RESUMEN
Background: Exposure-response (E-R) models were developed for the primary endpoint of hospitalization or death in COVID-19 patients from the Phase 3 portion of the MOVe-OUT study (Clinicaltrials.gov NCT04577797). Beyond dose, these models can identify other determinants of response, highlight the relationship of virologic response with clinical outcomes, and provide a basis for differential efficacy across trials. Method(s): Logistic regression models were constructed using a multi-step process with influential covariates identified first using placebo arm data only. Subsequently the assessment of drug effect based on drug exposure was determined using placebo and molnupiravir (MOV) arm data. To validate the models, the rate of hospitalization/death was predicted for published studies of COVID-19 treatment. All work was performed using R Version 3.0 or later. Result(s): A total of 1313 participants were included in the E-R analysis, including subjects having received MOV (N=630) and placebo (N=683). Participants with missing baseline RNA or PK were excluded (79 from MOV and 16 from placebo arms). The covariates shown to be significant determinants of response were baseline viral load, baseline disease severity, age, weight, viral clade, and co-morbidities of active cancer and diabetes. Day 5 and Day 10 viral load were identified as strong on-treatment predictors of hospitalization/death, pointing to sustained high viral load as driving negative outcomes. Estimated AUC50 was 19900 nM*hr with bootstrapped 95% C.I. of (9270, 32700). In an external validation exercise based on baseline characteristics, the E-R model predicted the mean (95% CI) placebo hospitalization rates across trials of 9.3% (7.6%, 11.7%) for MOVe-OUT, 7.2% (5.3%, 9.8%) for the nirmatrelvir/ritonavir EPIC-HR trial, and 3.2% (1.9%, 5.5%) for generic MOV trials by Aurobindo and Hetero, consistent with the differing observed placebo rates in these trials. The relative reduction in hospitalization/death rate predicted with MOV treatment (relative to placebo) also varied with the above patient populations. Conclusion(s): Overall, the exposure-response results support the MOV dose of 800 mg Q12H for treatment of COVID-19. The results further support that many clinical characteristics impacted hospitalization rate beyond drug exposures which can vary widely across studies. These characteristics also influenced the magnitude of relative risk reduction achieved by MOV in the MOVe-OUT study.
RESUMEN
Background: Viral dynamics models provide mechanistic insights into viral disease and therapeutic interventions. A detailed, mechanistic model of COVID-19 was developed and fit to data from molnupiravir (MOV) trials to characterize the SARS-CoV-2 viral dynamics in MOV-treated and untreated participants and describe the basis for variation across individuals. Method(s): An Immune-Viral Dynamics Model (IVDM) incorporating mechanisms of viral infection, viral replication, and induced innate and adaptive immune response described the dynamics of viral load (VL) from pooled data from MOV Phase 2 and 3 trials (N=1958). Population approaches were incorporated to estimate variation across individuals and to conduct an extensive covariate analysis. Nineteen parameters in a system of five differential equations described SARS-CoV-2 viral dynamics in humans. Six population parameters were successfully informed through fitting to observed trial data while the remaining parameters were fixed based on literature values or calibrated via sensitivity analysis. Result(s): Final viral dynamics and immune response parameters were all estimated with high certainty and reasonable inter-individual variabilities. The model captured the viral load profiles across a wide range of subpopulations and predicted lymphocyte dynamics without using this data to inform the parameters, suggesting inferred immune response curves from this model were accurate. This mechanistic representation of COVID-19 disease indicated that the processes of cellular infection, viral production, and immune response are in a time-varying, non-equilibrium state throughout the course of infection. MOV mechanism of action was best described as an inhibitory process on the infectivity term with estimated AUC50 of 10.5 muM*hr. Covariates identified included baseline viral load on infectivity and age, baseline disease severity, viral clade, baseline viral load, and diabetes on immune response parameters. Greater variation was identified for immune parameters than viral kinetic parameters. Conclusion(s): These findings show that the variation in the human response (e.g., immune response) is more influential in COVID-19 disease than variations in the virus kinetics. The model indicates that immunocompromised patients (due to HIV, organ transplant, active cancer, immunosuppressive therapies) develop an immune response to SARS-CoV-2, albeit more slowly than in immunocompetent, and MOV is effective in further reducing viral loads in the immunocompromised.
RESUMEN
BACKGROUND: Exposure-response (E-R) analysis supported molnupiravir phase 3 dose selection based on viral load (VL) and mechanism of action (MOA) markers from phase 2.1 This analysis evaluated how well these biomarkers predict the E-R for hospitalization or death in phase 3. METHOD(S): The following E-R models were developed and compared: (1) logistic regression of the primary outcome (hospitalization or death) from phase 3, (2) VL change from baseline (CFB) from phase 2 and 3, and (3) low frequency nucleotide substitutions (LNS), a measure of MOA, from phase 2. Individual estimates of exposure were derived from population PK modeling of sparse samples collected in all patients. All work was performed using R v3.0 or later. RESULT(S): All E-R relationships were best represented by an Emax model with AUC50 estimates of 19,900, 10,260, and 4,390 nM*hr for hospitalization, day 5 VL CFB, and LNS mutation rate, respectively. Normalized E-R relationships were overlaid, illustrating consistency in E-R shape (Figure). Plasma NHC AUC0-12 was identified as the PK driver. Patients at 800 mg achieved near maximal response. CONCLUSION(S): E-R results support the dose of 800 mg Q12H for treatment of COVID-19. E-R relationships for MOA and virology biomarkers were consistent with the clinical E-R. (Figure Presented).
RESUMEN
BACKGROUND: The goal of this analysis was to investigate the relationship of molnupiravir pharmacokinetics (PK) and clinical outcomes (primary endpoint of hospitalization or death) in patients with COVID-19 in the phase 3 cohort of MOVe-OUT (clinicaltrials.gov NCT04577797).1 METHODS: Logistic regression models were constructed using a multi-step process with influential covariates identified first using placebo arm data only and subsequently assessment of drug effect as a function of exposures evaluated using placebo and MOV arm data. Individual estimates of exposure were derived from population PK modeling of sparse samples collected in all patients. All work was performed using R v3.0 or later. RESULT(S): A total of 1,313 participants were included in the exposure-response (E-R) analysis, including subjects on MOV (N = 630) and placebo (N = 683). Participants with missing PK or baseline RNA were excluded (79 from MOV and 16 from placebo arms). The covariates shown to be significant determinants of response were baseline viral load, baseline disease severity, age, weight, viral clade, active cancer, and diabetic risk factors. An additive AUC-based Emax model with a fixed hill coefficient of 1 best represented exposure-dependency in drug effect. Estimated AUC50 was 19,900 nM*hr with bootstrapped 95% confidence interval of (9,270, 32,700). Patients at 800 mg achieved near maximal response, which was larger than the response projected for 200 or 400 mg. CONCLUSION(S): Overall, the E-R results support the MOV dose of 800 mg Q12H for treatment of COVID-19. Many patient characteristics, beyond drug exposures, impacted the risk of hospitalization or death.
RESUMEN
Background. Molnupiravir (MOV) is an orally administered ribonucleoside prodrug of β-D-N4-hydroxycytidine (NHC) against SARS-CoV-2. Here we present viral dynamics analysis of Phase 2 clinical virology data to inform MOV Phase 3 study design and development strategy. Methods. An Immune-Viral Dynamics Model (IVDM) was developed with mechanisms of SARS-CoV-2 infection, replication, and induced immunity, which together describe the dynamics of viral load (VL) during disease progression. Longitudinal virology data from ferret studies (Cox, et al. Nat. Microbiol 2021:6-11) were used to inform IVDM, which was further translated to human by adjusting parameter values to capture clinical data from MOVe-IN/MOVe-OUT studies. Different placements of drug effects (on viral infectivity vs. productivity) and representations of immune response were explored to identify the best ones to describe data. A simplified 95% drug effect was implemented to represent a highly effective dose of MOV. Results. IVDM showed data were best described when MOV acts on viral infectivity, consistent with the error catastrophe mechanism of action. A cascade of innate and adaptive immune response and a basal level activation enabled durable immunity and continued viral decay after treatment end. IVDM reasonably describes VL and viral titer data from animals and humans. Influence of MOV start time was explored using simulations. Consistent with the ferret studies, simulations showed when treatment is started within the first week post infection, MOV reduces viral growth, resulting in a lower and shortened duration of detectable VL. When started later (e.g. >7 days since symptom onset), the magnitude of drug effect is substantially diminished in a typical patient with an effective immune response which reduces VL prior to treatment start. Further work is needed to model response in patients with longer term infection, where MOV drug effects may have more persistent utility. Conclusion. A COVID-19 IVDM developed using multiscale MOV virology data supports drug action on viral infectivity and importance of interplay of treatment and immune response and can describe infection time course and drug effect. IVDM provided mechanistic interpretations for VL drug effect in clinical studies.