Purpose An integrated population pharmacokinetic (popPK) model was developed to describe the pharmacokinetics (PK) of tyrosine kinase inhibitor cabozantinib in healthy volunteers (HVs) and patients with various cancer types and to identify any differences in cabozantinib PK across these populations. Methods Plasma concentration data used to develop the popPK model were obtained from nine clinical trials (8072 con- centrations from 1534 HVs or patients) of cabozantinib in HVs and patients with renal cell carcinoma (RCC), medullary thyroid carcinoma (MTC), glioblastoma multiforme, castration-resistant prostate cancer, or other advanced malignancies. Results PK data across studies were adequately characterized by a two-compartment disposition model with dual first- and zero-order absorption processes and first-order elimination. Baseline demographic covariates (age, weight, gender, race, and cancer type) were generally predicted to have a small-to-moderate impact on apparent clearance (CL/F). However, MTC cancer type did show an approximately 93% higher CL/F relative to HVs following chronic dosing, resulting in approximately 40–50% lower predicted steady-state cabozantinib plasma concentrations. Conclusion This popPK analysis showed cabozantinib CL/F values to be higher for patients with MTC and may account for the higher dosage required in this patient population (140-mg) to achieve plasma exposures comparable to those in patients with RCC and other tumor types administered a 60-mg cabozantinib tablet dose. Possible factors that may underlie the higher cabozantinib clearance observed in MTC patients are discussed. Keywords Cabozantinib · Population pharmacokinetics · Cancer types Introduction cabozantinib capsule formulation (Cometriq ) is approved at a dose of 140-mg-free base equivalents (FBE) daily in the Cabozantinib is a tyrosine kinase inhibitor (TKI) targeting USA for the treatment of progressive metastatic medullary multiple receptor tyrosine kinases implicated in tumor angi- thyroid cancer (MTC) and in the European Union (EU) for ogenesis, invasion, and metastasis, including MET (hepato- the treatment of progressive, unresectable locally advanced cyte growth factor receptor), VEGFR2 (vascular endolethial or metastatic MTC [2, 3]. The cabozantinib tablet formula- growth factor receptor 2), AXL (GAS6 receptor), and RET tion (Cabometyx ) was subsequently approved at a dose of (glial cell-derived neurotrophic factor receptor) . The 60-mg FBE daily in the USA for the treatment of renal cell carcinoma (RCC) following anti-angiogenic therapy and in the EU following prior VEGF-targeted therapy [4, 5]. Cabo- zantinib tablets are also being evaluated in a pivotal clinical Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s0028 0-018-3581-0) contains study in patients with hepatocellular carcinoma at a 60-mg supplementary material, which is available to authorized users. FBE daily dose . The cabozantinib tablet formulation (Cabometyx) and * Steven Lacy capsule formulation (Cometriq) were not bioequivalent firstname.lastname@example.org following a single 140-mg dose in HVs ; the geometric Exelixis Inc., 210 East Grand Avenue, South San Francisco, least-squares mean (GLSM) for maximal plasma concentra- CA 94080-0511, USA tion (C ) was 19% higher for the tablet formulation and max Ann Arbor Pharmacometrics Group, Inc., Ann Arbor, MI, the upper 90% confidence interval for the GLSM ratio for USA Vol.:(0123456789) 1 3 1072 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 C (131.65%) slightly exceeded the 125% bioequivalence an estimated effective half-life of 55 h. The inter-subject max acceptance limit. However, the GLSM values for tablet variability (IIV) for CL/F (CV%) was 35%. and capsule formulations were similar (< 10% difference) A popPK analysis of cabozantinib was subsequently for both area under the plasma concentration–time curve performed using data collected from 282 patients with (AUC and AUC ) measures, and the 90% CIs were RCC and 63 normal HVs following oral administration 0−t 0−∞ 100–115% around the GLSM ratios. of doses of 20-, 40-, and 60-mg . A two-compartment The 140-mg FBE cabozantinib capsule dose used in the disposition model with dual (fast and slow) lagged first- pivotal phase III study in patients with MTC was based on order absorption processes adequately characterized the the maximally tolerated dose identified in a phase I study of concentration–time profile of cabozantinib in HVs and cabozantinib in patients with MTC and other solid tumors patients with RCC. The mean CL/F and terminal-phase . In the pivotal phase III study, 79% of MTC patients volume of distribution (V ) predicted for a typical White (169 of 214) who received the 140-mg FBE cabozantinib male subject were 2.23 L/h (90% CI 2.13, 2.34) and 319 L capsule dose eventually dose-reduced . Two protocol- (SE%: 2.7%), respectively, resulting in an estimated ter- defined cabozantinib dose reductions were allowed: from minal plasma half-life of approximately 99 h. The IIV for 140- to 100-mg/day, and from 100- to 60-mg/day. Forty-two CL/F was 46%. These popPK modeling analyses indicated percent of MTC patients received 60-mg/day as their final that cabozantinib CL/F was approximately twofold lower dose . Exposure–response (ER) modeling suggested that in RCC patients than in MTC patients, which is consist- the cabozantinib dose reductions from 140- to 100-mg and ent with the comparable observed steady-state exposures from 100- to 60-mg were not projected to result in a marked (C ) in RCC and CRPC patients administered a trough,ss reduction in progression free survival (PFS) or in tumor 60-mg tablet dose and in MTC patients administered a lesion regrowth in patients with MTC [11, 12]. 140-mg capsule dose . Based on these apparent differ - The 60-mg FBE cabozantinib tablet dose evaluated in ences in cabozantinib PK observed across cancer patients the phase III study in patients with RCC was based on find- with different tumor types, an integrated popPK model was ings from a phase I study in patients with RCC of improved developed with the pooled PK data from different patient tolerance to study drug and evidence of clinical activity in populations and HVs to evaluate the potential impact of patients who had dose-reduced from 140- to 60-mg . patient population, formulations, and doses on the PK of Dose reductions to a 40- or a 20-mg daily dose were permit- cabozantinib. ted in the pivotal phase III study in RCC patients to maintain treatment in response to drug-related adverse events (AEs). Although lower than the 140-mg capsule dose adminis- tered to MTC patients, the 60-mg cabozantinib tablet dose Methods was associated with a high percentage of dose reductions in both the phase III study in RCC patients (62%; 206 of Study design and data 330)  and in a phase III study in patients with castrate- resistant prostate cancer (CRPC) (74%; 505 of 682) . The popPK analysis was conducted using data from nine ER modeling suggested cabozantinib exposures associated clinical studies of cabozantinib including two phase I studies with a simulated 60-mg dose in RCC patients would result in HVs , a phase I study in cancer patients with advanced in slightly greater decreases in PFS, median percent change malignances , phase II studies in patients with GB  of tumor size from baseline, and best overall response rate and CRPC [19, 20], and phase III studies in patients with (based on target lesion) relative to simulated 40- or 20-mg RCC , MTC , or CRPC . The results of most starting doses . of these studies have been previously published; and a A popPK analysis was previously performed on pooled summary of the study designs, dosages, and PK sampling data for 289 cabozantinib-treated cancer patients (includ- schemes is presented in Table 1. All studies were con- ing MTC) receiving daily administration of the cabo- ducted following the ethical principles of the Declaration zantinib capsule formulation at a dose of 140-mg FBE/ of Helsinki and Good Clinical Practice guidelines. Written day, except for five subjects that were dosed at 200-mg informed consent was obtained from all patients and HVs. FBE/day . The data were adequately described by a 1-compartment model with first-order absorption and first-order elimination with a small lag time. The mean Bioanalytical methods CL/F and apparent volume of distribution of the central compartment (Vc/F) values estimated for a typical White Plasma cabozantinib concentrations were measured using a male subject were 4.42 [standard error (SE)%: 2.98%)] validated liquid chromatographic-tandem mass spectrometry L/h and 349 (SE%: 2.73%) L, respectively, resulting in 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 1073 Table 1 Summary of clinical studies included in the integrated population pharmacokinetic model of cabozantinib Study no. (Reference) Design Patient population Cabozantinib dose Planned pharmacokinetic sampling XL184-001  Phase 1, nonrandomized, Mixed malignancies 140- or 200-mg Days 1 and 19: pre-dose, 0.5, open-label FIH study 1, 2, 4, 8, and 24 h Day 5: pre-dose and 4 h Days 15 and 29: pre-dose XL184-010  Phase 1, crossover BE Healthy volunteer 140-mg Pre-dose, 0.5, 1,2,3, 4, 5, 6, study of tablet and 8, 10, 12, 14, 24, 48, 72, capsule 120, 168, 240, 288, 336, 408, and 504 h XL184-020  Phase 1, PK of tablet Healthy volunteer 20-, 40-, 60-mg Pre-dose, 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12, 14, 24, 48, 72, 120, 168, 240, 288, 336, 408, and 504 h XL184-201  Phase 2, multicenter, open Progressive glioblastoma 140-mg QD Each Cycle 28 days label multiforme Cycle 1: pre-dose, and 4 h on Days 1 and 15 Cycle 2: pre-dose and 4 h on Days 29 and 43 Cycle 3 and beyond: pre-dose on day 1 XL184-203 [19, 20] Phase 2, randomized dis- Castration-resistant prostate RDT: 100-mg QD RDT: pre-dose after “even” continuation study cancer NRE: 40- or 100-mg weeks after week 12 lead- QD in, or early termination or adverse event NRE; pre-dose week 1 day 1; pre-dose end of week 3 and 6, 12, 18, and 24, unsched- uled, early termination or adverse event XL184-301  Phase 3, randomized, Metastatic medullary thy- 140-mg QD Cycle 1, day 1: pre-dose and double-blind, placebo- roid cancer 2, 4, and 6 h controlled Cycle 2, day 29: pre-dose and 2, 4, and 6 h XL184-306 Phase 3, randomized, Castration-resistant prostate 60-mg QD Week 1 day 1, Week 4 day 1 [NCT01522443] double-blind, controlled cancer Week 7 day 1, Week 13 day 1 versus prednisone XL184-307  Phase 3, randomized, Castration-resistant prostate 60-mg QD End of Week 3 and End of double-blind, controlled cancer Week 12 versus prednisone XL184-308  Phase 3, randomized, con- Renal cell carcinoma 60-mg QD Days 29 and 57 approxi- trolled versus everolimus mately eight or more hours after prior evening’s dose BE bioequivalence, FIH first-in-human, QD once daily, RDT randomized discontinuation trial, NRE nonrandomized expansion method. The lower limit of quantitation (LLOQ) was 0.5 ng/ Missing PK drug concentrations, if any, were docu- mL . mented and excluded from the analysis. Drug concentra- tions which were below the level of quantification (BLQ) Analysis of data files were retained in the analysis data set but excluded from the analysis, because the number of BLQ samples was Source data in SAS format included information such as small (< 1%). PK sample concentrations, PK sample dates and times, Baseline covariate values were assigned using covariate dose amounts with dates and times, and patient demo- information collected prior to the first dose of study medica- graphics and covariates. N ONMEM (Version 7) ready tion. Covariate values closest to the first dose of study medi- data sets were constructed using SAS (Version 9.3), S-plus cation were used first; however, if covariate information was (Version 8.2) or R (version 3.0.2). not available immediately before study drug administration (e.g., pre-dose on day 1), then covariate information from 1 3 1074 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 Table 2 Baseline demographics and covariates in each clinical study used in the integrated population pharmacokinetic model of cabozantinib Study 001 010 020 201 203 301 306 307 308 Total N of subjects 40 77 63 39 284 210 41 498 282 1534 Sex Male (%) 31 (77.5) 32 (41.6) 33 (52.4) 26 (66.7) 284 (100) 146 (69.5) 41 (100) 498 (100) 222 (78.7) 1313 (85.6) Female (%) 9 (22.5) 45 (58.4) 30 (47.6) 13 (33.3) 0 (0) 64 (30.5) 0 (0) 0 (0) 60 (21.3) 221 (14.4) Race White (%) 35 (87.5) 74 (96.1) 62 (98.4) 33 (84.6) 246 (86.6) 188 (89.5) 34 (82.9) 380 (76.3) 231 (81.9) 1283 (83.6) Black (%) 2 (5) 2 (2.6) 1 (1.6) 3 (7.7) 15 (5.3) 1 (0.5) 4 (9.8) 9 (1.8) 5 (1.8) 42 (2.7) Asian (%) 1 (2.5) 0 (0) 0 (0) 1 (2.6) 14 (4.9) 9 (4.3) 1 (2.4) 1 (0.2) 19 (6.7) 46 (3.0) Other (%) 2 (5) 1 (1.3) 0 (0) 1 (2.6) 9 (3.2) 5 (2.4) 2 (4.9) 3 (0.6) 16 (5.7) 39 (2.5) NA (%) 0 (0) 0 (0) 0 (0) 1 (2.6) 0 (0) 7 (3.3) 0 (0) 105 (21.1) 11 (3.9) 124 (8.1) Population Healthy (%) 0 (0) 77 (100) 63 (100) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 140 (9.1) CRPC (%) 0 (0) 0 (0) 0 (0) 0 (0) 284 (100) 0 (0) 41 (100) 498 (100) 0 (0) 823 (53.7) RCC (%) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 282 (100) 282 (18.4) MTC (%) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 210 (100) 0 (0) 0 (0) 0 (0) 210 (13.7) GB (%) 0 (0) 0 (0) 0 (0) 39 (100) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 39 (2.5) Other (%) 40 (100) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 40 (2.6) Formulation Capsule (%) 40 (100) 75 0 (0) 39 (100) 284 (100) 210 (100) 0 (0) 0 (0) 0 (0) 648 (42.2) Tablet (%) 0 (0) 63 (100) 0 (0) 0 (0) 0 (0) 41 (100) 498 (100) 282 (100) 959 (62.5) Body weight Range (kg) 53.4–116 46.1–106 58.1–113.5 52–125.3 50.3–182.9 30.4–137.9 57.5–190.7 49.7–140 48.1–155.7 30.4–190.7 Mean (SD) 82.8 (15.9) 71.9 (11.5) 76.4 (11.8) 81.4 (18.3) 90.2 (18.6) 72.9 (18) 89.3 (23.1) 83.3 (14.0) 81.9 (17.0) 82.1 (17.3) Mean 83.0 71.9 76.5 79.4 87.8 71.3 84.8 82.3 80.4 81.0 Age Range (yrs) 23–71 18–55 19–54 20–67 43–87 20–86 48–79 35–87 32–86 18–87 Mean (SD) 56.0 (11.0) 39.3 (9.7) 36.9 (8.6) 48.6 (13.5) 66.3 (8.8) 54.7 (13.3) 64.8 (6.4) 68.7 (7.6) 61.6 (9.5) 61.3 (13.1) Median 57 39 38 53 67 55 65 69 62 64 CRPC castrate resistant prostate cancer, GB glioblastoma multiforme, MTC medullary thyroid cancer, N number, NA not available, RCC renal cell carcinoma, SD standard deviation Unknown mixed cancer type in Study 001 Study 010 is a cross-over study of capsule versus tablet formulations. The total percentage of subjects on tablet and capsule do not add up to 100% due to due to the crossover design in which each subject received both formulations. Six subjects in study 307 had missing weight information a previous visit (e.g., screening) was used. A summary of Base model the demographic characteristics and relevant covariates for HVs and cancer patients included in the integrated popPK The PK Base Model was developed initially using only the analysis is found in Table 2. exposure data from the HV and RCC studies, and then sub- sequently with the full integrated data set. Structural models Population PK model evaluated were one- and two-compartment disposition mod- els with first-order elimination, first-order absorption, and Analyses were performed using the nonlinear mixed effect absorption lag time. The previous reports indicated that the modeling as implemented in NONMEM (version 7.3 ICON concentration–time profile showed multiple peaks suggesting Development Solutions, Ellicott City, MD, USA). Estima- enterohepatic recirculation or multiple absorption sites [7, 21]; tion methods used included first-order conditional estimation therefore, other models were considered such as dual lagged with interaction (FOCEI), iterative two-stage (ITS), stochas- rs fi t-order absorption models or transit compartment models tic approximation expectation maximization (SAEM), and with increasing number of transit compartments. importance sampling method (IMP). 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 1075 Inter-individual variability (IIV) of the PK parameters was where θ and θ are fixed-effect parameters and χ is the REF x ij incorporated using a log normal random effects model: indicator variable with values of 1 or 0. To prevent negative parameter values in simulations, θ is the log of the frac- ( i) = ⋅ e , (1) i T tional change in the typical value for a categorical covariate. where θ is the individual value of the PK parameter (e.g., Plots of the individual random effect values versus covar - CL/F), θ is the typical value of the parameter, and η is T i iate values were generated to evaluate if inclusion of the the inter-individual random effect assumed to have a nor - covariate effects reduced or eliminated trends in the random mal distribution with a mean of zero and variance of ω . effects/PK parameters. In addition, box plots of the η values The vector of IIV random effects had a variance–covariance versus dose and study were generated to evaluate adequacy matrix Ω. A full-block Ω was estimated. Reductions to full- of pooling studies for this analysis. block covariance structure were considered if instability in At each key step in the model development, a complete the model was encountered. battery of diagnostic plots was generated. Standard good- Residual variability (RV), a composite measure of assay ness-of-fit plots were used to assess lack-of-fit. Different error, dose/sample time collection errors, model misspeci- structural models were considered if the initial model did fication, and any other unexplained variability within a sub- not adequately describe the integrated cabozantinib concen- ject, was initially modeled using the log-transformed addi- tration–time data. tive error model: ln(Y )= ln(C )+ , (2) ij ij ij Posterior predictive check where Y denotes the observed drug concentration for the ith ij individual at time t , C denotes the corresponding predicted j ij An internal posterior predictive check (PPC) was performed concentration based on the PK model, and ε denotes the ij to assess the predictive performance of the popPK mod- residual random variable, which is assumed to have a normal els . A smoothed parametric bootstrap procedure was distribution with a zero mean and variance σ . Other residual implemented to account for uncertainty in the parameter error models were explored if patterns were observed in the estimates. A total of 500 sets of population parameter val- individual weighted residual (IWRES) versus individual pre- ues were generated using the multivariate normal distribu- dicted value (IPRED) plot. tion with the mean vector set to the population parameter estimates and the covariance matrix set to that of the final model. These values were used to simulate a data set repli- Covariate model cating study design, sample size, and covariate distributions from the observed data set. The PPC statistics including the The covariate analysis was performed using a full model approach [22, 23]. Covariates were pre-specified based upon median, and 10th and 90th percentiles were computed at nominal time points for both the observed and each simu- clinical judgment and mechanistic plausibility and included: age, weight, sex, race, and population (cancer patient type lated data set. Prediction intervals were constructed based on the 5th and 95th percentiles of the simulated distributions including RCC, CRPC, MTC, GB, and advanced malignancy or HV) on CL/F and Vc/F. The full model was constructed of the PPC statistics. by including simultaneously all pre-specified covariates of interest into the base model. The relationship between continuous covariates and the typical value of PK parameters was described using power Results models: Data ij = , (3) TV, ij REF The final data set contained a total of 8072 plasma concen- REF trations from 1534 patients and HVs. All BLQ samples were excluded from the analysis, since the percentage of samples where θ and θ are the fixed-effect parameters and x REF x REF that were BLQ was small (< 1%). is a reference value of the covariate χ . The approximate ij The majority of subjects were male (85.6%), as four of median value was used for χ . The relationship between REF the nine studies enrolled patients with CRPC, and white categorical covariate and typical value of PK parameters was (83.6%). Body weights and ages were generally consist- modeled as follows: ent across the studies, except for the HV studies which had = ⋅ exp( ⋅ x ), TV, ij REF x ij (4) younger subjects due to the exclusion criteria. The single- dose clinical pharmacology studies in HVs and the phase 1 3 1076 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 I safety study that enrolled patients with advanced mixed covariates were also significant, but their effect was less malignancies (XL184-001) contained intensive sampling than cancer-type covariates. Goodness-of-fit plots showed which allowed for full characterization of the cabozantinib all three models provided reasonable fit to the data, but there PK profile. The phase II and III clinical studies in cancer was some lack-of-fit between observed and predicted geo- patients administered cabozantinib daily provided sparse PK metric means concentrations for MTC patients for the BASE sampling data. and FMECT models, with MTC patients having a higher estimated CL/F (Fig. 1). Only after including cancer type Population pharmacokinetic modeling as a covariate on CL/F (FM model) did the trend between IIR on CL/F and cancer patient population resolve (Fig. 2). Supplemental Table 1 lists the key steps in development Thus, the FM model was determined to be the final model. of the popPK model. Initially, the base popPK model was Cabozantinib PK parameter estimates for all three models developed utilizing only exposure data from studies enroll- are shown in Table 3. For the FM model, the transformed ing HVs or RCC patients. A two-compartment disposition estimates (90% CI) for CL/F and Vc/F were 2.478 (2.257, model and a dose-dependent dual absorption model con- 2.721) L/h and 187 (156.3, 223.9) L, respectively. Demo- taining two lagged first-order absorption processes (dose- graphic covariates (age, weight, sex, and race) generally dependent fast and slow) adequately characterized the cabo- showed minimal effect on CL/F and Vc/F, although race zantinib PK data. However, numerical problems and long covariate Black did result in an approximately 30% increase run times were encountered when the model was fit to the in CL/F. Cancer-type covariates RCC, CRPC, GB and Other fully integrated data set that contained all study populations showed minimal effects on CL/ F and Vc/F, whereas patients and covariates [including demographics (age, weight, sex, with MTC cancer type were predicted to have approximately and race) and cancer population effects (RCC, CRPC, MTC, 93% higher CL/F relative to HVs. Thus, when compared GB, and advanced malignancies)] (full model). Therefore, to HVs at the same dosage, patients with MTC would the dual first-order absorption model was replaced by a have approximately 40 and 50% lower steady-state maxi- model with dual first-order and zero-order absorption [full mal (C ) and minimal (C ) exposures, respectively max,ss min,ss modified (FM) model]. The first-order absorption process (Fig. 3). including a lag time and a dose-dependent effect on the The predictive performance of the three models for absorption rate constant (Ka) was described using a power patients with MTC stratified by day of study demonstrated model. In addition, capsule formulation was included as a that the lack-of-fit was most apparent on day 29 of the study structural covariate on Ka and overall bioavailability based for the BASE and FMECT models (Supplemental Fig. 2). on prior knowledge from the capsule-tablet bioequivalence These findings suggested that the day 29 concentration data study (XL184-020; ). and reduced accumulation relative to that expected from The FM model was stable and adequately described HVs are largely responsible for driving the increase in CL/F the PK data in the fully integrated cabozantinib PK data for patients with MTC. To confirm this finding, an addi- set across different studies (Supplemental Fig. 1). Close tional ad hoc run was performed using the FM model to a inspection of the model output suggested that the magni- re-fit data set including only day 1 data. In this model, the tude of demographic and cancer population-specific covari- population effect for MTC was − 0.312 (90% CI − 0.824, ate effects on cabozantinib PK was small, except for MTC 0.201), suggesting that CL/F on day 1 for MTC patients patient population who had a substantially higher estimated was not significantly different from HVs. Model prediction CL/F. To determine the significance of covariate effects, using this ad hoc model run remained reasonable for PK and the MTC covariate effect on CL/F , two additional ad data on day 1, but, when the predictions from the ad hoc hoc model runs were performed relative to the FM model: model were applied to day 29, the simulated data were much (1) all covariates except for dose on k12 and capsule on Ka higher than observed PK data for MTC patients (Supple- and F1 were removed (BASE) and (2) cancer-type covariates mental Fig. 3). Using the full data set and the FM model on CL/F and Vc/F were removed (FMECT). The OFV was which included a covariate for MTC patients, the fit was increased 401.2 units when all covariates (i.e., demographic substantially improved on day 29, while an acceptable fit and cancer type) were removed (comparing BASE to FM), was maintained on day 1 (Supplemental Fig. 2). suggesting significant effects of demographic covariates and cancer types together. The OFV was 305.9 units higher when cancer-type covariates were excluded (comparing FMECT Discussion to FM), indicating a significant effect of one or more cancer types on the PK of cabozantinib. Furthermore, the OFV for Cabozantinib is a TKI approved for the treatment of MTC BASE model was 95.4 units higher than the OFV for the and RCC [2–5]. Although the formulations and dosages are FMECT model, suggesting that some other demographic die ff rent for MTC (140-mg/day Cometriq) and RCC (60-mg/ 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 1077 Fig. 1 Comparison of goodness-of-fit plots for patients with medul- mean observed, typical individual predicted (PREDs), and individu- lary thyroid cancer on day 1 and day 29 of study XL184-301. Solid ally predicted (IPREDs) concentrations, respectively blue, red-dashed, and green-dashed lines correspond to geometric day Cabometyx) and dose adjustments and interruptions that cabozantinib CL/F (CV%) in MTC patients [4.4 L/h were allowed in the respective phase III studies, the result- (35%)] was twofold higher than in RCC patients [2.2 L/h ant cabozantinib steady-state exposures in the pivotal phase (46%)] [11, 17], suggesting an apparent difference in cabo- III studies were comparable for the two patient populations zantinib clearance in patients with different tumor types. . Findings from popPK analyses subsequently showed 1 3 1078 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 Fig. 2 Inter-individual random effect (Eta) on CL/F versus subject clearance, CRPC castrate-resistant prostate cancer, GB glioblastoma population. The boxes represent median and 25th and 75th percen- multiforme, HV healthy volunteer, MTC medullary thyroid cancer, tiles. The bars represent 5th and 95th percentiles The open circles OTH other cancer types in Study XL184-001, POP population, RCC represent individual values outside the 5th and 9th percentiles. CL renal cell carcinoma To examine the extent to which demographic covariates population. Possible reasons for the large increase in cabo- could explain heterogeneity in the PK parameters across zantinib clearance at steady state for MTC patients evident cancer populations, an integrated population PK analysis in the integrated popPK analysis were explored, including of cabozantinib was conducted using exposure data from differences in treatment-emergent AEs, concomitant medica- HVs and cancer patients with different types of malignan- tions, and administered cabozantinib dose. cies (ie, MTC, RCC, CRPC, GB). This analysis included Diarrhea is a common treatment-related AE in cancer data from nine clinical studies (three phase I, two phase II patients receiving cabozantinib [8, 9, 13–15], and the 140- and four phase III) for a total of 8072 cabozantinib con- mg dose administered to MTC patients may be anticipated centration records from 1534 subjects. A two-compartment to result in a higher incidence and/or severity of diarrhea model with first-order elimination and a dual absorption than a 60-mg dose given to RCC and CRPC subjects. As (first-order + zero-order) process adequately described the cabozantinib is considered to undergo enterohepatic recir- observed cabozantinib PK data. culation , a decrease in the absorption fraction of cabo- The FM model which incorporated demographic covari- zantinib typically reabsorbed via enterohepatic reabsorption ates (age, body weight, sex, and race) and type of cancer due to treatment-related diarrhea may result in an apparent malignancy (RCC, CRPC, MTC, GB, and other malignan- increase in cabozantinib clearance. The severity of diarrhea cies) on cabozantinib CL/F and Vc/F was evaluated. While and possible effect on clearance would be anticipated to be most covariate effects (including patient demographics) greater in MTC patients administered a higher cabozantinib included in the FM model had small-to-moderate effects on dose (140-mg) than that given to patients with other tumor cabozantinib PK parameters and exposure metrics, MTC types (60-mg). However, there was no marked difference in cancer-type led to a > 90% increase in CL/F. Ad hoc analy- Grade 3/4 diarrhea in the subjects enrolled in the cabozan- ses showed that the cabozantinib concentrations at day 29 tinib arm of the pivotal MTC study administered a 140-mg were primarily driving the increase in CL/F in MTC patients dose [16% (34 of 214); ] and in the cabozantinib arm of in pivotal phase III study XL184-301. MTC patients had the pivotal RCC study administered a 60-mg dose [13% (43 lower steady-state plasma concentrations at day 29 than of 311); ]. anticipated for the given dose relative to patients with other In a separate popPK analysis , MTC patients were cancer types or HVs and suggested that lower observed reported to have higher (67% greater) oral clearance accumulation could be due to higher clearance in this patient for another TKI (motesanib) relative to patients with 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 1079 Table 3 Parameter estimates for the final integrated population pharmacokinetic model of cabozantinib in patients with different cancer types Parameter Base model (BASE) Full model excluding cancer Full model (FM) type (FMECT) Transformed Estimate (90% CI) Transformed estimate (90% CI) Transformed estimate (90% CI) − 1 Ka (h ) 0.804 (0.576, 1.123) 0.846 (0.606, 1.182) 0.979 (0.679, 1.411) Duration of absorption for the zero-order 2.435 (1.966, 3.016) 2.441 (92.096, 2.843) 2.4 (2.01, 2.866) absorption process (h) Cl/F (L/h) 2.457 (2.396, 2.519) 2.553 (2.482, 2.625) 2.478 (2.257, 2.721) Vc/F (L) 157.178 (142.879, 172.95) 146.713 (131.894, 163.204) 187.0 (156.3. 223.9) Q/F (L/h) 30.154 (27.743, 32.786) 30.118 (27.883, 32.525) 31.213 (28.732, 33.92) Vp/F (L) 188.666 (176.091, 202.148) 193.605 (182.546, 205.203) 195.1 (183.3, 207.9) ALAG1 (h) 0.789 (0.763, 0.815) 0.777 (0.752, 0.804) 0.784 (0.757, 0.812) Fraction of dose in the first absorption depot 0.847 (0.805, 0.881) 0.840 (0.803, 0.8720) 0.854 (0.819, 0.884) F1 Dose-dependent Ka 0.566 (0.199–0.934) 0.585 (0.201, 0.969) 0.677 (0.268, 1.085) Covariates Capsule on Ka − 0.211 (− 0.541, 0.354) − 0.300 (− 0.599, 0.224) − 0.579 (− 0.783, − 0.183) Capsule on overall relative oral − 0.189 (− 0.205, − 0.173) − 0.183 (− 0.199, − 0.167) − 0.144 (− 0.162, − 0.126) availability Age on CL/F − 0.273 (− 0.367, − 0.178) − 0.162 (− 0.281, − 0.042) Female on CL/F – − 0.233 (− 0.29, − 0.172) − 0.230 (− 0.286, − 0.17) Race (Black) on CL/F – 0.249 (0.085, 0.439) 0.301 (0.139, 0.486) Race (Asian) on CL/F – − 0.118 (− 0.233, 0.013) − 0.078 (− 0.192, 0.052) Race (Other) on CL/F – − 0.029 (− 0.161, 0.124) − 0.007 (− 0.0136, 0.414) Weight on CL/F – − 0.248 (− 0.373, − 0.122) − 0.028 (− 0.148, 0.092) RCC on CL/F – – − 0.129 (− 0.217, − 0.033) CRPC on CL/F – – − 0.009 (− 0.11, 0.103) MTC on CL/F – – 0.928 (0.738, 1.136) GB on CL/F – – 0.216 (0.02, 0.449) Other malignancies on CL/F – – 0.178 (0.003, 0.384) Age on Vc/F − 0.277 (− 0.459, − 0.095) − 0.012 (− 0.247, 0.223) Female on Vc/F – 0.165(0.023, 0.327) 0.11 (− 0.033, 0.276) Race (Black) on Vc/F – − 0.065 (− 0.362, 0.372) − 0.022 (− 0.334, 0.438) Race (Asian) on Vc/F – 0.125 (− 0.197, 0.576) 0.05 (− 0.278, 0.528) Race (Other) on Vc/F – − 0.018 (− 0.321, 0.422) − 0.059 (− 0.382, 0.435) Weight on Vc/F – 0.798 (0.513, 1.083) 1.019 (0.72, 1.318) RCC on Vc/F – – − 0.63 (− 0.853, − 0.069) CRPC on Vc/F – – − 0.241 (− 0.395, − 0.049) MTC on Vc/F – – − 0.07 (− 0.232, 0.125) GB on Vc/F – – − 0.569 (− 0.72, − 0.337) Other malignancies on Vc/F – – − 0.186 (− 0.372, 0.055) Variance – – – ALAG, absorption lag time, CI confidence interval, CL/F apparent clearance, CRPC castrate-resistant prostate cancer, F1 fraction of dose split to the first absorption depot in a dual absorption model, GB Glioblastoma multiforme, Ka absorption rate constant, MTC medullary thyroid cancer, Q/F apparent flow parameter between compartments, RCC renal cell carcinoma Vc/F apparent volume of distribution of the central compart- ment, Vp/F apparent volume of distribution of the peripheral compartment Anti-logit transformation was used to obtain F1 For categorical covariates (e.g., capsule), transformed estimates correspond to fractional change from the reference level untransfor med values d 2 2 Untransformed full model variance estimates (90% CI) σ2 = 0.118 (0.114, 0.122); ω _Ka = 2.063 (1.579, 2.548); ω _CL/F = 0.202 (0.185, 2 2 2 2 0.218); ω _Vc/F = 0.233 (0.193, 0.273); ω _F1 = 0.466 (0.385, 0.546); ω _CL/F:Vc/F = 2.475 (1.923, 3.028), where ω is the variance–covari- ance matrix (Ω) of the inter-individual random effects (η) in the pharmacokinetic parameter, and σ the variance–covariance matrix of the intra- individual random effects (ε) in the measurements 1 3 1080 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 Fig. 3 Impact of covariates on steady-state cabozantinib CL/F, C ▸ min and C relative to a reference White, male, 80 kg, 60 year-old max healthy subject. CL/F, apparent clearance, Cmax,ss maximum plasma concentration at steady state, Cmin,ss minimum plasma concentration at steady state, CRPC castrate-resistant prostate cancer, GB glioblas- toma multiforme, HAGE a 79-year-old subject, HWT a subject with body weight of 112 kg, LAGE a 36-year-old subject, LWT a subject with body weight of 56 kg, MTC medullary thyroid cancer, RCC renal cell carcinoma differentiated thyroid cancer (DTC), in conjunction with a higher baseline incidence rate of diarrhea (68 and 6% in MTC and DTC cohorts, respectively). Similar to cabozan- tinib, patients’ disease type best accounted for inter-patient variability in motesanib CL/F of all covariates tested. However, incorporating diarrhea into the popPK model did not result in a significant improvement in the model fit, after accounting for the patients’ disease type, and there was no difference in motesanib CL/F among MTC patients with severe, moderate, and mild diarrhea. In addition, both the motesantib and cabozantinib popPK analyses showed a minimal effect on Vc/F in MTC patients, whereas a reduction in oral bioavailability due to diarrhea would be expected to result in increases in both CL/F and Vc/F. The mechanistic basis for the difference in motesanib CL/ F between DTC and MTC patients was not identified. Increased cabozantinib clearance in MTC patients at steady state could be related to treatment-emergent hypocalcemia, particularly in advanced MTC patients who undergo thyroidectomy when the parathyroid glands are also partially or completely removed resulting in decreased plasma parathyroid hormone levels. Hypocal- cemia may affect drug clearance indirectly via stimula- tion of active vitamin D metabolite 1,25 dihydroxyvitamin D (1,25(OH) D ) synthesis, and subsequent induction of 3 2 3 CYP3A4 by 1α,25(OH) D [27, 28]. Since cabozantinib is 2 3 metabolized by CYP3A4 , hypocalcemia was consid- ered as a potential contributing factor in reducing cabozan- tinib clearance in MTC patients. Although clinical labora- tory-defined hypocalcemia was identified in 52% of MTC patients receiving cabozantinib in the pivotal phase III study XL184-301 , and in fewer MTC patients receiv- ing placebo in the same study (27%), overall no evidence of altered calcium levels was noted in patients with MTC compared to other cancers to suggest that hypocalcemia was responsible for increased cabozantinib clearance in this population. Differences in concomitant medication use, particularly administration of strong CYP3A4 inducers in MTC patients, could have resulted in the increased cabozantinib clearance observed in the MTC patient population. However, only 1.4% of patients (3 of 207 total) were reported to have used a concomitant strong CYP3A4 inducer in the MTC phase III study of cabozantinib . Cabozantinib is also a substrate 1 3 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 1081 of eu ffl x transporter MRP2 [ 25], so concomitant administra- both CL/F and Vc/F resulted in an adequate fit to the tion of an MRP2 inducer could potentially increase cabozan- data. The magnitude of most demographic and popula- tinib clearance by enhancing hepatic and/or intestinal drug tion-specific covariate effects on cabozantinib PK was MRP2-mediated transport activity. Although overall use of small, except for MTC patient population who had a concomitant MRP2 inducers was not documented for MTC substantially larger estimated cabozantinib CL/F. Thus, patients in study XL184-301, only 5.5% of MTC subjects model-related and PK sampling differences do not appear (12 of 219) administered cabozantinib were reported to have to underlie the higher CL/F values in MTC patients evi- received MRP2 inducer (and moderate CYP3A4 inducer) dent at steady state. dexamethasone. Cabozantinib plasma clearance (CL/F) may also appear to be higher if oral bioavailability (F) decreased with Conclusion increasing cabozantinib dose. The approved cabozantinib dose for MTC patients (140-mg) is higher than the dose In summary, results from the integrated popPK analysis approved for RCC patients and dose generally admin- indicate that compared with other cancer patient groups istered to non-MTC patient populations (60-mg), and (RCC, CRPC, and GB), MTC patients clear cabozantinib steady-state CL/F in the MTC popPK analysis (4.4 L/h) faster and thus have lower dose-normalized steady-state was higher than that determined in the RCC popPK analy- plasma exposures. Cabozantinib PK appears to be time- sis (2.2 L/h). However, no decrease in cabozantinib oral varying in MTC patients, as day 1 CL/F values were lower bioavailability was evident in a cross-study analysis indi- and comparable to those in non-MTC cancer patient popu- cating generally dose-linear PK for tablet and capsule for- lations. Several possible factors may underlie the higher mulations over a broad dose range (20–140 mg) . In cabozantinib clearance observed in MTC patients; how- addition, lower cabozantinib exposures associated with ever, an exact cause has yet to be identified. Based on higher relative CL/F in MTC patients were only observed the integrated popPK analysis, non-MTC cancer patient at steady state (day 29) and not at day 1. cohorts (including RCC patients) appear to have compa- Alternatively, estimates of CL/F values from MTC rable cabozantinib clearance to that of HVs. subjects that tolerate a 140-mg daily cabozantinib dose Acknowledgements The authors wish to respectfully thank the medical may be higher than the overall study population if they professionals and patients that participated in the clinical trial of cabo- ref lect a sub-population that tolerates this higher dose at zantinib reported in this manuscript. The authors also wish to acknowl- steady state based on a faster relative intrinsic clearance. edge Susan K. Paulson, Ph.D., for her assistance in the preparation of this manuscript. This assistance was fully funded by Exelixis, Inc. In the MTC popPK analysis, high drop-out rate or early discontinuation was also considered to possibly explain Funding The studies described in this manuscript were supported by the lower day 29 concentrations in MTC patients relative Exelixis, Inc. to HVs . If subjects with low clearance and higher exposures dropped out or discontinued the study early Compliance with ethical standards due to treatment-emergent AEs, only those subjects with higher clearances resulting in lower, more tolerable expo- Conflict of interest Steven Lacy and Linh Nguyen are stockholders sures would remain. This scenario is unlikely considering and current employees of Exelixis, Inc. Dale Miles was an employee of Exelixis, Inc. when this work was performed, and is currently em- 79% of the patients in the MTC popPK analysis contrib- ployed at Genentech, Inc. Jace Nielsen, Bei Yang, and Matt Hutmacher uted PK samples on both days 1 and 29. are employees of Ann Arbor Pharmacometrics Group (A2PG), Inc. Finally, a more detailed PK sampling of the termi- who designed and conducted the popPK modeling reported in this nal elimination phase was included in the RCC popPK manuscript and was funded by Exelixis, Inc. Bei Yang is currently employed by Luoxin Biotechnology (Shanghai) Co., Ltd. The authors analysis (up to 504-h post-dose in HVs) than in the MTC contributed significantly to the design, conduct, analyses, and inter - popPK analysis where the final PK sample was taken at pretation of the data, and were involved in the preparation, review, and approximately 24-h post-dose. As cabozantinib has a rela- approval of this manuscript. tively long plasma terminal half-life (HV mean: 118 h Ethical approval The clinical studies were conducted in accordance ), plasma clearance could have been underestimated with the World Medical Association Declaration of Helsinki, the Inter- in the MTC popPK analysis based on a more limited PK national Conference on Harmonisation Tripartite Guideline for Good data collection of terminal elimination phase. However, Clinical Practice, and all applicable local regulations. Study protocols the integrated popPK model developed subsequently and informed consent documents were reviewed and approved by the Institutional Review Board (IRB) of participating institutions, and included exposure data from patients with different tumor informed consent was obtained from all participants before any study- types and HVs; the addition of all covariates, including specified procedures were undertaken. demographic (age, weight sex, and race) and population (RCC, CRPC, MTC, GB, and advanced malignancies) on 1 3 1082 Cancer Chemotherapy and Pharmacology (2018) 81:1071–1082 Open Access This article is distributed under the terms of the Crea- 14. Choueiri TK, Escudier B, Powles T et al (2016) Cabozantinib tive Commons Attribution 4.0 International License (http://creat iveco versus everolimus in advanced renal cell carcinoma (METEOR): mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- final results from a randomized, open-label phase 3 trial. Lancet tion, and reproduction in any medium, provided you give appropriate Oncol 17(7):917–927 credit to the original author(s) and the source, provide a link to the 15. Smith M, De Bono J, Stemberg C et al (2016) Phase III study of Creative Commons license, and indicate if changes were made. cabozantinib in previously treated metastatic castration-resistant prostate Cancer: COMET-1. J Clin Oncol 34(25):3005–3013 16. Singh H, Brave M, Beaver JA et al (2016) U.S. food and drug administration approval: cabozantinib for treatment of advanced renal cell carcinoma. Clin Cancer Res 23(2):330–335 References 17. European Medicines Agency (2016) European Public Assessment Report (EPAR) for Cabometyx. http://www.ema.europ a.eu/docs/ en_GB/docum ent_libr a r y/EP AR_-_Publi c_asses sment _r epor t/ 1. Yakes FM, Chen J, Tan J et al (2011) Cabozantinib (XL184), a human /00416 3/WC500 21407 0.pdf. Accessed 18 Oct 2017 novel MET and VEGFR2 inhibitor, simultaneously suppresses 18. Schiff D, Desjardins A, Cloughesy T et al (2016) Phase 1 dose metastasis, angiogenesis, and tumor growth. Mol Cancer Ther escalation trial of the safety and pharmacokinetics of cabozantinib 10(12):2298–2308 concurrent with temozoloide and radiotherapy or temozolomide 2. Cometriq (cabozantinib) capsules (2012) US prescribing after radiotherapy in newly diagnosed patients with high-grade information. Exelixis Inc., South San Francisco. https ://www. gliomas. Cancer 122(4):582–587 accessdat a.fda.go v/drugsatfda _docs/label /2012/20375 6lbl.pdf . 19. Smith MR, Sweeney CJ, Corn PG et al (2014) Cabozantinib in Accessed 05 Apr 2018 chemotherapy-pretreated metastatic castration-resistant prostate 3. Cometriq (cabozantinib) capsules (2016) Summary of product cancer: results of a phase II nonrandomized expansion study. J characteristics. Ipsen Pharma, Boulogne-Billancourt, France. Clin Oncol 32(30):3391–3399 https:// www.medicin es.org.uk/emc/product/ 4408. Accessed 05 20. Smith DC, Smith MR, Sweeney C et al (2013) Cabozantinib in Apr 2018 patients with advanced prostate cancer: results of a phase II ran- 4. Cabometyx™ (cabozantinib) tablets (2016) US prescribing infor- domized discontinuation trial. J Clin Oncol 31(4):412–419 mation Exelixis Inc., South San Francisco. h t t p s : / / w w w . a c c e s 21. Lacy S, Hsu B, Miles D et al (2015) Metabolism and disposition 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. of cabozantinib in healthy male volunteers and pharmacologic Accessed 05 Apr 2018 characterization of its major metabolites. Drug Metab Dispos 5. Cabometyx (cabozantinib) tablets (2016) Summary of product 43:1190–1207 characteristics. Ipsen Pharma, Boulogne-Billancourt, France. 22. Kowalski KG, Hutmacher MM (2014) Covariate selection in phar- https ://www.medic ines.org.uk/emc/produ ct/4331. Accessed 05 macometric analyses: a review of methods. Br J Clin Pharmacol Apr 2018 79:132–147 6. Study of Cabozantinib (XL184) vs Placebo in Subjects With 23. Gastonguay M (2011) Full covariate models as an alternative to Hepatocellular Carcinoma Who Have Received Prior Sorafenib. methods relying on statistical significance for inferences about ClinicalTrials.gov Identifier NCT01908426. https://clinicaltrials. covariate effects: a review of methodology and 42 case studies. gov/ct2/show/NCT01908426?term=cabozantinib&rank=10 PAGE 20:Abstr 2229 7. Nguyen L, Benrimoh N, Xie Y et al (2016) Pharmacokinetics 24. Yano Y, Beal SL, Scheiner LB (2001) Evaluating pharmacoki- of cabozantinib tablet and capsule formulations in healthy adult netic/pharmacodynamics models using the posterior predictive subjects. Anticancer Drugs 27(7):669–678 check. J Pharmacokinet Pharmacodyn 28:171–92 8. Kurzock R, Sherman SI, Ball DW et al (2011) Activity of XL184 25. Lacy S, Miles D, Nguyen L (2017) Clinical pharmacokinetics (cabozantinib), an oral tyrosine kinase inhibitor, in patients with and pharmacodynamics of cabozantinib. Clin Pharmacokinet medullary thyroid cancer. J Clin Oncol 29(19):2660–2666 56:477–491 9. Elisei R, Schlumberger MJ, Müller SP et al (2013) Cabozantinib 26. Lu J-F, Claret L, Sutjandra L et al (2010) Population pharmacoki- in progressive medullary cancer. J Clin Oncol 31(29):3639–3646 netic/pharmacodynamics modeling for the time course of tumor 10. EMA (2013) European Public Assessment Report (EPAR) shrinkage by motesanib in thyroid cancer patients. Cancer Chem- for Cometriq. http://www .ema.eur op a.eu/docs/en_GB/docum other Pharmacol 66:1151–1158 ent_libra ry/EPAR_-_Publi c_asses sment _repor t/human /00264 0/ 27. Wang Z, Schuetz EG, Xu Y et al (2013) Interplay between vitamin WC500 16370 5.pdf. Accessed 18 Oct 2017 D and drug metabolizing enzyme CYP3A4. J Steroid Biochem 11. Miles D, Jumbe S, Lacy S et al (2016) Population pharmacokinetic Mol Biol 136:54–58 model of cabozantinib in patients with medullary thyroid carci- 28. Drocourt L, Ourlin J-C, Pascussi J-M et al (2002) Expression of noma and its application to an exposure-response analysis. Clin CYP3A4, CYP2B6 and CYP2C9 is regulated by the vitamin D Pharmacokinet 55(1):93–105 receptor pathway in primary human hepatocytes. J Biol Chem 12. Miles DR, Wada DR, Jumbe NL et al (2016) Population phar- 277:25125–25132 macokinetic/pharmacodynamic modeling of tumor growth kinet- 29. Nguyen L, Holland J, Miles D et al (2015) Pharmacokinetic ics in medullary thyroid cancer patients receiving cabozantinib. (PK) drug interaction studies of cabozantinib: effect of CYP3A4 Anticancer Drugs 27(4):328–341 inducer rifampin and inhibitor ketoconazole on cabozantinib 13. Choueiri TK, Pal SK, McDermott DF et al (2014) A Phase 1 study plasma PK, and effect of cabozantinib on CYP2C8 probe substrate of cabozantinib (XL184) in patients with renal cell carcinoma. rosiglitazone plasma PK. J Clin Pharmacol 55(9):1012–1023 Ann Oncol 25(8):1603–1608 1 3
Cancer Chemotherapy and Pharmacology – Springer Journals
Published: Apr 23, 2018
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
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
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.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera