Influence of parameter accuracy on pharmacokinetic analysis of hyperpolarized pyruvate

Purpose: To explore the effects of noise and error on kinetic analyses of tumor metabolism using hyperpolarized [1-¹³C] pyruvate. Methods: Numerical simulations were performed to systematically investigate the effects of noise, the number of unknowns, and error in kinetic parameter estimates on kinetic analysis of the apparent rate of chemical conversion from hyperpolarized pyruvate to lactate (k_PL). A pharmacokinetic model with two physical and two chemical pools of hyperpolarized spins was used to generate and analyze the synthetic data. Results: The reproducibility of k_PL estimates worsened quickly when peak signal-to-noise ratio for hyperpolarized pyruvate was below approximately 20. The accuracy of k_PL estimates was most sensitive to errors in high excitation angles, the vascular blood volume fraction (v_b), and the rate of pyruvate extravasation (k_ve), and was least sensitive to errors in the T₁ of pyruvate. When v_b and/or k_ve were fit as additional unknowns, the accuracy of k_PL estimates suffered, and when the vascular input function of pyruvate was also fit, the reproducibility of k_PL estimates worsened. Conclusions: The accuracy and precision of k_PL estimates improve substantially for peak signal-to-noise ratio above approximately 20. Accurate estimates of perfusion parameters (combinations of v_b, k_ve, and the pyruvate vascular input function) and transmit calibration at high excitation angles have the greatest effect on the accuracy of kinetic analyses. Magn Reson Med 79:3239–3248, 2018.