Model-informed drug development of autologous CAR-T cell therapy: Strategies to optimize CAR-T cell exposure leveraging cell kinetic/dynamic modeling.

Autologous Chimeric antigen receptor (CAR-T) cell therapy has been highly successful in the treatment of aggressive hematological malignancies and is also being evaluated for the treatment of solid tumors as well as other therapeutic areas. A challenge, however, is that up to 60% of patients do not sustain a long-term response. Low CAR-T cell exposure has been suggested as an underlying factor for a poor prognosis. CAR-T cell therapy is a novel therapeutic modality with unique kinetic and dynamic properties. Importantly, "clear" dose-exposure relationships do not seem to exist for any of the currently approved CAR-T cell products. In other words, dose increases have not led to a commensurate increase in the measurable in vivo frequency of transferred CAR-T cells. Therefore, alternative approaches beyond dose titration are needed to optimize CAR-T cell exposure. In this paper, we provide examples of actionable variables - design elements in CAR-T cell discovery, development, and clinical practice, which can be modified to optimize autologous CAR-T cell exposure. Most of these actionable variables can be assessed throughout the various stages of discovery and development as part of a well-informed research and development program. Model-informed drug development approaches can enable such study and program design choices from discovery through to clinical practice and can be an important contributor to cell therapy effectiveness and efficiency.

Model-based prediction of progression-free survival in patients with first-line renal cell carcinoma using week 8 tumor size change from baseline.

PURPOSE: To assess the link between early tumor shrinkage (ETS) and progression-free survival (PFS) based on historical first-line metastatic renal cell carcinoma (mRCC) data. METHODS: Tumor size data from 921 patients with first-line mRCC who received interferon-alpha, sunitinib, sorafenib or axitinib in two Phase III studies were modeled. The relationship between model-based estimates of ETS at week 8 as well as the baseline prognostic factors and PFS was tested in multivariate log-logistic models. Model performance was evaluated using simulations of PFS distributions and hazard ratio (HR) across treatments for the two studies. In addition, an external validation was conducted using data from an independent Phase II RCC study. The relationship between expected HR of an investigational treatment vs. sunitinib and the differences in ETS was simulated. RESULTS: A model with a nonlinear ETS-PFS link was qualified to predict PFS distribution by ETS quartiles as well as to predict HRs of sunitinib vs. interferon-alpha and axitinib vs. sorafenib. The model also performed well in simulations of an independent study of axitinib (external validation). The simulations suggested that if a new investigational treatment could further reduce the week 8 ETS by 30 % compared with sunitinib, an expected HR [95 % predictive interval] of the new treatment vs. sunitinib would be 0.59 [0.46, 0.79]. CONCLUSION: A model has been developed that uses early changes in tumor size to predict the HR for PFS differences between treatment arms for first-line mRCC. Such a model may have utility in predicting the outcome of ongoing studies (e.g., as part of interim futility analyses), supporting early decision making and future study design for investigational agents in development for this indication.

Modeling and simulations relating overall survival to tumor growth inhibition in renal cell carcinoma patients.

PURPOSE: To assess the link between tumor growth inhibition (TGI) and overall survival (OS) based on historical renal cell carcinoma (RCC) data. To illustrate how simulations can help to identify TGI thresholds based on target OS benefit [i.e., hazard ratio (HR) compared with standard of care] to support new drug development in RCC. METHODS: Tumor size (TS) data were modeled from 2552 patients with first-line or refractory RCC who received temsirolimus, interferon, sunitinib, sorafenib or axitinib in 10 Phase II or Phase III studies. Three model-based TGI metrics estimates [early tumor shrinkage (ETS) at week 8, 10 or 12, time to tumor growth (TTG) and growth rate] as well as baseline prognostic factors were tested in multivariate lognormal models of OS. Model performance was evaluated by posterior predictive check of the OS distributions and hazard ratio across treatments. RESULTS: TTG was the best TGI metric to predict OS. However, week 8 ETS had a satisfactory performance and was employed in order to maximize clinical utilization. The week 8 ETS to OS model was then used to simulate clinically relevant ETS thresholds for future Phase II studies with investigational treatments. CONCLUSIONS: The published OS model and resultant simulations can be leveraged to support Phase II design and predict expected OS and HR (based on early observed TGI data obtained in Phase II or Phase III studies), thereby informing important mRCC development decisions, e.g., Go/No Go and dose regimen selection.

Pharmacokinetics, safety, and tolerability of voriconazole in immunocompromised children.

The pharmacokinetics of voriconazole in children receiving 4 mg/kg intravenously (i.v.) demonstrate substantially lower plasma exposures (as defined by area under the concentration-time curve [AUC]) than those in adults receiving the same therapeutic dosage. These differences in pharmacokinetics between children and adults limit accurate prediction of pediatric voriconazole exposure based on adult dosages. We therefore studied the pharmacokinetics and tolerability of higher dosages of an i.v.-to-oral regimen of voriconazole in immunocompromised children aged 2 to <12 years in two dosage cohorts for the prevention of invasive fungal infections. The first cohort received 4 mg/kg i.v. every 12 h (q12h), then 6 mg/kg i.v. q12h, and then 4 mg/kg orally (p.o.) q12h; the second received 6 mg/kg i.v. q12h, then 8 mg/kg i.v. q12h, and then 6 mg/kg p.o. q12h. The mean values for the AUC over the dosing interval (AUCτ) for 4 mg/kg and 6 mg/kg i.v. in cohort 1 were 11,827 and 22,914 ng.h/ml, respectively, whereas the mean AUCτ values for 6 mg/kg and 8 mg/kg i.v. in cohort 2 were 17,249 and 29,776 ng.h/ml, respectively. High interpatient variability was observed. The bioavailability of the oral formulation in children was approximately 65%. The safety profiles were similar in the two cohorts and age groups. The most common treatment-related adverse event was increased gamma glutamyl transpeptidase levels. There was no correlation between adverse events and voriconazole exposure. In summary, voriconazole was tolerated to a similar degree regardless of dosage and age; the mean plasma AUCτ for 8 mg/kg i.v. in children approached that for 4 mg/kg i.v. in adults, thus representing a rationally selected dosage for the pediatric population.

Assessing the efficiency of mixed effects modelling in quantifying metabolism based drug-drug interactions: using in vitro data as an aid to assess study power.

The clinical assessment of metabolic drug-drug interactions (mDDI) may involve population-based pharmacokinetic (POPPK) assessment as part of Phase 3 clinical trials. The elements of such POPPK study design have not been linked to prior information from in vitro experiments. Using in vitro-in vivo extrapolation techniques, implemented within Simcyp algorithms, the influence of POPPK study design (sample size, concentration-time data points, proportion of the population receiving a concomitant medication (COMED)) was studied in relation to the inhibitory potency of the each COMED. Steady-state concentrations of a candidate compound (compound X; mainly metabolized by cytochrome P450 enzymes, CYP3A4 and CYP2D6) in the presence and absence of COMEDs (including ketoconazole, fluconazole, quinidine and paroxetine as inhibitors) were analysed using non-linear mixed effect modelling (NONMEM). The NONMEM operator was blind to the nature of the COMEDs and the inhibitory effects on model parameters were classified as either statistically (p>0.01 for a change in objective function) or kinetically (COMED effect>2 fold) significant. Using a population study size of 2000, no false-positive cases were identified and, except in one case, no false-negative interaction was observed when >2.5% of patients had received an interacting COMED. The findings increase the confidence in the ability of the mixed effects modelling approach to identify 'true' interactions. However, they also emphasize the importance of study design and the potential value of using pre-clinical information from in vitro studies. Recent US Food and Drug Administration guidance on mDDI has put more emphasis on the use of in vitro systems for detecting and anticipating such effects. Combining these data with the framework of non-linear mixed effect modelling seems a natural progression in the field of assessing mDDI.

Population pharmacokinetic analysis of voriconazole plasma concentration data from pediatric studies.

Voriconazole is a potent triazole with broad-spectrum antifungal activity against clinically significant and emerging pathogens. The present population pharmacokinetic analysis evaluated voriconazole plasma concentration-time data from three studies of pediatric patients of 2 to <12 years of age, incorporating a range of single or multiple intravenous (i.v.) and/or oral (p.o.) doses. An appropriate pharmacokinetic model for this patient population was created using the nonlinear mixed-effect modeling approach. The final model described voriconazole elimination by a Michaelis-Menten process and distribution by a two-compartment model. It also incorporated a statistically significant (P < 0.001) influence of the CYP2C19 genotype and of the alanine aminotransferase level on clearance. The model was used in a number of deterministic simulations (based on various fixed, mg/kg of body weight, and individually adjusted doses) aimed at finding suitable i.v. and p.o. voriconazole dosing regimens for pediatric patients. As a result, 7 mg/kg twice a day (BID) i.v. or 200 mg BID p.o., irrespective of body weight, was recommended for this patient population. At these doses, the pediatric area-under-the-curve (AUC) distribution exhibited the least overall difference from the adult AUC distribution (at dose levels used in clinical practice). Loading doses or individual dosage adjustments according to baseline covariates are not considered necessary in administering voriconazole to children.

A non-parametric method to analyse time-course of effect in the absence of pharmacokinetic data: application to inhaled bronchodilators.

In spite of the extensive use of long-acting beta(2)-agonist (LABA) bronchodilators in asthma, the actual mechanism of their in vivo duration of action is not well understood, primarily due to limitations of standard pharmacokinetic-pharmacodynamic (PKPD) analysis methodologies. We have developed a novel method of analysing lung efficacy vs. time profiles for LABAs that can be used to provide comparative information on the lung PK. We hypothesised that for compounds that do not differ in their PK at the site of PD action, but differ in their in vivo potencies, the relationship between the area under the effect curve (AUEC) and the observed maximum effect (OME) at different doses is described by the same sigmoid curve. We have illustrated this property for standard PKPD models by obtaining analytical solution and through simulations. Anaesthetised dog in vivo effect vs. time profiles were gathered for six inhaled LABA candidates that differ in their in vitro potencies. Neither lung nor systemic PK was available for any compound. Analysis of the AUEC vs. OME data, derived from the efficacy profiles, using nonlinear mixed effects modelling indicated that for four compounds, the observed differences in in vivo duration of action was due to differences in their in vivo potencies and not because of lung PK differences. Therefore, it was concluded that for these compounds, characterisation of lung PK was unlikely to differentiate their PKPD characteristics. Thus, the proposed approach helped focus resources during translational research leading to lead candidate selection.

A pharmacokinetic-pharmacodynamic model for the quantitative prediction of dofetilide clinical QT prolongation from human ether-a-go-go-related gene current inhibition data.

BACKGROUND: QT prolongation is an important biomarker of the arrhythmia torsades de pointes and appears to be related mainly to blockade of delayed inward cardiac rectifier potassium currents. The aim of this study was to quantify the relationship between in vitro human ether-a-go-go-related gene (hERG) potassium channel blockade and the magnitude of QT prolongation in humans for the class III antiarrhythmic dofetilide. METHODS: The in vitro affinity and activity of dofetilide were determined in recombinant cell cultures expressing the hERG channel, and the QT-prolonging effect of dofetilide was assessed in 5 clinical studies (80 healthy volunteers and 17 patients with ischemic heart disease). A population pharmacokinetic-pharmacodynamic analysis of the in vitro and in vivo data was performed in NONMEM by use of the operational model of pharmacologic agonism to estimate the efficiency of transduction from ion channel binding to Fridericia-corrected QT response. RESULTS: A 3-compartment pharmacokinetic model with first-order absorption characterized the time course of dofetilide concentrations. On the basis of an in vitro potency of 5.13 ng/mL for potassium current inhibition and predicted unbound dofetilide concentrations, the estimated transducer ratio (tau) of 6.2 suggests that the QT response plateaus before currents are fully blocked. In our study population, 10% hERG blockade corresponds to a QT prolongation of 20 ms (95% confidence interval, 12-32 ms). With long-term dofetilide administration, tolerance develops with a half-life of 4.7 days. CONCLUSIONS: The current mechanism-based pharmacokinetic-pharmacodynamic model quantified the relationship between in vitro hERG channel blockade and clinical QT prolongation for dofetilide. This model may prove valuable for assessing the risk of QT prolongation in humans for other drugs that selectively block the hERG channel on the basis of in vitro assays and pharmacokinetic properties.

Models for time-varying covariates in population pharmacokinetic-pharmacodynamic analysis.

AIM: If appropriately accounted for in a pharmacokinetic (PK)-pharmacodynamic (PD) model, time-varying covariates can provide additional information to that obtained from time-constant covariates. The aim was to present and apply two models applicable to time-varying covariates that capture such additional information. METHODS: The first model estimates different covariate-parameter relationships for within- and between-individual variation in covariate values, by splitting the standard covariate model into a baseline covariate (BCOV) effect and a difference from baseline covariate (DCOV) effect. The second model allows the magnitude of the covariate effect to vary between individuals, by inclusion of interindividual variability in the covariate effect. The models were applied to four previously analysed data sets. RESULTS: The models were applied to 10 covariate-parameter relationships and for three of these the first extended model resulted in a significant improvement of the fit. Even when this model did not improve the fit significantly, it provided useful information because the standard covariate model, which assumes within- and between-patient covariate relationships of the same magnitude, was only supported by the data in four cases. The inclusion of BCOV was not supported in two cases and DCOV was unnecessary in three cases. In one case, significantly different, nonzero, relationships were found for DCOV and BCOV. The second extended model was found to be significant for four of the 10 covariate-parameter relationships. CONCLUSIONS: On the basis of the examples presented, traditionally made simplifications of covariate-parameter relationships are often inadequate. Extensions to the covariate-parameter relationships that include time-varying covariates have been developed, and their appropriateness and benefits have been described.

Assessment of the relative in vivo potency of the hydroxylated metabolite of darifenacin in its ability to decrease salivary flow using pooled population pharmacokinetic-pharmacodynamic data.

AIMS: To describe the population pharmacokinetic-pharmacodynamic relationship between darifenacin (UK-88,525) and its hydroxylated metabolite (UK-148,993), and the reduction in salivary flow (SF, a M3-mediated response). This enabled an estimation of the in vivo potency of the metabolite to decrease SF relative to that of the parent drug. METHODS: A total of 262 individuals were pooled from 11 Phase 1 studies and one Phase 2 study. A comparison was made between a series of pharmacodynamic models (direct-effect, indirect-effect, link and binding model) using NONMEM. RESULTS: The binding model yielded the best description of the decrease in SF by fully accounting for the time course of the pharmacodynamic effect. An internal validation exercise demonstrated the robustness of this model. Covariate analysis identified a circadian rhythm in SF. This model, with confidence intervals (CI) determined by likelihood profiling, indicated that the relative potency of the metabolite to darifenacin to reduce SF was 11.1% (95% CI 3.8, 19.6). This implied that the metabolite was ninefold less potent than darifenacin in vivo. Accounting for the unbound fraction of darifenacin (2%) and its metabolite (13%), the in vivo protein binding-corrected relative potency was estimated to be 2.1%, indicating that the metabolite was 50-fold less potent than the parent drug. The model supported the assumption that no other metabolites contributing to the impairment of the SF were formed during first-pass, and that the development of sensitization or tolerance was not evident over time. The validation process indicated that the i.v.-oral crossover study was necessary for the estimation of the relative potency. CONCLUSIONS: Population modelling of darifenacin and its hydroxylated metabolite yielded individual pharmacokinetic predictions that could be used to assess the in vivo potency of the metabolite to decrease SF relative to that of the parent drug. The metabolite had a negligible effect on SF.

Population pharmacokinetic modelling of darifenacin and its hydroxylated metabolite using pooled data, incorporating saturable first-pass metabolism, CYP2D6 genotype and formulation-dependent bioavailability.

AIMS: A model describing the population pharmacokinetics of darifenacin and its hydroxylated metabolite was developed from a combined analysis of 18 studies. The relationships between explanatory covariates and pharmacokinetic parameters were explored. METHODS: Plasma concentration data from 337 individuals were pooled from 17 Phase 1 studies (median 28/33 darifenacin/metabolite observations per healthy subject), and one Phase 2 study (median 7/7 darifenacin/metabolite observations per subject) encompassing one intravenous and five different oral formulations (1-45 mg). RESULTS: Non-linear Mixed Effects Models (NONMEM Version VI) described both the population pharmacokinetics of darifenacin and its hydroxylated metabolite with a two-compartment disposition model with first order absorption. The values (mean +/- standard error of the mean) for clearance (CL) and volume of distribution of the central compartment were 40.2 +/- 2.0 l h-1 and 34.7 +/- 4.6 l h-1, respectively, in a typical male CYP2D6 homozygote-extensive metabolizer (Hom-EM). The absolute bioavailability (F) of darifenacin in a Hom-EM after doses of 7.5, 15 or 30 mg extended release formulation (CR) was 15, 19 and 25%, respectively. Factors influencing F were formulation (70-110% higher for CR compared with immediate release following equivalent daily doses), CYP2D6 genotype [heterozygote-extensive metabolizers (Het-EM) and poor metabolizers (PM) experienced 40 and 90%, respectively, higher exposure than Hom-EM irrespective of dose administered] and saturable first-pass metabolism (dose nonlinearity 1.05-1.43-fold). Race affected F, which was 56% lower in Japanese males. The CYP3A4 inhibitors ketoconazole and erythromycin increased F to approximately 100% and ketoconazole decreased CL by 67.5%. CL was 31% lower in females and 10% lower at night. Formulation affected the metabolite absorption/formation rate. Ketoconazole and erythromycin administration resulted in a decrease of 61.2 and 28.8% in exposure to the metabolite, respectively. The covariates race, gender and circadian rhythm accounted for only approximately half of the variability in the estimated exposures to darifenacin. CONCLUSIONS: The pooled analysis provided a descriptive integration of all characteristics and covariates of the pharmacokinetics of darifenacin and its metabolite, enabling interpolation and extrapolation of these key factors.

A population pharmacokinetic analysis of sildenafil citrate in patients with erectile dysfunction.

AIMS: To analyse the pharmacokinetics of sildenafil citrate in patients with erectile dysfunction in order to characterize covariate relationships and assist in the development of rational dosage strategies. METHODS: A population pharmacokinetic sampling strategy was incorporated into five phase III clinical study protocols. Overall, 2077 patients, 1335 of whom received sildenafil, were asked to take an additional dose of study drug before their scheduled clinic visits on four or five occasions throughout the study duration. A single plasma sample was obtained at random times postdose (range 1--7 h), and a total of 4582 samples were assayed (average 3.4 samples per individual). RESULTS: For the population average patient (age 58 years; aspartate transaminase [AST], 24 IU l(-1); weight, 87 kg; not receiving CYP3A4 potential inhibitors), typical values for sildenafil (mean +/- SE) were 58.5 +/- 1.4 l h(-1) for apparent clearance (CL/F), 310 +/- 6.92 l for volume of distribution (V/F), and 2.6 +/- 0.176 h(-1) for first-order absorption constant (ka). The value for ka is associated with meal consumption within 2 h predose, at all other times ka was equivalent to an instantaneous bolus administration. The interindividual variabilities were 29% for CL/F, 20% for V/F, and 210% for ka. Over a dose range of 25--100 mg sildenafil, the pharmacokinetics exhibited dose proportionality. There was evidence of nonproportionality (40% increase on average) in relative bioavailability with respect to the 200-mg dose (P<0.001) relative to the other doses. Age, AST concentration, and co-administration with CYP3A4 potential inhibitors significantly influenced CL/F of sildenafil (P<0.001, for each relationship). For age and AST, the extent of the linear relationships (extrapolated from population average values) included a 4% decrease in CL/F for every decade increase and a 6% decrease in CL/F for every 10-unit increase, respectively. Following co-administration of CYP3A4 potential inhibitors, a 14% decrease in CL/F was estimated. Only body weight was found to significantly (P<0.001) influence V/F (a 6% increase in V/F for every 10-kg increase). CONCLUSIONS: The pharmacokinetics of, and covariate influences on, sildenafil in patients with erectile dysfunction were shown to be consistent with those demonstrated in phase I volunteer studies.