Figure 5

Fitting and validating the computational vascular tumour model with Treatment Schedule 2 (TS2) experimental data. Result of fitting the extended Argyri et al. model to experimental data from TS2 HCT-116 xenograft study. Bayesian model selection was used and the threshold model was found to be the most probable given the data. Parameters are sampled from posteriors displayed in Supplemental Fig. 3. Specifically, parameters values are λ₁ = 0.085 day⁻¹, λ₂ = 2.51 × 10⁻³ day⁻¹, c = 0.60 mg/(day·mm^3p·kg), α = 5.59 × 10⁻³ mg/(mm^3p·kg), d = 0.80 day⁻¹ mm⁻², Bk₁₂ = 0.45 day⁻¹, Bk₂₁ = 0.008 day⁻¹, Bk_e = 0.78 day⁻¹, Fk₁₂ = 0.05 day⁻¹, Fk₂₁ = 0.22 day⁻¹, Fk_e = 0.50 day⁻¹, K_F = 1.65 day⁻¹ mm⁻², α₂ = 720.8 mg/nl, p₁ = 12.24. (A) Shows how the experimentally observed tumour volume varies over a time period of 56 days where the bars represent the standard error of the mean (B) shows how the corresponding best fit model solutions varies over the same time period (solid lines) as well as predicted model solutions for combination treatments (dashed lines). (C) Shows boxplots corresponding to tumour volumes recorded at 56 days. The coloured dashed line corresponds to the mean and the solid black line corresponds to the median. (D) Upper plot shows the bvz and FOLFOX drug concentrations in the plasma for Treatment Schedule 1 as predicted by the model and the lower plot shows the predicted bvz and FOLFOX drug concentrations in the plasma for TS2 as predicted by the model. Initial conditions for tumour volume were chosen to be the first non-zero value from experimental data and initial conditions for the carrying capacity was taken to be this value multiplied by 5. The total normalised root mean squared error for the vehicle data, bvz data and FOLFOX data is 0.24.