer duration for genotype Ia HCV population to revert back to WT. Second, when compared to WT, the higher prevalence of low-level resistant variants detected at a week post-dosing is explained by the model by the higher prevalence of these variants and higher mutational rates from double mutants for genotype Ia subjects. Third, the modeling approach was able to capture different dynamics near to and far from the limit of replication space T. Thus, the above modeling approach allowed us to estimate fitness using more data not useable with the logarithmic ratio of fitness (LRF) method, including those from subjects who had rebounded (12/26 subjects), who had missing clonal sequencing at dosing day 14 because of the sequencing detection limit (7/26 subjects), whose WT was undetectable at day 14, and all the sequencing data (with the highest accuracy) taken between 3-6 months.

Telaprevir

[0143] VX-950 is described in PCT Publication Numbers WO 02/018369, WO 2006/050250 and WO/2008/144072, with reference to the following structural formula, or a pharmaceutically acceptable salt thereof:

(D

[0144] Other descriptions of VX-950 can be found in PCT Publication Numbers WO 07/098270 and WO 08/106151.

[0145] Accordingly, one embodiment of this invention provides a therapeutic regimen comprising administering to a patient VX-950 and a pharmaceutically acceptable carrier. The amount of VX-950 in these pharmaceutical compositions can be from about 100 mg to about 1500 mg, from about 300 mg to about 1500 mg, from about 300 mg to about 1250 mg, about 450 mg, about 750 mg, or about 1250 mg. Each of these pharmaceutical compositions can be administered, e.g., once, twice, or three times per day. Each of these compositions can be in one or more dosage forms (e.g., ampule, capsule, cream, emulsion, fluid, grain, drop, injection, suspension, tablet, powder). Each of these pharmaceutical compositions can be administered by one or more routes (e.g., orally, by infusion, by injection, topically, or parenterally) as considered appropriate by a skilled person in the art and depending on the dosage form.

[0146] Another aspect of this invention provides a method for treating or preventing HCV infections of a patient which includes administering to the patient VX-950. [0147] In some embodiments, the amount of VX-950 is at least about 300 mg (e.g., at least about 450 mg, at least about 500 mg, at least about 750 mg, at least about 1250 mg, or at least about 1500 mg). In some embodiments, the amount of VX-950 is no more than about 1500 mg (e.g., no more than about 1250 mg, no more than about 750 mg, no more than about 450 mg, no more than about 500 mg, or no more than about 300 mg).

[0148] It should be understood that these lower and upper amounts may be combined to provide preferred dose ranges for administering VX-950. For example, in some embodiments, VX-950 is administered in an amount from about 300 mg to about 1250 mg or from about 300 mg to about 1500 mg.

[0149] In some embodiments, VX-950 is administered in an amount of about 450 mg. In other embodiments, VX-950 is administered in an amount of about 500 mg. In other embodiments, VX-950 is administered in an amount of about 600 mg. In other embodiments, VX-950 is administered in an amount of about 750 mg. In other embodiments, VX-950 is administered in an amount of about 1000 mg. In other embodiments, VX-950 is administered in an amount of about 1250 mg. [0150] In any of these embodiments, the specified amount of VX-950 is administered once a day. Alternatively, the amount of VX-950 is administered twice a day (e.g., BID; ql2h). Alternatively, the amount of VX-950 is administered three times a day (e.g., TID; q8h). Further, VX-950 may be administered with or without food. [0151] As would be recognized, it is advantageous to have flexible dosing schedules. Accordingly, in another embodiment of this invention, the administration is 3 times per day, but not every 8 hours, optionally with meals. In certain embodiments, VX- 950 is administered with food.

[0152] This invention also provides a method for providing VX-950 to a patient in need thereof, which includes administering to the patient an oral dose of a composition comprising VX-950, wherein said dose provides to the patient an average plasma concentration (C_avg) of VX-950 of at least about 750 ng/mL after the administration. In some embodiments, the C_avg of VX-950 is about 1000 ng/mL or about 1250 ng/mL. In some embodiments, said dose essentially contains about 750 mg of VX-950. In some embodiments, the C_avg is obtained/ attained within 3 hours after administration, preferably 2 hours, more preferably 1 hour after administering. In a preferred form of these embodiments, the C_avg of VX-950 is maintained over about 24 hours, and preferably over 12 weeks.

[0153] In another aspect, this invention provides a method for treating HCV infection in a patient by administering at least one dosage form comprising VX-950 over a 24- hour period, wherein the dosage form is administered to maintain a trough plasma VX-950 level minimum of about 750 ng/ml over the 24-hour period. [0154] In certain forms of this embodiment, the dosage form is administered to maintain a trough plasma VX-950 level minimum of about 800 ng/mL, preferably about 900 ng/ml over the 24 hour period, and more preferably about 1000 ng/mL over the 24 hour period.

[0155] In certain preferred embodiments a therapeutically effective plasma concentration is obtained and a certain trough level is maintained. These methods are particularly useful for treating a human suffering from HCV infection by administering a VX-950 formulation, wherein the trough VX-950 plasma level is maintained at a minimum of about 750, 800, 900, or 1000 ng/mL over a 24 hour period. Without being bound by theory, trough levels of more than about 1500 ng/mL are thought to be not required by this invention. Accordingly, trough levels of about 750, 800, 900, 1000 ng/mL to about 1500 ng/mL (particularly 1000 to about 1500) are within the scope of this invention.

[0156] Also provided is a dosage form for delivering VX-950 to a human, wherein the dosage form comprises VX-950, said dosage form when administered at least once during a 24 hour period maintains a trough plasma VX-950 level that is at least about 750 ng/mL, 800 ng/mL, 900 ng/mL, or 1000 ng/mL over the 24 hour period to about 1500 ng/mL (particularly 1000 ng/mL to about 1500 ng/mL) over the 24 hour period. [0157] Ideally, when a method of this invention involves treating a patient infected with HCV, the method involves achieving, relatively rapidly, a therapeutically effective plasma concentration of VX-950 and then maintaining the trough level such that an effective therapeutic response is achieved. An effective therapeutic response is, preferably, one or both of a) achieving a sustained viral response; and b) achieving undetectable HCV RNA in the plasma by at least 12 weeks (12 weeks or more). As used herein, HCV RNA being "undetectable" means that the HCV RNA is present in less than 10 IU/mL as determined by assays currently commercially available, and preferably as determined by the Roche COBAS TaqMan™ HCV/HPS assay. [0158] The relatively rapid drop in plasma concentration may be obtained by administering a loading dose to a patient. In one embodiment, the loading dose is about 1250 mg of VX-950.

[0159] In certain dosage forms of this invention, the dosage form (other than the dosage form used to administer the loading dose) contains about 750 mg of VX-950 and the dosage form is administered once every 8 hours (i.e., q8h). [0160] In certain embodiments, the VX-950 dosage form is administered once every 8 hours.

[0161] In certain embodiments, the VX-950 dosage form is administered once every 12 hours.

[0162] In certain embodiments, the treatment duration with VX-950 is shorter than the current standard of care.

[0163] In certain embodiments, VX-950 is administered for less than about 12 weeks (or less than 12 weeks).

[0164] In certain embodiments, VX-950 is administered for about 8-12 weeks (or 8-12 weeks).

[0165] In certain embodiments, VX-950 is administered for about 10 weeks (or 10 weeks).

[0166] Modeling data indicate that administration with VX-950 may eradicate wild- type virus within 10 weeks.

[0167] In certain embodiments, VX-950 is administered for less than about 10 weeks. [0168] In certain embodiments, VX-950 is administered for about 2 weeks. [0169] Applicants have demonstrated that SVR was achieved in a patient receiving a 2 week treatment of VX-950.

[0170] In other embodiments, VX-950 is administered for less than about 8 weeks (or about 8 weeks or 8 weeks), less than about 6 weeks (or about 6 weeks or 6 weeks), or less than about 4 weeks (or about 4 weeks or 4 weeks).

[0171] In certain embodiments, a method according to this invention involves the treatment of a patient infected with genotype 1 Hepatitis C virus. Genotype 1 HCV infection is the most difficult strain of HCV to treat and the most prevalent strain in the United States.

[0172] Applicants have also demonstrated that administration of VX-950 decreases neopterin and ALT levels in vivo. AST (aspartate aminotransferase) levels were also decreased upon administration of VX-950. ALT is an enzyme that is present in liver cells; when liver cells are damaged or inflamed, ALT leaks from the cell into the blood. Blood ALT levels are useful as a marker of liver inflammation or damage. [0173] Neopterin (6-d-erythro-trihydroxypropylpteridine) is a pteridine derivative that is produced during the metabolism of guanosine triphosphate (GTP). Neopterin is produced primarily by monocytes and macrophages upon activation by interferon gamma or interferon alfa and is a marker of inflammation. Neopterin levels are frequently elevated in chronic HCV infection. The expected plasma level of neopterin in healthy individuals is between 3.1 and 7.7 nmol/1.

[0174] Accordingly, applicants determined the changes in serum neopterin concentration as a marker of monocyte/macrophage activity during administration of an inhibitor of (HCV) NS3^»4A protease. As described herein, VX-950 was administered for 14 days in a randomized, double blind, placebo controlled, multiple- dose study in 34 patients infected with HCV genotype 1. Patients received VX-950 450 mg q8h (n=10), 750 mg q8h (n=8), 1250 mg ql2h (n=10), or placebo (n=6). Serum neopterin concentrations were measured by a quantitative competitive ELISA

(ELItest Neopterin, Brahms, Hennigsdorf, Germany) at pretreatment, at day 7 and 14, and at day 10 of follow-up. The lower limit of detection (LLD) was 2 nmol/1. HCV RNA was assessed at frequent intervals during the study by real-time PCR (COBAS^® 1 8

TaqMan HCV Test; linear dynamic range of 3.0 x 10 to 2.0 x 10 HCV RNA IU/ml; LLD of 10 HCV RNA IU/ml; Roche Diagnostics, Branchburg, NJ). [0175] During administration of VX-950, every patient demonstrated at least 2-log₁₀ drop in viral load in all dose groups. In the 750 mg q8h dose group, mean HCV RNA dropped 3.6 logio at day 3, and 4.3 logio at day 14. In the 450 mg q8h and 1250 mg ql2h dose groups, maximal effect was seen at day 3 to day 7 followed by an increase in mean viral load between day 7 and day 14. Mean viral loads increased in all dose groups during follow-up. Advantageously, both HCV treatment naϊve and previously treated patients benefit from the methods of this invention. Both prior-treated patients and treatment naϊve patients responded to VX-950. For the avoidance of doubt, patients that may be treated according to the methods of this invention include those where HCV treatment has not been tried or has failed, including non-responding, rebound, relapse, and breakthrough patients.

[0176] Baseline neopterin was elevated in 23/34 patients (mean 9.33 nmol/L; upper limit of normal (ULN) 7.7 nmol/l). In the 750 mg dose group the decrease in neopterin compared to baseline and to placebo became significant at day 14 (750 mg q8h dose group baseline v day 14 10.48 ± 0.84 nmol/L v 7.32 ± 0.48 nmol/L P = 0.0104, Mann Whitney test; 750 mg q8h dose group v placebo day 14 7.32 ± 0.48 nmol/l v 9.81 ± 1.36 nmol/l P = 0.0036, unpaired two-tailed T test). Mean neopterin levels were within normal values at day 14 only in the 750 mg q8h dose group. In the 450 mg q8h dose group and the 1250 mg ql2h dose group, decreases in mean neopterin levels were smaller. Mean neopterin levels did not change in the placebo group. Mean neopterin levels increased in all dose groups during follow-up. [0177] The serum alanine aminotransferase (ALT) level can be measured using commercially available methods. Mean ALT levels, elevated at baseline, decreased during dosing in all groups. Mean ALT levels increased, returned toward baseline, in all dose groups during follow up.

[0178] Although HCV RNA increased in the 450 mg dose group and 1250 mg dose group after day 7, neopterin and especially ALT continued to decrease. Changes in mean neopterin concentration correlated with decline in HCV RNA and ALT levels during dosing of VX-950. Maximal decline in mean neopterin concentration was in the 750 mg q8h dose group at day 14. This was also the dose group with maximal reductions in HCV RNA at day 14. After day 7 in the 450 mg q8h and 1250 mg ql2h dose groups, ALT and neopterin levels decreased while HCV RNA levels increased.

These data suggest that inhibition of HCV replication by VX-950 results in a marked decline in systemic inflammatory activity associated with viral infection.

[0179] VX-950 also ameliorates elevated ALT levels in an animal model (see WO

2005/025517). Specifically, expression of WT-HCV protease-SEAP in SCID mice results in elevated ALT levels that can be ameliorated by treatment with VX-950.

Expression of WT-HCV protease alone in SCID mice also results in time and dose dependent elevation of ALT levels.

[0180] Accordingly, this invention provides a method for decreasing (including normalizing) ALT levels in a patient. The method includes administering to the patient in need thereof a therapeutically effective amount of VX-950 (e.g., about 1350 mg daily, about 2250 mg daily, or about 2500 mg daily). The patient can be infected with HCV or not infected with HCV.

[0181] In some embodiments, VX-950 is administered daily at about 450 mg or at about 750 mg every 8 hours, or at about 1250 mg every 12 hours.

[0182] Another aspect of this invention provides methods for treating or preventing one or more of liver damage, liver inflammation, steatosis, fatty liver, NAFLD, NASH, alcoholic steatosis, and Reye's syndrome in a patient that is either HCV positive or

HCV negative.

[0183] Also within the scope of this invention are methods for hepatoprotection in a patient that is either HCV positive or negative.

[0184] Applicants have also demonstrated that VX-950 blocks immune evasion in vitro.

[0185] VX-950 restores IFNβ dependent gene expression in Sendai virus infected

Huh7 cells. IFNβ promoter activity decreases in response to Sendai virus stimulation in the presence of WT HCVpro. VX-950 overcomes the WT HCVpro mediated suppression of IFNβ promoter activation.

[0186] Furthermore, NS3/4A is known to be involved in evasion of innate defenses, by e.g., TRIF-dependent mechanisms (as well as viral polyprotein processing). This immune evasion leads to viral persistence. Accordingly, a compound that inhibits both viral polyprotein processing and evasion of innate defenses is desirable.

Advantageously, VX-950 has been shown to do both. In particular, VX-950 inhibits in vitro cleavage of TRIF, which is a TLR3 adaptor protein.

[0187] Without being bound by theory, modeling suggests that VX-950 inhibits TRIF cleavage by NS3 protease. TRIF binds to non-prime side of the NS3 protease active site. VX-950 binds to the same non-prime side of the active site as TRIF and blocks

TRIF cleavage.

[0188] It has been shown that two VX-950 viral variants, A156T and A156V, show reduced ability to cleave either TRIF or 4A/4B. Because these viral variants are less fit, they are inefficient at both viral polyprotein processing and viral persistence.

Without being bound by theory, this is related to steric hindrance of A 156V affecting binding to 4A/4B and TRIF substrates.

[0189] This indicates that VX-950 acts as both a direct antiviral and as an inhibitor of immune evasion. Accordingly, this invention also provides methods of inhibiting

HCV protease mediated evasion of host defenses.

[0190] These results together with the in vivo data disclosed herein indicate the effectiveness of VX-950 as a monotherapy.

[0191] The amounts of VX-950 according to this invention are administered in a single dosage form or in more than one dosage form. If in separate dosage forms, each dosage form is administered about simultaneously. For the avoidance of doubt, for dosing regimens calling for dosing more than once a day, one or more pill or dose may be given at each time per day (e.g., 1 pill, three times per day or 3 pills, three times per day). Most embodiments of this invention will employ at least 2 pills per dose).

[0192] As would be realized by skilled practitioners, if a method of this invention is being used to treat a patient prophylactically, and that patient becomes infected with

Hepatitis C virus, the method may then treat the infection. Therefore, one embodiment of this invention provides methods for treating or preventing a Hepatitis

C infection in a patient. [0193] In addition to treating patients infected with Hepatitis C, the methods of this invention may be used to prevent a patient from becoming infected with Hepatitis C. Accordingly, one embodiment of this invention provides a method for preventing a Hepatitis C virus infection in a patient comprising administering to the patient a composition or dosage form according to this invention.

[0194] If pharmaceutically acceptable salts of compounds are utilized in these compositions, those salts are preferably derived from inorganic or organic acids and bases. Included among such acid salts are the following: acetate, adipate, alginate, aspartate, benzoate, benzene sulfonate, bisulfate, butyrate, citrate, camphorate, camphor sulfonate, cyclopentane-propionate, digluconate, dodecylsulfate, ethanesulfonate, fumarate, glucoheptanoate, glycerophosphate, hemisulfate, heptanoate, hexanoate, hydrochloride, hydrobromide, hydroiodide, 2- hydroxyethanesulfonate, lactate, maleate, methanesulfonate, 2-naphthalenesulfonate, nicotinate, oxalate, pamoate, pectinate, persulfate, 3 -phenyl-propionate, picrate, pivalate, propionate, succinate, tartrate, thiocyanate, tosylate and undecanoate. Base salts include ammonium salts, alkali metal salts, such as sodium and potassium salts, alkaline earth metal salts, such as calcium and magnesium salts, salts with organic bases, such as dicyclohexylamine salts, N-methyl-D-glucamine, and salts with amino acids such as arginine, lysine, and so forth.

[0195] Also, the basic nitrogen-containing groups may be quaternized with such agents as lower alkyl halides, such as methyl, ethyl, propyl, and butyl chloride, bromides and iodides; dialkyl sulfates, such as dimethyl, diethyl, dibutyl and diamyl sulfates, long chain halides such as decyl, lauryl, myristyl and stearyl chlorides, bromides and iodides, aralkyl halides, such as benzyl and phenethyl bromides and others. Water or oil-soluble or dispersible products are thereby obtained. [0196] The compounds utilized in the compositions and methods of this invention may also be modified by appending appropriate functionalities to enhance selective biological properties. Such modifications are known in the art and include those which increase biological penetration into a given biological system (e.g., blood, lymphatic system, central nervous system), increase oral availability, increase solubility to allow administration by injection, alter metabolism and alter rate of excretion.

[0197] Pharmaceutically acceptable earners that may be used in these compositions include, but are not limited to, ion exchangers, alumina, aluminum stearate, lecithin, serum proteins, such as human serum albumin, buffer substances such as phosphates, glycine, sorbic acid, potassium sorbate, partial glyceride mixtures of saturated vegetable fatty acids, water, salts or electrolytes, such as protamine sulfate, disodium hydrogen phosphate, potassium hydrogen phosphate, sodium chloride, zinc salts, colloidal silica, magnesium trisilicate, polyvinyl pyrrolidone, cellulose-based substances, polyethylene glycol, sodium carboxymethylcellulose, polyacrylates, waxes, polyethylene-polyoxypropylene-block polymers, polyethylene glycol and wool fat.

[0198] According to a preferred embodiment, the compositions of this invention are formulated for pharmaceutical administration to a mammal, particularly a human being.

[0199] Such pharmaceutical compositions of the present invention (as well as compositions for use in methods, combinations, kits, and packs of this inventions) may be administered orally, parenterally, sublingually, by inhalation spray, topically, rectally, nasally, buccally, vaginally or via an implanted reservoir. The term "parenteral" as used herein includes subcutaneous, intravenous, intramuscular, intraarticular, intra-synovial, intrasternal, intrathecal, intrahepatic, intralesional and intracranial injection or infusion techniques. Preferably, the compositions are administered orally or intravenously. More preferably, the compositions are administered orally.

[0200] Sterile injectable forms of the compositions of and according to this invention may be aqueous or oleaginous suspension. These suspensions may be formulated according to techniques known in the art using suitable dispersing or wetting agents and suspending agents. The sterile injectable preparation may also be a sterile injectable solution or suspension in a non-toxic parenterally acceptable diluent or solvent, for example as a solution in 1,3-butanediol. Among the acceptable vehicles and solvents that may be employed are water, Ringer's solution and isotonic sodium chloride solution. In addition, sterile, fixed oils are conventionally employed as a solvent or suspending medium. For this purpose, any bland fixed oil may be employed including synthetic mono- or di-glycerides. Fatty acids, such as oleic acid and its glyceride derivatives are useful in the preparation of injectables, as are natural pharmaceutically-acceptable oils, such as olive oil or castor oil, especially in their polyoxyethylated versions. These oil solutions or suspensions may also contain a long-chain alcohol diluent or dispersant, such as carboxymethyl cellulose or similar dispersing agents which are commonly used in the formulation of pharmaceutically acceptable dosage forms including emulsions and suspensions. Other commonly used surfactants, such as Tweens, Spans and other emulsifying agents or bioavailability enhancers which are commonly used in the manufacture of pharmaceutically acceptable solid, liquid, or other dosage forms may also be used for the purposes of formulation.

[0201] The pharmaceutical compositions of this invention may be orally administered in any orally acceptable dosage form including, but not limited to, capsules, tablets, pills, powders, granules, aqueous suspensions or solutions. In the case of tablets for oral use, carriers that are commonly used include lactose and corn starch. Lubricating agents, such as magnesium stearate, are also typically added. For oral administration in a capsule form, useful diluents include lactose and dried cornstarch. When aqueous suspensions are required for oral use, the active ingredient is combined with emulsifying and suspending agents. If desired, certain sweetening, flavoring or coloring agents may also be added. Acceptable liquid dosage forms include emulsions, solutions, suspensions, syrups, and elixirs.

[0202] Alternatively, the pharmaceutical compositions of this invention may be administered in the form of suppositories for rectal administration. These may be prepared by mixing the agent with a suitable non-irritating excipient which is solid at room temperature but liquid at rectal temperature and therefore will melt in the rectum to release the drug. Such materials include cocoa butter, beeswax and polyethylene glycols.

[0203] The pharmaceutical compositions of this invention may also be administered topically, especially when the target of treatment includes areas or organs readily accessible by topical application, including diseases of the eye, the skin, or the lower intestinal tract. Suitable topical formulations are readily prepared for each of these areas or organs.

[0204] As is recognized in the art, pharmaceutical compositions may also be administered in the form of liposomes.

[0205] Applicants have demonstrated that VX-950 is orally bioavailable. Accordingly, preferred pharmaceutical compositions of this invention are formulated for oral administration.

[0206] Administrations in connection with this invention can be used as a chronic or acute therapy. The amount of active ingredient that may be combined with the carrier materials to produce a single dosage form will vary depending upon the host treated and the particular mode of administration. A typical preparation will contain from about 5% to about 95% active compound (w/w). Preferably, such preparations contain from about 20% to about 80% active compound.

[0207] Upon improvement of a patient's condition, a maintenance dose of a compound, composition or combination of this invention may be administered, if necessary.

Subsequently, the dosage or frequency of administration, or both, may be reduced, as a function of the symptoms, to a level at which the improved condition is retained when the symptoms have been alleviated to the desired level, treatment should cease.

Patients may, however, require intermittent treatment on a long-term basis upon any recurrence of disease symptoms.

[0208] It should also be understood that a specific dosage and treatment regimen for any particular patient will depend upon a variety of factors, including the activity of the specific compound employed, the age, body weight, general health, sex, diet, time of administration, rate of excretion, drug combination, the judgment of the treating physician and the severity of the particular disease being treated, prior treatment history, co -morbidities or concomitant medications, baseline viral load, race, duration of diseases, status of liver function and degree of liver fibrosis/cirrhosis, and the goal of therapy (eliminating circulating virus per-transplant or viral eradication). The amount of active ingredients will also depend upon the particular described compound and the presence or absence and the nature of the additional anti-viral agent in the composition.

[0209] According to another embodiment, the invention provides a method for treating a patient infected with a virus characterized by a virally encoded NS 3/4 A serine protease that is necessary for the life cycle of the vims by administering to said patient a pharmaceutically acceptable composition of this invention. Preferably, the methods of this invention are used to treat a patient suffering from a HCV infection. Such treatment may completely eradicate the viral infection or reduce the severity thereof. Preferably, the patient is a mammal. More preferably, the patient is a human being. [0210] The dosages herein are preferably for use in vivo. Nevertheless, this is not intended as a limitation to using of these amounts of VX-950 for any purpose. In yet another embodiment the present invention provides a method of pre-treating a biological substance intended for administration to a patient comprising the step of contacting said biological substance with a pharmaceutically acceptable composition comprising a compound of this invention. Such biological substances include, but are not limited to, blood and components thereof such as plasma, platelets, subpopulations of blood cells and the like; organs such as kidney, liver, heart, lung, etc; sperm and ova; bone marrow and components thereof, and other fluids to be infused into a patient such as saline, dextrose, etc.

[0211] This invention also provides a process for preparing a composition comprising VX-950, or a pharmaceutically acceptable salt thereof, and a pharmaceutically acceptable carrier, adjuvant, or vehicle comprising the step of combining the VX-950, or the pharmaceutically acceptable salt thereof, and the pharmaceutically acceptable carrier, adjuvant, or vehicle, wherein the dosage of VX-950 in the composition is in accordance with any embodiment of this invention. An alternative embodiment of this invention provides a process wherein the composition comprises one or more additional agent as described herein.

[0212] Pharmaceutical compositions may also be prescribed to the patient in "patient packs" containing the whole course of treatment in a single package, usually a blister pack. Patient packs have an advantage over traditional prescriptions, where a pharmacist divides a patient's supply of a pharmaceutical from a bulk supply, in that the patient always has access to the package insert contained in the patient pack, normally missing in traditional prescriptions. The inclusion of a package insert has been shown to improve patient compliance with the physician's instructions. [0213] It will be understood that the administration of the combination of the invention by means of a single patient pack, or patient packs of each formulation, containing within a package insert instructing the patient to the correct use of the invention is a desirable additional feature of this invention.

[0214] According to a further aspect of the invention is a pack including VX-950 (in dosages according to this invention) and an information insert containing directions on the use of the combination of the invention. Any composition, dosage form, therapeutic regimen or other embodiment of this invention may be presented in a pharmaceutical pack. In an alternative embodiment of this invention, the pharmaceutical pack further comprises one or more of additional agent as described herein. The additional agent or agents may be provided in the same pack or in separate packs.

[0215] Another aspect of this involves a packaged kit for a patient to use in the treatment of HCV infection or in the prevention of HCV infection (or for use in another method of this invention), comprising: a single or a plurality of pharmaceutical formulation of each pharmaceutical component; a container housing the pharmaceutical formulation(s) during storage and prior to administration; and instructions for carrying out drug administration in a manner effective to treat or prevent HCV infection.

[0216] Accordingly, this invention provides kits for the simultaneous or sequential administration of a dose of VX-950 (and optionally an additional agent). Typically, such a kit will comprise, e.g. a composition of each compound and optional additional agent(s) in a pharmaceutically acceptable carrier (and in one or in a plurality of pharmaceutical formulations) and written instructions for the simultaneous or sequential administration.

[0217] In another embodiment, a packaged kit is provided that contains one or more dosage forms for self administration; a container means, preferably sealed, for housing the dosage forms during storage and prior to use; and instructions for a patient to carry out drug administration. The instructions will typically be written instructions on a package insert, a label, and/or on other components of the kit, and the dosage form or forms are as described herein. Each dosage form may be individually housed, as in a sheet of a metal foil-plastic laminate with each dosage form isolated from the others in individual cells or bubbles, or the dosage forms may be housed in a single container, as in a plastic bottle. The present kits will also typically include means for packaging the individual kit components, i.e., the dosage forms, the container means, and the written instructions for use. Such packaging means may take the form of a cardboard or paper box, a plastic or foil pouch, etc.

[0218] A kit according to this invention could embody any aspect of this invention such as any composition, dosage form, therapeutic regimen, or pharmaceutical pack. [0219] The packs and kits according to this invention optionally comprise a plurality of compositions or dosage forms. Accordingly, included within this invention would be packs and kits containing one composition or more than one composition. [0220] Although certain exemplary embodiments are depicted and described below, it will be appreciated that compounds of this invention can be prepared according to the methods described generally above using appropriate starting materials generally available to one of ordinary skill in the art.

[0221] In order that this invention be more fully understood, the following preparative and testing examples are set forth. These examples are for the purpose of illustration only and are not to be construed as limiting the scope of the invention in any way. [0222] VX-950 may be prepared in general by methods known to those skilled in the art (see, e.g., WO 02/18369). Any suitable formulations known in the art can be used in the invention. For example, formulations described in WO 2005/123075, WO 2007/109604, WO 2007/109605 and WO 2008/080167 can be employed in the invention. A specific formulation that can be used in the invention is exemplified in Example 6. Other specific examples include:

VX-950 49.5 wt%

HPMC 40 cp 49.5 wt %

SLS 1 wt % VX-950 49.5 wt%

HPC 49.5 wt % SLS 1 wt %

VX-950 49.5 wt% PVP K30 49.5 wt % SLS 1 wt %

VX-950 Solid Dispersion

% (w/w) Ingredient

49.5 VX-950 Spray-dried

49.5 PVP K29/32 from a MeCl₂ 1 SLS solution

wherein HPMC (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin- Etsu Metolose, HPMC60SH50), (Hypromellose Acetate Succinate, HG grade, Shin- Etsu Chemical Co.), HPC (hydroxypropyl cellulose), PVP (polyvinylpyrrolidone) and SLS (Sodium Lauryl Sulfate) are as described in WO 2005/123075. In certain embodiments, the solid dispersion shown above can be suspended in a 1% HPMC, 0.002% simethicone solution (1 wt% HPMC, 0.002 wt% simethicone and 99 wt% water). Additional examples include 1 :1 VX950: PVPK30, 1 wt% SLS (Refreshed Tox.); Niro-49 wt% HPMCAS/1 wt% SLS/1 wt% SDBS/ 49% VX-950; 40.5 wt% PVP-VA/10 wt% ETPGS/49.5 wt% VX-950; 40.5 wt% HPMC/10 wt% ETPGS/49.5 wt% VX-950; 49 wt% VX950, 49 wt% HPMCAS, 1 wt% SLS, 1 wt% SDBS; and 49 wt% VX950, 16 wt% HPPh, 33 wt% HPC, 1 wt% SLS, wt% SDBS, wherein PVPK30 (Polyvinyl Pyrrolidone K30), SDBS (sodium dodecyl benzene sulfonate), HPMCAS (Hydroxypropyl Methylcellulose Acetate Succinate), Vitamin ETPGS, PVP (polyvinylpyrrolidone) and SLS (Sodium Lauryl Sulfate), and details of the preparation of these formulations can be found in WO 2005/123075. Additional examples include those described in WO 2007/109604: a solid dispersion comprising 55 wt% VX-950, 24.4 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 19.6 wt% HPMC-60SH (Hydroxypropyl Methylcellulose 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 55 wt% VX-950, 14.7 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 29.3 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 60 wt% VX-950, 24.4 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 14.6 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 65 wt% VX-950, 17 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 17 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 70 wt% VX-950, 9.7 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 19.3 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 60 wt% VX-950, 39 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 49.5 wt% VX-950, 24.5 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 24.5 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 83 wt% VX-950, 8 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 8 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Laui-yl Sulfate (SLS); a solid dispersion comprising 49.5 wt% VX-950, 24.5 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 24.5 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 70 wt% VX-950, 14.5 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 14.5 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 65 wt% VX-950, 14.6 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 19.4 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 65 wt% VX-950, 9.7 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 24.3 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 60 wt% VX-950, 19.5 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 19.5 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 60 wt% VX-950, 14.6 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 24.4 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 70 wt% VX-950, 9.7 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 19.3 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 49.5 wt% VX-950, 24.5 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 24.5 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 83 wt% VX-950, 8 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 8 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 49.5 wt% VX-950, 49.5 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 83 wt% VX-950, 16 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 82.44 wt% VX-950, 15.89 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), and 1.67 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 49.5 wt% VX-950, 24.75 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 24.75 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS). a solid dispersion comprising 60 wt% VX-950, 24.6 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), 14.4 wt% HPMC-60SH (Hydroxypropyl Methylcellulose, 60SH 5OcP, Biddle Sawyer or Shin-Etsu Metolose, HPMC60SH50), and 1 wt% Sodium Lauryl Sulfate (SLS); a solid dispersion comprising 60 wt% VX-950, 39 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), and 1 wt% Sodium Lauryl Sulfate (SLS); and a solid dispersion comprising 49.5 wt% VX-950, 49.5 wt% HPMCAS-HG (Hydroxypropyl Methylcellulose Acetate Succinate, JPE, Biddle Sawyer or Shin-Etsu HPMCAS-HG grade), and 1 wt% Sodium Lauryl Sulfate (SLS).

Details of the preparation of these solid dispersions are described in WO 2007/109604.

Additional specific examples include tablet formulations containing a spray dried dispersion of VX-950, which are described in WO 2007/109604:

Additional specific examples include tablet formulations described in

WO2008/080167: VX95Q SD Tableting Experiment Design (Potency: 250 mg VX950)

Trial# A Formulation

Note: VX 950 SD Lot 02

Potency: 250 mg VX950

Trial# C Formulation

Note: VX 950 SD Lot 02

Potency: 250 mg VX950

Trial# F Formulation

Note: VX 950 SD Lot 02

Potency: 250 mg VX950 [0223] All cited documents are incorporated herein by reference. [0224] In order that this invention be more fully understood, the following preparative and testing examples are set forth. These examples are for the purpose of illustration only and are not to be construed as limiting the scope of the invention in any way.

Examples Example 1

[0225] Hepatitis C virus (HCV) genotype 1 variants resistant to protease inhibitors have been observed in clinical trials. The in vivo fitness of these variants was estimated in subjects dosed with monotherapy of an HCV protease inhibitor, telaprevir. [0226] The fitness of these valiants was estimated using a multi-variant viral dynamic model in which variants were selected by competition for shared limited replication space. Fitness was represented in the absence of telaprevir by different variant production rate constants and in the presence of telaprevir by additional antiviral blockage by telaprevir. Model parameters, including rate constants for viral production, clearance, and effective telaprevir concentration, were estimated from 1) plasma HCV RNA levels of subjects before, during, and after dosing, 2) post-dosing prevalence of plasma variants from subjects, and 3) sensitivity of variants to telaprevir in replicon cells.

[0227] The multi-variant viral kinetic model of HCV variants was used to measure data from patients dosed with TVR alone for 14 days. The model was parameterized using a scheme that integrated all available data including pre-, during-, and post- dosing plasma HCV RNA and post-dosing variant prevalence measured by clonal sequencing. Emergence of resistant variants were modeled by increased availability of healthy target cells as wild-type HCV rapidly disappeared when dosing started. [0228] Thirty-four subjects with HCV genotype 1 were enrolled in Study VX04-950- 101, a randomized, double-blind, placebo-controlled, 14-day, multi-dose, phase Ib trial. Subjects received placebo (n=6) or one of the following dosages of telaprevir administered as a suspension: 450 mg every 8 hours (n=10), 750 mg every 8 hours (n=8), or 1250 mg every 12 hours (n=10). Subjects' baseline characteristics were provided in Table 9. Variants were detected using clonal sequencing. For the model parameterization described here, data from 26 of the 28 subjects dosed with telaprevir were used. No variants were detected in one subject, and therefore this subject was excluded from further analysis. Estimation results in another subject with 8 variants did not converge to a global optimum - a standard requirement for computationally rigorous estimation; that subject was also excluded.

Table 9: Subjects characteristics³

a AU subjects were HTV negative

Table 10: Summary of the estimation results of multiple model scenarios applied to data of Subject L /, production rate ratio estimated from modeling method; RF_{\ i2}, relative fitness estimated from the first two of three clonal sequencing data points in each subject; RF₂₃, relative fitness estimated from the last two clonal sequencing data points.

Variants / RF 1.2 RF 2,3

Table 11. Descriptions of model variables and parameters.

Names Descriptions

(dot above) a variable time-derivative of a state variable

T healthy target cells, or replication 'space' s target cell synthesis rate d target cell degradation rate constant β infection rate constant

Vi or VJ plasma virion "i" or "j" with characterized amino-acid substitution(s) and different sensitivities to telaprevir in vitro h Vj-infected cells p production rate constant of wild-type (WT) tti_j,, mutation rates from V₁ to Vi f, ratio of production rates of a variant to WT c plasma virion clearance rate constant

Si production blockage factor of telaprevir to variant /

[TVR] effective telaprevir concentration for the observed inhibitions to WT and variants, deduced from the sensitivity curve measured in replicon cells

IC₅Oj IQo of variant / to telaprevir as measured in rep! icon cells hi HiU coefficient of exponentiation to represent the inhibition curve of variants by telaprevir as measured in replicon cells δo infected-cell clearance rate constant in subjects dosed with pegIFN and RBV δj additional infected-cell clearance rate constant in subjects dosed with telaprevir Table 12. Estimates of variants replication rate relative to wild-type HCV (f) and corresponding predictions of their pre-dosing prevalence.

Variants Geno NT ICsn n mean (SD) median (range)/ Pre-dosing type changes relative relative SD within prevalence (mean, from WT to WT production estimation [lower, upper]) rate /

R155M Ia 1 5.5 (L) 2 0.01 (n.a.) n.a. 1.2 [n.a.]- 10^"4

T54A Ia 1 6.3 (L) 15 0.55 (0.24) 0.JO [0.03-8.62] 2.6 [1.3,7O]-IO^"4

T54S Ia₁Ib 1 ND(L) 2 0.58 (0.04) I. II [1.01-1.21] 2.8 [n.a.]-l 0^"4

V36M Ia I 7.0(L) 12 0.68(0.16) 0.03 [0.01-2.44] 3.7 [2.0, 13] -10^'4

R155K Ia I 7.4(L) 12 0.66(0.17) 0.04 [0.01-2.79] 3.4 [1.8, 12] -10^"4

V36A Ia₇Ib 1 7.4(L) 20 0.49(0.21) 0.04 [0.02-7.01] 2.3 [1.4, 1O]-IO^"4

A156S Ib I 9.6(L) 2 0.17(0.07) 3.26 [0.09-6.43] 1.4 [J.2, 1.6] -10^"4

R155T Ia 1 19.8(H) 4 0.22(0.13) 0.06 [0.01-0.08] 1.5 [n.a.]- 10^""

V36M/R15 ia 2 «62(H) 9 0.51 (0.14) O.I 2 [0.04-0.66] 8.8 [4.0,4O]-IO^'7

5K

A156T Ia₁Ib 1 >62(H) 7 0.14(0.09) 0.06 [0.04-6.57] 1.4 [1.2, 1.6] -10^'4

A156V Ib J >62(H) 5 0.10(0.08) 0.04 [0.00-5.88] 1.3 [1.2, 1.5] -I0^'4

V36M/T54 Ia 2 ND(H) 2 0.31 (0.20) 0.40 [0.18-0.61] 4.1 [n.a.]-I0^'7 S

L= variant with low resistance to teiaprevir H= variant with high resistance to teiaprevir

Only variants detected for more than one subject are shown

IC₅₀ values were measured in replicon cells; (L) low-level resistant variants; (H) high- level resistant variants

Number of subjects from whose data variant fitness was estimated

Standard deviation for each fitness estimate computed in the neighborhood of optimal parameter values

95% confidence intervals were computed only for variants whose fitness was estimated from more than 5 subjects

In the estimation, the IC₅₀ of T54S was assumed to be the same as that of T54A

In the estimation, the IC₅₀ of V36M/T54S was assumed to be the same as that of V36M/R155K

Claims

What Is Claimed Is:

1. A method of modeling treatment of a hepatitis C infected patient with a protease inhibitor, comprising the step of: quantifying the patient's response to one or more dosing regimens of said protease inhibitor with a viral dynamic model employing one or more of Equations 1, 2, 3, 4(a), 4(b), 4(c) and 5.

2. The method of claim 1, wherein the viral dynamic model includes parameters for hepatitis C genotype Ia or Ib.

3. The method of claims 1 , wherein the protease inhibitor is a NS3/4A protease inhibitor.

4. The method of claim 3, wherein the NS3/4A protease inhibitor is telaprevir.

5. The method of any one of claims 1-4, wherein the viral dynamic model simulates rebound kinetics of a resistant HCV variant employing Equations 6-7.

6. The method of any one of claims 1-5, wherein the viral dynamic model employs normalizing a plasma viron value with a baseline value.

7. The method of any one of claims 1-6, wherein the viral dynamic model is implemented numerically by employing Equations 9-13.

8. The method of any one of claims 1 -7, wherein the viral dynamic model calculates a mutation rate by exponentiating the rate of 1.2 x 10^"4 per nucleotide position per replication cycle for a single mutation by the number of nucleotide mutations.

9. The method of any one of claims 1-8, wherein the viral dynamic model employs a mutation rate (m) in a range of between 1.2 x 10^~5 changes/site/cycle and 1.2 x 10^"3 changes/site/cycle.

10. The method of any one of claims 1-9, wherein the viral dynamic model is optimized with the maximum likelihood objective function defined with Equation (16).

11. The method of claim 10, wherein the optimization is solved as a two-step process consisting of integration and optimization steps, wherein the integration is performed by employing the staggered corrector sensitivity analysis method and the optimization is performed by employing the nonlinear programming/NLP approach.

12. The method of any one of claims 1-11, wherein the dynamic viral model calculates an initial decline of HCV variant levels upon dosing of the protease inhibitor by employing Equation 17.

13. The method of any one of claims 1-12, wherein the dynamic viral model calculates the fitness of variants relative to the fitness of the most-fit variant employing Equation (18) or (19).

14. The method of claim 13, wherein the dynamic viral model calculated the fitness of variants relative to the fitness of the most-fit variant at specific time points using a time-derivative of the HCV RNA levels.

15. A method of adjusting the dosing level of a composition comprising a protease inhibitor administered to a patient, the method comprising: measuring plasma hepatitis C RNA levels from a patient; utilizing the measured hepatitis C RNA levels in a multi- variant kinetic model to calculate the responsiveness of the patient to the administered composition comprising said protease inhibitor; and comparing the calculated responsiveness to a predetermined responsiveness to compositions comprising said protease inhibitor.

16. The method of claim 15 further comprising adjusting the dosing level of the composition comprising a protease inhibitor administered to a patient based upon the comparison of the calculated responsiveness to the predetermined responsiveness.

17. The method of claim 15, wherein the multi- variant kinetic model accounts for one or more of hepatitis C genotype 1 resistant variants selected from R155M, T54A, T54S, V36M, R155K, V36A, A156S, R155T, V36M/R155K, A156T, A156V, and V36M/T54S.

18. The method of claim 15, wherein utilizing the measured hepatitis C RNA levels in a multi- variant kinetic model to calculate the responsiveness of the patient to the administered composition comprising a protease inhibitor includes determining the fitness.

19. The method of claim 15, wherein the plasma hepatitis C RNA levels from a patient are measured within the first 20 days of administration.

20. The method of claim 15, wherein the measured hepatitis C RNA levels are utilized in the multi-variant kinetic model to calculate the initial responsiveness of the patient to the administered composition comprising a protease inhibitor.

21. The method of claim 15, wherein the initial responsiveness is compared to a predetermined responsiveness and based upon that comparison calculating a concentration of the protease inhibitor to be subsequently administered.

22. The method of claims 15-21, wherein the protease inhibitor is a NS3/4A protease inhibitor.

23. The method of claim 22, wherein the NS3/4A protease inhibitor is telaprevir.

24. The method of any one of claims 15-23, wherein the multi- variant kinetic model employs one or more of Equations 1, 2, 3, 4(a), 4(b), 4(c) and 5.

25. The method of claim 24, wherein the viral dynamic model simulates rebound kinetics of a resistant HCV variant employing Equations 6-7.

26. The method of any one of claims 24-25, wherein the viral dynamic model is implemented employing normalizing a plasma viron values with baseline values.

27. The method of any one of claims 24-26, wherein the viral dynamic model is implemented numerically employing Equations 9-13.

28. The method of any one of claims 24-27, wherein the viral dynamic model calculates a mutation rate by exponentiating the rate of 1.2 x 10^"4 per nucleotide position per replication cycle for a single mutation by the number of nucleotide mutations.

29. The method of any one of claims 24-28, wherein the viral dynamic model employs a mutation rate (m) in a range of between 1.2 x 10^~5 changes/site/cycle and 1.2 x 10^"3 changes/site/cycle.

30. The method of any one of claims 24-29, wherein the viral dynamic model is optimized with the maximum likelihood objective function defined with Equation (16).

31. The method of claim 30, wherein the optimization is solved as a two-step process consisting of integration and optimization steps, wherein the integration is performed by employing the staggered corrector sensitivity analysis method and the optimization is performed by employing the nonlinear programrning/NLP approach.

32. The method of any one of claims 24-31, wherein the dynamic viral model calculates an initial decline of HCV variant levels upon dosing of the protease inhibitor by employing Equation 17.

33. The method of any one of claims 24-32, wherein the dynamic viral model calculates the fitness of variants relative to the fitness of the most-fit variant employing Equation (18) or (19).

34. The method of claim 33, wherein the dynamic viral model calculated the fitness of variants relative to the fitness of the most-fit variant at specific time points using a time-derivative of the HCV RNA levels.

35. A computer system for modeling treatment of a hepatitis C infected patient with a protease inhibitor, comprising a computer-readable medium storing a computer program for quantifying a patient's response to one or more dosing regimens of said protease inhibitor with a viral dynamic model employing one or more of Equations 1, 2, 3, 4(a), 4(b), 4(c) and 5 to provide quantified patient's response to the dosing regimens.

36. The computer system of any one of claims 32-35, wherein the viral dynamic model includes parameters for hepatitis C genotype Ia or Ib.

37. The computer system of claims 32-36, wherein the protease inhibitor is a NS3/4A protease inhibitor.

38. The computer system of claim 37, wherein the NS3/4A protease inhibitor is telaprevir.

39 The computer system of any one of claims 32-38, wherein the viral dynamic model simulates rebound kinetics of a resistant HCV variant employing Equations 6-7.

40. The computer system of any one of claims 32-39, wherein the viral dynamic model is implemented employing normalizing a plasma viron values with baseline values.

41. The computer system of any one of claims 32-40, wherein the viral dynamic model is implemented numerically employing Equations 9-13.

42. The computer system of any one of claims 32-41, wherein the viral dynamic model calculates a mutation rate by exponentiating the rate of 1.2 x 10^"4 per nucleotide position per replication cycle for a single mutation by the number of nucleotide mutations.

43. The computer system of any one of claims 32-40, wherein the viral dynamic model employs a mutation rate (m) in a range of between 1.2 x 10^"5 changes/site/cycle and 1.2 x 10^"3 changes/site/cycle.

44. The computer system of any one of claims 32-40, wherein the viral dynamic model is optimized with the maximum likelihood objective function defined with Equation (16).

45. The computer system of claim 44, wherein the optimization is solved as a two- step process consisting of integration and optimization steps, wherein the integration is performed by employing the staggered corrector sensitivity analysis method and the optimization is performed by employing the nonlinear programming/NLP approach.

46. The computer system of any one of claims 32-45, wherein the dynamic viral model calculates an initial decline of HCV variant levels upon dosing of the protease inhibitor by employing Equation 17.

47. The computer system of any one of claims 32-46, wherein the dynamic viral model calculates the fitness of variants relative to the fitness of the most- fit variant employing Equation (18) or (19).

48. The computer system of claim 47, wherein the dynamic viral model calculated the fitness of variants relative to the fitness of the most-fit variant at specific time points using a time-derivative of the HCV RNA levels.