To examine the correspondence between these simulations and measurements in vivo, a cohort of mice were scanned using protocols with parameter combinations that the simulations suggested would introduce varying levels of measurement bias. The results (Fig. 5-5) show that when TR=2-s and 𝜃=60o,
exchange is significantly underestimated compared to TR=2=s, 𝜃 =20o (P=0.035), where relatively
accurate measurements are expected. Interestingly, the exchange measured with TR=2-s and 𝜃 =60o
was ~50% lower than the values measured at TR=2-s and 𝜃 =20o. This closely matches the predicted bias
of ~60% seen in Figure 4-3. Additionally, bias is reduced again with TR=7-s and 𝜃 =60o which also agrees
Figure 4-3. Notably the variance is higher under these conditions, which is likely due at least in part to the increased uncertainty in the injection time due to the longer sampling intervals.
Figure 5-5. Comparison of in vivo vs. simulated kinetic data analysis from data acquired using different acquisition parameter combinations. a) Anatomical image of a mouse bearing an anaplastic thyroid tumor and the slice used for dynamic HP spectroscopy. b) Dynamic metabolite curves of the same animal scanned with excitation TR=2-s and 𝜃 =20o (top), TR=2-s and 𝜃 =60o (middle), and TR=7-s and 𝜃
=60o (bottom). c) k
pl values from animals scanned with TR=2-s and 𝜃 =20o (n=3), TR=2-s and 𝜃 =60o
(n=4), and TR=7-s and 𝜃 =60o (n=4). Data acquired with TR=2-s and 𝜃 =60o significantly underestimates
kpl compared to the other two groups (P<0.035).
Section 5.4 Discussion
These simulation results are of limited use in isolation and require validation in physical systems. However, physical systems for repeated controlled hyperpolarized studies are not yet well developed and some inherent challenges remain. When the study endpoint is the characterization of chemical exchange, a dynamic chemical reaction will be needed. Additionally, the system will need to be able to repeatedly carry out the reaction of interest in some controllable manner. These two requirements make working in living systems practically challenging. In order to move away from living systems, a novel dynamic chemical exchange phantom was developed where exchange rates could be controlled. It demonstrated an improved repeatability over in vivo systems. This system was used to validate
simulation predictions that did not assume pyruvate delivery by native vasculature.
This phantom system provides new capabilities for experimental development and validation with distinct advantages over single-tracer injections, static multi-compartment thermal equilibrium phantoms, and in-vivo models. The platform provides dynamic evolution of HP tracer signals through chemical exchange in a manner that is consistent with that observed in target biology and can be tuned to mimic different disease conditions. The spatial characteristics of the phantom are known a priori, allowing rigorous evaluation of data encoding, acquisition, and reconstruction algorithms. This is especially important when considering data reduction strategies that are designed to address key limitations in the measurement of hyperpolarized tracers but that blur traditional definitions of spatial and temporal resolution in the observation of dynamic processes. Static phantoms are useful for confirming some functionality, but do not create the dynamic conditions that could lead to artifacts in
reconstruction algorithms that are based to any extent on the assumption of a stationary subject. Assessment using in vivo models is challenging because of biological heterogeneity and the evolution of target processes in diseases such as cancer that can progress rapidly and increase within-group
variations even in a matter of days. With this platform, acquisitions can be readily repeated, at arbitrary intervals, to extract statistical measures of image properties. The system has a known distribution of metabolites, and could be designed with multiple compartments73 with reaction rates tuned to simulate
different tissues or disease states in parallel. This platform is ideal for exploration of thresholds for detectability of pathologies that may not be evident in 1H MRI, for early testing of new sequences to
ensure preservation of spatial and temporal accuracy, and even for regular quality assurance scans to confirm that similar acquisition, reconstruction, and analysis parameters lead to similar data over time both within and between laboratories and institutions.
Hyperpolarized contrast agents are relatively new, and research into the best practices for signal acquisition, reconstruction, and analysis is ongoing. This dynamic phantom will enable robust,
reproducible, and tunable baseline measurements, providing a benchmark through which experimental strategies can be compared and optimized. This system catalyzes the final step in aerobic glycolysis, the conversion of pyruvate into lactate, without the need for animal subjects, human subjects, or cell suspensions that can increase the cost and the variability of technical measurements. The 14.5-19% variation that we observed is a result of many factors. LDH is sensitive to a range of experimental factors
77; small variations in temperature, pH, or even time from thawing to injection can affect the enzyme
activity and therefore the rate of the reaction. To ensure that the reaction progresses to completion, which is truest to in vivo studies, NADH has to be in excess and thus the rate of the reaction will depend on pyruvate concentration. In this work, the injection of a small amount of hyperpolarized pyruvate was performed by hand, potentially leading to unnecessarily high variations in the final concentration of
pyruvate. This variability can be reduced by utilizing automated injections that are currently under development.
A crucial step in the translation of powerful new imaging technologies into routine preclinical and clinical use is the establishment of well-defined reference standards93 to provide a common
reference against which experimental circumstances can be compared. This reference can be used to ensure comparable results across platforms, laboratories, and institutions, and to aid in study design and execution. This dynamic single enzyme phantom helps fill this critical need. The physical structure of the phantom can be tailored to more closely approximate preclinical or clinical applications, and the rate of the reaction can be controlled through multiple compartments in a spatially-dependent manner to simulate a wide range of disease states. This phantom platform represents a flexible and powerful tool to aid in the development, optimization, validation, and certification of techniques, processes, and instrumentation that are crucial to ensure the successful and efficient clinical translation of powerful new imaging capabilities afforded by MRSI of hyperpolarized tracers such as [1-13C]-pyruvate.
Using the phantom system, the simulation prediction from chapter 4, namely that a low rate of conversion, high excitation angles and rapid repetition times would suppress the apparent production of hyperpolarized lactate, was confirmed. Tuning the phantom system to match the low conversion rate used in the simulations showed a remarkable correlation in the expected mean 𝑘𝑝𝑙′ measured and the
signal evolution curves. This shows that the dynamic enzyme phantom system was an ideal model to validate the simulation architecture in the simplest case, where endogenous vasculature delivery is ignored. Additionally, the in vivo studies show strong agreement with the simulation predictions demonstrating the validity of the simulation architecture to account for perfusion. In aggregate, these results serve as a strong validation of the simulation architecture and support the dual ideas that simulation of hyperpolarized studies is a useful method for developing and optimizing acquisitions.