During myocardial ischaemia, an increase in anaerobic glycolysis leads to acidosis and a reduction in intracellular pH (pH_i). Transient acidosis is beneficial, reducing ATP consumption and cardiac contractility, but sustained ischaemia results in elevated Ca²⁺ levels and myocardial damage. In vivo pH_i measurements can be made by exploiting carbonic anhydrase (CA) mediated exchange between HCO₃^- and CO₂ generated from hyperpolarized [1-¹³C]pyruvate, but low ¹³CO₂ signal (approximately 10% of myocardial H¹³CO₃^-) is often at the detection limit, which has to date restricted this measurement to whole-heart assessment[1,2]. In this abstract, we investigate the feasibility of in vivo myocardial pH_i mapping by combining spectrally-selective excitation on the HCO₃^- and CO₂ resonances with a 3D imaging acquisition.

In vivo study. Male Wistar rats (n=5, weight 500±20 g, HR 380±20 bpm) were scanned prone (Agilent 7T, horizontal bore) using a volume Tx birdcage and 2-channel Rx surface array (Rapid Biomedical). ECG-gated, segmented ¹H GRE images were used for anatomical reference (TR 3.3 ms, TE 1.3 ms, FOV 60x60 mm², matrix 128x128). Continuous dobutamine infusion (n=3, weight 530±20 g, HR 450 bpm, 100 μg/kg/min, 15 min) was used to increase cardiac workload in separate experiments. ECG-gated ¹H cine imaging was used to assess cardiac function.

Polarization. [1-¹³C]pyruvate (14 M) was polarized with trityl radial (OX63, 15 mM) for 1 hour in a prototype DNP hyperpolarizer[3]. Dissolution with NaOH solution resulted in 2 mL/80 mM pre-polarized [1-¹³C]pyruvate, which was injected over 20 seconds via tail vein.

RF pulse design. Fig. 1a shows the effect of a minimum-phase SLR RF pulse, designed to excite HCO₃^- (160 ppm) and CO₂ (124.5 ppm) individually, while suppressing [1-¹³C]pyruvate (170 ppm) (duration 10 ms, TBW 6, 95% excitation: 1.5 ppm, 10^-5 suppression: 6.5 ppm, HCO₃^-: 20°, CO₂: 70°). The spectra are summed over one minute of data acquisition. Fig. 1b shows ECG-gated spectra (TR 1 s) used to demonstrate the SNR gain from large FA excitation and to determine the optimal timing window for imaging of HCO₃^- and CO₂ in the healthy heart. Excitation was alternated between all metabolites (hard pulse, 50 μs, 10°), of HCO₃^- only, or of CO₂ only.

Imaging pulse sequence. Fig. 2a shows the 3D stack-of-spiral ECG-gated pulse sequence used to image cardiac HCO₃^- and CO₂ (FOV 60x60x40 mm³, readout duration 6 ms, resolution 4.5x4.5x5.0 mm², 8 phase encodes, TE 4.3 ms). Sequential CO₂ and HCO₃^- volumes were acquired; the TR differed between the two resonances to allow for regeneration of CO₂ from the larger HCO₃^- resonance (TR 500 ms for CO₂, 150 ms for HCO₃^-). Fig. 2b shows whole-heart ECG-gated spectra acquired using this scheme with 500 ms TR for both resonances, demonstrating that following large flip angle excitation, the CO₂ resonance rapidly regenerates from the HCO₃^- pool. The scan was started 10 seconds after the start of injection.

Image reconstruction and data analysis. Following FFT in k_z, non-Cartesian k-space samples were gridded by NUFFT. Images were corrected for variable flip angle. Linear interpolation was used to align the HCO₃^- and CO₂ images in time. pH_i was estimated using the Henderson-Hasselbalch equation with pK_a = 6.15. pH_i maps were masked using voxels which reached both 20% of the maximum CO₂ and HCO₃^- intensities in time.

Fig. 3 shows dynamic imaging of HCO₃^- and CO₂ in a mid-ventricular slice. Fig. 4 shows volumetric images of HCO₃^- and CO₂ in the rodent heart, summed over the first 20 seconds of data acquisition, at rest and following dobutamine stress. Fig. 5 shows mid-ventricular pH_i maps (20 seconds post-injection) in multiple subjects, demonstrating uniform myocardial pH_i = 7.15±0.04 (mean±SD, n=3) across the whole heart, at rest, consistent with global pH_i measurements in rats and pigs[1,2]. During high cardiac workload, pH_i decreased to 6.90±0.06 (P<0.05, paired t-test).

SNR may be improved by using a shorter duration excitation for CO₂. This requires TE correction which can be performed using a ¹H-T₂* map. Currently, in vivo validation is impossible; in ³¹P-MRS, the pH-sensitive P_i peak overlaps with 2,3-DPG in blood. It may be possible to validate in the perfused heart. Future studies will focus on detecting focal changes in myocardial pH_i, for example, in stress-inducible ischemic states where viable, stunned tissue is locally hypoxic due to reduced perfusion.

We show the feasibility of mapping intracellular pH in the rodent heart in less than a minute, by combining spectrally selective excitation with a rapid 3D imaging readout, and exploiting the rapid regeneration of CO₂ due to CA-mediated exchange.