Discussion

ONO-7579 is an oral pan-TRK inhibitor that is expected to be highly effective and selective in inhibiting phosphorylation of all TRKs (Kawamoto et al., 2018). In this article, a relationship between pTRKA in tumors and antitumor effect in the murine KM12 xenograft model was defined using a PK/PD modeling approach to contribute to the selection of a recommended dose for clinical studies. There are two major approaches to apply PK/PD modeling to characterize in vivo exposure, biomarker response, and antitumor efficacy (Yamazaki et al., 2016). One is a two-step approach, in which a PK/PD model and a PK/efficacy model are developed in parallel, followed by comparison of the PK/PD-PK/efficacy relationships. The other is an integrated modeling approach, in which PK/PD/efficacy models are developed to simultaneously characterize the relationships of drug exposures to in vivo antitumor efficacy through biomarker responses. Both approaches are used to characterize molecularly targeted agents, and we considered that if a PD marker could be regarded as a single factor to control antitumor efficacy, the relationship between PD and efficacy would be clearly explained using an integrated modeling approach. The KM12 cell line is a colorectal cancer cell line harboring the TPM3-NTRK1 fusion gene (Vaishnavi et al., 2013), and pTRKA initiates the activation of downstream intracellular signaling pathways (Deinhardt and Chao, 2014). In addition, larotrectinib, a first-generation pan-TRK inhibitor, inhibits KM12 cell proliferation and pTRKA at approximately the same concentrations as in in vitro studies (Pandre et al., 2018). Therefore, inhibition of tumor growth after ONO-7579 administration can be explained directly by changes in pTRKA; consequently, an integrated modeling approach was used.

Plasma ONO-7579 concentration was well described with an oral one-compartment model. Only data about plasma ONO-7579 concentrations in xenograft mice ≥2 hour after administration were obtained, and the absorption rate constant could not be estimated without the data for normal mice. No differences in pharmacokinetics were found between normal and xenograft mice, and the model was considered to accurately describe absorption in xenograft mice. Tumor ONO-7579 concentrations were well described with an additional compartment, which did not influence plasma concentrations measurably. Although transfer parameters have no physiologic meaning, a description of the changes is required to consider pTRK behavior in the target. Tumor ONO-7579 concentrations were higher than plasma concentrations. It is the unbound fraction of a drug that normally penetrates into tissues: the more unbound drug in plasma, the more easily the drug penetrates into tissues. Our in-house data showed that plasma protein binding of ONO-7579 in mice was 99.94%, which was extremely high; it was also shown that ONO-7579 was sufficiently distributed to tumor target tissue. One possible explanation is the enhanced permeability and retention effect. In solid cancers, compounds with a high molecular weight, such as albumin-bound drugs, leak preferentially into tumor tissue through permeable tumor vessels and are then retained in the tumor because of reduced lymphatic drainage (Matsumura and Maeda, 1986). Another possible explanation is the lysosomotropism of basic compounds. Lipophilic compounds with weak base properties readily diffuse across cell membranes, and when these drugs enter the acidic lumen of lysosomes, they become protonated and are trapped in the lysosomes (Goldman et al., 2009). ONO-7579 is also a weakly basic compound having an aminopyridine moiety in its structure and has high lipophilicity. Therefore, these mechanisms might have contributed to the accumulation of ONO-7579 in the tumor.

Although a delay in distribution from plasma to tumor was found, tumor concentrations and pTRKA levels were well described with a direct inhibitory E_max model. An indirect response model (Dayneka et al., 1993) was also considered, which did not perform better, and no temporal delay in inhibiting pTRKA was incorporated. Since phosphorylation of receptor tyrosine kinase is the first phase of signal transduction, it was considered reasonable to have a direct correlation with ONO-7579 concentration in the target tumor site.

There have been many reports about tumor growth models in xenograft mice (Mould et al., 2015), and tumor growth in the simplest model can be described with an exponential function. Tumor volume increased in a logarithmic manner over 12 days of ONO-7579 administration, and a model describing growth saturation (Simeoni et al., 2004; Kogame et al., 2013) was not used. In addition, a slight reduction in tumors was observed in all treatment groups, including the control group, at observation 2 days after ONO-7579 administration. A placebo effect is considered unlikely in the model animals. The cause of the temporary reduction in tumor volume was unknown; however, mice were only assigned to groups 3 days after tumor cells were grafted and ONO-7579 administration was started. The remaining graft cell suspension (0.1 ml) and/or inflammation at the injection site on grafting might have biased the measurement of tumor volume 2 days after administration. In developing a PD/efficacy model, a significant reduction in the objective function (ΔOFV = −123.385) was noted when temporal effect up to day 2 was also incorporated into K_tg. Therefore, a model was developed with the temporal effect incorporated. Since predicted values in a final model accurately reflected observed values, the antitumor effect of ONO-7579 was considered properly evaluated.

The developed PD/efficacy model showed that ONO-7579 had a minimal antitumor effect at pTRKA inhibition rates ≤60%, but ONO-7579 started to sharply increase the antitumor effect at pTRKA inhibition rates >60%. These results indicate that there is a threshold pTRKA inhibition rate above which the antitumor effect is exerted. Interestingly, this “switch-like” behavior has also been reported in similar analyses conducted with other kinase inhibitors, such as cobimetinib, a mitogen-activated protein/extracellular signal–regulated kinase kinase inhibitor studied in A375 xenografts (Wong et al., 2012); GDC-0879, a v-raf murine sarcoma viral oncogene homolog B1 (B-RAF) kinase inhibitor studied in A375 xenografts (Wong et al., 2009); GDC-0834, a Bruton’s tyrosine kinase inhibitor studied in a rat collagen–induced arthritis model (Liu et al., 2011); and vismodegib, a hedgehog-signaling pathway inhibitor studied in D5123 xenografts and medulloblastoma allografts (Wong et al., 2011).

Although the strategy for the development of novel kinase inhibitors appears to follow the historical MTD-based development, many drugs issued postmarketing requirements/postmarketing commitments by the US Food and Drug Administration for dose optimization efforts due to safety concerns (Lu et al., 2016). PD biomarker monitoring in clinical development guides the optimized dose associated with biologic activity. For example, ibrutinib and trametinib have been approved for non-MTD label doses by this strategy (Lu et al., 2016). Similarly, measuring pTRKA in tumors would allow us to propose a biologically effective dose lower than the MTD. From another point of view, considering that drug resistance has been found in TRK inhibitors in development (Drilon et al., 2016; Russo et al., 2016), a dosage in which complete inhibition of pTRK is maintained throughout a dosing interval would be desirable to prevent the occurrence of drug resistance. Based on our results, a dosage required to fully (≥91.5%) and continuously inhibit pTRKA at the steady-state trough can be easily calculated and proposed for use in clinical studies of ONO-7579. In addition, the relationship between pTRKA inhibition and antitumor effect could be applied to other TRK inhibitors because their downstream signal is considered the same.

In conclusion, the PK/PD/efficacy models we developed in this study successfully identified a relationship between pTRKA inhibition and antitumor effect in a murine xenograft tumor model and demonstrated that pTRKA in tumors can be an effective biomarker to rationally design dosing regimens for early-stage clinical studies of ONO-7579.