Pulse sequence considerations for quantification of pyruvate‐to‐lactate conversion kPL in hyperpolarized 13C imaging. (21st January 2019)

• Record Type: Journal Article
• Authors: Chen, Hsin‐Yu; Gordon, Jeremy W.; Bok, Robert A.; Cao, Peng; von Morze, Cornelius; van Criekinge, Mark; Milshteyn, Eugene; Carvajal, Lucas; Hurd, Ralph E.; Kurhanewicz, John; Vigneron, Daniel B.; Larson, Peder E.Z.
• Is Part Of: NMR in biomedicine. Volume 32:Number 3(2019)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2019-01-21
• Subjects: cancer imaging -- hyperpolarized C‐13 pyruvate -- kinetic modeling -- pulse sequence; Nuclear magnetic resonance -- Periodicals; Magnetic Resonance Spectroscopy -- Periodicals; 574
• DOI: 10.1002/nbm.4052
• Languages: English
• ISSNs: 0952-3480

Abstract

Abstract : Hyperpolarized 13 C MRI takes advantage of the unprecedented 50 000‐fold signal‐to‐noise ratio enhancement to interrogate cancer metabolism in patients and animals. It can measure the pyruvate‐to‐lactate conversion rate, k PL, a metabolic biomarker of cancer aggressiveness and progression. Therefore, it is crucial to evaluate k PL reliably. In this study, three sequence components and parameters that modulate k PL estimation were identified and investigated in model simulations and through in vivo animal studies using several specifically designed pulse sequences. These factors included a magnetization spoiling effect due to RF pulses, a crusher gradient‐induced flow suppression, and intrinsic image weightings due to relaxation. Simulation showed that the RF‐induced magnetization spoiling can be substantially improved using an inputless k PL fitting. In vivo studies found a significantly higher apparent k PL with an additional gradient that leads to flow suppression ( k PL, FID‐Delay, Crush / k PL, FID‐Delay = 1.37 ± 0.33, P < 0.01, N = 6), which agrees with simulation outcomes (12.5% k PL error with Δ v = 40 cm/s), indicating that the gradients predominantly suppressed flowing pyruvate spins. Significantly lower k PL was found using a delayed free induction decay (FID) acquisition versus a minimum‐ T E version ( k PL, FID‐Delay / k PL, FID = 0.67 ± 0.09, P < 0.01, N = 5), and the lactate peak had broader linewidth than pyruvate (Δ ω lactate /Δ ω pyruvate = 1.32 ± 0.07, P < 0.000 01, N = 13). This illustrated that lactate's T 2 *, shorter than that of pyruvate, can affect calculated k PL values. We also found that an FID sequence yielded significantly lower k PL versus a double spin‐echo sequence that includes spin‐echo spoiling, flow suppression from crusher gradients, and more T 2 weighting ( k PL, DSE / k PL, FID = 2.40 ± 0.98, P < 0.0001, N = 7). In summary, the pulse sequence, as well as its interaction with pharmacokinetics and the tissue microenvironment, can impact and be optimized for the measurement of k PL . The data acquisition and analysis pipelines can work synergistically to provide more robust and reproducible k PL measures for future preclinical and clinical studies. Abstract : This study investigated three pulse sequence components and parameters that can affect estimates of k PL in HP‐ 13 C MRI. In vivo animal tumor studies showed significant impact on k PL with crusher‐gradient induced flow suppression, and intrinsic image weighting due to relaxation, corroborated by signal‐model simulations. RF‐induced magnetization spoiling can be substantially improved using an inputless k PL fitting. These outcomes suggested that the pulse sequence, as well as its interaction with pharmacokinetics and tissue microenvironment, can impact the measurement of k PL .