Population pharmacokinetics and toxicity analysis of high-dose methotrexate in patients with central nervous system lymphoma
Anyue Yin (1), Fleur A. de Groot (2), Henk-Jan Guchelaar (1), Marcel Nijland (3), Jeanette K. Doorduijn (4), Daan J. Touw (5), Thijs Oude Munnink (5), Brenda C.M. de Winter (6), Lena E. Friberg (7), Joost S.P. Vermaat (2), Dirk Jan A.R. Moes (1)
(1) Department of Clinical Pharmacy and Toxicology, Leiden University Medical Center, Leiden, the Netherlands, (2) Department of Hematology, Leiden University Medical Center, Leiden, The Netherlands, (3) Department of Hematology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands, (4) Department of Hematology, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam, The Netherlands, (5) Department of Clinical Pharmacy and Pharmacology, University Medical Center Groningen, the Netherlands, (6) Department of Hospital Pharmacy, Erasmus MC, University Medical Center Rotterdam, Rotterdam, the Netherlands, (7) Department of Pharmacy, Uppsala University, Uppsala, Sweden
Introduction: High-dose methotrexate (HD-MTX) is the backbone of polychemotherapy for central nervous system lymphoma (CNSL)[1]. Although HD-MTX dose is based on body surface area (BSA), significant inter- and intra-individual variability in its pharmacokinetics (PK) is observed[2-4]. Furthermore, the unpredictable occurrence of acute toxicity, including kidney dysfunction and hepatotoxicity, remains a major concern in HD-MTX treatment[1-3]. Currently, studies on potential baseline risk factors for renal and hepatotoxicity in CNSL patients are limited, and identifying an MTX exposure threshold for toxicity with a model-based approach is desired for further HD-MTX treatment improvement.
Objectives: (1) To characterize and explain the variability in HD-MTX PK in CNSL patients. (2) To identify baseline predictive factors to the occurrence of renal and hepatotoxicity during HD-MTX treatment and to explore the exposure threshold for toxicity.
Methods: Routinely monitored serum methotrexate (MTX) concentrations and HD-MTX dosing information were collected retrospectively from patients with CNSL from 3 Dutch academic hospitals. Acute event of toxicity was defined according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0, and ≥ grade 1 toxicity was defined as a toxicity event. A population PK model was developed with NONMEM (version 7.4.4) and the covariate effects were investigated with the stepwise covariate modeling (SCM) function. Inter-occasion variability (IOV) of clearance (CL) was also incorporated. Data below the quantification limit (7.8 %) were omitted.
Toxicity data were analyzed with a logistic regression model. Individual PK parameters from the final PK model were applied to estimate the MTX exposure metrics of interest. Potential baseline and exposure-related predictors were investigated. A forward inclusion process was performed when investigating baseline predictors and factors that resulted in a change in OFV < -3.84 was considered to be significant (p< 0.05, degrees freedom = 1).
The predictability and stability of the final PK model were evaluated with goodness-of-fit (GOF) plots, prediction-corrected visual predictive check (pcVPC), and a bootstrap. The adequacy of the toxicity model was evaluated with a visual predictive check (VPC).
Results: In total 1460 MTX samples from 110 patients (81 patients with primary CNSL (73.6%)) were available for analysis. A two-compartment population PK model with first-order elimination adequately described the data. Estimated glomerular filtration rate (eGFR), treatment group, albumin, alkaline phosphatase, and body weight were identified as significant covariates. The coefficient of variation (CV%) of random inter-individual variability (IIV) and IOV for CL decreased from 29.2% and 23.1% to 15.5% and 12.3%, respectively, after covariate inclusions. The predictability and stability of the final PK model were confirmed by the model evaluation results.
Among the included patients, 51 (46.4%) and 75 (68.2%) patients developed acute renal and hepatotoxicity during at least one administration cycle, respectively. eGFR (range: 40.2 - 158.7 mL/min/1.73m2, maximum predicted probability change (maxΔP) = -0.929) and sex (for females, ΔP = -0.103) were identified as significant baseline predictors for renal toxicity, and MTX dose (mg/m2) (range 1500 - 8000 mg/m2, maxΔP = 0.86) was the strongest baseline predictor for hepatotoxicity. The exposure metrics of MTX, i.e. the area under the concentration-time curve from 24 hours after drug administration to infinity (AUC24-∞) and concentration at 24 hours (C24h), correlated with renal toxicity only. AUC24-∞ > 109.5 μmol/L*h and C24h > 8.64 μmol/L showed to be potential exposure thresholds predicting a high risk of renal toxicity.
Conclusions: A population PK model was developed which well characterized the PK profile of HD-MTX in CNSL patients. The toxicity analysis identified eGFR, sex, and MTX dose (mg/m2) as baseline predictors for acute renal and hepatotoxicity, respectively. The correlation between MTX exposure metrics and renal toxicity was identified and potential exposure thresholds were suggested. These findings hold the potential for further individualizing HD-MTX dosage and preventing acute organ toxicity, which could potentially improve the outcomes of HD-MTX therapy in CNSL patients.
References:
[1] Fallah, J., L. Qunaj, and A.J. Olszewski, Therapy and outcomes of primary central nervous system lymphoma in the United States: analysis of the National Cancer Database. Blood Advances, 2016. 1(2): p. 112-121.
[2] Schmiegelow, K., Advances in individual prediction of methotrexate toxicity: a review. Br J Haematol, 2009. 146(5): p. 489-503.
[3] Howard, S.C., et al., Preventing and Managing Toxicities of High-Dose Methotrexate. The Oncologist, 2016. 21(12): p. 1471-1482.
[4] Benz-de Bretagne, I., et al., Urinary coproporphyrin I/(I + III) ratio as a surrogate for MRP2 or other transporter activities involved in methotrexate clearance. Br J Clin Pharmacol, 2014. 78(2): p. 329-42.