Received: 9 November 2023 |Revised: 30 April 2024 |Accepted: 7 May 2024
DOI: 10.1002/psp4.13160
A R T I C L E
Population pharmacokinetics of imetelstat, a first-in-class oligonucleotide telomerase inhibitor
MarioGonzález- Sales1| Ashley L. Lennox2,3| Fei Huang2| Chandra Pamulapati2| Ying Wan2| Libo Sun2| Tymara Berry2| Melissa Kelly Behrs2| Faye Feller2|
Peter N. Morcos4
1Modeling Great Solutions
Pharmaceutical Research & Studies,
FZE, Dubai, United Arab Emirates
2Geron Corporation, Parsippany, New
Jersey, USA
3Allucent, Cary, North Carolina, USA
4Morcos Pharmaceutical Consulting,
LLC, Marlboro, New Jersey, USA
Correspondence
Mario González-Sales, Modeling Great Solutions Pharmaceutical Research
-
Studies, FZE, Dubai, United Arab Emirates.
Email: mario@modelinggreatsolutions. com
Funding information
Geron Corporation
Abstract
Imetelstat is a novel, first-in-class, oligonucleotide telomerase inhibitor in development for the treatment of hematologic malignancies including lower-risk myelodysplastic syndromes and myelofibrosis. A nonlinear mixed-effects model was developed to characterize the population pharmacokinetics of imetelstat and identify and quantify covariates that contribute to its pharmacokinetic variabil- ity. The model was developed using plasma concentrations from 7 clinical studies including 424 patients with solid tumors or hematologic malignancies who received single-agent imetelstat via intravenous infusion at various dose levels (0.4-11.7 mg/kg) and schedules (every week to every 4 weeks). Covariate analysis included factors related to demographics, disease, laboratory results, renal and hepatic function, and antidrug antibodies. Imetelstat was described by a two-compartment, nonlinear disposition model with saturable binding/distribu- tion and dose-andtime-dependent elimination from the central compartment. Theory-based allometric scaling for body weight was included in disposition pa- rameters. The final covariates included sex, time, malignancy, and dose on clear- ance; malignancy and sex on volume of the central compartment; and malignancy and spleen volume on concentration of target. Clearance in females was modestly lower, resulting in nonclinically relevant increases in predicted exposure relative to males. No effects on imetelstat pharmacokinetics were identified for mild-to- moderate hepatic or renal impairment, age, race, and antidrug antibody status. All model parameters were estimated with adequate precision (relative standard error < 29%). Visual predictive checks confirmed the capacity of the model to describe the data. The analysis supports the imetelstat body-weight-based dosing approach and lack of need for dose individualizations for imetelstat-treated patients.
This is an open access article under the terms of the Creative CommonsAttribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
© 2024 Geron Corporation. CPT: Pharmacometrics & Systems Pharmacology published by Wiley Periodicals LLC on behalf of American Society for Clinical Pharmacology and Therapeutics.
CPT Pharmacometrics Syst Pharmacol. 2024;00:1-14. | www.psp-journal.com | | | 1 |
2 | | GONZÁLEZ-SALES et al. | |
Study Highlights
WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?
Imetelstat is a novel oligonucleotide telomerase inhibitor that has been investigated in hematologic malignancies and solid tumors. In phase III, imetelstat demonstrated statistically significant superior 8-week red blood cell transfusion independence rate versus placebo in lower-risk myelodysplastic syndromes. Another phase III study is ongoing in myelofibrosis.
WHAT QUESTION DID THIS STUDY ADDRESS?
A population PK (popPK) model using data from patients with hematologic malignancies and solid tumors was developed to characterize imetelstat PK proper- ties, identify and quantify PK variability, and inform the need for therapeutic individualization.
WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?
Imetelstat popPK was well described by a saturable binding/distribution model with dose- and time-dependent elimination. Covariates in the final model included body weight, disease factors, sex, dose, and time. The analysis supported body-weight-based dosing and no-dose individualization based on sex, age, race, and mild-to-moderate hepatic or renal impairment.
HOW MIGHT THIS CHANGE DRUG DISCOVERY, DEVELOPMENT,
AND/OR THERAPEUTICS?
This comprehensive popPK analysis adds to the pool of knowledge on oligonu- cleotide PK properties and helps optimize the appropriate use of imetelstat.
21638306, 0, Downloaded from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions
INTRODUCTION
Imetelstat (GRN163L) is a novel, first-in-class,13-mer N3′ → P5′ thio-phosphoramidate oligonucleotide with a palmitoyl group attached to the 5′ terminus through an aminoglycerol linker that acts as a potent and specific inhibitor of telomerase.1 It has a nucleotide sequence that is complementary to, and specifically binds with high affinity to, the template region of the RNA component of human telomerase, which lies in the catalytic site of human telomerase reverse transcriptase, resulting in competitive inhibition of human telomerase reverse tran- scriptase enzymatic activity that prevents telomere bind- ing.1,2 While sharing some structural similarity with other oligonucleotides, the mechanism of action of imetelstat is not antisense; it does not target mRNA of any gene and does not activate RNase-H-based degradation of its target sequence. The antitumor activity of imetelstat has been demonstrated in comprehensive preclinical in vitro and in vivo studies in various cancer cell lines, xenograft models, and primary patient samples.3-8 Cumulatively, results of nonclinical studies established antiprolifera- tive proof-of-concept activity for imetelstat; correlated pharmacokinetic (PK) exposure, pharmacodynamic effect (target engagement leading to inhibition of telomerase ac- tivity), and tumor growth inhibition in vivo in xenograft murine models; and indicated that higher imetelstat doses
were associated with greater plasma exposure and target engagement.
The clinical development of imetelstat investigated its safety, tolerability, and clinical activity when administered as an intravenous infusion using various doses and schedules as a single agent and in combination with systemic anticancer agents in hematologic malignancies and solid tumors. Phase I dose escalation and phase II studies established imetelstat 9.4 mg/kg as the maximum dose associated with an acceptable safety profile.9-11 Clinical PK results for imetelstat showed greater than dose proportional exposures across 0.4-11.7 mg/kg and minimal accumulation with multiple dosing with investigated once weekly to every-4-week dosing schedules.10 A phase
- dose ranging study in patients with myelofibrosis (MF) evaluated imetelstat 4.7 and 9.4 mg/kg administered every 3 weeks and demonstrated dose- and exposure- dependent pharmacodynamic (telomerase activity inhi- bition) and efficacy (symptom response rate and overall survival), with safety results (grade ≥3 adverse events, cy- topenias, and investigated laboratory abnormalities) that appeared generally similar across the investigated dose/ exposure range supporting selection of the imetelstat dosing regimen of 9.4 mg/kg every 3 weeks for the ongo- ing MF phase III study, MYF3001 (NCT04576156).9,12,13 The clinical efficacy and safety of imetelstat for treatment of anemia in patients with lower-risk myelodysplastic
(https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PopPK OF IMETELSTAT | |3 | |
21638306, 0,
syndromes (LR-MDS) were established in the phase II/III study, MDS3001 (NCT02598661), that investigated imetelstat 7.5 mg/kg every 4 weeks.14,15 The phase II portion showed that imetelstat treatment achieved durable rates of red blood cell transfusion independence and acceptable tolerability in pretreated LR-MDS patients.14 The results of the randomized, double-blind,placebo- controlled phase III study confirmed these findings and demonstrated that imetelstat 7.5 mg/kg every 4 weeks compared with placebo resulted in statistically significant superior rates of 8-week red blood cell transfusion independence (39.8% vs. 15.0%; p< 0.001).15 The ime- telstat clinical safety profile was manageable, with the most common grade ≥3 adverse events being thrombocy- topenia and neutropenia and showing similar low rates of high-grade bleeding and high-grade infection events between imetelstat and placebo.15
Oligonucleotides represent a novel modality with promising therapeutic application.16-19 As oligonucle- otides are distinct from classical small molecules and therapeutic biologics (e.g., monoclonal antibodies), PK characteristics are expected to be unique for this class of agents.20 This analysis comprehensively investigated the population pharmacokinetics (popPK) of imetelstat across its clinical development program. Results of this analysis aim to characterize imetelstat PK properties, identify and quantify sources of PK variability, and investigate the potential need for individualized dosing recommendations.
METHODS
Clinical studies
The imetelstat popPK model was developed based on data collected from patients receiving imetelstat as a single agent pooled from seven studies (Table 1). The doses in the included clinical studies ranged from imetelstat 0.4-11.7 mg/kg (expressed in terms of imetelstat sodium). Serial PK samples were collected from 28 patients in Study CP04-151, 74 patients in study CP05-101, 20 patients in study CP14B015, 28 patients in Study MYF2001, and 11 patients in Study MDS3001. Sparse samples were collected from patients in Studies CP14A004, CP14B013, and CP14B015 and from most of the patients in Studies MYF2001 and MDS3001.
All studies were performed in accordance with principles of the Declaration of Helsinki and were approved by the human investigational review board/ethics committee of each trial center, as required by the International Conference on Harmonization Guidelines for Good Clinical Practice. Informed consent was obtained from each patient before any study procedure was performed.
Bioanalytical methods
Blood samples were collected using lavender-top (K2EDTA) vacutainer collection tubes to prepare plasma. The plasma samples were analyzed for imetelstat concentration using a validated, Good Laboratory Practices, hybridization method that involved competition of imetelstat and a 3′-labeled di- goxigenin analogue for a complementary oligonucleotide sequence containingbiotinatthe3′end,withanenzyme-linked immunoassay used for the detection. The resulting complex was captured onto the surface of a neutravidin-coated mi- crotiter plate. The measurement of the digoxigenin-labeled probe was performed after reaction with antidigoxigenin antibody conjugated to alkaline phosphatase, which catalyzed the fluorescent AttoPhos® substrate (Promega, Madison, WI). Fluorescence intensity was measured using a fluorescence plate reader. Because the assay is competitive, the response is inversely proportional to the amount of imetelstat present in the calibration standards, quality control samples, and study sample. Concentrations were reported in terms of imetelstat sodium (molecular weight of 4896 g/mol). The lower limits of quantification of imetelstat in human plasma ranged from 0.367 to 0.588 μg/mL.
Dataset assembly
The nonlinear mixed-effects modeling dataset was prepared based on individual study imetelstat plasma concentration data, dosing records, and covariates. The number of imetelstat plasma concentration observations below the limit of quantification was 374 (8.55%). Thus, the M3 method was used as described elsewhere.21,22
Base structural model
Imetelstat concentrations were expressed in molar units for the model development. Models were estimated using a sequential process that implied iterative two-stage, stochastic approximation expectation maximization, and Monte Carlo important sampling (with expectation step only [EONLY = 1]) algorithms. Preliminary modeling efforts considered one- and two-compartment models following zero-order input from intravenous infusion. Based on visual inspection of imetelstat plasma concentration- time profiles and historical knowledge of greater than dose proportional exposures across the investigated dose range, the modeling efforts considered various nonlinear disposition models. Nonlinear models explored during structural model development included target-mediated drug disposition models using quasi-equilibrium approximation, quasi-steady-state approximation, or Michaelis-Menten
Downloaded from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 | | GONZÁLEZ-SALES et al. | |
TABLE 1 Listing of studies in PopPK analysis.
21638306, 0, Downloaded
Study number NCT (N) Phase
CP04-151 NCT00124189 (N= 22) phase I
CP05-101 NCT00310895 (N= 72) phase I
CP14A004
NCT00594126 (N= 19) phase I
CP14B015
NCT01243073 (N= 20) phase II
CP14B013
NCT01242930 (N= 13) phase II
MYF2001
NCT02426086 (N= 107) phase II
Study design/ population
Open-label, sequential dose cohort, dose- escalation, monotherapy study Chronic lymphoproliferative disease
Open-label, sequential dose cohort, dose- escalation, monotherapy study R/R solid tumor malignancies
Open-label, sequential dose cohort, dose- escalation, monotherapy study R/R multiple myeloma
Open-label, monotherapy study Essential thrombocythemia or polycythemia vera
Open-label study of imetelstat alone or in combination with lenalidomide maintenance therapy Multiple myeloma
Randomized, single- blind, dose-finding, monotherapy study Intermediate-2-risk or high-risk MF R/R to JAK inhibitor
Imetelstat starting dose/ number of patients treated with imetelstat
6-h i.v. infusion weekly: 20, 40, 80, 160, and 240 mg/m2
2-h i.v. infusion at the following frequencies and dose levels:
- Weekly in each 4-week cycle: 160 and 200 mg/m2
- Days 1 and 8 in 3-week cycle: 200 mg/m2
2-h i.v. infusion at the following frequencies and dose levels:
- Weekly in each 4-week cycle: 0.4, 0.8, 1.6, 3.2, and 4.8 mg/kg
- Days 1 and 8 in each 3- week cycle: 4.8, 6, 7.5, 9.4, and 11.7 mg/kg
- Day 1 in each 4-week cycle: 9.4 and 11.7 mg/kg
2-h i.v. infusion at the following frequencies and dose levels:
- Weekly in each 3-week cycle: 3.2 4.8, 6.0, and 7.2 mg/kg
- Days 1 and 8 in each 3-week cycle: 6.0 mg/kg
2-h i.v. infusion weekly in each 4-week cycle: 7.5, 9.4 mg/kg
2-h i.v. infusion on days 1 and 8 in each 4-week cycle: 7.5 and 9.4 mg/kg
2-h i.v. infusion once every 3-week cycle: 4.7 and 9.4 mg/kg
PK and immunogenicity assessments
Blood samples for plasma PK analysis were collected on day 1 of cycles 1 and 2 preinfusion; 1 h after SOI; at EOI; and at 1, 2, 3, 6, 10, 14, 18, and 24 h postinfusion. Samples were drawn at the end of study drug infusion on all other infusion days. (PK sampling after 2 h postinfusion was optional for cohorts 7 and 8 only.)
Blood samples for PK analysis were collected on day 1 of cycles 1 and 2 before SOI. Limited PK profiles were obtained for a subset of patients (University of Chicago) before the SOI; 1 h post-SOI; 5 min post-EOI; and 3, 4, 6, 8, and 24 h post-SOI. Extensive PK profiles were obtained in a subset of patients (Wayne State University) preinfusion; 1 h post-SOI; 5 min post-EOI; and 3, 4, 6, 8, 12, 16, 20, and 24 h post-SOI
Blood samples for PK analysis were collected for cycle 1 week 1 on day 1 prior to SOI, at EOI, 1 and 2 h post-EOI, and on days 2 and 4. For cycle 1 week 2, samples were collected prior to SOI, at EOI, and 1 h post-EOI. For cycle 1 week 3 and cycle 2 weeks 1 to 3, samples were collected prior to SOI and at EOI. For cycles 3 to 6 weeks 1 to 3, samples were collected prior to SOI.
Blood samples for plasma PK analysis were collected at the following timepoints: preinfusion, at the EOI, and 1 and 2 h postinfusion on day 1 of cycle 1. At nominal cycles 3, 6, 9, and 12, additional plasma samples were obtained preinfusion, EOI, and 1 and 2 h postinfusion.
Blood samples for plasma PK analysis were collected before the infusion, at the EOI, and 1 and 2 h after EOI (within 10 min; document collection time) on day 1 of cycle 1, and plasma was collected on day 15 of cycle 2.
Blood samples for plasma PK analysis were collected at the following timepoints:
Serial sampling: 0 h (preinfusion) and 1, 2, 3-5,6-10,12-16, and 18-24 h postdose on day 1 of cycle 1, and at 2 h every cycle starting from cycle 3.
Sparse sampling: 2, 3-6,8-12, and 16-24 h (if feasible) postdose on day 1 of cycle 1 and 2, 3-6, and 8-12 h (if feasible) postdose on day 1 of cycle 2, and at 2 h every cycle starting from cycle 3. Sparse sampling (dose escalation): 2, 3-6,8-12, and 16-24 h (if feasible) postdose on day 1 of first cycle receiving 9.4 mg/kg and at 2 h every cycle starting from cycle 3.
Blood samples for immunogenicity assessments were collected predose on day 1 of cycles 1, 3, and all subsequent cycles until EOT, at EOT, and first visit of posttreatment follow-up (if feasible).
from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PopPK OF IMETELSTAT | |5 | |
TABLE 1 (Continued)
21638306, 0, Downloaded
Study number | Imetelstat starting dose/ | |
NCT (N) | Study design/ | number of patients treated |
Phase | population | with imetelstat |
MDS3001 | Two-part | 2-h i.v. infusion once every |
NCT02598661 | monotherapy study: | 4-week cycle: 7.5 mg/kg |
(N= 171) | phase II, open-label, | |
phase II/III | single-arm design; | |
phase III, double- | ||
blind, randomized, | ||
placebo-controlled | ||
design | ||
Transfusion- | ||
dependent, low-or | ||
intermediate-1-risk | ||
MDS that is R/R to | ||
ESA treatment |
PK and immunogenicity assessments
Plasma samples were to be collected for PK and immunogenicity assessments at the following timepoints:
Rich sampling (PK analysis)
In part 1, a subset of patients had serial sampling: 0 h (preinfusion) and 1, 2, 3, 4, 6-8,22-26 h postdose on day 1 of cycle 1, and at 2 h every cycle starting from cycle 3.
Sparse sampling (PK analysis): Two, 4-5, and 6-7 h postdose on day 1 of cycle 1 and 2 h postdose on day 1 of all subsequent cycles, as soon as possible after IRR is observed (part 2 only), at EOT (part 1 only), and at first posttreatment follow-up (part 1 only).
Blood samples for immunogenicity assessments were collected predose on day 1 of cycle 1 and then every three cycles thereafter, as soon as possible after IRR is observed (part 2 only), at EOT, and first posttreatment follow-up.
from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library
Abbreviations: ADA, antidrug antibodies; EOI, end of infusion; EOT, end of treatment; ESA, erythropoiesis-stimulating agent; HER2, human epidermal growth factor receptor 2; i.v., intravenous; IRR, infusion-related reaction; JAK, Janus kinase; MDS, myelodysplastic syndromes; MF, myelofibrosis; N, number of patients in the final population PK analysis dataset; NCT, National Clinical Trial; PD, pharmacodynamics; PK, pharmacokinetics; R/R, relapsed and/or refractory; SOI, start of infusion. Doses are expressed in terms of imetelstat sodium.
where θi is the estimated parameter for patient i, θTV is the typical population value of the parameter, and ηi represents the IIV (ETA) random effects for individual i and were assumed to be normally distributed according to η~ N(0,ω) with covariance matrix for interindividual random effects (Ω).
Residual (unexplained) variability was modeled using an additive error model after logarithmic transformation of both observations and model predictions (Equation 3):
LnYobs,ij = LnYpred,ij + ij | (3) |
on [22/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-
AEF = | WTAllometric Exponent |
70 |
In this model, Yobs,ij denotes the observed concentration for the ith individual at time tj; Ypred,ij denotes the corresponding individual predicted concentration (IPRED); and ij is the residual unexplained variability term for the additive error model in the log scale, with a mean of 0 and variance
- of coefficient of variation (CV%) = sqrt(exp(21)−1) · 100.31
conditions) on Wiley Online Library for
where AEF represents the allometric exponent factor in a patient with a specific body weight (WT). The typical body weight for patient treated with imetelstat was assumed to be 70 kg.
Statistical model
Interindividual variability (IIV) was estimated as an exponential error model according to Equation 2:
i = TV ⋅ exp i(2)
Covariate analysis
The impact of patient demographic assessments (age, sex, race, and body weight), disease-related evaluations (base- line spleen volume, malignancy type, hepatic impairment classified according to the criteria of the National Cancer Institute Organ Dysfunction Working Group,32 and renal impairment classified according to creatinine clearance values), clinical laboratory measurements (creatinine clearance, albumin, bilirubin, aspartate aminotransferase, and alanine aminotransferase), antidrug antibodies
rules of use; OA articles are governed by the applicable Creative Commons License
GONZÁLEZ-SALES et al.
precision of estimates, condition number, and the magnitude of interindividual and residual variability were all considered for final model selection.
Model evaluation
21638306, 0, Downloaded from https://ascpt.onlinelibrary
- TV,i = TV ⋅ exp SLPk ⋅ COVk,i
- SLPk
- TV,i = TV ⋅ CCOVk,i
The following plots were created to assess the goodness of fit of the model to the data: observed versus population
-
and individual predicted values, conditional weighted re- siduals versus population predictions versus time, and in- dividual weighted residuals versus individual predictions versus time.
To verify that the final model adequately predicted both the central tendency and the variability of the ob- served data, simulation-based diagnostics were conducted to generate visual predictive checks (VPCs).22 Simulations (n= 200) were performed using the final model and final model parameters. A graphical comparison was made be- - tween the observed data and the model-predicted predic- tion intervals of 2.5th, 50th, and 97.5th percentiles over time. VPC output was stratified by relevant factors (e.g., malignancy type) to evaluate the ability of the final model to adequately characterize imetelstat over the relevant range of each stratification variable.
.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions
CCOVk,i = | COVk,i | (6) |
COVref,k | ||
Stepwise forward addition was conducted to develop the full covariate model. Covariates were added to the model based on prespecified hypothesis and diagnostic plots. Covariates were added one at a time. A decrease in the objective function value (OFV) of at least 6.635 (p≤ 0.01) and/or improvement in diagnostic plots was required for retention of the covariate. After evaluation of all important covariates, a stepwise backward elimination was conducted. During backward elimination, multiple models were run (i.e., full covariate model minus each covariate effect eliminated in- dividually) in steps as data permitted. Each covariate effect was evaluated for elimination based on an increase in the OFV of at least 10.83 (p≤ 0.001, with 1 degree of freedom [df]). The least significant covariate was selected for deletion at each step. After deletion of the least significant covariate relationship, a new full model was run, and covariates were once again tested for elimination. This process was rerun until the final model contained all covariates associated with statistically significant increases in the OFV. The decision to include a covariate was not based solely on the change in the OFV. In addition, goodness-of-fit plots, Akaike information criteria for non-nested models (where appropriate), the
Evaluation of clinical relevance of covariates
The final popPK model was used to simulate individual cycle 1 concentration−time profiles in the LR-MDS patients receiving a 2-h infusion of imetelstat 7.5 mg/kg (n= 170) using individual post hoc PK parameters to derive secondary PK parameters (maximum concentration [Cmax], area under the curve [AUC], and apparent half-life [t½]). Apparent t½ was estimated as the time required for maximum plasma concentrations to be reduced by 50%.
In addition, the final popPK model was used to simulate 1000 virtual patients to investigate the clinical relevance of the identified covariates for each category. Clinical relevance was informed by the difference between the predicted value for a parameter of exposure (e.g., cycle 1 AUC and Cmax) at the 5th percentile of a covariate and the 95th percentile relative to the median (for continuous variables) or against the category of reference (for categorical covari- ates). The use of cycle 1 AUC and Cmax is justified given that no appreciable accumulation has been observed for imetelstat with the intermittent dosing schedules. Forest plots were used to provide a visual representation of the significant covariate effects on imetelstat exposure. Exposure variations were considered relative to default 80%-125% boundaries when compared with the reference population.
(https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PopPK OF IMETELSTAT | |7 | |
21638306, 0,
Software
The modeling dataset was assembled using R version 4.1.3 (R Core Team, http://www.r-project.org) and R Studio Workbench (R Studio Team, Boston, MA). The popPK analysis was performed using NONMEM® program version 7.4 (ICON Development Solutions, Ellicott City, MD) and Pirana version 2.9.8 (Certara, Princeton, NJ). VPCs were conducted using R version 4.1.3.
RESULTS
Patient characteristics at baseline and covariates
The final analysis dataset included 424 patients from seven clinical studies who contributed a total of 4375 imetelstat plasma concentrations (Table 1). The summary statistics for baseline covariates included in the analysis dataset are presented in Table S1. Overall, in the analysis dataset, 247 (58.3%) patients were male, and the mean (range) baseline body weight was 77.2 kg (44.0-161 kg). Baseline spleen volume was only available from patients with MF (Study MYF2001). This covariate was investigated only in patients with available data. ADA status data were available from two studies (MYF2001 and MDS3001), which contributed most of the data to the overall dataset (65.6% of patients). ADA status was categorized as negative (reference) or positive for ADA-evaluable subjects, or as missing for those without ADA data. This categorization approach enabled the objective of the analysis to assess the impact of ADA positivity relative to the reference negative category. The distribution of baseline covariates reflects the eligibility criteria and patient populations enrolled in the studies included in the analysis.
FIGURE 1 Imetelstat popPK model. Bmax, Total concentration of target; CL, clearance; Kback, transfer rate constant from peripheral to central compartments; Kint, internalization rate constant; Koff, dissociation rate constant; Kon, binding rate constant; popPK, population pharmacokinetics; Vc, volume of central compartment.
The residual variability was modeled by an additive error model after logarithmic transformation of both observations and individual model predictions. Theory-based al- lometric exponents for body weight of 1, 0.75, and −0.25 were used for Vc, CL, and Kback, respectively. Alternative disposition models were investigated but did not improve model fit (i.e., no improvement in objective func- tion). These included models with linear elimination (∆OFV = +659) and nonlinear elimination including target-mediated drug disposition models using the quasi- equilibrium approximation (∆OFV = +225.6), the quasi-steady-state approximation (∆OFV = +997.3), or a model with Michaelis-Menten elimination (∆OFV = +319.2) when compared with the base model. Therefore, the base model was a two-compartment nonlinear model with saturable binding/distribution to the peripheral compart- ment, theory-based allometric exponents for body weight, and IIV quantified on CL, Vc, Bmax, and residual variability.
Downloaded from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley
Base structural model
Base models investigated one- and two-compartment structural models with increasing complexity based on known distribution properties of chemically modified oligonucleotides, visual inspection of the imetelstat PK profiles, and evidence of nonlinearity across the wide dose range (Table S2). The PK of imetelstat was best characterized by a two-compartment nonlinear disposition model with saturable binding/distribution to the peripheral compartment (Figure 1). The data supported the quantification of IIV on CL (∆OFV = −1610.1), Vc (∆OFV = −453.8), Bmax (∆OFV = −368.9), and residual variability (∆OFV = −2028.4). However, the data did not support the quantification of IIV on Kback (∆OFV = −6.0), Kint (∆OFV = −3.7), Kon (∆OFV = −3.7), or Koff (∆OFV = −4.9).
Covariate analysis
During the forward addition process, the following covariate effects were found to be statistically significant (p≤ 0.01) and were simultaneously included in the full covariate model: spleen volume on Bmax (∆OFV = −14.2; df = 1), MF malignancy on Bmax (∆OFV = −116.1; df = 1), time-variant ADA on CL (∆OFV = −48.6; df = 2), dose on CL (∆OFV = −28.0; df = 1), MF malignancy on CL (∆OFV = −18.8; df = 1), sex on CL (∆OFV = −31.9; df = 1), sex on Vc (∆OFV = −9.5; df = 1), time on CL (∆OFV = −140.3; df = 1), creatinine clearance on CL (∆OFV = −11.4; df = 1), and multiple myeloma malignancy on Vc (∆OFV = −8.9; df = 1). In the backward elimination process, the following covariates were eliminated
Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 | | GONZÁLEZ-SALES et al. | |
21638306, 0,
from the full covariate model because of lack of statistical significance at p≤ 0.001 level: the effect of time-variant ADA on CL (∆OFV = +7.7; df = 2), and the effect of cre- atinine clearance on CL (∆OFV = +8.7; df = 1). After covariate evaluation, further improvement in the OFV was achieved by the quantification of the correlation between the random effects of CL and Vc (∆OFV = −40.9; df = 1).
Removal of the fixed theory-based allometric exponents for body weight from CL, Vc, and Kback resulted in an increase in OFV (∆OFV = +65.2), supporting the importance of body weight as a covariate on imetelstat disposition.
Effects of other baseline covariates including mild-to- moderate hepatic impairment, mild-to-moderate renal impairment, liver enzymes (aspartate aminotransferase and alanine aminotransferase), bilirubin, age, and race did not have a statistically significant or meaningful impact on imetelstat PK.
Model evaluation
Parameter estimates for the final popPK model are presented in Table 2. All parameters were estimated with adequate precision. For fixed and random effects, the relative standard errors were <29% and <11%, respectively. Goodness-of-fit plots for the final popPK model showed no major deficiencies; residuals were randomly distributed around the line of identity and concentrated with the [−4, +4] interval (Figure 2, Figure S7). Mean and median ETA values were centered at 0 (Figures S8 and S9). The condition number of the final model was 23.9. The VPC for imetelstat for the overall population and specifically for patients with LR-MDS illustrate that the final model was appropriate to adequately describe imetelstat plasma concentrations and their respective variabilities in cancer patients (Figure 3 and Figure S12, respectively; see Figures S13-S17 for VPCs for other significant covariates).
Evaluation of clinical relevance of covariates
The model-predicted imetelstat concentration-time profile by relevant covariates is presented in Figure S18. Secondary PK parameters derived from simulated ime- telstat cycle 1 profiles for patients with LR-MDS receiving imetelstat 7.5 mg/kg show geometric mean (CV%) Cmax of 89.5 μg/mL (27.3%), AUC0-28d of 559 h μg/mL (43.2%), and apparent t½ of 4.9 h (43.2%).
A forest plot illustrating the magnitude of effect of the final covariates on derived secondary PK parameters, AUC, and Cmax, when considering a male patient weighing 70 kg with LR-MDS malignancy receiving imetelstat 7.5 mg/kg
as reference is provided in Figure 4. Overall, the direction of covariate effects was consistent with expected changes based on observed data (Figures S19-S21,Table S2), and the magnitude of covariate effects was generally modest. Using the imetelstat body-weight-based dosing approach, LR-MDS patients weighing 107 or 52 kg receive approximately 50% higher or 30% lower absolute imetelstat dose than the 70-kg reference patient. Nonetheless, the model- predicted exposures showed limited imetelstat exposure variations at these extremes of body weight (Figure 4).
DISCUSSION
Oligonucleotide therapeutics have emerged as a novel modality and offer the potential to therapeutically approach intracellular targets.16 The PK properties of oligo- nucleotide therapeutics have been characterized based on knowledge gained from chemical properties (i.e., backbone chemistry and chemical modifications), non- clinical studies, and individual clinical programs.17,33-39 Here, we report a comprehensive, popPK model for im- etelstat, a novel oligonucleotide telomerase inhibitor being investigated in hematologic malignancies based on data collected from seven single-agent clinical studies across a wide dose and schedule range in patients with solid tumors or hematologic malignancies.
The PK of imetelstat was described by a two- compartment disposition model with saturable binding/ distribution to a peripheral compartment and dose-andtime-dependent elimination from the central compart- ment. For the typical patient weighing 70 kg, CL, Vc, and Bmax were estimated to be 0.969 L/h, 3.91 L, and 15.5 μmol/L, respectively. The unique structural PK model for imetelstat, albeit empirical in nature, may provide mechanistic insights into the disposition characteristics of imetelstat. The saturable binding and dose- dependent elimination indicate nonlinear disposition properties for imetelstat. Saturable disposition processes have also been described for other oligonucleotides and may be partially explained by saturation of uptake receptors that facilitate distribution from plasma and/or sequestration within tissue.35 This is corroborated by the finding of increased CL and lower exposure in patients with MF (Figure 4), who have enlarged spleens and presumably greater uptake receptor capacity. The differences in exposure seen between patients with LR-MDS and patients with MF were identified during imetelstat development and contributed to the dose differences used in the two pivotal studies (7.5 mg/kg for MDS3001 and 9.4 mg/kg for MYF3001) (Figure S19). Spleen volume was a significant covariate on Bmax, although the effect is not considered to be clinically relevant given
Downloaded from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PopPK OF IMETELSTAT | |9 | |
21638306,
TABLE 2 Parameter estimates for the final PopPK model of imetelstat.
Shrinkage | ||||
Parameter | Unit | Value | RSE (%) | (%) |
CL, clearance from central | L/h/70 kg | 1.00 | 3.50 | - |
compartment | ||||
Vc, central volume of | L/70 kg | 4.08 | 2.55 | - |
distribution | ||||
Kback, transfer rate from | 1/h/70 kg | 0.0253 | 7.58 | - |
peripheral to central | ||||
compartment | ||||
Bmax, total concentration of | μmol/L | 15.0 | 7.08 | - |
target | ||||
Kint, internalization rate | L/h/70 kg | 0.103 | 9.08 | - |
constant | ||||
Kon, binding rate constant | L2/(μmol/L h) | 0.159 | 8.52 | - |
Koff, dissociation rate constant | L/h | 0.609 | 10.7 | - |
Effect of spleen volume on | 0.772 | 27.3 | - | |
Bmax | ||||
Effect of MF malignancy on | 1.44 | 7.25 | - | |
Bmax | ||||
Effect of MM malignancy on | −0.233 | 28.3 | - | |
Vc | ||||
Effect of dose on CL | −0.401 | 9.69 | - | |
Effect of MF on CL | 0.511 | 10.9 | - | |
Effect of sex on CL | −0.299 | 17.1 | - | |
Effect of time on CL | 5880 | 6.37 | - | |
Effect of sex on Vc | −0.122 | 27.3 | - | |
IIV on CL | CV% | 43.7 | 4.36 | 11.4% |
Correlation between ETA on | r | 0.545 | 10.9 | - |
CL and Vc | ||||
IIV on Vc | CV% | 25.7 | 6.02 | 20.4% |
IIV on Bmax | CV% | 45.8 | 9.44 | 44.5% |
IIV residual variability | CV% | 54.6 | 3.98 | 16.1% |
Residual variability | CV% | 21.8 | 3.82 | 12.9% |
0, Downloaded from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)
Abbreviations: Bmax, total concentration of target; CL, clearance; CV, coefficient of variation; IIV, interindividual variability; Kback, transfer rate from peripheral to central compartment; Kint, internalization rate constant; Koff, dissociation rate constant; Kon, binding rate constant; MF, myelofibrosis; MM, multiple myeloma RSE, relative standard error; Vc, central volume of distribution.
on Wiley Online
the negligible differences in AUC and Cmax at extremes of spleen volume. Nonlinear disposition has also been observed for imetelstat in NCA-based analyses of clinical studies in patients with solid tumors, ET/PV, or CLD (Table S2) and in nonclinical species, rabbit and monkey [Geron, data on file]. During the model development pro- cess, investigation of conditional weighted residuals versus time revealed some small evidence of bias that was accounted for by incorporating a time effect on CL. The covariate effect of time on CL, which was coded through a function that described that baseline CL decreased over time, resulted in better model fit; the physiological mechanism of the effect is unknown. Changes in CL over time may occur because of improvement in individual disease
status, as suggested for monoclonal antibodies approved for oncology indications.40-42 Imetelstat disposition is significantly influenced by baseline spleen volume in patients with MF, and spleen volume is reduced over time in those who respond to imetelstat treatment.9,13 Interestingly, apparent CL was strongly associated with efficacy for luspatercept, a therapeutic protein investigated in the MDS population, suggesting possible CL associations with severity of disease/anemia in MDS and providing support for these postulations.43 However, it cannot be excluded that the identified time effect in this analysis may reflect fitting criteria to improve model description of the available data. The identification of dose as a covariate likely reflects additional nonlinearity in
Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 | | GONZÁLEZ-SALES et al. | |
FIGURE 2 Goodness-of-fit lots for the final popPK model of imetelstat. CLD, chronic lymphoproliferative disease; ET/PV, essential thrombocythemia/polycythemia vera; MDS, myelodysplastic syndromes; MF, myelofibrosis; popPK, population pharmacokinetics.
disposition not fully accounted for by the saturable bind- | effects on CL in addition to saturable binding/distribu- |
ing/distribution function in the model. Attempts to con- | tion resulted in model instability, prohibitive run times, |
sider concentration-dependent (i.e., Michaelis-Menten) | and poor parameter precision (i.e., relative standard |
21638306, 0, Downloaded from https://ascpt.onlinelibrary.wiley.com/doi/10.1002/psp4.13160, Wiley Online Library on [22/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Attachments
- Original Link
- Original Document
- Permalink
Disclaimer
Geron Corporation published this content on 07 May 2024 and is solely responsible for the information contained therein. Distributed by Public, unedited and unaltered, on 29 May 2024 19:06:03 UTC.