Longitudinal Sensitivity to Change of MRI-based Muscle Cross-Sectional Area vs. Isometric Strength Analysis in Osteoarthritic Knees With and Without Structural Progression – Pilot Data from the Osteoarthritis Initiative

Abstract

Object:

Biomechanical measurement of muscle strength represents established technology in evaluating limb function. Yet, analysis of longitudinal change suffers from relatively large between-measurement variability. Here we determine the sensitivity to change of MRI-based measurement of thigh muscle anatomical cross sectional areas (ACSAs) vs. isometric strength in limbs with and without structural progressive knee osteoarthritis (KOA), with focus on the quadriceps.

Materials and Methods:

Of 625 “Osteoarthritis Initiative” participants with radiographic KOA 20 had MRI cartilage and radiographic joint space width loss in the right knee isometric muscle strength measurement, and axial T1-weighted spin-echo acquisitions of the thigh. Muscle ACSAs were determined from manual segmentation at 33% femoral length (distal to proximal).

Results:

In progressor knees, the reduction in quadriceps ACSA between baseline and two-year follow-up was −2.8±7.9% (standardized response mean [SRM] =−0.35), and it was −1.8±6.8% (SRM=−0.26) in matched, non-progressive KOA controls. The decline in extensor strength was more variable than that in ACSAs, both in progressors (−3.9±20%; SRM=−0.20) and in non-progressive controls (−4.5±28%; SRM=−0.16).

Conclusion:

MRI-based analysis of quadriceps muscles ACSAs appears to be more sensitive to longitudinal change than isometric extensor strength, and is suggestive of greater loss in limbs with structurally progressive KOA than in non-progressive controls.

INTRODUCTION

Biomechanical measurement of muscle strength is commonly used to evaluate an important component of limb and muscle function, specifically in context of knee osteoarthritis (KOA) incidence and progression [1–4]. Recommendations by the Osteoarthritis Research Society International include quadriceps muscle strengthening for the clinical management of KOA [5]. Although exercise is known to modify muscle strength [6–9], and high quadriceps muscle strength is thought to protect against the onset of symptomatic KOA [1,10,11], it is currently unclear whether such intervention can alter the incidence or structural progression [1,10,12,13] of KOA, and thus has structure- (or disease-) modifying potential [1,14,15]. Further, measurement of (longitudinal change in) muscle strength suffers from large between-measurement variability [16–19], is influenced by the voluntary muscle activation [20,21], and may hence be relatively insensitive to the measurement of small changes over time. In contrast, quantitative measurement of thigh muscle volume and anatomical cross sectional areas (ACSAs) has been shown to be highly sensitive to change, for instance in context of short term exercise intervention [22].

A recent study [23] was the first to examine the longitudinal change in quadriceps muscle volume occurring in KOA. This study reported rates of change of muscle volume in knees with symptomatic and radiographically definite KOA, i.e. Kellgren Lawrence grades (KLG) 2–4 (−2.1±5.5% over two years) that were similar to those in subjects with risk factors for KOA, but without definite radiographic KOA (KLG 0/1: −1.5±4.4%). The authors concluded that quadriceps ACSAs may decline as a function of aging, independently of radiographic KOA status. However, it is well known that amongst subjects with KOA many remain in a static phase of the disease, where no or little structural (radiographic or other) change occurs, and only few knees actually display structural progression, such as loss of radiographic joint space width (JSW) [24] or cartilage loss [25,26]. Yet, it is conceivable that muscle mass is deteriorating more strongly in those who are structurally progressing than in those who exhibit static phases of the disease, and that therefore limbs with progressive KOA may be a better model to study concurrent change in muscle status than unselected limbs with radiographic KOA.

The purpose of this pilot study was to evaluate the sensitivity to longitudinal change of MRI-based measurement of thigh muscle ACSAs vs. those of isometric strength. This comparison was conducted in limbs with and without concurrent structurally progressive KOA. Structural progression was defined by MRI (quantitative cartilage loss) AND radiography (loss of JSW in fixed flexion views). Given the more prominent role of the quadriceps in KOA [1,10,11,22], as well as its greater responsiveness to immobilization [27], training intervention [22], and pain [28] in comparison to other thigh muscles, the primary focus of our analysis was on the quadriceps and extensor strength. However, other thigh muscles (hamstrings, adductors) and flexor strength were also evaluated. Further, the study was designed to also examine whether baseline MRI-measurement thigh muscle (specifically quadriceps) ACSAs were more strongly associated with structural progression status in KOA than biomechanical measurement of isometric muscle strength.

MATERIALS AND METHODS

Study design and sample selection

Clinical and imaging data were drawn from the Osteoarthritis Initiative (OAI) database. General inclusion criteria for the OAI have been published and are publicly available (http://oai.epi-ucsf.org/datarelease/) [29]. Informed consent was obtained from all 4796 participants and the study was approved by the local ethics committees.

For the current matched prospective cohort study, OAI knees with and without medial structural progression were selected based on longitudinal data on quantitative cartilage loss obtained between baseline and one year follow-up using 3 Tesla MRI [30,31], and on JSW loss obtained with fixed flexion radiography during the same interval [32]. Cartilage thickness change had been studied in the right knees of an OAI subsample of 725 participants [29,30] who were KLG 2–4 according to the site radiographic readings, using a double oblique coronal fast low angle shot (FLASH) MRI sequence. This MRI sequence has previously been validated in context of technical accuracy of cartilage thickness measurement [33–35]. Radiographic JSW [32,36] was used to ensure that apparent change in cartilage loss was not due to MRI-specific precision errors or artifacts, and that structural progression could be confirmed by a second independent method. Of the 725 right knees studied with MRI, 624 also had quantitative JSW measurement (Fig. 1) at baseline and follow-up (kXR_QJSW_Duryea00 and 01) [32]. Using the smallest detectable change method (SDC) [37] and test-retest data obtained at baseline and follow-up [38], knees with a reduction in medial femorotibial compartment cartilage thickness larger than expected by chance (>102μm) and those with a reduction in medial minimum JSW (mJSW) larger than the SDC threshold (328 μm) based on OAI reliability data at baseline and follow-up were identified (Fig. 1). Knees without structural progression were defined as those without a reduction in either medial femorotibial cartilage thickness (MFTC_ThCtAB) or medial mJSW greater than the above thresholds. Knees that showed no medial but lateral cartilage loss in MRI (SDC threshold = 92μm) were excluded from non-progressor control knees (Fig. 1). Knees with KLG 1 and 4 (release 0.5., central readings) were excluded, the former because they represented a small subcohort with doubtful radiographic KOA status that cannot be addressed as definite knee osteoarthritis, and the latter because of the low or no baseline mJSW and small dynamic window to measure radiographic change. Knee with and without structural progression were matched 1:1 by the same sex and baseline KLG, body height±3cm, body mass index (BMI)±5kg/m², and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain scores±5 (scale from 0–20).

Fig. 1:

Sample selection: Schematic showing the selection process of knees with and without cartilage loss. MRI: Magnetic Resonance Imaging; JSW: Joint Space Width; SDC: Smallest Detectable Change

Measurement of thigh anatomical muscle cross sectional areas (ACSAs)

Of all knees with structural progression we selected those that had axial T1-weighted spin-echo thigh MR image data (TR=600ms, TE=13ms, slice thickness=5mm, in plane resolution=0.98/0.98mm); preferably at baseline and two-year follow-up (release 0.E.1/3.E.1)(Fig. 1). Specifics of the imaging protocol and analysis of thigh muscle ACSAs have been described previously [22,28,29]: In brief, 15 axial contiguous 5mm slices were acquired from a slice located 10cm proximal to the distal femoral epiphysis, and then extending 7.5cm proximally (www.oai.ucsf.edu/datarelease/operationsmanuals.asp). Due to the fixed 10cm distance between the distal femoral epiphysis and the distal MR image, the position of the 15 images varied between subjects, depending on femoral length and body height [28]. To ensure an anatomically consistent location, an image positioned at a relative femoral length of 33% (distal to proximal) was selected based on body height [28,39]. This image was used to determine muscle ACSAs of the quadriceps, hamstrings, and adductors (Fig. 2), given that single slice analysis of muscle ACSAs has been shown to display a high correlation to total thigh muscle volume [40] and to be similarly sensitive to pain-related muscle loss as the analysis of several MRI slices (i.e. the p value for side differences of quadriceps ACSA between painful and painless limbs ranged from 0.00001 to 0.00008 when analyzing single slices, and was 0.00003 when analyzing three contiguous MRI slices) [28]. Using custom software developed at our institution (T.D., W.W.), ACSAs of all muscles were segmented manually by a trained user (M.S.), excluding adipose tissue, vessels, and fibrous tissue between the muscle groups (Fig. 2), [28]. The OAI did not provide test-retest image data, but the precision for similar measurements in another study [41] was 1.7% for the quadriceps, 3.4% for the hamstrings, and 9.9% for the adductors (root mean square [RMS] coefficient of variation [CV%] for test-retest with repositioning between acquisitions). Using OAI data, the intra-reader and inter-reader reliability for measurements of quadriceps volume (from 15 slices) was reported to be <1% (RMS CV%), when processing the same image at two different occasions (one reader), or by two different readers, but without repositioning between test and retest acquisitions [23].

Fig. 2:

Slice selection and image analysis: Coronal localizer MR image to select the axial thigh MR images. Segmentation of the muscle groups (quadriceps, hamstrings, and adductors) are outlined in the right thigh

Measurements of muscle strength

Measurement of the maximal isometric extensor and flexor strength at baseline (and at two-year follow-up [http://www.oai.ucsf.edu/datarelease/forms.asp; release 0.2.2. and 3.2.1.]) was performed using the “Good Strength Chair” (Metitur Oy, Jyvaskyla, Finland) [28,42]: For the measurement, participants were sitting on the chair in an upright position, with the examined limb fixed at 60° flexion angle. The force sensor was positioned in an anatomical consistent location, 2 cm above the top of the calcaneus bone, and the pelvis was fixed with a seat belt. The measurements were acquired by pushing (flexion) and pulling (extension) as strongly as possible. Isometric strength was measured in Newton (N) ( [43]). Test-retest reliability (Pearson product moment correlation coefficients) in the order of r=0.92 for knee extension and of r=0.88 for flexion has been reported previously for this device [16].

Statistical analyses

For reasons outlined above, the primary focus of this pilot study was to determine the rate and sensitivity of the two-year change in quadriceps ACSA in knees with concurrent structurally progressive KOA, in relation to change in a biomechanical measure of isometric strength. Comparison between two-year follow-up vs. baseline was performed using a paired t-test. The standardized response mean (SRM), i.e. the mean longitudinal (two-year) change divided by the standard deviation of the change (across participants), was used as a measure of effect size. A secondary focus was the comparison of the sensitivity to change of ACSAs and isometric strength in knees without structural progression. Further, percent rates of change [(follow-up value – baseline value)/ baseline value] in ACSA, isometric strength (and descriptively for cartilage thickness, mJSW loss and pain) were directly compared between progressors and matched non-progressors using a paired t-test. On an exploratory basis, we also evaluated longitudinal change in ACSAs (and isometric strength) of other thigh muscles. Further, we explored whether baseline measures of ACSA (and strength) were associated with structural progression status in KOA, again with particular focus on the quadriceps. Given recent recommendations in the literature, the latter analyses were stratified between men and women [1,12].

RESULTS

Sample selection and demographic results

Of knees with complete longitudinal MRI and radiographic data (n=625), 134 displayed medial femorotibial cartilage loss in MRI greater than the SDC threshold, 141 medial mJSW loss in radiography, and 54 loss (structural progression) in both methods at the same time (Fig. 1). Of the 46 KLG 2/3 knees that showed progression, 26 had complete information on demographics, isometric muscle strength, baseline and two year thigh MRI acquisitions, and 6 additional cases had only baseline data (Fig. 1). 404 knees did not have medial femorotibial cartilage loss in either MRI or radiography, and 340 of these also did not show cartilage loss in the lateral compartment (MRI). Of these, 229 were KLG 2/3 (Fig. 1), and of these 158 had complete information on demographics, isometric muscle strength, and baseline as well as two year thigh MRI acquisitions (Fig. 1). For 20 of the 26 progressor knees with longitudinal thigh MRI acquisitions and for three of the six additional knees with only baseline acquisitions, a suitable match of a non-progressor knee could be identified based on the above matching criteria.

The demographics of the progressor and non-progressor subsamples are shown in Table 1. Progressor knees exhibited a cartilage thickness loss of −0.24±0.13mm vs. +0.03±0.09mm in non-progressors knees (p=2 × 10⁻⁶), and a reduction in mJSW of −0.93±0.76mm vs. +0.04±0.27 (p=2 × 10⁻⁵). The WOMAC pain score increased from baseline to two-year follow-up in progressor knees (+1.3±2.5 units) but did decrease slightly in non-progressor knees (−0.2±1.6 units); the difference in longitudinal change reached statistical significance (p=0.047). Neither progressors (p=0.92) nor non-progressors (r=0.99) displayed a significant change in BMI over the 2 year observation period.

Table 1:

Demographics of progressors and non-progressors in the cross-sectional and in the longitudinal sample studied

	Longitudinal subsample (n=20)		Cross sect. subsample (n=23)
	Progress.	Non-progress.	Progress.	Non-progress.
	Mean±SD	Mean±SD	Mean±SD	Mean±SD
Age (y)	63.2±7.9	65.3±9.4	62.6±7.6	65.9±9.2
Body height (cm)	165.8±7.9	165.7±7.5	165.0±7.9	165.0±7.5
BMI (kg/m2)	30.0±5.1	30.0±5.3	30.0±4.8	29.9±5.0
Sex (F/M)	13/7	13/7	16/7	16/7
KLG (2/3)	12/8	12/8	14/9	14/9
WOMAC Pain	2.6±2.5	1.8±2.2	3.0±2.7	2.3±2.6
Score [0–20]

BMI = body mass index; F = female; M = male; KLG = Kellgren Lawrence grade; WOMAC = Western Ontario Mac Master, Progress. = progressor knees

Longitudinal comparisons

The reduction in quadriceps ACSAs in limbs with concurrent structurally progressive KOA was −2.8±7.9% [mean±SD]; p=0.12), the effect size (SRM) being −0.35 (Table 2). Loss of extensor muscle strength in the same limbs was of somewhat greater magnitude, but also was substantially more variable across participants (−3.9±20%; p=0.55; SRM=−0.20;. The change in quadriceps ACSA in limbs without structurally progressive KOA was −1.8±6.8% (p=0.30), the SRM being −0.26 (Table 2). Loss of extensor muscle strength in the same limbs was of greater magnitude than that in ACSAs, but was again more variable (−4.5±28%; p=0.16; SRM=−0.16. Direct comparison of the rates of change in ACSA (and strength) between limbs with and without structurally progressive KOA did not attain statistical significance in this small sample (p=0.61 [ACSA] and 0.94 [strength], respectively; Table 2). There only was a very weak correlation between change in quadriceps ACSA and extensor strength in the progressor (r=+0.21) and in the non-progressor group (r=+020), and these correlations did not reach statistical significance (p>0.38).

Table 2:

Longitudinal comparisons between the thigh muscle status in knees with (n=20) and without progression (n=20)

	Knees with Progression			Knees without Progression
	Change (%)	SRM	p-value	Change (%)	SRM	p-value
	Mean±SD		(BL→FU)	Mean±SD		(BL→FU)
Quad ACSA (cm2)	−2.8±7.9	−0.35	0.12	−1.8±6.8	−0.26	0.30
Quad Strength (N)	−3.9±20	−0.20	0.55	−4.5±28	−0.16	0.16
Hams ACSA (cm2)	−0.9±5.2	−0.17	0.38	−2.8±7.8	−0.36	0.08
Flexor Strength (N)	−4.6±34	−0.14	0.26	−13±37	−0.35	0.02
Add. ACSA (cm2)	+4.7±38	+0.12	0.54	+6.5±33	+0.20	0.63
Women (n=13)
Quad ACSA (cm2)	−1.2±7.1	−0.17	0.53	−0.9±6.3	−0.15	0.69
Quad strength (N)	−4.9±24	−0.20	0.67	−7.6±34	−0.22	0.14
Men (n=7)
Quad ACSA (cm2)	−5.6±9.0	−0.63	0.16	−3.3±7.9	−0.41	0.35
Quad strength (N)	−2.0±11	−0.18	0.65	+1.2±12	+0.10	0.91

SRM = standardized response mean; p-value was calculated with a paired t-test; BL = baseline, FU = follow up; ACSA = anatomical cross sectional area; Quad = quadriceps, Hams = hamstrings, Add. = adductors;.

In limbs without structurally progressive KOA, the flexor strength decreased significantly (−13±37%, p=0.02), but again showed a great range of variability; the changes in progressors was also highly variable (−4.6±34%; p=0.26) and was not statistically significant. Changes in hamstring or adductor ACSAs were not significant in either sample and did not differ significantly between limbs with and without structurally progressive KOA (Table 2). A trend towards greater rates of decline of quadriceps ACSA in progressor vs. non-progressor limbs was noted both in women and men, with the SRM being greater (i.e. more negative) than in non-progressors, and greater in men than in women (Table 2).

Cross-sectional comparisons

No statistically significant differences between baseline ACSAs or strength were observed between limbs with and without structurally progressive KOA (Table 3). When stratifying for sex, quadriceps ACSA tended to be smaller in women with structural progression compared to those without progression, and the difference reached borderline statistical significance (p=0.10; Table 3). Quadriceps strength also was slightly lower in women with structurally progressive KOA than in those without, but the difference was far from being statistically significant (p=0.42; Table 3). Male limbs with progression tended to have greater quadriceps ACSAs and strength than those without progression, but again the difference did not reach statistical significance (Table 3).

Table 3:

Cross-sectional pairwise comparisons between the thigh muscle status in knees with (n=23) and without progression (n=23)

	Knees with	Knees without	Difference between groups
	Progression	Progression	95% CI	percent	p
	Mean±SD	Mean±SD	(absolute)	(%)	(t-test)
ACSA (cm2)	45.8±14.4	46.5±10.7	(−5.2/3.7)	−1.5%	0.7
Quad strength (N)	342±112	353±112	(−64/41)	−3.1%	0.65
Hams ACSA (cm2)	31.8±9.1	30.4±7.8	(−1.6/4.4)	+4.6%	0.33
Flexor strength (N)	139±61.5	136±64.2	(−29/36)	+2.2%	0.83
Add. ACSA (cm2)	10.6±6.7	9.7±4.9	(−1.3/3.1)	+9.3%	0.40
Female (n=16)
Quad ACSA (cm2)	38.7±9.3	41.8±7.4	(−6.8/0.7)	−7.4%	0.10
Quad strength (N)	293±91.1	318±88.7	(−88/39)	−7.8%	0.42
Male (n=7)
Quad ACSA (cm2)	62.1±9.9	57.4±9.3	(−8.9/18)	+8.2%	0.43
Quad strength (N)	453±64.9	435±123	(−100/137)	+4.1%	0.72

CI = confidence interval, p-value was calculated with a paired t-test

^*p< 0.05

ACSA = anatomical cross sectional area; Quad = quadriceps, Hams = hamstrings, Add. = adductors.

DISCUSSION

The objective of this study was to compare the sensitivity to longitudinal change of an MRI-based measurement of thigh muscle with that of isometric strength, in limbs with and without structural progressive KOA. Further we examined whether baseline MRI-measures of thigh muscle ACSAs were more strongly associated with structural progression than a biomechanical measurement of isometric strength. Given the important role of the quadriceps in KOA [1,10,11,22] and its greater responsiveness to immobilization, training intervention, and pain [22,27,28] in comparison with other thigh muscles, the primary focus of our analysis was on the quadriceps and extensor strength. We found a somewhat greater sensitivity to longitudinal change (i.e. a more negative SRM) for the MRI-based measure of quadriceps ACSA (−0.35) than for isometric strength (−0.20), mainly due to the smaller between-person variability of the change for ACSA than of that for isometric strength. The rate of change (and sensitivity to change) were smaller in knees that did not display structural progression in MRI or radiography, but the sensitivity to change in that group was still greater for the MRI-based measures of quadriceps ACSA (SRM=−0.26) than for isometric strength (SRM=−0.16).

The main limitation of this pilot study is its limited sample size, and the results should be confirmed in a larger sample. However, a strength of our approach was that two independent imaging methods were used to identify limbs with and without concurrent structurally progressive KOA, and to our knowledge, this is the first time that a combined criterion of quantitative MRI and radiography has been used in a selection process. The rates of cartilage thickness and mJSW loss (−0.24mm/−0.93mm) in our progressor knees were 6–10-fold greater than the mean values observed across a subsample of KLG2–4 knees in the OAI (−0.04mm/−0.09mm), highlighting that the use of a combined MRI and radiographic SDC threshold was very effective in identifying knees with structurally progressive diseases. In contrast, the non-progressor knees did not show any loss at all in either cartilage thickness or mJSW. Whether or not a reduction in quadriceps ACSA would also be observed when only applying an MRI-based criterion of structural progression, but not both, remains to be determined in future studies. Using only radiographic progression as a criterion and exploring the entire OAI data set, we did not find strong evidence that (change in) isometric muscle strength precedes or is associated with structural (radiographic) progression of KOA [44].

We deliberately used isometrics “force” (N) as an outcome, and not the moment (Nm), although the length of the lever arm between the force transducer and the joint center during the measurement was known. The reason for this was that the force transducer was positioned in an anatomically consistent position on the shank above the calcaneus, and that hence it can be assumed that the distance (lever arm) between the transducer and the joint center was approximately proportional to the distance between the quadriceps tendon and the joint center. Therefore, in order to estimate the isometric force generated by the muscle, we felt it was appropriate to use the force generated at the measurement transducer rather than the moment. Yet, a limitation of this study (and the OAI in general) is that only isometrics strength measurements were obtained, but not isokinetic ones, which have been frequently obtained in the literature and may be more reflective of limb function [45]

The two-year rate of change of quadriceps ACSA observed here in limbs with structurally progressive KOA (−2.8±7.9) was slightly greater than that reported by Beattie et al. [23] in a subsample (n=45) of KLG2–4 knees from the OAI cohort that was not specifically selected to display structural progression (−2.1±5.5%). The limbs without structurally progressive KOA, in contrast, showed somewhat smaller rates of change (−1.8±6.8%) to those observed by Beattie et al. [23]. The sensitivity to change (SRM) observed here in limbs with structurally progressive KOA (−0.35) was similar to that reported by Beattie et al. [23] for unselected KLG2–4 knees (−0.38) and for knees with risk factors of KOA, but without radiographic signs of KOA (−0.34). The two-year rate of quadriceps ACSAs in our study was of greater magnitude than that observed for total thigh muscle cross sectional areas in the Health ABC study [19], when accounting for the difference in the length of the observation interval. However, similar to our results, the Health ABC study reported a greater rate, but also greater variability, of muscle strength than muscle ACSA, indicating that muscle quality (strength per area) declines with age. In our study, progressor and non-progressor knees were matched for similar levels of baseline pain; however, the increase in WOMAC scores in progressor knees during the observation interval may have contributed to why the progressor limbs lost more quadriceps ACSA than those without progressive KOA, as the latter had a slight decline in WOMAC pain scores.

Measurement of (isometric) muscle strength is influenced by the ability or willingness to exert maximal voluntary muscle activation, of which both may be compromised in KOA [3,20,21,46]. Further biomechanical strength testing cannot be performed under all clinical situations, e.g. not immediately after trauma or surgery [27]. MRI-based analysis of muscle ACSAs appears to represent a potentially more objective, robust, and sensitive outcome than biomechanical measures of isometric strength, with the current data confirming less within-person variability in a longitudinal study. The application of MRI-based measurement of change in muscle (ACSA) is hence not limited to the study of knee OA, but may be useful in context of other basic research or clinical questions

In the Health ABC study, the within-participant test-retest reproducibility for extensor strength was reported to be around 11% (CV%) [19], whereas much smaller reproducibility errors have been described for MRI-based measurement of quadriceps muscle ACSAs (RMS CV%=1.7%) [41].. Please note that the acquisition protocol of the OAI covers only a distal portion of the thigh, in which the adductors have a small ACSA (Table 3) [40], and at which their ACSA may vary quite strongly with small differences in image position [40]. The non-significant trend of increasing adductor ACSAs during the observation period should therefore not be overestimated. Also, we observed a statistically significant reduction in flexor strength in knees without, but not in those with, structural progression: The inter-subject-variability of longitudinal change in flexor strength, however, was very high (8 times the mean change in progressors and 2.9 time the mean change in non-progressors), so that this may likely represent a chance finding which needs to be explored further in a larger cohort.

Previous work has suggested that greater baseline quadriceps strength may protect against the increase in joint space narrowing, i.e. worsening of KOA, in women, but not in men [1,12]. Our cross-sectional findings support this idea and are suggestive of smaller quadriceps ACSAs in female progressor than in non-progressor limbs, but not in men. Despite the relatively small sample size of the current pilot study, the difference in quadriceps ACSAs between female progressors and non-progressors reached borderline significance, whereas measures of strength discriminated less between both groups. These results encourage further exploration and use of MRI-based measures of muscle ACSAs (rather than only biomechanical measures of strength), as outcomes in epidemiological studies that look at the temporal and potentially causal relationship between muscle status and KOA, or in intervention studies that attempt to maintain muscle mass and strength, in order to prevent the deleterious consequences of sarcopenia in general [47].

CONCLUSION

In conclusion, MRI-based analysis of quadriceps ACSAs appears to be more sensitive to longitudinal change than biomechanical measurement of isometric extensor strength and is suggestive of greater rates of longitudinal decline in limbs with structurally progressive KOA than in non-progressive controls.