Reference Values for Water-Specific T1, Intermuscular and Intramuscular Fat Content in Skeletal Muscle at 2.89 T

Abstract

Background: MRI offers quantification of proton density fat fraction (PDFF) and tissue characteristics with T1 mapping. The influence of age, sex, and the potential confounding effects of fat on T1 values in skeletal muscle in healthy adults are insufficiently known.

Purpose: To determine the accuracy and repeatability of a saturation-recovery chemical-shift encoded multiparametric approach (SR-CSE) for quantification of T1_Water and muscle fat content, and establish normative values (age, sex) from a healthy cohort.

Study type: Prospective observational; phantoms (NiCL₂-agarose T1 phantoms with no fat content; gadolinium T1 phantoms with mixed fat-water content).

Populations: A total of 130 healthy community-dwelling adults (63 male, 18-76 years) free of chronic health conditions that require regular prescription medication, and with no contraindications to MRI.

Field strength/sequence: 2.89 T; gradient echo sequences including saturation-recovery chemical-shift encoded T1 mapping (SR-CSE); MOLLI; SASHA; CSE; and single voxel spectroscopy.

Assessment: SR-CSE provided T1_Water and PDFF maps for assessment of intramuscular (MF_Intra), intermuscular (MF_Inter), and subcutaneous fat and muscle volumes (thigh, paraspinal muscles). Comparison with MOLLI/SASHA T1 mapping.

Statistical tests: Univariable and multivariable linear regression, general linear models, Bland and Altman, coefficient of variation (CV). P-value <0.05 was considered statistically significant.

Results: Phantom and in vivo validation studies showed excellent accuracy of SR-CSE T1_Water and PDFF vs. values from reference standards and repeatability CVs between 0.2% and 2.6% for T1_Water, R2*, MF_Inter, MF_Intra, subcutaneous fat and muscle volumes. Mean T1_Water was 36 msec significantly higher in females (1445 ± 23 msec vs. 1409 ± 22 msec), with no age-effect (P = 0.35). Females had significantly higher values for MF_Inter (10.4% ± 4.8% vs. 7.1% ± 2.9%) and MF_Intra (2.6% ± 1.0% vs. 2.3% ± 0.8%), both of which increased with age, secondary to lower muscle volume. MOLLI and SASHA T1 values had a fat-related bias of 21.7/35.0 msec per 1% increase in fat fraction (MFF_Intra), in vivo, and a constant bias of -319.8/+35.6 msec, respectively.

Data conclusion: SR-CSE provides accurate (vs. phantoms) and repeatable assessment of water-specific T1 values and muscle and fat volumes. Conventional methods (SASHA, MOLLI) have a significant fat-modulated T1-bias. T1_Water values are higher in females with no significant age dependence.

Plain language summary: We developed and tested the accuracy of a new MRI approach to measure tissue damage in skeletal muscle using a method called T1 mapping. The approach also provided matching images of fat within the muscle. We measured T1 values and muscle fat volumes in the thighs of 130 healthy adults to define normal values in healthy people and to understand if these values are influenced by age, sex, or weight. We found that our MRI technique accurately measured T1 values and fat volumes within muscle and we defined normal ranges of values, which were different in healthy males and females.

Level of evidence: 2 TECHNICAL EFFICACY: Stage 1.

Keywords: MOLLI; MRI; PDFF; SASHA; SR‐CSE; T1 mapping; skeletal muscle.

FIGURE 1

Thigh and paraspinal muscle slice prescriptions. (a) Standardized thigh slice locations for all in vivo studies (five slices). (b) Paraspinal muscle slice locations (three slices). (c) Illustrative locations of 30 regions of interest on SR‐CSE, SASHA, and MOLLI T1 maps used for method comparison.

FIGURE 2

Phantom validation of SR‐CSE T1_Water and PDFF. (a) Bland–Altman plot of SR‐CSE T1 values vs. gradient‐echo T1 measurement (GRE) in NiCl₂ phantoms (water phantoms). The SR‐CSE T1_Water map is shown. (b) SR‐CSE proton‐density fat fraction (PDFF) and T1_Water images of agar fat‐water phantoms with increasing fat content. (c) Comparison of PDFF assessed by SR‐CSE and ¹H spectroscopy in the five tubes from (b). (d) Comparison of SR‐CSE T1_Water values in the five tubes with increasing PDFF from (b) with spectroscopic T1 evaluation.

FIGURE 3

Illustrative calculated SR‐CSE images from a male (a–d) and female (e–h) participant: (a) Water and fat separated images (nonsaturation recovery), (b) machine learning segmentation, (c) T1_Water maps and (d) intramuscular fat fraction (MFF_Intra) maps, with matched images in (e–h) for the female participant. T1_Water and MFF_Intra maps include pixels in the muscle segmentation region. One of five acquired slices is shown.

FIGURE 4

In vivo validation of SR‐CSE‐derived intramuscular fat fraction (MFF_Intra). (a) Example of a single‐voxel ¹H MRS spectra with fitting for intramyocellular (IMCL) and extramyocellular (EMCL) fat pools which together make up MFF_Intra = IMCL + EMCL (blue: acquired spectrum; green: residual after fitting). (b) Scatter plot showing the agreement between MFF_Intra derived from SR‐CSE and ¹H spectroscopy.

FIGURE 5

Influence of fat content on SASHA and MOLLI T1 bias in skeletal muscle. Comparison of T1 bias (SASHA or MOLLI T1 minus SR‐CSE T1_Water) as a function of MFF_Intra, measured with the SR‐CSE acquisitions, from a total of 277 regions of interest from nine healthy participants.

FIGURE 6

Association of muscle T1_Water with sex and age. (a) Individual T1_Water values (with overlay of mean and 95% CI) for males and females. (b) Scatter plot showing T1_Water values for males and females as a function of increasing age. Dotted line corresponds to sex‐specific mean values shown in (a).

FIGURE 7

Agreement for muscle T1_Water and MFF_Intra in the thigh and paraspinal muscles. (a) Scatter plot showing agreement between values for MFF_Intra and (c) T1_Water in the erector spinae vs. mid‐thigh; Solid black line and adjacent dotted lines represent mean regression line and corresponding 95% confidence interval, respectively, while the solid red‐line represents the line of identity. (b) Values for MFF_Intra and (d) muscle T1_Water grouped according to sex for each muscle region (thigh or erector spinae). Overlay and error bars represent mean and 95% confidence interval, respectively, with P‐values for comparison of males vs. females is shown above.