4291

Making the Transfer: Exploiting Hyperpolarized Bicarbonate-CO2 Exchange for Increased Signal-to-Noise pH Imaging

David E Korenchan¹, Jeremy W Gordon¹, Sukumar Subramaniam¹, Renuka Sriram¹, Peder E Z Larson^1,2, Dan B Vigneron^1,2, Robert R Flavell¹, and John Kurhanewicz^1,2

¹Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States, ²Bioengineering, University of California, Berkeley and University of California, San Francisco, San Francisco, CA, United States

Synopsis

Hyperpolarized (HP) [¹³C]bicarbonate MR imaging can map pH in vivo, but images generally suffer from low CO₂ signal-to-noise ratio (SNR). However, rapid bicarbonate-CO₂ chemical exchange can increase CO₂ SNR via exchange-mediated polarization transfer. We exploit this phenomenon for HP [¹³C]bicarbonate imaging to boost CO₂ SNR by 2.2-fold at pH 7.6, where CO₂ SNR is lowest in the physiologic range, by acquiring and summing multiple transients. Tip angles and delays are chosen using a priori knowledge of exchange rate to increase SNR while mitigating pH error. This approach can potentially improve imaging SNR in vivo for studying extracellular acidosis in cancer.

Purpose

To increase CO₂ signal-to-noise ratio (SNR) for hyperpolarized (HP) [¹³C]bicarbonate pH imaging by exploiting rapid bicarbonate-CO₂ chemical exchange

Methods

Bicarbonate-CO₂ exchange measurements: Non-HP sodium [¹³C]bicarbonate (~20mM) was added to 100mM phosphate plus 7.55µg/mL carbonic anhydrase II (CAII) at pH 6.7. Overall bicarbonate-CO₂ exchange rate was quantified via ¹³C selective inversion as previously described¹ at 37°C on an 11.7T NMR spectrometer. Additionally, HP bicarbonate-CO₂ exchange rates were measured in transgenic adenocarcinoma of the mouse prostate (TRAMP) tumors in a vertical-bore 14T MR scanner via slice-selective ¹³C-NMR spectroscopy plus selective CO₂ saturation, as previously described^2,3 (2-band Gaussian pulse⁴, 20° tip, 10mm slice, 2s TR, 15.5s pre-acquisition delay).

Bicarbonate-CO₂ exchange simulations: HP bicarbonate+CO₂ z-magnetization/signal evolution was simulated using Bloch-McConnell equations, varying pH, exchange rate, T₁, tip angles, TR, and number of excitations. pH was calculated from tip angle-corrected bicarbonate/CO₂ signals using a modified Henderson-Hasselbalch equation, assuming pK_a = 6.17². In determining imaging parameters to maximize CO₂ SNR, a |pH error| ≤ 0.05 and minimum bicarbonate:CO₂ SNR ratio of 1:5 was enforced over pH 6.3-7.8.

Phantom pH imaging, data processing: HP [¹³C]bicarbonate generated via HP [1-¹³C]1,2-glycerol carbonate decarboxylation⁴ was diluted 1:6 in three tubes containing 100mM phosphate, pH 6.4-7.6, and 7.55µg/mL CAII. A 2D echo-planar imaging (EPI) sequence (7° bicarbonate, 63° CO2, 2mm in-plane, 8mm slice, 400ms TR, 39 transients, total time 16s) with spectral-spatial excitation pulses (6.7ms pulsewidth, 300Hz spatial bandwidth, 300Hz spectral pass-band, 3624Hz spectral band separation) was then performed in a vertical-bore 14T MR scanner kept at 37°C. Using custom MATLAB scripts, raw imaging data were Fourier-transformed and pH maps calculated as described above. Regions of interest (ROIs) were drawn within each tube and in a noisy region outside the tubes, and maximum voxel CO₂ SNR and average pH were plotted for each ROI as a function of the number of transients summed.

Results

The overall rate constant for CAII-catalyzed bicarbonate-CO₂ exchange in phantoms was 5.51s^-1 (Figure 1). Bloch-McConnell simulation predicted that by using a 9-fold higher CO₂ excitation than bicarbonate, rapid exchange boosts CO₂ SNR by a factor of 2.2-4.2, depending on pH, as compared with no exchange (Figure 2). This motivated further simulation to identify a sampling scheme to maximize CO₂ SNR and limit pH error. A parameter scheme of [α_bicarb,α_CO2,TR,N_excitations] = [7°,63°,400ms,39] was thus predicted to increase CO₂ SNR by 2.9-fold over a single 90° acquisition at pH 7.8 while keeping |pH error| < 0.05 (Figure 3). This scheme was implemented in a pH phantom with HP [¹³C]bicarbonate using a frequency-selective 2D EPI approach. The number of summed transients to maximize CO₂ SNR depended on tube pH, and average pH remained constant over time except for the low-pH tube (Figure 4), likely due to CO₂ signal depletion. In order to evaluate the approach for use in vivo, bicarbonate-CO₂ exchange kinetics were measured in n=3 TRAMP mice (Figure 5). Average tumor pH was 7.16±0.14, and the forward in vivo exchange rate, k_bc, was 0.15±0.05s^-1, agreeing with measurements in human colorectal carcinoma cell xenografts³. This corresponds with an overall exchange time constant of 1.56±0.04s^-1.

Discussion

We and others have previously demonstrated that exciting bicarbonate with a lower tip angle than CO₂ preserves HP signal for subsequent excitations^4,5. However, Bloch-McConnell simulation predicts that this unequal sampling also causes HP magnetization “shuttling” from bicarbonate to CO₂ via chemical exchange over multiple acquisitions (Figure 1). This phenomenon can be exploited in a 2D EPI approach, given a priori knowledge of exchange rate, in which averaging multiple images together improves CO₂ SNR, particularly for high-pH voxels, without compromising pH accuracy. Importantly, the SNR can be plotted as a function of summed transients on a per-voxel basis in order to determine where the maximum occurs, which will vary with voxel pH (Figure 4). However, pH measurement is only accurate if the injected HP bicarbonate is given sufficient time to equilibrate. Our results suggest that TRAMP prostate tumors have sufficiently rapid exchange, reaching equilibrium in ~2-2.5s (3-4 time constants). By optimizing imaging parameters for this exchange rate and an in vivo T₁ of 10s^2,6, the CO₂ SNR can theoretically increase by 53% at pH 7.6. For TRAMP tumors, pH 7.16, the predicted SNR gain is 45% (data not shown).

Conclusions

By exploiting a priori knowledge of bicarbonate-CO₂ exchange, we have demonstrated significant gains in achievable CO₂ SNR, particularly at high-pH values where CO₂ signal is lower. Future studies will utilize these techniques for in vivo pH imaging.

Acknowledgements

The authors thank all Flavell and Kurhanewicz Lab members.

Grants: R01-CA166655; R01-EB016741; R21-CA-0121429; P41-EB013598; DOD-CA-110032; DOD-PC140571P4.

References

Dahlquist, F. W., Longmuir, K. J. & Vernet, Du, R. B. Direct Observation of Chemical Exchange by a Selective Pulse NMR Technique. J Magn Reson 406–410 (1975).
Gallagher, F. A. et al. Magnetic resonance imaging of pH in vivo using hyperpolarized 13C-labelled bicarbonate. Nature 453, 940–943 (2008).
Gallagher, F. A. et al. Carbonic Anhydrase Activity Monitored In Vivo by Hyperpolarized 13C-Magnetic Resonance Spectroscopy Demonstrates Its Importance for pH Regulation in Tumors. Cancer Res 75, 4109–4118 (2015).
Korenchan, D. E. et al. Dynamic nuclear polarization of biocompatible 13C-enriched carbonates for in vivo pH imaging. Chem Commun 52, 3030–3033 (2016).
Scholz, D. J. et al. Quantified pH imaging with hyperpolarized 13C-bicarbonate. Magn Reson Med 73, 2274–2282 (2014).
Wilson, D. M. et al. Multi-compound polarization by DNP allows simultaneous assessment of multiple enzymatic activities in vivo. J Magn Reson 205, 141–147 (2010).

Figures

Figure 1: Measurement of bicarbonate-CO₂ exchange in a phantom via selective CO₂ inversion and ¹³C NMR dynamic spectroscopy. (a-b) Spectra showing [¹³C]bicarbonate and ¹³CO₂ resonances at (a) 0.03s and (b) 0.54s after selective CO₂ inversion. (c) Exponential fitting of (bicarbonate - CO₂) signal difference in order to obtain k_ex = k_bc + k_cb = 5.51s^-1. T₁ contribution to decay is negligible (R₁ ~ 0.03s^-1 at 11.7T).

Figure 2: Bloch-McConnell simulated gain in CO₂ signal due to bicarbonate-CO₂ exchange as a function of pH. CO₂ signal improves most at high pH values, where more HP signal resides in the bicarbonate pool. Simulation parameters: 2.78° bicarbonate tip angle, 25° CO₂ tip angle, k_ex = 5.51s^-1, TR = 67ms, T₁ = 30s (both metabolites), 64 excitations.

Figure 3: Bloch-McConnell simulation for phantom conditions of CO₂ signal-to-noise ratio at pH 7.8 as a function of bicarbonate/CO₂ tip angles, TR, and number of excitations. For each plot, the unplotted parameters are set to the values that maximize CO₂ SNR. SNR = 0 for parameter combinations that lead to |pH error| > 0.05 unit and/or bicarbonate:CO₂ SNR ratio < 1:5 over simulated pH range. Simulation parameters: T₁ = 30s (both metabolites), k_ex = 4.26s^-1 (adjusted for pH 7.8).

Figure 4: Frequency-selective 2D EPI phantom pH imaging with HP [¹³C]bicarbonate: (a) Bicarbonate, CO₂ and calculated pH images with N_exc = 5 and 39 summed transients. Data are zero-filled 4x for display. More transients improve SNR for mid- and high-pH tubes but reduce SNR for the low-pH tube due to heavy CO₂ sampling. (b-c) Plots of (b) maximum voxel CO₂ SNR and (c) average pH for each tube as a function of summed transients. Decreases in bicarbonate+CO₂ SNR with more transients lead to pH inaccuracy in the low-pH tube. (e) Maximum percentage gains in SNR for each pH tube.

Figure 5: Measurement of in vivo bicarbonate-CO₂ exchange in transgenic murine prostate cancer. (a) Coronal ¹H image of a TRAMP mouse showing tumor and slice placement for dynamic ¹³C NMR spectroscopy. (b) Slice-selective dynamic ¹³C NMR spectra from tumor, showing faster [¹³C]bicarbonate signal decay with selective ¹³CO₂ saturation. Average tumor pH was calculated from spectra to be 7.26. (c) Fitting bicarbonate signal decays before and after selective ¹³CO₂ saturation determines both T₁ decay and forward exchange rate constant, k_bc. Using the average tumor pH, the overall exchange rate constant is calculated as k_ex = k_bc(1 + 10^pH-pKa) = 1.53s^-1.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)

4291