Absolute quantification of carnosine

IOP PUBLISHING

PHYSICS IN MEDICINE AND BIOLOGY

Phys. Med. Biol. 52 (2007) 6781–6794

doi:10.1088/0031-9155/52/23/001

Absolute quantification of carnosine in human calf muscle by proton magnetic resonance spectroscopy 1 ¨ Mahir S Ozdemir , Harmen Reyngoudt2, Yves De Deene3, 4 Hakan S Sazak , Els Fieremans1, Steven Delputte1, Yves D’Asseler1, Wim Derave5, Ignace Lemahieu1 and Eric Achten2 1

Department of Electronics and Information Systems, MEDISIP, Ghent University-IBBT-IBiTech, De Pintelaan 185 block B, B-9000 Ghent, Belgium 2 Department of Radiology, Ghent University Hospital, De Pintelaan 185, Ghent, Belgium 3 Department of Radiotherapy, Ghent University Hospital, De Pintelaan 185, Ghent, Belgium 4 Department of Statistics, Ege University, 35100 Bornova, Izmir, Turkey 5 Department of Movement and Sports Science, Ghent University, Watersportlaan 2, Ghent, Belgium E-mail: [email protected]

Received 30 May 2007, in final form 9 October 2007 Published 6 November 2007 Online at stacks.iop.org/PMB/52/6781 Abstract Carnosine has been shown to be present in the skeletal muscle and in the brain of a variety of animals and humans. Despite the various physiological functions assigned to this metabolite, its exact role remains unclear. It has been suggested that carnosine plays a role in buffering in the intracellular physiological pHi range in skeletal muscle as a result of accepting hydrogen ions released in the development of fatigue during intensive exercise. It is thus postulated that the concentration of carnosine is an indicator for the extent of the buffering capacity. However, the determination of the concentration of this metabolite has only been performed by means of muscle biopsy, which is an invasive procedure. In this paper, we utilized proton magnetic resonance spectroscopy (1 H MRS) in order to perform absolute quantification of carnosine in vivo non-invasively. The method was verified by phantom experiments and in vivo measurements in the calf muscles of athletes and untrained volunteers. The measured mean concentrations in the soleus and the gastrocnemius muscles were found to be 2.81 ± 0.57/4.8 ± 1.59 mM (mean ± SD) for athletes and 2.58 ± 0.65/3.3 ± 0.32 mM for untrained volunteers, respectively. These values are in agreement with previously reported biopsy-based results. Our results suggest that 1 H MRS can provide an alternative method for noninvasively determining carnosine concentration in human calf muscle in vivo. (Some figures in this article are in colour only in the electronic version)

0031-9155/07/236781+14$30.00 © 2007 IOP Publishing Ltd Printed in the UK

6781

6782

¨ M S Ozdemir et al

1. Introduction Gulewitch and Amiradzibi were the first to isolate carnosine from a Liebig’s meat extract (Gulewitch and Amiradzibi 1900). Since then, carnosine has been shown to be present in varying proportions in both vertebrate and non-vertebrate skeletal muscle as well as in the brain (Scriver et al 1983, Brown 1981). Despite the various functions assigned to carnosine in muscle, its precise physiological role remains obscure. Of the various hypotheses postulated, only that of intracellular physiological pHi buffering by neutralizing lactic acid produced in skeletal muscle during exercise is undisputed (Bate-Smith 1938). Although the relative contribution of carnosine in the buffering capacity in skeletal muscle is usually as low as 7% (Mannion et al 1992), muscle carnosine is probably the only buffer of which the content can be easily manipulated by diet and training. The muscle carnosine content is approximately three-fold higher in body builders than in vegetarian non-athletes. In the former population, the elevated muscle carnosine content represents 20% of the total muscle buffering capacity (Tallon et al 2005). Furthermore, several weeks of β-alanine supplementation can elevate the muscle carnosine content by 50–80%, which raises the relative contribution of carnosine in buffering to 15% and has a substantial impact on the development of muscle fatigue (Hill et al 2007, Derave et al 2007). This indicates that carnosine represents only a small fraction of the total buffer capacity, but the main fraction of the variable buffer capacity, which is more important in muscle physiology. Lactic acid produced within muscle gives rise to hydrogen ion release, thereby decreasing pHi , which is probably linked to muscle fatigue. Thus, a buffering function of carnosine has been suggested to stabilize pHi and has been assumed to increase the capacity for intensive exercise. Based on this, it has been proposed that the degree to which carnosine contributes to the total muscle buffer value would depend on the concentration of this metabolite present in the muscle. In addition, carnosine content in fast-twitching muscle fibers has been reported to be twice as high as that found in slow-twitching muscle fibers (Harris et al 1998). As was previously shown, a higher muscle buffer value is expected to lead to an increase in the muscle’s ability to maintain high intensity exercise performance (Suziki et al 2002, Hill et al 2007). It has also been demonstrated that carnosine in skeletal muscle can be increased by 80% and even more by 4–10 weeks of oral β-alanine supplementation (Harris et al 2006). This proposed benefit of carnosine uptake might be important as it can be used to augment high-intensity muscle performance required from athletes. Therefore, determining the carnosine concentration in muscles is of interest and an entirely satisfactory method for its estimation is lacking. To the best of our knowledge, no previous study has attempted to determine the concentration of carnosine non-invasively. Biopsy is the only technique currently utilized, which is invasive. Additionally, biopsy has some drawbacks: samples can only be acquired from superficial muscles, the detection of rapidly decaying components in the sample is inaccurate (Bottomley and Weiss 2001) and the results show large variation (Wendling et al 1996). Unlike biopsy, proton magnetic resonance spectroscopy (1 H MRS) provides the advantage of detecting the MR visible signals from a selected volume of interest localized practically anywhere in the body and shows less variability (Boesch et al 1997, Taylor et al 1992). 1 H MRS of human skeletal muscle has gained considerable interest recently as it provides information on a variety of metabolites (Schick et al 1993, Kreis and Boesch 1996, Kreis et al 1999, Boesch et al 1997, Bottomley et al 1997). Previously, 1 H MRS has been used to detect the carnosine signal in human muscles, thanks to imidazole protons giving rise to well-discerned resonances around 7 ppm (C4) and 8 ppm (C2), downfield of the water resonance (Pan et al 1988). In this study, we attempted to determine the concentration of the C2 resonance of carnosine in the human muscle by use of 1 H MRS on account of being less susceptible to orientation effects as compared to C4 (Kreis and Boesch 2000) and also

Absolute quantification of carnosine

6783

because it has a longer T2 (Pan et al 1988). The method is based on the phantom replacement technique by employing an external standard of known concentration scanned in an additional measurement (Buchli et al 1994). Correction for the relaxation effects, coil loading and temperature differences between in vivo and phantom measurements were taken into account. The method was verified by phantom experiments and in vivo measurements in the calf muscles of sprint-trained athletes and untrained volunteers. The effect of B1 inhomogeneity caused by standing waves (i.e. dielectric effects) on the quantification was also assessed using both phantom and in vivo measurements. 2. Material and methods 2.1. Magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS) measurements A 3 Tesla whole-body MR scanner (Siemens Trio, Erlangen), equipped with a knee coil, was used for imaging and recording of the spectra. All the spectra were acquired using a single voxel point-resolved spectroscopy (PRESS) localization with the following parameters: TR/TE = 2000/30 ms, voxel size = 12 × 30 × 40 mm3 , number of averages (NEX) = 256, 1024 data points with a spectral width of 1200 Hz. The total acquisition time was 8.4 min. A 1.5 l cylindrical bottle mimicking the calf filled with an aqueous solution of 50 mM carnosine was used as an external reference. 1% sodium chloride (NaCl) and 0.2% gadolinium DTPA were added to obtain physiological NaCl concentration and realistic relaxation times, respectively. The full-width at half-maximum (FWHM) of the water resonance was minimized to be in the range of 4–7 Hz for all phantom experiments. In each in vivo measurement, the right leg of each subject was positioned and was firmly immobilized in the knee coil such that the gastrocnemius muscle was in the center of the coil. The left leg was supported outside the coil to improve the comfort and thus minimize the movement. Voxels were placed using T1 weighted gradient-echo MR images in three orthogonal planes acquired with 3 mm slice thickness and TR/TE = 250/2.52 ms. Spectra were acquired from the soleus and gastrocnemius muscles in the right leg. The FWHM of the water signal was minimized to be in the range of 20–25 Hz in all in vivo measurements. Three professional male athletes and six untrained volunteers (four male, two female) of whom informed consent had been acquired were employed for the in vivo measurements. The study athletes were track-and-field athletes, specifically trained for 400 m running. All the athletes compete at regional or national level for several years and train 4 to 5 times weekly. 2.2. Signal linearity Quantitative MRS requires that the intensity of the recorded signal of a particular metabolite changes linearly with both the concentration of this metabolite and the volume of interest (VOI) from which the metabolite signal is obtained. Therefore, prior to performing quantification, we evaluated this by performing experiments in which signals were acquired from different phantoms of varying concentrations and for different VOI (i.e. cubic and non-cubic voxels) using the same PRESS sequence as described before. 2.3. Quantification The absolute quantification of the carnosine resonance was determined using the following equation:


6784

Ci =

Cr Si FiT1 FiT2 Sr FrT1 FrT2 pT pC

(mM),

(1)

where C is the concentration, S is the signal intensity, FT1 and FT2 are the T1 and T2 correction factors respectively and were calculated as follows: FT1 =

1 1 − exp(−TR /T1 )

FT2 = exp(TE /T2 ).

(2) (3)

The subscripts i and r denote the in vivo carnosine signal and reference carnosine signal to be measured, respectively. The concentration of the reference carnosine signal, Cr , is 50 mM. pT is the correction factor for the temperature difference between the phantom and the human body and pC is the correction factor for coil loading. The signal intensity, S, is defined as the peak area of carnosine resonances found by Lorentzian curve fitting performed in the frequency domain by the use of the iterative Levenberg–Marquardt algorithm. Postprocessing of the spectra included Fourier transformation of the FIDs followed by zero-order phase correction and baseline correction based on a fifth-order polynomial fitting to the spectral region lying outside the resonances of interest and subtracting it from the spectrum. Prior to Fourier transformation, no apodization or zero-filling was applied. 2.3.1. T1 and T2 effects. It was necessary to correct for the relaxation effects as the signals were not fully relaxed in the longitudinal direction and there was a signal loss in the transverse plane (TR = 2000 ms, TE = 30 ms). To this end, we measured the relaxation parameters, T1 and T2 , of the carnosine in vitro. The T1 relaxation time was measured by using a PRESS sequence with ten different repetition times (1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 6000 and 6500 ms) with TE of 30 ms. For the measurement of the T2 relaxation time, spectra were acquired at eight different TEs (30, 60, 90, 120, 150, 180, 240 and 350 ms) with TR of 4000 ms. T1 and T2 were calculated by fitting data points determined using the peak areas of metabolites to mono-exponential functions, Ms = M∞ (1 − exp(−TR /T1 )) and Ms = M0 exp(−TE/T2 ), respectively. Here, M0 is the magnetization in the longitudinal direction for TE = 0 ms. The correction of in vivo relaxation effects was performed by measuring the T2 in the soleus and gastrocnemius muscles of three volunteers as already described and using the mean value for the corresponding muscle. The correction of the T1 relaxation term was ignored as is explained in section 3.2. 2.3.2. The effect of temperature. The signal intensities were corrected for the effect of the temperature difference between the reference phantom used and the human body. This is on account of higher magnetization produced by the lower temperature (Tofts 2003). A correction factor, which is taken into account by pT as in equation (1), is therefore applied as the signal decreases by 6% between the room temperature (i.e. phantom temperature) and body temperature (Tofts 2003). 2.3.3. Coil loading. Differences in signal intensity due to coil loading were corrected by using the difference between the radio frequency (RF) power needed to obtain a 180◦ pulse in the phantom and that in vivo (Soher et al 1996). The signal was corrected as Sc = Su ∗ 10A/20 ,

(4)


6785

where Sc is the corrected signal, Su is the uncorrected signal and A is the difference in RF 2 Z power between two measurements, calculated in decibels. Power is calculated as Vtra ◦ where Vtra is the transmitter voltage required for the 180 pulse and Z is the impedance of the coil, which is 50 . We tested this correction procedure by correcting the signal intensities of a set of phantoms of different NaCl concentrations resulting in different coil loading. We used 20 mM, 500 ml cylindrical creatine phantoms of different NaCl concentrations (0.5%, 1%, 1.5% and 2% by weight). 2.3.4. B1 field inhomogeneity. Performing accurate quantification requires knowledge of the spatial distribution of the B1 field. The B1 field was measured by the double-angle method (Insko and Bolinger 1993), which measures the flip angle distribution within the sample, which is an indirect measure of the B1 of interest. Two gradient-echo images, I1 and I2 , with corresponding flip angles of θ and 2θ are recorded. T1 -dependent effects were avoided by using a long TR value (10 s). The ratio of these gradient-echo images (I1 /I2 ) is sin(θ )/sin(2θ ). Therefore, the flip angle distribution is given by θ = arccos(I2 /2I1 ).

(5)

The images were obtained by the gradient-echo pulse sequence with the following parameters: TE = 10 s, NEX = 1, slice thickness = 5 mm and θ was 35◦ . 3. Results 3.1. Signal linearity R 2 values showing the goodness of the fit of signal linearity as a function of concentration and the size of VOI were found to be 0.992, 0.999 (cubic VOI) and 0.976 (non-cubic VOI). This indicates excellent linearity of the signal intensities with the concentration and VOI of interest. 3.2. T1 and T2 effects T1 and T2 relaxation times of the C2 proton (8 ppm) of carnosine in vitro were found to be 560 ± 20 ms and 66.22 ± 9.3 ms, respectively. The in vivo T2 values (mean ± inter-subject SD) of soleus and gastrocnemius muscles were found to be 107.54 ± 25.34 and 92.11 ± 21.04, respectively. It should be noted that due to the relatively long TR of 2000 ms, the in vitro carnosine signal recovers its magnetization in the longitudinal direction by 97.1%. Given the fact that in vivo relaxation times are generally shorter than those of in vitro, this signal recovery is expected to be even higher in vivo. Therefore, the signal was not corrected for T1 relaxation. For T2 , however, the signal decays by almost 37% for a TE of 30 ms. Thus, correction for T2 relaxation is important and was taken into account. 3.3. Coil loading The effect of different coil loading on the acquired signal intensity is illustrated in figure 1 showing the signals acquired from the phantoms containing different amounts of NaCl, leading to different coil loading. In figure 1, it can be noted that signal intensities (peak areas) decrease as the amount of NaCl increases, resulting in an increase in coil loading (low quality factor) (Tofts 2003). Note also in figure 1 that transmitter voltages recorded for the given pulse length and flip angle increase as the amount of NaCl increases.


6786 80

1300

1200

1100 70 1000

transmitter voltage (Vtra)

signal intensity (a.u)

75

65 V

900

tra

signal signal

c

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

60

amount of NaCl (% weight)

Figure 1. The relation between the signal intensities (before (signal) and after correction (signalc )), NaCl concentration and RF power (Vtra ) required.

The recorded signals before (signal) and after correction (signalc ) are also shown in figure 1. The maximum relative deviation in signal intensities recorded from different phantoms was 22.14% before correction. After correction, the maximum signal deviation was as low as 3.41%. 3.4. B1 field inhomogeneity Whether B1 inhomogeneity has any effect on the measured carnosine concentration in different muscles and on the variations between two successive measurements was tested by performing B1 mapping in 1.5 l cylindrical aqueous phantoms containing different amounts of NaCl. The amount of NaCl was varied in concentrations of 0.1%, 0.33%, 0.5% and 1% by weight so as to change RF penetration into the sample, thereby the B1 inhomogeneity caused by RF standing waves (dielectric effects) (Tofts 1994). We also used an oil phantom since oil has a low dielectric constant ( = 2–3) (Tofts et al 1997), and is therefore expected to increase √ the wavelength of the RF field in the sample as the wavelength is inversely proportional to (Tofts 1994). Figure 2(a) shows the flip angle profile along the centerline in the axial plane obtained from each phantom. As can be seen, the largest deviation in B1 is seen in the distilled water phantom, showing the dome-shaped flip angle profile as a result of the standing waves generated. The mean flip angle and the corresponding standard deviation (SD) in this phantom were found to be 48.41 ± 3.41◦ . Note that the profile tends to be gradually flatter with an increasing amount of NaCl within the sample. It is important to point out that the best profile is obtained in the phantom containing 1% NaCl ( 30.23 ± 1.82◦ ), which corresponds to the physiological NaCl concentration. A homogenous B1 field is also found in the oil phantom. The B1 inhomogeneity can also be seen more clearly in figure 2, where 2D sagittal flip angle maps of the distilled water and the 1% NaCl-doped phantoms are shown. Note that the dome-shaped profile is also present in the longitudinal direction. In order to investigate the effect of the standing wave-induced B1 inhomogeneity on the in vivo results obtained, we performed in vivo B1 mapping in an athlete and an untrained volunteer.


6787 flip angle

55

52

distilled water 0.1% NaCl 0.33% NaCl 0.5 % NaCl 1% NaCl corn oil

50

50 48 46

45

flip angle

44 42

40

40 38

35

36 30

34 32

25 0

10

20

30

40

50

60

position along center line

(b)

(a) flip angle 33 32 31 30 29 28 27 26 25 24

(c)

Figure 2. Comparison of flip angle profiles of phantoms of different electrical conductivities. (a) Axial flip angle profile for phantoms containing different amounts of NaCl, (b) sagittal flip angle profile of the distilled water phantom and (c) the sagittal flip angle profile of a 1% NaCl-doped water phantom.

Figures 3(a) and (b) show the in vivo flip angle maps using the axial gradient-echo images obtained from an athlete and an untrained volunteer, respectively. Note that despite some fluctuations caused by the anatomy, flip angle profiles are markedly flat within the muscles. Figure 3(c) presents the flip angle map of the same volunteer as in figure 3(b), obtained using the sagittal images. The flat flip angle profile can also be noticed in this map. The corresponding mean and SD values of flip angles extracted from a region covering soleus and gastrocnemius muscles in the athlete and the volunteer were calculated to be 30.86 ± 1.87◦ and 31.41 ± 1.15◦ , respectively. Figure 3(d) shows a comparison of the in vivo centerline flip angle profiles as in figures 3(a) and (b) with the 1% NaCl-doped water phantom. It should be noted that due to the correspondence between the NaCl concentration in this phantom and in vivo conditions, there is a strong agreement among the profiles plotted. This indicates that measured signal intensities do not suffer from standing waves-induced B1 inhomogeneity. The effect of B1 inhomogeneity on the spectra was also investigated. This was achieved by comparing the creatine signal in the spectra acquired from the concentric and the eccentric voxels localized within a 500 ml spherical phantom. Figure 4(a) shows the diagram in which


6788 flip angle

flip angle

35

35

30

30

25

25

20

20

15

15

10

10

5

5

0

0

(b)

(a) flip angle 60

1% NaCl athlete untrained volunteer

55

35

50

30 45

25

40

flip angle

20

35 30

15 25

10

5

20 15 10 0

10

20

30

40

50

60

profile along the center line

(c)

(d)

Figure 3. Flip angle maps of an athlete, untrained volunteer and the comparison of the profile with that of the 1% NaCl-doped water phantom. (a) In vivo flip angle profile in the calf muscle of an athlete, (b) in vivo flip angle profile in the calf muscle of an untrained volunteer, (c) in vivo sagittal flip angle map in the calf muscle of an untrained volunteer and (d) centerline flip angle profile of (a), (b) and the 1% NaCl-doped water phantom.

the locations of the voxels are indicated along with the coronal slice in which B1 maps were acquired. To compare the effect of B1 inhomogeneity on the spectra, we recorded spectra from two 20 mM creatine phantoms containing 0.5% and 2% NaCl by weight, respectively. The B1 and proton density (PD) maps were acquired from the coronal slice passing through the center of the spherical phantoms. PD maps were calculated using the same GRE sequence with TR/TE = 10 000/4 ms and a flip angle of 8◦ . As seen in figure 4(c), the B1 map is markedly flat for the phantom doped with 2% NaCl (30.25 ± 1.51◦ ), as expected. There is a slight increment toward the edges of the phantom, which is due to noise. In contrast, B1 shows the standing wave effect in the center of the phantom containing 0.5% NaCl (figure 4(b)) (32.14±2.31◦ ). Figure 4(d) shows the difference between the PD profiles along the centerline in these phantoms, indicating the effect of B1 inhomogeneity. We compared the signal intensities of the creatine peak in the spectra in each phantom, acquired from concentric and eccentric voxels, respectively. The corresponding relative signal differences compared to the signal from the central voxel were found to


6789 flip angle 37 36 35 34 33 32 31 30 29 28 27

(b)

(a) flip angle 700

33 600

32

31

30

proton density (a.u)

500

400

300

29

200

28

100 2% NaCl − doped 0.5% NaCl − doped

27

0 0

10

20

30

40

50

60

position along center line

(c)

(d)

Figure 4. Flip angle maps and proton density images of phantoms containing 0.5% and 2% NaCl, respectively. (a) Schematic showing the concentric (C) and eccentric (E) voxel localizations and the position of the coronal slice in which B1 maps were calculated, (b) flip angle profile of the phantom containing 0.5% NaCl, (c) flip angle profile of the phantom containing 2% NaCl and (d) proton density map of the phantom as in (c).

be −13.21% and 0.59% for the 0.5% and 2% NaCl-doped phantoms, respectively. These changes were also found to be in agreement with the averaged PD values in corresponding voxels as seen in figure 4(d). These results indicate that B1 inhomogeneity has a significant effect on the spectral signal intensity. Owing to the absence of standing waves in the 2% NaCl-doped phantom, the signal change is around 1%. This signal change can be caused by the transmit/receive profile of the coil, which was assessed by moving the 50 mM carnosine phantom to different locations inside the coil. The spectra obtained from the phantom showed that the recorded signal intensities varied by less than 5%. Therefore, B1 inhomogeneity effects, which might be due to the transmit or receive field of the coil, were neglected in this study. 3.5. In vitro quantification The quantification protocol was evaluated using phantoms of different metabolite concentrations. We used two 500 mL phantoms with carnosine concentrations of 50 mM and 68 mM. To simulate different coil loadings, the electrical conductivities of the phantoms were


6790

Table 1. Measured concentration of carnosine in athletes (A) and untrained volunteers (V) in soleus and gastrocnemius muscles. (mM)

A1 A2 A3 V1 V2 V3 V4 V5 V6

Sol1 (mM)

Sol2 (mM)

VarSol %

Gast1 (mM)

Gast2 (mM)

VarGast %

2.7 2.3 3.43 2.67 1.75 2.76 3.71 2.29 2.3

2.96 2.06 3.26 2.35 1.68 2.86 3.61 2.19 2.35

9.6 10.4 4.9 11.9 4 3.6 2.6 4.3 2.1

3.75 4.02 6.64 3.63 3.31 3.22 3.72 3.02 2.91

3.82 3.66 6.73 4.1 3.35 2.83 3.44 2.44 3.87

1.8 8.9 1.3 12.9 1.2 12.1 7.5 19.2 32.9

made to be different by adding NaCl concentrations of 1.5% and 2% by weight, respectively. These values were chosen in order to provide flat B1 profiles along with different coil loadings. Using the 50 mM carnosine phantom as a concentration standard, the concentration of the other phantom was found to be 68.80 mM, which indicates an error of 1.17% only. To investigate the precision (reproducibility) of the experiments, we performed eight consecutive spectroscopic measurements on a phantom in two separate sessions after removing and replacing the phantom in the second session. The precision was expressed as the coefficient of variation (CV) and is calculated by SD/mean, where SD is the standard deviation obtained from the four consecutive measurements in each session and mean is the mean value of the acquired signal intensity. CV were found to be 3.03% and 4.09% for the first and second sessions, respectively. 3.6. In vivo quantification The values of the quantified carnosine concentration in the soleus (Sol) and gastrocnemius (Gast) muscles of three athletes (A) and six untrained volunteers (V) are listed in table 1. These values were obtained after the necessary corrections for relaxation times, coil loading and temperature differences, as described above. In order to assess the reproducibility, we performed a second measurement after removing and repositioning each volunteer into the scanner. The first and second measurements in each muscle type are represented by superscripts 1 and 2, respectively, and Var is the relative change between the corresponding values. To minimize the localization problems, we marked a spot on the leg using the laser indicating the center of the coil. Figure 5(a) shows a T1 -weighted axial image of one of the athletes in which the soleus and gastrocnemius muscles can be noted. The fitted and residual spectra acquired from the gastrocnemius muscle from this athlete are shown in figures 5(b) and (c), respectively. Taking the average of two consecutive measurements in each muscle, the measured mean concentrations in the soleus and the gastrocnemius muscles are found to be 2.81 ± 0.57/4.8 ± 1.59 mM and 2.58 ± 0.65/3.3 ± 0.32 mM for athletes and untrained volunteers, respectively. We also looked at the variation in signal intensity (in arbitrary units) of C4 resonance between successive measurements (data not shown). The maximum variation was found to be as high as 60.8% for athletes and 101.2% for untrained volunteers. This high variation in C4 resonance indicates the influence of orientation-dependent effects on this resonance as compared to the C2 resonance, as was previously reported (Kreis and Boesch 2000). It should be noted that carnosine concentration was found to be higher in gastrocnemius than in soleus muscle in all cases, except for one volunteer V4. The p-value acquired by paired


6791

soleus

gastrocnemius

(a)

(b)

(c)

Figure 5. T1 -weighted axial image showing the soleus and gastrocnemius muscles (a) and a fitted spectrum acquired from the gastrocnemius muscle (b). (c) Residual after the fit as in (b).

t-test was found to be 0.006. Furthermore, the paired difference between gastrocnemius and soleus values is greater in the athlete group than in the volunteer group (independent t-test, p-value = 0.045; Levene’s test for equality of variance, p-value = 0.186). These values are in agreement with previously reported biopsy-based values (Hill et al 2007). 4. Discussion In this work, we performed absolute quantification of carnosine in human calf muscle noninvasively by means of 1 H MRS. The effect of the signal relaxation, coil loading and temperature difference between the reference external standard (i.e. phantom) and in vivo were taken into account. We included three athletes and six untrained volunteers in the in vivo measurements. The quantified in vivo carnosine concentrations are in agreement with the reported biochemical range of 2–20 mM (Scriver et al 1983) and reported biopsy-based results indicating concentrations of about 4.9 mM in untrained subjects (≈19.9 mmol kg−1 dm) (Hill et al 2007) and 10 mM in body builders (≈43 mmol kg−1 dm) (Tallon et al 2005). Additionally, the measured carnosine concentrations were found to be higher in the gastrocnemius muscle than in the soleus muscle (p-value = 0.006). This is in agreement with the reported results based on biopsy, and the difference in carnosine concentration in gastrocnemius and soleus

6792


was attributed to the different fiber types present in the muscles in question (Hill et al 2007), which results in a higher capacity of gastrocnemius muscle for intense anaerobic exercise compared to the soleus muscle. As a result of the higher exercise capacity, the need for pHi buffering can be expected to be higher, hence the carnosine concentration also. Despite the small number of subjects which does not allow conclusive remarks to be made, the carnosine concentration in gastrocnemius seems to be higher in athletes (≈4.8) than that in untrained volunteers (≈3.3) (p-value = 0.045). This is expected as gastrocnemius is a phasic muscle that is used during sprint exercise. Thus, the sprint-trained athletes only train their gastrocnemius and not their soleus. This is also in concordance with previous reports on higher concentration of carnosine in animals with a capacity for prolonged intense exercise or in those exposed to frequent periods of increased hypoxia (Abe 2000, Crush 1970). In order to evaluate the effect of the dielectric properties of the subject on the accuracy of the quantification, B1 maps were acquired in phantoms of different NaCl concentrations and in the calf muscle of an athlete and an untrained volunteer. Due to the increased wavelength of the electromagnetic field introduced to the samples, we observed a flat flip angle profile in the phantom containing 1% NaCl concentration. This is equivalent to the in vivo NaCl concentration. As the NaCl concentration decreases, the standing wave effect is more drastically noticed. This effect was the worst for the distilled water phantom. This also supports the previous work (Tofts 1994), indicating the pitfall of using non-doped water phantoms in MR. The flat flip angle profile in the in vivo maps is in accordance with the 1% NaCl-doped phantom showing a flat profile. This suggests that the variations between successive measurements and carnosine concentrations quantified in soleus and gastrocnemius muscles were not brought about by the B1 inhomogeneity. The high variation observed in the signal intensity of C4 resonance can be attributed to orientation-dependent effects on this resonance (Kreis and Boesch 2000). Based on our reproducibility measurements, as compared to C2 resonance, C4 resonance was found to be very sensitive to orientation-dependent dipolar coupling effects on account of the imperfect repositioning of a subject. Thus, the C2 resonance is a better reference for the quantification of carnosine in human calf muscle as it changes less with orientation. Nonetheless, a potential signal contamination from amide protons to C2 resonance which in turn might lead to higher signal intensity of C2 resonance (Alonso et al 1990, Schröder L and Bachert 2003, Schröder et al 2004) should be remembered. Also, the importance of the correct repositioning of a subject should still be kept in mind while using this resonance for the absolute quantification of carnosine in human calf muscle in vivo. Albeit it is not very likely in this study owing to the large chemical shift difference between water signal (4.7 ppm) and C2 and C4 (8–7 ppm) resonances, there is a potential in signal loss of these resonances which might be introduced by frequency selective water suppression pulses (van der Veen et al 2000). This can be particularly important in the presence of significant B0 inhomogeneity as would be the case of large volume coverage (i.e. MRSI of calf muscle). In the current study, C2 and C4 resonances did not display satellite peaks as reported elsewhere (Schröder and Bachert 2003), which might be due to higher field strength (i.e. 3T) and TE (30 ms) used in this study. Therefore, the least-squares fit performed included only C2 and C4 singlets without taking the satellite peaks into account. However, we cannot rule out that there might have been contributions owing to invisible satellite peaks, giving rise to variations in the quantified values of carnosine. Another source of error in the reported values arises from the T2 values used for the absolute quantification. The inter-subject variation (SD) in T2 values measured in three volunteers was as high as 23%. Thus, based on our simulations, the error in the quantified carnosine concentrations might be as high as 0.5 mM, which is not negligible. This suggests that if the intent is to conduct absolute quantification, the acquisition of individual T2 values should be attempted as opposed to common practice


6793

of adopting literature values. One should note that owing to the use of the mean T2 of three untrained volunteers, the reported values of other individuals in the current study are also prone to this source of error. 5. Conclusion In this study, it is shown that 1 H MRS can be used for the absolute quantification of carnosine in the human calf muscle in vivo. The measured carnosine concentrations were in the same range as the reported concentrations acquired by biopsy. As a non-invasive technique with high accuracy, 1 H MRS can be used as an alternative to biopsy. We believe that 1 H MRS might play a role in understanding the physiological functions of carnosine in the human body. References Abe H 2000 Role of histidine-related compounds as intracellular proton buffering constituents in vertebrate muscle Biochemistry 65 757–65 Alonso J, Arus C, Westler W M and Markley J 1990 Two-dimensional spectra of intact tissue: homonuclear Hartmann– Hahn spectroscopy provides increased sensitivity and information content as compared to COSY Magn. Reson. Med. 15 142–51 Bate-Smith E C 1938 The buffering of muscle in rigor: protein, phosphate and carnosine J. Physiol. 92 336–43 Boesch C, Slotboom J, Hoppeler H and Kreis R 1997 In vivo determination of intra-myocellular lipids in human muscle by means of localized 1 H-MRS Magn. Reson. Med. 37 484–93 Bottomley P A, Lee Y and Weiss R G 1997 Total creatine in muscle: imaging and quantification with proton MR spectroscopy Radiology 204 403–10 Bottomley P A and Weiss R G 2001 Noninvasive localized MR quantification of creatine kinase metabolites in normal and infarcted canine myocardium Radiology 219 411–8 Brown C E 1981 Interactions among carnosine, anserine, ophidine and copper in biochemical adaptation J. Theor. Biol. 88 245–56 Buchli R, Ernst M and Boesiger P 1994 Comparison of calibration strategies for the in vivo determination of absolute metabolite concentrations in the human brain by 31 P MRS NMR Biomed. 7 225–30 Crush K G 1970 Carnosine and related substances in animal tissues Comput. Biochem. Physiol. 34 3–30 ¨ Derave W, Ozdemir M S, Harris R C, Pottier A, Reyngoudt H, Koppo K, Wise J A and Achten E 2007 Betaalanine supplementation augments muscle carnosine content and attenuates fatigue during repeated isokinetic contraction bouts in trained sprinters J. Appl. Physiol. 103 1736–43 Gulewitsch W and Amiradzibi S 1900 Uber das carnosine, eine neue organische Base des Fleischextraktes Ber. Dtsch. Chem. Ges. 33 1902–3 Harris R C, Dunnett M and Greenhaff P L 1998 Carnosine and taurine contents in individual fibers of human vastus lateralis muscle J. Sports Sci. 16 639–43 Harris R C, Tallon M J, Dunnett M, Boobis L, Coakley J, Kim H J, Fallowfield J L, Hill C A, Sale C and Wise J A 2006 The absorption of orally supplied β-alanine and its effect on muscle carnosine synthesis in human vastus lateralis Amino Acids 30 279–89 Hill C A, Harris R C, Kim H J, Harris B D, Sale C, Boobis L, Kim C K and Wise J A 2007 Influence of βalanine supplementation on skeletal muscle carnosine concentrations and high intensity cycling capacity Amino Acids 32 225–33 Insko E K and Bolinger L 1993 Mapping of radiofrequency field J. Magn. Reson. Ser. A 103 82–5 Kreis R and Boesch C 1996 Spatially localized, one- and two-dimensional NMR spectroscopy and in vivo application to human muscle J. Magn. Reson. Ser. B 113 103–18 Kreis R and Boesch C 2000 Orientation dependence is the rule, not the exception in 1 H-MR spectra of skeletal muscle: the case of carnosine ISMRM Proc. Int. Soc. Magn. Reson. Med. 31 Kreis R, Jung B, Slotboom J, Felblinger J and Boesch C 1999 Effect of exercise on the creatine resonances in 1 H MR spectra of human skeletal muscle: the case of carnosine J. Magn. Reson. 137 350–7 Mannion A F, Jakeman P M, Dunnett M, Harris R C and Willan P L T 1992 Carnosine and anserine concentrations in the quadriceps femoris muscle of healthy humans Eur. J. Appl. Physiol. 64 47–50 Pan J W, Hamm J R, Rothman D L and Shulman R G 1988 Intracellular pH in human skeletal muscle by 1 H NMR Proc. Natl Acad. Sci. 85 7836–9

6794


Schick F, Eismann B, Jung W I, Bongers H, Bunse M and Lutz O 1993 Comparison of localized proton NMR signals of skeletal muscle and fat tissue in vivo: two lipid compartments in muscle tissue Magn. Reson. Med. 29 158–67 Schröder L and Bachert P 2003 Evidence for a dipolar-coupled AM system in carnosine in human calf muscle from in vivo 1 H NMR spectroscopy J. Magn. Reson. 164 256–69 Schröder L, Schmitz C and Bachert P 2004 Molecular dynamics and information on possible sites of interaction of intramyocellular metabolites in vivo from resolved dipolar couplings in localized 1 H NMR spectra J. Magn. Reson. 171 213–24 Scriver C R, Perry T L and Nutzenadel W 1983 The Metabolic Basis of Inherited Disease (New York: McGraw-Hill) pp 570–85 Soher B J, van Zijl P C M, Duyn J H and Barker P B 1996 Quantitative proton MR spectroscopic imaging of the human brain Magn. Reson. Med. 35 356–63 Suzuki Y, Ito O, Mukai N, Takahashi H and Takamatsu K 2002 High level of skeletal muscle carnosine contributes to the latter half of exercise performance during 30-s maximal cycle ergometer sprinting Japan. J. Physiol. 52 199–205 Tallon M J, Harris R C, Boobis L H, Fallowfield J L and Wise J A 2005 The carnosine content of vastus lateralis is elevated in resistance-trained bodybuilders J. Strength. Cond. Res. 19 725–9 Taylor R, Price T B, Rothman D L, Shulman R G and Shulman G I 1992 Validation of 13C NMR measurement of human skeletal muscle glycogen by direct biochemical assay of needle biopsy samples Magn. Reson. Med. 27 13–20 Tofts P 2003 Quantitative MRI of the Brain-Measuring Changes Caused by Diseases (Chichester: Wiley) Tofts P S 1994 Standing waves in uniform water phantoms J. Magn. Reson. Ser. B 104 143–7 Tofts P S, Barker G J, Dean T L, Gallagher H, Gregory A P and Clarke R N 1997 A low dielectric constant customized phantom design to measure rf coil nonuniformity Magn. Reson. Imaging 15 69–75 Wendling P S, Peters S J, Heigenhauser J F and Spriet L L 1996 Variability of triacylglycerol content in human skeletal muscle biopsy samples J. Appl. Physiol. 81 1150–5 van der Veen J W C, Weinberger D R, Tedeschi G, Frank J A and Duyn J H 2000 Proton MR spectroscopic imaging without water suppression Radiology 217 296–300