Detection of increased pyruvate dehydrogenase flux in the human heart during adenosine stress test using hyperpolarized [1-¹³C]pyruvate cardiovascular magnetic resonance imaging

We prospectively included healthy subjects aged over 18 years with no history of heart disease, no significant medical history, normal electrocardiogram (ECG) and normal echocardiography with a left ventricular (LV) ejection fraction (LVEF) > 50%. The study was conducted according to the Helsinki principles and approved by the National Committee on Health Research Ethics (2018-003533-15) and by the Danish Medicines Agency (2,019,123,690). Written informed consent was obtained from all participants before enrollment. Participants were recruited through local advertisement. The study was performed at the Department of Cardiology and at the MR-Research Centre, Aarhus University Hospital, Aarhus, Denmark between October 2019 and August 2020.

A resting study and a stress study were performed on two separate days (Fig. 1). Participants fasted for 8 to15 h prior to study. Standard blood counts and metabolic profiles before and 7 days after each study day was done. Echocardiography was performed at the first visit (VIVID e9, General Electric Healthcare, Chicago, Illinois, USA) to assess systolic and diastolic function. ECG was recorded to confirm sinus rhythm and exclude abnormalities. Participants ingested 75 g of oral glucose in 200 ml water to maximize PDH flux [6]. Cine CMR imaging was performed with the subjects in the supine position. HP [1-¹³C]pyruvate CMR was performed one hour after glucose ingestion. Before adenosine stress, participants were instructed to refrain from caffeine for 24 h prior to the test. For the stress study, continuous adenosine infusion was initiated 1 h after glucose ingestion at a rate of 140 mcg/kg/min through an 18G venous cannula in the left arm using an infusion pump (Alaris™ VP Plus Guardrails™ Denmark) [10]. Three minutes later, hyperpolarized [1-¹³C]pyruvate was injected through an 18G venous cannula in the right antecubital vein using a power injector (Medrad®, Bayer Healthcare, Denmark) and hyperpolarized CMR was undertaken. A three-slice cine‐LV function assessed LV function during the last minute of adenosine infusion.

The sterile [1-¹³C]pyruvate was polarised in a clinical SPINlab™ DNP polarizer (General Electric Healthcare) and release criteria were as previously described [14]. Proton images were acquired using an eight‐channel cardiac array receiver coil (General Electric Healthcare) or the built-in body coil. The same setup was used for each individual at both study dates. Balanced steady-state free precession (bSSFP) was used for cine LV function assessment. The sequence parameters were: TE 2.4 ms, TR 5.1 ms, flip angle 55°, acquisition matrix 200 × 160, FOV 400 × 400 mm², in-plane resolution 2 × 2.5 mm², slice thickness 8 mm, recon matrix 512 × 512 and cardiac phases 30. HP [1‐¹³C]pyruvate CMR was undertaken with a transmit/receive Helmholtz loop-pair ¹³C coil (PulseTeq Limited, UK) or a transmit clamshell coil with a 16-channel array receive coil (Rapid Biomedical GmbH, Rimpar, Germany). The multiple radiofrequency (RF) coil data were combined using a singular value decomposition (SVD method) [11]. Transmit gain calibration was performed to adjust the RF power levels to the desired flip angles for each subject and ¹³C transmit power was calculated with a Bloch-Siegert method on a [¹³C]-bicarbonate phantom positioned in the coil sensitivity area and close to the imaging plane above the heart [12]. Transmit gain was measured to vary ~ 0.2 dB across subjects. The imaging frequency for the HP ¹³C-imaging was calculated from the proton frequency obtained in the individual heart [13]. Iterative ¹H B0 maps was obtained prior to ¹³C imaging to ensure a local B0 field of ± 15 Hz across the heart. The ¹H centre frequency in the heart was then used to calculate the ¹³C frequencies as the same shim was used. HP imaging was acquired in diastole using a cardiac-gated spectral-spatial (SPSP) excitation with spiral read-out acquisition: TE = 10 ms, FOV = 400 × 400 mm, matrix = 30 × 30, real in-plane pixel size = 13.3 × 13.3 mm and one short axis slice with 30 mm slice thickness [14, 15]. Excitation pulses increase the decay of HP signal. We used a spiral acquisition scheme with an 8° flip angle for the precursor pyruvate to preserve signal and 90° for downstream metabolites lactate, bicarbonate and alanine to maximize metabolite signal. During each R-R interval, we used an excite-read in the diastolic phase for pyruvate and one other metabolite. Time per image was 110 ms and image read-out was 45 ms. We obtained 240 images. Thus, we obtained 120 pyruvate images and 40 images of each metabolite with a time resolution of three heart beats. (Fig. 2).

Reconstruction of the ¹³C data and analysis of the DICOM images were done using Segment version 3.1 R8215 (http://​segment.​heiberg.​se). ROIs outlining the myocardium was firstly drawn on the corresponding heart phase on the cine images and thereafter transferred to the ¹³C images. Here the ROIs were reduced to cover 5% less in endo and epi direction to reduce signal outside the myocardium. All images were tracked manually through the course of the acquisition to minimize motion due to breathing.

The metabolite signal was analysed at the mid-LV level as a mean signal from the myocardium. The area under the curve (AUC) from the peak pyruvate signal and ten time frames forward for [1‐¹³C]pyruvate and each of the metabolites were used to calculate the total myocardial signal using MATLAB (The Mathworks, Natick, Massachusetts, USA) [16, 17]. The exchange rate constants for the conversion of [1¹³C]pyruvate to [¹³C]lactate (k_PL), [¹³C]bicarbonate (k_PB) and [¹³C]alanine (k_PA) were measured as previously described [18, 19].

Statistical analyses were made in GraphPad Prism Windows, (version 8.0.0, GraphPad Software, San Diego, California, USA). Normality was assessed using the Shapiro-Wilks test. Results are presented as mean ± standard deviation (SD). Paired t-tests were used to compare mean values from the two study days. One-tailed p-values were used for time to peak and first order moment pyruvate perfusion. Otherwise two-tailed p-values were used. A p-value < 0.05 was considered statistically significant.

Six healthy subjects, 29 ± 7 years were studied. Echocardiography showed normal diastolic function and CMR showed normal systolic function. Mean body mass index (BMI), mean glycated haemoglobin (HbA1c), mean fasting glucose were similar and within normal range on both study days (Table 1). Time from oral glucose to hyperpolarized [1-¹³C]pyruvate injection was 1 h ± 5 min. During the adenosine stress test, the heart rate (HR) increased from 65 ± 13 to 108 ± 11 bpm (p < 0.001) and the rate-pressure product (mmHg*bpm/10³) increased from 8.2 ± 1.1 to 12.3 ± 1.2 (p = 0.03) (Table 2).

Participants received a total of two successful injections of [1-¹³C]pyruvate. The mean level of polarization was 25 ± 9% and pH was 7.7 ± 0.1. Time from dissolution to injection was < 150 s. No adverse events occurred.

The obtained data of hyperpolarized [1-¹³C]pyruvate had a signal-to-noise ratio (SNR) of 151 ± 67 at rest and 289 ± 189 during stress. The metabolite SNR were increased from rest to stress: 7 ± 3 to 19 ± 9 for [1-¹³C]lactate, 7 ± 3 to 12 ± 5 for [¹³C]bicarbonate and 9 ± 3 to 17 ± 13 for [1-¹³C]alanine (Fig. 4). The time-to-peak (TTP) for metabolites increased from rest to stress: 13 ± 4 s to 16 ± 7 s (p = 0.3) for [1-¹³C]lactate, 13 ± 4 s to 17 ± 7 s for [¹³C] (p = 0.1) and 6 ± 3 to 11 ± 9 s (p = 0.3) for [1-¹³C]alanine. An example of HP [1-¹³C]pyruvate CMR imaging and temporal dynamics are shown in Fig. 3.

TTP perfusion and first order moment (FM) perfusion of [1-¹³C]pyruvate were assessed in the myocardium at the mid-LV level. TTP was significantly reduced from 6.2 ± 2.8 to 2.7 ± 1.3 s, p = 0.02, and FM was reduced from 17.8 ± 5.5 to 7.1 ± 2.0 s, p = 0.005. The k_PL increased from 11 ± 9 *10^–3 s⁻¹ to 20 ± 10 *10^–3 s⁻¹, p = 0.04. The k_PB increased from 4 ± 4 *10^–3 s⁻¹ to 12 ± 7 *10^–3 s⁻¹, p = 0.008. Finally, the k_PA increased from 5 ± 3 *10^–3 s⁻¹ to 16 ± 9 *10^–3 s⁻¹, p = 0.06 (Figs. 4, 5). We found a positive and significant correlation of HR and increase in k_PL (p = 0.02), k_PB (p = 0.002) and k_PA (p = 0.04) (Fig. 6).

Participants were fasted for 8 to 15 h and oral glucose loading was used to maximize PDH flux and standardize metabolic conditions in both the rest and stress study [16, 20, 21]. Adenosine is a validated and widely used pharmacological stress test in cardiac imaging [10, 22, 23]. We found significant increases in HR, rate pressure product and cardiac output during the adenosine stress. The HP observations and the haemodynamic response must reflect both the systemic and coronary effects of adenosine. Adenosine induces coronary and systemic vasodilatation and reflex tachycardia [24]. Semi-parametric measures of the large blood pool component of [1-¹³C]pyruvate was used to evaluate the hemodynamic response to adenosine. We found a significant reduction in the TTP and FM for [1-¹³C]pyruvate in the mid-LV myocardium (inversely correlated to perfusion [25]) consistent with increased HR and thus pyruvate delivery. The increased myocardial signal of pyruvate during stress, suggest an enhanced myocardial [1-¹³C]pyruvate uptake likely due to coronary vasodilation, consistent with previous experimental findings [26]. However, unfortunately we cannot distinguish how much of the increased [1-¹³C]pyruvate signal is derived from increased myocardial uptake and how much is caused by increased vascular signal due to vasodilation. In this regard, it is interesting that we observed a tendency towards increased [1-¹³C]alanine signal which has been suggested as a surrogate measure for pyruvate uptake [16]. We observed a statistically significant increases in PDH flux as well as [1-13C]pyruvate to [1-¹³C] lactate exchange during stress. The relative increase in [1-¹³C]lactate signal is explained by [1-¹³C]pyruvate to [1-¹³C]lactate exchange, which depend on the delivery of pyruvate, on the rate of pyruvate uptake and subsequently on the concentration of LDH and its substrate pool sizes. The relative increase in PDH flux was larger, demonstrating that the healthy heart can increase oxidative energy production during moderate stress. This agrees with previous animal studies and invasive human studies, showing activation of PDH and increased pyruvate oxidation in response cardiac stress [27]. Interestingly, a similar metabolite dynamic pattern (TTP) of the individual metabolites was observed, which seem to contradict the increased metabolic conversion (TTP of the individual metabolite dynamic curve is inversely correlated with the apparent rate constant of the individual metabolites) [28]. It is important to note that the temporal dynamics of the metabolites are 3 times lower than that of pyruvate and thus, could at least partly explain this surprising finding. Also, given the voxel size, partial volume effects may influence these results. Finally, due to coronary vasodilation, coronary vasculature may be a significant part of the pyruvate signal in the myocardium and thus may influence TTP. Further studies are needed to solve these issues.

In line with previous studies [10‐14], we found that all participants tolerated HP [1-¹³C]pyruvate well and all participants returned for the second examination. Furthermore, we found no adverse reactions when HP [1-¹³C]pyruvate infusion was combined with adenosine infusion.