Eric Masson

Harnessing Meta‐analysis to Refine an Oncology Patient Population for Physiology‐Based Pharmacokinetic Modeling of Drugs

Certain oncology compounds exhibit fundamental pharmacokinetic (PK) disparities between healthy and malignant conditions. Given the effects of tumor‐associated inflammation on enzyme and transporter expression, we performed a meta‐analysis of CYP‐ and transporter‐sensitive substrate clinical PK to quantitatively compare enzyme and transporter abundances between healthy volunteers (HV) and cancer patients (CP). Hepatic and intestinal CYP1A2, CYP2C19, and CYP3A4 abundance were subsequently adjusted via Simcyps sensitivity analysis tool. Of the 11 substrates we investigated, seven displayed marked exposure differences >1.25‐fold between CP and HV. Although CP studies are limited, meta‐analysis‐based reduction in CYP1A2, CYP2C19, and CYP3A4 enzyme abundances in a virtual oncology population effectively captures CP‐PK for caffeine, theophylline, midazolam, simvastatin, omeprazole, and a subset of oncology compounds. These changes allow extrapolation from HV to CP, enhancing predictive capability; therefore, conducting simulations in this CYP‐modified oncology (MOD‐CP) population provides a more relevant characterization of CP‐PK.

Use of a cocktail probe to assess potential drug interactions with cytochrome P450 after administration of belatacept, a costimulatory immunomodulator

Aim This open‐label study investigated the effect of belatacept on cytokine levels and on the pharmacokinetics of caffeine, losartan, omeprazole, dextromethorphan and midazolam, as CYP probe substrates after oral administration of the Inje cocktail in healthy volunteers. Methods Twenty‐two evaluable subjects received the Inje cocktail on Days 1, 4, 7 and 11 and belatacept infusion on Day 4. Results Since belatacept caused no major alterations to cytokine levels, there were no major effects on CYP‐substrate pharmacokinetics, except for a slight (16–30%) increase in omeprazole exposure, which was probably due to omeprazole‐mediated, time‐dependent CYP inhibition. Belatacept did not cause major alterations in the pharmacokinetics, as measured by the geometric mean ratios and associated 90% confidence interval for area under the plasma concentration ‐time curve from time zero to infinity on Day 7 comparing administration with and without belatacept for caffeine (1.002 [0.914, 1.098]), dextromethorphan (1.031 [0.885, 1.200]), losartan (1.016 [0.938, 1.101)], midazolam (0.968 [0.892, 1.049]) or their respective metabolites. Conclusions Therefore, no dose adjustments of CYP substrates are indicated with belatacept coadministration.

Population pharmacokinetic and exposure simulation analysis for cediranib (AZD2171) in pooled Phase I/II studies in patients with cancer

AIMS A population pharmacokinetic (PK) model was developed for cediranib to simulate cediranib exposure for different doses, including comedication with strong uridine glucuronosyl transferase/P-glycoprotein inducers such as rifampicin, in cancer patients. METHODS Plasma concentrations and covariates from 625 cancer patients after single or multiple oral cediranib administrations ranging from 0.5 to 90 mg in 19 Phase I and II studies were included in the analysis. Stepwise covariate modelling was used to develop the population PK model. The final model was used to simulate cediranib exposure in cancer patients to evaluate cediranib target coverage and the need for dose adjustment for covariates or coadministration with rifampicin. RESULTS A two-compartment model with sequential zero- and first-order absorption and first-order elimination adequately described the cediranib concentration-time courses. Body weight and age were identified as having statistically significant impact on cediranib PK, but only <21% impact on AUC and maximum concentrations. Simulated lower bounds of 90% prediction interval or median of unbound cediranib concentrations after cediranib 15 or 20 mg exceeded the IC50 for vascular endothelial growth factor receptors-1, -2 and -3. Exposures of cediranib 20 or 30 mg with coadministration of rifampicin were comparable to those of 15 or 20 mg, respectively, without coadministration. CONCLUSIONS No covariate was identified to require dose adjustment for cediranib. Cediranib exposure following 15 or 20 mg daily dose administration is adequate overall for inhibition of in vitro estimated vascular endothelial growth factor receptor-1, -2 and -3 activities. An increase in cediranib dose may be needed for cediranib coadministered with strong uridine glucuronosyl transferase/P-glycoprotein inducers such as rifampicin.

Physiologically Based Pharmacokinetic Modeling to Evaluate the Systemic Exposure of Gefitinib in CYP2D6 Ultrarapid Metabolizers and Extensive Metabolizers

Gefitinib is a selective inhibitor of epidermal growth factor receptor (EGFR) tyrosine kinase and is used for the treatment of non‐small‐cell lung cancer (NSCLC) with activating EGFR mutations. Gefitinib is metabolized by CYP2D6 and CYP3A4. This analysis compared the systemic exposure of gefitinib after oral administration in CYP2D6 ultrarapid metabolizers (UM) vs extensive metabolizers (EM). Physiologically based pharmacokinetic (PBPK) modeling was conducted using a population‐based simulator. The model was calibrated using itraconazole‐gefitinib clinical drug‐drug interaction data and validated with gefitinib data in CYP2D6 EM vs poor metabolizers (PM). System components of the PBPK model were evaluated using published clinical metoprolol pharmacokinetic data in CYP2D6 UM, EM, and PM. Our PBPK model predicted a gefitinib geometric least‐squares mean area under the plasma concentration‐time curve (AUC) from time 0 to 264 hours (AUC(0‐264)) ratio in the presence vs absence of itraconazole of 1.85, similar to the ratio of 1.78 from clinical study data. Predicted and observed metoprolol geometric least‐squares mean AUC(0‐24) ratios in UM vs EM were also similar (0.46 and 0.55, respectively), suggesting that system components related to CYP2D6 in the PBPK model were properly established. In addition, the PBPK model also captured gefitinib pharmacokinetic profiles in CYP2D6 polymorphic populations. The final PBPK model predicted a decrease in gefitinib AUC of ∼39% in CYP2D6 UM vs EM. Such changes in exposure will have limited impact as the reduced exposure is still above gefitinibs in vitro IC90 for EGFR activating mutations in NSCLC patients.

Evaluation of classical clinical endpoints as surrogates for overall survival in patients treated with immune checkpoint blockers: a systematic review and meta-analysis

PurposeClassical clinical endpoints [e.g., objective response rate (ORR), disease control rate (DCR), and progression-free survival (PFS)] may not be appropriate for immune checkpoint blockers (ICBs). We evaluated correlations between these endpoints and overall survival (OS) for surrogacy.MethodsRandomized controlled trials (RCTs) of solid tumors patients treated with ICBs published between 01/2005 and 03/2017, and congress proceedings (2014–2016) were included. Arm-level analyses measured 6-month PFS rate to predict 18-month OS rate. Comparison-level analyses measured ORR odds ratio (OR), DCR OR, and 6-month PFS hazard ratio (HR) to predict OS HR. A pooled analysis for single-agent ICBs and ICBs plus chemotherapy vs chemotherapy was conducted. Studies of single-agent ICBs vs chemotherapy were separately analyzed.Results27 RCTs involving 61 treatment arms and 10,300 patients were included. Arm-level analysis showed higher 6- or 9-month PFS rates predicted better 18-month OS rates for ICB arms and/or chemotherapy arms. ICB arms had a higher average OS rate vs chemotherapy for all PFS rates. Comparison-level analysis showed a nonsignificant/weak correlation between ORR OR (adjusted R2 = − 0.069; P = 0.866) or DCR OR (adjusted R2 = 0.271; P = 0.107) and OS HR. PFS HR correlated weakly with OS HR in the pooled (adjusted R2 = 0.366; P = 0.005) and single-agent (adjusted R2 = 0.452; P = 0.005) ICB studies. Six-month PFS HR was highly predictive of OS HR for single-agent ICBs (adjusted R2 = 0.907; P < 0.001), but weakly predictive in the pooled analysis (adjusted R2 = 0.333; P = 0.023).ConclusionsPFS was an imperfect surrogate for OS. Predictive value of 6-month PFS HR for OS HR in the single-agent ICB analysis requires further exploration.

Abstract LB-C14: Using Drug Exposure as a Metric for Predicting Clinical Response to Targeted Cancer Therapeutics from Preclinical Efficacy: A Retrospective Preclinical to Clinical Correlation

Objective The objective of this retrospective analysis was to evaluate the correlation between preclinical and clinical markers of efficacy, using available monotherapy data for AstraZeneca targeted oncology drugs. Methods For each drug, preclinical/clinical parameters were identified, including; dose/dosing schedule, tumor model,pharmacokinetic parameters (AUC, t1/2), clinically relevant dose (CRD), tumor growth inhibition (TGI), objective response rate (ORR), disease control rate (DCR), mean tumor shrinkage (MTS), and protein binding. Drugs were first examined to compare the concentrations (Cav) seen at CRDs to preclinical thresholds defined by the concentration required for 50% target inhibition (IC50). Each drug had a number of associated IC50 values, and in vivo values were preferentially used as the concentration target when available. Free Cav values were calculated from AUCs, and a ratio of Cav/IC50 > 1 was used to determine whether the clinical drug exposure exceeded the preclinical efficacy target concentration. The second objective was to determine the correlation between preclinical and clinical efficacy. This was accomplished by estimating murine TGI for each drug at either the allometrically-scaled, mouse-equivalent CRD or at the free exposure (CRE) seen in humans using a variety of dose-response models. Linear regression, weighted by the number of patients from whom clinical data was available, was performed on each comparison, and ANOVA was used to test the significance of each relationship using R v.3.2.2. Results Seventeen targeted drugs with clinical efficacy data were identified. A total of 9/15 (60%) drugs with available data exceeded a free Cav /IC50 ratio of 1 (ratio range of target attaining drugs: 3.6-63.3), indicating that target concentration had been achieved in those drugs. Tumor growth inhibition predicted by the mouse-equivalent CRD ranged from 0.5-155.7% across 13 drugs, while it ranged from 53.4-179.9% across 9 drugs for which TGI could be predicted from the CRE. The weighted correlation between CRD-TGI and clinical ORR was significant (r = 0.71, p Conclusions Comparison of preclinical concentration targets and clinical exposures suggests that the CRD is only achieving the preclinical target in just over half of the drugs, highlighting the importance of choosing a relevant in vivo preclinical concentration to target clinically. Conversely, a high correlation between CRD-based TGI and clinical ORR was observed suggesting that this marker may be more appropriate for prediction of clinical efficacy from preclinical data. Overall, this study demonstrated a correlation between dose-predicted TGI and clinical efficacy. Although this is a small data set, this study confirmed the importance of setting an efficacy threshold preclinically before moving into the clinic with oncology targeted drugs. Citation Format: Matthew Linakis, James Yates, Eric Masson, Ganesh Mugundu. Using Drug Exposure as a Metric for Predicting Clinical Response to Targeted Cancer Therapeutics from Preclinical Efficacy: A Retrospective Preclinical to Clinical Correlation. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr LB-C14.