Sule Sahin1,2, Shuyu Tang3, Manushka Vaidya2, and Peder E.Z. Larson2
1Graduate Program in Bioengineering, University of California, Berkeley and University of California, San Francisco, Berkeley, CA, United States, 2Radiology, University of California, San Francisco, San Francisco, CA, United States, 3HeartVista, Inc., Los Altos, CA, United States
Synopsis
An alternate to AUC ratio, fitting pyruvate to lactate rate constants
(kPL) can be a powerful tool for quantification of hyperpolarized [1-13C]pyruvate
studies. In this work, a model was developed to fit kPL values to a novel
acquisition method where lactate was acquired with a stack-of-spiral bSSFP
sequence. The model was utilized to fit kPL on three sets of in vivo data:
healthy rat kidneys, mouse prostate tumors and human kidney tumors. It was
shown that the fit kPL values matched those fit using an established GRE
fitting method for complimentary GRE-acquired data sets.
Introduction
In hyperpolarized [1-13C]pyruvate imaging, one method of
quantification for the modality is determining kinetic rate constants between
pyruvate and its’ metabolites (lactate, alanine etc.) to characterize tissue
metabolism. Recently, Tang et al. acquired metabolite-specific 3D images using
a stack-of-spiral balanced steady state free precession (bSSFP) sequence to
acquire lactate1. This resulted in a 2.5-fold SNR improvement
for lactate imaging. The acquired data were quantitatively evaluated by
observing area under the curve (AUC) ratio but has yet to be fit for the
pyruvate to lactate kinetic rate constant (kPL) due to additional relaxation
mechanisms and metabolic conversion during the bSSFP acquisition1. In this work, a model was developed and used
to fit ROI-averaged time course data acquired using this bSSFP sequence on in
vivo datasets in healthy rat kidneys, mouse prostate tumors, and human kidney
tumors. Mapping the rate constants of these studies may give us new metabolic
information, better specificity of healthy and cancerous tissues, and allow us
to take advantage of the higher SNR achieved using this novel acquisition
method. Methods
Two
different acquisition methods were used for hyperpolarized [1-13C]pyruvate
imaging. In “lactate-bSSFP” experiments (figure 1 and 2), lactate was
acquired with a 3D stack-of-spiral bSSFP sequence while pyruvate was acquired
with a 3D spoiled gradient-recalled echo (GRE) sequence. The flip angles for
pyruvate and lactate were 3° and 60° for the animal studies and 20° and 60°,
respectively, for the human studies. In “GRE-all” experiments, both pyruvate
and lactate were acquired with a 3D GRE sequence. The flip angles in this experiment for pyruvate
and lactate for the animal studies were 3° and 7.67°, and for the human
studies, 20° and 30°, respectively1. The images
acquired were dynamic with 30 time points and 4 second temporal resolution for the
animal studies and 3.5 second temporal resolution for the human studies.
A
model was developed to simulate the “lactate-bSSFP” experiment based on
simplifications to the Bloch equations in matrix form. Assumptions made to
simplify the model include no off-resonance, instantaneous RF pulses, metabolic
conversion only in the longitudinal magnetization components and a negligible lactate
to pyruvate rate constant. To model the bSSFP acquisition of lactate, the
acquisition was broken up into two parts (figure 2 and 3). In (A), the RF pulses
were modelled as a rotation matrix about the y-axis with flip angle α. Only the
transverse (x and y) components of the lactate signal were considered since the
pyruvate transverse magnetization is spoiled with GRE acquisition. In (B), the
T1 and T2 relaxation and metabolic conversion over the rest of TR was considered.
The state matrix, S, and
the equation for part (B) in figure 3 were derived from the differential
equations relating the rate constants to the rates of the metabolites2. For each
excitation these two steps were repeated, and the magnitude values were averaged
to determine one value for each time point. The pyruvate signal was modelled
with a similar state matrix but without considering T2 relaxation and
simplified using the cosine and sine of the flip angle.
This model was transformed
into a fitting function using nonlinear least squares. The fitting was
evaluated on three sets of “lactate-bSSFP” experiment data: healthy rat kidneys
(N=6, 3 rats with two kidneys each), transgenic adenocarcinoma of mouse
prostate (TRAMP) (N=3), and human patients with kidney tumors (N=3) to
determine kPL values. ROI-averaged (tumor ROIs, for the TRAMP and human
patients) time courses were used for the fitting (figure 4) with fixed T1 values
of 20 seconds for lactate and pyruvate and fixed T2 value of 1 second for lactate3,4. As a control,
both “lactate-bSSFP” and “GRE-all” experiments were also fit for kPL values
using a inputless GRE kPL fitting function2. MATLAB and
Python were used for modelling and fitting5. Results and Conclusions
The bSSFP fitting for “lactate-bSSFP”
experiment data was successful and the average R squared value for pyruvate and
lactate over all the data was 0.87 and 0.943 respectively (figure 4). Additionally,
the kPL values from fitting “lactate-bSSFP” experiment data with the bSSFP
model better matched the kPL values of the “GRE-all” experiment data fit with
the GRE model compared to fitting “lactate-bSSFP” experiment data with the GRE
model (figure 5). This was as expected due to mismatch between the model and
data acquisition, motivating the use of this bSSFP model for “lactate-bSSFP” experiments.
This work describes a
novel fitting model for lactate acquired with a stack-of-spiral bSSFP sequence
in hyperpolarized [1-13C]pyruvate imaging and demonstrates that the
attained kPL values are equivalent to those attained if GRE acquisition and
fitting were used, allowing for quantification of kinetic rates with this
highly efficient pulse sequence.Acknowledgements
This work was supported by NIH Grant P41EB013598 and American Cancer
Society Grant RSG‐18‐005‐01‐CCE.References
- Tang,
S. et al. A metabolite‐specific 3D stack‐of‐spiral bSSFP sequence for
improved lactate imaging in hyperpolarized [1‐ 13 C]pyruvate studies
on a 3T clinical scanner. Magn. Reson. Med. 84, 1113–1125 (2020).
- Larson, P. E. Z. et al.
Investigation of analysis methods for hyperpolarized 13C-pyruvate metabolic MRI
in prostate cancer patients. NMR Biomed. 31, 1–17 (2018).
- Bahrami, N., Swisher, C. L., Von Morze,
C., Vigneron, D. B. & Larson, P. E. Z. Kinetic and perfusion modeling of
hyperpolarized (13)C pyruvate and urea in cancer with arbitrary RF flip angles.
Quant. Imaging Med. Surg. 4, 24–32 (2014).
- Yen, Y. F. et al. T2 relaxation
times of 13C metabolites in a rat hepatocellular carcinoma model measured in
vivo using 13C-MRS of hyperpolarized [1-13C]pyruvate. NMR Biomed. 23,
414–423 (2010).
- Hyperpolarized-MRI-Toolbox.
Available online at: https://github.com/LarsonLab/hyperpolarized-mri-toolbox DOI: 10.5281/zenodo.1198915