Núria Buil-Bruna, Lorena de Pablo-Maiso, Sara Zalba, MarÃa J Garrido, Iñaki F Trocóniz
Department of Pharmacy and Pharmaceutical Technology, School of Pharmacy, University of Navarra, Pamplona 30180, Spain
Objectives: Several pharmacodynamic models describing in vivo tumor growth have been reported in literature. However, relatively few models exist for in vitro cell growth [1-3]. This is partly due to current experimental protocols being generally designed for empirical analyses which offer poor prospects for in vivo prediction. Here we evaluate three alternative in vitro study designs for developing (semi)mechanistic cell growth models to predict the antitumor effect of Oxaliplatin (L-OHP).
Methods: Three different study designs were performed on SW480 cells, a human colon carcinoma cell line. The first study (S1) represented a typical in vitro study characterised by continuous drug exposure up to 72 hours, usually performed to obtain empirical statistics (i.e. EC50). In the second (S2) and third (S3) studies cells were exposed to L-OHP for periods of time between 4 and 24 hours. After each exposure time, L-OHP was removed and cells were supplemented with drug-free growth medium. S3 also included a second exposure of L-OHP. The Alamar Blue technique used allowed for multiple measurements of cell count over a period of 168 hours. Data from the three studies were analysed independently with NONMEM VII. To assess the predictive performance of each model, an external validation simulation exercise was performed for each study using the models developed for the remaining two studies. Results were summarised based on the prediction errors computed as the mean absolute performance error (MAE).
Results: The Gompertz model was used to describe the proliferation growth of SW480 cells. The effect of L-OHP was incorporated as an activation of delayed drug induced signal, which was described using four signal transduction (transit) compartments. This process reflected the inhibition of cell proliferation followed by an apoptotic death. All models successfully described their own study data. The lowest MAE resulted when models developed for S2 and S3 were used to evaluate model predictive performance while the worst results were those found after S1 model based simulations. A model integrating data from the three studies is currently under development.
Conclusions: The models obtained with S2 and S3 showed the best predictive performance. Therefore, we recommend in vitro cell growth studies be performed, if possible, according to the proposed design to facilitate the estimation of predictive (semi)mechanistic tumor cell growth models.
References:
[1] Del Bene F, Germani M, De Nicolao G, Magni P, Re CE, Ballinari D, et al. A model-based approach to the in vitro evaluation of anticancer activity. Cancer Chemother Pharmacol 2009 Apr;63(5):827-836
[2]Lobo ED, Balthasar JP. Pharmacodynamic modeling of chemotherapeutic effects: application of a transit compartment model to characterize methotrexate effects in vitro. The AAPS Journal 2002;4(4):212-222
[3]Moreno D, Troconiz IF, Enguita M, Bandres E, Garcia-Foncillas J, Garrido MJ. Semi-mechanistic description of the in-vitro antiproliferative effect of different antitumour agents. J Pharm Pharmacol 2008 Jan;60(1):77-82
Reference: PAGE 21 () Abstr 2576 [www.page-meeting.org/?abstract=2576]
Poster: Oncology