Tumor Growth Inhibition In Models Research Articles

Vismodegib Efficacy in Advanced Basal Cell Carcinoma Maintained with 8-Week Dose Interruptions: A Model-Based Evaluation

Long-term use of vismodegib is associated with treatment-emergent adverse drug reactions in clinical trials of patients with locally advanced basal cell carcinoma (laBCC) and metastatic BCC (mBCC). Treatment interruptions have been recommended for management of adverse drug reactions, but there is no clear guidance on the duration of those interruptions. A model-based evaluation was conducted to assess the impact of treatment interruptions on vismodegib efficacy. A tumor growth-inhibition model was developed and model-based simulations were performed to assess the proportion of patients achieving ≥30% reduction from baseline in tumor size (ie, proportion of responders) for various dosing schedules. The most conservative simulated scenario compared an intermittent dosing schedule of 12 weeks of vismodegib treatment followed by 8-week dosing interruption (repeated four times for a total duration of 80 weeks) versus 80 weeks of continuous dosing, which predicted a decrease in responders of -4.7% from 90.7% in laBCC and -3.7% from 63.0% in mBCC. These results demonstrated maintenance of vismodegib efficacy with the intermittent dosing schedule and support the recently approved recommendations of treatment interruptions of up to 8 weeks to manage tolerability in laBCC or mBCC patients in the vismodegib US prescribing information.

Abstract 3022: Using translational tumor growth inhibition modeling approach and population PK analysis to predict efficacious doses for KN026, a HER2 bispecific antibody

Abstract Background: KN026 is a bispecific antibody simultaneously targeting the extracellular domains II and IV of the human HER2. It blocks both ligand-dependent and ligand-independent HER2 signaling pathway. The IgG1 Fc fragment of KN026 binds FcRγIIIa mediates potent ADCC and inhibits tumor cell proliferation. Aim: The goal of the present work is to predict efficacious doses for KN026 using a translational tumor growth inhibition modeling approach based on tumor growth and KN026 exposure data from mouse, followed by a population PK analysis based on KN026 concentration data from patients in a first-in-human (FIH) clinical trial. Methods: A tumor growth inhibition model was developed to fit tumor volume time course data from NCI-N87 and Calu-3 xenografts. A translational tumor growth inhibition model was next developed by adjusting the tumor growth component to more realistically reflect tumor growth dynamics as reported in HER2-positive breast cancer patients. Model parameters intrinsic to KN026 efficacy (Emax, EC50) were assumed constant across species. To determine the target concentration, simulations were performed for different initial tumor volumes and tumor doubling time in humans under different KN026 exposures. Next, intensively sampled pharmacokinetics data from 20 patients in a FIH trial of KN026 were used to develop the human population PK model. KN026 concentration-time profiles in humans were then simulated across a variety of candidate dosing regimens and compared with the projected target concentration from translational tumor growth inhibition modeling. Recommended efficacious doses and dosing schedules were predicted if steady state trough concentration from more than 90% simulated subjects achieve the target concentration. Results: A tumor growth inhibition model adequately described the data from xenograft models. To build the translational tumor growth inhibition model, a literature-documented tumor growth equation more relevant to the observed tumor growth dynamics in breast cancer patients was used to replace the growth component in formula 1 originally fitted to mouse data. Simulation results from the translational tumor growth inhibition model show that tumor stasis can be achieved at the KN026 trough concentration lower than 20 μg/mL. More aggressive tumors will take longer time to achieve tumor stasis under the same given concentration. A two-compartment PK model incorporating body weight as a covariate on both volume of distribution and clearance describes human PK data well. Efficacious steady state dose levels were predicted to be 20 mg/kg Q2W and 30 mg/kg Q3W. Loading doses with a higher frequency were predicted to have the advantage of maximizing initial tumor killing. Conclusions: The use of the modeling and simulation approach to aid decision in drug development has become a popular tool. Here we present a translational PK-PD modeling framework incorporating preclinical tumor growth data and clinical PK data to inform dose selection strategy for KN026 in patients with HER2-positive solid tumors. Furthermore, simulation results suggest that loading doses with higher dosing frequency will likely be beneficial. These model predictions will be further validated by emerging PK and efficacy data from ongoing clinical trials of KN026. Citation Format: Dongmei Ji, Yuan Xiong, Jing Yang, Pilin Wang, Jiarong Wang, Fei Yang, Junfang Xu. Using translational tumor growth inhibition modeling approach and population PK analysis to predict efficacious doses for KN026, a HER2 bispecific antibody [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3022.

Abstract 3026: Evaluation of brain pharmacokinetics (PK) and tumor growth inhibition of PARP inhibitors in mouse xenograft models using semi-mechanistic PK/pharmacodynamic (PD) modeling

Abstract OBJECTIVE: The PARP inhibitors rucaparib, niraparib, olaparib, and veliparib have shown remarkable antitumor activity against various cancer types. We evaluated the brain penetration of PARP inhibitors in mice and the anti-tumor effects of selected PARP inhibitors in orthotopic and intracranial BRCA1-mutant triple-negative breast cancer (TNBC) MDA-MB-436 mouse models. PK/PD models were used to characterize the systemic/brain PK and tumor growth inhibition. METHODS: In non-tumor-bearing NOD-SCID female mice, plasma concentrations were determined at selected time points following an oral dose of 2, 5, 15, 50, or 150 mg rucaparib. The antitumor efficacy of rucaparib (2, 5, 15, 50, or 150 mg BID for 28 days) was evaluated in the MDA-MB-436 orthotopic model. In addition, a single dose of rucaparib (150 mg), niraparib (50 mg), olaparib (150 mg), or veliparib (150 mg) was given to male CD-1 mice; plasma, brain, and cerebrospinal fluid (CSF) concentrations of these compounds were determined at various time points. Intracranial efficacy was evaluated in luciferase-expressing MDA-MB-436 tumors following treatment of rucaparib (150 mg BID) or niraparib (50 mg QD) for 57 days. Tumor burden was evaluated using bioluminescence measurements. Semi-mechanistic PK and tumor growth inhibition models were developed to evaluate systemic, brain, and CSF concentration-time profiles for all 4 PARP inhibitors, and to evaluate tumor inhibition for rucaparib and niraparib using NONMEM software. RESULTS: All of the PARP inhibitors showed limited brain penetration in mice with an intact blood-brain barrier. However, rucaparib and niraparib demonstrated antitumor efficacy in a BRCA1-mutant, intracranial, TNBC mouse model. Brain and CSF concentrations were modeled as separate physiologically relevant compartments with PARP inhibitors being able to distribute to both brain and CSF compartments from the systemic compartment following oral administration. The model characterized plasma, brain, and CSF drug concentrations simultaneously. A modified Simeoni model with a 1st-order tumor growth rate adequately described the dose-dependent tumor inhibition at different doses of rucaparib in the orthotopic tumor model. The effects of brain tumor inhibition by rucaparib and niraparib were sufficiently described by a simpler model with a 1st-order tumor growth rate and a 2nd-order tumor inhibition rate that was proportional to brain drug concentrations. CONCLUSIONS: Systemic and brain concentrations, as well as orthotopic and intracranial tumor inhibition data, following oral administration in mice, were adequately described using semi-mechanistic models. Further studies are warranted to understand the intracranial distribution and antitumor activity of PARP inhibitors. Citation Format: Michelle Liao, Feng Jin, Minh Nguyen, Liliane Robillard, Andrew D. Simmons, Thomas C. Harding, Jim J. Xiao. Evaluation of brain pharmacokinetics (PK) and tumor growth inhibition of PARP inhibitors in mouse xenograft models using semi-mechanistic PK/pharmacodynamic (PD) modeling [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3026.

Use of translational modeling and simulation for quantitative comparison of PF-06804103, a new generation HER2 ADC, with Trastuzumab-DM1

A modeling and simulation approach was used for quantitative comparison of a new generation HER2 antibody drug conjugate (ADC, PF-06804103) with trastuzumab-DM1 (T-DM1). To compare preclinical efficacy, the pharmacokinetic (PK)/pharmacodynamic (PD) relationship of PF-06804103 and T-DM1 was determined across a range of mouse tumor xenograft models, using a tumor growth inhibition model. The tumor static concentration was assigned as the minimal efficacious concentration. PF-06804103 was concluded to be more potent than T-DM1 across cell lines studied. TSCs ranged from 1.0 to 9.8 µg/mL (n = 7) for PF-06804103 and from 4.7 to 29 µg/mL (n = 5) for T-DM1. Two experimental models which were resistant to T-DM1, responded to PF-06804103 treatment. A mechanism-based target mediated drug disposition (TMDD) model was used to predict the human PK of PF-06804103. This model was constructed and validated based on T-DM1 which has non-linear PK at doses administered in the clinic, driven by binding to shed HER2. Non-linear PK is predicted for PF-06804103 in the clinic and is dependent upon circulating HER2 extracellular domain (ECD) concentrations. The models were translated to human and suggested greater efficacy for PF-06804103 compared to T-DM1. In conclusion, a fit-for-purpose translational PK/PD strategy for ADCs is presented and used to compare a new generation HER2 ADC with T-DM1.

Pharmacokinetic-Pharmacodynamic-Efficacy Modeling of ONO-7579, a Novel Pan-Tropomyosin Receptor Kinase Inhibitor, in a Murine Xenograft Tumor Model.

The orally available and novel small molecule ONO-7579 (N-{2-[4-(2-amino-5-chloropyridin-3-yl)phenoxy]pyrimidin-5-yl}-N'-[2-(methanesulfonyl)-5-(trifluoromethyl)phenyl]urea) is a highly potent and selective pan-tropomyosin receptor kinase (TRK) inhibitor. The objective of the present study was to characterize the pharmacokinetic (PK), pharmacodynamic (PD), and antitumor efficacy relationships of ONO-7579 in mice xenografted with a human colorectal cancer cell line, KM12 (harboring the tropomyosin 3 (TPM3) -neurotrophic tyrosine receptor kinase 1 fusion gene), via a PK/PD modeling approach. Plasma and tumor concentrations of ONO-7579, tumor levels of phosphorylated TPM3-TRKA (pTRKA), and tumor volumes in the murine model were measured with a single or multiple dose of ONO-7579 (0.06-0.60 mg/kg) administered once daily. The PK/PD/efficacy models were developed in a sequential manner. Changes in plasma concentrations of ONO-7579 were described with an oral one-compartment model. Tumor concentrations of ONO-7579 were higher than plasma concentrations, and changes in ONO-7579 tumor concentrations were described with an additional tumor compartment that had no influence on plasma concentrations. pTRKA in tumors was described with a direct Emax model, and the tumor ONO-7579 concentration causing 50% of the maximum effect was estimated to be 17.6 ng/g. In addition, a pTRKA-driven tumor growth inhibition model indicated that ONO-7579 started to sharply increase the antitumor effect at pTRKA inhibition rates >60% and required >91.5% to reduce tumors. In conclusion, the developed PK/PD/efficacy models revealed a "switch-like" relationship between pTRKA inhibition rate and antitumor effect in a murine KM12 xenograft model, demonstrating that pTRKA in tumors could serve as an effective biomarker for scheduling the dose regimen in early-stage clinical studies. SIGNIFICANCE STATEMENT: In recent years, clinical development of TRK inhibitors in patients with neurotrophic tyrosine receptor kinase fusion-positive solid tumors has been accelerated. This research found that phosphorylated TRKA was a useful biomarker for explaining the antitumor efficacy of TRK inhibitors using a pharmacokinetic/pharmacodynamic modeling approach in xenograft mice. This finding suggests a rational dosing regimen in early-stage clinical studies for ONO-7579 (N-{2-[4-(2-amino-5-chloropyridin-3-yl)phenoxy]pyrimidin-5-yl}-N'-[2-(methanesulfonyl)-5-(trifluoromethyl)phenyl]urea), a novel pan-TRK inhibitor.

Exposure-Response Analyses and Disease Modeling for Selection and Confirmation of Optimal Dosing Regimen of Isatuximab in Combination Treatment in Patients with Multiple Myeloma

Background: Isatuximab (Isa) is an anti-CD38 monoclonal antibody with multiple modes of action for killing tumor cells through direct tumor targeting and immune cell engagement. Exposure-Response (E-R) analysis and disease modelling of tumor burden were performed to evaluate the relationship between Isa exposure and efficacy outcomes. This analysis also supports the Isatuximab dosing regimen selection/confirmation when administered in combination in relapsed/refractory multiple myeloma (RRMM) patients. Methods: To determine the optimal dose for the Phase 3 ICARIA-MM study, E-R analyses were conducted in 44 and 52 RRMM evaluable patients from two Phase 1 trials of Isa in combination with pomalidomide/dexamethasone (Pd) (NCT02283775) or with lenalidomide/dexamethasone (Rd) (NCT01749969), respectively. Isa was administered intravenously at doses from 3 to 20 mg/kg every 2 weeks or weekly for 4 weeks then every 2 weeks (QW/Q2W). In the E-R analyses, several Isa exposure parameters were tested to evaluate if they were predictors of response (partial response [PR] or better). Baseline covariates were also considered in the model to evaluate potential confounding effects. In addition, a second type of analysis was performed on 2 pooled datasets of 153 (single agent and combination with Pd) and 162 (single agent and combination with Rd) evaluable RRMM patients: disease response was also captured by modeling the dynamics of the serum M-protein using a joint model of tumor growth inhibition and drop-out. Trial simulations were then performed to evaluate different dosing regimens of interest. To confirm the dose, E-R analyses were conducted in 297 evaluable RRMM patients from the phase 3 trial (ICARIA-MM; NCT02990338) comparing Isa-Pd (N=148) and Pd (N=149). Isa was administered intravenously at 10 mg/kg QW/Q2W in the Isa-Pd arm. Cox and Weibull (Progression-Free Survival [PFS]) or logistic regression (overall response rate [ORR]) models were used. Results: In the two combination Phase 1 trials, Isa appeared to be well tolerated with no clear dose response relationship between 10 and 20 mg/kg. E-R analyses suggested that higher Isa exposure represented by log Ctrough at 4 weeks (CT4W) was associated with increased ORR. Both models predicted an increased ORR with decreased beta-2 microglobulin and lower number of prior lines of therapy (Rd only), together with higher log CT4W. Clinical trial simulations showed that the probability of success to reach the targeted ORR was high with 10 mg/kg QW/Q2W for both Pd (83% for ORR ≥60%) and Rd (96% for ORR ≥50%). Serum M-protein reduction was greater at doses ≥10 mg/kg compared to doses <10 mg/kg, with minimal differences between 10 and 20 mg/kg QW/Q2W. These analyses supported the rationale for 10 mg/kg QW/Q2W selection for the Phase 3 combination study. In the Phase 3 ICARIA-MM study, the E-R analysis showed that higher CT4W was associated with higher ORR (predicted ORR below and above the median CT4W being 52% and 72%, respectively). The final model also included time since diagnosis and Revised International Staging System (R-ISS) stage at baseline and an interaction between time since diagnosis and R-ISS; ORR increased as time since diagnosis increased and decreased with higher R-ISS stage. For PFS, in addition to CT4W, the model included plasmacytomas, serum albumin level and R-ISS at baseline. The predicted distribution of PFS indicated that patients with greater Isa CT4W had longer PFS, patients with R-ISS Stage III and plasmacytomas had shorter PFS. Matched analyses to Pd indicated that even patients with concentrations ≤median CT4W benefited from Isa treatment (PFS hazard ratio: 0.773). Conclusion: Model-based drug development was successfully applied to support Phase 3 Isa dosing regimen selection for use in combination with Pd in RRMM patients. E-R and clinical data from the Phase 3 ICARIA-MM study confirmed the efficacy and safety of 10 mg/kg QW/Q2W dose/regimen selected. Disclosures Rachedi: Sanofi: Employment. Koiwai:Sanofi: Employment. Gaudel-Dedieu:Sanofi: Employment. Sebastien:Sanofi: Employment. Thai:Sanofi: Employment. El-Cheikh:Sanofi: Employment. Brillac:Sanofi: Employment. Fau:Sanofi: Employment. Nguyen:Sanofi: Employment. Liu:Sanofi: Employment. Campana:Sanofi: Employment. van de Velde:Sanofi: Employment. Semiond:Sanofi: Employment.

Quantitative mathematical modeling of clinical brain metastasis dynamics in non-small cell lung cancer

Brain metastases (BMs) are associated with poor prognosis in non-small cell lung cancer (NSCLC), but are only visible when large enough. Therapeutic decisions such as whole brain radiation therapy would benefit from patient-specific predictions of radiologically undetectable BMs. Here, we propose a mathematical modeling approach and use it to analyze clinical data of BM from NSCLC. Primary tumor growth was best described by a gompertzian model for the pre-diagnosis history, followed by a tumor growth inhibition model during treatment. Growth parameters were estimated only from the size at diagnosis and histology, but predicted plausible individual estimates of the tumor age (2.1–5.3 years). Multiple metastatic models were further assessed from fitting either literature data of BM probability (n = 183 patients) or longitudinal measurements of visible BMs in two patients. Among the tested models, the one featuring dormancy was best able to describe the data. It predicted latency phases of 4.4–5.7 months and onset of BMs 14–19 months before diagnosis. This quantitative model paves the way for a computational tool of potential help during therapeutic management.

PF645 EXPOSURE‐RESPONSE ANALYSIS & DISEASE MODELING FOR SELECTION OF OPTIMAL DOSING REGIMEN OF ISATUXIMAB AS SINGLE AGENT IN PATIENTS WITH MULTIPLE MYELOMA

Background:Isatuximab is an anti‐CD38 monoclonal antibody with multiple modes of action for killing tumor cells through direct tumor targeting and immune cell engagement [1].Aims:Exposure‐Response (E‐R) analysis and disease modelling of tumor burden were performed to evaluate the relationship between isatuximab exposure and efficacy outcome and to support dosing regimen selection for isatuximab as a single agent in relapsed/refractory multiple myeloma (RRMM) patients.Methods:The E‐R analyses were conducted in 194 RRMM patients evaluable for pharmacokinetics (PK) from two monotherapy trials, a phase 1/2 trial (NCT01084252)[2] and a phase 1 trial (NCT02514668). Isatuximab was administered intravenously at doses from 1 to 20 mg/kg once a week or every 2 weeks. Several isatuximab exposure parameters were tested and their association with the probability of achieving an objective response (CR, VGPR or PR) was examined by logistic regression modeling. Baseline covariates were also considered in the model to reduce their potential confounding effects. Disease progression was captured in a subset of 122 evaluable patients with the dynamics of the serum M‐protein and accounted for dropout using a joint model. Trial simulations were then performed to evaluate different dosing regimens of interest using both models.Results:The E‐R relationship was best described by Emax model, in which Ctrough at 4 weeks and percent of bone marrow plasma cells were significant predictors of overall response rate (ORR). Patients with bone marrow plasma cells lower than 50% were more likely to respond. For a given bone marrow plasma cell value, higher probability of response to treatment was obtained with higher Ctrough at 4 weeks. Longitudinal data of M‐protein was adequately described by a tumor growth inhibition model [3,4] and provided more insights into the response of patients over time. Therefore, a high loading dose of 20 mg/kg weekly over 4 weeks was chosen for maximizing the tumor response and a maintenance dose of 20 mg/kg every 2 weeks appeared sufficient to sustain efficacy. Clinical trial simulation demonstrated that this dosing regimen presented a probability of success of 76% to reach 30% ORR (5,000 trials with 100 patients) and would allow 52% reduction of serum M‐protein from baseline at 2 months of treatment. In addition, isatuximab appeared to be well tolerated at this dose level.Summary/Conclusion:Model‐based drug development has been successfully applied to support phase II isatuximab monotherapy dosing regimen selection in RRMM patients. This approach increases both the robustness of dose selection decision making and the chances of success for future clinical trials.

Modeling to capture bystander-killing effect by released payload in target positive tumor cells

BackgroundAntibody-drug conjugates (ADCs) are intended to bind to specific positive target antigens and eradicate only tumor cells from an intracellular released payload through the lysosomal protease. Payloads, such as MMAE, have the capacity to kill adjacent antigen-negative (Ag–) tumor cells, which is called the bystander-killing effect, as well as directly kill antigen-positive (Ag+) tumor cells. We propose that a dose-response curve should be independently considered to account for target antigen-positive/negative tumor cells.MethodsA model was developed to account for the payload in Ag+/Ag– cells and the associated parameters were applied. A tumor growth inhibition (TGI) effect was explored based on an ordinary differential equation (ODE) after substituting the payload concentration in Ag+/Ag– cells into an Emax model, which accounts for the dose-response curve. To observe the bystander-killing effects based on the amount of Ag+/Ag– cells, the Emax model is used independently. TGI models based on ODE are unsuitable for describing the initial delay through a tumor–drug interaction. This was solved using an age-structured model based on the stochastic process.Resultsβ∈(0,1] is a fraction parameter that determines the proportion of cells that consist of Ag+/Ag– cells. The payload concentration decreases when the ratio of efflux to influx increases. The bystander-killing effect differs with varying amounts of Ag+ cells. The larger β is, the less bystander-killing effect. The decrease of the bystander-killing effect becomes stronger as Ag+ cells become larger than the Ag– cells. Overall, the ratio of efflux to influx, the amount of released payload, and the proportion of Ag+ cells determine the efficacy of the ADC. The tumor inhibition delay through a payload-tumor interaction, which goes through several stages, may be solved using an age-structured model.ConclusionsThe bystander-killing effect, one of the most important topics of ADCs, has been explored in several studies without the use of modeling. We propose that the bystander-killing effect can be captured through a mathematical model when considering the Ag+ and Ag– cells. In addition, the TGI model based on the age-structure can capture the initial delay through a drug interaction as well as the bystander-killing effect.

ATIM-13. ASUNERCEPT PLUS RADIOTHERAPY IN RELAPSED GLIOBLASTOMA. UPDATE ON FIVE YEARS OVERALL SURVIVAL OF STUDY NCT01071837 AND DEVELOPMENT OF A POPULATION-PK - TUMOR GROWTH INHIBITION - SURVIVAL MODEL

Asunercept (APG101) is an Fc-fusion protein consisting of the extracellular domain of human CD95 (APO-1/Fas) and the Fc domain of IgG1. Asunercept is in clinical development for glioblastoma (GB). Based on PK-data from study NCT01071837 a population pharmacokinetic (PopPK) - tumor growth inhibition (TGI) model was developed and extended to a survival model describing the effect of radiotherapy (RT) or RT + asunercept on the overall survival (OS) of GB-patients. The objective of the model was to identify the best descriptor (i.e. asunercept exposure, tumor size determined by MRT) for survival and to quantify the effect of CpG2 methylation in the CD95 ligand promoter on patient survival. Model development was performed using non-linear mixed effects modeling with NONMEM 7.3 in a stepwise procedure. Firstly, a PopPK–TGI model was developed. Several tumor growth models were tested (e.g. exponential, sequential exponential-linear). Secondly, a survival model was developed and linked to the PopPK-TGI model. For the PopPK-TGI model, data from 84 patients were available contributing to 314 tumor measurements. Glioblastoma growth was best described by an exponential growth model with an average doubling time of 90 days. Asunercept exposure showed a significant inhibitory effect on the tumor growth rate. Tumor size was identified to significantly influence survival. Incorporation of CpG2 CD95L promotor methylation further improved the model: the survival in asunercept-treated patients was prolonged with lower CpG2 methylation status. A PopPK-TGI-Survival model for asunercept and radiotherapy treated patients was developed. A clear inhibitory effect of asunercept exposure was observable on tumor growth resulting in an increased survival. A recent update on OS of study NCT01071837 revealed that 7% of Asunercept+RT treated 2nd-line GB patients were alive after 5 years compared to 0% in patients treated with RT alone.

Abstract 4264: Mathematical modeling of differential effects of sunitinib on primary tumor and metastatic growth

Abstract Sunitinib is a drug with anti-angiogenic activity used in the treatment of patients with metastases from renal cell carcinoma or gastrointestinal tumors. However, despite clear efficacy in reducing established tumor growth, recent preclinical studies have shown limited, or even opposing, efficacies in preventing metastatic spread. In this work, we evaluated a mathematical metastatic model to describe primary tumor and metastatic dynamics in response to sunitinib in a clinically relevant mouse model of spontaneous metastatic breast cancer that develops after surgical removal of an orthotopically implanted primary tumor. Mice received either vehicle or sunitinib in the neoadjuvant (presurgical) setting according to different schedules. The experimental dataset comprises measurements of primary tumor kinetics, metastatic burden and pre-surgical molecular and cellular biomarkers, including vascular cell Ki67 and CD31 expression, circulating tumor cells (CTCs) and myeloid derived suppressor cells (MDSCs). Estimation of the model's parameters was performed using a mixed-effects population approach. To describe tumor growth under sunitinib treatment, a simple ordinary differential equation model of tumor growth inhibition was used. Population fits obtained modeling the effect of treatment only on primary tumor growth described well the experimental data of all the treated groups considered. On the contrary, simulations of treatment also on metastasis could not reproduce the behavior of the data. By inserting in the model the available biomarkers as covariates, through a Wald-test, measurements of Ki67+/CD31+, CTCs and granulocytic MDSCs were found significantly correlated with the model parameter expressing the metastatic aggressiveness of the tumor. Together, these mathematical models confirm a differential effect of sunitinib on primary (localized) tumors compared to secondary (metastatic) disease. Our results suggest that Ki67+/CD31+, CTCs and MDSCs measurements might help in predicting metastatic potential and thus aid in predicting benefit in overall survival for preoperative antiangiogenic treatments. Citation Format: Chiara Nicolò, Michalis Mastri, Amanda Tracz, John M. Ebos, Sébastien Benzekry. Mathematical modeling of differential effects of sunitinib on primary tumor and metastatic growth [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4264.

Modeling tumor growth inhibition and toxicity outcome after administration of anticancer agents in xenograft mice: A Dynamic Energy Budget (DEB) approach

Host features, such as cell proliferation rates, caloric intake, metabolism and energetic conditions, significantly influence tumor growth; at the same time, tumor growth may have a dramatic impact on the host conditions. For example, in clinics, at certain stages of the tumor growth, cachexia (body weight reduction) may become so relevant to be considered as responsible for around 20% of cancer deaths. Unfortunately, anticancer therapies may also contribute to the development of cachexia due to reduced food intake (anorexia), commonly observed during the treatment periods. For this reason, cachexia is considered one of the major toxicity findings to be evaluated also in preclinical studies. However, although various pharmacokinetic-pharmacodynamic (PK-PD) tumor growth inhibition (TGI) models are currently available, the mathematical modeling of cachexia onset and TGI after an anticancer administration in preclinical experiments is still an open issue. To cope with this, a new PK-PD model, based on a set of tumor-host interaction rules taken from Dynamic Energy Budget (DEB) theory and a set of drug tumor inhibition equations taken from the well-known Simeoni TGI model, was developed. The model is able to describe the body weight reduction, splitting the cachexia directly induced by tumor and that caused by the drug treatment under study. It was tested in typical preclinical studies, essentially designed for efficacy evaluation and routinely performed as a part of the industrial drug development plans. For the first time, both the dynamics of tumor and host growth could be predicted in xenograft mice untreated or treated with different anticancer agents and following different schedules. The model code is freely available for downloading at http://repository.ddmore.eu (model number DDMODEL00000274).

Model-Based Evaluation of Radiation and Radiosensitizing Agents in Oncology.

Radiotherapy is one of the major therapy forms in oncology, and combination therapies involving radiation and chemical compounds can yield highly effective tumor eradication. In this paper, we develop a tumor growth inhibition model for combination therapy with radiation and radiosensitizing agents. Moreover, we extend previous analyses of drug combinations by introducing the tumor static exposure (TSE) curve. The TSE curve for radiation and radiosensitizer visualizes exposure combinations sufficient for tumor regression. The model and TSE analysis are then tested on xenograft data. The calibrated model indicates that the highest dose of combination therapy increases the time until tumor regrowth 10‐fold. The TSE curve shows that with an average radiosensitizer concentration of 1.0 μg/mL the radiation dose can be decreased from 2.2 Gy to 0.7 Gy. Finally, we successfully predict the effect of a clinically relevant treatment schedule, which contributes to validating both the model and the TSE concept.

Abstract 519: Antitumor activity of MET antibody emibetuzumab (LY2875358) in combination with EGFR inhibitors in erlotinib resistant (ER) xenograft mouse models

Abstract Background: MET amplification (amp) is a resistance mechanism to EGFR TKI treatment. Emibetuzumab, a bivalent MET antibody (Ab) blocks HGF binding to MET and internalizes the receptor. Combination of emibetuzumab with EGFR TKIs (erlotinib, AZD9291, CO1686) or EGFR Ab (necitumumab, cetuximab) was evaluated in 3 ER xenograft models. Methods: Model 1: ER cell line HCC827ERL with high focal MET amp, high pMET, EGFR ex19 del (no T790M) was created from parental HCC827 NSCLC (EGFR ex19 del, EGFR amp, no MET amp) by increasing concentration of erlotinib in vitro over 7 months. Model 2: ER cell line HCC827-A8 was derived from HCC827 parental xenograft tumor serially passed in vivo with long term treatment of gefitinib and erlotinib. HCC827-A8 cells express high focal MET amp, high pMET/AXL (Western blot) while retaining EGFR ex19 del (no T790M). Model 3: LU0858 was an ER patient-derived NSCLC xenograft tumor, with focal MET amp, EGFR L858R (no T790M). MET amp and EGFRmt was determined by FISH and LNA-PCR sequencing respectively. Compound dosing: emibetuzumab 20 mg/kg qw; necitumumab 4 mg/kg or 20 mg/kg biw; cetuximab 4 mg/kg biw; erlotinib 25 mg/kg qd; 5 mg/kg AZD9291 qd; 30 mg/kg CO1686 bid. Results: EGFR inhibitors, but not emibetuzumab showed significant single agent anti-tumor effect in xenograft tumors derived from non-MET amp HCC827 parental cells. In MET amp ER models, single agent emibetuzumab resulted in tumor growth inhibition in Model 1 (T/C= 51.7%-61.0%, p<0.05) and 3 (T/C=2.8%, p<0.05)] but no tumor regression, and no anti-tumor effect in Model 2. Where evaluated, EGFR inhibitors showed no anti-tumor effect in the 3 ER models as monotherapy, except necitumumab (20 mg/kg) in Model 1 (T/C = 36.2%, p<0.05). However, combination of emibetuzumab with AZD9291, CO1686, necitumumab (20 mg/kg), or erlotinib resulted in 80.4%, 58.2%, 44.4%, 69.1% tumor regression respectively (p<0.001) in Model 1, while emibetuzumab + cetuximab (4 mg/kg) resulted in tumor stasis (T/C=0.2%, p<0.05). In Model 2, emibetuzumab + AZD9291 resulted in tumor stasis (T/C = 12.9%, p<0.05). In Model 3, emibetuzumab + necitumumab (20 mg/kg) resulted in 80.1% tumor regression (p<0.001). Conclusion: The three erlotinib resistant models with MET amp and retaining sensitizing EGFRmt (ex19 del or L858R), and no acquired T790M were found resistant to other EGFR inhibitors (Abs and TKIs). Emibetuzumab in combination with either EGFR TKI or Ab showed anti-tumor activity in MET amp ER xenograft models including tumor regression in 2 out of 3 models. The combination of emibetuzumab with erlotinib is being evaluated in NSCLC patients with EGFR activating mutation (NCT01897480). Citation Format: Suzane L. Um, Victoria L. Peek, Jennifer R. Stephens, Jessica A. Baker, Holly K. Cannon, Joel D. Cook, Isabella H. Wulur, Roger Agyei, Sudhakar Chintharlapalli, Robert J. Evans, William J. Feaver, Lysiane Huber, Linda N. Lee, Ling Liu, Liandong Ma, Ruslan Novosiadly, Volker Wacheck, Sau-Chi Betty Yan. Antitumor activity of MET antibody emibetuzumab (LY2875358) in combination with EGFR inhibitors in erlotinib resistant (ER) xenograft mouse models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 519. doi:10.1158/1538-7445.AM2017-519

Abstract 5049: Application of population pharmacokinetic and exposure-response modeling for DS-8201a, a HER2-targeting ADC, predicts 50% ORR in patients with heavily pretreated breast cancer

Abstract Background: Antibody-drug conjugates (ADC) provide a wider therapeutic index by allowing specific targeting of cytotoxic agents to tumor cells. DS-8201a is a novel HER2-targeting ADC which contains a high average drug-to-antibody-ratio (DAR) of 7-8 molecules of a topoisomerase I inhibitor (exatecan derivative, DXd) per antibody. Preclinical data demonstrated a broader antitumor activity of DS-8201a than T-DM1 (the only FDA approved HER2-targeting ADC), including efficacy against T-DM1 resistant and HER2 low-expressing tumors. The first-in-human (FIH) study, comprising a dose escalation (Part 1) and expansion (Part 2), is being conducted in patients with HER2+ breast cancer (BC), gastric cancer (GC), and other HER2 expressing tumors. This analysis was aimed to assess DS-8201a population pharmacokinetics (POPPK) and exposure-response (E-R) relationships for efficacy and key safety endpoints using data available so far, and to support Phase 2 dose selection. Methods: Preliminary data were obtained from Part 1 portion of the FIH study, in which a total of 23 patients with BC or GC received intravenous DS-8201a doses ranging from 0.8 to 8.0 mg/kg tri-weekly. Firstly, the POPPK model of DS-8201a and DXd in serum was developed and the individual-predicted exposures were derived. An exposure-driven, tumor growth inhibition (TGI) model was then fit to the longitudinal tumor sum of longest diameter (SLD) data. Additionally, the E-R relationships for thrombocytopenia (TCP) and gastrointestinal (GI) toxicity were assessed using a cell lifespan model and a logistic regression model, respectively. Results: A 2-compartment PK model with parallel linear and nonlinear elimination and a 1-compartment model with linear release and elimination best characterized the serum concentration-time profiles for DS-8201a and DXd, respectively. The TGI model for tumor SLD indicated a DS-8201a-exposure-dependent drug efficacy across the dose range tested. Model-based simulations showed that predicted % of patients exceeding pre-clinical efficacious DS-8201a concentration (4260 ng/mL) at trough was 86.7 to 95.9% during 6.4 mg/kg tri-weekly dosing, and predicted objective response rate (ORR) was approximately 50.5% in the Part 1 patient population with BC or GC. The cell lifespan modelling for TCP indicated that increased DS-8201a exposure was associated with a greater risk of grade 3 or 4 TCP but the predicted probability was low for doses of 6.4 mg/kg or below. No to minimal relationships with exposures of DS-8201a or DXd were observed for the GI toxicity. Conclusion: A series of model-based simulations demonstrated that 50% ORR was expected at 6.4 mg/kg DS-8201a tri-weekly, with limited increases in the risk of TCP and GI toxicity relative to lower doses, which supports the selection of this dose for Phase 2. Citation Format: Kazutaka Yoshihara, Seiko Endo, Kenji Tamura, Toshihiko Doi, Taro Tokui. Application of population pharmacokinetic and exposure-response modeling for DS-8201a, a HER2-targeting ADC, predicts 50% ORR in patients with heavily pretreated breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5049. doi:10.1158/1538-7445.AM2017-5049

Predictability of survival outcome by early tumor shrinkage (ETS) at 8 weeks with cutoff value of 20% in metastatic colorectal cancer (mCRC) patients treated with mFOLFOX6 plus bevacizumab (bev) or FOLFIRI plus bev as first-line chemotherapy (1st-CTx): Validation in the phase III WJOG4407G trial.

729 Background: We reported that ETS at 8 weeks with cutoff value of 20% is the best candidate for predicting overall survival (OS) in mCRC patients who received 1st-CTx in the retrospective study (Kito, ESMO, 2015), where ETS was estimated by the tumor growth inhibition model (Claret, JCO, 2013). The objective of the study was to validate our proposed evaluation time and cutoff value for ETS in the phase III WJOG4407G trial which showed equivalent efficacy of mFOLFOX6+bev and FOLFIRI+bev as 1st-CTx for mCRC (Yamazaki, Ann Oncol, 2016). Methods: Tumor shrinkage ratio (TSR), which is the relative change in the sum of the longest diameters of target lesions according to the RECIST between before and during 1st-CTx, was calculated using actually measured tumor size data. Cutoff values of TSR for ETS and allowance ranges for measurement time of tumor size were evaluated. The association between ETS and OS was investigated by Cox regression model. In multivariate analyses, the general prognostic factors of mCRC, such as ECOG performance status (PS), prior history of adjuvant chemotherapy (AC), baseline tumor size (bTS), serum alkaline phosphatase (ALP) and white blood cell (WBC), were adjusted. Results: Of 402 enrolled patients, 305 patients had tumor size data measured within 8 +- 2 weeks. Patient characteristics were: mFOLFOX6+bev/FOLFIRI+bev 153/152; PS 0/1 244/61; AC Yes/No 49/256; median bTS 79 mm (range 10-512); median ALP 274 IU/L (110-1,200); median WBC 6400/mm3(3,030–12,000). In multivariate analyses, ETS at 8 weeks with cutoff value of 20% was most significantly associated with OS showing hazard ratio (HR) of 0.53 (95%CI: 0.38-0.74, p < 0.001). Moreover, HR of ETS was 0.47 (95%CI: 0.33-0.67, p < 0.001) in using the data during 8 +- 1 weeks (n = 257), while HR was 1.03 (95%CI: 0.41-2.62, p = 0.95) in using the data earlier or later than 8 +- 1 weeks (6 to 7 or 9 to 10 weeks) (n = 48). Conclusions: This analysis supported that ETS with cutoff value of 20% using data within 8 +- 1 weeks is optimal for predicting OS in mCRC patients who received 1st-CTx. Clinical trial information: UMIN000001396.

Exposure\u2013Response and Tumor Growth Inhibition Analyses of the Monovalent Anti-c-MET Antibody Onartuzumab (MetMAb) in the Second- and Third-Line Non-Small Cell Lung Cancer

The phase III trial comparing onartuzumab + erlotinib vs. erlotinib in the second- and third-line non-small cell lung cancer (NSCLC) did not meet its primary endpoint of overall survival (OS). The objective was to assess whether doses higher than the phase III dose (15 mg/kg) might yield better efficacy without compromising the safety profile. Data were from 636 patients from the phase II and III NSCLC studies. Tumor growth inhibition (TGI) models were fit to longitudinal tumor size data to estimate individual TGI metrics including time to tumor re-growth (TTG). Cox regression models were developed for time-to-event endpoints (progression-free survival (PFS), OS, and TTG) to investigate relationships with baseline prognostic factors and onartuzumab exposure. Incidence of adverse events was modeled by logistic regression. In the final models, higher onartuzumab exposure was associated with longer PFS, but not with longer OS. Longer OS was associated with higher baseline albumin, longer TTG, smaller number of metastatic sites, female gender, lower ECOG score, and younger age. TTG was the only TGI metric retained in the final OS model. Onartuzumab exposure was not significantly associated with TTG after adjusting for prognostic factors. Higher Cmin was associated with increased incidence of infusion reactions and peripheral edema. Higher onartuzumab exposure was not significantly associated with improved OS after adjusting for prognostic factors and TTG, and there was a trend of unknown clinical significance toward increased incidence of infusion reactions and peripheral edema. These results did not support testing higher onartuzumab doses.

CD47 Monoclonal Antibody SRF231 Is a Potent Inducer of Macrophage-Mediated Tumor Cell Phagocytosis and Reduces Tumor Burden in Murine Models of Hematologic Malignancies

The transmembrane protein CD47 is an immunoglobulin superfamily member involved in multiple cellular processes, including cell migration, adhesion and T cell function. The interaction between CD47 and signal regulatory protein alpha (SIRPα), an inhibitory protein expressed on macrophages, prevents phagocytosis of CD47-expressing cells. CD47 was originally identified as a tumor antigen on human ovarian cancer and was subsequently shown to be expressed on multiple human tumor types, including both hematologic and solid tumors. In many hematologic cancers, high CD47 expression is associated with poor clinical outcomes. CD47 is also expressed at low levels on virtually all non-malignant cells, and loss of expression or changes in membrane distribution may serve as markers of aged or damaged cells, particularly on red blood cells (RBC). High expression of CD47 on tumors blocks phagocytic uptake, subsequent antigen cross-presentation and T cell activation, which collectively contribute to tumor immune evasion. Agents that block the CD47-SIRPα interaction can restore phagocytic uptake of CD47+ target cells and lower the threshold for macrophage activation, which can enhance the efficacy of therapeutic antibodies with ADCC-enabling activity. We developed and characterized a CD47 blocking antibody and evaluated its activity in multiple hematologic models.SRF231 is a fully human monoclonal antibody that binds with high affinity to human CD47 and blocks the CD47-SIRP□ interaction. SRF231 promotes macrophage-mediated phagocytic clearance of several hematologic primary tumor samples and cell lines in vitro. For example, SRF231 increases phagocytosis of Raji tumor cell line targets with an EC50 of ~300 ng/ml. Enhanced phagocytosis is preferential for tumor cells over normal leukocytes and RBC. Tumor cell phagocytosis can be enhanced in the presence of opsonizing antibodies (e.g., anti-CD20 Ab) when co-administered with SRF231.In vivo efficacy of SRF231 was assessed in several preclinical murine xenograft models of hematologic malignancies. Notably, SRF231 administration led to profound tumor growth inhibition in models of multiple myeloma, diffuse large B cell lymphoma and Burkitt's lymphoma as a single agent and in combination with opsonizing antibodies. In the Raji xenograft model, single agent therapy leads to 112% tumor growth inhibition. This anti-tumor activity is at least partially dependent on macrophages, as depletion of macrophages via clodronate administration leads to reduced tumor growth inhibition. Tumor-associated macrophage (TAM) numbers and polarization status are also modulated by SRF231 treatment.In summary, the CD47 mAb SRF231 induces robust tumor cell phagocytosis and tumor clearance both alone and in combination with opsonizing antibodies in pre-clinical models of myeloma and lymphoma. These properties warrant further development of SRF231 in hematologic malignancies. SRF231 is currently in IND-enabling studies and is expected to enter clinical trials in 2017. DisclosuresNo relevant conflicts of interest to declare.

Pharmacokinetic/pharmacodynamic modeling of etoposide tumor growth inhibitory effect in Walker-256 tumor-bearing rat model using free intratumoral drug concentrations

The purpose of this study was to establish a population pharmacokinetic/pharmacodynamic (PK/PD) model linking etoposide free tumor and total plasma concentrations to the inhibition of solid tumor growth in rats. Walker-256 tumor cells were inoculated subcutaneously in the right flank of Wistar rats, which were randomly divided in control and two treated groups that received etoposide 5 or 10mg/kg i.v. bolus every day for 8 and 4days, respectively, and tumor volume was monitored daily for 30days. The plasma and intratumoral concentrations-time profiles were obtained from a previous study and were modeled by a four-compartment population pharmacokinetic (popPK) model. PK/PD analysis was conducted using MONOLIX v.4.3.3 on average data and by mean of a nonlinear mixed-effect model. PK/PD data were analyzed using a modification of Simeoni Tumor Growth Inhibition (TGI) model by introduction of an Emax function to take into account the concentration dependency of k2variable parameter (variable potency). The Simeoni TGI-Emax model was capable to fit schedule-dependent antitumor effects using the tumor growth curves from the control and two different administered schedules. The PK/PD model was capable of describing the tumor growth inhibition using total plasma or free tumor concentrations, resulting in higher k2max (maximal potency) for free concentrations (25.8mL·μg−1·day−1 - intratumoral vs. 12.6mL·μg−1·day−1 total plasma). These findings indicate that the plasma concentration may not be a good surrogate for pharmacologically active free tumor concentrations, emphasizing the importance of knowing drug tumor penetration to choose the best antitumor therapy.

Tumor Static Concentration Curves in Combination Therapy

Combination therapies are widely accepted as a cornerstone for treatment of different cancer types. A tumor growth inhibition (TGI) model is developed for combinations of cetuximab and cisplatin obtained from xenograft mice. Unlike traditional TGI models, both natural cell growth and cell death are considered explicitly. The growth rate was estimated to 0.006 h−1 and the natural cell death to 0.0039 h−1 resulting in a tumor doubling time of 14 days. The tumor static concentrations (TSC) are predicted for each individual compound. When the compounds are given as single-agents, the required concentrations were computed to be 506 μg · mL−1 and 56 ng · mL−1 for cetuximab and cisplatin, respectively. A TSC curve is constructed for different combinations of the two drugs, which separates concentration combinations into regions of tumor shrinkage and tumor growth. The more concave the TSC curve is, the lower is the total exposure to test compounds necessary to achieve tumor regression. The TSC curve for cetuximab and cisplatin showed weak concavity. TSC values and TSC curves were estimated that predict tumor regression for 95% of the population by taking between-subject variability into account. The TSC concept is further discussed for different concentration-effect relationships and for combinations of three or more compounds.