| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Musculoskeletal Imaging |
1 From the Division of Musculoskeletal Imaging (M.T., B.J.T., M.E.J., D.I.R., W.E.P.) and Institute for Technology Assessment (E.F.H.), Massachusetts General Hospital and Harvard Medical School, 15 Parkman St, WACC 515, Boston, MA 02114. Received September 27, 2004; revision requested December 2; revision received December 6; accepted January 12, 2005. Supported in part by NIH grant M01 RR01066. Address correspondence to M.T. (e-mail: mtorriani{at}hms.harvard.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Institutional review board approval and written informed consent were obtained for this Health Insurance Portability and Accountability Act–compliant study. The authors examined the anterior tibial muscles of 27 healthy subjects aged 19–48 years (12 men, 15 women; mean age, 25 years) by using single-voxel short-echo-time point-resolved 1H MR spectroscopy. During a first visit, the subjects underwent 1H MR spectroscopy before and after being repositioned in the magnet bore, with voxels carefully placed on the basis of osseous landmarks. Measurements were repeated after a mean interval of 12 days. All spectra were fitted by using Java-based MR user interface (jMRUI) and LCModel software, and lipid peaks were scaled to the unsuppressed water peak (at 4.7 ppm) and the total creatine peak (at approximately 3.0 ppm). A one-way random-effects variance components model was used to determine intraday and intervisit coefficients of variation (CVs). A power analysis was performed to determine the detectable percentage change in lipid measurements for two subject sample sizes.
RESULTS: Measurements of the IMCL methylene protons peak at a resonance of 1.3 ppm scaled to the unsuppressed water peak (IMCLW) that were obtained by using jMRUI software yielded the lowest CVs overall (intraday and intervisit CVs, 13.4% and 14.4%, respectively). The random-effects variance components model revealed that nonbiologic factors (equipment and repositioning) accounted for 50% of the total variability in IMCL quantifications. Power analysis for a sample size of 20 subjects revealed that changes in IMCLW of greater than 15% could be confidently detected between 1H MR spectroscopic measurements obtained on different days.
CONCLUSION: 1H MR spectroscopy is feasible for repeatable quantification of IMCL concentrations in longitudinal studies of muscle metabolism.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The variability of 1H MR spectroscopic IMCL quantifications has been examined in limited subsets of patients in broader studies (7,8,11). It is important to note that investigators in only a few studies have assessed the repeatability and variability of 1H MR spectroscopic IMCL quantifications in healthy subjects. Technical factors, data analysis methods, and an individual's physical activity, diet, and medication use may cause variations in 1H MR spectroscopic IMCL quantifications and thus potentially affect the sensitivity of these measurements in the detection of changes during longitudinal studies. Thus, the purpose of our study was to prospectively determine the repeatability and variability of tibialis anterior IMCL quantifications performed by using 1.5-T 1H MR spectroscopy in healthy subjects.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Subject Selection
Between November 2002 and June 2003, a total of 27 healthy subjects were recruited by means of an advertisement posted at our institution's clinical research Web site. Twelve men and 15 women (mean age, 25 years; age range, 19–48 years) were recruited and required to complete a five-item questionnaire that was developed specifically for this study. The included questions concerned the subjects' current medication regimens, medical and surgical histories, nutritional patterns, and physical activity levels. Individuals who had a history of endocrine and/or metabolic disease (including but not limited to type 2 diabetes, hyperlipidemia, Cushing disease, hypo- or hyperthyroidism, and polycystic ovarian syndrome); were obese (body mass index > 30); were known to be pregnant; were undergoing treatment involving medication use; were taking hormones (other than birth control pills), steroids, or dietary supplements; were undergoing an exercise training program; and/or were elite athletes (ie, running more than 50 miles a week) were excluded from the study.
We performed screening for asymptomatic hyperglycemia and/or hyperlipidemia in all subjects by determining their fasting serum glucose and lipid profile levels. Individuals with a fasting glucose level higher than 110 mg/dL (6.11 mmol/L), a total cholesterol level higher than 200 mg/dL (5.17 mmol/L), a triglycerides level higher than 150 mg/dL (1.69 mmol/L), and/or a low-density lipoprotein cholesterol level higher than 130 mg/dL (3.36 mmol/L) were excluded from the study. All parameters were determined by using previously published methods (12).
The mean interval between the blood tests and the first 1H MR spectroscopic examination was 21 days (range, 3–57 days). All subjects had fasting glucose and lipid profile levels within the normal limits. The mean body mass index was 23.0 ± 0.5 (standard error of mean), with no significant difference between the male and female subjects (P = .2, 
2 analysis). For practical reasons, the subjects were encouraged to undergo both MR spectroscopic examinations within a 2-week interval. The mean interval between the first and second visits for 1H MR spectroscopy was 12 days (range, 4–56 days).
1H MR Spectroscopic Technique
Data collection was monitored by either of two experienced musculoskeletal radiologists with 4 (M.T.) and 5 (D.I.R.) years of experience with spectroscopic techniques. MR spectroscopy was performed in the 27 subjects who were considered to be eligible for the study by using a 1.5-T system (Signa LX, software version 8.3; GE Medical Systems, Milwaukee, Wis). After the subjects had fasted for 8 hours overnight, 1H MR spectroscopy of the anterior tibial muscle was performed between 7:00 and 8:00 AM and repeated up to 8 weeks later during a second visit. During each visit for MR spectroscopy, localizer and 1H MR spectroscopic images were obtained before and after the subject was repositioned in the magnet bore (ie, with complete removal of the lower extremity from the coil). All subjects were instructed to maintain their usual physical activities and dietary habits for the duration of the study but to avoid excessive physical effort (ie, moderate or vigorous exercise) and a high-fat diet 72 hours before undergoing imaging.
Each subject was positioned feet first in the magnet bore, and the right calf was placed in a standard radiofrequency transmit-receive extremity coil. A triplanar gradient-echo localizer pulse sequence involving an echo time of 1.6 msec and a repetition time of 49.0 msec was performed. Transverse T1-weighted MR images (600/14 [repetition time msec/echo time msec]; section thickness, 4 mm; intersection gap, 1 mm; matrix, 128 x 128; one signal acquired; field of view, 16 cm) of the proximal two-thirds of the calf were obtained by using the proximal tip of the fibula as the osseous landmark. The first section was always obtained at the level of the proximal fibular tip, and the prescription stack was propagated distally by using the section thickness and spacing parameters just described.
Single-voxel MR spectroscopic data were acquired by using a point-resolved spatially localized spectroscopic pulse sequence with 3000/25, 32 acquisitions, and eight acquired signals. In all cases, a 15 x 15 x 15-mm (3.4-mL) voxel was placed on the transverse T1-weighted section showing the largest cross-sectional area of the muscle, with visible interstitial tissue, fat, and/or vessels avoided. Oblique voxels and spatial presaturation bands were not used. For each voxel placement, automated optimization of gradient shimming, water suppression, and transmit-receive gain, followed by manual adjustment of the central frequency, was performed, and water line widths of 10–12 Hz were obtained. Water presaturation was used for the metabolite spectrum acquisitions, and unsuppressed water spectra of the same voxel were obtained for each MR spectroscopic examination. Neither the metabolite levels nor the unsuppressed water levels were corrected for T1 and T2 relaxation times.
To ensure consistent positioning of the patient during the follow-up examination, the transverse section used for voxel placement (counted from the proximal fibular tip) was annotated in a log book and screen captured with voxel overlays and x-y coordinates. The acquisition time per spectrum was about 3 minutes, and each imaging session lasted approximately 45 minutes. All 27 subjects completed the entire MR spectroscopic protocol.
1H MR Spectroscopic Data Analysis
All 1H MR spectroscopic data were fitted (by M.T., B.J.T., and M.E.J.) by using two distinct procedures (Figure). In one procedure, Java-based MR user interface (jMRUI; developed by A. van den Boogaart, Katholieke Universiteit Leuven, Leuven, Belgium) software (14) was used for analysis in the time domain directly on free-induction decays. The spectra were filtered for removal of residual water by using the Hankel-Lanczos single-variable decomposition method and apodized with a 1.25-Hz Gaussian function. Metabolite signals were analyzed by using the Advanced Magnetic Resonance (AMARES) fitting algorithm within jMRUI (15). Six resonances were described with the assumption of Gaussian line shapes: (a) the IMCL methyl protons peak at a resonance of 0.9 ppm, (b) the extramyocellular lipid (EMCL) methyl protons peak at a resonance of 1.1 ppm, (c) the IMCL methylene protons peak at a resonance of 1.3 ppm, (d) the EMCL methylene protons peak at a resonance of 1.5 ppm, (e) the total creatine (ie, free creatine plus phosphocreatine) methyl peak at a resonance of approximately 3.0 ppm, and (f) the trimethylamines (ie, choline plus carnitine) peak at a resonance of 3.2 ppm. We incorporated prior knowledge into the fitting algorithm by using previously published criteria (11). The zero-order phase correction was estimated by using the AMARES algorithm, and the first-order phase correction was fixed to zero. Unsuppressed water spectra obtained from the same voxel were used as the internal reference for the relative quantification of the metabolite resonances. All nonwater resonances were removed from the unsuppressed free-induction decays by using the Hankel-Lanczos single-variable decomposition method. The water signal peak at 4.7 ppm was described by using one Gaussian line shape and analyzed by using the AMARES algorithm.
|
|
Only the data for the IMCL methylene protons peak at 1.3 ppm and the EMCL methylene protons peak at 1.5 ppm were used for statistical analyses. Both the jMRUI- and the LCModel-derived lipid content estimates were scaled to the unsuppressed water peak (IMCLW for IMCL methylene protons peak at 1.3 ppm scaled to the unsuppressed water peak; EMCLW for EMCL methylene protons peak at 1.5 ppm scaled to the unsuppressed water peak) and to the creatine peak at approximately 3.0 ppm (IMCLCR for IMCL methylene protons peak at 1.3 ppm scaled to the creatine peak; EMCLCR for EMCL methylene protons peak at 1.5 ppm scaled to the creatine peak) and were expressed in institutional units. Spectra in which the jMRUI model yielded poor residuals or did not facilitate reliable identification of the IMCL methylene proton peak were excluded from statistical analysis and LCModel data fitting.
Statistical Analyses
Mean lipid quantification values, standard deviations, and coefficients of variation (CVs) were obtained for each subject and each visit (for MR spectroscopic examination). To estimate the variability caused by different sources (equipment, repositioning, analysis error, and biologic variation), we converted the data obtained from multiple observations per subject to a percentage difference from each mean value to generate values with constant variance and analyzed these data by using a one-way analysis of variance random-effects model (variance components analysis). The random-effects model was used to determine estimates of intraday variability—that is, the measurement variability attributable to subject repositioning and equipment performance between images obtained on the same day—and intervisit variability—that is, the measurement variability attributable to biologic change, subject repositioning, and equipment performance between images obtained on different days.
Power analysis was performed to determine the effect of variance component measurements on study design. The intraday variance component and the total variance component measurements were used to determine the detectable percentage change in lipid measurements for different sample sizes. The mean IMCL concentrations measured during the first and second visits (for 1H MR spectroscopy) were compared by using 2 analysis. P < .05 indicated statistical significance. Statistical analyses were performed by using JMP Statistical Database software (SAS Institute, Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The intraday variability and the intervisit variability of all calculated metabolite ratios are listed in Table 1. The random-effects variance components model also yielded estimated values of the sources of variability in our study. With jMRUI data fitting, nonbiologic factors (ie, repositioning, equipment stability, and data analysis method) accounted for a mean of 50% (IMCLW, 46%; IMCLCR, 54%) of the total variability of IMCL measurements.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Our study results show that reproducible lipid ratio measurements can be obtained with single-voxel 1H MR spectroscopy of the tibialis anterior by examining a large cohort of healthy individuals. Our data also indicate that jMRUI quantification of IMCL scaled to unsuppressed water in the tibialis anterior yields acceptable CVs. Compared with the jMRUI algorithm, the LCModel algorithm yielded higher intravisit and total CVs for IMCLW measurements and lower CVs for IMCLCR measurements. These data represent estimates of detectable differences in serial 1H MR spectroscopic measurements in a group of individuals and indicate that better indexes of intravisit variability can be obtained by using the jMRUI model. In a sample size of 20 subjects, a 10% change in IMCLW due to subject repositioning would be confidently detected on 1H MR spectroscopic images obtained on the same day and a 15% change in IMCLW would be confidently detected on images obtained on different days. These results suggest that 1H MR spectroscopy may be sensitive in most examinations of ICML content variations.
Previous studies have revealed elevations in IMCL concentration above the threshold for a detectable change in IMCLW described by using jMRUI in our study. In exercise studies (3,17), same-day changes in IMCL of 20%–26% have been observed. Compared with the IMCL concentrations measured in healthy subjects, 57%–84% increases in IMCL levels in patients who were insulin resistant (6), 130%–140% increases in patients with human immunodeficiency virus–related lipodystrophy (18), and 210% increases in obese patients (2) have been demonstrated. Our results represent initial parameters by which study designs can be optimized for the detection of IMCL changes during exercise or pharmacologic interventions.
The random-effects model revealed that with use of the jMRUI algorithm, 50% of the total variability in IMCL measurements could be attributed to nonbiologic factors (eg, instrumental instability and voxel positioning). Because these factors are amenable to technical optimization, an increase in their proportion is desirable and can be achieved with strict control of the imaging parameters.
Our 1H MR spectroscopic protocol was geared toward clinical implementation, and we used a reproducible voxel localization technique to minimize imperfect relocalization as a substantial source of measurement variability. Small changes in voxel position should not cause a considerable variation in IMCL signal intensity on same-day images because the signal is derived from homogeneously distributed protons. On the other hand, EMCL protons are distributed heterogeneously and the reproducibility of IMCL methylene resonance measurements is highly affected by the angular dependence of adjacent EMCL resonances. A voxel displacement of a few millimeters can change the EMCL signal amplitude by an order of magnitude, making EMCL variability a sensitive indicator of consistent voxel placement (7) and an important factor in the decreased reliability of IMCL measurements.
Our jMRUI analysis results showed that the total variability of EMCLW (26.7%) was within the previously reported ranges of 10%–50% (19) and 4%–74% (3). This finding suggests that the implementation of a rigorous voxel repositioning protocol, such as that used in our study, can have an important role in ensuring measurement consistency in serial 1H MR spectroscopic examinations.
Metabolite values were calculated as ratios and expressed in institutional units, which are easily implemented and widely described in the literature. We preferred to use this method rather than absolute quantification with phantom calibration, which requires time-consuming and usually impractical measurements of in vivo T1 and T2 relaxation times. Although corrections for relaxation may lead to higher accuracy in metabolite concentration measurements, the use of short echo times helps to minimize the effects of T2 relaxation changes, and long repetition times induce weak saturation, which minimizes the effects of T1 relaxation changes (20). In addition, short-echo-time acquisitions yield better signal-to-noise ratios than do longer-echo-time acquisitions and thus appear to be preferable for quantification of lipid concentrations in muscle tissue.
The variability of metabolite concentrations between imaging visits is of critical importance in follow-up studies performed to examine the effects of pharmacologic, dietary, and exercise interventions. Besides having a nonbiologic component, intervisit variability includes natural biologic variations in metabolite concentrations, which are difficult to control with technical optimization alone. Results of previous studies (7,17,21,22) have shown changes in IMCL concentrations after exercise and the accumulation of IMCL in the muscle of nonexercising individuals during physiologic states of increased plasma-free fatty acids. Efforts to control physiologic changes during longitudinal studies should be introduced to minimize the effect of these changes on IMCL concentration variability.
Our study subjects underwent 1H MR spectroscopy after fasting for 8 hours overnight and were encouraged to maintain stable nutritional and exercise habits throughout their participation in the study but particularly 72 hours before each imaging examination. Prescribed diets and strict exercise protocols may help to decrease the contribution of fluctuating physiologic IMCL concentrations to changes detected during longitudinal studies.
There were several limitations in our study. First, we acquired data from a healthy population without strictly enforcing specific diet and exercise regimens. In a more controlled setting, lower intervisit variability may be achieved as a function of diminished physiologic fluctuation in IMCL concentrations. Therefore, our results serve as references of variability primarily for protocols with a study design similar to ours. Another shortcoming was the use of metabolite ratios. This method does not enable the absolute quantification of IMCL (eg, in millimoles per kilogram), which may yield more realistic data for energy balance calculation and for comparisons between patient groups.
Our study was also limited because it was focused on the variability of IMCL concentrations in the anterior tibial muscle only, and, thus, the results may not directly apply to the other calf muscles. Peak fitting procedures and variability indexes may be affected by differences in muscle fiber angle, fiber composition, and lipid concentrations. On the other hand, the challenge of spectral fitting may be offset by the fact that measurements obtained in muscles with higher IMCL content and slow-twitch oxidative fibers better reflect changes in insulin dynamics (23,24). These observations support the recommendation that the study design be tailored to address specific physiologic hypotheses. Although the tibialis anterior contains a higher proportion of fast-twitch glycolytic fibers, it seems reasonable to recommend that measurements in this muscle be included in longitudinal 1H MR spectroscopic studies owing to the parallel fiber arrangement with optimal lipid peak separation and good repeatability indexes. The addition of 1H MR spectroscopic measurements in slow-twitch muscles (eg, soleus) may yield additional insight with regard to IMCL dynamics with a minimal increase in the total imaging time. Further studies with patient populations may be necessary to determine which calf muscle is the most appropriate for evaluation of specific conditions that affect IMCL concentrations and insulin sensitivity.
In summary, our study results demonstrate that 1H MR spectroscopy of the anterior tibial muscle yields reliable measurements of IMCL content in healthy subjects and that repeatable quantifications can be performed by using this method.
|
FOOTNOTES |
|---|
Abbreviations: CV = coefficient of variation • EMCL = extramyocellular lipid • EMCLCR = EMCL methylene protons peak at 1.5 ppm scaled to the creatine peak • EMCLW = EMCL methylene protons peak at 1.5 ppm scaled to the unsuppressed water peak • IMCL = intramyocellular lipid • IMCLCR = IMCL methylene protons peak at 1.3 ppm scaled to the creatine peak • IMCLW = IMCL methylene protons peak at 1.3 ppm scaled to the unsuppressed water peak
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, M.T.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, M.T., B.J.T., M.E.J.; clinical studies, M.T., B.J.T., M.E.J.; experimental studies, M.T.; statistical analysis, M.T., E.F.H.; manuscript editing, M.T., B.J.T., M.E.J., D.I.R., W.E.P.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. W. van der Meer, S. Hammer, J. W.A. Smit, M. Frolich, J. J. Bax, M. Diamant, L. J. Rijzewijk, A. de Roos, J. A. Romijn, and H. J. Lamb Short-Term Caloric Restriction Induces Accumulation of Myocardial Triglycerides and Decreases Left Ventricular Diastolic Function in Healthy Subjects Diabetes, December 1, 2007; 56(12): 2849 - 2853. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. W. van der Meer, J. Doornbos, S. Kozerke, M. Schar, J. J. Bax, S. Hammer, MSc, J. W. A. Smit, J. A. Romijn, M. Diamant, L. J. Rijzewijk, et al. Metabolic Imaging of Myocardial Triglyceride Content: Reproducibility of 1H MR Spectroscopy with Respiratory Navigator Gating in Volunteers Radiology, October 1, 2007; 245(1): 251 - 257. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Hadigan, J. Liebau, M. Torriani, R. Andersen, and S. Grinspoon Improved Triglycerides and Insulin Sensitivity with 3 Months of Acipimox in Human Immunodeficiency Virus-infected Patients with Hypertriglyceridemia J. Clin. Endocrinol. Metab., November 1, 2006; 91(11): 4438 - 4444. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Torriani, B. J. Thomas, R. B. Barlow, J. Librizzi, S. Dolan, and S. Grinspoon Increased intramyocellular lipid accumulation in HIV-infected women with fat redistribution J Appl Physiol, February 1, 2006; 100(2): 609 - 614. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
