Article
A Pharmacometric Model Characterizing the Time Course of the Adverse Events in Advanced Non-Small Cell Lung Cancer Patients Treated With Erlotinib
Search Medline for
Authors
Published: | September 25, 2014 |
---|
Outline
Text
Aim: Changing the dosing strategies for kinase inhibitors was speculated to increase the benefits for lung cancer patients [1], [2]. An important factor to be considered with changing doses is the incidence of adverse events (AE). Our aim was to build exposure-driven models characterizing the time courses of transition probabilities between the different grades of skin and gastrointestinal AE, the most frequently encountered AE with the tyrosine kinase inhibitor erlotinib. Such models can help in understanding and predicting the incidence of AE in relation to exposure changes.
Method: Data was provided from stage-IV non-small cell lung cancer (NSCLC) patients (n=39) first-treated with erlotinib (150 mg/day) [3]. 102 incidents of skin AE and 61 incidents of GIT AE of different grades were recorded. A Markov chain-type model was used to predict the likelihood of observing a patient at any of the AE intensity grades or dropping out (due to death or other reasons). This was achieved using a compartmental structure (Figure 1 [Fig. 1]) where 4 compartments represent the AE grades (absent, mild, moderate, and severe), and a fifth compartment to account for dropping out (death or other reasons). The probabilities were left to flow between compartments and the first-order rate constants corresponding to the transition probability constants were estimated. Linear, exponential and Emax models were tested for describing the exposure-AE relationships. Demographics, mutational statuses, additional radiotherapy intervention, and laboratory values were evaluated as covariates. The analysis was performed by estimating the likelihood of the data using the Laplacian estimation method in NONMEM (version 7.3).
Results: Transitioning to higher grades of skin and GIT AE were found to increase linearly with erlotinib plasma concentration. None of the covariates tested were found to affect the probabilities of transitioning between different AE grades. In contrast to other findings [4], experiencing rash was not significantly correlated with positive survival outcomes (p=0.113).Visual predictive checks based on 200 simulation datasets demonstrated the adequate predictive performance of the models.
Conclusion: Exposure-driven models characterizing the proportions of the patients with different severity levels of skin or GIT AE over time were developed. The models adequately simulated the longitudinal observed AE data and thereby will be further used in simulation studies to investigate the safety of the proposed dosing regimens for kinase inhibitors.
References
- 1.
- Chmielecki J, et al. Optimization of dosing for EGFR-mutant non-small cell lung cancer with evolutionary cancer modeling. Sci Transl Med. 2011;3(90):90ra59.
- 2.
- Foo J, et al. Effects of pharmacokinetic processes and varied dosing schedules on the dynamics of acquired resistance to erlotinib in EGFR-mutant lung cancer. J Thorac Oncol. 2012;7(10):1583-93.
- 3.
- Zander T, et al. Early prediction of nonprogression in advanced non-small-cell lung cancer treated with erlotinib by using [(18)F]fluorodeoxyglucose and [(18)F]fluorothymidine positron emission tomography. J Clin Oncol. 2011;29:1701-8.
- 4.
- Lee SM, et al. First-line erlotinib in patients with advanced non-small-cell lung cancer unsuitable for chemotherapy (TOPICAL): a double-blind, placebo-controlled, phase 3 trial. Lancet Oncol. 2012;13(11):1161-70.