Evaluation of clinical dosing of E7820 from preclinical and clinical data using a biomarker
RJ Keizer, Y Funahashi, T Semba, J Wanders, JH Beijnen, JHM Schellens, ADR Huitema
Dept of Pharmacy & Pharmacology, the Netherlands Cancer Institute / Slotervaart Hospital, Amsterdam, NL; Eisai Limited, London, UK; Eisai Limited, Japan
Introduction
The novel anti-cancer agent E7820 inhibits angiogenesis by inhibition of mRNA expression of α2-integrin.[1] It has shown to induce tumor remission in preclinical experiments[2], and is now tested in phase I.[3] It was hypothesized that the inhibition of α2-integrin measured on platelets would provide a measure of pharmacological target modulation, and thereby a possible biomarker for drug efficacy.
Objectives
- establish target levels of α2-integrin inhibition correlated with tumor stasis in mice
- to investigate if the dose regimen proposed from the phase I study (based on acceptable toxicity) would result in sufficient inhibition of α2-integrin expression to reach this target
Methods
Tumor growth experiments were performed in 25 mice bearing a transplanted pancreatic KP-1 tumor, studying dosing at 0- 25 mg/kg over a period of 21 days. In the clinical study, E7820 was administered daily for 28 days, followed by a washout period of 7 days prior to starting subsequent cycles. Both in preclinical and clinical experiments, α2-integrin levels were measured on platelets by FITC-conjugated anti-integrin staining in combination with flow cytometry. PK-PD models were developed in NONMEM VI, using FOCE-I. Modeling and simulation was performed according to a pre-specified analysis plan, and construction of the PK-PD model was performed sequentially. First, the E7820 plasma concentrations were correlated to the inhibition of α2-integrin expression. Next, a tumor growth model based on unperturbed tumor growth experiments was fitted. The parameter estimates for the α2-integrin model were then fixed, and the model predicted expression levels were used to drive the tumor growth model.
Preclinical modeling
Longitudinal description of the α2-integrin expression level on platelets was modeled as a turnover model with an inhibitory effect on input rate (kin) of E7820 plasma concentration, described by a linear equation or by the Hill equation. An inhibitory effect of plasma exposure on kin was considered a mechanistically more plausible model than a stimulatory effect on kout, as the pharmacological effect of E7820 is mediated through the inhibition of mRNA expression.
Several tumor growth models were evaluated including exponentials models, Gompertz models [4], and a tumor growth model introduced by Simeoni et al.[5] First, unperturbed growth was described, after which it was attempted to describe the effect of inhibition of α2-integrin. The effect of inhibition of α2-integrin expression on one of the relevant growth rate parameters in the tumor growth models was incorporated as linear or Emax relationships. For both the model for α2-integrin and the tumor growth model, it was evaluated if the available data supported the estimation of between-mice variation of the parameters. The incorporation of effect compartments to delay the effects of drug on α2-integrin, or α2-integrin on tumor growth inhibition, and the development of resistance to drug were also evaluated. For both the α2-integrin and the tumor growth model, an exponential residual error model was used.
As at all dose levels tumor growth was observed despite initial remission, achieving a tumor size at t = 21 days lower or equal than the tumor size at baseline was defined as tumor stasis. To allow for discrepancies in tumor sensitivity between cell lines, two targets were set at 50% and 90% of mice achieving tumor stasis. Using the combined preclinical model, the relative inhibition of α2-integrin expression at steady state that correlated with these targets (Iinh,50, Iinh,90) were calculated using simulations over a dose range of 50 to 200 mg/kg bid.
Clinical modeling
A PK model for E7820 in patients had been constructed earlier. Briefly, this consisted of a one compartment model, with first-order elimination and absorption.[6] This model was then linked to a turnover model describing the inhibition of α2-integrin expression in patients. Next, simulations of profiles of α2-integrin expression from the clinical PK-PD model evaluated the levels of inhibition of α2-integrin expression that were achieved using these regimens. To evaluate if efficacy can be expected from the clinical regimens, these inhibition levels were then compared to the α2-integrin targets required for tumor stasis defined from the preclinical experiments.
Results
Preclinical modeling
From preclinical experiments, 119 α2-integrin measurements and 210 tumor size measurements were available. An Emax model was used to describe the relationship between plasma concentration and the effect on the input rate in the α2-integrin turnover model. An exponential tumor growth best described unperturbed tumor growth and growth inhibition in mice due to the inhibition of α2-integrin expression. The exponential tumor growth model was extended with a term describing initial slow growth, possibly due to the transplanted tumor not being fully embedded in its surrounding tissue. The tumor growth equation thus was described by: dT/dt = -α·(1-eβt)·T - EI·T, with T describing tumor size in mm, t describing time in days, and α and β describing tumor growth rate and initial growth resistance rate. The factor EI described the effect of inhibition of α2-integrin expression on tumor growth and was described by a sigmoid Emax model. Both the α2-integrin and tumor growth model provided good fit as judged by visual predictive checks. Simulations from the combined model for α2-integrin inhibition and tumor growth resulted in values for Iinh,50 and Iinh,90 of 14.7% and 17.9%, respectively.
Clinical modeling
The clinical dataset consisted of 462 α2-integrin level measurements at 209 unique time-points from 29 patients, collected from up to 9 treatment cycles. Although considerable between patient variation was observed in baseline expression levels and response to drug, the observed α2-integrin expression levels in patients and between patient variation could be described adequately using the turnover model. Simulation of clinical regimens from this PK-PD model showed that a dose of 100 mg qd, which was the maximum tolerable dose level in the phase I study, α2-integrin expression was inhibited more strongly than the Iinh,50 in >95% of patients, and more strongly than the Iinh,90 in >50% of patients. Doses of 50 mg qd or lower resulted in ≤ 50% of patients expected to reach the Iinh,50.
Conclusion
The relative level of α2-integrin inhibition that corresponded to tumor stasis in 50% or 90% of mice was only moderate (<20%). These levels of inhibition could be met in patients using doses of 100 mg for the majority of patients, but this analysis suggest that doses of 50 mg qd or lower are unlikely to achieve efficacy. These results also indicate that the level of α2-integrin expression on platelets relative to baseline may be a valuable clinical biomarker for the drug effect on tumor growth, warranting further investigation.
References
[1] Y. Funahashi, N.H. Sugi, T. Semba, Y. Yamamoto, S. Hamaoka, N. Tsukahara-Tamai, Y. Ozawa, A. Tsuruoka, K. Nara, K. Takahashi, T. Okabe, J. Kamata, T. Owa, N. Ueda, T. Haneda, M. Yonaga, K. Yoshimatsu, and T. Wakabayashi, "Sulfonamide derivative, E7820, is a unique angiogenesis inhibitor suppressing an expression of integrin alpha2 subunit on endothelium," Cancer Res, vol. 62, 2002, pp. 6116-6123.
[2] T. Semba, Y. Funahashi, N. Ono, Y. Yamamoto, N.H. Sugi, M. Asada, K. Yoshimatsu, and T. Wakabayashi, "An angiogenesis inhibitor E7820 shows broad-spectrum tumor growth inhibition in a xenograft model: possible value of integrin alpha2 on platelets as a biological marker," Clin Cancer Res, vol. 10, 2004, pp. 1430-1438.
[3] M.M. Mita, A.C. Mita, A. Ricart, J. Rodon, E. Till, C. Okereke, P. Desai, E. Zang, E.K. Rowinsky, and C. Takimoto, "Phase I study of an anti-angiogenic agent with a novel mechanism of action E7820: Safety, pharmacokinetics (PK) and pharmacodynamic (PD) studies in patients (pts) with solid tumors," J Clin Oncol (Meeting Abstracts), vol. 24, Jun. 2006, p. 3048.
[4] A. Iliadis and D. Barbolosi, "Optimizing drug regimens in cancer chemotherapy by an efficacy-toxicity mathematical model," Computers and Biomedical Research, an International Journal, vol. 33, Jun. 2000, pp. 211-226.
[5] M. Simeoni, P. Magni, C. Cammia, G. De Nicolao, V. Croci, E. Pesenti, M. Germani, I. Poggesi, and M. Rocchetti, "Predictive pharmacokinetic-pharmacodynamic modeling of tumor growth kinetics in xenograft models after administration of anticancer agents," Cancer research, vol. 64, Feb. 2004, pp. 1094-101.
[6] R. Keizer, M. Zamacona, M. Jansen, D. Critchley, J. Wanders, J. Beijnen, J. Schellens, and A. Huitema, "Application of population pharmacokinetic modeling in early clinical development of the anticancer agent E7820," Investigational New Drugs, Aug. 2008.
