Integrative Pharmacokinetic-Pharmacodynamic Modeling and Simulation of Amenamevir (ASP2151) for Treatment of Recurrent Genital Herpes
Abstract
Amenamevir is a novel antiviral agent that targets the viral helicase-primase complex. Although its dose-dependent efficacy has been demonstrated in non-clinical studies, a clear dose-response relationship has not been observed in humans. To address this discrepancy, a pharmacokinetic/pharmacodynamic (PK/PD) model was developed to explore the potential factors contributing to the observed differences and to better understand the immune-related healing process in human subjects. The model incorporated a non-linear kinetic representation of a virtual number of virus plaques as a built-in biomarker. Lesion score was selected as the endpoint for antiviral efficacy, and a logit model analysis was used to handle the ordered-categorical nature of this endpoint.
The modeling results suggested that changes in lesion scores over time could be explained by two key factors within the logit model: the change in the number of virus plaques to represent drug effect, and the time elapsed to represent immune-mediated healing. In humans, the pharmacodynamic response to amenamevir appeared to be largely dose-independent, suggesting that the immune response played a dominant role in lesion healing. These findings highlight the possibility that immune-driven healing can obscure the observable efficacy of antiviral drugs in clinical studies, particularly in conditions such as genital herpes. The proposed PK/PD model provides a valuable framework for understanding the pharmacodynamic behavior of antiviral agents whose effects may be masked by natural immune processes.
Keywords
amenamevir, pharmacokinetic/pharmacodynamic, pharmacometrics, genital herpes, modeling and simulation
Introduction
Genital herpes is primarily caused by herpes simplex virus type 2 (HSV-2), and less commonly by type 1 (HSV-1). It is one of the most widespread sexually transmitted infections, affecting nearly 50 million American adults and adolescents, which represents approximately 20% of the population. Among individuals infected with HSV-2, up to 80% experience recurrent outbreaks within the first year following the initial episode.
Amenamevir is a new class of antiviral drug that inhibits the viral helicase-primase complex, a crucial component for HSV DNA replication. This mechanism makes amenamevir a promising treatment option for genital herpes. The pharmacokinetic/pharmacodynamic approach for antiviral drug development parallels that of antibiotics, where in vitro IC50 values are often used to predict effective in vivo doses using animal models. In laboratory studies, amenamevir has shown superior antiviral potency against HSV-1 and HSV-2 compared to aciclovir.
A phase 2 clinical trial was conducted in the United States to evaluate the safety and efficacy of multiple amenamevir dose regimens in comparison with valaciclovir and placebo in the treatment of recurrent genital herpes. In this trial, patients initiated treatment at the first sign of recurrence with either amenamevir (100, 200, or 400 mg daily for 3 days or a single 1200 mg dose), placebo (3-day course), or valaciclovir (500 mg twice daily for 3 days). The dose regimens were selected based on prior pharmacokinetic data to maintain plasma concentrations of amenamevir above 200 ng/ml, referred to as T200.
Out of 695 randomized patients, data from 437 patients with confirmed recurrent infection were included in the efficacy analysis. The primary endpoint was the time to lesion healing. All active treatment groups, including the valaciclovir group, showed shorter times to lesion healing compared with placebo. However, a clear dose-response relationship was not observed. Apart from the 100 mg amenamevir group, all treatment arms showed a greater proportion of aborted lesions relative to placebo.
A population pharmacokinetic analysis was performed using 957 plasma samples collected from 273 participants in the clinical trial. The plasma concentration-time profiles of amenamevir were best described by a one-compartment model with first-order absorption. An exploratory analysis was undertaken to identify pharmacokinetic parameters correlated with clinical response, focusing on individual PK parameters derived from the final model. The variable T200 showed some correlation with time to lesion healing and viral shedding, consistent with results from animal models. However, these correlations were not statistically significant, likely due to the dose-dependency of T200 itself.
In the same clinical study, dose-dependent efficacy was not evident based on the primary endpoint, despite earlier animal studies suggesting such a relationship. To reconcile these conflicting findings between preclinical and clinical data, a unified PK/PD modeling approach was applied to both human and animal data. This model used the number of virus plaques as a biomarker to represent the antiviral action of amenamevir over time, and lesion score was used as the clinical efficacy endpoint. Additionally, the model incorporated an immune-related healing component to capture the contribution of natural recovery in the absence of drug effect.
Typically, new drug candidates are screened in animal studies to assess efficacy, safety, and pharmacokinetics before advancing to human trials. Amenamevir demonstrated a clear, dose-dependent antiviral effect in guinea pigs, but such a relationship was not replicated in humans. Understanding this inconsistency is a complex yet common issue in drug development. In this report, a bi-directional translational modeling approach was applied to bridge non-clinical and clinical findings using the same PK/PD framework for both species. This approach highlights the importance of integrating drug action and host immune responses when evaluating antiviral drug efficacy.
Materials and Methods
Analysis Data from Clinical and Non-Clinical Studies
This study involved a population pharmacokinetic and pharmacodynamic analysis using pharmacokinetic and pharmacodynamic data obtained from earlier clinical and non-clinical investigations. The PK/PD model developed in this research was structured based on prior observations suggesting a relationship between amenamevir concentration and viral activity. It was hypothesized that amenamevir exposure could influence the suppression of lesion scores. The model was constructed under the assumption that the progression in the number of virus plaques over time would be suppressed by amenamevir and that lesion score progression would be determined by a virtual count of virus plaques. The analysis re-used existing data from previous research as follows:
1. Virus plaque assay data were used to develop the virus plaque pharmacodynamic model.
2. Plasma concentration data of amenamevir from guinea pigs and humans were used to establish the pharmacokinetic model.
3. Lesion score data from both species were incorporated into a logit model to evaluate pharmacodynamic outcomes.
A summary of the experimental methods used in the earlier studies is described below.
Antiviral Compound and Sample Measurement
Amenamevir, a compound with a molecular weight of 482.55, was synthesized by Astellas Pharma Inc. Plasma concentrations of amenamevir were quantified using a validated liquid chromatography-tandem mass spectrometry method following FDA guidelines. The measurements were performed by Covance Laboratories, Ltd. The lower limit of quantification for amenamevir was 5 nanograms per milliliter when using 0.1 milliliter of plasma. Any concentrations below this limit were excluded from the model analysis and treated as zero.
Viruses and Cell Lines
The herpes simplex virus strains used in this study were clinically isolated in the United States and supplied by collaborators. Additional viral strains and cell lines were obtained from Rational Drug Design Laboratories. Human embryonic fibroblast cells and Vero cells were cultured in Eagle’s minimum essential medium supplemented with 10 percent fetal bovine serum, 100 units per milliliter penicillin G, and 100 micrograms per milliliter streptomycin. HSV-1 and HSV-2 strains were propagated in HEF cells maintained in medium containing 2 percent fetal bovine serum.
Virus Plaque Reduction Assay
The antiviral effect of amenamevir on HSV-1 and HSV-2 was assessed using a standard plaque reduction assay, allowing evaluation of both concentration- and time-dependent effects. HEF cells were grown in multiwell plates until a monolayer developed. After removal of the culture medium, the cells were infected with HSV-1 and incubated for one hour at 37 degrees Celsius. Following incubation, the cells were washed and treated with different concentrations of amenamevir. Treatment continued until plaques were clearly visible. Cells were fixed with 10 percent formalin in phosphate-buffered saline and stained with 0.02 percent crystal violet. The number of plaques was counted using a light microscope. Amenamevir concentrations tested were 0.01, 0.03, 0.1, 0.3, 1, 3, 10, and 30 micromolar. Incubation times were 6, 8, and 24 hours. This data was used to simulate time-course profiles of plaque formation in both guinea pigs and human subjects in conjunction with pharmacokinetic profiles.
Pharmacokinetic Study in Guinea Pigs
All animal procedures were carried out following the ethical guidelines of the Animal Ethical Committee of Yamanouchi Pharmaceutical Co., Ltd., currently part of Astellas Pharma Inc. Female Hartley guinea pigs were obtained from Charles River Laboratories in Kanagawa, Japan. Oral doses of amenamevir in methylcellulose suspension were administered at 0.3, 1.0, and 3.0 milligrams per kilogram. Three animals were included in each dosing group, all four weeks old at the time of dosing. Blood samples were taken at 0.25, 0.5, 1, 2, 4, 8, 12, and 24 hours after administration for plasma analysis.
Pharmacokinetic Study in Humans
This study used estimated population pharmacokinetic parameters for amenamevir based on plasma concentration data collected during screening and once-daily clinical visits from Days 1 to 4 in Study CL-101. These estimates were used for further pharmacokinetic modeling and simulations.
Lesion Score Measurement in Guinea Pigs
Antiviral activity of amenamevir was evaluated in guinea pigs by scoring lesion severity. Female Hartley guinea pigs, four weeks old, were infected intravaginally on Day 0 with HSV-2 strain G using a cotton swab soaked in phosphate-buffered saline containing virus. This strain had a viral load of 1.25 × 10⁵ plaque-forming units per milliliter and caused lesions in nearly all control animals. Amenamevir was administered orally at doses of 0 (placebo), 1, 3, 10, or 30 milligrams per kilogram twice daily for five days. The prophylactic regimen began three hours post-inoculation, while the therapeutic regimen started four days after inoculation. Each treatment group included ten animals. Disease symptoms were assessed daily over 21 days using a 0 to 6 composite score that evaluated severity of vaginitis and neurological symptoms:
Score 0: no signs of infection
Score 1: localized, barely visible vesicles
Score 2: vesicles involving 10 to 50 percent of the area
Score 3: vesicles involving 50 to 100 percent of the area
Score 4: ulcers involving 10 to 50 percent of the area
Score 5: severe ulcers involving 50 to 100 percent of the area
Score 6: hind limb paralysis or death
Lesion scores were recorded daily. For pharmacokinetic and pharmacodynamic modeling, lesion scores on Day 3 were used as the baseline, as no symptoms were observed before that day. The model assumed that the disease did not influence the pharmacokinetics of amenamevir.
Lesion Score Measurement in Humans
Time to healing of all lesions was the primary efficacy endpoint in Study CL-101. Herpes recurrence was defined as any recurrence below the umbilicus and above the knees. Healing time was calculated as the interval from therapy initiation to complete re-epithelialization of all lesions, excluding aborted lesions. Aborted lesions were defined by the presence of symptoms like pain, tingling, itching, and burning that did not progress beyond the macule or papule stage. Healed lesions were those with no crusts, erosions, depressions, or ulcerations. Residual erythema without preceding lesions was also considered healed. Lesion classification was conducted during clinical examinations on Days 1 through 6 and, if necessary, on Days 8 and 10. If symptoms were still present on Day 17, an additional examination was performed. Lesions were scored as follows for logit model analysis:
Score 0: healed lesion
Score 1: crust
Score 2: vesicle, pustule, or ulcer
Score 3: macule or papule
Lesion scores were assumed to follow a unidirectional pattern of 0, 3, 2, 1, and back to 0. Patients with persistent scores of 0 were classified as having aborted lesions. This clinical study was conducted in compliance with the Declaration of Helsinki, Good Clinical Practice standards, International Conference on Harmonization guidelines, and all applicable legal and regulatory requirements.
Analysis Models and Simulation Data
This section outlines the analysis conditions applied in the study.
Population Modeling
A logit model analysis was performed using a non-linear mixed effects modeling approach. The software used for modeling was NONMEM Version VI Level 1.0. Outputs from the modeling process were further processed graphically using SAS Version 8.2, Release 8.02. Model selection was based on evaluating goodness-of-fit, primarily considering differences in log-likelihood values derived from the objective function. A log-likelihood difference greater than 3.84 between models with one degree of freedom was considered statistically significant, corresponding to a p-value less than 0.05. The modeling process in NONMEM was carried out using the first-order conditional estimation method with interaction.
Model Validation
Validation of the final models developed for guinea pig pharmacokinetics and virus plaque dynamics was performed using a visual predictive check. Simulations were carried out generating one thousand hypothetical values for amenamevir concentrations and virus plaque counts at each time point. From these simulations, median values and 95 percent prediction intervals were derived and compared to the observed data for both amenamevir concentrations and virus plaque counts as a validation measure.
Pharmacokinetic Model in Guinea Pigs
A linear single-compartment model with an absorption lag time and first-order absorption was selected based on a visual inspection of observed data to describe the pharmacokinetics in guinea pigs. Key pharmacokinetic parameters defined in the model included the absorption rate constant (ka), the apparent distribution volume (V/F), and the apparent total clearance (CL/F), where F represents the oral bioavailability fraction. Mixed-effects modeling was conducted using NONMEM, and a log-normal distribution was assumed for inter-individual variability in each pharmacokinetic parameter.
The model assumes each parameter Pj in the jth subject follows a log-normal distribution:
Pj = θj × exp(ηj)
Here, θj denotes the population mean of the parameter, and ηj is a random variable representing inter-individual variability, assumed to be normally distributed with a mean of zero and a variance of ω².
Residual error in the model was accounted for using a proportional error model described as:
Yij = Cij × (1 + εij)
Yij represents the ith observed concentration in the jth subject, Cij is the model-predicted concentration, and εij is the residual error, assumed to be normally distributed with a mean of zero and a variance of σ².
Pharmacokinetic Model in Humans
A pharmacokinetic model for amenamevir in patients with genital herpes had been previously developed using a single-compartment model with first-order absorption. The model was further refined to account for extremely low concentration data points (LCPs) observed during the absorption phase by introducing an additional absorption rate constant, ka,LCP. Using NONMEM, the following mean parameter estimates and their respective relative standard errors (%RSE) were obtained: apparent clearance (CL/F) of 13.8 L/h (4.28%), apparent distribution volume (V/F) of 143 L (4.31%), absorption rate constant (ka) of 0.874 h⁻¹ (15.70%), and an absorption rate constant for LCP (ka,LCP) of 0.00107 h⁻¹ (63.30%).
Relative bioavailability values at different doses, assuming 100 mg as a reference with bioavailability set to 1.0, were estimated as follows: 0.982 (6.02%) for 200 mg, 0.874 (5.74%) for 400 mg, and 0.706 (6.15%) for 1200 mg.
Simulation of Pharmacokinetic Profiles in Guinea Pigs and Humans
Simulations of plasma concentration profiles for amenamevir were performed in both guinea pigs and humans. For guinea pigs, population mean PK parameters were used, while for humans, individual post-hoc estimates of PK parameters were utilized to simulate the time-course of virus plaque numbers. Data points identified as LCPs were excluded from these simulations, as the LCP modeling approach was specifically designed to explain such outlier values and provide more accurate estimation of ka using only the non-LCP data. A total of 33 LCP data points out of 928 (3.56%) were omitted, and this exclusion was considered to have negligible impact on the simulation results.
Dosing regimens used in the simulations were as follows:
Guinea pigs received 0, 1, 3, 10, or 30 mg/kg of amenamevir twice daily for 5 days.
Humans received 0, 100, 200, or 400 mg once daily for 3 days, or a single dose of 1200 mg.
Pharmacodynamic Model for Virus Plaques
A two-compartment model was developed to describe the concentration- and time-dependent dynamics of virus plaque formation. The model includes two compartments representing amenamevir-effective and amenamevir-ineffective virus activity. These compartments are linked by first-order rate constants kinact and kact. A virus replication cycle was integrated into the model framework.
The increase in the number of virus plaques was modeled using a fixed first-order rate constant kin of 0.0569 h⁻¹, calculated from ln(60)/72, reflecting an increase to 60 plaque-forming units (pFU) at 72 hours, based on study design. The drug effect was modeled using a non-linear Michaelis-Menten function with parameters Emax (maximum effect) and EC50 (drug concentration at half-maximal effect).
The dynamics of virus plaque numbers were described by the following differential equations:
dV(2)/dt = kin × V(1) − kinact × V(2) + kact × V(3) − (Emax × CP) / (EC50 + CP)
dV(3)/dt = kinact × V(2) − kact × V(3)
In this model, V(1) denotes the number of plaques in the input compartment, V(2) is the effective drug compartment, and V(3) is the ineffective drug compartment. The initial values were set at 60 for V(1) and 0 for both V(2) and V(3). CP represents the plasma concentration of amenamevir. No inter-individual variability was incorporated in this model. The residual variability was modeled using an additive error model defined as:
Yij = Cij + εij
Yij is the observed number of plaques, Cij is the model prediction, and εij is the residual error assumed to follow a normal distribution with a mean of zero and a variance of σ².
Simulation of Virus Plaque Profiles
Simulations were performed to generate virtual profiles of virus plaques in both guinea pigs and humans, using simulated amenamevir concentrations combined with virus kinetic parameters. The initial number of virus plaques was set to 60 pFU. Virus plaque administration was designated at Day 0 (0 hours), while amenamevir administration started on Day 1 (24 hours). For the human simulations, lesion scores recorded on Day 1 were used as initial values. This approach aligned with the design of a Phase 2 clinical study, where patients self-initiated treatment and returned to the clinic within 24 hours after their initial dose.
Pharmacokinetic/Pharmacodynamic (PK/PD) Model
Logit Model Analysis of Lesion Score
Logit model analysis was applied to ordered-categorical lesion score data. The lesion score, denoted as Yijk = Y(tijk), represents the lesion score at the kth time point tijk, in the ith individual of the jth treatment group. In the guinea pig study, lesion scores ranged from 0 to 4, while in the human study, scores ranged from 0 to 3. Higher scores (5 and 6) were not included for guinea pigs due to the absence of observed data for these values. The logit model expresses the cumulative probability of having a lesion score equal to or less than a given value yijk as a cumulative distribution function:
F(y; θ, η) = Pr{Y ≤ y; θ, η} = exp\[g(y; θ, η)] / (1 + exp\[g(y; θ, η)])
Here, θ represents a vector of parameters, and η is a random variable. The probability of observing a specific lesion score y is given by the difference in cumulative probabilities:
Pr{Y = y} = Pr{Y ≤ y} − Pr{Y ≤ y − 1}
with Pr{Y < 0} defined as zero and Pr{Y ≤ n} as one. The logit function g(yijk; θ, η) is defined as a sum over thresholds: g(y; θ, η) = Σ θm × Qm(y) + Eff where Qm(yijk) equals 1 if yijk is greater than or equal to m−1, and 0 otherwise. In guinea pigs, lesion scores increased almost monotonically in the placebo group, but began to decrease after Day 2 in the amenamevir-treated groups, showing clear dose dependence. In humans, lesion scores decreased steadily over time in both treatment and placebo groups, and dose dependence was less apparent. This suggested that lesion score reduction could not be explained solely by changes in virus plaques, which reflect amenamevir's antiviral effects. Therefore, a time component representing immune system healing was incorporated into the model in addition to the virus plaque component. The drug efficacy term Eff was modeled as the sum of virus plaque and time components: Eff = β1 × Virus Plaque + β2 × Time + η Here, Virus Plaque and Time are fixed effects representing the number of virus plaques and the elapsed time since virus plaque increase, respectively. This model was developed theoretically without conducting a covariate step. Maximum Likelihood Estimation was used for model fitting and posterior prediction. The individual estimated lesion score for subject i at time t was calculated as the weighted sum of probabilities for each lesion score: Lesion score\_i,t = Σ m × Pr\_i,t{Y = m} with m ranging from 0 to 4 for guinea pigs and 0 to 3 for humans. Results Pharmacokinetic Model in Guinea Pigs Model validation showed that most observed values fell within the 95% prediction interval. Population mean pharmacokinetic parameters for guinea pigs were estimated successfully. Pharmacodynamic Model for Virus Plaques Population mean parameters for the pharmacodynamic model describing virus plaque profiles were estimated. The estimated EC50 was 127 ng/mL, which is slightly lower than the in vitro expected effective concentration of 200 ng/mL, indicating consistency between in vitro antiviral activity and the virus kinetic model. Logit Model for Lesion Score Final logit models for lesion scores in guinea pigs and humans were defined as follows: In guinea pigs: Lesion score = 0: Logit = 6.76 + Eff Lesion score = 1: Logit = 9.87 + Eff Lesion score = 2: Logit = 14.04 + Eff Lesion score = 3: Logit = 19.93 + Eff where Eff = −0.265 × Virus Plaque − 0.0334 × Time In humans: Lesion score = 0: Logit = −5.21 + Eff Lesion score = 1: Logit = −3.35 + Eff Lesion score = 2: Logit = −0.01 + Eff where Eff = −0.0247 × Virus Plaque + 0.0424 × Time In these models, a decrease in lesion score represents a beneficial effect, while an increase indicates worsening. The virus plaque component had a negative effect on lesion scores in both species, indicating it contributed to improvement. However, the time component, representing immune system healing, had opposing effects: it was negative (worsening) in guinea pigs but positive (improving) in humans. In guinea pigs, both components contributed to worsening lesion scores, whereas in humans, the virus plaque component worsened scores but the immune response component improved them. Estimated mean lesion score profiles aligned well with observed data in both species. Additional simulations demonstrated the probability distributions of lesion scores over time and across doses, confirming dose- and time-dependent patterns. All model parameters were statistically significant, with 95% confidence intervals excluding zero. Discussion In this study, an empirical pharmacokinetic/pharmacodynamic (PK/PD) model for a helicase-primase inhibitor used in genital herpes patients was developed. The model indicated that the progression of lesion scores was not directly dependent on the pharmacokinetics of amenamevir but was primarily influenced by the virtual number of virus plaques. This finding suggests a time-dependent antiviral mechanism of amenamevir. Pharmacokinetic analysis of amenamevir concentrations in guinea pigs demonstrated a linear PK profile. However, previous human studies showed dose-dependent bioavailability, with bioavailability decreasing as the dose increased. This discrepancy may be due to differences in the dosage forms between species; amenamevir was administered as a methylcellulose solution to guinea pigs but in tablet form to humans. Given that amenamevir has poor solubility, which significantly impacts its passive diffusion, this difference in formulation may explain the observed species differences. When allometric scaling was applied, the clearance per bioavailability (CL/F) in humans was estimated to be lower than in guinea pigs, although the exact cause of this interspecies variation remains unclear. The current study concentrated on the relationship between amenamevir concentrations and virus plaque data. The virtual number of virus plaques, obtained through a plaque reduction assay, was used as a marker to explain the PK/PD mechanism of amenamevir. The model's goodness of fit for virus plaque data was not entirely satisfactory, especially at higher concentration exposures around 6 to 8 hours. Various alternative models that incorporated both time-dependent and dose-dependent components were tested to improve the fit, but none yielded significant improvement. The reason for this discrepancy at higher concentrations remains unresolved. Despite this, the current model is considered adequate for simulating the efficacy of twice-daily or once-daily dosing regimens because it provided acceptable results under 24-hour exposure conditions, which established a threshold concentration necessary to maintain amenamevir’s efficacy throughout the day. The virus plaque time profile in vivo was simulated based on a viral kinetic model derived from in vitro data. Several uncertainties remain, including the appropriateness of using the same viral kinetic model across species, whether plasma concentration serves as the best surrogate marker in different species, and species-specific differences in amenamevir's antiviral efficacy. Lesion score categories in humans were not predefined in the clinical study protocol; thus, four categories were originally defined in this analysis. In human patients, lesion scores showed a monotonic decline, starting from a higher score and decreasing as lesions healed. Since most patients healed without recurrence, this monotonic decrease suggests that the defined lesion scores are suitable for modeling purposes. Patients whose lesion scores remained zero throughout the study period, indicating aborted lesions, were excluded from the model analysis because it was not possible to determine whether the absence of symptoms was due to the drug's effect or other factors. Therefore, the PD model was developed solely based on patients who exhibited symptoms. Lesion scores in both guinea pigs and humans were explained using similar models with two fixed-effect parameters: the amount of virus plaques and elapsed time. Previous research demonstrated that continuous exposure to amenamevir above a certain concentration is essential to prevent viral replication. This was further confirmed in studies employing multiple dosing, a common approach in antibiotic research. During amenamevir’s clinical development, the effective concentration (EC50) value derived from non-clinical studies was directly applied to clinical dosing rationale. However, this non-clinical EC50 underestimated the clinical cure effect, and a clear dose-response relationship was not observed in clinical studies. In the present study, PK/PD models incorporating virus plaque and time components were developed for both species to explain lesion score time-course profiles. The inclusion of a virtual virus plaque kinetic profile allowed linking of amenamevir’s PK profile with lesion score changes, and the logistic model incorporating virus plaques and elapsed time effectively explained the dose- and time-dependent PD profiles. These findings suggest that the virtual number of virus plaques can serve as an integrated biomarker. While the fixed effect related to virus plaque efficacy was negative in both species, the time component reflecting immune system effects was negative in guinea pigs but positive in humans. HSV-2 can damage the central nervous system, which may impair the immune response. The current findings imply that the immune system might be weakened by viral infection in guinea pigs, although supporting evidence is limited. Differences in the time component between species may be attributed to varying experimental conditions and differential immune responses. In the guinea pig model, animals were infected with a lethal dose of HSV to ensure infection, resulting in increased lesion severity over time. Conversely, in humans, the immune system likely functions adequately to reduce lesion severity. The virus plaque component had a substantial effect in guinea pigs, as efficacy clearly depended on amenamevir dose while the placebo group showed saturation of effect. In humans, the virus plaque component’s effect was comparatively smaller, and drug effects on lesion scores were less pronounced. The pharmacodynamic effect in humans was nearly dose-independent, with immune system-mediated healing probably driving lesion score reduction. These results suggest that drug effects may be masked in diseases where healing is primarily mediated by the immune response, such as genital herpes. Thus, the proposed PK/PD model is particularly useful for explaining drug relationships in diseases that are self-resolving.
In traditional antibiotic and antiviral kinetic analyses, drug-pathogen interactions are often assumed to be independent of host species. In this study, viral kinetic parameters from in vitro data were applied to both guinea pigs and humans. However, obtaining viral kinetic data in humans is not feasible, and evaluating interspecies differences remains challenging.
When developing drugs for diseases that resolve naturally, the absence of natural healing in animal disease models may result in efficacy estimates that differ from clinical outcomes. From an efficacy prediction perspective in non-clinical studies, animal models without natural healing that clearly demonstrate drug potency may be suitable. However, these models risk misjudging clinical endpoints. Therefore, natural healing should be considered during animal model development.
Conclusions
This PK/PD modeling approach based on a bi-directional translational strategy is valuable not only for exploring new candidates in the non-clinical stage but also for analyzing clinical data. This modeling and simulation approach provides a unique method to connect non-clinical and clinical data during the development of drugs for herpes simplex virus infection.