Magnetic Resonance Imaging 21 (2003) 561–566
The effect of orientation on quantification of muscle creatine by 1H MR spectroscopy Fabao Gaoa, Paul A. Bottomleyb, Cheryl Arnoldb, Robert G. Weissa,* b
a Cardiology Division, Department of Medicine, The Johns Hopkins University School of Medicine, Baltimore, MD, USA MR Research Division, Department of Radiology, The Johns Hopkins University School of Medicine, Baltimore, MD, USA
Abstract Creatine is a central energy metabolite whose N-CH3 group can be detected with 1H MR spectroscopy (1H MRS) with relatively high sensitivity. Prior studies suggest that muscle fiber orientation can influence the appearance of other resonances attributed to total creatine (CR). Our purpose was to determine whether muscle fiber orientation affects muscle CR concentration quantification by 1H MRS with the commonly used N-CH3 resonance at 3.0 ppm. Skeletal muscle CR was quantified with water-referenced 1H MRS in normal subjects with different forearm muscle orientations relative to the static magnetic field at 1.5T. There were no significant differences in mean total [CR] in two different series of experiments separately including two orthogonal orientations and four orientations (0o, 30o, 60o, 90o) of the forearm relative to the static field using either short (TE ⫽ 15 ms) or long (TE ⫽ 100 ms) echo times for voxels containing or centered on the same tissues. Subtle differences in CR line-width and T2 correction factors were observed with orientation. These observations are consistent with the primary hypothesis that careful water-referenced [CR] quantification, accounting for T2 effects and using the N-CH3 peak at 3.0ppm, is not affected by muscle orientation. © 2003 Elsevier Inc. All rights reserved. Keywords: Total creatine; 1H MRS; Skeletal muscle; Concentration; Orientation; Quantification
1. Introduction Creatine (Cr) plays a central role in muscle, brain, and myocardial energy metabolism. By means of the creatine kinase (CK) reaction, high-energy phosphates are reversibly transferred from ATP to Cr to form creatine phosphate (PCr) and ADP. While [ATP] and [PCr] can be measured in humans non-invasively by 31P magnetic resonance spectroscopy (MRS), total creatine (PCr ⫹ Cr ⫽ CR) can be detected with 1H MRS. The N-methyl protons of CR at 3.0 ppm in the 1H MR spectrum can be detected with high sensitivity and have served as a validated means for reliably quantifying CR concentration in brain [1] as well as skeletal [2-4] and cardiac [5] muscle with water-referenced 1H MRS. However, recent studies suggest that several resonances in the 2.7-3.6 ppm region of the 1H MR spectrum as well as lipids may be affected by the orientation of muscle fibers, relative to the static magnetic field [6-8]. Some of these
resonances have recently been attributed to creatine [9], which raises concerns about the reliable quantification of muscle creatine with 1H MRS. However to date no quantitative studies of the effects of muscle orientation on CR concentration determined from the commonly used 3.0 ppm N-methyl resonance with 1H MRS have been reported. Because CR is a key energetic compound that can be used to identify myopathies and irreversible myocardial injury [5], it is critical to assess whether orientation effects are indeed present that could spuriously affect muscle CR quantification by 1H MRS. The aim of these studies was to evaluate the effect of muscle orientation on CR quantification using image-guided, water-referenced 1H MRS of the same muscle regions in the human forearm.
2. Materials and methods 2.1. Subjects
* Corresponding author. Tel.: ⫹1-410-955-1703; fax: ⫹1-410-9555996 E-mail address:
[email protected] (R. Weiss). 0730-725X/03/$ – see front matter © 2003 Elsevier Inc. All rights reserved. doi:10.1016/S0730-725X(03)00073-0
Studies were approved by an institutional review committee for human investigation and were conducted on a 1.5T Signa whole body system (GE, Milwaukee). The fore-
562
F. Gao et al. / Magnetic Resonance Imaging 21 (2003) 561–566
arms of 15 normal subjects with no history of peripheral vascular, myopathic, or systemic disease were studied. Two series of orientation studies were separately performed, with voxels fixed relative to the muscle or with voxels fixed relative to the center of a muscle region. In the first series, the same voxel physically containing the same tissue in a given muscle region was studied in two orthogonal orientations of the long axis of the forearm relative to the static magnetic field. In the second series, a voxel of the same size and orientation relative to the magnet was centered over the same point in the forearm and studied with four orientations (0°, 30°, 60°and 90°) of the forearm relative to the static field. In each instance, the forearm was re-oriented after data acquisition, imaging repeated, and the arm re-positioned as necessary to match the same position identified by fiducial markers before spectroscopy was repeated. To avoid potential systematic effects, the sequence of the orientations studied was randomly varied between subjects in both two- and four-orientation studies. 2.2. Data acquisition The volunteers were positioned in the prone position with their right forearm centered over a five-inch imaging/ spectroscopy surface coil with fiducial markers attached in at least two positions on the arm. Multi-slice, spin-echo images were acquired with the following acquisition parameters: echo train length 8, echo time (TE) 17 ms, repetition time (TR) 2000 ms, field of view (FOV) 16 cm, slice thickness 10 mm, inter-slice spacing 2 mm, 256 ⫻ 192 matrix and auto shimming. The homogeneity of the magnetic field was then optimized by volume selective shimming over a selected voxel using the water signal. Spatially localized 1H magnetic resonance (MR) spectra were then acquired from muscle voxels of mean size 1.4 ⫻ 1.4 ⫻ 1.0 cm with a stimulated echo acquisition mode (STEAM CSI) sequence applied with and without water suppression for absolute quantification of CR concentration. Spectra were acquired with TR of 1600 ms and acquired with short (TE ⫽ 15 ms) and long (TE ⫽ 100 ms) echo times to account for T2 effects. To obtain satisfactory signal-to-noise ratios, 128 signals were summed leading to a total acquisition time of 3.3 min per spectrum for suppress acquisitions and 32 signals summed for non-suppressed acquisitions. The water resonance in unsuppressed spectra was used as a chemical shift reference with an assumed value of 4.68 ppm. Total creatine was determined from the 3.0 ⫾ 0.1 ppm peak in the water-suppressed spectra. Data processing was performed by automated fitting in the frequency-domain with custom software running on a Sun workstation [10]. 2.3. Data processing Peak areas of creatine and water were determined from the areas of Gaussian line-shapes that were automatically fitted to the peaks using a simplex routine. In a subset of
subjects, the data were analgyed using both Gaussian and Lorentzian fitting. Both approaches generated similar CR concentrations although the Gaussian approach provided better fits and was therefore used for data analysis. Total muscle creatine concentration was quantified from the relaxation-corrected ratio of the fitted areas of the CR peak in the suppressed spectrum to the water peak of the corresponding unsuppressed spectrum, using the previously validated method [2], as given by the equation: 关CR兴 ⫽
再
冎
2 ⫻ S CR ⫻ 关W兴 F CR ⫻ E CR ⫻ 3 ⫻ SW FW ⫻ EW
(1)
Here, SCR is the 3.0ppm resonance area for total creatine, SW is the water MR signal area; FCR and FW are the T1 saturation correction factors; ECR and EWsimilarly account for the signal decay occurring during period TE and are generally taken to be E ⫽ exp (TE/T2e), with T2e the effective T2; [W] is the water concentration in the tissue; and the factors of 2 and 3 account for the three protons on the N-methyl group of CR and the two protons of water. The concentration of water was assumed to equal 42.4 mol/kg wet weight, based on a mean adult muscle tissue water content of 77 ⫾ 3% [2,11]. To determine whether orientation affected the observed line-widths and/or T2 corrections for the 3.0 ppm CR peak, the unfiltered line-widths, fitted by the automated fitting routine, and the T2 correction factors were also documented at the different orientations and echo times. 2.4. Statistical Test Statistical comparisons of CR measured at different forearm orientations were performed with paired t tests for the two-orientation series and with an analysis of variance for the four-orientation series. The results are presented as mean ⫾ SD. P values ⬍0.05 were considered significant.
3. Results The first series of studies investigated the same muscle tissue in two orthogonal orientations of the forearm and was conducted on seven normal subjects (6 men, one woman; mean age 38 ⫾ 3 years). Representative images and 1H MR spectra are shown in Fig. 1. There was no significant difference in mean muscle creatine concentration between the 0o and 90o orientations for either the short (36.4 ⫾ 6.2 and 36.8 ⫾ 3.6 mmol/kg wet wt, p ⫽ 0.8) or long TE acquisitions (36.4 ⫾ 7.8 and 37.1 ⫾ 4.7 mmol/kg wet wt, p ⫽ 0.8, Fig. 2). In addition, there were no significant differences in mean line-width or T2 corrections factors (ECR/EW) for the 3.0ppm CR resonance at the two orientations for either the short or long TE acquisitions (Fig. 3-4). The second series of investigations studied four forearm orientations in eight normal subjects (5 men, 3 women;
F. Gao et al. / Magnetic Resonance Imaging 21 (2003) 561–566
563
Fig. 1. Representative 1H MR images (right panel): upper is image from the zero degree orientation, and the lower is image from the 90 degree orientation. Localized spectra (left panel) is spectra (TE ⫽ 100 ms) from two orthogonal forearm orientations relative to the static magnetic field. The calculated [CR] were 34.6 mmol/kg wet wt and 30.3 mmol/kg wet wt for 0o and 90oCR. ⫽ Total creatine, Cho. ⫽ Choline.
mean age 39 ⫾ 2 years). Differences in CR peak height were occasionally noted between some orientations, but such differences were not consistently observed. There was
no significant difference in mean [CR] between the 0o, 30o, 60o, and 90o orientations with TE 15 ms (40.7 ⫾ 5.3, 37.9 ⫾ 1.8, 41.7 ⫾ 4.1 and 36.2 ⫾ 5.4 mmol/kg wet wt, p ⬎
Fig. 2. Effect of muscle orientation on [CR]: two-orientation study showed no significant differences in at creatine concentration for either the short or long echo time (TE ⫽ 15 ms, upper panel “A.” and 100 ms lower panel “B.,” p ⬎ 0.05).
Fig. 3. The effect of orientation of CR line-width: no significant differences in CR linewidth were observed for the short and long echo time (TE ⫽ 15 ms, upper panel “A.” and 100 ms lower panel “B.,” p ⬎ 0.05).
564
F. Gao et al. / Magnetic Resonance Imaging 21 (2003) 561–566
Fig. 4. The effect of orientation on ECR/EwECR/Ew were not significantly different for both short TE (first voxel, p ⫽ 0.37, second voxel p ⫽ 0.84, upper panel “A.”) and long TE (first voxel, p ⫽ 0.86, second voxel p ⫽ 0.53, lower panel “B.”) between the two orthogonal orientations.
0.05), or with TE 100 ms (40.7 ⫾ 5.3, 37.9 ⫾ 1.8, 41.7 ⫾ 4.1 and 36.2 ⫾ 5.5 mmol/kg wet wt, p ⬎ 0.05, Fig. 5). However, differences in mean line-width of the 3.0 ppm CR resonance were observed in the four orientation series that were significant at the short TE (TE ⫽ 15 ms) but not significant at the long (TE ⫽ 100 ms) acquisitions (Fig. 6). In addition, ECR/EW varied among the orientations and the differences were more striking for the long TE (4.1 ⫾ 2.2,
Fig. 6. The effect of orientation of CR line-width: the differences were significant for the short echo time (TE ⫽ 15 ms, upper panel “A.”) acquisitions and were p ⬍ 0.05 for 0o vs 90o, the differences were not significant (p ⬎ 0.05) for the long echo time (TE ⫽ 100 ms, lower “B.”) using Student-Newman-Keuls test.
1.6 ⫾ 0.5, 3.7 ⫾ 1.4, 3.2 ⫾ 1.2) than for the short TE acquisitions (0.93 ⫾ 0.08, 0.81 ⫾ 0.04, 0.92 ⫾ 0.05, 0.90 ⫾ 0.06, for 0o, 30o, 60o, and 90o, respectively). Overall differences in ECR/EW were significant (p ⬍ 0.05) for both short and long TE, with the 30o ECR/EW significantly lower than for the other orientations (Fig. 7).
4. Discussion
Fig. 5. Effect of orientation on muscle [CR]: the four-orientation study showed no significant differences in muscle CR with orientation for the short and long echo times (TE ⫽ 15ms, upper panel “A.”, F ⫽ 0.57, P⬎0.05; and TE ⫽ 100ms, lower panel “B.”, F ⫽ 0.54, P⬎0.05) using analysis of variance.
In this study the effects of forearm muscle orientation relative to the static magnetic field on the quantification of CR with water-referenced 1H MRS were studied at 1.5 T. Muscle CR was quantified noninvasively using corrections for T1 and T2 effects, as originally described [2]. Prior studies with water-referenced 1H MRS of muscle CR have generated comparable [CR] values, consistent with invasive biopsy data and the 3.0 ppm peak being 100% MR visible, as well as demonstrating little change in [CR] among different muscle groups [2,4,5]. No significant orientation effects on CR concentrations measured from the CR N-CH3 resonance at 3.0 ppm were observed in the two series of studies. These included image-guided exams of the same tissue with two orthogonal orientations as well as exams of identically sized voxels centered on the same tissue in four orientations. However, some differences in the line-width and T2 correction factors were noted with orientation as well as associated changes in peak height, possibly attributable to dipolar coupling [8,9] and/or to anisotropic diffu-
F. Gao et al. / Magnetic Resonance Imaging 21 (2003) 561–566
Fig. 7. The effect of orientation on ECR/EW. Overall there were significant differences in ECR/EW with orientation for both short TE (p ⬍ 0.01, upper panel “A. ”) and long TE (p ⬍ 0.05, lower panel “B.”) for 30o vs 0o, 30o vs 60o and 30o vs 90o.
sion in the myocytes. The latter is suggested by diffusion measures in muscle showing CR diffusion is rapid although slower than water [12]. Taken together these data demonstrate that [CR] quantification in muscle with water-referenced 1H MRS is not altered by forearm muscle orientation although some care with T2 correction is warranted. 1 H MRS has also been used to study lipid metabolism. Recent studies demonstrate orientation effects on muscle lipid resonances and these have been attributed to dipoledipole coupling in intramyocellular and extramyocellular lipids [13-15]. Subsequent studies have reported orientation effects as well for several small resonances in the 1H MR spectrum between 2.6 and 3.9 ppm, initially labeled X1-X4 [6,7]. Several of these resonances (X1 and X2) exhibited orientation effects interpreted as dipole-dipole coupling interactions [6], and were later attributed to CR based on creatine loading studies [9]. However, the assignments were not unequivocal since creatine loading also resulted in changes in lipid signals and since X peak assignments included corrections for significant water accumulation that were not detected by MRI [9]. In addition, CR loading over time could result in cellular changes that include accumulation of proteins or other by-products, in addition to the spectral changes in myocellular lipids that were reported [9], and there may be reflected in the X peaks. On the other hand, the 3.0 ppm resonance has been clearly linked to CR by many investigators [1,4,5,16] and has not been quantitatively studied before for the effects of orientation on CR concentration measurements. Qualitative differences in the height of the 3.0 ppm CR resonance with muscle orientation
565
are apparent in some but not all previously published spectra [6,7]. In studies of two orthogonal orientations of the same voxel in this work, no significant differences in [CR], linewidths, or ECR/EW were observed. In studies of four orientations of identically centered voxels, no significant differences in mean [CR] determinations were detected, although differences in linewidth and ECR/EW correction were apparent. While in this study such variations did not systematically affect mean [CR] measurements, they may nevertheless represent a source of scatter in the data. Similarly, individual variations in muscle fiber orientations relative to the forearm may be a confounding factor, although the muscle fibers studied all run substantially parallel to the forearm axis in the region studied. Note that the two- and four-orientation studies differ subtly in that the former compared substantially identical portions of muscle tissue, and was therefore more timeconsuming to accomplish while the latter compared voxels centered on the same region whose content may vary with muscle orientation. In any event, all of the data from both two- and four-orientation studies demonstrate that waterreferenced 1H MRS measurements of muscle [CR] using appropriate T2 corrections, are not dependent on muscle orientation relative to the magnetic field.
Acknowledgments Support by NIH Grants 1R01-HL61912 and 1R01HL56882 is gratefully acknowledged. .
References [1] Kreis R, Ernst R, Ross BD. Absolute quantitation of water and metabolites in the human brain. II. Metabolite concentrations. J Mag Res 1993;B102:9 –19. [2] Bottomley PA, Lee Y, Weiss RG. Total creatine in muscle: imaging and quantification with proton MR spectroscopy. Radiology 1997; 204:403–10. [3] Kreis R, Kamber M, Koster M, et al. Creatine supplementation—part. II: in vivo magnetic resonacne spectroscopy. Med Sci Sports Exerc 1999;31:1770 –7. [4] Hwang J-H, Pan JW, Heydari S, Hetherington HP, Stein DT. Regional differences in intramyocellular lipids in humans observed by in vivo 1H-MR spectroscopic imaging. J Appl Physiol 2001;90:1267– 74. [5] Bottomley PA, Weiss RG. Creatine depletion in non-viable, infarcted myocardium measured by noninvasive MRS. Lancet 1998;351: 714 – 8. [6] Kreis R, Boesch C. Liquid-crystal-like structures of human muscle demonstrated by in vivo observation of direct dipolar coupling in localized proton magnetic resonance spectroscopy. J Mag Res 1994; B104:189 –92. [7] Kreis R, Boesch C. Spatialy localized, one- and two dimensional NMR spectroscopy and in vivo application to human muscleJ Mag Res 1996;B113:103–18.
566
F. Gao et al. / Magnetic Resonance Imaging 21 (2003) 561–566
[8] Ntziachristos V, Kreis R, Boesch C, Quistorff B. Dipolar resonance frequency shifts in 1H MR spectra of skeletal muscle: confirmation in rats at 4.7 T in vivo and observation of changes postmortem. Magnetic Resonance in Medicine 1997;38:33–9. [9] Kreis R, Koster M, Kamber M, Hoppeler H, Boesch C. Peak assignment in localized 1H MR spectra of human muscle based on oral creatine supplementation. Magnetic Resonance in Medicine 1997;37:159 – 63. [10] Barker PB, Sibisi S. Non-linear least squares analysis of in vivo 31P NMR data. Soc Magn Res Med (Book of Abstracts) 1990;9:1089. [11] Forsberg AM, Nilsson E, Werneman J, Bergstrom J, Hultman E. Muscle composition in relation to age and sex. Clin Sci 1991;81:249 –56. [12] Liess C, Radda GK, Clarke K. Metabolite and water apparent diffusion coefficients in the isolated rat herat: effects of ischemia. Magn Reson Med 2000;44:208 –14.
[13] Schick F, Eismann B, Jung W-I, Bongers H, Bunse M, Lutz O. Comparison of localized proton NMR signals of skeletal muscle and fat tissue in vivo: two lipid compartments in muscle tissue. Magnetic Resonance in Medicine 1993;29:158 – 67. [14] Boesch C, Kreis R. Observation of intramyocellular lipids by 1Hmagnetic resonance spectroscopy. Ann N Y Acad Sci 2000;904:25– 31. [15] Krssak M, Petersen KF, Dresner A, DiPietro L, Vogel SM, Rothman DL, Shulman GI, Roden M. Intramyocellular lipid concentrations are correlated with insulin sensitivity in humans: a 1H NMR spectroscopy study. Diabetologia 1999;42:113– 6. [16] Yoshizaki K, Seo Y, Nishikawa H. High-resolution proton magnetic resonance spectra of muscle. Biochim Biophys Acta 1981;678:283– 91.