Association of Vastus Lateralis Diffusion Properties with in vivo Quadriceps Contractile Function in Premenopausal Women

Abstract

Diffusion Tensor Imaging (DTI) parameters correlate with muscle fiber composition, but it is unclear how these relate to in vivo contractile function.

Purpose:

To determine the relationship between DTI parameters of the vastus lateralis (VL) and in vivo knee extensor contractile.

Methods:

Thirteen healthy, premenopausal women underwent magnetic resonance imaging of the mid-thigh to determine patellar tendon moment arm length and quadriceps cross-sectional area. Fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) of the VL were determined using Diffusion Tensor Imaging (DTI). Participants underwent an interpolated twitch (ITT) experiment before and after a fatiguing concentric-eccentric isokinetic knee extension (60°∙s⁻¹). During the ITT, supramaximal electrical stimuli were delivered to elicit twitch responses from the knee extensors before, during, and after a maximal voluntary isometric contraction (MVIC). Knee extensor specific tension during twitch and MVIC were calculated from isometric torque data. Pearson’s correlations were used to determine the relationship between muscle contractile properties and DTI parameters.

Results:

MD and RD were moderately correlated with peak twitch force, and rate of force development. FA and AD were moderately inversely related to percent change in MVIC following exercise.

Conclusion:

MD and RD are associated with in vivo quadriceps twitch properties, but not voluntary strength which may reflect the mechanical properties of constituent fiber types. FA and AD appear to relate to MVIC strength following fatiguing exercise.

Introduction

Muscle tissue quality is a major determinant of contractile performance measures such as strength, rate of force development, and fatigue¹. Evidence suggests that strength declines more rapidly than muscle cross-sectional area, suggesting that loss of muscle contractile quality precedes loss of muscle tissue². Non-invasive assessments of muscle tissue composition and structure may provide important quantification of muscle contractile quality.

Diffusion Tensor Imaging (DTI) is a novel Magnetic Resonance Imaging (MRI)-based technique that provides quantifiable information of muscle architecture such as fiber orientation^3–8. DTI evaluates the anisotropic properties of tissues such as white matter and striated muscle, by measuring rates of water diffusion in the principle axes. These rates of diffusion are called the principal eigenvalues λ1, λ2, and λ3 and are usually ordered so that λ1 > λ2 > λ3. Typically, water diffuses more readily along the direction of axial symmetry and less readily in the two directions perpendicular to this direction. From the principal eigenvalues one can derive a variety of other useful diffusion parameters. The most commonly reported DTI parameters include: Fractional Anisotropy, Mean Diffusivity, Radial Diffusivity, and Axial Diffusivity. Fractional Anisotropy is calculated as a fraction derived from the ratio of DTI eigenvalues λ1, λ2, and λ3. This index of diffusion asymmetry exists on a continuum implying completely isotropic diffusion (value of 0) to completely anisotropic diffusion (value of 1). Mean diffusivity represents the diffusivity of water throughout the tissue in all directions and is calculated as the average of the three eigenvalues. Radial Diffusivity represents diffusivity across anisotropic tracts and is calculated as the average of the two eigenvalues perpendicular to the axial direction (λ2, and λ3). Lastly, Axial Diffusiviy quantifies the speed that water diffuses along the axial direction of the tracts and is measured as the eigenvalue in the axial direction (λ1).

Skeletal muscle is highly anisotropic due to the parallel arrangement of myofibrils and muscle fibers. DTI has been previously used for muscle fiber tracking in human skeletal muscle^9–11. DTI of skeletal muscle is sensitive to several demographic factors such as age, sex, and body weight as well as intrinsic factors such as joint angle, ischemia, and muscle damage¹². To date, it is unclear how DTI parameters relate to muscle contractile performance.

Importantly, DTI parameters of skeletal muscle have previously been associated with indices of muscle strength and fiber type. Klupp et al.¹³ demonstrated that spinal extension to flexion strength ratio was moderately correlated with Mean Diffusivity and RD, but not AD or Fractional Anisotropy, of paraspinal lumbar muscles in young, healthy subjects. Since voluntary strength is predicated upon contractile properties of muscle as well as voluntary recruitment of motor units, it remains to be seen how DTI parameters specifically relate to contractile function of skeletal muscle.

Through a combination of DTI analysis and fine needle muscle biopsy, Scheel et al.¹⁴ showed that Fractional Anisotropy (positively) and Mean Diffusivity and Radial Diffusivity (inversely) were strongly correlated with proportion of type I muscle fibers in the soleus of 12 young, healthy males. It is well established that type I fibers demonstrate lesser peak force and slower contraction velocity¹⁵ than type II fibers. Additionally, post-activation potentiation appears to be inversely related to myosin heavy chain type I expression and proportion of type I muscle fibers^16,17. Therefore, these data suggest that higher levels of Mean Diffusivity and Radial Diffusivity and lower levels of Fractional Anisotropy would be associated with greater contractile force and speed. Indeed, a follow up paper by this group demonstrated that mechanical power of the soleus was strongly inversely correlated with Fractional Anisotropy and positively correlated with RD during maximal voluntary isokinetic exercise¹⁸. However, these authors did not report measures of muscle force, rate of force development, post-activation potentiation, or voluntary recruitment.

The interpolated twitch technique (ITT)^19–21 is a unique method for assessing muscle contractile properties and neural recruitment in vivo. In the ITT, a supramaximal electrical stimulus elicits a twitch response before, during, and after a maximal voluntary isometric contraction. Although the limitations and interpretation of ITT data have been the subject of scientific debate^22–25, this technique allows quantification of twitch properties and maximal voluntary strength as well as an index of “voluntary activation”. Voluntary activation is calculated as the percent of peak twitch torque which is recruited during an MVIC. Although this value should not be conflated with the proportion of motor units recruited or degree of motor unit activation, voluntary activation provides a valuable index for neural recruitment.

The purpose of this study was to determine how DTI parameters relate to in vivo contractile function of the quadriceps. Due to data indicating Fractional Anisotropy is directly related, and Mean Diffusivity and Radial Diffusivity are inversely related, to proportion of type I muscle fibers, we hypothesized that Fractional Anisotropy would be inversely related to quadriceps force and rate of force development and that Mean and Radial Diffusivity would be positively related to force and rate of force development. Furthermore, given the fatigue-resistant nature of type I muscle fibers, we hypothesized Fractional Anisotropy would be inversely related to fatigue and that Mean and Radial Diffusivity would be positively related to quadriceps fatigue.

Materials and Methods

Subjects

Thirteen, healthy, premenopausal women volunteered to participate in this study. Due to well-described differences in DTI and muscle contractile function between sexes and with differing ages¹², we limited our study cohort to premenopausal women. Research procedures were approved by the university’s medical institutional review board. All subjects provided written, informed consent prior to participation in accordance with the policies of the university’s Office of Research Integrity.

Following consent, all subjects completed a Physical Activity Readiness Questionnaire and health history form to identify existing contraindications to exercise²⁶. All subjects were free from cardiovascular, respiratory, and metabolic disease; free from low back and lower extremity injury, and were not taking medication that may inhibit muscle strength or activity. Subjects were weight-stable (<10% fluctuation in body weight) and not participating in structured cardiovascular or resistance exercise for at least six months prior to participating in this study. Women who were post-menopausal, were currently or had been pregnant within the previous six months were excluded from the study.

Research design

Subjects reported to the research laboratory three times over the course of the study. At intake, body composition, mid-thigh MRI, and muscle strength testing familiarization were completed. At least seven days separated the familiarization and final testing session to limit potential effects of muscle damage or fatigue from the familiarization exercise. Session one consisted of obtaining informed consent, completion of health and physical activity questionnaires, anthropometric measurements, and a total body dual-energy X-ray absorptiometry (DXA) scan to assess body composition. During session two, subjects underwent MRI scan of the mid-thigh on the self-reported dominant leg (preferred leg to kick a ball). Following the MRI, subjects completed the familiarization of muscle strength testing protocols using an isokinetic dynamometer. During the final testing session, subjects performed an ITT experiment.

Anthropometrics and body composition

Standing height was measured on a wall-fixed stadiometer (SecaModel 216, Seca, Ontario, CA) to the nearest 0.1 cm. Body weight was measured to the nearest 0.01 kg using a calibrated electric scale (American Scales & Equipment Company, Inc., Baltimore, MD). These measures were completed with the subjects wearing light-weight clothing containing no metal and without shoes.

Body composition was measured using total body DXA scans performed and analyzed by a single trained investigator with standardized procedures²⁷ using a GE Lunar Prodigy (Lunar Inc., Madison, WI; software version 13.10). In accordance with state and university procedures and policies, all subjects completed a urine pregnancy test (McKesson Corp., San Francisco, CA) (ICON 20 hCG, Beckman Coulter, Inc., Brea, CA) immediately prior to DXA scanning. Urine specific gravity was analyzed by a digital USG refractometer (ATAGO 4410 PAL-10S, ATAGO Co., LTD) and required to be above 1.003 for accuracy of pregnancy testing. Total body composition measures of absolute and relative fat mass were determined for each participant.

Magnetic resonance imaging

MRI was conducted using a Siemens PRISMA 3T, a Tim System (Siemens Healthineers, Malvern, PA). Subjects lay supine, with a 14.5cm thick pad under the knees to increase subject comfort and to slightly flex the knee (~10°) to elongate and straighten the quadriceps. A spine array coil was placed on the MR-scanner table and a flexible body coil was wrapped around the subject’s upper thigh and centered over mid-thigh (half the distance between the greater trochanter and lateral epicondyle of the femur) of the dominant leg. Straps were placed around the flexible coil to secure its placement and around the heel to reduce movement of the lower extremities. Prior to scanning, subjects were provided with a pillow, blanket, and ear plugs to promote comfort.

All scans were unilateral, 6mm slices of the mid-thigh. Imaging began with a three-plane localizer(s) to determine position and orientation of the subject’s leg. In order to place image sections uniformly among subjects, a fish oil lozenge was placed on the subjects’ mid-thigh and the orientation of the central image was aligned accordingly. Following previously published scan techniques²⁸, axial images were acquired superior and inferior to the central slab while matching the slight flexion of the knee in the magnet bore. A sagittal plane, 3D T1-weighted MR image, centered on the patella was used to measure patellar tendon moment arm length. Additionally, four packets of T1 Dixon sequences were collected in 6mm axial slices and used to assess cross-sectional area (CSA) of the quadriceps.

Analysis of MRI was completed using ImageJ software (64-bit Java 1.8.0_172) in accordance with previously published literature²⁹ (Figure 1A–B). For CSA analysis, regions of interest were manually drawn around the entire quadriceps muscle on a single image using the polygon selection function in ImageJ. A threshold function (manually adjusted for each image) was applied to remove non-muscle tissue (i.e. intermuscular fat) from CSA measures. We selected one image from the T1-weighted, sagittal 3D set which best visualized the patellar tendon. Patellar tendon moment arm length (dPT) was then measured as the perpendicular distance between the patellar tendon and the tibio-femoral contact point as previously described^30,31(Figure 1C). Patellar tendon moment arm length was used to calculate quadriceps specific tension during twitch and voluntary isometric contractions. Importantly, it has previously been shown that passive patellar tendon moment arm length is similar among the range of knee joint angles used in this study³⁰.

Figure 1. Representative Magnetic Resonance Imaging (MRI) scans from one subject.

A: Cross-sectional MRI of the mid thigh with quadriceps muscles outlined and annotated. B. Thresholded cross-sectional MRI of mid-thigh. Dark areas were used for quadriceps CSA quantification to eliminate non-muscle tissue such as intermuscular adipose tissue. C: Longitudinal MRI showing patellar tendon moment arm (dPT) as the perpendicular distance between the patellar tendon and the tibio-femoral contact point. D: cross-sectional view of mid thigh muscle fiber tractography from DSI studio. VL: vastus lateralis, VI: vastus intermedius, RF: Rectus Femoris, VM: vastus medialis, SA: Sartorius, AL: Adductor Longus, AM: Adductor Magnus, BF: Biceps Femoris, ST: Semitendinosus, SM: Semimembranosus.

We collected DTI images similar to that previously described by Noehren et al.²⁸. Single-shot, echo planar imaging (EPI) acquisition parameters were TR / TE = 3000/64.8 ms with number of signal excitations of 4. The DTI acquired 27 gradient directions at b = 500 s∙mm⁻² and 4 repetitions at b = 0. Eleven slices, each 6 mm thick were acquired with a sequence which employed a GRAPPA factor of 2, a field of view of 192 mm, and acquisition matrix of 96 × 96 to produce image voxels with dimensions of 2 × 2 × 6 mm³. The EPI phase-encoding direction was posterior-to-anterior to shift signal from non-fat suppressed subcutaneous adipose tissue in the anterior direction and away from the VL muscle in the DT images. DSI Studio^11,32 was used to analyze the raw DTI images and derive Fractional Anisotropy (FA), Mean Diffusivity (MD), Radial Diffusivity (RD), and Axial Diffusivity (AD) of the quadriceps. DTI values did not differ among quadriceps muscles. Coefficient of variation among quadriceps muscles for anisotropic parameters were: FA: 5.35 ± 2.65, MD: 2.68 ± 1.31, RD: 3.60 ± 1.80, and AD: 2.39 ± 1.21%. Since the VL was the largest of the quadriceps muscles (40.5 ± 7.9% of all quadriceps fiber tracks) and imaging was optimized for this muscle, data presented are limited to the VL. Representative cross-sectional DTI images from one subject can be found in Figure 1D.

Immediately after MRI assessment, subjects were familiarized with the isokinetic dynamometer muscle strength procedures. Subjects performed three maximal voluntary isometric contractions (MVIC) of the knee extensors with a knee angle of 60° flexion. Each trial was separated by 2-minute rest. Next, all subjects performed one set of 10 repetitions of the fatiguing exercise (coupled isokinetic concentric-eccentric knee extension at 60°∙s⁻¹). The familiarization session and final testing session were separated by at least seven days to mitigate potential effects of fatigue or acute muscle damage. Isokinetic dynamometer settings were identical between familiarization and testing sessions.

Interpolated twitch (ITT)

Peak twitch force (PTF), rate of force development (RFD), relaxation rate (RR), MVIC specific tension, PAP, and voluntary activation of the quadriceps were measured using an ITT experiment. Participants were seated in an upright position (approximately 85° hip flexion), with the dominant leg secured at a knee joint angle of 60° flexion on an isokinetic dynamometer (Biodex S3). The axis of rotation of the knee was aligned with the axis of rotation of the dynamometer. The ankle was secured with a padded strap just above the lateral malleolus and test leg was secured by a strap across the distal thigh. To minimize movement of the upper body, chest and abdominal straps were placed and secured during testing.

Two self-adhesive gel electrodes (12.7 cm × 6.35 cm Dura-Stick Plus Stimulating Gel Electrodes), connected to a constant-voltage electrical muscle stimulator (Digitimer DS7A, Hertfordshire, UK), were placed over the femoral nerve, immediately distal to the inguinal fold and at the distal end of the quadriceps above the patella. Optimal current for the ITT experiment was determined using an individualized dose-response procedure immediately prior to ITT. During the dose-response procedure, the electrical current was initially set at 10 mA for the first stimulus and increased by 10 mA, with a minimum of 5 s between stimuli until a plateau in torque output was achieved. In accordance with previously standardized study procedures¹⁹, current was set to 120% of that needed to evoke maximal twitch torque. The interpolated twitch experiment consisted of a train of eight constant voltage (400 V) square wave doublet (200 μs, 10 ms delay) electrical stimuli applied at a frequency of 0.33 Hz (Digitimer DG2A Train Delay Generator, Hertfordshire, UK). Three twitches were delivered to the fully relaxed quadriceps. After a one second rest period, subjects performed a 5 s MVIC with an electrically-evoked superimposed twitch. After the MVIC, four post-contraction twitches were delivered to the relaxed muscle. To ensure post-contraction twitches were delivered in the fully relaxed state, only the final three post-twitches were used for data analysis. This procedure was performed in triplicate before and after a fatiguing bout of exercise with one-minute separating trials. Two subjects refused to complete the ITT following the fatiguing exercise bout.

Concentric-eccentric fatiguing exercise

Subjects performed three sets of 10 maximal-effort coupled concentric and eccentric knee extensions at 60°∙s⁻¹. Exercise was completed on the isokinetic dynamometer with 30 s rest between sets. Due to an inability to generate sufficient torque to trigger eccentric action of the dynamometer, range of motion was set from 45–105° flexion. Thus, each set lasted a total of 20s and total exercise time was 60s.

Data Acquisition and Processing

All torque data were gravity corrected, streamed through LabVIEW software (National Instruments, Austin, TX), and time synchronized with the signal from the muscle stimulator. Torque output and electrical stimulation data were processed using a custom written MATLAB program. Torque data were high-pass filtered at 10Hz. Individual twitch responses were isolated. Peak twitch torque was defined as the highest torque value achieved within 150ms of electrical stimulation. Rate of torque development and relaxation rate were calculated in the linear phase of the twitch response between 15 and 85% of peak twitch torque. MVIC was defined as the peak torque produced prior to the superimposed electrical stimulus. The superimposed twitch torque (SITT) was defined as the maximal torque achieved within 150ms after the electrical stimulus. Representative torque and electrical stimulation tracings from one ITT experiment can be found in Figure 2A–B to illustrate workflow of the ITT and calculations from this procedure.

Figure 2. Representative tracing from the interpolated twitch (ITT) experiment from one subject.

A: Interpolated twitch torque tracing. Supramaximal electrical stimuli were delivered at rest (pre-MVIC twitches), during (superimposed twitch torque, SITT), and after (post-MVIC twitches) a maximal voluntary isometric contraction (MVIC). Solid black line indicates torque tracing, grey line indicates electrical muscle stimulation (e-stim). B: Expanded view of first pre-MVIC twitch. Peak Twitch Force (PTF) was defined as the highest force value after dividing by patellar tendon moment arm length and normalizing to cross-sectional area. Rate of force development (RFD) and relaxation rate (RR) were calculated in the linear phases between 15 and 85% peak twitch torque (horizontal dashed lines) after dividing by patellar tendon moment arm length and normalizing to cross-sectional area. Post-activation potentiation was defined as the percent difference in peak pre-and post-MVIC twitch torques.

Torque data were divided by dPT and normalized to quadriceps CSA to deliver measures of PTF, RFD, RR, and MVIC specific tension during the ITT experiment. PTF, RFD, and RR were calculated using the pre-MVIC twitches. PAP was calculated as the percent difference in peak twitch torque before and after MVIC. Voluntary activation was calculated as: (1 – (SITT – MVIC)/Post-twitch torque) × 100¹⁹.

Statistical Analyses

Paired-samples t-test were used to compare effects of fatiguing exercise on muscle contractile function (α = 0.05). Cohen’s D was used as an indicator of effect size and interpreted as small (d < 0.20), medium (0.20 < d < 0.79), or large (d ≥ 0.80). Pearson correlations (r) were used to evaluate DTI parameters and twitch properties. Based on previously published data^9,33 a weak correlation was defined as r < 0.40, moderate correlation was defined as 0.40 ≤ r < 0.70, and strong correlation was defined as r ≥ 0.70. All statistical analyses were made using SPSS software (Version 27.0, SPSS, Inc. Chicago, IL).

Results

Effects of exercise on in vivo muscle contractile function

Subject characteristics, body composition, and MRI data are presented in Table 1. As shown in Table 2, twitch RFD and MVIC specific tension were significantly lesser following fatiguing exercise with large effect sizes. PTF tended to be lower following exercise, but this failed to reach statistical significance. Neither PTF, PAP, nor Voluntary activation were significantly different between pre- and post-exercise measures.

Table 1. Subject Characteristics.

Subjects were presumably healthy, premenopausal women (N = 13). Body composition was determined by dual energy x-ray absorptiometry (DXA). Quadriceps cross-sectional area (CSA) was determined as the anatomical area of the quadriceps after removing area of non-muscle tissues.

Parameter	Mean		SD	Range
Age (y)	29.6	±	5.6	22.0	–	40.0
Height (m)	1.64	±	0.06	1.54	–	1.74
Weight (kg)	80.9	±	20.1	49.7	–	109.2
Body Mass Index (kg∙m−2)	29.8	±	6.5	18.5	–	38.7
Body Composition (% fat mass)	42.2	±	9.2	21.9	–	55.0
Quadriceps CSA (cm2)	49.3	±	9.2	35.0	–	67.9
Fractional Anisotropy (FA)	0.305	±	0.038	0.268	–	0.404
Mean Diffusivity (MD)	1.187	±	0.045	1.118	–	1.283
Axial Diffusivity (AD)	1.625	±	0.066	1.513	–	1.784
Radial Diffusivity (RD)	0.968	±	0.055	0.893	–	1.078

Table 2. Isometric quadriceps contractile performance before and after fatiguing isokinetic exercise.

Isometric contractile performance (60° knee flexion) was assessed through an interpolated twitch experiment before and one minute after completion of a fatiguing exercise protocol consisting of three sets of 10 maximal-effort concentric and eccentric isokinetic (60°∙s⁻¹) knee extension over a range of 60° with 30s rest between sets. Peak twitch force, rate of force development, and relaxation rate were determined from resting twitch data. Post-activation Potentiation was calculated as the % change in peak twitch force before and after MVIC. MVIC: Maximal Voluntary Isometric Contraction. Significant differences (two-tailed paired t-test) are denoted by bold text.

Parameter	Time	Mean		SD	Min	Max	p-value	Cohen’s D
Peak Twitch Force (N∙cm−2)	Pre	24.2	±	9.0	12.2	40.2	0.114	0.495
	Post	20.0	±	7.7	9.9	36.6
Rate of Force Development (N∙cm−2∙s−1)	Pre	242.3	±	97.8	103.6	390.1	0.013	0.850
	Post	177.9	±	62.8	85.1	313.2
Potentiation (% Peak twitch force)	Pre	7.33	±	25.9	−29.6	52.9	0.953	−0.017
	Post	9.29	±	11.2	−5.0	29.1
MVIC Specific Tension (N∙cm−2)	Pre	89.2	±	20.3	63.5	123.6	0.002	1.151
	Post	71.3	±	14.7	43.6	90.8
Voluntary Activation (%)	Pre	92.1	±	6.9	74.2	99.0	0.978	0.087
	Post	92.7	±	5.4	81.2	99.0

Associations between DTI parameters of the VL and quadriceps contractile function

Pearson correlations between DTI parameters and contractile function are presented in Table 3. Graphical depictions of the relationship between DTI parameters and twitch properties are presented in Figure 3. PTF was moderately positively correlated with MD and RD, but was not significantly correlated with FA or AD. RFD was moderately positively correlated with MD and RD. RFD tended to be inversely related to FA, but not AD. MVIC specific tension, PAP, and voluntary activation were not significantly correlated with any DTI parameters. FA and AD, but not MD or RD, were moderately correlated to % change in MVIC following fatiguing exercise. No other significant correlations were found between % change in contractile function and DTI parameters.

Table 3. Pearson correlations between DTI parameters and muscle contractile function before and after fatiguing exercise.

PRE: correlation between muscle contractile function before fatiguing exercise. POST: correlation between DTI parameters and % change following fatiguing knee extension. Significant values are denoted by bold text.

Parameter			FA	MD	AD	RD
Peak Twitch Force (PTF)	PRE	r	−0.493	0.565	0.075	0.644
		p-value	(0.087)	(0.044)	(0.808)	(0.018)
	POST	r	−0.113	0.051	−0.033	0.081
		p-value	(0.727)	(0.876)	(0.918)	(0.801)
Rate of Force Dev (RFD)	PRE	r	−0.546	0.580	0.051	0.677
		p-value	(0.053)	(0.038)	(0.869)	(0.011)
	POST	r	−0.257	0.135	−0.062	0.202
		p-value	(0.421)	(0.675)	(0.847)	(0.529)
Post-Activation Potentiation (PAP)	PRE	r	0.157	0.142	0.319	−0.016
		p-value	(0.609)	(0.642)	(0.288)	(0.958)
	POST	r	0.067	0.020	0.004	0.022
		p-value	(0.837)	(0.951)	(0.990)	(0.946)
MVIC Specific Tension	PRE	r	0.099	0.167	0.237	0.061
		p-value	(0.747)	(0.585)	(0.436)	(0.844)
	POST	r	0.585	−0.130	0.628	0.216
		p-value	(0.046)	(0.688)	(0.029)	(0.499)
Voluntary Activation	PRE	r	0.075	−0.428	−0.270	−0.360
		p-value	(0.808)	(0.145)	(0.373)	(0.227)
	POST	r	−0.529	−0.113	−0.516	0.169
		p-value	(0.094)	(0.741)	(0.104)	(0.619)

Figure 3. Scatterplot of Diffusion Tensor Imaging properties and peak twitch force (A-D) and rate of force development (E-H) in the pre-exercise period.

A,E: Fractional Anisotropy, B,F: Mean Diffusivity, C,G: Axial Diffusivity, D,H: Radial Diffusivity.

Discussion

We sought to determine how DTI parameters of the VL relate to in vivo muscle contractile properties of premenopausal women. To our knowledge, this is the first study to describe the relationship between muscle DTI parameters and intrinsic contractile properties. Our data demonstrate that MD and RD, but not FA and AD, are moderately correlated with PTF and RFD, but not voluntary strength, PAP, or voluntary activation. Since these twitch parameters are independent of the central nervous system, our data suggest that muscle diffusivity is related to the intrinsic ability of muscle to generate force.

Previous data indicate that type I fiber proportion is positively correlated with FA and inversely correlated with MD and RD¹⁴. Scheel et al.¹⁸ further demonstrated that voluntary isokinetic power was associated with FA and RD. Our results demonstrate that PTF was positively related to MD and RD, but not FA or AD. These data partially support our hypothesis. Importantly, MD is the average of eigenvalues λ1, λ2, and λ3 where AD is equal to λ1 and RD is equal to the mean of λ2, and λ3. The lack of association between AD and twitch properties suggests that diffusivity along the axis of the muscle is not an important determinant of twitch properties, but that diffusivity perpendicular to the axis may be an important determinant of contractile properties. This also explains why RD displayed stronger correlations with twitch properties than did MD. Our data (collected from a single knee extensor muscle), suggest that RD may reflect the contractile properties of the constituent fiber types. Although we noted no differences among DTI parameters among the knee extensors, more work is needed to confirm the relationship between DTI parameters and muscle contractile properties. This is especially true since fascicle orientation and elastic elements (both parallel and series) may affect the ability to assess this relationship at a whole-muscle level.

Although MD and RD were significantly related to twitch properties before exercise, we noted no significant correlation between these parameters and PAP or percent change in twitch properties following fatiguing exercise. Type II muscle fibers are associated with a greater PAP response and fatigability than type I fibers^16,17. Therefore, we expected PAP and % change in contractile properties would be related to FA, MD, and RD. However, PAP was not significantly correlated to any DTI parameters. PAP is largely determined by phosphorylation of the regulatory light chain of myosin which positions the myosin head closer to the actin filament and reduces the number of super-relaxed myosin heads and persists for several minutes following exercise³⁴. It is possible that PAP is more dependent on this mechanism and is less sensitive to differences in diffusion parameters. This mechanism may also explain why resting PTF was unchanged following exercise. Therefore, it appears DTI measures may not be sensitive enough to monitor mechanisms that lead to fatigue.

It should be noted that DTI parameters change acutely in response to muscle fatigue and damage. Indeed, MD, FA, and RD are significantly altered in as little as 24–48h^5,35 following marathon running and return to baseline within 2 weeks. Rockel et al.³⁶ conducted 5 min of resisted in-bore dorsiflexion-eversion foot motion and noted significant increases in MD of active, but not inactive, muscle in the immediate post-exercise period. Therefore, it appears that DTI properties of muscle change during exercise, possibly due to hyperemia or sarcomere disruption. It is possible that post-exercise DTI may be more sensitive than pre-exercise DTI measures to evaluate fatigability of muscle.

Interestingly, our data demonstrate no relationship between DTI parameters and MVIC specific tension in the pre-exercise period. These data appear to conflict with those of Klupp et al.¹³ who showed that spinal extension to spinal flexion ratio was moderately correlated with paraspinal MD and RD. This may partially be explained by methodological differences: our data exclusively analyzed knee extension force rather than a muscle strength ratio.

In the present study, exercise reduced MVIC force, but not peak twitch force or voluntary activation. Thus, the mechanism of fatigue is unclear. The lack of change in peak twitch force suggest reduced MVIC was driven by reduced motor unit recruitment in the present study. Females tend to have a greater proportion of type I muscle fibers than males³⁷ which may explain the apparent fatigue resistance of our subjects’ muscle. Importantly, voluntary activation (as assessed by ITT) is calculated as the percent of peak twitch torque that is elicited during a maximal voluntary isometric contraction. As previously discussed in the literature, ITT-based measures of voluntary activation due not directly assess the proportion of motor units or degree to which motor units are activated during a maximal voluntary effort^22–25. Thus, ITT data are viewed by some as a qualitative assessment of neural recruitment. Thus, the ITT technique may not be sensitive enough to detect changes in neural recruitment that occur during this fatiguing exercise. As contractile force increases, so does firing frequency of individual motor units³⁸. Advanced analysis of motor unit recruitment and firing frequency through EMG signal decomposition may help establish the mechanism of fatigue during this type of exercise.

Interestingly, both FA and AD were positively associated with percent change in MVIC specific tension in the post-exercise period. We conclude that FA and AD may better represent fatigability of muscle under tetanic conditions than MD or RD. Future research is needed to determine how DTI parameters relate to muscle force production during tetanic contractions.

We found no association between AD and contractile properties at rest. These data appear to agree with previous findings showing no relationship between AD and spinal extension/flexion strength ratio³⁹, fiber type distribution¹⁴, or plantarflexor power¹⁸. Based on these findings, we conclude that, despite its essential role in determining fiber tractography, Therefore, it is possible that AD may be a better index of muscle fatigue or tetanic contractile force than other DTI parameters.

Limitations

We calculated MVIC specific tension relative to anatomical CSA (measured perpendicular to the thigh at the largest portion along the length of the muscle) rather than the gold standard physiological CSA (measured perpendicular to fascicle orientation) which is a better index of parallel contractile elements. However, Erkshine et al.⁴⁰ previously showed that normalizing quadriceps force to anatomical CSA produced similar values as normalizing to physiological CSA. It should also be noted that we measured dPT, CSA, and VL DTI with the leg only slightly bent whereas contractile function measures were made at 60° flexion. Although our specific tension data compare favorably to that previously calculated for the tibialis anterior⁴¹ and quadriceps¹⁷, these methodological differences may explain, in part, the different values we observed for MVIC specific tension compared to others⁴⁰. Unfortunately, limitations in bore dimensions prevented us from performing MRI scanning with the knee flexed to 60° which is the knee joint angle that is optimal for assessing maximal strength of knee extensors⁴².

Despite previous reports of high correlations between FA and RD with muscle fiber type, we noted only moderate correlations between RD and quadriceps twitch properties at rest. It should be noted that DTI measures the anisotropic properties of tissues relative to the axis of symmetry. Thus, DTI parameters reflect diffusion along and perpendicular to the muscle fascicles, which is not synonymous with the net direction of force along the whole muscle. We postulate that diffusion properties may be more closely related to contractile mechanics of fascicles than measures of whole muscle contractile function. It is likely that the complex arrangement of 3D fascicle orientation among the quadriceps muscles resulted in a weaker correlation between diffusion properties and our measures of whole muscle contractile function. Future work should aim to discover how DTI parameters relate to contractile mechanics of muscle fascicles.

It should also be mentioned that our DTI analysis was limited to a limited mid-thigh region of VL. Although DTI parameters were similar among all quadriceps muscles, it is possible DTI of the whole VL may be more representative of all constituent fiber types in this muscle. Recent data indicate that fiber type does not systematically vary longitudinally throughout the VL, but that individual biopsy samples differ substantially in fiber type distribution⁴³. Our analysis represents a large volume (127.1 ± 32.4 cm³) of the VL and may, therefore, better represent the composition of this muscle than relying on a single muscle biopsy. Still, more work is needed to establish the relationship between DTI parameters and muscle fiber type to allow non-invasive assessment of fiber type in muscles which are inaccessible by needle biopsy.

Perspective

Our data demonstrate that DTI parameters of the VL, specifically MD and RD, moderately relate to in vivo contractile function of quadriceps at rest. We speculate that this is mediated by differences in diffusion properties among type I and type II fibers. Furthermore, we show that FA and AD are moderately related to fatigability under voluntary tetanic contractions. Assessing DTI may provide valuable clinical insights in the structure and contractile quality of muscle. It is well established that DTI parameters change with age, body weight, and training status¹². Identifying early changes in muscle diffusion properties may allow for interventions to restore muscle quality before permanent changes in contractile function are made. Therefore, future studies should aim to discover how muscle DTI may be used as a clinically meaningful biomarker for non-invasive, passive assessment of muscle contractile function. Although our sample size compares favorably to others which have investigated DTI parameters of skeletal muscle, studies with greater number of subjects will be needed to establish the feasibility of this technique as a clinically meaningful biomarker. The readily available, non-invasive nature of DTI makes this method a potentially useful tool for assessing muscle contractile function in settings where direct assessment of contractile properties is not possible.