Limits...
Modeling and predicting optimal treatment scheduling between the antiangiogenic drug sunitinib and irinotecan in preclinical settings.

Wilson S, Tod M, Ouerdani A, Emde A, Yarden Y, Adda Berkane A, Kassour S, Wei MX, Freyer G, You B, Grenier E, Ribba B - CPT Pharmacometrics Syst Pharmacol (2015)

Bottom Line: Through a log-likelihood ratio test, we conclude that there is a potential synergistic interaction between sunitinib when administered in combination with irinotecan in preclinical settings.Model simulations were then compared to data from a follow-up preclinical experiment.We conclude that the model has predictive value in identifying the therapeutic window in which the timing between the administrations of these two drugs is most effective.

View Article: PubMed Central - PubMed

Affiliation: Inria, project-team Numed, Ecole Normale Supérieure de Lyon Lyon France.

ABSTRACT
We present a system of nonlinear ordinary differential equations used to quantify the complex dynamics of the interactions between tumor growth, vasculature generation, and antiangiogenic treatment. The primary dataset consists of longitudinal tumor size measurements (1,371 total observations) in 105 colorectal tumor-bearing mice. Mice received single or combination administration of sunitinib, an antiangiogenic agent, and/or irinotecan, a cytotoxic agent. Depending on the dataset, parameter estimation was performed either using a mixed-effect approach or by nonlinear least squares. Through a log-likelihood ratio test, we conclude that there is a potential synergistic interaction between sunitinib when administered in combination with irinotecan in preclinical settings. Model simulations were then compared to data from a follow-up preclinical experiment. We conclude that the model has predictive value in identifying the therapeutic window in which the timing between the administrations of these two drugs is most effective.

No MeSH data available.


Related in: MedlinePlus

Experiment 2. A comparison of the predictions of the noninteraction and interaction models. (a) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 4 (combination sunitinib with irinotecan on day 2). (b) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 5 (combination sunitinib with irinotecan on day 15). (c) Observed tumor sizes vs. the simulated tumor sizes for the noninteraction model. (d) Observed tumor sizes vs. the simulated tumor sizes for the interaction model.
© Copyright Policy - creativeCommonsBy-nc-nd
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4759705&req=5

psp412045-fig-0004: Experiment 2. A comparison of the predictions of the noninteraction and interaction models. (a) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 4 (combination sunitinib with irinotecan on day 2). (b) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 5 (combination sunitinib with irinotecan on day 15). (c) Observed tumor sizes vs. the simulated tumor sizes for the noninteraction model. (d) Observed tumor sizes vs. the simulated tumor sizes for the interaction model.

Mentions: Figure3 provides a schematic view of the sunitinib‐irinotecan combination model, while Figure4 shows a comparison of the noninteraction and the interaction models. In Figure4a, we show a comparison of the simulations of both the noninteraction and interaction models along with the data of Group 4 (combination sunitinib with irinotecan on day 2). In Figure4b, we show a comparison of the simulations of both the noninteraction and interaction models along with the data of Group 5 (combination sunitinib with irinotecan on day 15). While the argument can be made for both models accurately describing the data of Group 4, it is clear that the noninteraction model does not capture the phenomenon observed for the data of Group 5. We continue by quantifying the observation that the interaction model more accurately describes our data. We show the observed tumor sizes vs. the simulated tumor sizes for the noninteraction model (Figure4d) and interaction model (Figure4c). We then consider the least squares linear fit of these data. Visually, it is clear that the regression line of the interaction model (Figure4d) is more closely aligned with the identity line than the noninteraction model (Figure4c). This is evidence that the interaction model might be better suited to our data. Since the interaction model is nested within the noninteraction model, a log‐likelihood ratio test was performed to determine if the difference in likelihoods is statistically significant. The result of the test shows a change in likelihood equal to −5.99. This indicates that the interaction model fits the data significantly better (P < 0.05) than the noninteraction model. The simulation of Experiment 2 with the interaction model is shown in Figure5, where we see good concordance between model and data in all five arms of the experiment.


Modeling and predicting optimal treatment scheduling between the antiangiogenic drug sunitinib and irinotecan in preclinical settings.

Wilson S, Tod M, Ouerdani A, Emde A, Yarden Y, Adda Berkane A, Kassour S, Wei MX, Freyer G, You B, Grenier E, Ribba B - CPT Pharmacometrics Syst Pharmacol (2015)

Experiment 2. A comparison of the predictions of the noninteraction and interaction models. (a) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 4 (combination sunitinib with irinotecan on day 2). (b) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 5 (combination sunitinib with irinotecan on day 15). (c) Observed tumor sizes vs. the simulated tumor sizes for the noninteraction model. (d) Observed tumor sizes vs. the simulated tumor sizes for the interaction model.
© Copyright Policy - creativeCommonsBy-nc-nd
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4759705&req=5

psp412045-fig-0004: Experiment 2. A comparison of the predictions of the noninteraction and interaction models. (a) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 4 (combination sunitinib with irinotecan on day 2). (b) Comparison of the simulations of both the noninteraction and interaction models along with the data of Group 5 (combination sunitinib with irinotecan on day 15). (c) Observed tumor sizes vs. the simulated tumor sizes for the noninteraction model. (d) Observed tumor sizes vs. the simulated tumor sizes for the interaction model.
Mentions: Figure3 provides a schematic view of the sunitinib‐irinotecan combination model, while Figure4 shows a comparison of the noninteraction and the interaction models. In Figure4a, we show a comparison of the simulations of both the noninteraction and interaction models along with the data of Group 4 (combination sunitinib with irinotecan on day 2). In Figure4b, we show a comparison of the simulations of both the noninteraction and interaction models along with the data of Group 5 (combination sunitinib with irinotecan on day 15). While the argument can be made for both models accurately describing the data of Group 4, it is clear that the noninteraction model does not capture the phenomenon observed for the data of Group 5. We continue by quantifying the observation that the interaction model more accurately describes our data. We show the observed tumor sizes vs. the simulated tumor sizes for the noninteraction model (Figure4d) and interaction model (Figure4c). We then consider the least squares linear fit of these data. Visually, it is clear that the regression line of the interaction model (Figure4d) is more closely aligned with the identity line than the noninteraction model (Figure4c). This is evidence that the interaction model might be better suited to our data. Since the interaction model is nested within the noninteraction model, a log‐likelihood ratio test was performed to determine if the difference in likelihoods is statistically significant. The result of the test shows a change in likelihood equal to −5.99. This indicates that the interaction model fits the data significantly better (P < 0.05) than the noninteraction model. The simulation of Experiment 2 with the interaction model is shown in Figure5, where we see good concordance between model and data in all five arms of the experiment.

Bottom Line: Through a log-likelihood ratio test, we conclude that there is a potential synergistic interaction between sunitinib when administered in combination with irinotecan in preclinical settings.Model simulations were then compared to data from a follow-up preclinical experiment.We conclude that the model has predictive value in identifying the therapeutic window in which the timing between the administrations of these two drugs is most effective.

View Article: PubMed Central - PubMed

Affiliation: Inria, project-team Numed, Ecole Normale Supérieure de Lyon Lyon France.

ABSTRACT
We present a system of nonlinear ordinary differential equations used to quantify the complex dynamics of the interactions between tumor growth, vasculature generation, and antiangiogenic treatment. The primary dataset consists of longitudinal tumor size measurements (1,371 total observations) in 105 colorectal tumor-bearing mice. Mice received single or combination administration of sunitinib, an antiangiogenic agent, and/or irinotecan, a cytotoxic agent. Depending on the dataset, parameter estimation was performed either using a mixed-effect approach or by nonlinear least squares. Through a log-likelihood ratio test, we conclude that there is a potential synergistic interaction between sunitinib when administered in combination with irinotecan in preclinical settings. Model simulations were then compared to data from a follow-up preclinical experiment. We conclude that the model has predictive value in identifying the therapeutic window in which the timing between the administrations of these two drugs is most effective.

No MeSH data available.


Related in: MedlinePlus