Kinetic Modeling of ¹⁸F-(2S,4R)4-Fluoroglutamine in Mouse Models of Breast Cancer to Estimate Glutamine Pool Size as an Indicator of Tumor Glutamine Metabolism

Proposed compartment model of ¹⁸F-Gln.

Static images of TNBC and ER+ tumor xenografts and response to glutaminase inhibition (post–CB-839 image is right-hand image of each pair). Tumor and blood time–activity curves demonstrate adequate temporal sampling of radiotracer uptake over time. Spheric VOIs used are drawn on each tumor. %ID = percentage injected dose.

Single-compartment model fit (A and C) and double-compartment model fit (B and D) of same representative time–activity curves for ER+ and TNBC mice before treatment. Double-compartment fit separately shows time course of both reversible and trapped compartments. %ID = percentage injected dose.

Logan graphical analysis (A and C) and Patlak graphical analysis (B and D) of same representative time–activity curves for ER+ and TNBC mice before treatment.

Changes in volume of distribution, as estimated by a 1-compartment model, for both TNBC and ER+ mice who underwent treatment of tumor xenografts with GLS inhibitor (left) or placebo (right). VEH = vehicle solution.

(A) Comparison of estimation of volume of distribution for 1- and 2-compartment models after adding Poisson noise to time–activity curves and fitting each noise realization with corresponding model. Boxes reflects 25th, 50th, and 75th percentiles of data, and whiskers are extent of data after outlier removal (1.5 × interquartile range, where interquartile range = 75th–25th percentiles). Diamonds show true V_D before noise is added to fit curve. True value for ER+ mouse before treatment is essentially infinite because of near-zero k₄. (B) Comparison of true vs. estimated V_D for 1- and 2-compartment models with trapping, where diagonal line is line of identity. 1C = 1-compartment; 2C = 2-compartment.