Author + information
- Received July 9, 2002
- Revision received November 19, 2002
- Accepted May 7, 2003
- Published online November 5, 2003.
- Ichiro Nakae, MD*,
- Kenichi Mitsunami, MD†,* (, )
- Tomoko Omura, MD*,
- Takahiro Yabe, MD*,
- Takayoshi Tsutamoto, MD*,
- Shinro Matsuo, MD*,
- Masayuki Takahashi, MD*,
- Shigehiro Morikawa, MD‡,
- Toshiro Inubushi, PhD‡,
- Yasuyuki Nakamura, MD*,
- Masahiko Kinoshita, MD* and
- Minoru Horie, MD*
- ↵*Reprint requests and correspondence:
Dr. Kenichi Mitsunami, Department of General Medicine, Medical Coordination Center, Shiga University of Medical Science, Seta, Otsu 520-2192, Japan.
Objectives This study noninvasively examined total creatine (CR) of the myocardium in dilated cardiomyopathy (DCM) or hypertrophic cardiomyopathy (HCM) using proton magnetic resonance spectroscopy (1H-MRS).
Background Abnormalities in CR metabolism in failing hearts have been reported. A biochemical study suggested that myocardial metabolic changes are very similar in DCM and HCM despite the different heart failure (HF) mechanisms.
Methods Using cardiac-gated 1H-MRS with magnetic resonance image (MRI)-guided point-resolved spectroscopy (PRESS) localization, we quantitatively measured septal CR. Patients with either DCM (n = 11) or HCM (n = 7) and age-matched normal subjects (n = 14) were examined.
Results Myocardial CR was significantly lower in DCM patients (16.1 ± 4.5 μmol/g wet weight [range 10.2 to 22.9], p < 0.05) than that in subjects with normal hearts (27.6 ± 4.1 μmol/g [range 21.4 to 36.2]). Myocardial CR in HCM patients (22.6 ± 8.1 μmol/g [range 12.2 to 34.5]) was significantly lower than that in subjects with normal hearts (p < 0.05) but was significantly higher than that in DCM patients (p < 0.05). In 18 patients with either DCM or HCM, myocardial CR correlated positively with left ventricular ejection fraction (LVEF) (y = 0.22x + 9.8, r = 0.73, p = 0.0006) but correlated negatively with plasma B-type natriuretic peptide (BNP) levels (y = −0.012x + 22.4, r = −0.54, p = 0.022).
Conclusions This study showed that 1H-MRS can noninvasively detect CR depletion associated with the severity of HF in cardiomyopathy.
Abnormalities in creatine (CR) metabolism of the heart (1–8)and other organs (9–11)in patients with failing hearts have been demonstrated by 31P magnetic resonance spectroscopy (31P-MRS) or biochemical methods. It was demonstrated by using 31P-MRS that the phosphocreatine (PCr)-to-adenosine triphosphate (PCr/ATP) ratio is lower in the failing human myocardium (1–3). In patients with dilated cardiomyopathy (DCM), this ratio correlated well with the severity of heart failure (HF) estimated from the New York Heart Association (NYHA) classification (3). In patients with hypertrophic cardiomyopathy (HCM), some studies indicated that myocardial PCr/ATP ratios are lower than normal (4–7). We previously reported that absolute concentrations of both myocardial PCr and ATP were lower than normal in HCM patients, as determined by 31P-MRS (8). By biochemical methods, Kalsi et al. (12)reported that the changes in myocardial energy metabolism are very similar in biopsy specimens from DCM and HCM patients before heart transplantation despite the different HF mechanisms. The myocardial CR pool was reduced in both HCM and DCM patients (12,13).
Recently, total CR was quantitatively measured in the human heart by proton magnetic resonance spectroscopy (1H-MRS). In patients with a myocardial infarction, Bottomley and Weiss (14)demonstrated that myocardial CR was decreased in nonviable infarcted regions using 1H-MRS with stimulated-echo acquisition mode (STEAM) localization.
In the present study, we measured the CR concentration in the human myocardium by using 1H-MRS with point-resolved spectroscopy (PRESS) localization, another method of spatial localization (15). Using this method, we examined the CR concentration in non-ischemic dysfunctional hearts of DCM and HCM patients, which, to our knowledge, has never been reported with the use of 1H-MRS. We further assessed the left ventricular ejection fraction (LVEF) and plasma B-type natriuretic peptide (BNP) levels as indicators of HF severity.
The study protocol was approved by the Ethical Committee of Shiga University of Medical Science. The MRS studies were done with a GE 1.5-T Signa imaging/spectroscopy system (General Electric Medical Systems, Milwaukee, Wisconsin). Subjects were examined in the supine position. The Signa 1.5-T general-purpose flex coil (GPFLEX, GE Medical Systems) was wrapped around the chest so as to be centered over the heart. Cardiac-gated spin-echo magnetic resonance imaging (MRI) was performed to determine the location of localized volume elements (voxels) for MRS investigation. Voxels were localized to 8 cm3(2 × 2 × 2 cm) in the interventricular septum by the PRESS method (15). This voxel size was chosen to yield suitable signal-to-noise ratios. The PRESS flip angles were adjusted to achieve the maximum intensity for the echo signal; therefore, the angles were set to 90° for the first pulse and 180° for the second and third pulses. Automatic shimming was performed. The spectral acquisition was gated once per two cardiac cycles using plethysmography or electrocardiography. The acquisition parameters included an echo time (TE) of 25 ms and a repetition time (TR) of 1.4 to 2.9 s. At acquisition, the initial data set of 16 signals was collected without water suppression for the water resonance, and then a data set of 128 signals was collected with water suppression for the CR resonance. A chemical shift-selective (CHESS) sequence was used to suppress the water signal (16). The chemical shift of water was taken as 4.75 ppm, as referred to in the Signa Horizon LX PSD Manual. Spectral peaks were identified with known chemical shifts: cholines at 3.2 ppm, CR at 3.0 ppm, and lipids at 0.9 to 1.4 ppm (14,17).
Quantitation of CR spectra of the myocardium was performed using water signals without suppression as an internal concentration reference (14,17,18). The integrated areas of the resonances of CR and water were measured as reported previously in 31P-MRS studies (8,19). Analysis was performed with a home-built automatic data processing station, using the simplex technique (20). Each peak was well fitted to an 80% Gaussian and 20% Lorentzian line. All signal areas were corrected for relaxation losses according to the following formula (14,17):where Snis the total CR or water (W) magnetic resonance signal area in the tissue; S*nis the corrected Sn; Enand Fnare correction factors for effective spin-spin (T2e) and spin-lattice (T1) relaxation effects, respectively; TR is the pulse sequence repetition period; and TE is the echo time. The T1 values of myocardial CR and W were estimated using the spectra of five healthy volunteers acquired at TRs of 1.5 and 6 s using PRESS without cardiac gating. The obtained T1 values were 1.48 ± 0.33 s and 1.21 ± 0.33 s (n = 5) for CR and W, respectively. Spectral acquisition was gated once per two cardiac cycles in this study: TR (s) = 120/heart rate (beats/min). Thus, Fcr/Fwis 1.09 when the heart rate is 60 beats/min, where Fcrand Fware T1 correction factors for CR and W, respectively, as defined in the formula (N = CR or W). The T2e values of myocardial CR and W were obtained from the spectra of five volunteers acquired at TEs of 25 to 45 ms. The obtained T2e values were 135 ± 13 ms and 33.1 ± 6.1 ms (n = 5) for CR and W, respectively. Thus, the value of Ecr/Ewis 0.565 at TE = 25 ms, where Ecrand Eware T2e correction factors for CR and W, respectively, as defined in the formula (N = CR or W). The concentration of total CR in tissue filling a voxel was calculated from the tissue water concentration according to the following equation: The numbers “2” and “3” in the equation indicate the two protons of water and the three protons of the N-methyl group of CR, respectively. The concentration of water was measured as 55.5 mol/l, and the myocardial tissue water content was measured as 72.7% by weight (18,21). In this study, therefore, the myocardial tissue water concentration was calculated by the following equation: In 22 healthy volunteers (age 26 to 76 years), myocardial CR contents were measured. There was no history suggestive of cardiac or respiratory disease. They showed no pathologic findings on the electrocardiogram and no cardiomegaly on roentgenologic examination of the chest. The acquisition time was ∼5 min per single examination when the heart rate was 60 beats/min. The repeated CR examinations yielded a coefficient of variation of 7.4% (n = 8). The correlation between age and myocardial CR concentration was examined.
Patients with idiopathic DCM (n = 11) or HCM (n = 7) in NYHA functional class I, II, or III and 14 age-matched healthy volunteers [n = 14, NML(1)] were included in this study (Table 1). None of the patients showed evidence of ischemia on the basis of coronary arteriograms or thallium-201 stress imaging using a treadmill test. The average period between the detection of disease and examination by MRS in DCM patients was 54 ± 58 months, whereas the average period in HCM patients was 117 ± 84 months. In HCM patients, four patients had non-obstructive HCM, one patient had obstructive HCM, and two patients with HCM at the dilated phase were examined. Only one of the HCM patients had atrial fibrillation when this study was performed. All the DCM patients had a clinical history of severe HF (NYHA class IV), whereas two of the HCM patients had a history of NYHA class IV. The study involving 1H-MRS was not performed in the acute phase of CHF, because of clinical instability such as orthopnea. Thus, the MRS study was performed when clinical stability was confirmed. The NYHA functional class was determined when the MRS study was performed. Of the 11 patients with DCM, 9 patients were treated with digitalis, 11 with diuretics, 10 with beta-blocker, 9 with angiotensin-converting enzyme inhibitors, and two with an angiotensin II antagonist. Of the seven HCM patients, four were treated with beta-blocker, one with calcium antagonist, three with diuretics, one with digitalis, one with an angiotensin-converting enzyme inhibitor, and two with an angiotensin II antagonist. All the patients gave their written, informed consent before the study.
BNP measurements in patients with DCM or HCM
In the 18 patients with either DCM or HCM, the plasma BNP levels were measured with a commercially available specific immunoradiometric assay kit for human BNP (Shionoria BNP kit, Shionogi, Osaka, Japan). Plasma BNP was also measured on a day close to the MRS study, and NYHA functional class in CHF patients at this time was the same as that at the MRS study. Left ventricular (LV) diastolic and systolic dimensions and LVEF were assessed by ultrasonic echocardiography.
Data are expressed as the mean ± SD. Single comparisons were performed with the Student unpaired ttest. Multiple comparisons were performed by one-way analysis of variance followed by the Fisher least significant difference test. A p value <0.05 was considered statistically significant.
MRS study in normal subjects
Representative data of MRI-guided 1H-MRS obtained from a normal volunteer are shown in Figure 1. We measured the myocardial CR in 22 healthy humans (age 26 to 76 years) by 1H-MRS [NML(2)], non–age-matched larger normal control group). The CR peak was observed at 3.03 ± 0.08 ppm. As shown in Figure 2, there was no significant correlation between age and myocardial CR concentration. The mean myocardial CR value was 28.9 ± 4.4 μmol/g in ages <40 years (n = 9), 28.6 ± 3.8 μmol/g in ages ≥40 years (n = 13), and 28.7 ± 4.0 μmol/g in total (n = 22).
MRS study in patients with DCM or HCM
Myocardial CR concentrations measured by 1H-MRS was compared among three age-matched groups: 11 patients with DCM, seven patients with HCM, and 14 normal controls, age-matched normal control group [NML(1)]. Fourteen age-matched normal controls overlapped with the 22 normal subjects in Figure 2. The clinical characteristics are summarized in Table 1. There were no significant differences in age, gender, body weight, and heart rate among the three groups. The LV diastolic dimension was significantly smaller in the HCM group (51.4 ± 10.1 mm [range 41.0 to 66.0], p < 0.01) than in the DCM group (69.2 ± 10.2 mm [range 56.0 to 87.0]), and the LV systolic dimension was also significantly smaller in the patients with HCM (35.0 ± 13.1 mm [range 22.0 to 58.0], p < 0.01) than in patients with DCM (60.4 ± 11.4 mm [range 46.0 to 77.0]). The LVEF was significantly higher in patients with HCM (59.8 ± 20.2% [range 25.5 to 82.6%], p < 0.01) than in the patients with DCM (27.2 ± 10.9% [range 15.4 to 43.9%]). The representative 1H magnetic resonance spectrum obtained from a patient with DCM is shown in Figure 3. The representative spectrum from a patient with HCM is shown in Figure 4, and further, the spectrum at the dilated phase is shown in Figure 5. These spectra are shown at the same scale relative to water (Figs. 1, 3, 4, and 5). The CR peaks were observed at 3.01 ± 0.03 ppm in the DCM group, 3.00 ± 0.03 ppm in the HCM group, and 3.02 ± 0.06 ppm in the NML(1) group. As shown in Figure 6, the myocardial CR concentrations in DCM (16.1 ± 4.5 μmol/g wet weight [range 10.2 to 22.9], p < 0.05) were significantly lower than those in the NML(1) group (27.6 ± 4.1 μmol/g wet weight [range 21.4 to 36.2]). The CR values in the NML(2) group are also shown in Figure 6. Myocardial CR concentrations in patients with HCM (range 12.2 to 34.5) ranged between those in NML(1) and those in the DCM. Two cases of HCM at the dilated phase had lower values of CR (12.2 and 15.6 μmol/g), similar to the values found in DCM, whereas two cases of asymptomatic HCM had higher values of CR (28.0 and 34.5 μmol/g), close to the values found in the NML(1). Thus, myocardial CR in the HCM group (22.6 ± 8.1 μmol/g wet weight) was significantly lower than that in NML(1) (p < 0.05) but was significantly higher than that in the DCM group (p < 0.05). The two panels in Figure 7show the relationship between LVEF and myocardial CR concentration, and the relationship between the plasma BNP level and myocardial CR concentration in 18 patients with either DCM or HCM. Mean plasma BNP levels in DCM and HCM patients were 321 ± 353 (n = 11) and 320 ± 259 pg/ml (n = 7), respectively. There was a positive correlation between LVEF and myocardial CR (y = 0.22x + 9.8, r = 0.73, p = 0.0006), but there was a negative correlation between plasma BNP levels and myocardial CR (y = −0.012x + 22.4, r = −0.54, p = 0.022). When the analysis was done separately for the DCM and HCM groups, LVEF did not significantly correlate with CR in the DCM group (p = 0.20); there was a borderline significance between LVEF and CR in the HCM group (p = 0.06); BNP did not significantly correlate with CR in the DCM group (p = 0.11); but BNP correlated negatively (r = −0.88, p = 0.01) with CR in the HCM group.
Our study demonstrated that myocardial CR was lower in non-ischemic dysfunctional hearts with either DCM or HCM than in ischemic dysfunctional hearts. Myocardial CR concentrations correlated with the severity of HF estimated from LVEF and plasma BNP levels: the more severe the HF, the lower the CR concentration.
1H-MRS study in normal subjects
We measured myocardial CR by cardiac-gated 1H-MRS with MRI-guided PRESS localization. Formerly, it had been reported that cardiac lipids and CR were observed by 1H-MRS with PRESS localization (22). More recently, quantitative measurements of CR in human hearts by 1H-MRS with STEAM localization were reported (14). These values obtained from 1H-MRS were confirmed by biochemical measurements at biopsy (CR: 21.8 ± 2.9 μmol/g) in an animal study (23). In our preliminary study, we obtained less noisy spectroscopy with the PRESS method than with the STEAM method. The CR signals were most clearly observed in the septum, probably because they were not influenced by the lipids in the epicardium or by the air in the lung. Thus, we acquired CR signals by selected localization of the septum using the PRESS method. Our CR value measured in non-diseased myocardium (27.6 ± 4.1 μmol/g) was in close agreement with the value (28 ± 6 μmol/g) in a previous study by Bottomley and Weiss (14).
We examined the correlation between age and the myocardial CR concentration in normal volunteers. In the age range studied, the myocardial CR concentration did not correlate significantly with age. Because the number of healthy volunteers more than 65 years of age was small, however, further examination may be needed to clarify the influence of aging on the CR levels in human hearts (8,24).
In this study, voxels were localized to 2 × 2 × 2 cm in the interventricular septum. This size may be larger than the wall thickness, especially in DCM patients. Thus, this voxel may include the water of the blood in the ventricle, which may cause CR values to be underestimated. However, spin-echo MRI showed no signal intensity in the chamber. Similar to previous studies (14,23), the integrated signal intensity in the large portion of the PRESS voxel that intersects the blood-filled chamber was small, owing to the “black blood” properties of the sequence. Even if the voxel includes a blood space of 30% in total volume, the underestimated CR value is <8%. This difference should not substantively affect our finding of low CR levels in cardiomyopathy.
Furthermore, the T1 and T2e values were obtained in our separate study of five healthy volunteers in the age range of 29 to 52 years. The corrections based on the T1 and T2e values obtained from five healthy volunteers were applied to all subjects including HF patients and higher aged people in this study, although the T1 and T2e values may change with aging or disease. Thus, these corrections may cause some error in the patients and aging subjects. In addition, we assumed that tissue water contents in the diseased hearts were the same as those in healthy hearts.
1H-MRS study in cardiomyopathy
In our study, myocardial CR in the patients with either DCM or HCM was lower than in age-matched normal subjects [NML(1)]. The CR concentration correlated positively with LVEF but negatively with plasma BNP levels. A previous 31P-MRS study revealed that the myocardial PCr/ATP ratio correlated with the severity of HF estimated from the NYHA class in DCM patients (3). Later, it was shown that the PCr/ATP ratio correlated with LVEF in patients with DCM (25). Thus, the myocardial PCr/ATP ratio in the 31P-MRS study and also the CR levels in the present study were closely related to the severity of failing hearts.
Creatine metabolism is essential for normal cardiac function and viability. Creatine is transported against a concentration gradient from the blood into the myocytes (26). Adenosine triphosphate is the sole substrate for myofibrillar ATPase and is absolutely required for muscle contraction. The creatine kinase (CK) reaction is important for the rapid resynthesis of ATP (27–29). Phosphoryl transfer from PCr to adenosine diphosphate (ADP) is catalyzed by CK: PCr + ADP + H+⇌ ATP + CR.
Nascimben et al. (30)examined human myocardium samples and showed that the CR content and total CK activity were both lower than normal in the failing myocardium. A recent study showed that one mechanism of CR depletion in DCM is caused by downregulating the Na+-CR co-transporter, which acts when CR is taken up from the blood (13). The CR level measured by 1H-MRS probably reflects the degree of myocardial cellular damage in the diseased heart. A loss in CR results in a loss in PCr. Using 31P-MRS, it was previously reported that myocardial PCr/ATP was a predictor of total and cardiovascular mortality in patients with DCM (31). In that study, 39 patients with DCM were followed up for ∼2.5 years, and 8 of 20 patients with a low PCr/ATP ratio died, all from cardiovascular causes. Thus, the myocardial PCr/ATP ratio in the 31P-MRS study and also the CR levels in the 1H-MRS study may provide important prognostic information on the diseased heart.
In patients with HCM, myocardial CR concentrations ranged between those in the NML groups and those in the DCM group. Myocardial CR concentrations of the hypertrophic regions in two patients with asymptomatic HCM were normal. In one case of obstructive HCM, myocardial CR was well preserved. In this case, diffuse hypertrophy of the LV wall (thickness 16 to 20 mm), including the outflow tract, was observed; LVEF was 73%. Our study suggested that the myocardial CR levels in HCM patients are well preserved at the early stage of disease or when normal contractile function is maintained while hypertrophy increases. In two patients who had already advanced to the dilated phase of HCM, however, myocardial CR concentrations had obviously decreased to a degree similar to that in DCM patients. These cases exhibited typical symptoms of HF, such as dyspnea on exertion, and had higher plasma BNP values.
The mean value of CR was higher in the HCM group than in the DCM group. Indeed, it is unclear whether this difference is due to the difference in pathologic mechanisms between DCM and HCM. When the present study was performed, NYHA functional class and average values of plasma BNP were similar between the DCM and HCM groups. However, LVEF was higher in the HCM group. All patients had a clinical history of NYHA class IV in the DCM group, but only two patients had such a history in the HCM group. Kalsi et al. (12)reported that myocardial metabolic changes are very similar in biopsy specimens from DCM and HCM patients (both NYHA class IV) despite the different HF mechanisms. Thus, the difference in CR concentrations between DCM and HCM patients appears to be mainly due to the difference in the severity of HF.
31P-MRS and 1H-MRS
Recently, a combined 31P/1H-MRS approach was attempted in the animal heart and provided useful information (23). To date, it is difficult to perform combined 31P and 1H-MRS in patients with HF, because the total examination time required is over 2 h in our laboratory. If combined 31P and 1H-MRS can be applied to humans, however, we can obtain more detailed information, which would be valuable in the elucidation of pathologic mechanism(s) in diseased hearts.
Our results suggest that noninvasive measurement of myocardial CR by 1H-MRS is useful in the assessment of HF severity.
- B-type natriuretic peptide
- dilated cardiomyopathy
- hypertrophic cardiomyopathy
- heart failure
- 1H- and 31P-MRS
- proton and phosphorus-31 magnetic resonance spectroscopy, respectively
- left ventricular
- left ventricular ejection fraction
- magnetic resonance imaging
- magnetic resonance spectroscopy/spectroscopic
- age-matched normal control group
- non–age-matched larger normal control group
- New York Heart Association
- point-resolved spectroscopy
- stimulated-echo acquisition mode
- Received July 9, 2002.
- Revision received November 19, 2002.
- Accepted May 7, 2003.
- American College of Cardiology Foundation
- Neubauer S.,
- Krabe T.,
- Schindler R.,
- et al.
- Jung W.I.,
- Sieverding L.,
- Breuer J.,
- et al.
- Massie B.,
- Conway M.,
- Yonge R.,
- et al.
- Lee C.W.,
- Lee J.-H.,
- Kim J.-J.,
- et al.
- Neubauer S.,
- Remkes H.,
- Spindler M.,
- et al.
- Yabe T.,
- Mitsunami K.,
- Inubushi T.,
- Kinoshita M.
- Snyder WS, Cook MJ, Nasset ES, et al. Report of the task group on reference man. International Commission on Radiological Protection, no. 23. Oxford: Pergamon Press, 1984:116
- Neubauer S.,
- Horn M.,
- Pabst T.,
- et al.
- Ingwall J.S.
- Bittl J.A.,
- Ingwall J.S.
- Zimmer H.-G.,
- Trendelenburg C.,
- Kammermeier H.,
- et al.
- Nascimben L.,
- Ingwall J.S.,
- Pauletto P.,
- et al.
- Neubauer S.,
- Horn M.,
- Cramer M.,
- et al.