ABSTRACT Fruquintinib is approved, including in the United States and Europe, for previously treated metastatic colorectal cancer, regardless of biomarker status. Using in vitro and clinical data, physiologically based pharmacokinetic (PBPK) models were developed for fruquintinib and M11, a major metabolite of fruquintinib. The models were applied to evaluate the potential impact of strong/moderate/weak cytochrome P450 (CYP) and 3A4 (CYP3A4) inhibitors and inducers on the pharmacokinetics of fruquintinib and M11. In addition, the PBPK models were applied to evaluate the potential for P‐glycoprotein or breast cancer resistance protein (BCRP)‐mediated drug–drug interactions, with fruquintinib and M11 as precipitants (inhibitors). The simulations indicated that the changes in plasma exposure of fruquintinib area under the curve (AUC) geometric mean ratio (GMR) during administration with strong/moderate/weak inhibitors of CYP3A4 are negligible (< 1.25), and M11 interactions are weak with strong inhibition. Simulated changes in plasma exposure of fruquintinib in the presence of rifampin (a strong CYP3A4 and moderate CYP2C9 inducer) were > 0.20, but ≤ 0.50 (no significant change in M11). Fruquintinib AUC GMRs were > 0.50, but ≤ 0.80 with moderate CYP3A4 inducers, while the ratio fell between > 0.80 and 1.00 in the presence of weak CYP3A4 inducers. The effects of fruquintinib on the pharmacokinetics of P‐glycoprotein or BCRP substrates were not clinically significant. These analyses support avoiding concomitant use of strong, and if possible, moderate CYP3A4 inducers with fruquintinib. In addition, no dose adjustments for fruquintinib are required in the presence of CYP3A4 inhibitors, or for medications that are substrates of P‐glycoprotein or BCRP when used concomitantly with fruquintinib.
ABSTRACT Exposure‐Response (ER) analyses are key to optimizing dosing regimens for the safe and effective use of drugs. The choice of modeling approach should depend on the questions being asked and on the method's ability to appropriately characterize the ER relationship. While logistic regression (LR) is a common choice to evaluate ER relationships for binary endpoints, limitations can affect its reliability. Specifically, it cannot explicitly account for time‐varying exposure, necessitating the use of summary exposure metrics that may bias its results. Longitudinal models can address some time‐dependent factors known to affect LR analyses, but these require greater resource investment. As a result, clearer guidance is needed for selecting appropriate exposure metrics in LR and for determining when the additional complexities of longitudinal models are warranted. In this study, we compared Markov modeling with LR using systematically varied simulations designed to evaluate multiple strengths of the ER relationship, representative pharmacokinetics for small and large molecules, early and late event onset, high and low dose modification rates, different total sample sizes, and multiple exposure metrics. Overall, Markov modeling demonstrated a superior ability to identify ER relationships, especially those of intermediate strength and in smaller sample sizes, and tended to yield less biased estimates of the drug effect. Results did not suggest advantages of event‐based metrics for LR. This study is the first to demonstrate the effectiveness of a longitudinal model relative to LR for ER analyses, and it introduces a simulation framework that guides method and exposure metric selection across common drug development scenarios.
ABSTRACT AZD2796 is a humanized anti‐LILRB2 monoclonal antibody designed to promote immune activation by inhibiting LILRB2 on myeloid cells. Because AZD2796 does not bind cynomolgus monkey or rodent LILRB2, conventional approaches to first‐in‐human dose selection were limited. An affinity‐based minimum anticipated biological effect level (MABEL) approach would predict a very low starting dose despite the potential for substantial target‐mediated drug disposition (TMDD). To support dose selection, pharmacokinetic and receptor occupancy (PK‐RO) models were developed using AZD2796 pharmacokinetics in cynomolgus monkeys and TMDD data estimated from a cynomolgus‐reactive surrogate anti‐LILRB2 antibody. An alternative approach using neonatal Fc receptor‐ and LILRB2‐transgenic mouse models was also explored. The cynomolgus surrogate model predicted substantial TMDD at low doses and supported an intravenous starting dose of 30 mg every 3 weeks (Q3W) that was predicted to maintain peripheral LILRB2 occupancy. The model further predicted that approximately 300 mg Q3W would maintain > 95% tumor LILRB2 occupancy under conservative assumptions regarding target burden and tumor exposure. Model performance was supported by the model's ability to reproduce the clinically observed PK and RO characteristics of other anti‐LILRB2 antibodies. The transgenic mouse approach was less robust and generally predicted higher dose requirements. These findings demonstrate that surrogate antibody PK‐RO modeling can support the selection of a pharmacologically active first‐in‐human starting dose for therapeutic antibodies that lack cross‐reactivity with standard preclinical species.
ABSTRACT Antisense oligonucleotides are small synthetic DNA or RNA molecules that can be used to modulate gene expression, particularly to block the production of unwanted proteins. This makes them potentially powerful drugs to contrast cancer growth. Drug development is a long process that includes preclinical and clinical studies. Preclinical animal studies are designed to assess drug kinetics and safety before progressing to clinical trials, as they may predict the therapeutic dose to be used in the “first in human” (FIH) studies. This is usually achieved by applying allometric scaling methods that allow predicting the human pharmacokinetic (PK) parameters from those estimated in animals. Selecting the appropriate scaling formula reduces the risk of translational uncertainty and supports the design of FIH studies. The aim of this study is to compare the accuracy of different allometric scaling methods in predicting human PK of the antisense oligonucleotide LNA‐i‐miR‐221, a powerful inhibitor of microRNA miR‐221, which was found to be deregulated in several malignancies, favoring cancer progression. To do that, we exploited the availability of PK data of the drug in both Sprague Dawley rats and humans and we found that the best scaling method for both clearance and distribution volume is the simple allometry, with predictions laying within the 2‐fold and 4‐fold error interval, respectively. This study contributes to identifying the most accurate scaling methods among the evaluated single‐species approaches to extrapolate human pharmacokinetic for LNA‐i‐miR‐221, possibly offering a starting point for future studies involving antisense oligonucleotides.
ABSTRACT Cenobamate causes dose‐dependent cytochrome P450 (CYP)2B6 and CYP3A4 induction and CYP2C19 inhibition, raising concern over multi‐pathway drug–drug interactions (DDIs) in patients with epilepsy treated with cardiovascular co‐medications. We developed a physiologically based pharmacokinetic (PBPK) model for cenobamate and verified it using reported DDIs for cenobamate 200 mg using three probe substrates: bupropion (CYP2B6), midazolam (CYP3A4), and omeprazole (CYP2C19). The verified model predicted exposure changes (area under the curve ratio [AUCR] and maximum concentration [Cmax] ratio with/without cenobamate) for eight drugs, including anticoagulants and antiplatelets (apixaban, cilostazol, clopidogrel, and rivaroxaban), and antihypertensives (amlodipine, bisoprolol, carvedilol, and nifedipine). Regarding anticoagulants and antiplatelets, for apixaban, AUCR/Cmax ratios were 0.78/0.92 (100 mg), 0.71/0.87 (200 mg), and 0.63/0.83 (400 mg), respectively. For cilostazol, parent exposure decreased (0.67/0.85, 0.55/0.77, 0.45/0.70), while the active metabolite OPC‐13015 was predicted to increase in Cmax. For clopidogrel, parent exposure was unchanged, but formation of 2‐oxo‐clopidogrel and clopidogrel‐H4 were decreased and increased, respectively. For rivaroxaban, the AUCR/Cmax ratios were 0.78/0.90, 0.69/0.85, and 0.61/0.80. Among antihypertensives, amlodipine showed decreased AUCR/Cmax ratios (0.57/0.66, 0.45/0.55, 0.37/0.47). The exposure changes of bisoprolol were minor (AUCR 0.84–0.92, Cmax ratio 0.95–0.98); AUCR/Cmax ratios of carvedilol fell to 0.73/0.80, 0.63/0.71, and 0.55/0.64; and nifedipine decreased markedly (0.41/0.54, 0.29/0.41, 0.21/0.32). In conclusion, cenobamate induced CYP2B6 weakly and CYP3A4 moderately, and it inhibited CYP2C19 moderately to drive dose‐dependent decreases in exposure for seven of eight drugs with minimal change for bisoprolol. The verified PBPK framework enables risk assessment in scenarios not yet clinically tested.
