Dynamic phosphorus MR spectroscopy (³¹P-MRS), allows direct estimation of mitochondrial capacity. Typically, only rough localization by the sensitive volume of the surface coil is used, due to high SNR, methodological simplicity and robustness. However, such localization often mixes signals from several muscle groups that are involved differently in the exercise performed (e.g., soleus and gastrocnemius during plantar flexion)¹. To overcome this issue, single-muscle localization techniques (e.g., semi-LASER¹ or DRESS²) can be used. Moreover, in order to measure mitochondrial metabolism in several muscles simultaneously, ³¹P-MRS imaging, using lengthy exercise protocols (i.e., gated ³¹P-MRSI³), or frequency-selective ³¹P-MRI^4-6, providing only limited amount of information in comparison to MRSI, have been proposed.

To overcome the low spatial and temporal resolution of the standard ³¹P-MRSI, caused by slow Cartesian readout, we aimed to develop a ³¹P-MRSI sequence using rapid readout, for spatially resolved quantification of mitochondrial capacity in the human calf muscles during plantar flexion at 7T.

To accelerate the acquisition and, hence, increase the temporal resolution of dynamic ³¹P-MRSI scans, we have implemented a constant-density spiral spectroscopic readout^7,8 (Fig.1). To compensate for the increased dwelltime with increasing matrix size, temporal interleaving was employed. The number of temporal interleaves was adjusted to the targeted matrix size (i.e., n=5 for 14x14 matrix), making sure that the readout bandwidth remained >1.45 kHz (i.e., 12 ppm). The readout bandwidth/dwelltime was set to minimize the deadtime between adjacent spirals as this optimizes SNR. Spiral samples were two-dimensionally gridded with a Kaiser-Bessel kernel and the entire spiral-trajectory calculation, gradient-delay correction and data reconstruction was implemented on our 7T MR system (Siemens Healthcare, Erlangen, Germany), equipped with an SC72CD gradient system featuring 70 mT/m nominal gradient strength and 200 mT/m/s maximum slew rate.

Ten healthy individuals (7M/3F, age: 28.3±4.1 years, BMI: 22.7±2.6 kg.m^-2) participated in the study. Subjects were positioned supine on a plantar-flexion ergometer (Ergospect, Innsbruck, Austria) with a ¹H/³¹P surface RF-coil (10 cm diameter, Rapid Biomedical, Rimpar, Germany) fixed to the right calf. The proposed 2D-MRSI sequence was used for the dynamic protocol (rest/exercise/recovery – 2/6/6 minutes) and its parameters were set as follows: TE*=1 ms; TR=2 s; temporal resolution=10 s; acquisition bandwidth (BW)=1450 Hz; matrix size=14x14; FOV=200x200 mm; slice thickness=30 mm, and the nominal voxel size was 14.3x14.3x30 mm³ (i.e., 6.1 mL). During the exercise phase volunteers performed plantar flexions at a workload of ~35% of maximal voluntary contraction, once every TR.

All spectra were analyzed using AMARES in jMRUI. The γ-ATP peak was used as a concentration reference (8.2 mM). The parameters of oxidative metabolism (e.g., time-constant of PCr recovery [τ_PCr], initial PCr recovery rate [V_PCr] and mitochondrial capacity [Q_max]) were calculated and compared between three muscle groups (gastrocnemius medialis [GM], gastrocnemius lateralis [GL] and soleus [SOL]) by a one-way ANOVA and a Bonferroni post-hoc test.

Representative spectra acquired at rest and at the end of exercise from each muscle group are given in Fig.2. To demonstrate the importance of localization, a voxel combining GM and SOL tissue is also visualized (green). The mixture of signals arising from differently active muscles causes a splitting of the Pi signal, representing two different pH values. Voxels directly between adjacent muscles have been, therefore, avoided in our analysis. The temporal and spatial resolution of our spiral-MRSI sequence enabled mapping of the time evolution of ³¹P metabolites in all visible muscles simultaneously. Sample time courses of the PCr and Pi signals for GL, GM and SOL are depicted in Fig.3. Readily visible lower PCr depletion in the SOL muscle in comparison to GM and GL has been also found statistically significant (p<0.001). Similarly, lower VPCr and Qmax values have been calculated for SOL muscle, however this can be explained by minimal participation of SOL in the performed exercise and, hence, low activation of its mitochondria. All measured parameters of oxidative metabolism are given in Table 1.

In comparison to our approach, gated ³¹P-MRSI for dynamic experiments³ suffered from slow Cartesian readout, which significantly limited the spatial resolution and even required a prolonged exercise protocol. Due to recent development, spectrally-selective ³¹P-MRI allows slightly higher spatial (1.6-3.0 ml)^5,6 and similar temporal resolution (~10 s) at 7T, and even enables quantification of pH⁶. However, ³¹P-MRI still provides only limited amount of information in comparison to MRSI techniques, e.g., no ATP signal for concentration quantification.