Anzeige
Erschienen in:
24.04.2019 | Original Article
Application of time-dependent modeling for the exposure–efficacy analysis of ceritinib in untreated ALK-rearranged advanced NSCLC patients
Erschienen in: Cancer Chemotherapy and Pharmacology | Ausgabe 3/2019
Einloggen, um Zugang zu erhaltenAbstract
Purpose
Ceritinib 750 mg/day was approved for the treatment of patients with untreated anaplastic lymphoma kinase (ALK)-rearranged non-small cell lung cancer (NSCLC) based on ASCEND-4 study. The objective of this article is to introduce the use of time-dependent modeling approach in the updated exposure–efficacy analysis of ceritinib for the first-line indication.
Methods
Exposure–efficacy analyses, including data from 156 patients, were first conducted using time-independent logistic regression model for response of complete or partial response and Cox regression model for progression-free survival (PFS). The exposure measure used was average Ctrough, which is defined as the geometric mean of all evaluable Ctrough for each patient. To further investigate the impact of exposure measure on exposure–efficacy analyses, a time-dependent modeling approach was used, where exposure at different time intervals was associated with the corresponding response endpoints in a longitudinal manner.
Results
With exposure measure being average Ctrough, it was observed that higher exposure was associated with reduced efficacy in terms of response (odds ratio = 0.77) and PFS [hazard ratio (HR) = 1.12]. These time-independent models do not account for the impact of time-varying concentration due to dose modifications. Subsequently, a new time-dependent modeling approach was used, where exposure and efficacy were associated longitudinally in the analyses. The results showed that the odds ratio of response became 1.07, and the HR of PFS became 1.04, indicating no apparent reverse relationship between exposure and efficacy across the exposure range studied.
Conclusion
The drug effect on efficacy in clinical trials could be better characterized using time-dependent exposure–response models.