NMR allows to investigate multiple aspects of physiological parameters in vivo such as regional perfusion, blood and tissue oxygenation, intracellular pH or high-energy phosphate metabolism. Classically, NMR acquisition schemes rarely explore more than a few biological parameters which are often measured in separate experimental sessions, adding experimental variability¹ on biological processes which are already multifactorial^2,3. The benefit of interleaved multinuclear dynamic NMR imaging and spectroscopy has been demonstrated and exploited for decades, but only on modified prototype scanners^4-8. This has limited the impact of such an approach in spite of the wealth of additional information it brings on muscle physiological and biochemical regulation^1,4,9-11. With the latest hardware configuration available on advanced commercial scanners, obstacles have been lifted. Here we evaluated an interleaved pulse sequence combining the NMR acquisition of a ¹H image and ³¹P spectrum in a clinical system without any hardware modifications done by the customer. T₂* changes induced by the blood oxygen level dependent (BOLD) response as well as the phosphocreatine (PCr) and inorganic phosphate (Pi) concentrations were monitored during and after exercise in the calf muscle.

Experimental Setup

Two healthy male subjects (29±1 years old) participated in this study. The protocol was approved by the local ethical committee. NMR examinations were done on the right calf muscle. An amagnetic pneumatic ergometer interfaced with LabVIEW (National Instruments, Austin, Texas, USA) was used during plantar flexions (starting resistance 70 N, 10 N increase every 90s).Experiments were done on a 3-T/60-cm bore Siemens Magnetom Prisma MR system (Siemens Healthcare, Erlagen, Germany) equipped with the multi-nuclear option. Contrary to previous models from this constructor, the ¹H and X-nuclei RF channels are separated, allowing an independent reception for each nucleus. A dual-tuned ¹H/³¹P flex transceiver coil was used (RAPID Biomedical GmbH, Rimpar, Germany). The coil was wrapped around the calf with the 11-cm-diamater surface ³¹P coil facing the gastrocnemius muscle.

Interleaved NMR

Muscle BOLD response to exercise and energy metabolism was studied within the same paradigm by interleaving a ³¹P MRS non-localized acquisition and a GRE MRI module within a single repetition. A data set was generated every 3 s over a total scan duration of 15 mins. Exercise began after 1 min and lasted 5 mins (1/3 Hz pedaling frequency). Each data set consisted of a ³¹P MRS spectrum (0.5-ms-long square pulse, 0.2 ms acquisition delay, 1024 complex points, 4 kHz Bandwidth) and a T₂*-weighted ¹H Fast low-angle shot (FLASH) image (10 mm thickness, 256x64 matrix size, 1.5x1.5 mm in-plane resolution, 330 Hz/Pixel). Images were acquired with alternating TEs (= 5, 15, 25 ms).

Data Analysis

The raw interleaved data was first exported and then separated and reconstructed into imaging and spectroscopy data sets using in-house Matlab routines (The MathWorks, MA, USA). No motion correction was applied. ³¹P spectra were zero-filled to 2048 points and zero- and first-order phase corrections were applied. Normalized concentrations were obtained by integrating the area under the curve of PCr (0±0.45 ppm) and Pi (5±0.45 ppm) and dividing them by the PCr integral at rest. T₂* maps were estimated from each consecutive set of 3 images and were fitted with a 2-parameter monoexponential function. Average T₂* values were measured over two regions-of-interest (ROI) in the gastrocnemius muscle.

Figure 1 shows a single ³¹P spectrum. Figure 2 shows the evolution of PCr and Pi during the experiment, depicting a depletion of PCr to 55% at the end of the exercise. Figure 3 displays the time courses of the average T₂* values in the internal and external gastrocnemius, increasing from 25.2 ms at rest to 26.4 ms and 27.7 ms during recovery, respectively. Figure 4 shows T₂*-maps before and after exercise, evidencing the solicited muscles during plantar flexions.

A ¹H/³¹P interleaved pulse sequence was developed on a modern clinical scanner without the need of hardware modifications from the customer. Although the sequence was relatively simple, no major obstacles are foreseen for future extensions including ASL perfusion, ¹H spectroscopy of deoxymyoglobin, lactate editing or localized ³¹P MRS. Motion correction¹³ and cardiac triggering¹⁴ could in principle also be included to improve perfusion measurements quality during exercise.

The necessity of hardware modifications^4-8,12 has severely hampered the diffusion and exploitation of multiparametric acquisitions techniques by specialists of muscle physiology and metabolism as well as by clinical investigators, despite the numerous possible applications^8-11. Having the possibility to run synchronous multinuclear sequences on unmodified clinical systems is expected to accelerate and encourage their inclusion in clinical studies as well as in providing new insights into muscular energy regulation processes.