Comparison of single-voxel 1H-cardiovascular magnetic resonance spectroscopy techniques for in vivo measurement of myocardial creatine and triglycerides at 3T

Abstract

Background: Single-voxel proton cardiovascular magnetic resonance spectroscopy (¹H-CMRS) benefits from 3 T to detect metabolic abnormalities with the quantification of intramyocardial fatty acids (FA) and creatine (Cr). Conventional point resolved spectroscopy (PRESS) sequence remains the preferred choice for CMRS, despite its chemical shift displacement error (CSDE) at high field (≥ 3 T). Alternative candidate sequences are the semi-adiabatic Localization by Adiabatic SElective Refocusing (sLASER) recommended for brain and musculoskeletal applications and the localized stimulated echo acquisition mode (STEAM). In this study, we aim to compare these three single-voxel ¹H-CMRS techniques: PRESS, sLASER and STEAM for reproducible quantification of myocardial FA and Cr at 3 T. Sequences are compared both using breath-hold (BH) and free-breathing (FB) acquisitions.

Methods: CMRS accuracy and theoretical CSDE were verified on a purposely-designed fat-water phantom. FA and Cr CMRS data quality and reliability were evaluated in the interventricular septum of 10 healthy subjects, comparing repeated BH and free-breathing with retrospective gating.

Results: Measured FA/W ratio deviated from expected phantom ratio due to CSDE with all sequences. sLASER supplied the lowest bias (10%, vs -28% and 27% for PRESS and STEAM). In vivo, PRESS provided the highest signal-to-noise ratio (SNR) in FB scans (27.5 for Cr and 103.2 for FA). Nevertheless, a linear regression analysis between the two BH showed a better correlation between myocardial Cr content measured with sLASER compared to PRESS (r = 0.46; p = 0.03 vs. r = 0.35; p = 0.07) and similar slopes of regression lines for FA measurements (r = 0.94; p < 0.001 vs. r = 0.87; p < 0.001). STEAM was unable to perform Cr measurement and was the method with the lowest correlation (r = 0.59; p = 0.07) for FA. No difference was found between measurements done either during BH or FB for Cr, FA and triglycerides using PRESS, sLASER and STEAM.

Conclusion: When quantifying myocardial lipids and creatine with CMR proton spectroscopy at 3 T, PRESS provided higher SNR, while sLASER was more reproducible both with single BH and FB scans.

Keywords: 3 T; CMR spectroscopy; Cardiac metabolism; Creatine; Lipids; Proton magnetic resonance spectroscopy; SLASER.

Fig. 1

a Spectra from the homemade phantom containing fat acquired with three different sequences (PRESS, sLASER and STEAM) showing the resonances of water (W, 4.7 ppm) and fatty acid (FA, 1.2 ppm). b Acquisitions were realized without water suppression using a voxel of 20 × 20 × 30mm³ (dashed line) that surrounded the fat-containing cylindrical tube. Excitation was prescribed through-plane, and both refocusing pulses were prescribed in-plane, according to the dotted box

Fig. 2

a Examples of spectra obtained on the same healthy subject with the same volume of interest placed on the interventricular septum (c). Black curves represent the spectra assessed during one breath hold (BH). Red curves represent the spectra assessed during free breathing (FB). BH spectra tended to exhibit higher noise levels than FB. b The fitting processing was designed to quantify trimethyl amide (TMA, 3.2 ppm), creatine (CR, 3.1 ppm) and the total myocardial triglyceride (TG) resonance, which includes fatty acids (FA, 0.9, 1.3, and 1.6 ppm) and unsaturated fatty acids (UFA, 2.1 and 2.3 and 2.8 ppm). Residual signal enables to evaluate the adequacy of the model

Fig. 3

a Average of water linewidths measured from 10 subjects from non-water suppressed spectrum assessed during BH (white) and FB (red) for PRESS, sLASER and STEAM. b Average of FA linewidths measured from 10 subjects from water suppressed spectrum assessed during BH (white) and FB (red) for PRESS, sLASER and STEAM. Data expressed as mean ± SD. PRESS: BH, n = 20 & FB n = 10; sLASER: BH, n = 20 & FB n = 10; STEAM: BH, n = 9 for Cr, n = 14 for FA & FB n = 8 for both Cr and FA

Fig. 4

Graphs of mean Cr (blue) and FA (orange) SNR for PRESS, sLASER and STEAM. Data expressed as mean ± SD. PRESS: BH, n = 20 & FB n = 10; sLASER: BH, n = 20 & FB n = 10; STEAM: BH, n = 9 for Cr, n = 14 for FA & FB n = 8 for both Cr and FA. *: p < 0.05; **: p < 0.01 and ***: p < 0.001 compared with the same metabolite (Cr or FA) measured in BH with PRESS. #: p < 0.05 and ##: p < 0.01 compared with the same SNR metabolite (Cr or FA) measured during FB with PRESS. No difference was found between sLASER and STEAM

Fig. 5

Bland–Altman plots showing reproducibility between two repeated ¹H-CMRS measures within a single CMR session in the 10 subjects regarding myocardial creatine (a) and fatty acids (b). Metabolite contents were acquired using two single breath-hold methods with PRESS (n = 10), sLASER (n = 10) and STEAM (n = 6). The solid line represents the mean of differences between levels obtained with the repeated BH methods, the dashed lines indicate the confidence intervals ± 1.96 SD. Peak at 1.2 ppm was used to calculate FA levels. BH: breath hold; Cr: creatine; FA: fatty acids; SD: standard deviation; W: water

Fig. 6

Linear regression analysis showing the correlation between myocardial creatine (a) and fatty acids (b) content relative to water in two acquisitions in breath-hold with PRESS (n = 10), sLASER (n = 10) and STEAM (n = 6). BH: breath-hold; Cr: creatine; FA: fatty acids; FB: free-breathing; W: water

Fig. 7

Linear regression analysis showing the correlation between myocardial creatine (a) and fatty acid (b) content relative to water in acquisitions realized in breath-hold and free-breathing with PRESS and sLASER. BH: breath-hold; Cr: creatine; FA: fatty acids; FB: free-breathing; W: water

Fig. 8

Ratio of signals from fatty acids to water given by PRESS and STEAM compared to that given by sLASER. The dashed red line represents the linear regression analysis between ratio measured with PRESS and sLASER or STEAM and sLASER. The dashed grey lines indicate unity