Supplementary Information from Functional Parameters Derived from Magnetic Resonance Imaging Reflect Vascular Morphology in Preclinical Tumors and in Human Liver Metastases

Supplementary methods and data. Supplementary Table 1. Equations of chosen models fit using ordinary least squares regression with the best Bayesian Information Criterion for iAUC90, Ktrans, and BATfrac in untreated preclinical tumors (cohort 1). Goodness of fit for each model was assessed by two measures, residuals vs fitted and Q-Q plots, which are shown in Supplementary Figure 4. Supplementary Table 2. Evaluation of model fits for iAUC90, Ktrans, and BATfrac generated from cohort 1 applied to preclinical data from cohort 2 treated with control IgG or with anti-VEGFR2 antibody DC101. Goodness of fit for each model was assessed by two measures, residuals vs fitted and Q-Q plots, which are shown in Supplementary Figures 6-7. Supplementary Table 3. Equations of chosen linear models with best Bayesian Information Criterion for iAUC90 and Ktrans in clinical tumors. Supplementary Figure 1. Signal-to-noise ratio on DCE-MRI is affected > 5 mm away from surface coil. Supplementary Figure 2. Longitudinal, repeated imaging with CT curbs tumor growth. Supplementary Figure 3. Vascular parameters (vessel length (a), tortuosity (b), and radius (c)) correlate poorly with volume of MC38 and FaDu tumors. Supplementary Figure 4. For each MR parameter of untreated preclinical tumors of cohort 1, the chosen linear model has normal residuals, as assessed by residual vs fitted plots and Q-Q plots, demonstrating goodness of fit. Supplementary Figure 5. For each MR parameter of untreated preclinical tumors of cohort 2, the tested linear model has normal residuals, as assessed by residual vs fitted plots and Q-Q plots, demonstrating goodness of fit. Supplementary Figure 6. Goodness-of-fit, as assessed by residual vs fitted plots and by Q-Q plots, for linear models applied to DC101-treated preclinical tumors of cohort 2. The chosen linear models for iAUC and BAT have residuals that exhibit departures from normality, indicating that the treatment altered these parameters. Supplementary Figure 7. For each MR parameter of clinical tumors, the chosen linear models demonstrate normal residuals, indicating goodness-of-fit.