Pe d i a t r i c I m a g i n g • O r i g i n a l R e s e a r c h Ponrartana et al. Water-Fat MRI and Diffusion-Tensor Imaging of Pediatric Lower Extremity Muscles
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Pediatric Imaging Original Research
Repeatability of Chemical-ShiftEncoded Water-Fat MRI and Diffusion-Tensor Imaging in Lower Extremity Muscles in Children Skorn Ponrartana1 Kristine E. Andrade1 Tishya A. L. Wren1,2 Leigh Ramos-Platt 3 Houchun H. Hu1,4 Stefan Bluml1 Vicente Gilsanz1,2 Ponrartana S, Andrade KE, Wren TAL, et al. Keywords: diffusion-tensor imaging, fat quantification, MRI, muscle DOI:10.2214/AJR.13.11081 Received April 13, 2013; accepted after revision August 4, 2013. Support for this study was provided by the Society of Pediatric Radiology Research and Education Foundation (to S. Ponrartana). Partial support for this study was provided by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant R01HD059826 to T. A. L. Wren) by the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health (grant K25DK089731 to H. H. Hu), and by a Zumberge Award from the University of Southern California Office of the Provost and Research (to H. H. Hu). 1 Department of Radiology, Children’s Hospital Los Angeles, MS #81, 4650 Sunset Blvd, Los Angeles, CA 90027. Address correspondence to S. Ponrartana ([email protected]). 2 Department of Orthopaedic Surgery, Children’s Hospital Los Angeles, Los Angeles, CA. 3 Department of Neurology, Children’s Hospital Los Angeles, Los Angeles, CA. 4 Department of Electrical Engineering, Viterbi School of Engineering, University of Southern California, Los Angeles, CA.
WEB This is a web exclusive article. AJR 2014; 202:W567–W573 0361–803X/14/2026–W567 © American Roentgen Ray Society
OBJECTIVE. The purpose of this study was to assess the repeatability of water-fat MRI and diffusion-tensor imaging (DTI) as quantitative biomarkers of pediatric lower extremity skeletal muscle. SUBJECTS AND METHODS. MRI at 3 T of a randomly selected thigh and lower leg of seven healthy children was studied using water-fat separation and DTI techniques. Musclefat fraction, apparent diffusion coefficient (ADC), and fractional anisotropy (FA) values were calculated. Test-retest and interrater repeatability were assessed by calculating the Pearson correlation coefficient, intraclass correlation coefficient, and Bland-Altman analysis. RESULTS. Bland-Altman plots show that the mean difference between test-retest and interrater measurements of muscle-fat fraction, ADC, and FA was near 0. The correlation coefficients and intraclass correlation coefficients were all between 0.88 and 0.99 (p < 0.05), suggesting excellent reliability of the measurements. Muscle-fat fraction measurements from water-fat MRI exhibited the highest intraclass correlation coefficient. Interrater agreement was consistently better than test-retest comparisons. CONCLUSION. Water-fat MRI and DTI measurements in lower extremity skeletal muscles are objective repeatable biomarkers in children. This knowledge should aid in the understanding of the number of participants needed in clinical trials when using these determinations as an outcome measure to noninvasively monitor neuromuscular disease.
T
he ability to detect major histologic features associated with muscle injury (i.e., fiber size, number and architecture, fatty infiltration, edema, and fibrosis) using MRI techniques, including chemical-shift-encoded water-fat MRI (hereafter referred to as “water-fat MRI”) and diffusion-tensor imaging (DTI), has the potential to enhance the assessment of neuromuscular disorders [1, 2]. With water-fat MRI, the separation of individual water and fat contributions in each voxel provides muscle-fat fraction measures related to disease severity in neuromuscular disorders [3, 4]. Likewise, the ability of DTI to assess differences in diffusion of water in elongated tissue structure, such as muscle fibers [5], has been proposed as a potential tool to study skeletal muscle. Measurements of diffusion directionality, expressed as fractional anisotropy (FA), or of mean diffusivity expressed as apparent diffusion coefficient (ADC) [6], have been applied to characterize muscle architecture [7], evaluate muscle injury [8] and regeneration [9], and identify the effects of denervation on muscle structure [10].
Children represent a unique challenge for quantitative imaging because of their large variability in size and body mass composition. Moreover, pediatric neuromuscular disorders represent a large and heterogeneous group of diseases that include both abnormalities in the peripheral nerves and abnormalities in muscle tissue. Longitudinal clinical evaluation of these conditions has been hindered by the lack of objective noninvasive measures. The current clinical standard of care does not involve imaging but rather an assessment of muscle strength and flexibility through tests performed by a physician. Although manual muscle tests are simple, safe, and inexpensive, they are dependent on patient effort and are limited by intra- and interexaminer variability [11, 12]. Functional testing (such as the time required to walk a set distance) is also effort dependent and changes with fatigue throughout the day [11, 12]. Measurements of creatine phosphokinase levels can vary tremendously and do not always correlate with disease activity. Finally, muscle biopsies, which have
AJR:202, June 2014 W567
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Ponrartana et al.
A
B
C
Fig. 1—6-year-old healthy girl. A–C, Normal cross-sectional images from mid thigh were obtained using water-fat MRI. Shown are fat-only image (A), water-only image (B), and color-coded muscle fat fraction map (C).
long been recognized as the reference standard to monitor disease and interventions, are invasive and difficult to justify [12]. The noninvasive and objective nature of musclefat fraction and DTI measurements to detect early muscle disease is a significant advantage over current techniques for tracking the progression or regression of subclinical disease over time. Although quantitative fat fraction and diffusion measurements have been extensively applied to the liver and brain, respectively [13–15], the repeatability of these quantitative biomarkers in evaluating muscle specifically in children has not been reported, to our knowledge. Short-term reproducibility of muscle-fat fraction and DTI measurements, which is key for determining the length of and the number of participants in a trial, is influenced by a multitude of factors, including the operator, equipment, and population studied. Even when these factors are controlled, muscle-fat fraction and DTI measurements vary greatly with methods
A W568
of analysis, and available data indicate substantial age-specific differences in measures. The aim of this study was to evaluate the repeatability of quantitative measures of muscle-fat fraction from water-fat MRI and ADC and FA from DTI in pediatric subjects. Both repeatability between tests (test-retest reliability) and repeatability between raters (interrater reliability) were assessed in the lower extremity muscles. Subjects and Methods Study Population The study participants were seven healthy a symptomatic children (five girls and two boys; 8.9 ± 3.5 years old; age range, 5–15 years) recruited from the population surrounding Children’s Hospital Los Angeles. The investigation protocol for this prospective study was compliant with HIPAA and was approved by the hospital’s institutional review board for clinical investigations. Informed consent was obtained from parents or guardians of all the subjects. Possible candidates were asked about their age and medical history.
B
Subjects were excluded if they had a diagnosis of any underlying disease or illness, had been ill for longer than 2 weeks during the previous 6 months, or had been admitted to the hospital at any time during the previous 3 years.
MRI Protocol All MRI scans were performed on a 3-T MRI scanner (Achieva R3.2, Philips Healthcare). For all subjects, a baseline study of a randomly (e.g., left or right) selected thigh and lower leg was obtained; thereafter, the subject was taken off the table and repositioned, and the same lower extremity was rescanned to obtain test-retest measurements. A 16-channel torso array was used for imaging. Water-fat MRI was performed using the multiecho mDIXON (Philips Healthcare) pulse sequence [16]. Imaging parameters and setup for the 3D Cartesian spoiled gradient-echo sequence were as follows: supine axial acquisition; TR/TE, 10/1.48; echo spacing, 1.2 ms; number of TEs, six; nonflyback (bipolar) readout gradients, 170 slices with 1-mm isotropic spatial resolution; FOV, 140 mm × 140 mm; flip angle, 3°; bandwidth, 1.3 kHz/
Fig. 2—6-year-old healthy girl. A and B, Normal cross-sectional and tractographic images from mid thigh were obtained using diffusion-tensor imaging. Shown are color fractional anisotropy map image (A) and representative colorcoded tractographic image (B) of rectus femoris (red) and semitendinosus (green) muscles. A = anterior, F = foot, H = head, L = left, P = posterior, R = right.
AJR:202, June 2014
Water-Fat MRI and Diffusion-Tensor Imaging of Pediatric Lower Extremity Muscles TABLE 1: Characteristics of the Study Group Total Apparent Diffusion Coefficient (× 10 −3 mm2 /s)
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Total Muscle-Fat Fraction (%) Patient No. Age (y) 1
15
Total Fractional Anisotropy
Sex
Rater 1
Rater 2
Rater 1
Rater 2
Rater 1
Rater 2
Female
8
8
1.57
1.56
0.31
0.33
2
8
Female
3
3
1.54
1.54
0.36
0.36
3
6
Female
4
4
1.52
1.49
0.35
0.38
4
12
Male
6
6
1.54
1.54
0.35
0.34
5
7
Male
4
4
1.57
1.57
0.34
0.36
6
5
Female
3
3
1.59
1.59
0.37
0.36
7
9
Female
5
5
1.47
1.44
0.40
0.37
Note—All data are averages of the values of each muscle of the lower extremity.
pixel; SENSE acceleration, 2; and number of signals averaged, one. The average scan time for the mDIXON sequence was approximately 2 minutes. The DTI sequence used a multislice spinecho single-shot echo-planar imaging sequence covering the same anatomic volume of the mDIXON scan. Imaging parameters were as follows: TR/TE, 2.479/43; partial Fourier, 70%; adiabatic fat suppression using inversion recovery; 1.5 mm in-plane resolution; 6-mm slices; 15 diffusion directions plus one baseline; b = 0, 250, and 500 s/ mm 2 ; and SENSE acceleration, 2. The scan time was approximately 6 minutes.
Data Processing Two radiologists independently reviewed the initial and repeat MRI datasets. For both datasets, each rater was instructed to select image slices that provided maximal transverse cross-sectional area of each muscle to be measured. Regions of interest (ROIs) were drawn circumferentially around the cross-sectional area of the muscle studied using a freehand technique, with instructions to avoid the edge of the muscle to limit partial volume averaging. The following muscles were evaluated: thigh (gluteus maximus, rectus femoris, vastus medialis, lateralis, and intermedius, semimembranosus,
TABLE 2: Mean Muscle-Fat Fraction, Apparent Diffusion Coefficient, and Fractional Anisotropy Values by Muscle Mean Muscle-Fat Fraction (%)
Mean Apparent Diffusion Coefficient (× 10 −3 mm2 /s)
Rater 1
Rater 2
Rater 1
Gluteus maximus
7
7
1.39
1.42
0.42
0.42
Rectus femoris
4
4
1.53
1.49
0.40
0.40
Vastus lateralis
4
3
1.52
1.51
0.38
0.37
Vastus medialis
4
4
1.65
1.61
0.35
0.34
Vastus intermedius
4
4
1.62
1.60
0.35
0.35
Biceps femoris
6
5
1.52
1.50
0.35
0.35
Semitendinosus
6
6
1.57
1.51
0.35
0.36
Semimembranosus
6
6
1.49
1.46
0.35
0.36
Adductorsa
5
5
1.62
1.59
0.32
0.32
Anterior tibialis
5
4
1.51
1.50
0.39
0.39
Posterior tibialis
5
4
1.78
1.75
0.34
0.34
Peroneus longus
6
5
1.47
1.46
0.37
0.37
Soleus
5
4
1.55
1.52
0.28
0.29
Gastrocnemius
5
5
1.39
1.38
0.33
0.33
Muscle
Rater 2
Mean Fractional Anisotropy Rater 1
Rater 2
aValues for adductor longus and adductor magnus were combined because of difficulty separating the muscles
on imaging.
semitendinosus, biceps femoris, and combined adductors) and lower leg (anterior and posterior tibialis, peroneus longus, gastrocnemius, and soleus). The mDIXON pulse sequence and reconstruction technique used was a six-echo generalization of the traditional in- and opposed-phase two-echo Dixon water-fat imaging technique. Image reconstructions yielded coregistered fat, water, and quantitative fat fraction images (Fig. 1). Fat fraction maps accurately reflect the underlying proton-density ratios between fat and the sum of water and fat on a range of 0–100%, when using a multipeak spectral model of fat and a small excitation flip angle [17]. The multiecho water-fat algorithms can jointly estimate fat fraction and T2* per voxel. Average muscle-fat fraction values were measured for the ROI in each of the muscles using imaging software (OsiriX, Pixmeo). DTI datasets were analyzed on an offline workstation using commercially available processing software (FiberTrak, Philips Healthcare). DTI studies yielded color FA maps, which were coregistered with mDIXON images for anatomic reference, and fiber tracking was performed with a single ROI line propagation technique (Fig. 2). Tracking was launched from a seed ROI from which a line was propagated in both the retrograde and anterograde directions according to the main eigenvector at each voxel. The following tractography parameters were chosen in the software: FA threshold, 0.12; direction threshold, 6.75°. Once fiber tracking was performed, mean ADC and FA values were automatically generated by the software.
Statistical Analysis Statistical analyses were performed using Stata software (version 9, StataCorp). Values for age, muscle-fat fraction, ADC, and FA (for the first and second measurements) are expressed as the mean ± SD. The test-retest and interrater repeatability of muscle-fat fraction, ADC, and FA were assessed by calculating the Pearson correlation coefficient (r), intraclass correlation coefficient (ICC), and Bland-Altman analysis [18]. The criteria of Portney and Watkins [19] were used to judge the strength of the correlation coefficients, as follows: little to no relationship (r ≤ 0.25), fair degree of relationship (r = 0.26–0.50), moderate-to-good relationship (r = 0.51–0.75), and good-to-excellent relationship (r ≥ 0.76). ICC results were interpreted according to the guidelines by Fleiss [20] as excellent relationship (r > 0.75), fair-to-good relationship (r = 0.40–0.75), and poor relationship (r < 0.40).
Results Table 1 describes the age and sex of each subject and the total muscle-fat fraction, ADC, and FA values of each lower
AJR:202, June 2014 W569
Ponrartana et al.
1.96 SD
0.2
0.20
Difference (× 10-3 mm2/s)
Difference (× 10-3 mm2/s)
0.3
0.1 Mean
0.0
0.01
–0.1 –1.96 SD –0.2
–0.18
–0.3
0.2
1.96 SD 0.16
0.1
Mean 0.0
0.02 –1.96 SD
–0.1
–0.11
–0.2 –0.3
1.0
1.2
1.4
1.6
1.8
2.0
2.2
1.0
1.2
Mean ADC (× 10-3 mm2/s)
1.4
1.6
1.8
2.0
2.2
Mean ADC (× 10-3 mm2/s)
A 0.2
0.1
Difference
Difference
0.2
1.96 SD 0.07 Mean
0.0
0.00
0.1 1.96 SD 0.05 Mean
0.0
0.00 –1.96 SD
–1.96 SD
–0.05
–0.07
–0.1 0.2
0.3
0.4
–0.1
0.5
0.2
0.3
0.4
Mean FA
0.6
0.5
Mean FA
B 4
2.5 2.0
3 1.96 SD 2
2.2
1 Mean
0
0.1
–1 –2
1.39
1.0 0.5
Mean
0.0
0.12
–0.5
–1.96 SD
–1.0
–1.9
–1.5
–3
1.96 SD
1.5 Difference (%)
Difference (%)
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
0.3
–1.96 SD –1.15
–2.0 0
2
4
6
8
10
12
Mean Muscle-Fat Fraction (%)
14
16
0
2
4
6
8
10
14
16
Mean Muscle-Fat Fraction (%)
C Fig. 3—Bland-Altman plots of test-retest (left column) and interrater (right column) measurements. A–C, Graphs show mean apparent diffusion coefficient (ADC) (A), fractional anisotropy (FA) (B), and muscle-fat fraction (C). Each symbol represents different subject, and each data point represents mean value from each individual muscle measured.
W570
AJR:202, June 2014
Water-Fat MRI and Diffusion-Tensor Imaging of Pediatric Lower Extremity Muscles
Muscle-Fat Fraction Test 2 (%)
Muscle-Fat Fraction Rater 2 (%)
8 r = 0.96; p < 0.0001 ICC = 0.98
7
6
5
4
3
r = 0.89; p < 0.0001 ICC = 0.88
7
6
5
4
3 3
4
5
6
7
8
3
4
5
Muscle-Fat Fraction Rater 1 (%)
6
7
8
Muscle-Fat Fraction Test 1 (%)
A 1.8
1.8 r = 0.98; p < 0.0001 ICC = 0.94
r = 0.94; p < 0.0001 ICC = 0.75
1.7 ADC Test 2
ADC Rater 2
1.7
1.6
1.5
1.6
1.5
1.4
1.4
1.3
1.3 1.3
1.4
1.5
1.6
1.7
1.8
1.3
1.4
1.5
ADC Rater 1
1.6
1.7
1.8
ADC Test 1
B 0.44
0.44 r = 0.99; p < 0.0001 ICC = 0.93
r = 0.88; p < 0.0001 ICC = 0.73
0.40 FA Test 2
0.40 FA Rater 2
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
8
0.36
0.32
0.36
0.32
0.28
0.28 0.28
0.32
0.36 FA Rater 1
0.40
0.44
0.28
0.32
0.36
0.40
0.44
FA Test 1
C Fig. 4—Scatter plots of correlation between interrater (right column) and test-retest (left column) measurements. A–C, Graphs show mean apparent diffusion coefficient (ADC) (A), fractional anisotropy (FA) (B), and muscle-fat fraction (C), with r and intraclass correlation of coefficient (ICC) statistics included in each graph.
AJR:202, June 2014 W571
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Ponrartana et al. extremity. Total muscle-fat fraction, ADC, and FA values represent the average of the two readings for the 14 individual muscles studied at baseline and follow-up. Values for each muscle are shown in Table 2. Figure 3 depicts the Bland-Altman-plots showing differences between test-retest and interrater measurements of muscle-fat fraction, ADC, and FA for the individual muscles in each subject. The largest difference in muscle-fat fraction measurements between the two raters was 2.1% (mean, 0.5%), compared with 3.7% (mean, 0.8%) between the two tests. For ADC, the largest differences between the two raters and the two tests were 0.11 × 10–3 mm2/s (mean, 0.02 × 10–3 mm2/s) and 0.11 × 10–3 mm2/s (mean, 0.03 × 10–3 mm2/s), respectively. Lastly, the largest differences in FA measurements between the two raters were 0.27 (mean, 0.05) and 0.29 (mean, 0.07) between the two tests. On average, the mean differences were within 2 SD, indicating the absence of systematic bias. Figure 4 summarizes the test-retest and interrater correlation coefficient and ICC statistics for muscle-fat fraction, ADC, and FA measures. Interrater repeatability was excellent, with the correlation and ICC between raters very high for mean musclefat fraction (r = 0.96; ICC = 0.96), ADC (r = 0.98; ICC = 0.94), and FA (r = 0.99; ICC = 0.93). Test-retest repeatability was also mostly excellent for mean musclefat fraction (r = 0.89; ICC = 0.88), ADC (r = 0.94; ICC = 0.75), and FA (r = 0.88) with the exception of the ICC for FA, which showed fair-to-good agreement (ICC = 0.73). Interrater agreement was consistently better than test-retest comparisons (p = 0.04). A significant positive correlation between muscle-fat fraction measures and age was observed, with older subjects showing higher muscle-fat fraction throughout the lower extremities (r = 0.92; p < 0.01) (Table 3). The muscle-fat fractions of all the muscles were significantly correlated with age with the exception of the posterior tibialis muscle (p = 0.08) and gastrocnemius muscle (p = 0.10); the gluteus maximus muscle had the strongest correlation (r = 0.96), whereas the gastrocnemius muscle had the weakest (r = 0.66). Although the correlation of mean FA with age tended to be negative, this correlation was weak and not statistically significant when evaluated by individual muscles. No significant association was seen between mean or individual muscle ADC measures and age.
W572
Discussion Our study was designed to determine the variability and repeatability of waterfat MRI and DTI measurements of skeletal muscle architecture and composition in healthy children. We found that measures of muscle fat by water-fat MRI and of muscle diffusivity (ADC) and directionality of diffusion (FA) by DTI are highly repeatable and can potentially serve as objective biomarkers of disease progression in pediatric neuromuscular disorders. Greater agreement was observed between raters than between tests, suggesting that technical factors, such as patient motion, scan positioning, and magnetic field homogeneity, are a larger source of variability than observer bias in ROI selection. Among the different MRI measures, muscle-fat fraction had the highest interrater and test-retest agreement, which may indicate that water-fat MRI is less prone to sources of variability than DTI. MRI measures of skeletal muscle architecture and composition can be useful in the pediatric population, most notably in monitoring disease progression longitudinally in neuromuscular disorders such as Duchenne muscular dystrophy [21, 22]. Currently, the standard for clinical evaluation of muscle function is manual muscle and functional testing. However, these tests provide only a semiquantitative assessment that can be dependent on the individual performing the evaluation and the effort exerted by the person being tested. Not only is MRI more objective and quantitative, MRI has a greater sensitivity to small changes and can evaluate individual muscles within muscle function groups. In addition to being highly repeatable in the pediatric population, water-fat MRI and DTI are also advantageous for this age group because they are relatively fast scans that can be easily performed in awake children as young as 5 years old. As a result, these repeatable quantitative MRI measures may enable better patient monitoring, titration of therapy, and functional prognosis in patients with neuromuscular disorders, possibly facilitating future clinical trials. Although our study was not powered to assess the normal changes in muscle composition with growth and development, we unexpectedly found a significant positive association between muscle-fat fraction and age, which suggests increasing fat infiltration throughout life. Although ample data indicate that adipose tissue deposition occurs with aging and with neuromuscular disease [11, 23,
TABLE 3: Correlation Coefficients of Muscle-Fat Fraction and Age by Muscle Muscle
r
Gluteus maximus
0.96
Rectus femoris
0.74
Vastus lateralis
0.87
Vastus medialis
0.92
Vastus intermedius
0.86
Biceps femoris
0.83
Semitendinosus
0.79
Semimembranosus
0.88
Adductors
0.72
Anterior tibialis
0.82
Posterior tibialisa
0.71
Peroneus longus
0.74
Soleus
0.80
Gastrocnemiusa
0.66
ap > 0.05.
24], a positive association between age and skeletal muscle fat has not previously been shown in healthy children. We also found that, in our small cohort of children, there is no significant relationship between age and FA or ADC. These results are consistent with previous adult studies on ADC and FA values for leg musculature [6, 8, 25–27]. This study has several notable limitations. First, the number of subjects examined was small. Although we included an age range that encompassed pre- and peripubertal stages, the number of subjects was not powered to detect small differences. Further studies are needed to confirm these MRI metrics in a larger pediatric cohort to determine normal muscle-fat fraction, ADC, and FA values in skeletal muscle and to confirm differences in muscle-fat fraction with age reported in this study. Second, intrarater repeatability of the same examination was not studied. However, variability between positioning (test-retest) and between raters (interrater) are more clinically relevant causes of variation. The waterfat MRI technique is also limited in its ability to detect intramyocellular lipid, because water-fat MRI identifies any fat signal within the sampled voxel, whether intracellular or extracellular. In this work, we did not also use MR spectroscopy to investigate intramyocellular lipid because it would have required significantly longer scan sessions. Finally, possible confounding variables of muscle fat infiltration and architecture, such as body mass in-
AJR:202, June 2014
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Water-Fat MRI and Diffusion-Tensor Imaging of Pediatric Lower Extremity Muscles dex, pubertal stage, nutrition, and degree of physical activity, were not included in our secondary analysis. Prospective longitudinal analysis is needed to clearly establish any association between age and muscle histology. In conclusion, we showed excellent repeatability of muscle-fat fraction, ADC, and FA values, with muscle-fat fraction having the best agreement between different raters and repeated tests. In addition, our study revealed a statistically significant association between age and muscle fat infiltration. Larger pediatric cohorts are needed to corroborate our findings and to determine normal muscular muscle-fat fraction and ADC and FA values during growth and development. Nevertheless, by providing repeatable objective measurements reflective of muscle architecture and composition, MRI offers the potential for use as a biomarker of disease severity and treatment response. Acknowledgments We thank Patricia Aggabao, Mercy Landaverde, and Lisa Villanueva for their assistance with this study. References 1. Wren TAL, Bluml S, Tseng-Ong L, Gilsanz V. Three-point technique of fat quantification of muscle tissue as a marker of disease progression in Duchenne muscular dystrophy: preliminary study. AJR 2008; 190:[web]W8–W12 2. Gaeta M, Scribano E, Mileto A, et al. Muscle fat fraction in neuromuscular disorders: dual-echo dual-flip-angle spoiled gradient-recalled MR imaging technique for quantification—a feasibility study. Radiology 2011; 259:487–494 3. Dock W, Happak W, Grabenwoger F, Toifl K, Bittner R, Gruber H. Neuromuscular diseases: evaluation with high-frequency sonography. Radiology 1990; 177:825–828 4. Dixon WT. Simple proton spectroscopic imaging. Radiology 1984; 153:189–194
5. Basser PJ, Jones DK. Diffusion-tensor MRI: theory, experimental design and data analysis—a technical review. NMR Biomed 2002; 15:456–467 6. Sinha S, Sinha U, Edgerton VR. In vivo diffusion tensor imaging of the human calf muscle. J Magn Reson Imaging 2006; 24:182–190 7. Heemskerk AM, Strijkers GJ, Vilanova A, Drost MR, Nicolay K. Determination of mouse skeletal muscle architecture using three-dimensional diffusion tensor imaging. Magn Reson Med 2005; 53:1333–1340 8. Zaraiskaya T, Kumbhare D, Noseworthy MD. Diffusion tensor imaging in evaluation of human skeletal muscle injury. J Magn Reson Imaging 2006; 24:402–408 9. Heemskerk AM, Damon DM. Diffusion tensor MRI assessment of human skeletal muscle architecture. Curr Med Imaging Rev 2007; 3:152–160 10. Zhang J, Zhang G, Morrison B, Mori S, Sheikh KA. Magnetic resonance imaging of mouse skeletal muscle to measure denervation atrophy. Exp Neurol 2008; 212:448–457 11. Marden FA, Connolly AM, Siegel MJ, Rubin DA. Compositional analysis of muscle in boys with Duchenne muscular dystrophy using MR imaging. Skeletal Radiol 2005; 34:140–148 12. Scott OM, Hyde SA, Goddard C, Dubowitz V. Quantitation of muscle function in children: a prospective study in Duchenne muscular dystrophy. Muscle Nerve 1982; 5:291–301 13. Brander A, Kataja A, Saastamoinen A, et al. Diffusion tensor imaging of the brain in a healthy adult population: normative values and measurement reproducibility at 3 T and 1.5 T. Acta Radiol 2010; 51:800–807 14. Mashhood A, Railkar R, Yokoo T, et al. Reproducibility of hepatic fat fraction measurement by magnetic resonance imaging. J Magn Reson Imaging 2013; 37:1359–1370 15. Reeder SB, Cruite I, Hamilton G, Sirlin CB. Quantitative assessment of liver fat with magnetic resonance imaging and spectroscopy. J Magn Reson Imaging 2011; 34:spcone 16. Eggers H, Brendel B, Duijndam A, Herigault G.
Dual-echo Dixon imaging with flexible choice of echo times. Magn Reson Med 2011; 65:96–107 17. Bley TA, Wieben O, Francois CJ, Brittain JH, Reeder SB. Fat and water magnetic resonance imaging. J Magn Reson Imaging 2010; 31:4–18 18. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1(8476): 307–310 19. Portney LG, Watkins MP. Foundations of clinical research: application to practice, 2nd ed. Upper Saddle River, NJ: Prentice Hall, 1999 20. Fleiss J. The design and analysis of clinical experiments. New York, NY: Wiley, 1986 21. Hollingsworth KG, Garrood P, Eagle M, Bushby K, Straub V. MR imaging in Duchenne muscular dystrophy: longitudinal assessment of natural history over 18 months. Muscle Nerve 2013; 48:586–588 22. Wokke BH, Bos C, Reijnierse M, et al. Comparison of Dixon and T1-weighted MR methods to assess the degree of fat infiltration in Duchenne muscular dystrophy patients. J Magn Reson Imaging 2013; 38:619–624 23. Beattie KAMN, Ramadan K, Inglis D, Maly MR. Longitudinal changes in intermuscular fat volume and quadriceps muscle volume in the thighs of women with knee osteoarthritis. Arthritis Care Res (Hoboken) 2012; 64:22–29 24. Marcus RL, Addison O, Kidde JP, Dibble LE, Lastayo PC. Skeletal muscle fat infiltration: impact of age, inactivity, and exercise. J Nutr Health Aging 2010; 14:362–366 25. Kermarrec E, Budzik JF, Khalil C, Le Thuc V, Hancart-Destee C, Cotten A. In vivo diffusion tensor imaging and tractography of human thigh muscles in healthy subjects. AJR 2010; 195:[web] W352–W356 26. Galbán CJ, Maderwald S, Uffmann K, Ladd ME. A diffusion tensor imaging analysis of gender differences in water diffusivity within human skeletal muscle. NMR Biomed 2005; 18:489–498 27. Budzik JF, Le Thuc V, Demondion X, Morel M, Chechin D, Cotten A. In vivo MR tractography of thigh muscles using diffusion imaging: initial results. Eur Radiol 2007; 17:3079–3085
AJR:202, June 2014 W573
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Pediatric Imaging Original Research
Repeatability of Chemical-ShiftEncoded Water-Fat MRI and Diffusion-Tensor Imaging in Lower Extremity Muscles in Children Skorn Ponrartana1 Kristine E. Andrade1 Tishya A. L. Wren1,2 Leigh Ramos-Platt 3 Houchun H. Hu1,4 Stefan Bluml1 Vicente Gilsanz1,2 Ponrartana S, Andrade KE, Wren TAL, et al. Keywords: diffusion-tensor imaging, fat quantification, MRI, muscle DOI:10.2214/AJR.13.11081 Received April 13, 2013; accepted after revision August 4, 2013. Support for this study was provided by the Society of Pediatric Radiology Research and Education Foundation (to S. Ponrartana). Partial support for this study was provided by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant R01HD059826 to T. A. L. Wren) by the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health (grant K25DK089731 to H. H. Hu), and by a Zumberge Award from the University of Southern California Office of the Provost and Research (to H. H. Hu). 1 Department of Radiology, Children’s Hospital Los Angeles, MS #81, 4650 Sunset Blvd, Los Angeles, CA 90027. Address correspondence to S. Ponrartana ([email protected]). 2 Department of Orthopaedic Surgery, Children’s Hospital Los Angeles, Los Angeles, CA. 3 Department of Neurology, Children’s Hospital Los Angeles, Los Angeles, CA. 4 Department of Electrical Engineering, Viterbi School of Engineering, University of Southern California, Los Angeles, CA.
WEB This is a web exclusive article. AJR 2014; 202:W567–W573 0361–803X/14/2026–W567 © American Roentgen Ray Society
OBJECTIVE. The purpose of this study was to assess the repeatability of water-fat MRI and diffusion-tensor imaging (DTI) as quantitative biomarkers of pediatric lower extremity skeletal muscle. SUBJECTS AND METHODS. MRI at 3 T of a randomly selected thigh and lower leg of seven healthy children was studied using water-fat separation and DTI techniques. Musclefat fraction, apparent diffusion coefficient (ADC), and fractional anisotropy (FA) values were calculated. Test-retest and interrater repeatability were assessed by calculating the Pearson correlation coefficient, intraclass correlation coefficient, and Bland-Altman analysis. RESULTS. Bland-Altman plots show that the mean difference between test-retest and interrater measurements of muscle-fat fraction, ADC, and FA was near 0. The correlation coefficients and intraclass correlation coefficients were all between 0.88 and 0.99 (p < 0.05), suggesting excellent reliability of the measurements. Muscle-fat fraction measurements from water-fat MRI exhibited the highest intraclass correlation coefficient. Interrater agreement was consistently better than test-retest comparisons. CONCLUSION. Water-fat MRI and DTI measurements in lower extremity skeletal muscles are objective repeatable biomarkers in children. This knowledge should aid in the understanding of the number of participants needed in clinical trials when using these determinations as an outcome measure to noninvasively monitor neuromuscular disease.
T
he ability to detect major histologic features associated with muscle injury (i.e., fiber size, number and architecture, fatty infiltration, edema, and fibrosis) using MRI techniques, including chemical-shift-encoded water-fat MRI (hereafter referred to as “water-fat MRI”) and diffusion-tensor imaging (DTI), has the potential to enhance the assessment of neuromuscular disorders [1, 2]. With water-fat MRI, the separation of individual water and fat contributions in each voxel provides muscle-fat fraction measures related to disease severity in neuromuscular disorders [3, 4]. Likewise, the ability of DTI to assess differences in diffusion of water in elongated tissue structure, such as muscle fibers [5], has been proposed as a potential tool to study skeletal muscle. Measurements of diffusion directionality, expressed as fractional anisotropy (FA), or of mean diffusivity expressed as apparent diffusion coefficient (ADC) [6], have been applied to characterize muscle architecture [7], evaluate muscle injury [8] and regeneration [9], and identify the effects of denervation on muscle structure [10].
Children represent a unique challenge for quantitative imaging because of their large variability in size and body mass composition. Moreover, pediatric neuromuscular disorders represent a large and heterogeneous group of diseases that include both abnormalities in the peripheral nerves and abnormalities in muscle tissue. Longitudinal clinical evaluation of these conditions has been hindered by the lack of objective noninvasive measures. The current clinical standard of care does not involve imaging but rather an assessment of muscle strength and flexibility through tests performed by a physician. Although manual muscle tests are simple, safe, and inexpensive, they are dependent on patient effort and are limited by intra- and interexaminer variability [11, 12]. Functional testing (such as the time required to walk a set distance) is also effort dependent and changes with fatigue throughout the day [11, 12]. Measurements of creatine phosphokinase levels can vary tremendously and do not always correlate with disease activity. Finally, muscle biopsies, which have
AJR:202, June 2014 W567
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Ponrartana et al.
A
B
C
Fig. 1—6-year-old healthy girl. A–C, Normal cross-sectional images from mid thigh were obtained using water-fat MRI. Shown are fat-only image (A), water-only image (B), and color-coded muscle fat fraction map (C).
long been recognized as the reference standard to monitor disease and interventions, are invasive and difficult to justify [12]. The noninvasive and objective nature of musclefat fraction and DTI measurements to detect early muscle disease is a significant advantage over current techniques for tracking the progression or regression of subclinical disease over time. Although quantitative fat fraction and diffusion measurements have been extensively applied to the liver and brain, respectively [13–15], the repeatability of these quantitative biomarkers in evaluating muscle specifically in children has not been reported, to our knowledge. Short-term reproducibility of muscle-fat fraction and DTI measurements, which is key for determining the length of and the number of participants in a trial, is influenced by a multitude of factors, including the operator, equipment, and population studied. Even when these factors are controlled, muscle-fat fraction and DTI measurements vary greatly with methods
A W568
of analysis, and available data indicate substantial age-specific differences in measures. The aim of this study was to evaluate the repeatability of quantitative measures of muscle-fat fraction from water-fat MRI and ADC and FA from DTI in pediatric subjects. Both repeatability between tests (test-retest reliability) and repeatability between raters (interrater reliability) were assessed in the lower extremity muscles. Subjects and Methods Study Population The study participants were seven healthy a symptomatic children (five girls and two boys; 8.9 ± 3.5 years old; age range, 5–15 years) recruited from the population surrounding Children’s Hospital Los Angeles. The investigation protocol for this prospective study was compliant with HIPAA and was approved by the hospital’s institutional review board for clinical investigations. Informed consent was obtained from parents or guardians of all the subjects. Possible candidates were asked about their age and medical history.
B
Subjects were excluded if they had a diagnosis of any underlying disease or illness, had been ill for longer than 2 weeks during the previous 6 months, or had been admitted to the hospital at any time during the previous 3 years.
MRI Protocol All MRI scans were performed on a 3-T MRI scanner (Achieva R3.2, Philips Healthcare). For all subjects, a baseline study of a randomly (e.g., left or right) selected thigh and lower leg was obtained; thereafter, the subject was taken off the table and repositioned, and the same lower extremity was rescanned to obtain test-retest measurements. A 16-channel torso array was used for imaging. Water-fat MRI was performed using the multiecho mDIXON (Philips Healthcare) pulse sequence [16]. Imaging parameters and setup for the 3D Cartesian spoiled gradient-echo sequence were as follows: supine axial acquisition; TR/TE, 10/1.48; echo spacing, 1.2 ms; number of TEs, six; nonflyback (bipolar) readout gradients, 170 slices with 1-mm isotropic spatial resolution; FOV, 140 mm × 140 mm; flip angle, 3°; bandwidth, 1.3 kHz/
Fig. 2—6-year-old healthy girl. A and B, Normal cross-sectional and tractographic images from mid thigh were obtained using diffusion-tensor imaging. Shown are color fractional anisotropy map image (A) and representative colorcoded tractographic image (B) of rectus femoris (red) and semitendinosus (green) muscles. A = anterior, F = foot, H = head, L = left, P = posterior, R = right.
AJR:202, June 2014
Water-Fat MRI and Diffusion-Tensor Imaging of Pediatric Lower Extremity Muscles TABLE 1: Characteristics of the Study Group Total Apparent Diffusion Coefficient (× 10 −3 mm2 /s)
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Total Muscle-Fat Fraction (%) Patient No. Age (y) 1
15
Total Fractional Anisotropy
Sex
Rater 1
Rater 2
Rater 1
Rater 2
Rater 1
Rater 2
Female
8
8
1.57
1.56
0.31
0.33
2
8
Female
3
3
1.54
1.54
0.36
0.36
3
6
Female
4
4
1.52
1.49
0.35
0.38
4
12
Male
6
6
1.54
1.54
0.35
0.34
5
7
Male
4
4
1.57
1.57
0.34
0.36
6
5
Female
3
3
1.59
1.59
0.37
0.36
7
9
Female
5
5
1.47
1.44
0.40
0.37
Note—All data are averages of the values of each muscle of the lower extremity.
pixel; SENSE acceleration, 2; and number of signals averaged, one. The average scan time for the mDIXON sequence was approximately 2 minutes. The DTI sequence used a multislice spinecho single-shot echo-planar imaging sequence covering the same anatomic volume of the mDIXON scan. Imaging parameters were as follows: TR/TE, 2.479/43; partial Fourier, 70%; adiabatic fat suppression using inversion recovery; 1.5 mm in-plane resolution; 6-mm slices; 15 diffusion directions plus one baseline; b = 0, 250, and 500 s/ mm 2 ; and SENSE acceleration, 2. The scan time was approximately 6 minutes.
Data Processing Two radiologists independently reviewed the initial and repeat MRI datasets. For both datasets, each rater was instructed to select image slices that provided maximal transverse cross-sectional area of each muscle to be measured. Regions of interest (ROIs) were drawn circumferentially around the cross-sectional area of the muscle studied using a freehand technique, with instructions to avoid the edge of the muscle to limit partial volume averaging. The following muscles were evaluated: thigh (gluteus maximus, rectus femoris, vastus medialis, lateralis, and intermedius, semimembranosus,
TABLE 2: Mean Muscle-Fat Fraction, Apparent Diffusion Coefficient, and Fractional Anisotropy Values by Muscle Mean Muscle-Fat Fraction (%)
Mean Apparent Diffusion Coefficient (× 10 −3 mm2 /s)
Rater 1
Rater 2
Rater 1
Gluteus maximus
7
7
1.39
1.42
0.42
0.42
Rectus femoris
4
4
1.53
1.49
0.40
0.40
Vastus lateralis
4
3
1.52
1.51
0.38
0.37
Vastus medialis
4
4
1.65
1.61
0.35
0.34
Vastus intermedius
4
4
1.62
1.60
0.35
0.35
Biceps femoris
6
5
1.52
1.50
0.35
0.35
Semitendinosus
6
6
1.57
1.51
0.35
0.36
Semimembranosus
6
6
1.49
1.46
0.35
0.36
Adductorsa
5
5
1.62
1.59
0.32
0.32
Anterior tibialis
5
4
1.51
1.50
0.39
0.39
Posterior tibialis
5
4
1.78
1.75
0.34
0.34
Peroneus longus
6
5
1.47
1.46
0.37
0.37
Soleus
5
4
1.55
1.52
0.28
0.29
Gastrocnemius
5
5
1.39
1.38
0.33
0.33
Muscle
Rater 2
Mean Fractional Anisotropy Rater 1
Rater 2
aValues for adductor longus and adductor magnus were combined because of difficulty separating the muscles
on imaging.
semitendinosus, biceps femoris, and combined adductors) and lower leg (anterior and posterior tibialis, peroneus longus, gastrocnemius, and soleus). The mDIXON pulse sequence and reconstruction technique used was a six-echo generalization of the traditional in- and opposed-phase two-echo Dixon water-fat imaging technique. Image reconstructions yielded coregistered fat, water, and quantitative fat fraction images (Fig. 1). Fat fraction maps accurately reflect the underlying proton-density ratios between fat and the sum of water and fat on a range of 0–100%, when using a multipeak spectral model of fat and a small excitation flip angle [17]. The multiecho water-fat algorithms can jointly estimate fat fraction and T2* per voxel. Average muscle-fat fraction values were measured for the ROI in each of the muscles using imaging software (OsiriX, Pixmeo). DTI datasets were analyzed on an offline workstation using commercially available processing software (FiberTrak, Philips Healthcare). DTI studies yielded color FA maps, which were coregistered with mDIXON images for anatomic reference, and fiber tracking was performed with a single ROI line propagation technique (Fig. 2). Tracking was launched from a seed ROI from which a line was propagated in both the retrograde and anterograde directions according to the main eigenvector at each voxel. The following tractography parameters were chosen in the software: FA threshold, 0.12; direction threshold, 6.75°. Once fiber tracking was performed, mean ADC and FA values were automatically generated by the software.
Statistical Analysis Statistical analyses were performed using Stata software (version 9, StataCorp). Values for age, muscle-fat fraction, ADC, and FA (for the first and second measurements) are expressed as the mean ± SD. The test-retest and interrater repeatability of muscle-fat fraction, ADC, and FA were assessed by calculating the Pearson correlation coefficient (r), intraclass correlation coefficient (ICC), and Bland-Altman analysis [18]. The criteria of Portney and Watkins [19] were used to judge the strength of the correlation coefficients, as follows: little to no relationship (r ≤ 0.25), fair degree of relationship (r = 0.26–0.50), moderate-to-good relationship (r = 0.51–0.75), and good-to-excellent relationship (r ≥ 0.76). ICC results were interpreted according to the guidelines by Fleiss [20] as excellent relationship (r > 0.75), fair-to-good relationship (r = 0.40–0.75), and poor relationship (r < 0.40).
Results Table 1 describes the age and sex of each subject and the total muscle-fat fraction, ADC, and FA values of each lower
AJR:202, June 2014 W569
Ponrartana et al.
1.96 SD
0.2
0.20
Difference (× 10-3 mm2/s)
Difference (× 10-3 mm2/s)
0.3
0.1 Mean
0.0
0.01
–0.1 –1.96 SD –0.2
–0.18
–0.3
0.2
1.96 SD 0.16
0.1
Mean 0.0
0.02 –1.96 SD
–0.1
–0.11
–0.2 –0.3
1.0
1.2
1.4
1.6
1.8
2.0
2.2
1.0
1.2
Mean ADC (× 10-3 mm2/s)
1.4
1.6
1.8
2.0
2.2
Mean ADC (× 10-3 mm2/s)
A 0.2
0.1
Difference
Difference
0.2
1.96 SD 0.07 Mean
0.0
0.00
0.1 1.96 SD 0.05 Mean
0.0
0.00 –1.96 SD
–1.96 SD
–0.05
–0.07
–0.1 0.2
0.3
0.4
–0.1
0.5
0.2
0.3
0.4
Mean FA
0.6
0.5
Mean FA
B 4
2.5 2.0
3 1.96 SD 2
2.2
1 Mean
0
0.1
–1 –2
1.39
1.0 0.5
Mean
0.0
0.12
–0.5
–1.96 SD
–1.0
–1.9
–1.5
–3
1.96 SD
1.5 Difference (%)
Difference (%)
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
0.3
–1.96 SD –1.15
–2.0 0
2
4
6
8
10
12
Mean Muscle-Fat Fraction (%)
14
16
0
2
4
6
8
10
14
16
Mean Muscle-Fat Fraction (%)
C Fig. 3—Bland-Altman plots of test-retest (left column) and interrater (right column) measurements. A–C, Graphs show mean apparent diffusion coefficient (ADC) (A), fractional anisotropy (FA) (B), and muscle-fat fraction (C). Each symbol represents different subject, and each data point represents mean value from each individual muscle measured.
W570
AJR:202, June 2014
Water-Fat MRI and Diffusion-Tensor Imaging of Pediatric Lower Extremity Muscles
Muscle-Fat Fraction Test 2 (%)
Muscle-Fat Fraction Rater 2 (%)
8 r = 0.96; p < 0.0001 ICC = 0.98
7
6
5
4
3
r = 0.89; p < 0.0001 ICC = 0.88
7
6
5
4
3 3
4
5
6
7
8
3
4
5
Muscle-Fat Fraction Rater 1 (%)
6
7
8
Muscle-Fat Fraction Test 1 (%)
A 1.8
1.8 r = 0.98; p < 0.0001 ICC = 0.94
r = 0.94; p < 0.0001 ICC = 0.75
1.7 ADC Test 2
ADC Rater 2
1.7
1.6
1.5
1.6
1.5
1.4
1.4
1.3
1.3 1.3
1.4
1.5
1.6
1.7
1.8
1.3
1.4
1.5
ADC Rater 1
1.6
1.7
1.8
ADC Test 1
B 0.44
0.44 r = 0.99; p < 0.0001 ICC = 0.93
r = 0.88; p < 0.0001 ICC = 0.73
0.40 FA Test 2
0.40 FA Rater 2
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
8
0.36
0.32
0.36
0.32
0.28
0.28 0.28
0.32
0.36 FA Rater 1
0.40
0.44
0.28
0.32
0.36
0.40
0.44
FA Test 1
C Fig. 4—Scatter plots of correlation between interrater (right column) and test-retest (left column) measurements. A–C, Graphs show mean apparent diffusion coefficient (ADC) (A), fractional anisotropy (FA) (B), and muscle-fat fraction (C), with r and intraclass correlation of coefficient (ICC) statistics included in each graph.
AJR:202, June 2014 W571
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Ponrartana et al. extremity. Total muscle-fat fraction, ADC, and FA values represent the average of the two readings for the 14 individual muscles studied at baseline and follow-up. Values for each muscle are shown in Table 2. Figure 3 depicts the Bland-Altman-plots showing differences between test-retest and interrater measurements of muscle-fat fraction, ADC, and FA for the individual muscles in each subject. The largest difference in muscle-fat fraction measurements between the two raters was 2.1% (mean, 0.5%), compared with 3.7% (mean, 0.8%) between the two tests. For ADC, the largest differences between the two raters and the two tests were 0.11 × 10–3 mm2/s (mean, 0.02 × 10–3 mm2/s) and 0.11 × 10–3 mm2/s (mean, 0.03 × 10–3 mm2/s), respectively. Lastly, the largest differences in FA measurements between the two raters were 0.27 (mean, 0.05) and 0.29 (mean, 0.07) between the two tests. On average, the mean differences were within 2 SD, indicating the absence of systematic bias. Figure 4 summarizes the test-retest and interrater correlation coefficient and ICC statistics for muscle-fat fraction, ADC, and FA measures. Interrater repeatability was excellent, with the correlation and ICC between raters very high for mean musclefat fraction (r = 0.96; ICC = 0.96), ADC (r = 0.98; ICC = 0.94), and FA (r = 0.99; ICC = 0.93). Test-retest repeatability was also mostly excellent for mean musclefat fraction (r = 0.89; ICC = 0.88), ADC (r = 0.94; ICC = 0.75), and FA (r = 0.88) with the exception of the ICC for FA, which showed fair-to-good agreement (ICC = 0.73). Interrater agreement was consistently better than test-retest comparisons (p = 0.04). A significant positive correlation between muscle-fat fraction measures and age was observed, with older subjects showing higher muscle-fat fraction throughout the lower extremities (r = 0.92; p < 0.01) (Table 3). The muscle-fat fractions of all the muscles were significantly correlated with age with the exception of the posterior tibialis muscle (p = 0.08) and gastrocnemius muscle (p = 0.10); the gluteus maximus muscle had the strongest correlation (r = 0.96), whereas the gastrocnemius muscle had the weakest (r = 0.66). Although the correlation of mean FA with age tended to be negative, this correlation was weak and not statistically significant when evaluated by individual muscles. No significant association was seen between mean or individual muscle ADC measures and age.
W572
Discussion Our study was designed to determine the variability and repeatability of waterfat MRI and DTI measurements of skeletal muscle architecture and composition in healthy children. We found that measures of muscle fat by water-fat MRI and of muscle diffusivity (ADC) and directionality of diffusion (FA) by DTI are highly repeatable and can potentially serve as objective biomarkers of disease progression in pediatric neuromuscular disorders. Greater agreement was observed between raters than between tests, suggesting that technical factors, such as patient motion, scan positioning, and magnetic field homogeneity, are a larger source of variability than observer bias in ROI selection. Among the different MRI measures, muscle-fat fraction had the highest interrater and test-retest agreement, which may indicate that water-fat MRI is less prone to sources of variability than DTI. MRI measures of skeletal muscle architecture and composition can be useful in the pediatric population, most notably in monitoring disease progression longitudinally in neuromuscular disorders such as Duchenne muscular dystrophy [21, 22]. Currently, the standard for clinical evaluation of muscle function is manual muscle and functional testing. However, these tests provide only a semiquantitative assessment that can be dependent on the individual performing the evaluation and the effort exerted by the person being tested. Not only is MRI more objective and quantitative, MRI has a greater sensitivity to small changes and can evaluate individual muscles within muscle function groups. In addition to being highly repeatable in the pediatric population, water-fat MRI and DTI are also advantageous for this age group because they are relatively fast scans that can be easily performed in awake children as young as 5 years old. As a result, these repeatable quantitative MRI measures may enable better patient monitoring, titration of therapy, and functional prognosis in patients with neuromuscular disorders, possibly facilitating future clinical trials. Although our study was not powered to assess the normal changes in muscle composition with growth and development, we unexpectedly found a significant positive association between muscle-fat fraction and age, which suggests increasing fat infiltration throughout life. Although ample data indicate that adipose tissue deposition occurs with aging and with neuromuscular disease [11, 23,
TABLE 3: Correlation Coefficients of Muscle-Fat Fraction and Age by Muscle Muscle
r
Gluteus maximus
0.96
Rectus femoris
0.74
Vastus lateralis
0.87
Vastus medialis
0.92
Vastus intermedius
0.86
Biceps femoris
0.83
Semitendinosus
0.79
Semimembranosus
0.88
Adductors
0.72
Anterior tibialis
0.82
Posterior tibialisa
0.71
Peroneus longus
0.74
Soleus
0.80
Gastrocnemiusa
0.66
ap > 0.05.
24], a positive association between age and skeletal muscle fat has not previously been shown in healthy children. We also found that, in our small cohort of children, there is no significant relationship between age and FA or ADC. These results are consistent with previous adult studies on ADC and FA values for leg musculature [6, 8, 25–27]. This study has several notable limitations. First, the number of subjects examined was small. Although we included an age range that encompassed pre- and peripubertal stages, the number of subjects was not powered to detect small differences. Further studies are needed to confirm these MRI metrics in a larger pediatric cohort to determine normal muscle-fat fraction, ADC, and FA values in skeletal muscle and to confirm differences in muscle-fat fraction with age reported in this study. Second, intrarater repeatability of the same examination was not studied. However, variability between positioning (test-retest) and between raters (interrater) are more clinically relevant causes of variation. The waterfat MRI technique is also limited in its ability to detect intramyocellular lipid, because water-fat MRI identifies any fat signal within the sampled voxel, whether intracellular or extracellular. In this work, we did not also use MR spectroscopy to investigate intramyocellular lipid because it would have required significantly longer scan sessions. Finally, possible confounding variables of muscle fat infiltration and architecture, such as body mass in-
AJR:202, June 2014
Downloaded from www.ajronline.org by Library Of Medicine on 06/05/14 from IP address 128.103.149.52. Copyright ARRS. For personal use only; all rights reserved
Water-Fat MRI and Diffusion-Tensor Imaging of Pediatric Lower Extremity Muscles dex, pubertal stage, nutrition, and degree of physical activity, were not included in our secondary analysis. Prospective longitudinal analysis is needed to clearly establish any association between age and muscle histology. In conclusion, we showed excellent repeatability of muscle-fat fraction, ADC, and FA values, with muscle-fat fraction having the best agreement between different raters and repeated tests. In addition, our study revealed a statistically significant association between age and muscle fat infiltration. Larger pediatric cohorts are needed to corroborate our findings and to determine normal muscular muscle-fat fraction and ADC and FA values during growth and development. Nevertheless, by providing repeatable objective measurements reflective of muscle architecture and composition, MRI offers the potential for use as a biomarker of disease severity and treatment response. Acknowledgments We thank Patricia Aggabao, Mercy Landaverde, and Lisa Villanueva for their assistance with this study. References 1. Wren TAL, Bluml S, Tseng-Ong L, Gilsanz V. Three-point technique of fat quantification of muscle tissue as a marker of disease progression in Duchenne muscular dystrophy: preliminary study. AJR 2008; 190:[web]W8–W12 2. Gaeta M, Scribano E, Mileto A, et al. Muscle fat fraction in neuromuscular disorders: dual-echo dual-flip-angle spoiled gradient-recalled MR imaging technique for quantification—a feasibility study. Radiology 2011; 259:487–494 3. Dock W, Happak W, Grabenwoger F, Toifl K, Bittner R, Gruber H. Neuromuscular diseases: evaluation with high-frequency sonography. Radiology 1990; 177:825–828 4. Dixon WT. Simple proton spectroscopic imaging. Radiology 1984; 153:189–194
5. Basser PJ, Jones DK. Diffusion-tensor MRI: theory, experimental design and data analysis—a technical review. NMR Biomed 2002; 15:456–467 6. Sinha S, Sinha U, Edgerton VR. In vivo diffusion tensor imaging of the human calf muscle. J Magn Reson Imaging 2006; 24:182–190 7. Heemskerk AM, Strijkers GJ, Vilanova A, Drost MR, Nicolay K. Determination of mouse skeletal muscle architecture using three-dimensional diffusion tensor imaging. Magn Reson Med 2005; 53:1333–1340 8. Zaraiskaya T, Kumbhare D, Noseworthy MD. Diffusion tensor imaging in evaluation of human skeletal muscle injury. J Magn Reson Imaging 2006; 24:402–408 9. Heemskerk AM, Damon DM. Diffusion tensor MRI assessment of human skeletal muscle architecture. Curr Med Imaging Rev 2007; 3:152–160 10. Zhang J, Zhang G, Morrison B, Mori S, Sheikh KA. Magnetic resonance imaging of mouse skeletal muscle to measure denervation atrophy. Exp Neurol 2008; 212:448–457 11. Marden FA, Connolly AM, Siegel MJ, Rubin DA. Compositional analysis of muscle in boys with Duchenne muscular dystrophy using MR imaging. Skeletal Radiol 2005; 34:140–148 12. Scott OM, Hyde SA, Goddard C, Dubowitz V. Quantitation of muscle function in children: a prospective study in Duchenne muscular dystrophy. Muscle Nerve 1982; 5:291–301 13. Brander A, Kataja A, Saastamoinen A, et al. Diffusion tensor imaging of the brain in a healthy adult population: normative values and measurement reproducibility at 3 T and 1.5 T. Acta Radiol 2010; 51:800–807 14. Mashhood A, Railkar R, Yokoo T, et al. Reproducibility of hepatic fat fraction measurement by magnetic resonance imaging. J Magn Reson Imaging 2013; 37:1359–1370 15. Reeder SB, Cruite I, Hamilton G, Sirlin CB. Quantitative assessment of liver fat with magnetic resonance imaging and spectroscopy. J Magn Reson Imaging 2011; 34:spcone 16. Eggers H, Brendel B, Duijndam A, Herigault G.
Dual-echo Dixon imaging with flexible choice of echo times. Magn Reson Med 2011; 65:96–107 17. Bley TA, Wieben O, Francois CJ, Brittain JH, Reeder SB. Fat and water magnetic resonance imaging. J Magn Reson Imaging 2010; 31:4–18 18. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1(8476): 307–310 19. Portney LG, Watkins MP. Foundations of clinical research: application to practice, 2nd ed. Upper Saddle River, NJ: Prentice Hall, 1999 20. Fleiss J. The design and analysis of clinical experiments. New York, NY: Wiley, 1986 21. Hollingsworth KG, Garrood P, Eagle M, Bushby K, Straub V. MR imaging in Duchenne muscular dystrophy: longitudinal assessment of natural history over 18 months. Muscle Nerve 2013; 48:586–588 22. Wokke BH, Bos C, Reijnierse M, et al. Comparison of Dixon and T1-weighted MR methods to assess the degree of fat infiltration in Duchenne muscular dystrophy patients. J Magn Reson Imaging 2013; 38:619–624 23. Beattie KAMN, Ramadan K, Inglis D, Maly MR. Longitudinal changes in intermuscular fat volume and quadriceps muscle volume in the thighs of women with knee osteoarthritis. Arthritis Care Res (Hoboken) 2012; 64:22–29 24. Marcus RL, Addison O, Kidde JP, Dibble LE, Lastayo PC. Skeletal muscle fat infiltration: impact of age, inactivity, and exercise. J Nutr Health Aging 2010; 14:362–366 25. Kermarrec E, Budzik JF, Khalil C, Le Thuc V, Hancart-Destee C, Cotten A. In vivo diffusion tensor imaging and tractography of human thigh muscles in healthy subjects. AJR 2010; 195:[web] W352–W356 26. Galbán CJ, Maderwald S, Uffmann K, Ladd ME. A diffusion tensor imaging analysis of gender differences in water diffusivity within human skeletal muscle. NMR Biomed 2005; 18:489–498 27. Budzik JF, Le Thuc V, Demondion X, Morel M, Chechin D, Cotten A. In vivo MR tractography of thigh muscles using diffusion imaging: initial results. Eur Radiol 2007; 17:3079–3085
AJR:202, June 2014 W573
