CT based muscle density predicts muscle function and health-related quality of life in patients with idiopathic inflammatory myopathies

Abstract

Objective

To investigate the association of low-density (lipid-rich) muscle measured by computed tomography (CT) with skeletal muscle function and health-related quality of life in idiopathic inflammatory myopathies (IIMs).

Methods

Seventeen patients and ten healthy controls underwent CT of the mid-thigh to quantify high (30-100HU) and low density (0-29HU) skeletal muscle areas. Anthropometric measures, body composition, physical activity level, health-related quality of life, skeletal muscle strength, endurance and fatigue were assessed. Patients were compared against controls. The relationship of anthropometric, body composition and disease variables with measures of muscle function were examined using Spearman’s test on the patient group. Linear regression was used to assess the age-and disease-adjusted relationship of muscle quality to physical function and muscle strength.

Results

Patients had higher body fat% (p=0.042), trunk fat mass (p=0.042), android/gynoid fat (p=0.033) and mid-thigh low density muscle/total muscle area (p<0.001) compared to controls. Mid-thigh low density muscle/total muscle area was negatively correlated with self-reported physical function, strength and endurance; the SF-36 physical functioning (p=0.004), manual muscle testing (p=0.020), knee maximal voluntary isometric contraction/thigh mineral free lean mass (p<0.001) and the endurance step test (p<0.001), suggesting that muscle quality impacts function in IIM. Using multiple linear regression adjusted for age, global disease damage, and total fat mass, poor muscle quality as measured by mid-thigh low density muscle/total muscle area was negatively associated with SF-36 physical functioning (p= 0.009).

Conclusion

Mid-thigh low density muscle/ total muscle area is a good predictor of muscle strength, endurance and health-related quality of life as it pertains to physical functioning in patients with IIMs.

Introduction

Idiopathic inflammatory myopathies (IIM) are a group of systemic autoimmune diseases that affect skeletal muscle, resulting in proximal muscle weakness. IIMs are generally classified as polymyositis (PM), dermatomyositis (DM) and sporadic inclusion body myositis (sIBM). IIMs can also occur in association with other connective tissue diseases [overlap myositis (OM)]. PM, DM and OM are immune-mediated while sIBM is thought to be related to abnormal accumulation of proteins that leads to an inflammatory response. Despite improved survival of patients with inflammatory myopathies since the 1960s (1), a significant proportion of these patients have considerable morbidity due to damage caused by both the disease and its treatment (2).

Skeletal muscle strength is a function of both muscle mass (quantity) and specific force (quality). In patients with IIMs, the etiology of muscle weakness and reduced endurance is thought to be multifactorial and, largely influenced by patient specific characteristics, such as age, sex, race, pre-morbid health and co-morbidities; disease specific variables, such as disease subtype, activity and duration; and previous therapies with corticosteroids and immunosuppressants (3). In this context, factors intrinsic to muscle also play an important role. Alterations in muscle fiber type composition have been associated with persistent muscle weakness in patients with chronic DM/ PM (4). Inflammatory infiltrates, muscle atrophy and replacement of muscle tissue by fat or fibrosis can lead to loss of contractile area; although these changes are not always observed in the muscles of all patients with DM/ PM who have significant muscle weakness (4).

Computed tomography (CT)-derived measure of low density skeletal muscle area is a surrogate marker of muscle quality, and it reflects lipid-rich skeletal muscle (5). Low muscle density has been shown to be associated with increased disability and physical function impairment in patients with RA (6), but there are no published studies investigating this association in IIMs. We hypothesized that low density muscle measured by CT is also a characteristic of IIM, and that it is negatively associated with muscle function and health-related quality of life parameters.

Methods

Patients with IIM, 18 years of age or older, regularly followed in the Rheumatology and Neurology Clinics at the University of Kentucky between 2006 and 2012 were identified using physician service claim diagnosis, provided as International Classification of Diseases, Ninth Revision (ICD-9-CM) codes 710.3, 710.4 and 359.7. In order to increase our specificity of IIM case ascertainment, cohort members were further required to meet inclusion criteria for definite or probable PM/ DM or sIBM according to the Bohan and Peter criteria (7, 8) and Griggs’ criteria (9), respectively. For OM, cohort members had to meet criteria for their primary connective tissue disease and satisfy probable/ definite criteria for myositis (10). The healthy controls were recruited via advertisements and word-of-mouth from the general population to be comparable to the age, gender, weight and BMI of the patient group, although patients and controls were not individually matched. All experimental procedures were performed in accordance with the Institutional Review Board. Written informed consent was obtained from each subject prior to participation.

Demographic and clinical data (including medical history, physical exam findings, laboratory results and medications) were obtained directly from patients by a single study physician (B.Y.H.). Global disease activity and damage were assessed using visual analogue scales (VAS), ranging from 0-10 centimeters, on the Myositis Disease Activity Assessment Tool (MDAAT) and the Myositis Disease Damage Index (MDI) (11), respectively. Functional status was assessed using the 1991 revised American College of Rheumatology (ACR) classification criteria of functional status (ACR-FS) in Rheumatoid Arthritis (12). Uses of oral prednisone, intravenous methylprednisolone, methotrexate, azathioprine, hydroxychloroquine, mycophenolate mofetil, rituximab, intravenous gamma globulin(IVIG), tacrolimus, cyclosporine and adalimumab by patients during the observation period were recorded as “never”/ “ever”/ “current”. Cumulative prednisone dose per treatment month was calculated by dividing cumulative prednisone dose (in milligrams) by the number of months the patient was on this medication. Cumulative prednisone dose per study period was obtained by dividing the cumulative prednisone dose (in milligrams) by total study observation period in months. Disease course was defined as monocyclic, chronic polycyclic or chronic continuous, according to criteria used by Bronner et al. (13). In cases where the onset of illness was less than 2 years, disease course was classified as undefined (14). Study data were collected and managed using REDCap (15).

Participants underwent manual muscle testing (MMT-8) and muscle endurance testing using the Functional Index-2 (FI-2) (16). Frequency, intensity, type, and duration of weekly exercise were assessed using the International Physical Activity Questionnaire (IPAQ) (17). Participants also completed the Short-Form Health Survey Questionnaire, version-2 (SF-36v2), a widely used tool that measures quality of life, well-being and functional health of general and specific populations, including adult DM, PM and sIBM patients (11).

Maximal isometric strength testing

Strength was measured by determining each subject’s maximal voluntary isometric contraction (MVIC) using a dynamometer (Biodex System4 Quick-Set). MVIC in the right arm and leg (in the arm at an elbow angle of 90° and in the leg at a knee angle of 90°) were determined with the subject in a seated position (seat angle at 85°), and with the lateral elbow and femoral epicondyles aligned to the center of the dynamometer shaft, respectively. In order to minimize the use of muscles other than the knee extensors and elbow flexors, subjects were stabilized with shoulder straps and a waist strap. Three maximal practice trials were performed, followed by 3 maximal test trials. MVIC force was recorded as the highest force generated over 3 trials held for 4 seconds, with a 3-minute rest period between attempts. MVIC measurements were corrected for mineral-free lean mass.

Fatiguing exercise protocol

Immediately after completing the maximal isometric strength testing, participants started the fatiguing exercise protocol as previously described (18). Percent losses of MVIC immediately following the fatigue protocol (% fall MVIC post) and after 12 minutes of recovery (% fall MVIC recovery) were used as measures of fatigability.

Dual energy x-ray absorptiometry (DXA) scanning

Total and regional fat and lean mass were measured in each participant via DXA scan performed by study personnel trained in this procedure. Each subject underwent a total body and dominant proximal hip DXA scan using a Lunar Prodigy (GELunar Inc., Madison,WI) bone densitometer prior to any physical activity. All scans were analyzed by a certified investigator using the GE Lunar software version 10.0. Total body fat (kg), body fat percentage, total body mineral-free lean mass (kg), trunk fat (kg), appendicular fat (kg) and appendicular mineral-free mass(kg) were assessed. Fat-free mass index (kg/m²) and fat mass index (kg/m²) were calculated by dividing body fat-free mass and fat mass respectively, by body height squared.

Computed tomography (CT) of the thigh

Each participant underwent a CT scan of his or her right and left quadriceps muscle at the University of Kentucky by trained technologists. With the subject supine, 10mm thick cross-section scans of both legs were taken at the midpoint of the femur using a Siemens Sensation 40-slice scanner. CT images were used to quantify high and low density skeletal muscle and fat cross-sectional area of the right and left thighs of each subject using 100mA with a scanning time of 3s and a 512 × 512 matrix. The attenuation of computed tomography radiation is determined by the chemical composition of the tissue being imaged (5). Skeletal muscle and adipose tissue have widely different attenuation values on CT. Since attenuation values for skeletal muscle are positive and attenuation values for adipose tissue are negative, skeletal muscle with relatively lower attenuation contains proportionally more adipose tissue (19). Therefore, intramuscular adipose tissue corresponds to the area in the attenuation range of 0-29HU, and muscle area of normal fat content (high density skeletal muscle) corresponds to the area in the attenuation range of 30-100HU (20). All CT image segmentation and quantification were performed by the same blinded technician at the University of Kentucky using Image J software (21).

Laboratory assessment

Blood samples were collected and the samples were analyzed for CK (U/L; normal range 38-176) and aldolase (U/L; normal range < 8.1) levels in the medical center clinical lab.

Statistical analysis

Continuous variables were described with means/ standard deviations and medians/interquartile ranges, according to whether the variables were approximately normally distributed. Categorical variables were described with counts and percentages. Comparisons of continuous variables between patients and controls were made using either a parametric T-test or a nonparametric rank-sum test, while comparisons of categorical variables between patients and controls were made using either a chi-square or Fisher’s exact test. Spearman’s correlations were used to examine the relationships of age, anthropometric disease measures, prednisone use, body composition, muscle quality and global disease activity/damage with selected functional outcomes among patients.

Because a comparison of patients and controls on lean mass may be vulnerable to confounding effects from adiposity, we also fit an ANCOVA model in which two regression lines were computed to relate total lean mass to total fat mass, one for patients and one for controls. We then performed an F-test to assess whether the computed regression lines were significantly different from each other, and judged accordingly whether patients and controls differed significantly on total lean mass when accounting for total fat mass. We proceeded similarly to assess for possible confounding effects from age. We refer the reader to Tschop et al. (22) for an explanation on the use of ANCOVA modeling to address possible confounding in two-group comparisons involving biological data.

We also fit multivariable regression models for two functional outcomes, namely SF-36 physical functioning and knee extensor MVIC/ thigh mineral-free lean mass. Both of these functional outcomes were significantly correlated with age, and we wished to analyze how these functional outcomes related to global disease damage while adjusting for age as well as two body composition characteristics: the ratio of mid-thigh low-density muscle to total muscle area (because this was also significantly correlated with both outcomes), and total fat mass (due to concern about possible confounding effects of adiposity). In addition to estimating standardized regression coefficients for the various predictor variables and obtaining corresponding p-values, we also recorded R² and adjusted R² for each model as indices of how much variability in the functional outcomes was explained by the predictor variables.

Analyses were completed using SAS software (Version 9.3) and JMP 10.0 software. Statistical significance was defined by a p-value < 0.05.

Results

Participant Characteristics

We invited patients from a cohort of 102 patients with IIM to participate in the study. Fifty four (53%) responded and 1 (17%) completed the study visit with the CT of the thigh. The main reasons for refusal to participate were physical inability and travel distance. Patients who were included in the study were more frequently male, older, had longer duration of active disease from diagnosis and suffered more from the SIBM subtype compared to the patients that were not enrolled in the study. We also identified and enrolled ten healthy controls that were selected to be comparable to the age, gender, weight and BMI of the patients enrolled in the study.

Histopathological subtypes among patients were DM (n=6), sIBM (n=5), OM (n=3), and PM (n=3). The median disease duration was 4.4 years (2.4-12.9). As for clinical course, 6/17 patients had chronic polycyclic course (35%), 6/17 had chronic continuous course (35%), 2/17 had monocyclic course (12%) and 3/17 (18%) had a yet undefined course. The median global disease activity on a visual analogue scale (VAS) was 0.8 cm (0.65-4.52).The median global disease damage on a visual analogue scale (VAS) was 2.05cm (0.57-4.00). Median CK and aldolase values were 193.0U/L (61.5-574.5) and 7.6U/L (3.6-8.7). Extra-muscular manifestations reported in order of prevalence were dysphagia (47%); arthritis (41%); interstitial lung disease and dysphonia (18%); arrhythmia and erythroderma (12%); and pericarditis, cutaneous ulceration and calcinosis (6%). Functional status was well maintained in less than one quarter of patients (23.5%). Mild decline of functional status was observed in 17.6% of patients, and the overwhelming majority (58.8%) suffered from moderate-to-severe functional impairment. Hypertension (n=6) and hyperlipidemia (n=3) were the most common co-morbidities among patients. One patient suffered from diabetes mellitus. The most common immunomodulatory agents used by patients at the time of the assessment for the treatment of IIM were prednisone (59%), methotrexate (41%), hydroxychloroquine (18%), azathioprine (12%), mycophenolate mofetil (12%) and rituximab (12%). More than half of the patients were on prednisone at the time of the study. Methotrexate use was also prevalent among patients. The mean dose of prednisone was 2.8 mg/day (1.3). The median cumulative prednisone dose per treatment period and cumulative dose per study period were 92.73 mg/ month (17.33-675) and 50mg/month (12-190.16), respectively.

Body composition in patients and controls

Demographic, anthropometric and body composition characteristics among patients and controls are shown in Table1. Patients and controls were not significantly different in terms of age, sex, race and anthropometric measures (height, weight and BMI). The racial composition of the study group was predominantly whites. Mean BMI among patients and controls were in the obese and overweight categories, respectively. Patients had higher body fat percentage (p=0.042), trunk fat mass (p=0.042) and android/ gynoid fat ratio (p=0.033) as compared with controls. Measures of lean tissue mass by DXA were not significantly different between patients and controls. Using ANCOVA modeling, we concluded that patients and controls did not differ significantly on total lean mass while accounting for total fat mass, although there was a trend toward significance (p=0.074). More specifically, ANCOVA modeling yielded an estimated relationship of total lean mass = 56.9 – 0.32 total fat mass for controls versus total lean mass = 30.0 + 0.35 total fat mass for patients. We also discovered that patients and controls did not differ significantly on total lean mass while accounting for age (p=0.210). Moreover, we obtained an estimated relationship of total lean mass=76.0 – 0.57 age for controls versus total lean mass=46.0 – 0.05 age for patients. When evaluated by CT, mid-thigh high density muscle area (p=0.017) was significantly decreased, while mid-thigh low density muscle/ total muscle area (p<0.001) was significantly increased among patients. Figures1A and 1B illustrate the differences between CT images of the mid-thigh of a healthy control and a patient with sIBM.

Table 1.

Demographic, anthropometric and body composition characteristics of the study cohort

	Patients (N=17)	Mean (SD) or Median (IQR) Control (N=10)	P-value
Demographic Characteristics
Age	55.55 (17.26)	49.22 (10.57)	0.306
Female sex, no. (%)	11 (65)	7 (70)	0.317
Race
Caucasian, no. (%)	16 (94)	10 (100)	0.630
Anthropometric parameters
Height, m	1.68 (0.10)	1.70 (0.08)	0.522
Weight, kg	85.40 (19.66)	79.93 (11.88)	0.356
BMI, kg/m2	30.51 (7.22)	27.29 (3.57)	0.202
DXA-derived measures
Total body fat, kg	38.70 (24.33-46.43)	27.74 (20.53-36.82)	0.060
Body fat percentage	0.44 (0.38-0.52)	0.37 (0.26-0.45)	0.042*
Total body mineral free lean mass, kg	40.85 (35.65-48.25)	44.28 (40.15-59.41)	0.219
Trunk fat, kg	22.70 (13.21-26.07)	15.73(10.61-18.61)	0.042*
Thigh fat, kg	3.44 (2.61-5.08)	2.77 (2.10-4.33)	0.183
Thigh lean, hg	4.16 (3.48-4.81)	4.88 (4.17-6.30)	0.075
Android/ gynoid fat	0.60 (0.49-0.74)	0.47 (0.36-0.56)	0.033*
CT-derived measures
Total mid-thigh area, cm2	330.20 (311.05-492.70)	324.42 (308.67-467.56)	0.498
Mid-thigh high density muscle area, cm2	113.32 (74.76-146-68)	176.37 (124.00-222.55)	0.017*
Mid-thigh fat area, cm2	166.92 (145.94-323.75)	151.63 (88.75-198.77)	0.126
Mid-thigh low density muscle area, cm2	33.48 (22.24-44.58)	25.71 (18.51-37.54)	0.093
Mid-thigh low density muscle/total muscle area	0.24 (0.08)	0.13 (0.03)	<0.001***
Mid-thigh % SAT	0.50 (0.45-0.63)	0.46 (0.21-0.54)	0.152

^(*)p-value <0.05,

^(**)p-value <0.01,

^(***)p-value <0.001

Figure 1.

Differences in low density muscle content are depicted in the CT images of the mid-thigh as variations in gray contrasting color. A, 32-year-old healthy control, in whom the mid-thigh low density muscle area and low density muscle/ high density muscle measured 19.68 cm² and 0.10, respectively. B, 64-year-old patient with sIBM, in whom the mid-thigh low density muscle area and low density muscle/ high density muscle measured 46.67 cm² and 0.59, respectively.

Measures of health and muscle function among patients and controls

Patients reported significantly lower levels of physical activity compared to controls, measured by the total physical activity score (p=0.025), as shown in Table2. Patients also reported spending significantly less time walking (p=0.042) and performing moderate physical activity (p=0.012). In terms of health related quality of life measures, several dimensions of the SF-36, including physical functioning (p=0.002), role-physical functioning (p=0.002), social functioning (p=0.015), vitality (p=0.005) and general health perceptions dimensions (p=0.008), were significantly lower among patients. Patients demonstrated significantly greater skeletal muscle weakness [i.e. lower MMT-8 total score (p<0.001) and knee MVIC/ thigh mineral-free lean mass (<0.001)] and decreased muscle endurance in the lower extremities, measured by the maximum number of repetitions on the FI-2 [i.e. hip flexion (p=0.015), step test (p<0.001), heel lift (p=0.005) and toe lift(p=0.013)]. Among patients, the three weakest muscles in the MMT-8 were the gluteus maximus, gluteus medius and neck flexors. Patients had decreased muscle endurance in the neck flexors (FI-2 neck flexors) compared to controls, although this difference was not statistically significant. Patients did not differ statistically from controls in terms of muscle strength in the elbow flexors (elbow flexor MVIC/ arm mineral-free lean mass); muscle endurance in the upper extremities (maximum number of repetitions on the FI-2 shoulder flexion and abduction); or skeletal muscle fatigability (%knee extensor MVIC fall).

Table 2.

Comparison of health status and muscle function measures in patients with IIM versus healthy controls

	Patients	Mean (SD) or Median (IQR) Control	P-value
IPAQ	(N=15)	(N=10)
Total leisure MET, min/ wk	231 (0-1314)	963.75 (508.5-1257.75)	0.207
Total walking MET, min/wk	462 (0-2178)	1889.25 (680.62-5234.63)	0.042*
Total moderate MET, min/ wk	1080 (180-2040)	2820 (1815-4987.5)	0.012*
Total vigorous, min/ wk	0 (0-0)	240 (0-1140)	0.241
Total physical activity score	2238 (840-4540)	6432 (2529-9364.5)	0.025*
SF-36	(N=17)	(N=10)
Physical	45 (25-75)	97.5 (88.75-100)	0.002**
Role-physical	43.75 (25-71.87)	100 (82.81-100)	0.002**
Bodily pain	77.5 (45-95)	95 (84.37-100)	0.080
Social	75 (37.5-100)	100 (100-100)	0.015*
Mental health	85- (75-92.5)	85 (75-91.25)	0.959
Role-emotional	91.67 (75-100)	100 (91.67-100)	0.256
Vitality	43.75(25-62.5)	78.12 (65.62-87.5)	0.005**
General health perceptions	45 (32.5-65)	87.5 (71.25-95)	0.008**
MMT-8	(N=16)	(N=10)
	136.5 (115.5-145.5)	150 (150-150)	<0.001***
Fatigue Protocol - derived measures	(N=15)	(N=10)
Elbow flexor Pre-MVIC/ arm mineral free lean mass (Nm/kg*103)	1.62 (1.14)	2.18 (0.76)	0.185
Knee extensor Pre-MVIC/ thigh mineral free lean mass (Nm/kg*103)	17856.66 (9697.05)	34626.56 (8442.52)	<0.001***
% Knee extensor MVIC fall post	−29.51 (13.26)	−22.48 (10.47)	0.173
% Knee extensor MVIC fall recovery	−17.51 (13.10)	−15.40 (11.61)	0.684
Functional Index-2
	(N=17)	(N=10)
Shoulder flexion	45 (11.5-60)	60 (40-60)	0.152
Shoulder abduction	49 (20-60)	60 (6-30)	0.175
Head lift	20 (10-55)	59.5 (27.5-60)	0.062
Hip flexion	19 (2.5-34.5)	46 (23.75-60)	0.015*
	(N=16)	(N=10)
Step test	0 (0-32.5)	60 (56.5-60)	<0.001***
Heel lift	25.5 (12-92.25)	120 (107.5-120)	0.005**
Toe lift	25.5 (3.75-57.5)	94 (41-120)	0.013*

^(*)p-value <0.05,

^(**)p-value <0.01,

^(***)p-value <0.001

Correlations of body composition, age, disease measures and prednisone use with functional outcomes in IIM patients

The proportion of low-density muscle relative to total muscle area in the mid-thigh was negatively correlated with four measures of self-reported physical function, strength and endurance: the SF-36 physical functioning (ρ=−0.68, p=0.004), MMT-8 (ρ=−0.57, p=0.020), knee MVIC/ thigh mineral free lean mass (ρ=−0.81, p<0.001) and the FI-2 step test (ρ=−0.83, p<0.001), as shown in Table3, suggesting that muscle quality impacts function in IIM. Total body fat (ρ=−0.57, p=0.026), fat-mass-index(ρ=−0.53, p=0.041), body fat percentage(ρ=−0.55, p=0.034) and trunk fat (ρ=−0.59, p=0.021) were negatively correlated with muscle fatigability, measured by percent fall in strength during the fatiguing exercise protocol.

Table 3.

Correlations of age, anthropometric and disease measures, prednisone use and body composition with functional outcomes in IIM patients

	IPAQ moderate MET	SF-36 physical functioning	MMT-8	Knee extensor MVIC/ thigh mineral lean free mass	FI-2 Step test	% Fall MVIC Post
	ρ (p)	ρ (p)	ρ (p)	ρ (p)	ρ (p)	ρ (p)
Age, years	−0.01 (0.970)	−0.58 (0.015*)	−0.45 (0.076)	−0.57 (0.016*)	−0.42 (0.106)	0.35 (0.196)
Anthropometric Parameters
Weight, kg	−0.33 (0.234)	0.10 (0.708)	0.09 (0.736)	−0.10 (0.687)	0.19 (0.484)	−0.34 (0.216)
BMI, kg/m2	−0.31 (0.262)	0.12 (0.652)	0.33 (0.215)	0.14 (0.580)	0.21 (0.425)	−0.49 (0.066)
Disease characteristics
Disease duration, months	−0.07 (0.815)	−0.42 (0.104)	−0.19 (0.476)	−0.39 (0.122)	−0.32 (0.229)	0.39 (0.147)
CK, U/L	0.01 (0.979)	−0.09 (0.788)	−0.48 (0.110)	−0.46 (0.112)	−0.37 (0.230)	0.21 (0.537)
Aldolase, U/L	0.09 (0.803)	0.02 (0.958)	−0.35 (0.289)	−0.29 (0.354)	−0.43 (0.189)	0.53 (0.117)
Global disease activity (VAS), cm	−0.28 (0.305)	−0.21 (0.443)	−0.02 (0.930)	0.19 (0.462)	−0.12 (0.650)	−0.17 (0.555)
Global disease damage (VAS), cm	−0.21 (0.467)	−0.41 (0.132)	−0.66 (0.008**)	−0.76 (<0.001***)	−0.52 (0.048*)	0.53 (0.053)
Prednisone use
Cumulative prednisone dose, mg	0.08 (0.794)	−0.11 (0.695)	0.05 (0.862)	0.11 (0.684)	−0.23 (0.431)	0.07 (0.816)
Cumulative prednisone dose/ treatment period mg/ months	0.20 (0.508)	−0.07 (0.798)	−0.04 (0.892)	0.11 (0693)	−0.24 (0.411)	0.05 (0.872)
Cumulative prednisone dose/ observation period, mg/ months	0.20 (0.502)	−0.12 (0.684)	−0.04 (0.886)	0.09 (0.741)	−0.29 (0.305)	−0.03 (0.929)
DXA-derived measures
Total body fat, kg	−0.32 (0.242)	0.05 (0.854)	0.14 (0.596)	−0.02 (0.933)	0.06 (0.811)	−0.57 (0.026*)
Fat mass index, kg/m2	−0.34 (0.213)	0.02 (0.939)	0.20 (0.455)	0.01 (0.970)	0.04 (0.895)	−0.53 (0.041*)
Body fat percentage	−0.35 (0.205)	−0.18 (0.502)	0.04 (0.888)	−0.12 (0.639)	−0.26 (0.330)	−0.55 (0.034*)
Total body mineral free lean mass, kg	−0.20 (0.624)	0.36 (0.164)	0.21 (0.429)	−0.04 (0.881)	0.44 (0.086)	−0.03 (0.919)
Fat-free mass index, kg/m2	−0.20 (0.470)	0.40 (0.119)	0.49 (0.051)	0.20 (0.451)	0.54 (0.030*)	−0.19 (0.499)
Trunk fat, kg	−0.19 (0.494)	0.16 (0.552)	0.14 (0.604))	0.06 (0.830)	0.15 (0.580)	−0.59 (0.021*)
Android/ gynoid fat ratio	−0.23 (0.412)	0.10 (0.712)	−0.05 (0.866)	−0.06 (0.823)	0.03 (0.924)	−0.20 (0.467)
CT-derived measures
Total mid-thigh area, cm2	−0.51 (0.053)	0.02 (0.939)	0.14 (0.592)	−0.03 (0.911)	0.03 (0.915)	−0.17 (0.541)
Mid-thigh high density muscle area, cm2	0.06 (0.830)	0.80 (0.004**)	0.66 (0.005**)	0.85 (<0.001***)	0.80 (0.004**)	−0.14 (0.630)
Mid-thigh fat area, cm2	−0.67 (0.006**)	−0.30 (0.253)	−0.08 (0.781)	−0.33 (0.198)	−0.41 (0.130)	−0.19 (0.491)
Mid-thigh low density muscle area, cm2	0.00 (1.000)	0.037 (0.892)	0.09 (0.732)	0.06 (0.815)	−0.03 (0.580)	0.11 (0.704)
Mid-thigh low density muscle/ total muscle area	−0.24 (0.397)	−0.68 (0.004**)	−0.57 (0.020*)	−0.81 (<0.001***)	−0.83 (<0.001***)	0.06 (0.830)
Mid-thigh % SAT	−0.49 (0.063)	−0.30 (0.255)	−0.04 (0.879)	−0.39 (0.119)	−0.49 (0.061)	−0.33 (0.226)

^(*)p-value <0.05,

^(**)p-value <0.01,

^(***)p-value <0.001

Several patient characteristics were negatively associated with poor functional outcomes, as might be expected. Age was negatively correlated with self-reported physical function and leg strength (SF-36 ρ=−0.58, p=0.015; knee-extensor MVIC corrected for thigh mineral-free lean mass (ρ=−0.57, p=0.016). Similarly, global disease damage (VAS) was negatively correlated with several direct measures of muscle strength (total MMT-8 score ρ=−0.66, p=0.008, knee-extensor MVIC/thigh mineral-free lean mass (ρ=−0.76, p<0.001); endurance as quantified as the number of repetitions on the FI-2 step test(ρ=−0.52, p=0.048). We did not observe significant correlations of global disease activity, CK and aldolase blood levels or prednisone use with impairment of skeletal muscle function.

Correlations of mid-thigh muscle density by CT and global disease activity and damage

Low muscle quality measured by mid-thigh low-density muscle/ total muscle area was significantly correlated with global disease damage (ρ=−0.69, p=0.003), but not with global disease activity.

The multivariable linear regression model for SF-36 physical functioning had R²= 65.4% and adjusted R²= 51.6%. The estimated standardized coefficients for age, global disease damage, ratio of mid-thigh low-density muscle to total muscle area, and total fat mass were 0.514 (p=0.154), 0.531 (p=0.208), −1.650 (p=0.009), and 0.895 (p=0.033), respectively. To interpret these results, consider, for example, two hypothetical patients who differed by one standard deviation on the aforementioned ratio but were otherwise similar; these patients would be expected to differ by an estimated 1.650 standard deviations on SF-36 physical functioning, with the patient having the higher ratio exhibiting lesser physical functioning. For knee extensor MVIC/ thigh mineral-free lean mass, the multivariable regression model had R²= 70.2% and adjusted R²= 58.2%. The estimated standardized coefficients for age, global disease damage, ratio of mid-thigh low density muscle to total muscle area, and total fat mass were 0.007 (p=0.982), −0.185 (p=0.605), −0.691 (p=0.128), and −0.047 (p=0.875), respectively.

Because most of the estimated standardized coefficients were not statistically significant, and since our sample size was rather modest to support having four predictor variables in a model, we also applied a backward elimination algorithm to each multivariable linear regression model, whereby insignificant predictor variables were dropped one at a time until all remaining predictor variables were significant. When we did so, we found in both cases that the only remaining predictor variable was the ratio of mid-thigh low-density muscle to total muscle area (R²=52.7%, adjusted R²=49.3%, estimated coefficient = −0.726, p=0.002 for the first model [SF-36 physical functioning]; R²=70.5%, adjusted R²=68.4%, estimated coefficient=−0.839, p<0.001 for the second model [knee extensor MVIC/ thigh mineral-free lean mass]).

Discussion

In this study, we investigated the relationship between body composition, muscle quality and muscle function in patients with IIM. Poor muscle quality, as measured by mid-thigh low-density muscle/ total muscle area, was significantly and negatively associated with physical functioning after adjusting for age, global disease damage, and total fat mass; poor muscle quality was also significantly and negatively associated with muscle strength after backward elimination was applied to a regression model adjusting for the aforementioned potential confounders. To our knowledge, this is the first report of its kind in this patient population. These results suggest that skeletal muscle quality, and not quantity, may better predict muscle function among patients with IIMs. Increased skeletal muscle fat content, which is likely to account for low density muscle, may contribute to loss of specific force in IIMs, which has been reported to occur in age-related skeletal muscle wasting and rotator cuff tear (23, 24).

Patients did not differ statistically from controls in terms of skeletal muscle fatigability. This finding is consistent with a previous report in which muscle fatigability assessed by a non-volitional method was not statistically different among IIM patients and controls (25). However, we observed that among IIM patients, skeletal muscle fatigue was significantly correlated with total and trunk adipose tissue mass. This observation is in line with previous reports that obesity is associated with decreased, fatigue resistant, oxidative myofiber frequency (26). We also found that patients had decreased muscle endurance in the lower extremities, which has been previously reported among patients with PM and DM, and linked to low proportion of oxygen-dependent type I fibers compared with healthy individuals (27).

Total body fat mass, body fat percentage, trunk fat mass and ratio of android/ gynoid fat were significantly higher among IIM patients compared to controls, although these groups did not differ significantly in weight or BMI. Adiposity has been previously shown to be negatively associated with physical function, particularly in older adults (28). One study showed that while the amount of mineral-free upper leg mass was the main contributor to physical function in young women aged 20-30 years; in older women aged 64-80 years, relative adiposity and age were the strongest contributors to physical function (29). In our study, while patients and controls were comparable in terms of lean mass, patients had a statistically significantly higher proportion of mid-thigh low density muscle to total muscle area. These results indicate that IIM patients suffer from metabolic derangement leading to excess fat accumulation and loss of functional lean mass. This is similar to what has been described in rheumatoid cachexia, where muscle loss is often accompanied by increased fat mass and stable weight (30). Rheumatoid cachexia has been linked with increased disability and premature mortality in rheumatoid arthritis(31). Physical inactivity and chronic inflammation have been implicated as underlying mechanisms in rheumatoid cachexia (32), which are also likely operative in IIMs. Patients in our study reported significantly lower levels of physical activity compared to controls. TNF-α is an inflammatory cytokine that is expressed in skeletal muscle of patients with IIMs (33, 34), and it also promotes cachexia via activation of nuclear factor kappa B (NF-κB) (35, 36). Among our cohort of IIM patients, global disease damage, a reflection of disease severity and cumulative inflammatory burden, correlated well with measures of muscle strength and endurance, and moderately (though not significantly) with muscle fatigue.

In addition to the effects of chronic inflammation on body composition, muscle quality and function, patients with IIMs are also susceptible to the effects of aging and glucocorticoid use on these parameters (37). Although glucocorticoids are the cornerstone of therapy in IIMs, they can also have potentially inhibitory effects on skeletal muscle mass, myogenesis and immune responses that promote skeletal muscle regeneration following muscle injury (38). In our study we did not observe a significant correlation between cumulative prednisone use and impairment of skeletal muscle function, but increased age was significantly negatively correlated with SF-36 physical functioning and leg muscle strength.

Low muscle density has been shown to be associated with insulin resistance in obese nondiabetic patients (39), but there are no published studies that have investigated this in patients with IIMs. While we did not assess participants for insulin sensitivity, we suspect insulin resistance is highly prevalent among IIM patients due to glucocorticoid use, physical inactivity, obesity and chronic inflammation. Insulin is a potent anabolic hormone, and muscle atrophy of aging and type2 diabetes mellitus have been associated with insulin resistance (40).

Our study has several limitations and raises some interesting questions for further research. First, our study is cross-sectional, examining the relationship of muscle quality with muscle function, and therefore does not prove causation. Second, skeletal muscle involvement in IIMs can be patchy, but we only assessed muscle quality at the mid-thighs. Third, our sample size is modest; therefore small or even medium associations may not have been detectible. Fourth, although our multivariable modeling attempted to address possible confounding effects, our sample sizes were rather modest for detecting statistical significance in multivariable settings; thus, confounding cannot be definitively ruled out.

We have shown that mid-thigh low density muscle/ total muscle area predicts muscle strength, endurance and health-related quality of life; while retention of high density muscle is associated with better outcome with regard to physical functioning. We also demonstrated derangements in body composition and muscle quality among IIM patients, which could be contributing factors of muscle dysfunction in these patients. Additional biochemical and histopathological studies are needed to characterize low muscle density in order to determine the utility of therapeutic strategies aimed at improving muscle density.

Significance and Innovation.

• Poor muscle quality, as measured by the proportion of mid-thigh low density muscle relative to total muscle area, was significantly and negatively associated with physical functioning after adjusting for age, global disease damage, and total fat mass.

• Dual energy x-ray absorptiometry (DXA) derived measure of lean mass did not yield statistically significant correlations with muscle function. These results suggest that skeletal muscle quality, and not quantity, may better predict muscle function among patients with IIMs.