1785

Quantification of Low-Signal Metabolites in 7 T 31P Spectra of the Human Myocardium
Carla Valeria Fink1, Stefan Wampl1, Lorenz Kiss1, and Albrecht Ingo Schmid1
1Center for Medical Physics and Biomedical Engineering, High Field MR Center, Medical University of Vienna, Vienna, Austria

Synopsis

Keywords: Myocardium, Spectroscopy, MRS, Phosphorus

Motivation: In cardiac 31P-CSI MRS, the signals containing 31P are important markers of myocardial health. Optimizing metabolite quantification is essential for understanding myocardial bioenergetics and related pathologies.

Goal(s): The aim was to quantify an extended range of low-signal metabolites including PME, Pi, GPE, GPC, PEP, and NAD.

Approach: 36 of the existing 31P-CSI MRS volunteer data at 7T were analysed using an optimized and extended prior knowledge.

Results: The results, presented as chemical shift, line-width, amplitude, and their CRBs, demonstrated successful quantification of low-signal metabolites alongside with 2,3-DPG, PCr, and ATP in the same spectra, demonstrating the feasibility of quantifying these signals at 7T.

Impact: The 7T cardiac 31P-CSI MRS data provide valuable insight and demonstrates that low-signal metabolite quantification is possible. Looking at these signals in patient data may add to cardiac 31P MRS and could provide new markers of myocardial health.

Introduction

31P-MRS is a unique tool to analyse compounds that play a significant part in myocardial health 1. It provides insights into high-energy metabolites such as adenosine triphosphate (ATP) and phosphocreatine (PCr), which are important markers of myocardial disease 2. In most cases, only the most prominent resonances of ATP, PCr and DPG (blood) are reported, there are, however, additional peaks in the spectra (Figure 1). Here, these low-signal metabolites are quantified, which could provide additional valuable data to study myocardial health 3.

Methods

The study was approved by the local Ethics Board (EK Nr. 1670/2017) and conducted according to the Declaration of Helsinki in its latest version.
Data were obtained from a Magnetom 7 T MR scanner (Siemens Healthcare Diagnostics GmbH, Erlangen, Germany) with a dual-tuned cardiac surface coil (RAPID Biomedical GmbH, Rimpar, Germany), equipped with a 31P (𝑑 = 14 cm) and a 14×22 cm2 1H loop. 36 participants (18 f, 18 m) were scanned in supine position (age: 28.7 ± 4.9 years, BMI: 22.6 ± 2.1 kg/m2). CINE images were acquired as localizers for 3D CSI MR spectroscopy, as described in more detail elsewhere 4.
The previously used prior knowledge was expanded after revising the chemical shift and line-width information of the metabolites PCr, α-ATP, γ-ATP, 2,3-diphosphoglycerate (2,3-DPG) and inorganic phosphate (Pi). Additionally, the prior knowledge was extended to encompass metabolites phosphomonoesters (PME), glycerophosphocholine (GPC), glycerophosphoethanolamine (GPE), and nicotinamide adenine dinucleotide (NAD), with corresponding chemical shift and line-width data based on available literature and an average spectrum of about 50 scans, see Figure 1. The metabolite with an average chemical shift of about 2.0 ppm in the spectra was assumed to be phosphoenolpyruvate (PEP), supported by earlier studies 5.
Phases of neighbouring lines were constrained to be equal, as well as line-widths of singlets and multiplets, respectively. Fitting was performed in MATLAB using the open-source OXSA toolbox 6 and the AMARES fitting algorithm 7. Voxels in the interventricular septum were selected for fitting (Figure 2). Metabolite amplitudes were correct for partial saturation and blood contributions 4. The fitted voxel data, including calculated peak ratios, were assessed for significance, supported by Cramér-Rao bounds (CRB). Mean values and standard deviations were calculated for the respective chemical shifts. Line-widths and amplitudes were assessed to confirm data plausibility. Boxplot diagrams were created for the metabolite to γ-ATP ratios (Figure 3).

Results

For the re-evaluated data the average SNR of PCr is 27.7 ± 10.6, revealing high spectral quality. Out of a total of 178 spectra evaluated, 177 met the predefined SNR threshold (> 10), ensuring dataquality control. When considering the line-width, it can be observed that all metabolites fall within the defined boundaries in all datasets. The metabolite to γ-ATP ratio and chemical shifts are summarized in Table 1 and Figure 3. Additionally, the number of datasets excluded when assuming a PCr SNR > 10 and setting a relative CRB value of < 100 % was examined. The dataset already cut out based on SNR is excluded here, too. For PCr, α-ATP, γ-ATP and PEP, all datasets meet the criteria, 175 are included for DPG, 174 for Pi, 165 for GPC, 136 for GPE, 164 for NAD, and only 113 for PME.

Discussion & Conclusion

We showed that it is possible to fit low-signal metabolites (Pi, PME, GPE, GPC, PEP, NAD) within the same spectra alongside dominant metabolites (PCr, α-ATP, γ-ATP, and 2,3-DPG), with PME clearly being the smallest and most unreliable (6 different spectra in Figure 4 as examples). For high-signal metabolites, the results of the amplitudes and their associated CRB demonstrate robust and easily detectable signals, as long known 1-3. However, for low-signal metabolites, the difficulty in their detection is evident, yet in many cases, these signals are visible in 7 T spectra and are quantifiable. These findings highlight the challenges in quantifying low-signal metabolites, likely due to their inherently weaker signals, as evidenced by higher CRBs. Nonetheless, several of these metabolites are known to be related to e.g. tumour presence (phosphodiesters), which could be also elevated in inflammatory states 1. Others are important participants in energy metabolism (Pi,NAD) 1 and should therefore be included in data analysis, especially in patients. In summary, this study provides insights into the analysis of cardiac 31P-MRS, highlighting low-signal metabolite quantification challenges. The results emphasize the importance of precise dataquality control. While high-signal metabolites showed robust amplitudes and acceptable CRB values, further research and method validation is needed to improve the quantification of low-signal metabolites.

Acknowledgements

This study was supported by the Austrian Science Fund (FWF), P 35607.

References

  1. R. A. de Graaf, In Vivo NMR Spectroscopy: Principles and Techniques, 3rd ed. Newark: JohnWiley & Sons Incorporated, 2019. [Online]. Available: https://livivo.idm.oclc.org/login?url=https://ebookcentral.proquest.com/lib/zbmed-ebooks/detail.action?docID=5613474
  2. M. ten Hove and S. Neubauer, “MR spectroscopy in heart failure--clinical and experimentalfindings,” Heart failure reviews, vol. 12, no. 1, pp. 48–57, 2007, doi: 10.1007/s10741-007-9003-8 .
  3. C. T. Rodgers, W. T. Clarke, C. Snyder, J. T. Vaughan, S. Neubauer, and M. D. Robson,“Human cardiac 31P magnetic resonance spectroscopy at 7 Tesla,” Magnetic resonance inmedicine, vol. 72, no. 2, pp. 304–315, 2014, doi: 10.1002/mrm.24922 .
  4. S. Wampl et al., “Investigating the effect of trigger delay on cardiac 31P MRS signals,”Scientific reports, vol. 11, no. 1, p. 9268, 2021, doi: 10.1038/s41598-021-87063-8 .
  5. P. Sedivy, T. Dusilova, M. Hajek, M. Burian, M. Krššák, and M. Dezortova, “In Vitro 31P MRChemical Shifts of In Vivo-Detectable Metabolites at 3T as a Basis Set for a Pilot Evaluationof Skeletal Muscle and Liver 31P Spectra with LCModel Software,” Molecules (Basel,Switzerland), vol. 26, no. 24, 2021, doi: 10.3390/molecules26247571 .
  6. L. A. B. Purvis, W. T. Clarke, L. Biasiolli, L. Valkovič, M. D. Robson, and C. T. Rodgers,“OXSA: An open-source magnetic resonance spectroscopy analysis toolbox in MATLAB,”PloS one, vol. 12, no. 9, e0185356, 2017, doi: 10.1371/journal.pone.0185356 .
  7. L. Vanhamme, A. van den Boogaart, and S. van Huffel, “Improved method for accurate andefficient quantification of MRS data with use of prior knowledge,” Journal of magneticresonance (San Diego, Calif. : 1997), vol. 129, no. 1, pp. 35–43, 1997, doi:10.1006/jmre.1997.1244 .

Figures

Figure 1: 31P-MRS spectrum of the myocardium. This myocardial spectrum was produced by averaging over around 50 data sets. It served as motivation for a more in-depth exploration of low-signal metabolite quantification.

Figure 2: Localiser CINE image in cardiac short axis orientation and the CSI grid as overlay. Allvoxels were selected within the interventricular septum, as demonstrated here.

Figure 3: The boxplots display the metabolite to γ-ATP ratios, they are corrected for partial T1 saturation and blood pool contributions. The solid line in the box represents the median and the dashed line in the box shows the mean value of the data.

Figure 4: Cardiac 31P MR spectra from six different subjects are shown. These spectra serve as exemplars, showcasing the capability of quantifying low-signal metabolites alongside dominant metabolites within the same spectrum.


Table 1: Ratios and chemical shifts of the evaluated metabolites.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)
1785
DOI: https://doi.org/10.58530/2024/1785