Population exposure–response analysis of cabozantinib efficacy and safety endpoints in patients with renal cell carcinoma

Population exposure–response analysis of cabozantinib efficacy and safety endpoints in patients... Background In the phase III METEOR trial, tyrosine kinase inhibitor cabozantinib significantly improved progression-free survival (PFS), objective response rate (ORR), and overall survival compared to everolimus in patients with advanced renal cell carcinoma (RCC) who had received prior VEGFR inhibitor therapy. In METEOR, RCC patients started at a daily 60-mg cabozantinib tablet (Cabometyx™) dose but could reduce to 40- or 20-mg to achieve a tolerated exposure. Objectives and methods Exposure–response (ER) models were developed to characterize the relationship between cabo- zantinib at clinically relevant exposures in RCC patients enrolled in METEOR and efficacy (PFS and tumor response) and safety endpoints. Results Compared to the average steady-state cabozantinib concentration for a 60-mg dose, exposures at simulated 40- and 20-mg starting doses were predicted to result in higher risk of disease progression or death [hazard ratios (HRs) of 1.10 and 1.39, respectively], lower maximal median reduction in tumor size (− 11.9 vs − 9.1 and − 4.5%, respectively), and lower ORR (19.1 vs 15.6 and 8.7%, respectively). The 60-mg exposure was also associated with higher risk for selected adverse events (AEs) palmar-plantar erythrodysesthesia syndrome (grade ≥ 1), fatigue/asthenia (grade ≥ 3), diarrhea (grade ≥ 3), and hypertension (predicted HRs of 2.21, 2.01, 1.78, and 1.85, respectively) relative to the predicted average steady-state cabozantinib concentration for a 20-mg starting dose. Conclusion ER modeling predicted that cabozantinib exposures in RCC patients at the 60-mg starting dose would provide greater anti-tumor activity relative to exposures at simulated 40- and 20-mg starting doses that were associated with decreased rates of clinically relevant AEs. Keywords Cabozantinib · Exposure–response modeling · Renal cell carcinoma Introduction improved understanding of the molecular biology of this disease, including the involvement of pathways linked to the Renal cell carcinoma (RCC) accounts for approximately vascular endothelial growth factor receptor (VEGFR), mam- 2–3% of all malignancies in adults with about one-third of malian target of rapamycin (mTOR), and the programmed patients having metastatic disease at diagnosis [1–3]. Recent cell death (PD-1) receptor [4–6]. Therapeutic approaches for advances in the treatment of RCC have been made based on treatment of RCC include the VEGF antibody bevacizumab, mTOR inhibitors temsirolimus and everolimus, the PD-1 checkpoint inhibitor nivolumab, and tyrosine kinase inhibi- tors (TKI) such as sunitinib, pazopanib, sorafenib, levatinib, Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s0028 0-018-3579-7) contains and axitinib [7]. Although these therapies were significant supplementary material, which is available to authorized users. advancements in the treatment of RCC, disease progres- sion is common as resistance to these treatments eventually * Steven Lacy develops. slacy@exelixis.com Cabozantinib is an inhibitor of receptor tyrosine kinases Exelixis Inc., 210 East Grand Avenue, South San Francisco, including VEGFR2, and the tyrosine kinases MET (hepat- CA 94080-0511, USA ocyte growth factor receptor) and AXL (GAS6 receptor) Ann Arbor Pharmacometrics Group, Inc., Ann Arbor, MI, implicated in development of resistance to RCC therapy USA Vol.:(0123456789) 1 3 1062 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 [8–10]. In the pivotal phase III METEOR study, cabozan- endpoint, and overall survival (OS) and objective response tinib improved overall survival, decreased disease progres- rate (ORR) as secondary endpoints. A total of 658 patients sion, and increased objective response in patients with were randomized 1:1 to receive cabozantinib (n = 330) or advanced RCC who had received prior VEGFR-TKI treat- everolimus (n = 328). Randomization was stratified by num- ment [11]. The cabozantinib tablet formulation (Cabome- ber of prior VEGFR-TKI therapies and number of risk fac- tyx™) is approved at a 60-mg-free-base equivalent (FBE) tors per Memorial Sloan-Kettering Cancer Center (MSKCC) daily dosage for the treatment of patients with advanced criteria. Patients were ≥ 18 years of age with advanced or RCC in USA who have received prior anti-angiogenic ther- metastatic RCC with a clear-cell histology and measurable apy and in the European Union (EU) following prior VEGF- disease per RECIST. To manage AEs, dose modifications targeted therapy, with dosage adjustments to 40-mg FBE and were allowed that included dose interruptions and reduc- then 20-mg FBE permitted to manage adverse events (AEs) tions. The dose of cabozantinib could be reduced to 40-mg [12, 13]. In METEOR, 60% of patients treated with cabozan- and then to 20-mg from the starting dose of 60-mg. The dose tinib had at least one dose reduction and 70% required dose of everolimus could be reduced to 5-mg and then to 2.5-mg. modification (i.e., dose interruption, reduction, or increase). Radiographic assessments were performed at screening and The most frequent AEs leading to dose reduction were diar- every 8 weeks for the first year, and every 12 weeks thereaf- rhea (16%), palmar-plantar erythrodysesthesia syndrome ter. Safety was assessed in all patients who received at least (PPES; 11%), fatigue (10%), and hypertension (7.6%). one dose of study drug every 2 weeks for the first 8 weeks Cabozantinib capsule formulation (Come triq ) is and every 4 weeks thereafter. At the discretion of the Inves- approved at a dose of 140-mg FBE in USA for treatment of tigator, treatment could be continued after radiographic pro- patients with progressive, metastatic medullary thyroid can- gression. One blood sample for plasma cabozantinib con- cer (MTC), and in the EU for the treatment of progressive, centration determinations was collected at approximately 8 unresectable locally advanced or metastatic MTC, with dose or more hours after the prior evening’s dose on day 29 and reductions to 100-mg FBE then to 60-mg FBE to manage day 57 of the study. AEs [14, 15]. Exposure–response (ER) modeling in MTC patients showed an increased risk of time to the first dose Bioanalytical methods modification to be highly correlated with lower cabozantinib apparent clearance (CL/F) values; however, no clear associa- Plasma cabozantinib concentrations were measured using a tion was observed between time to first dose modification validated liquid chromatographic–tandem mass spectrom- and progression-free survival (PFS) [16]. In an integrated etry method. The lower limit of quantitation was 0.5 ng/mL population PK (popPK) analysis, MTC cancer type was [18]. shown to be a statistically significant covariate on cabozan- tinib CL/F, with values approximately twofold higher rela- Software and modeling strategy tive to patients with other types of malignancies (including RCC) [17]. Several possible factors may underlie the higher Time-to-event Cox proportional hazard (PH) models were cabozantinib clearance observed in MTC patients, including developed using SAS Version 9.3 (SAS Institute Inc., Cary differences in the incidence or severity of treatment-related NC). Longitudinal tumor size and time-to-event dose modifi- diarrhea or hypocalcemia, or use of concomitant medica- cation models were developed using NONMEM version 7.3 tions; however, an exact cause has yet to be identified [17]. (ICON Development Solutions, Ellicott City, MD). The sto- The current ER analyses were thus undertaken to better chastic approximation expectation maximization (SAEM)/ understand the relationship between cabozantinib exposure importance sampling (IMP) estimation method was used for and efficacy, safety, and the need for dose adjustments in the tumor size model and the Laplacian estimation method RCC patients. was used for the repeated time-to-event dose modification model. Differences in objective function values (OFVs) between competing models as well as standard goodness-of- Methods fit plots were used to identify the best fitting model among those evaluated. Study design and data Population pharmacokinetic model The ER analyses were conducted utilizing data from the phase III METEOR study of cabozantinib in patients with A popPK model was developed from nine clinical stud- RCC [11]. METEOR was a multicenter, randomized, con- ies, comprised of 1534 subjects. The detailed method for trolled trial of cabozantinib [60-mg FBE once a day (QD)] the popPK model development is described in a separate versus everolimus (10-mg QD), with PFS as the primary manuscript [17]. Individual exposures for subjects in the 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1063 METEOR study were predicted from this popPK model for importance of selected cabozantinib exposures on the rate use as the exposure metrics for the present ER analysis. of events in the survival/probability scale. Longitudinal sum of tumor diameter model Exposure response analysis for time‑to‑event endpoints Nonlinear mixed-effects modeling was used to develop a model to describe the relationship between longitudinal sum Progression‑free survival and safety endpoints of tumor diameter measurements and C . The sum of tumor avg diameter over time is described in Eq. 4 [19]: Time-to-event analyses were performed to characterize dY∕dt = Growth − Drug effect, (4) the ER relationship between cabozantinib exposure and where dY∕dt is the change in tumor diameter over time, clinical endpoints including PFS and six safety endpoints. and growth represents the increase of tumor diameter over The number of patients with events and the total num- time, which is independent of any drug effects. Drug effect ber of patients at risk are listed in Supplemental Table 1. represents the first-order drug induced decay. Linear and The safety endpoints evaluated included fatigue/asthenia nonlinear relationships were evaluated between cabozantinib (grade ≥ 3), PPES (grade ≥ 1), nausea or vomiting, diarrhea and decay rate. In addition, models accounting for resistance (grade ≥ 3), hypertension (systolic BP > 160 mmHg or dias- were also tested. tolic BP > 100 mmHg), and stomatitis (grade ≥ 3). The model with the best fit to the data had a first-order The extended Cox PH model was used to describe the growth rate, nonlinear cabozantinib drug ee ff ct, and a resist - relative hazard for all endpoints (efficacy, safety, and dose ance component. The model included exponential error modification) and to allow for time-varying cabozantinib models for inter-individual variability (IIV) and an additive exposure. The Cox PH model is represented by the equation: error model for residual variability. The model is defined h(t, X(t)) = h (t) ⋅ exp( ⋅ X(t)), (1) in Eq. 5: where h(t, X(t)) denotes the hazard at time t, h (t) is the −k ⋅t tol background hazard function, β is a vector of the regression ((k + k ⋅ e ) ⋅ C dmax dmaxtot avg dY = k ⋅ Y − ⋅ Y, coefficients, and X (t) is a matrix of covariates which may grow dt (EC + C ) 50 avg vary with time. A Cox PH model was fit for each endpoint (5) and three possible cabozantinib exposure measures: average where dY/dt is the change in tumor diameter per unit time, concentration (C ) over a pre-specified time interval rang- avg k is the first-order growth rate constant, k is the grow dmax ing from 1 day to 3 weeks prior to time t or from time 0 to maximum non-attenuating drug induced tumor decay rate, time t or area-under-the-plasma-concentration–time-curve k governs the maximum loss in the decay rate due to dmaxtot from time 0 to time t (AUC0 ). 0–t resistance, k is the rate constant which governs the rate of tol The impact of cabozantinib exposure on the relative haz- attenuation, EC is the cabozantinib concentration yielding ard was evaluated during base model development. Both one-half of the current tumor decay rate, and C is the indi- avg linear (Eq. 2) and nonlinear (Eq. 3) functional forms were vidual predicted daily average cabozantinib concentration. evaluated: Dose modification h(t, X (t)) = h (t) ⋅ exp( ⋅ X (t)), (2) ex o ex1 ex X (t) ex � � � The relationship (relative risk) between individual predicted h(t, X (t)) = h (t) ⋅ exp( ⋅ X (t)); X (t)= , o ex2 ex ex ex X (t)+ EC cabozantinib CL/F and the rate of dose modifications (i.e., ex 50 (3) reductions or interruptions) was assessed using the linear where X (t) or X′ (t) is time-varying cabozantinib exposure model (Eq. 2) and the nonlinear model(s) (Eq. 3). As CL/F ex ex measure, β represents the slope in the log-linear model, was log normally distributed, a linear model using log-trans- ex1 β represents the maximum drug effect in the E model, formed CL/F was also evaluated. ex2 max and EC represents a range of fixed values for the exposure Repeated time-to-event analyses were also performed at which half of the maximal effect is achieved. to characterize the ER relationship between cabozantinib Covariate effects were evaluated only for PFS using the C and the dose modification of any kind (DMAK) that avg SELECTION = SCORE option in the PHREG procedure considered dose escalations, reductions, and interruptions within SAS. Schwarz’s Bayesian Criterion (SBC) was cal- necessary for realistic simulations (Eq. 6). culated to find the most parsimonious or final model. A The date of the first dose of study medication repre- summary of the covariates evaluated is provided in Sup- sented “time 0” in these analyses. C was used in these avg plemental Table 2. The final model was used to evaluate the models. The number of events in a particular patient 1 3 1064 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 ranged from 0 to 52. The hazard for the DMAK model is Cox proportional hazard models defined by the following equation: for progression‑free survival 𝜆 = exp(𝜃 + 𝜃 ⋅ C ){if dose > 0}, base drug avg (6) A total of 315 patients were included in PFS analysis. Dif- ferent cabozantinib exposure measures were fit initially to = exp( −  ){if dose = 0}, base base-hold a linear Cox PH model; the time-varying average cabozan- where  represents the baseline log hazard,  , the base drug tinib concentration calculated over the 3 weeks prior to the change in log hazard per unit cabozantininb concentration, event t (C ) resulted in the lowest partial (negative 2 avg3w and  , the baseline log hazard for dose hold. The haz- base-hold log) likelihood (− 2LL) value and was the only statistically ard was dependent on whether a patient was currently on a significant exposure measure (p < 0.001). The relationship dose interruption (i.e., dose = 0). Cabozantinib exposure was between cabozantinib C and PFS was further assessed avg3w tested on the hazard for dose interruption, but this did not using nonlinear models over a range of EC values for result in a reduction in the OFV and was not included in the C (50–300 ng/mL). The nonlinear models resulted in avg3w final model. The baseline hazard during dose interruptions statistically significant reductions in − 2LL compared to is larger than the hazard during active treatment indicating the linear model; an EC value of 100 ng/mL resulted that there is an increased risk of a dose modification dur - in the best model fit and was used to illustrate the rela - ing dose interruptions. This is consistent with the relatively tionship between disease progression and cabozantinib short duration of the majority of dose interruptions. When concentrations. Figure 1 illustrates the impact of selected subjects are not on a dose interruption, increases in cabo- cabozantinib exposure values on the predicted survival zantinib concentration increase the instantaneous risk for a curves for PFS. These curves show the predicted fraction dose modification. The parameter estimates for the DMAK of subjects without disease progression or death. The typi- model are shown in Supplemental Table 3. cal individual predicted average steady-state cabozantinib concentration for 20-mg (375  ng/mL), 40-mg (750  ng/ mL), and 60-mg (1125  ng/mL) QD doses was used for Simulations exposure in the predictions and the predictions did not account for dose holds. Relative to reference steady-state Simulations were performed to compare the effects of plasma concentration at 60-mg, cabozantinib concentra- cabozantinib on longitudinal tumor size for a starting tions at lower simulated starting doses of 40- and 20-mg dose of 60-, 40-, and 20-mg. First, predictions for DMAK yielded hazard ratios (HRs) (95% lower–upper confidence analysis were based on Monte Carlo simulations (1000 limits) for disease progression or death [1.10 (1.07–1.12) patients) for trial duration of 365 days. Uncertainty in the and 1.39 (1.29–1.49), respectively] indicating an increased predictions was not computed. C was calculated interac- avg risk with decreasing plasma drug concentrations. tively based on dosing change predicted from the DMAK The most parsimonious model which was selected model. The tumor model was used to predict the change in as the final model contained covariates of cabozantinib tumor size over time for the 1000 subjects at 60-, 40-, and C , baseline Eastern Cooperative Oncology Group avg3w 20-mg starting dose groups. The median tumor diameter (ECOG) score ≥ 1, baseline sum of tumor diameter greater was tabulated by day for each treatment group. than the median, liver metastasis, high MET immuno- histochemistry (IHC) status, and elapsed time less than 3 months before progressive disease on prior TKI therapy. The parameter estimates for the final PFS model are listed Results in Supplemental Table 4. The impact of each covariate is pictured in Fig.  2. Patients with increasing cabozan- Population pharmacokinetic model tinib concentrations were predicted to have a decreased risk of progressive disease or death. Patients with baseline A two-compartment model with the first-order elimina- ECOG score ≥ 1, baseline sum of tumor diameter above tion and a dual absorption (first order + zero order) process median, liver metastasis, and high MET IHC status were adequately described the observed cabozantinib concen- predicted to have a higher baseline hazard ratio of progres- tration data. The results of the model and the covariates sive disease or death, but these effects were largely offset analyses are described in a separate manuscript [17]. The by higher cabozantinib exposure. At higher cabozantinib predicted exposure measures used in the ER analyses in concentrations, the HR for PFS was similar with or with- this report were derived from the post hoc PK parameters out these covariates. from this popPK model. 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1065 Fig. 1 Predicted survival curves for progression-free survival for selected values of average cabo- zantinib concentration. Typical individual predicted steady-state average cabozantinib concen- tration for the 20-mg (black), 40-mg (blue), and 60-mg (red) doses are 375, 750, and 1125 ng/mL, respectively. The solid line represents the fraction of subjects at each dose level without progress disease or death over time. The shaded areas represent 95% confidence intervals Fig. 2 Comparison of predicted hazard ratio for different covariate equal to 3 months for the previous TKI therapy. EC drug exposure effects for final progression-free survival model. The reference haz- producing 20% of the maximum effect, ECOG Eastern Cooperative ard represents an average cabozantinib concenration equal to 25  ng/ Oncology Group, MET IHC hepatocyte growth factor receptor protein mL (EC ), an ECOG score equal to zero, a baseline sum of tumor immunohistochemistry, SOD sum of diameters, TKI tyrosine kinase diameter that is below the median, no liver metastasis, an MET IHC inhibitor, PD progressive disease status equal to low, and a time to progressive disease greater than or 1 3 1066 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 HRs (95% lower–upper confidence limits) for PPES, fatigue/ Longitudinal tumor growth asthenia, hypertension, and diarrhea were 2.21 (1.60, 3.06), 2.01 (1.22, 3.31), 1.85 (1.33, 2.57), and 1.78 (1.08, 2.91), A total of 319 patients with 1637 evaluable tumor diameter measurements were included in the analysis. The parameter respectively, based on the predicted steady-stage average cabozantinib concentration for a 60-mg dose relative to estimates for the final tumor growth model is in Supplemen- tal Table 3. The attenuation half-life was 25.6 days for a typi- a 20-mg dose. For these respective AEs, lower predicted HRs (95% lower–upper confidence limits) were also evident cal patient, indicating that this component of tumor decay rate was near zero by 128 days. The EC is 251 ng/mL and based on the predicted steady-stage average cabozantinib concentration for a 60-mg dose relative to a 40-mg dose [i.e., the EC value is about 1000 ng/mL, suggesting that the rec- ommended 60-mg daily dosage was near the plateau of the 1.49 (1.27, 1.75), 1.42 (1.11, 1.82), 1.36 (1.15, 1.60), and 1.33 (1.04, 1.70)]. Statistically significant ER relationships dose–response curve as cabozantinib C for a 60-mg QD avg dose is 1125 ng/mL. Baseline tumor size variability between were not found for nausea/vomiting (grade ≥ 3) or stomati- tis (grade ≥ 3), but the frequencies of events for these two patients was large (CV = 72%). endpoints were small (n = 16 for nausea/vomiting and n = 10 for stomatitis). Cox proportional hazard models for safety endpoints Dose modifications For safety endpoints fatigue/asthenia, PPES, nausea/vom- 317 patients were included in the analysis of dose modi- iting, and diarrhea, the time-varying average cabozan- tinib concentration over 2 weeks prior to time t (C ) fications. The − 2LL for the linear model using log-trans- avg2w formed CL/F (log-linear CL/F model) was lower than the resulted in the lowest − 2LL. For hypertension and stoma- titis, the time-varying average concentration over the 24 h linear CL/F model and was similar to the − 2LL for the best nonlinear model. Residual diagnostics were similar for the prior to time t (C ) and concentration calculated from avg1d time zero to time equal to t (C ) resulted in the low- linear model using log CL/F and the best nonlinear model. avg0t Overall, the log-linear CL/F was selected over the best non- est − 2LL, respectively. Parameter estimates for the final models for each of these AEs are shown in Supplemental linear model. A statistically significant relationship was identified Table 4. An increase in average cabozantinib concentra- tions was associated with increased risk of PPES (grade ≥ 1; between individual predicted cabozantinib CL/F and the relative risk of dose modifications (p value < 0.0001). The Fig.  3), fatigue/asthenia (grade ≥ 3), hypertension (sys- tolic blood pressure > 160 mmHg or diastolic blood pres- parameter estimate for the dose modification model [β CL/F (SE) = − 1.27033 (0.177930)], indicating that the log HR sure > 100 mmHg), and diarrhea (grade ≥ 3). The predicted Fig. 3 Predicted accumula- tive hazards for palmar-plantar erythrodysesthesia (PPE) syndrome at specific, constant average cabozantinib concen- trations. Typical individual predicted steady-state average cabozantinib concentration for the 20-mg (black), 40-mg (blue), and 60-mg (red) doses is 375, 750, and 1125 ng/mL, respectively. The solid line rep- resents the accumulative hazard for PPE at each dose level over time. The shaded areas repre- sent 95% confidence intervals 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1067 Fig. 4 Predicted fractions of subjects without dose modi- fication for selected values of cabozantinib apparent clear- ance. The solid black line (shaded black areas represent 95% CI) represents the frac- tion of subjects without dose modification over time for CL/F of 1.3 L/h, the solid blue line (shaded blue areas represent 95% CI) represents the fraction of subjects without dose modification over time for CL/F of 2.3 L/h, and the solid red line (shaded red areas represent 95% CI) represents the fraction of subjects without dose modification over time for CL/F of 3.3 L/h. The hazard ratio of 0.281 (p < 0.0001) indicates 0.281 times less risk of dose modification for one unit increase of ln(CL) decreases with increasing log CL/F, was used to compute the data. Slight differences between the observed and simulated relative HR for different values of CL/F. Relative to a CL/F data sets may reflect the lack of a drop-out (based on pro- of 2.3 L/h, the HR (95% CI) for risk of dose modification for gressive disease or death) in the modeled analysis. Relative a lower CL/F value of 1.3 L/h was approximately two times to a 60-mg simulated starting dose, a lower percentage of greater [2.07 (1.60, 2.52)]. Figure 4 illustrates the impact of subjects were predicted to have dose-reduced at a 40-mg selected cabozantinib CL/F values on the predicted survival simulated starting dose at both the 6-month treatment time- curves; these curves show the predicted fraction without point (24 vs 45%) and 12-month treatment time-point (37 dose modifications over time for CL/F values of 1.3, 2.3, vs 64%). At 6 months, approximately 50% of subjects in the and 3.3 L/h. 60-mg starting dose group (observed and simulated) are still on the 60-mg dose. Simulations Next, the tumor growth model was used to simulate the time course of tumor diameters for each of the 1000 patients First, the dose modification model DMAK was used to sim- in the 60-, 40-, and 20-mg starting dose treatment groups. ulate longitudinal C for 1000 patients over a 12-month Patients in the 20- and 40-mg starting dose treatments avg period based on the changing events. To mimic dose change groups were predicted to have a smaller reduction (median scenarios in the observed data set, at the time of each simu- change from baseline = − 4.45 and − 9.1%, respectively) in lated event, the observed probability of a dose reduction, tumor size relative to the 60-mg (–11.9%) starting dose treat- interruption, or escalation based on the current dose was ment group (Fig. 5). used. If the current dose was 0 (representing a dose inter- To further assess the clinical relevance of the difference ruption), the observed probability of escalating to 20-, 40-, in tumor reduction between starting doses, the response to or 60-mg given the dose prior to the interruption was used. treatment was computed at baseline and every 8 weeks for The percentage of patients that were predicted to be on the 1 year using the longitudinal sum of tumor diameter pre- 20-mg, 40-mg, and 60-mg dosages after 6 and 12 months for dictions. The responses [complete response (CR), partial the observed data set for 60-mg and for the simulated 60-mg response (PR), stable disease, and progressive disease (PD)] and simulated 40-mg starting dose data sets are shown in were calculated using the criteria outlined in the METEOR Supplemental Table 5. study. In the simulated 60-mg starting dose group, a higher The simulated 60-mg starting dose yielded similar per- percentage of patients were predicted to achieve objective centages of patients that dose reduced to 40- or 20-mg after response (CR + PR) relative to the 40- and 20-mg starting 6 and 12 months on cabozantinib treatment. In general, the dose groups (19.1, 15.6, and 8.7%, respectively), whereas a simulated data set for 60-mg starting dose showed a similar lower percentage of patients were predicted to have PD (7.5, pattern for dose modification as the observed 60-mg dose 8.1, and 10.2%, respectively; Table 1). 1 3 1068 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 Fig. 5 Comparison of predicted median percent change from baseline tumor diameter for 20-, 40-, and 60-mg cabozantinib simulated starting doses Table 1 Percentage of simulated subjects (N = 1000) achieving each average cabozantinib concentrations of the modeled 20-, best overall response category 40-, and 60-mg dose levels (375, 750 and 1125 ng/mL, respectively). The nonlinear relationship was reflected in Best overall response 20-mg start- 40-mg start- 60-mg start- ing dose (%) ing dose (%) ing dose (%) the marginally increased HRs (1.10 and 1.39) for risk of progressive disease or death for simulated concentrations Complete response 0.10 0.00 0.00 at starting doses of 40- and 20-mg, respectively, vs 60-mg. Partial response 8.60 15.6 19.1 Covariates that were associated with an increase in rate Stable disease 81.1 76.3 73.4 of disease progression were baseline ECOG score of ≥ 1, Progressive disease 10.2 8.10 7.50 baseline tumor diameters above the median, presence of liver metastasis, MET IHC status designated as high, and having a < 3-month time elapsed before disease progression with prior TKI therapy. At clinically relevant cabozantinib Discussion concentrations, the effects of the covariates decreased, such that the HRs generally reflected primarily drug effect on In the phase III METEOR study, cabozantinib demon- PFS. However, subjects that had progressive disease prior strated significantly improved PFS [HR 0.51 (95% CI to 3 months on the previous TKI therapy showed a higher 0.41–0.62); median 7.4 vs 3.9 months; p < 0.0001], ORR HR compared to subjects that had progressive disease after [17% (13–22) vs 3% (2–6); p < 0.0001], and OS [HR 0.66 3 months on the previous TKI therapy for the same cabozan- (0.53–0.83); median 21.4 vs 16.5 months; p = 0.0003] ver- tinib plasma concentrations and appeared to be minimally sus everolimus in patients with advanced RCC who had affected by increasing cabozantinib plasma concentration. received prior VEGFR-TKI therapy [11]. A high percent- The nonlinear mixed-effect tumor growth model age (~ 60%) of RCC patients treated with cabozantinib in developed using target lesion tumor diameter measure- METEOR had at least one dose reduction from the 60-mg ments from RCC patients administered cabozantinib in FBE dose (to 40- or 20-mg FBE) in response to treatment- METEOR yielded an estimated EC value (251 ng/mL) emergent AEs. ER models were thus developed to describe 50 lower than the predicted steady-state average cabozantinib the relationship between cabozantinib exposures at doses concentrations for the starting dose levels of 60-, 40-, and evaluated in METEOR and measures of efficacy and safety 20-mg (1125, 750, and 375  ng/mL, respectively). The (AEs and the need for dose modifications) using data in corresponding predicted median percent change of tumor RCC patients enrolled in the pivotal phase III study. size from baseline (− 11.9, − 9.1, and − 4.5% respectively) A statistically significant relationship was identified and predicted ORR (19.1, 15.6, and 8.7%, respectively) between the rate of PFS (progressive disease or death) indicates the 60-mg cabozantinib starting dose which and the time-varying average cabozantinib concentration. provides relatively greater anti-tumor activity compared Increases in cabozantinib concentration were predicted to those predicted for lower simulated starting doses of to decrease the rate of PFS in a nonlinear manner, with 40- and 20-mg. These findings in RCC patients are con- an EC value (100 ng/mL) by the best nonlinear (E ) 50 max sistent with those of the nonlinear mixed-effect models model markedly lower than the predicted steady-state developed previously to describe the relationship between 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1069 cabozantinib exposure and target lesion tumor size in the Conclusions phase III EXAM study of patients with progressive meta- static MTC: the estimated EC values (range 58–79 ng/ In ER models, a 60-mg simulated starting dose resulted mL) were lower than the steady-state cabozantinib con- in improved PFS, reduced tumor growth, and increased centration at the 140-mg dose level (1640 ng/mL), and no ORR compared to a 40- or 20-mg simulated starting dose. marked decrease in target lesion regrowth was predicted as However, higher cabozantinib exposures resulting from a consequence of the two protocol-defined dose reductions lower cabozantinib CL/F are predicted to increase the to 100- and 60-mg [20, 21]. rate of dose modification, while reducing cabozantinib An increase in cabozantinib concentration was asso- exposure with dose reduction is projected to decrease the ciated with an increased risk of AEs fatigue/asthenia risk of individual clinically relevant AEs in RCC patients. (g r ade ≥ 3), PPES (g r ade ≥ 1), hyper tension (syst olic Overall, the ER analysis predicts that cabozantinib would BP > 160 mmHg or diastolic BP > 100 mmHg), and diar- be effective at the 60-mg starting dosage evaluated in rhea (grade ≥ 3), with predicted HRs for these AEs approx- METEOR as well as daily dosages of 40- and 20-mg that imately 1.4- and twofold higher at the predicted steady- resulted from dose reduction. state average cabozantinib concentration for a 60-mg dose Acknowledgements The authors wish to respectfully thank the medical relative to a 40- or 20-mg dose, respectively. Cabozantinib professionals and patients that participated in the clinical trial of cabo- showed moderately high inter-individual variability for zantinib reported in this manuscript. The authors also wish to acknowl- CL/F in subjects with various tumor types including RCC edge Susan K. Paulson, Ph.D., for her assistance in the preparation based on popPK modeling (percent coefficient of varia- of this manuscript. This assistance was fully funded by Exelixis, Inc. tion = approximately 46%  [22]); thus, the 60-mg starting Funding The studies described in this manuscript were supported by dose would provide RCC subjects with higher cabozan- Exelixis, Inc. tinib CL/F (and lower exposures relative to subjects with lower CL/F) the opportunity to achieve therapeutic con- Compliance with ethical standards centrations. For RCC subjects with lower cabozantinib CL/F, dose modification to 40- or 20-mg provides the Conflict of interest Steven Lacy and Linh Nguyen are stockholders opportunity to achieve a tolerated plasma exposure that and current employees of Exelixis, Inc. Dale Miles was an employee of also yields acceptable clinical activity. Exelixis, Inc. when this work was performed and is currently employed by Genentech, Inc. Jace Nielsen, Bei Yang, and Matt Hutmacher are Treatment-emergent AEs frequently lead to dose modifi- employees of Ann Arbor Pharmacometrics Group (A2PG), Inc. who cations in RCC patients enrolled in METEOR. Patients were designed and conducted the ER modeling reported in this manuscript permitted to modify or interrupt treatment at non-uniform and was funded by Exelixis, Inc. Bei Yang is currently employed by times over the course of the METEOR study, so a repeated Luoxin Biotechnology (Shanghai) Co., Ltd. The authors contributed significantly to the design, conduct, analyses, and interpretation of the time-to-event model was developed that incorporated all data, and were involved in the preparation, review, and approval of this dose changes (dose holds, reductions and increases) over manuscript. time and was used to calculate time-varying cabozantinib exposure based upon the dose changes. Ethical approval The clinical studies were conducted in accordance with the World Medical Association Declaration of Helsinki, the Inter- While simulations showed that the 20-, 40-, and 60-mg national Conference on Harmonisation Tripartite Guideline for Good cabozantinib starting dosages were all predicted to reduce Clinical Practice, and all applicable local regulations. Study protocols tumor growth, the 60-mg dose resulted in the greatest reduc- and informed consent documents were reviewed and approved by the tion in tumor growth, best ORR, and lowest rate of PD. In Institutional Review Board (IRB) of participating institutions, and informed consent was obtained from all participants before any study- the review of the New Drug Application for cabozantinib for specified procedures were undertaken. the treatment of patients with RCC [23], the Food and Drug Administration (FDA) addressed the issue regarding the Open Access This article is distributed under the terms of the Crea- appropriateness of dose selection given the high percentage tive Commons Attribution 4.0 International License (http://creat iveco of dose reductions in METEOR. Based on their review, FDA mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- concluded that: (1) most adverse reactions were successfully tion, and reproduction in any medium, provided you give appropriate managed with dose interruptions and supportive measures; credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. (2) ER modeling indicated that lower starting doses could possibly compromise activity of the drug with decreased response rates; and (3) the dose selection of 60-mg daily was adequate based on ER analyses and a safety profile that is acceptable for the patient population. The cabometyx label specifies dose reduction instructions for the RCC popula- tion [12]. 1 3 1070 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 14. Cometriq (cabozantinib) Capsules (2012) US prescribing References information. Exelixis, Inc., South San Francisco. https ://www. acces sdata .fda.gov/drugs atfda _docs/label /2012/20375 6lbl.pdf. 1. Gupta K, Miller JD, Li JZ et al (2008) Epidemiologic and socio- Accessed 5 Apr 2018 economic burden of metastatic renal cell carcinoma (mRCC): a 15. Cometriq (cabozantinib) Capsules (2016) Summary of product literature review. Cancer Treat Rev 34:193–205 characteristics. Ipsen Pharma, Boulogne-Billancourt, France. 2. Rini BI, Campbell SC, Escudier B (2009) Renal cell carcinoma. https ://www .medic ines.or g.uk/emc/pr odu ct/4408. Accessed 5 Lancet 373:1119–1132 Apr 2018 3. Siegel R, DeSantis C, Virgo K et al (2012) Cancer treatment 16. Miles D, Jumbe NL, Lacy S et al (2016) Population pharmacoki- and survivorship statistics, 2012. CA Can J Clin 62:220–241 netic model of cabozantinib in patients with medullary thyroid 4. Greef B, Eisen T (2016) Medical treatment of renal cancer: new carcinoma and its application to an exposure–response analysis. horizons. Br J Cancer 115:505–516 Clin Pharmacokinet 55(1):93–105 5. Batelli C, Cho DC (2011) mTOR inhibitors in renal cell carci- 17. Lacy S, Yang B, Nielsen J et al The population pharmacokinet- noma. Therapy 8:359–367 ics of cabozantinib in healthy volunteers and patients with vari- 6. Motzer RJ, Escudier B, McDermott DF et al (2015) Nivolumab ous tumor types. Cancer Chemother Pharmacol (manuscript versus everolimus in advanced renal-cell carcinoma. N Engl J Med accepted for publication) 373:1803–1813 18. Lacy S, Hsu B, Miles D et al (2015) Metabolism and disposition 7. Powles T, Staehler M, Ljungberg B et al (2016) Updated EAU of cabozantinib in healthy male volunteers and pharmacologic guidelines for clear cell renal cancer patients who fail VEFG tar- characterization of its major metabolites. Drug Metab Dispos geted therapy. Eur Urol 69:4–6 43:1190–1207 8. Yakes FM, Chen J, Tan J et al (2011) Cabozantinib (XL184), a 19. Ribba B, Holford NH, Magni P et al (2014) A review of mixed- novel MET and VEGFR2 inhibitor, simultaneously suppresses effects models of tumor growth and effects of anticancer drug metastasis, angiogenesis and tumor growth. Mol Cancer Ther treatment used in population analysis. CPT Pharmacometr Syst 19(12):2298–2308 Pharmacol 3:e113 9. Rankin EB, Fuh KC, Castellini L et al (2014) Direct regulation of 20. Miles DR, Wada DR, Jumbe NL et al (2016) Population phar- GAS6/AXL signaling by HIF promotes renal metastasis through macokinetic/pharmacodynamic modeling of tumor growth kinet- SRC and MET. Proc Natl Acad Sci USA 111(37):13373–13378 ics in medullary thyroid cancer patients receiving cabozantinib. 10. Bommi-Reddy A, Almeciga I, Sawyer J et  al (2008) Kinase Anticancer Drugs 27(4):328–341 requirements in human cells: III. Altered kinase requirements in 21. Center for Drug Evaluation and Research (CDER) (2012) Clinical VHL−/− cancer cells detected in a pilot synthetic lethal screen. pharmacology and biopharmaceutics review[s] for cabozantinib Proc Natl Acad Sci USA 105(43):16484–16489 [COMETRIQ]. http://www.acces sdata .fda.gov/drugs atfda _docs/ 11. Choueiri TK, Escudier B, Powles T et al (2016) Cabozantinib nda/2012/20375 6Orig 1s000 ClinP harmR .pdf. Accessed 18 Oct versus everolimus in advanced renal cell carcinoma (METEOR): final results from a randomized open-label, phase 3 trial. Lancet 22. European Medicines Agency (2016) European Public Assessment Oncol 17(7):917–927 Report (EPAR) for Cabometyx. http://www.ema.europ a.eu/docs/ 12. Cabometyx™ (cabozantinib) Tablets (2016) US prescribing infor- en_GB/docum ent_libr a r y/EP AR_-_Publi c_asses sment _r epor t/ mation. Exelixis, Inc., South San Francisco. https ://www.acces human /00416 3/WC500 21407 0.pdf. Accessed 18 Oct 2017 s d a t a . f d a . g ov / d r u g s a t f d a _ d o c s / l a b e l / 2 0 1 6 / 2 0 8 6 9 2 s 0 0 0 l b l . p d f. 23. Singh H, Brave M, Beaver JA et al (2017) U.S. Food and Drug Accessed 5 Apr 2018 Administration approval: cabozantinib for the treatment of 13. Cabometyx™ (cabozantinib) Tablets (2016) Summary of prod- advanced renal cell carcinoma. Clin Cancer Res 23(2):330–335 uct characteristics. Ipsen Pharma, Boulogne-Billancourt, France. https://www .medicines.or g.uk/emc/product/4331 . Accessed 5 Apr 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cancer Chemotherapy and Pharmacology Springer Journals

Population exposure–response analysis of cabozantinib efficacy and safety endpoints in patients with renal cell carcinoma

Free
10 pages

Loading next page...
 
/lp/springer_journal/population-exposure-response-analysis-of-cabozantinib-efficacy-and-KhAIc9794C
Publisher
Springer Berlin Heidelberg
Copyright
Copyright © 2018 by The Author(s)
Subject
Medicine & Public Health; Oncology; Pharmacology/Toxicology; Cancer Research
ISSN
0344-5704
eISSN
1432-0843
D.O.I.
10.1007/s00280-018-3579-7
Publisher site
See Article on Publisher Site

Abstract

Background In the phase III METEOR trial, tyrosine kinase inhibitor cabozantinib significantly improved progression-free survival (PFS), objective response rate (ORR), and overall survival compared to everolimus in patients with advanced renal cell carcinoma (RCC) who had received prior VEGFR inhibitor therapy. In METEOR, RCC patients started at a daily 60-mg cabozantinib tablet (Cabometyx™) dose but could reduce to 40- or 20-mg to achieve a tolerated exposure. Objectives and methods Exposure–response (ER) models were developed to characterize the relationship between cabo- zantinib at clinically relevant exposures in RCC patients enrolled in METEOR and efficacy (PFS and tumor response) and safety endpoints. Results Compared to the average steady-state cabozantinib concentration for a 60-mg dose, exposures at simulated 40- and 20-mg starting doses were predicted to result in higher risk of disease progression or death [hazard ratios (HRs) of 1.10 and 1.39, respectively], lower maximal median reduction in tumor size (− 11.9 vs − 9.1 and − 4.5%, respectively), and lower ORR (19.1 vs 15.6 and 8.7%, respectively). The 60-mg exposure was also associated with higher risk for selected adverse events (AEs) palmar-plantar erythrodysesthesia syndrome (grade ≥ 1), fatigue/asthenia (grade ≥ 3), diarrhea (grade ≥ 3), and hypertension (predicted HRs of 2.21, 2.01, 1.78, and 1.85, respectively) relative to the predicted average steady-state cabozantinib concentration for a 20-mg starting dose. Conclusion ER modeling predicted that cabozantinib exposures in RCC patients at the 60-mg starting dose would provide greater anti-tumor activity relative to exposures at simulated 40- and 20-mg starting doses that were associated with decreased rates of clinically relevant AEs. Keywords Cabozantinib · Exposure–response modeling · Renal cell carcinoma Introduction improved understanding of the molecular biology of this disease, including the involvement of pathways linked to the Renal cell carcinoma (RCC) accounts for approximately vascular endothelial growth factor receptor (VEGFR), mam- 2–3% of all malignancies in adults with about one-third of malian target of rapamycin (mTOR), and the programmed patients having metastatic disease at diagnosis [1–3]. Recent cell death (PD-1) receptor [4–6]. Therapeutic approaches for advances in the treatment of RCC have been made based on treatment of RCC include the VEGF antibody bevacizumab, mTOR inhibitors temsirolimus and everolimus, the PD-1 checkpoint inhibitor nivolumab, and tyrosine kinase inhibi- tors (TKI) such as sunitinib, pazopanib, sorafenib, levatinib, Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s0028 0-018-3579-7) contains and axitinib [7]. Although these therapies were significant supplementary material, which is available to authorized users. advancements in the treatment of RCC, disease progres- sion is common as resistance to these treatments eventually * Steven Lacy develops. slacy@exelixis.com Cabozantinib is an inhibitor of receptor tyrosine kinases Exelixis Inc., 210 East Grand Avenue, South San Francisco, including VEGFR2, and the tyrosine kinases MET (hepat- CA 94080-0511, USA ocyte growth factor receptor) and AXL (GAS6 receptor) Ann Arbor Pharmacometrics Group, Inc., Ann Arbor, MI, implicated in development of resistance to RCC therapy USA Vol.:(0123456789) 1 3 1062 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 [8–10]. In the pivotal phase III METEOR study, cabozan- endpoint, and overall survival (OS) and objective response tinib improved overall survival, decreased disease progres- rate (ORR) as secondary endpoints. A total of 658 patients sion, and increased objective response in patients with were randomized 1:1 to receive cabozantinib (n = 330) or advanced RCC who had received prior VEGFR-TKI treat- everolimus (n = 328). Randomization was stratified by num- ment [11]. The cabozantinib tablet formulation (Cabome- ber of prior VEGFR-TKI therapies and number of risk fac- tyx™) is approved at a 60-mg-free-base equivalent (FBE) tors per Memorial Sloan-Kettering Cancer Center (MSKCC) daily dosage for the treatment of patients with advanced criteria. Patients were ≥ 18 years of age with advanced or RCC in USA who have received prior anti-angiogenic ther- metastatic RCC with a clear-cell histology and measurable apy and in the European Union (EU) following prior VEGF- disease per RECIST. To manage AEs, dose modifications targeted therapy, with dosage adjustments to 40-mg FBE and were allowed that included dose interruptions and reduc- then 20-mg FBE permitted to manage adverse events (AEs) tions. The dose of cabozantinib could be reduced to 40-mg [12, 13]. In METEOR, 60% of patients treated with cabozan- and then to 20-mg from the starting dose of 60-mg. The dose tinib had at least one dose reduction and 70% required dose of everolimus could be reduced to 5-mg and then to 2.5-mg. modification (i.e., dose interruption, reduction, or increase). Radiographic assessments were performed at screening and The most frequent AEs leading to dose reduction were diar- every 8 weeks for the first year, and every 12 weeks thereaf- rhea (16%), palmar-plantar erythrodysesthesia syndrome ter. Safety was assessed in all patients who received at least (PPES; 11%), fatigue (10%), and hypertension (7.6%). one dose of study drug every 2 weeks for the first 8 weeks Cabozantinib capsule formulation (Come triq ) is and every 4 weeks thereafter. At the discretion of the Inves- approved at a dose of 140-mg FBE in USA for treatment of tigator, treatment could be continued after radiographic pro- patients with progressive, metastatic medullary thyroid can- gression. One blood sample for plasma cabozantinib con- cer (MTC), and in the EU for the treatment of progressive, centration determinations was collected at approximately 8 unresectable locally advanced or metastatic MTC, with dose or more hours after the prior evening’s dose on day 29 and reductions to 100-mg FBE then to 60-mg FBE to manage day 57 of the study. AEs [14, 15]. Exposure–response (ER) modeling in MTC patients showed an increased risk of time to the first dose Bioanalytical methods modification to be highly correlated with lower cabozantinib apparent clearance (CL/F) values; however, no clear associa- Plasma cabozantinib concentrations were measured using a tion was observed between time to first dose modification validated liquid chromatographic–tandem mass spectrom- and progression-free survival (PFS) [16]. In an integrated etry method. The lower limit of quantitation was 0.5 ng/mL population PK (popPK) analysis, MTC cancer type was [18]. shown to be a statistically significant covariate on cabozan- tinib CL/F, with values approximately twofold higher rela- Software and modeling strategy tive to patients with other types of malignancies (including RCC) [17]. Several possible factors may underlie the higher Time-to-event Cox proportional hazard (PH) models were cabozantinib clearance observed in MTC patients, including developed using SAS Version 9.3 (SAS Institute Inc., Cary differences in the incidence or severity of treatment-related NC). Longitudinal tumor size and time-to-event dose modifi- diarrhea or hypocalcemia, or use of concomitant medica- cation models were developed using NONMEM version 7.3 tions; however, an exact cause has yet to be identified [17]. (ICON Development Solutions, Ellicott City, MD). The sto- The current ER analyses were thus undertaken to better chastic approximation expectation maximization (SAEM)/ understand the relationship between cabozantinib exposure importance sampling (IMP) estimation method was used for and efficacy, safety, and the need for dose adjustments in the tumor size model and the Laplacian estimation method RCC patients. was used for the repeated time-to-event dose modification model. Differences in objective function values (OFVs) between competing models as well as standard goodness-of- Methods fit plots were used to identify the best fitting model among those evaluated. Study design and data Population pharmacokinetic model The ER analyses were conducted utilizing data from the phase III METEOR study of cabozantinib in patients with A popPK model was developed from nine clinical stud- RCC [11]. METEOR was a multicenter, randomized, con- ies, comprised of 1534 subjects. The detailed method for trolled trial of cabozantinib [60-mg FBE once a day (QD)] the popPK model development is described in a separate versus everolimus (10-mg QD), with PFS as the primary manuscript [17]. Individual exposures for subjects in the 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1063 METEOR study were predicted from this popPK model for importance of selected cabozantinib exposures on the rate use as the exposure metrics for the present ER analysis. of events in the survival/probability scale. Longitudinal sum of tumor diameter model Exposure response analysis for time‑to‑event endpoints Nonlinear mixed-effects modeling was used to develop a model to describe the relationship between longitudinal sum Progression‑free survival and safety endpoints of tumor diameter measurements and C . The sum of tumor avg diameter over time is described in Eq. 4 [19]: Time-to-event analyses were performed to characterize dY∕dt = Growth − Drug effect, (4) the ER relationship between cabozantinib exposure and where dY∕dt is the change in tumor diameter over time, clinical endpoints including PFS and six safety endpoints. and growth represents the increase of tumor diameter over The number of patients with events and the total num- time, which is independent of any drug effects. Drug effect ber of patients at risk are listed in Supplemental Table 1. represents the first-order drug induced decay. Linear and The safety endpoints evaluated included fatigue/asthenia nonlinear relationships were evaluated between cabozantinib (grade ≥ 3), PPES (grade ≥ 1), nausea or vomiting, diarrhea and decay rate. In addition, models accounting for resistance (grade ≥ 3), hypertension (systolic BP > 160 mmHg or dias- were also tested. tolic BP > 100 mmHg), and stomatitis (grade ≥ 3). The model with the best fit to the data had a first-order The extended Cox PH model was used to describe the growth rate, nonlinear cabozantinib drug ee ff ct, and a resist - relative hazard for all endpoints (efficacy, safety, and dose ance component. The model included exponential error modification) and to allow for time-varying cabozantinib models for inter-individual variability (IIV) and an additive exposure. The Cox PH model is represented by the equation: error model for residual variability. The model is defined h(t, X(t)) = h (t) ⋅ exp( ⋅ X(t)), (1) in Eq. 5: where h(t, X(t)) denotes the hazard at time t, h (t) is the −k ⋅t tol background hazard function, β is a vector of the regression ((k + k ⋅ e ) ⋅ C dmax dmaxtot avg dY = k ⋅ Y − ⋅ Y, coefficients, and X (t) is a matrix of covariates which may grow dt (EC + C ) 50 avg vary with time. A Cox PH model was fit for each endpoint (5) and three possible cabozantinib exposure measures: average where dY/dt is the change in tumor diameter per unit time, concentration (C ) over a pre-specified time interval rang- avg k is the first-order growth rate constant, k is the grow dmax ing from 1 day to 3 weeks prior to time t or from time 0 to maximum non-attenuating drug induced tumor decay rate, time t or area-under-the-plasma-concentration–time-curve k governs the maximum loss in the decay rate due to dmaxtot from time 0 to time t (AUC0 ). 0–t resistance, k is the rate constant which governs the rate of tol The impact of cabozantinib exposure on the relative haz- attenuation, EC is the cabozantinib concentration yielding ard was evaluated during base model development. Both one-half of the current tumor decay rate, and C is the indi- avg linear (Eq. 2) and nonlinear (Eq. 3) functional forms were vidual predicted daily average cabozantinib concentration. evaluated: Dose modification h(t, X (t)) = h (t) ⋅ exp( ⋅ X (t)), (2) ex o ex1 ex X (t) ex � � � The relationship (relative risk) between individual predicted h(t, X (t)) = h (t) ⋅ exp( ⋅ X (t)); X (t)= , o ex2 ex ex ex X (t)+ EC cabozantinib CL/F and the rate of dose modifications (i.e., ex 50 (3) reductions or interruptions) was assessed using the linear where X (t) or X′ (t) is time-varying cabozantinib exposure model (Eq. 2) and the nonlinear model(s) (Eq. 3). As CL/F ex ex measure, β represents the slope in the log-linear model, was log normally distributed, a linear model using log-trans- ex1 β represents the maximum drug effect in the E model, formed CL/F was also evaluated. ex2 max and EC represents a range of fixed values for the exposure Repeated time-to-event analyses were also performed at which half of the maximal effect is achieved. to characterize the ER relationship between cabozantinib Covariate effects were evaluated only for PFS using the C and the dose modification of any kind (DMAK) that avg SELECTION = SCORE option in the PHREG procedure considered dose escalations, reductions, and interruptions within SAS. Schwarz’s Bayesian Criterion (SBC) was cal- necessary for realistic simulations (Eq. 6). culated to find the most parsimonious or final model. A The date of the first dose of study medication repre- summary of the covariates evaluated is provided in Sup- sented “time 0” in these analyses. C was used in these avg plemental Table 2. The final model was used to evaluate the models. The number of events in a particular patient 1 3 1064 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 ranged from 0 to 52. The hazard for the DMAK model is Cox proportional hazard models defined by the following equation: for progression‑free survival 𝜆 = exp(𝜃 + 𝜃 ⋅ C ){if dose > 0}, base drug avg (6) A total of 315 patients were included in PFS analysis. Dif- ferent cabozantinib exposure measures were fit initially to = exp( −  ){if dose = 0}, base base-hold a linear Cox PH model; the time-varying average cabozan- where  represents the baseline log hazard,  , the base drug tinib concentration calculated over the 3 weeks prior to the change in log hazard per unit cabozantininb concentration, event t (C ) resulted in the lowest partial (negative 2 avg3w and  , the baseline log hazard for dose hold. The haz- base-hold log) likelihood (− 2LL) value and was the only statistically ard was dependent on whether a patient was currently on a significant exposure measure (p < 0.001). The relationship dose interruption (i.e., dose = 0). Cabozantinib exposure was between cabozantinib C and PFS was further assessed avg3w tested on the hazard for dose interruption, but this did not using nonlinear models over a range of EC values for result in a reduction in the OFV and was not included in the C (50–300 ng/mL). The nonlinear models resulted in avg3w final model. The baseline hazard during dose interruptions statistically significant reductions in − 2LL compared to is larger than the hazard during active treatment indicating the linear model; an EC value of 100 ng/mL resulted that there is an increased risk of a dose modification dur - in the best model fit and was used to illustrate the rela - ing dose interruptions. This is consistent with the relatively tionship between disease progression and cabozantinib short duration of the majority of dose interruptions. When concentrations. Figure 1 illustrates the impact of selected subjects are not on a dose interruption, increases in cabo- cabozantinib exposure values on the predicted survival zantinib concentration increase the instantaneous risk for a curves for PFS. These curves show the predicted fraction dose modification. The parameter estimates for the DMAK of subjects without disease progression or death. The typi- model are shown in Supplemental Table 3. cal individual predicted average steady-state cabozantinib concentration for 20-mg (375  ng/mL), 40-mg (750  ng/ mL), and 60-mg (1125  ng/mL) QD doses was used for Simulations exposure in the predictions and the predictions did not account for dose holds. Relative to reference steady-state Simulations were performed to compare the effects of plasma concentration at 60-mg, cabozantinib concentra- cabozantinib on longitudinal tumor size for a starting tions at lower simulated starting doses of 40- and 20-mg dose of 60-, 40-, and 20-mg. First, predictions for DMAK yielded hazard ratios (HRs) (95% lower–upper confidence analysis were based on Monte Carlo simulations (1000 limits) for disease progression or death [1.10 (1.07–1.12) patients) for trial duration of 365 days. Uncertainty in the and 1.39 (1.29–1.49), respectively] indicating an increased predictions was not computed. C was calculated interac- avg risk with decreasing plasma drug concentrations. tively based on dosing change predicted from the DMAK The most parsimonious model which was selected model. The tumor model was used to predict the change in as the final model contained covariates of cabozantinib tumor size over time for the 1000 subjects at 60-, 40-, and C , baseline Eastern Cooperative Oncology Group avg3w 20-mg starting dose groups. The median tumor diameter (ECOG) score ≥ 1, baseline sum of tumor diameter greater was tabulated by day for each treatment group. than the median, liver metastasis, high MET immuno- histochemistry (IHC) status, and elapsed time less than 3 months before progressive disease on prior TKI therapy. The parameter estimates for the final PFS model are listed Results in Supplemental Table 4. The impact of each covariate is pictured in Fig.  2. Patients with increasing cabozan- Population pharmacokinetic model tinib concentrations were predicted to have a decreased risk of progressive disease or death. Patients with baseline A two-compartment model with the first-order elimina- ECOG score ≥ 1, baseline sum of tumor diameter above tion and a dual absorption (first order + zero order) process median, liver metastasis, and high MET IHC status were adequately described the observed cabozantinib concen- predicted to have a higher baseline hazard ratio of progres- tration data. The results of the model and the covariates sive disease or death, but these effects were largely offset analyses are described in a separate manuscript [17]. The by higher cabozantinib exposure. At higher cabozantinib predicted exposure measures used in the ER analyses in concentrations, the HR for PFS was similar with or with- this report were derived from the post hoc PK parameters out these covariates. from this popPK model. 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1065 Fig. 1 Predicted survival curves for progression-free survival for selected values of average cabo- zantinib concentration. Typical individual predicted steady-state average cabozantinib concen- tration for the 20-mg (black), 40-mg (blue), and 60-mg (red) doses are 375, 750, and 1125 ng/mL, respectively. The solid line represents the fraction of subjects at each dose level without progress disease or death over time. The shaded areas represent 95% confidence intervals Fig. 2 Comparison of predicted hazard ratio for different covariate equal to 3 months for the previous TKI therapy. EC drug exposure effects for final progression-free survival model. The reference haz- producing 20% of the maximum effect, ECOG Eastern Cooperative ard represents an average cabozantinib concenration equal to 25  ng/ Oncology Group, MET IHC hepatocyte growth factor receptor protein mL (EC ), an ECOG score equal to zero, a baseline sum of tumor immunohistochemistry, SOD sum of diameters, TKI tyrosine kinase diameter that is below the median, no liver metastasis, an MET IHC inhibitor, PD progressive disease status equal to low, and a time to progressive disease greater than or 1 3 1066 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 HRs (95% lower–upper confidence limits) for PPES, fatigue/ Longitudinal tumor growth asthenia, hypertension, and diarrhea were 2.21 (1.60, 3.06), 2.01 (1.22, 3.31), 1.85 (1.33, 2.57), and 1.78 (1.08, 2.91), A total of 319 patients with 1637 evaluable tumor diameter measurements were included in the analysis. The parameter respectively, based on the predicted steady-stage average cabozantinib concentration for a 60-mg dose relative to estimates for the final tumor growth model is in Supplemen- tal Table 3. The attenuation half-life was 25.6 days for a typi- a 20-mg dose. For these respective AEs, lower predicted HRs (95% lower–upper confidence limits) were also evident cal patient, indicating that this component of tumor decay rate was near zero by 128 days. The EC is 251 ng/mL and based on the predicted steady-stage average cabozantinib concentration for a 60-mg dose relative to a 40-mg dose [i.e., the EC value is about 1000 ng/mL, suggesting that the rec- ommended 60-mg daily dosage was near the plateau of the 1.49 (1.27, 1.75), 1.42 (1.11, 1.82), 1.36 (1.15, 1.60), and 1.33 (1.04, 1.70)]. Statistically significant ER relationships dose–response curve as cabozantinib C for a 60-mg QD avg dose is 1125 ng/mL. Baseline tumor size variability between were not found for nausea/vomiting (grade ≥ 3) or stomati- tis (grade ≥ 3), but the frequencies of events for these two patients was large (CV = 72%). endpoints were small (n = 16 for nausea/vomiting and n = 10 for stomatitis). Cox proportional hazard models for safety endpoints Dose modifications For safety endpoints fatigue/asthenia, PPES, nausea/vom- 317 patients were included in the analysis of dose modi- iting, and diarrhea, the time-varying average cabozan- tinib concentration over 2 weeks prior to time t (C ) fications. The − 2LL for the linear model using log-trans- avg2w formed CL/F (log-linear CL/F model) was lower than the resulted in the lowest − 2LL. For hypertension and stoma- titis, the time-varying average concentration over the 24 h linear CL/F model and was similar to the − 2LL for the best nonlinear model. Residual diagnostics were similar for the prior to time t (C ) and concentration calculated from avg1d time zero to time equal to t (C ) resulted in the low- linear model using log CL/F and the best nonlinear model. avg0t Overall, the log-linear CL/F was selected over the best non- est − 2LL, respectively. Parameter estimates for the final models for each of these AEs are shown in Supplemental linear model. A statistically significant relationship was identified Table 4. An increase in average cabozantinib concentra- tions was associated with increased risk of PPES (grade ≥ 1; between individual predicted cabozantinib CL/F and the relative risk of dose modifications (p value < 0.0001). The Fig.  3), fatigue/asthenia (grade ≥ 3), hypertension (sys- tolic blood pressure > 160 mmHg or diastolic blood pres- parameter estimate for the dose modification model [β CL/F (SE) = − 1.27033 (0.177930)], indicating that the log HR sure > 100 mmHg), and diarrhea (grade ≥ 3). The predicted Fig. 3 Predicted accumula- tive hazards for palmar-plantar erythrodysesthesia (PPE) syndrome at specific, constant average cabozantinib concen- trations. Typical individual predicted steady-state average cabozantinib concentration for the 20-mg (black), 40-mg (blue), and 60-mg (red) doses is 375, 750, and 1125 ng/mL, respectively. The solid line rep- resents the accumulative hazard for PPE at each dose level over time. The shaded areas repre- sent 95% confidence intervals 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1067 Fig. 4 Predicted fractions of subjects without dose modi- fication for selected values of cabozantinib apparent clear- ance. The solid black line (shaded black areas represent 95% CI) represents the frac- tion of subjects without dose modification over time for CL/F of 1.3 L/h, the solid blue line (shaded blue areas represent 95% CI) represents the fraction of subjects without dose modification over time for CL/F of 2.3 L/h, and the solid red line (shaded red areas represent 95% CI) represents the fraction of subjects without dose modification over time for CL/F of 3.3 L/h. The hazard ratio of 0.281 (p < 0.0001) indicates 0.281 times less risk of dose modification for one unit increase of ln(CL) decreases with increasing log CL/F, was used to compute the data. Slight differences between the observed and simulated relative HR for different values of CL/F. Relative to a CL/F data sets may reflect the lack of a drop-out (based on pro- of 2.3 L/h, the HR (95% CI) for risk of dose modification for gressive disease or death) in the modeled analysis. Relative a lower CL/F value of 1.3 L/h was approximately two times to a 60-mg simulated starting dose, a lower percentage of greater [2.07 (1.60, 2.52)]. Figure 4 illustrates the impact of subjects were predicted to have dose-reduced at a 40-mg selected cabozantinib CL/F values on the predicted survival simulated starting dose at both the 6-month treatment time- curves; these curves show the predicted fraction without point (24 vs 45%) and 12-month treatment time-point (37 dose modifications over time for CL/F values of 1.3, 2.3, vs 64%). At 6 months, approximately 50% of subjects in the and 3.3 L/h. 60-mg starting dose group (observed and simulated) are still on the 60-mg dose. Simulations Next, the tumor growth model was used to simulate the time course of tumor diameters for each of the 1000 patients First, the dose modification model DMAK was used to sim- in the 60-, 40-, and 20-mg starting dose treatment groups. ulate longitudinal C for 1000 patients over a 12-month Patients in the 20- and 40-mg starting dose treatments avg period based on the changing events. To mimic dose change groups were predicted to have a smaller reduction (median scenarios in the observed data set, at the time of each simu- change from baseline = − 4.45 and − 9.1%, respectively) in lated event, the observed probability of a dose reduction, tumor size relative to the 60-mg (–11.9%) starting dose treat- interruption, or escalation based on the current dose was ment group (Fig. 5). used. If the current dose was 0 (representing a dose inter- To further assess the clinical relevance of the difference ruption), the observed probability of escalating to 20-, 40-, in tumor reduction between starting doses, the response to or 60-mg given the dose prior to the interruption was used. treatment was computed at baseline and every 8 weeks for The percentage of patients that were predicted to be on the 1 year using the longitudinal sum of tumor diameter pre- 20-mg, 40-mg, and 60-mg dosages after 6 and 12 months for dictions. The responses [complete response (CR), partial the observed data set for 60-mg and for the simulated 60-mg response (PR), stable disease, and progressive disease (PD)] and simulated 40-mg starting dose data sets are shown in were calculated using the criteria outlined in the METEOR Supplemental Table 5. study. In the simulated 60-mg starting dose group, a higher The simulated 60-mg starting dose yielded similar per- percentage of patients were predicted to achieve objective centages of patients that dose reduced to 40- or 20-mg after response (CR + PR) relative to the 40- and 20-mg starting 6 and 12 months on cabozantinib treatment. In general, the dose groups (19.1, 15.6, and 8.7%, respectively), whereas a simulated data set for 60-mg starting dose showed a similar lower percentage of patients were predicted to have PD (7.5, pattern for dose modification as the observed 60-mg dose 8.1, and 10.2%, respectively; Table 1). 1 3 1068 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 Fig. 5 Comparison of predicted median percent change from baseline tumor diameter for 20-, 40-, and 60-mg cabozantinib simulated starting doses Table 1 Percentage of simulated subjects (N = 1000) achieving each average cabozantinib concentrations of the modeled 20-, best overall response category 40-, and 60-mg dose levels (375, 750 and 1125 ng/mL, respectively). The nonlinear relationship was reflected in Best overall response 20-mg start- 40-mg start- 60-mg start- ing dose (%) ing dose (%) ing dose (%) the marginally increased HRs (1.10 and 1.39) for risk of progressive disease or death for simulated concentrations Complete response 0.10 0.00 0.00 at starting doses of 40- and 20-mg, respectively, vs 60-mg. Partial response 8.60 15.6 19.1 Covariates that were associated with an increase in rate Stable disease 81.1 76.3 73.4 of disease progression were baseline ECOG score of ≥ 1, Progressive disease 10.2 8.10 7.50 baseline tumor diameters above the median, presence of liver metastasis, MET IHC status designated as high, and having a < 3-month time elapsed before disease progression with prior TKI therapy. At clinically relevant cabozantinib Discussion concentrations, the effects of the covariates decreased, such that the HRs generally reflected primarily drug effect on In the phase III METEOR study, cabozantinib demon- PFS. However, subjects that had progressive disease prior strated significantly improved PFS [HR 0.51 (95% CI to 3 months on the previous TKI therapy showed a higher 0.41–0.62); median 7.4 vs 3.9 months; p < 0.0001], ORR HR compared to subjects that had progressive disease after [17% (13–22) vs 3% (2–6); p < 0.0001], and OS [HR 0.66 3 months on the previous TKI therapy for the same cabozan- (0.53–0.83); median 21.4 vs 16.5 months; p = 0.0003] ver- tinib plasma concentrations and appeared to be minimally sus everolimus in patients with advanced RCC who had affected by increasing cabozantinib plasma concentration. received prior VEGFR-TKI therapy [11]. A high percent- The nonlinear mixed-effect tumor growth model age (~ 60%) of RCC patients treated with cabozantinib in developed using target lesion tumor diameter measure- METEOR had at least one dose reduction from the 60-mg ments from RCC patients administered cabozantinib in FBE dose (to 40- or 20-mg FBE) in response to treatment- METEOR yielded an estimated EC value (251 ng/mL) emergent AEs. ER models were thus developed to describe 50 lower than the predicted steady-state average cabozantinib the relationship between cabozantinib exposures at doses concentrations for the starting dose levels of 60-, 40-, and evaluated in METEOR and measures of efficacy and safety 20-mg (1125, 750, and 375  ng/mL, respectively). The (AEs and the need for dose modifications) using data in corresponding predicted median percent change of tumor RCC patients enrolled in the pivotal phase III study. size from baseline (− 11.9, − 9.1, and − 4.5% respectively) A statistically significant relationship was identified and predicted ORR (19.1, 15.6, and 8.7%, respectively) between the rate of PFS (progressive disease or death) indicates the 60-mg cabozantinib starting dose which and the time-varying average cabozantinib concentration. provides relatively greater anti-tumor activity compared Increases in cabozantinib concentration were predicted to those predicted for lower simulated starting doses of to decrease the rate of PFS in a nonlinear manner, with 40- and 20-mg. These findings in RCC patients are con- an EC value (100 ng/mL) by the best nonlinear (E ) 50 max sistent with those of the nonlinear mixed-effect models model markedly lower than the predicted steady-state developed previously to describe the relationship between 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 1069 cabozantinib exposure and target lesion tumor size in the Conclusions phase III EXAM study of patients with progressive meta- static MTC: the estimated EC values (range 58–79 ng/ In ER models, a 60-mg simulated starting dose resulted mL) were lower than the steady-state cabozantinib con- in improved PFS, reduced tumor growth, and increased centration at the 140-mg dose level (1640 ng/mL), and no ORR compared to a 40- or 20-mg simulated starting dose. marked decrease in target lesion regrowth was predicted as However, higher cabozantinib exposures resulting from a consequence of the two protocol-defined dose reductions lower cabozantinib CL/F are predicted to increase the to 100- and 60-mg [20, 21]. rate of dose modification, while reducing cabozantinib An increase in cabozantinib concentration was asso- exposure with dose reduction is projected to decrease the ciated with an increased risk of AEs fatigue/asthenia risk of individual clinically relevant AEs in RCC patients. (g r ade ≥ 3), PPES (g r ade ≥ 1), hyper tension (syst olic Overall, the ER analysis predicts that cabozantinib would BP > 160 mmHg or diastolic BP > 100 mmHg), and diar- be effective at the 60-mg starting dosage evaluated in rhea (grade ≥ 3), with predicted HRs for these AEs approx- METEOR as well as daily dosages of 40- and 20-mg that imately 1.4- and twofold higher at the predicted steady- resulted from dose reduction. state average cabozantinib concentration for a 60-mg dose Acknowledgements The authors wish to respectfully thank the medical relative to a 40- or 20-mg dose, respectively. Cabozantinib professionals and patients that participated in the clinical trial of cabo- showed moderately high inter-individual variability for zantinib reported in this manuscript. The authors also wish to acknowl- CL/F in subjects with various tumor types including RCC edge Susan K. Paulson, Ph.D., for her assistance in the preparation based on popPK modeling (percent coefficient of varia- of this manuscript. This assistance was fully funded by Exelixis, Inc. tion = approximately 46%  [22]); thus, the 60-mg starting Funding The studies described in this manuscript were supported by dose would provide RCC subjects with higher cabozan- Exelixis, Inc. tinib CL/F (and lower exposures relative to subjects with lower CL/F) the opportunity to achieve therapeutic con- Compliance with ethical standards centrations. For RCC subjects with lower cabozantinib CL/F, dose modification to 40- or 20-mg provides the Conflict of interest Steven Lacy and Linh Nguyen are stockholders opportunity to achieve a tolerated plasma exposure that and current employees of Exelixis, Inc. Dale Miles was an employee of also yields acceptable clinical activity. Exelixis, Inc. when this work was performed and is currently employed by Genentech, Inc. Jace Nielsen, Bei Yang, and Matt Hutmacher are Treatment-emergent AEs frequently lead to dose modifi- employees of Ann Arbor Pharmacometrics Group (A2PG), Inc. who cations in RCC patients enrolled in METEOR. Patients were designed and conducted the ER modeling reported in this manuscript permitted to modify or interrupt treatment at non-uniform and was funded by Exelixis, Inc. Bei Yang is currently employed by times over the course of the METEOR study, so a repeated Luoxin Biotechnology (Shanghai) Co., Ltd. The authors contributed significantly to the design, conduct, analyses, and interpretation of the time-to-event model was developed that incorporated all data, and were involved in the preparation, review, and approval of this dose changes (dose holds, reductions and increases) over manuscript. time and was used to calculate time-varying cabozantinib exposure based upon the dose changes. Ethical approval The clinical studies were conducted in accordance with the World Medical Association Declaration of Helsinki, the Inter- While simulations showed that the 20-, 40-, and 60-mg national Conference on Harmonisation Tripartite Guideline for Good cabozantinib starting dosages were all predicted to reduce Clinical Practice, and all applicable local regulations. Study protocols tumor growth, the 60-mg dose resulted in the greatest reduc- and informed consent documents were reviewed and approved by the tion in tumor growth, best ORR, and lowest rate of PD. In Institutional Review Board (IRB) of participating institutions, and informed consent was obtained from all participants before any study- the review of the New Drug Application for cabozantinib for specified procedures were undertaken. the treatment of patients with RCC [23], the Food and Drug Administration (FDA) addressed the issue regarding the Open Access This article is distributed under the terms of the Crea- appropriateness of dose selection given the high percentage tive Commons Attribution 4.0 International License (http://creat iveco of dose reductions in METEOR. Based on their review, FDA mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- concluded that: (1) most adverse reactions were successfully tion, and reproduction in any medium, provided you give appropriate managed with dose interruptions and supportive measures; credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. (2) ER modeling indicated that lower starting doses could possibly compromise activity of the drug with decreased response rates; and (3) the dose selection of 60-mg daily was adequate based on ER analyses and a safety profile that is acceptable for the patient population. The cabometyx label specifies dose reduction instructions for the RCC popula- tion [12]. 1 3 1070 Cancer Chemotherapy and Pharmacology (2018) 81:1061–1070 14. Cometriq (cabozantinib) Capsules (2012) US prescribing References information. Exelixis, Inc., South San Francisco. https ://www. acces sdata .fda.gov/drugs atfda _docs/label /2012/20375 6lbl.pdf. 1. Gupta K, Miller JD, Li JZ et al (2008) Epidemiologic and socio- Accessed 5 Apr 2018 economic burden of metastatic renal cell carcinoma (mRCC): a 15. Cometriq (cabozantinib) Capsules (2016) Summary of product literature review. Cancer Treat Rev 34:193–205 characteristics. Ipsen Pharma, Boulogne-Billancourt, France. 2. Rini BI, Campbell SC, Escudier B (2009) Renal cell carcinoma. https ://www .medic ines.or g.uk/emc/pr odu ct/4408. Accessed 5 Lancet 373:1119–1132 Apr 2018 3. Siegel R, DeSantis C, Virgo K et al (2012) Cancer treatment 16. Miles D, Jumbe NL, Lacy S et al (2016) Population pharmacoki- and survivorship statistics, 2012. CA Can J Clin 62:220–241 netic model of cabozantinib in patients with medullary thyroid 4. Greef B, Eisen T (2016) Medical treatment of renal cancer: new carcinoma and its application to an exposure–response analysis. horizons. Br J Cancer 115:505–516 Clin Pharmacokinet 55(1):93–105 5. Batelli C, Cho DC (2011) mTOR inhibitors in renal cell carci- 17. Lacy S, Yang B, Nielsen J et al The population pharmacokinet- noma. Therapy 8:359–367 ics of cabozantinib in healthy volunteers and patients with vari- 6. Motzer RJ, Escudier B, McDermott DF et al (2015) Nivolumab ous tumor types. Cancer Chemother Pharmacol (manuscript versus everolimus in advanced renal-cell carcinoma. N Engl J Med accepted for publication) 373:1803–1813 18. Lacy S, Hsu B, Miles D et al (2015) Metabolism and disposition 7. Powles T, Staehler M, Ljungberg B et al (2016) Updated EAU of cabozantinib in healthy male volunteers and pharmacologic guidelines for clear cell renal cancer patients who fail VEFG tar- characterization of its major metabolites. Drug Metab Dispos geted therapy. Eur Urol 69:4–6 43:1190–1207 8. Yakes FM, Chen J, Tan J et al (2011) Cabozantinib (XL184), a 19. Ribba B, Holford NH, Magni P et al (2014) A review of mixed- novel MET and VEGFR2 inhibitor, simultaneously suppresses effects models of tumor growth and effects of anticancer drug metastasis, angiogenesis and tumor growth. Mol Cancer Ther treatment used in population analysis. CPT Pharmacometr Syst 19(12):2298–2308 Pharmacol 3:e113 9. Rankin EB, Fuh KC, Castellini L et al (2014) Direct regulation of 20. Miles DR, Wada DR, Jumbe NL et al (2016) Population phar- GAS6/AXL signaling by HIF promotes renal metastasis through macokinetic/pharmacodynamic modeling of tumor growth kinet- SRC and MET. Proc Natl Acad Sci USA 111(37):13373–13378 ics in medullary thyroid cancer patients receiving cabozantinib. 10. Bommi-Reddy A, Almeciga I, Sawyer J et  al (2008) Kinase Anticancer Drugs 27(4):328–341 requirements in human cells: III. Altered kinase requirements in 21. Center for Drug Evaluation and Research (CDER) (2012) Clinical VHL−/− cancer cells detected in a pilot synthetic lethal screen. pharmacology and biopharmaceutics review[s] for cabozantinib Proc Natl Acad Sci USA 105(43):16484–16489 [COMETRIQ]. http://www.acces sdata .fda.gov/drugs atfda _docs/ 11. Choueiri TK, Escudier B, Powles T et al (2016) Cabozantinib nda/2012/20375 6Orig 1s000 ClinP harmR .pdf. Accessed 18 Oct versus everolimus in advanced renal cell carcinoma (METEOR): final results from a randomized open-label, phase 3 trial. Lancet 22. European Medicines Agency (2016) European Public Assessment Oncol 17(7):917–927 Report (EPAR) for Cabometyx. http://www.ema.europ a.eu/docs/ 12. Cabometyx™ (cabozantinib) Tablets (2016) US prescribing infor- en_GB/docum ent_libr a r y/EP AR_-_Publi c_asses sment _r epor t/ mation. Exelixis, Inc., South San Francisco. https ://www.acces human /00416 3/WC500 21407 0.pdf. Accessed 18 Oct 2017 s d a t a . f d a . g ov / d r u g s a t f d a _ d o c s / l a b e l / 2 0 1 6 / 2 0 8 6 9 2 s 0 0 0 l b l . p d f. 23. Singh H, Brave M, Beaver JA et al (2017) U.S. Food and Drug Accessed 5 Apr 2018 Administration approval: cabozantinib for the treatment of 13. Cabometyx™ (cabozantinib) Tablets (2016) Summary of prod- advanced renal cell carcinoma. Clin Cancer Res 23(2):330–335 uct characteristics. Ipsen Pharma, Boulogne-Billancourt, France. https://www .medicines.or g.uk/emc/product/4331 . Accessed 5 Apr 1 3

Journal

Cancer Chemotherapy and PharmacologySpringer Journals

Published: Apr 17, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off