Accuracy of ultrasound, bioelectrical impedance analysis and anthropometry as point‐of‐care measurements of skeletal muscle mass in patients with inflammatory bowel disease

Summary

Background

Disturbance of skeletal muscle mass has clinically important implications in patients with inflammatory bowel disease (IBD), but accurate quantification requires radiation‐intense techniques.

Aims

We aimed to compare point‐of‐care muscle assessments and their change with therapy with those using reference‐standard whole‐body dual energy X‐ray absorptiometry (DXA).

Methods

Adult patients with IBD and healthy controls underwent prospective assessment of muscularity by ultrasound of the dominant arm and both thighs, bioelectrical impedance analysis (BIA), anthropometric measurements, and DXA. Patients with active IBD were assessed again ≥13 weeks after initiating biologic induction therapy.

Results

In 54 patients with IBD and 30 controls, all muscle assessments correlated significantly with DXA‐derived skeletal muscle index (SMI). In IBD, ultrasound of the arm and legs had the best agreement with DXA‐derived SMI (mean difference 0 kg/m², 95% limits of agreement −1.3 to 1.3), while BIA overestimated DXA‐derived SMI by 1.07 (−0.16 to +2.30) kg/m². In 17 patients who underwent biologic therapy, the percentage change in DXA‐derived SMI correlated significantly with the percentage change in all other muscle assessment techniques. Responders (n = 9) increased SMI from baseline to follow‐up when derived from DXA (mean 7.8–8.5 kg/m², p = 0.004), ultrasound of the arm and legs (300–343 cm², p = 0.021) and BIA (9.2–9.6 kg/m², p = 0.011).

Conclusions

Ultrasound of the arm and legs out‐performed other point‐of‐care methods in its accuracy of measuring muscle mass. All methods, except mid‐arm circumference, were responsive to therapy‐induced change. Ultrasound is the preferred non‐invasive test for measuring muscle mass in patients with IBD.

Muscle thickness ultrasound of the arm and legs was the most accurate point‐of‐care measurement of skeletal muscle index, as determined by dual‐energy X‐ray absorptiometry, in patients with inflammatory bowel disease.

1. INTRODUCTION

Altered body composition has been observed in patients with inflammatory bowel disease (IBD). ¹ Low muscle mass, or skeletal muscle mass index (SMI), has been associated with poor quality of life, increased need for surgery, poorer response to therapy and post‐surgical complications in patients with IBD. ² , ³ , ⁴ Nutritional optimisation and muscle strengthening, in addition to controlling the underlying inflammation, can correct low muscle mass and prevent such complications occurring. Timely detection is, therefore, of great potential importance. ⁵ However, methods used in routine clinical practice such as calculating body mass index (BMI) or applying malnutrition screening tools to identify patients at‐risk of low muscle mass can be misleading. ⁶ BMI cannot distinguish between levels of adiposity and muscle mass. Furthermore, application of reference body composition techniques, such as dual energy X‐ray absorptiometry (DXA), is costly, involves radiation and has limited accessibility with equipment housed in radiology departments. ⁷ Hence, there is a need for other muscle assessment techniques, which are portable, sensitive, safe and effective that can be used in a point‐of‐care clinical setting.

Ultrasound of large muscles, such as the quadriceps and biceps, has been increasingly studied in the identification of low muscle mass in healthy volunteers and elderly populations, as well as various pathologies such as lung disease and patients in intensive care units. ⁸ , ⁹ , ¹⁰ , ¹¹ However, there have been no studies evaluating the relationship between muscle mass assessed by ultrasound compared with a reference method such as DXA in an IBD cohort. Bioelectrical impedance analysis (BIA) is another bedside technique, which indirectly estimates body composition. It functions by a low‐voltage current passing through the body, where tissue resistance or impedance is then measured. Variables such as height, weight, age and sex are then used in empiric equations to estimate fat mass and fat‐free mass. ¹² BIA has been shown to correlate well with DXA‐measured muscle mass in patients with IBD, but agreement with DXA has differed between studies. ¹³ , ¹⁴ , ¹⁵ Finally, simple anthropometric measurements such as calf and mid‐arm circumference have also shown to correlate with muscle mass on DXA in middle‐aged and elderly individuals when performed by well‐trained professionals. ⁷ , ¹⁶ Limited studies in IBD have evaluated the utility of anthropometric measurements, especially in relationship to a reference method such as DXA. ¹⁷ , ¹⁸

The aims of this study were: (1) compare point‐of‐care muscle measurements (ultrasound, BIA, and anthropometry) to muscle mass assessed by a reference method (DXA) in patients with IBD and controls; (2) assess the ability of each point‐of‐care measurements to identify patients with low muscle mass assessed by DXA and (3) observe change in muscle mass over time, pre‐ and post‐biologic induction therapy, using all muscle measurements.

2. MATERIALS AND METHODS

2.1. Patients

Adult participants with IBD and sex‐ and age‐matched controls were recruited in this single‐centre, prospective, observational study between October 2020 and January 2022. Participants were excluded if they had other major medical conditions, including neuromuscular, rheumatological or orthopaedic conditions such as prosthetic joint replacements, malignancy or infection, gastrointestinal symptoms or diseases other than IBD, cognitive impairment, inability to provide informed consent in English, were currently pregnant or breastfeeding. Patient demographics were collected including date of birth, sex, disease activity and duration and all participants underwent fasting blood tests, including albumin and C‐reactive protein, at the time of assessment.

All participants underwent a one‐off assessment of body composition and a subgroup of patients with active IBD (faecal calprotectin, FCP ≥150 μg/g) were assessed again ≥13 weeks after initiating biologic induction therapy. Response to therapy was defined as FCP reduction of ≥50%. Concentrations of FCP were measured by Liaison® sandwich immunoassay (Diasorin Inc).

2.2. Whole‐body DXA

Whole‐body DXA (GE Lunar Prodigy) was used as the reference method for SMI measurement. All participants were fasted for at least 2 h prior with no water or other fluid included. SMI was calculated by summing all limb lean muscle mass (kg) and dividing by height squared (m²). Low SMI was defined as men ≤7.26 kg/m² and women ≤5.45 kg/m². ¹⁹

2.3. Ultrasound

Two trained investigators (A.L.N and M.B.) performed all muscle ultrasound imaging. A portable B‐mode ultrasound (Philips Lumify) with a multi‐frequency linear array transducer (4–12 MHz) was used for muscle measurements. Previously published protocols for landmarking, muscle and bone measurements were used in this study. ⁸ , ¹⁰

2.3.1. Mid‐upper arm bone length and landmarking

The bone length was measured from acromial process of the scapula to the olecranon process of the ulna and the half‐way point marked. ⁸ , ¹⁰ In supine position with the elbow extended, the patient's forearm was supinated and the shoulder abducted to 30°. All sonographic scanning was performed in this position at the half‐way point.

2.3.2. Bilateral thigh bone length and landmarking

In supine position, with knees extended and relaxed, the bone length for both thighs was measured between the anterior superior iliac spine and the upper pole of the patella. ⁸ , ¹⁰ All sonographic scanning was performed in this position at the half‐way point and two‐third point (closest to the patella) between these landmarks.

2.3.3. Ultrasound technique and imaging acquisition

The depth of the ultrasound for the arm was set at 4 cm and for the thighs at 7 cm. This was increased if the bone (humerus or femur) was unable to be visualised. Water‐soluble transmission gel was used between the transducer and skin, and minimal compression applied, so the skin remained convex on images ⁸ (Figure 1). Three images were saved at each landmark and uploaded to the National Institutes of Health (NIH) Image J software (version 1.52, US NIH) for analysis. The mean of the three measurements of each muscle thickness was used in the final analysis. For the arm and thighs, muscle thickness (cm) was measured from the subcutaneous adipose tissues‐muscle interface to the muscle‐bone interface of the humerus and femur, respectively. ⁸ , ¹⁰ , ²⁰ Muscle thickness ultrasound of the dominant arm was derived from the dominant arm muscle thickness (cm) multiplied by the humeral length (cm), resulting in a cm² value that was then used in the analysis. Muscle thickness ultrasound of the legs was the sum of the four points on the thighs (bilateral half‐way point and two‐thirds point) divided by 4, and multiplied by the right femur length, reported as cm². Muscle thickness ultrasound of the arm and legs was a sum of both previous calculations.

FIGURE 1.

Ultrasound images of the arm (left) and thigh (right). The line, labelled A, is the measure taken as the muscle thickness of the biceps and quadriceps, respectively.

Intra‐rater and inter‐rater reliability for the ultrasound protocol was assessed in five control participants. The protocol including landmarking five locations (dominant mid‐arm, right mid‐thigh, right 2/3‐thigh, left mid‐thigh and left 2/3‐thigh) and measuring the muscle thickness at each location was performed by two investigators (A.L.N and M.B.) consecutively on the same day. The first investigator used stick‐on labels for landmarks, which were then removed prior to the second investigator completing the muscle ultrasound protocol. All locations were included in the final statistical analysis.

2.4. Bioelectrical impedance analysis

A whole‐body, single‐frequency, segmental BIA device (Tanita BC‐418) was used to measure SMI in all participants. Participants were fasted for at least 2 h prior including no water or fluid, dressed in a hospital gown only and asked to empty their bladder prior to assessment. Participants stood with bare feet on the electrode panel and held electrodes in both hands, ensuring contact with thumbs and palms. Elbows were extended and arms abducted at 30°. Total body water and total fat‐free mass was calculated and recorded by the BIA device. A BIA‐predicted SMI was calculated by dividing the sum of upper and lower limb fat‐free mass (kg) by the square of the height (m²).

2.5. Anthropometric measurements

Measurements of height and weight (to the nearest 0.1 m and 0.1 kg, respectively) were performed with the patient in a hospital gown without footwear. BMI was calculated by weight (kg) divided by height squared (m²). Limb circumferences were measured with a cloth tape. Calf circumference was measured in standing position around the widest part of both calves and the mean between the two measurements recorded. Mid‐arm circumference was measured at half‐way point, as marked during the ultrasound protocol, while the arm was in a relaxed position and the patient seated. ²¹

2.6. Statistical analysis

All statistical analyses and graphs were performed with STATA 13 (StataCorp LLC) and GraphPad Prism 9 (Dotmatics). The a priori hypothesis tested was that low muscle mass was at least 5 times more common in patients with IBD than controls. ²² A power calculation determined a need for at least 20 participants to be able to reject the null hypothesis at a significance level of 0.05 (two‐sided) with a power of 0.80. Normality was tested with the Shapiro–Wilk test. Continuous variables were presented as mean and 95% confidence intervals (CI) or medians and interquartile (IQR) range depending on normality distribution. Categorical values as frequency and percentages. Differences between IBD patients and control demographics were compared using an unpaired t‐test or Wilcoxon rank sum test. Categorical values were compared with a chi‐squared test. Longitudinal assessments were compared using either paired two‐tailed t‐test or Wilcoxon signed‐rank test. The relationship between bedside muscle assessments and DXA‐derived SMI measurements were compared using either Pearson's or Spearman's correlation coefficients.

Linear regression was used to predict DXA‐derived SMI from ultrasound measures, calf circumference, mid‐arm circumference and BMI. A Bland–Altman plot analysis with 95% limits of agreement was then used to compare these predicted muscle assessments and BIA‐derived SMI to DXA‐derived SMI. ²³ Receiver‐operating characteristic analysis was used to determine a cut‐off value for each muscle assessment technique, by sex, for identifying low DXA‐derived SMI. A p ≤ 0.05 was considered significant.

The intra‐rater reliability and inter‐rater reliability were assessed using intra‐class correlation coefficient for the ultrasound protocol. Intra‐rater reliability was only performed for muscle thicknesses, which was repeated three times in each of the five locations. Inter‐rater reliability was performed for landmarking five locations and muscle thickness measurements at five locations. The following criterion was used to interpret intra‐class correlation coefficient: a value of 0.5–0.75 indicated moderate reliability, >0.75–0.90 indicated good reliability, and >0.90 indicated excellent reliability. ²⁴

2.7. Ethical considerations

All participants provided informed verbal and written consent. This study was approved by the Monash Health Ethics Research Committee (HREA 58293, local reference RES‐20‐034A).

3. RESULTS

3.1. Participants

For the cross‐sectional analysis, 54 patients with IBD and 30 age‐ and sex‐matched healthy controls were recruited. Of IBD patients invited to participate, three patients declined. Four patients did not complete their whole initial assessment due to emergency surgery (1), new diagnosis of diabetes (1) and inability to attend the appointment due to personal reasons (2). Thus, they were not included in the analysis. The IBD patients included 52 outpatients presenting for routine clinical review and two inpatients. Their demographic and clinical details are shown in Table 1. Of 54 patients with IBD, 56% had Crohn's disease, 44% had ulcerative colitis. Male patients comprised 61%, and median age was 30 (IQR 23–43) years. FCP was measured in 50 patients, of whom 28 had active (≥150 μg/g) and 22 inactive disease. Approximately one‐third of patients were on biologic therapy and a quarter were on corticosteroid‐based therapy at the time of the assessment. A higher proportion of patients with IBD had low DXA‐derived SMI, measured by DXA, compared to controls (19% vs 3%, p = 0.048). This included 7/54 (13%) of IBD patients within the normal BMI range. For men with IBD, 7/33 (21%) and for women with IBD, 3/21 (14%) had low SMI. Patients with IBD and controls had similar total body water (40.6 vs 39.2 kg, p = 0.849) measured by BIA, and differed in albumin levels (39 vs 40 g/L, p = 0.007).

TABLE 1.

Demographics and clinical data of patients with inflammatory bowel disease (IBD) and healthy controls. Active disease defined as faecal calprotectin ≥150 μg/g; four patients did not have faecal calprotectin available.

		All IBD (n = 54)	Controls (n = 30)	p‐value
Age (years)		30 (23–43)	35 (31–39)	0.115 d
Male sex, n (%)		33 (61%)	15 (50%)	0.324
Current smoking, n (%)		6 (12%)	2 (7%)	0.536
Disease duration (years)		5 (1–11)
Previous bowel resection, n (%)		6 (11%)
Diagnosis, n (%)
Crohn's disease, n (%)		30 (56%)
Location, n (%)
Ileal, colonic, ileocolonic		12 (40%), 5 (17%), 13 (43%)
Behaviour, n (%)
Inflammatory, fibrostenosing, penetrating		14 (47%), 4 (13%)
Perianal disease		12 (40%)
Ulcerative colitis, n (%)		24 (44%)
Extent, n (%)
Proctitis, left‐sided, pancolitis		2 (8%), 4 (17%), 18 (75%)
Disease activity, n (%)
Active		28 (52%)
Inactive		22 (41%)
Therapies, n (%)
5‐aminosalicylate		26 (49%)
Immunomodulators		26 (49%)
Steroids		14 (26%)
Biologics		19 (35%)
Laboratory findings
Serum albumin (g/L)		39 (36–41)	40 (39–42)	0.007 d
C‐reactive protein (mg/L)		3.6 (1.1–6)	0.6 (0.4–1.4)	<0.0001 d
Faecal calprotectin (μg/g)		443 (34–1210)
Body composition
Body mass index (kg/m2)		25 (22–27)	23 (22–26)	0.243 d
SMI a (DXA b ) (kg/m2)		7.4 (7–7.8)	7.7 (7.2–8.2)	0.388 e
Proportion with low SMI		10 (19%)	1 (3%)	0.048
Fat % (DXA)		33 (27–39)	28 (24–34)	0.022 e
Fat‐free mass (BIA c ) (kg)		56 (43–63)	54 (45–65)	0.859 d
Total body water (BIA) (kg)		40.6 (31.6–46.3)	39.2 (32.6–47.8)	0.846 d

^a Skeletal Muscle Index.

^b Dual energy X‐ray absorptiometry.

^c Bioelectrical impedance analysis.

^d Wilcoxon rank sum test.

^e Unpaired t‐test.

Seventeen patients with active IBD were studied before and at completion of biologic induction therapy. An additional eight patients did not complete their follow‐up assessment, due to an allergic reaction to the initial first biologic infusion and subsequent cessation of therapy (1), inability to attend the follow‐up appointment (5), failure to commence initial biologic infusion (1), and undergoing a knee replacement before follow‐up (1). These were not included in the pre‐ and post‐treatment analysis. Of the 17 patients included, nine were classified as responders and eight non‐responders. All responders were male, with a median age of 23 years, six had Crohn's disease and three had ulcerative colitis. Biologic therapies undertaken were infliximab (5/9), adalimumab (1/9), vedolizumab (2/9) and ustekinumab (1/9). Median FCP reduced from 1500 to 79 μg/g (p = 0.012) and C‐reactive protein reduced from 4.2 to 0.9 mg/L (p = 0.018) from baseline to follow‐up. Non‐responders were 63% men, median age of 26 years, five had Crohn's disease and three had ulcerative colitis. Biologic therapies undertaken were infliximab (2/8), adalimumab (1/8), vedolizumab (1/8) and ustekinumab (4/8). There were no significant changes in pathology markers including FCP (800 to 1552 μg/g, p = 0.889) and C‐reactive protein (9 to 6.6 mg/L, p = 0.484). Three (9%) ultrasound assessments were unable to visualise bone (due to obesity in two participants and poor image capture in one participant) and, therefore, no muscle measurements were possible. Hence, for ultrasound of the arm and legs, paired results of seven responders and seven non‐responders were analysed and for ultrasound of the arm, paired results of eight responders and eight non‐responders were analysed.

3.2. Intra‐ and inter‐rater reliability of sonographic measurements

Intra‐rater reliability for muscle thickness measurements of the entire ultrasound protocol revealed intra‐class correlation coefficient 0.98 (95% CI 0.96–0.99) and 0.99 (95% CI 0.98–0.99) for each single operator. Inter‐rater reliability for landmarking five areas on the limbs, revealed intra‐class correlation coefficient 0.99 (95% CI 0.98–1.00) and for muscle thickness measurements revealed intra‐class correlation coefficient 0.93 (95% CI 0.84–0.97).

3.3. Comparisons between DXA‐SMI and point‐of‐care muscle assessments

All muscle assessments, including ultrasound, BIA and anthropometric measurements, correlated significantly with DXA‐derived SMI in IBD participants (Figure 2) and in control participants (Table 2). For men (n = 33) and women (n = 21), respectively, DXA‐derived SMI correlated significantly with ultrasound of the arm (r = 0.74, r = 0.63), ultrasound of the arm and legs (r = 0.79, r = 0.72), BIA (r = 0.83, r = 0.72), calf circumference (r = 0.76, r = 0.82), mid‐arm circumference (r = 0.82, r = 0.71) and BMI (r = 0.78, r = 0.81).

FIGURE 2.

Correlation graphs comparing different muscle assessment methods to dual energy X‐ray absorptiometry (DXA)‐derived skeletal muscle index (SMI) in patients with inflammatory bowel disease (IBD). Pearson correlation coefficients were used for ultrasound and bioelectrical impedance analysis (BIA) graphs and Spearman's correlation coefficients were used for calf and mid‐arm circumference as well as body mass index (BMI).

TABLE 2.

The relationship in healthy controls of muscle assessment methods with dual energy X‐ray absorptiometry (DXA)‐derived skeletal muscle index (SMI) as the reference standard, including a correlation and Bland–Altman analysis.

Controls	Correlation analysis			Bland–Altman analysis
	Coefficient			Mean difference (kg/m2)	SD	95% limits of agreement
	All	Men (n = 15)	Women (n = 15)	All
Ultrasound (arm) (cm2)	0.87 b	0.70 b	0.84 c	0.00	0.63	−1.23 to 1.23
Ultrasound (arm and legs) (cm2)	0.88 b	0.64 b	0.82 b	0.00	0.61	−1.20 to 1.20
BIA a predicted SMI (kg/m2)	0.91 c	0.81 b	0.62 b	0.55	0.55	−0.53 to 1.62
Calf circumference (cm)	0.73 b	0.76 b	0.65 b	0.00	0.88	−1.73 to 1.73
Mid‐arm circumference (cm)	0.79 b	0.63 b	0.84 b	0.00	0.79	−1.55 to 1.55
Body mass index (kg/m2)	0.70 b	0.82 b	0.63 b	0.00	0.92	−1.81 to 1.81

Note: The 95% limits of agreement were smaller for healthy controls than patients with IBD for all muscle assessment methods. For all correlation analyses, the p < 0.05.

^a Bioelectrical impedance analysis.

^b Pearson correlation coefficient.

^c Spearman correlation coefficient.

Estimates of SMI using ultrasound correlated with DXA‐derived SMI similarly for patients with ulcerative colitis (n = 24) and Crohn's disease (n = 30) with ultrasound of the arm and legs being r = 0.89 for both groups, and ultrasound of the arm being r = 0.86 and r = 0.88, respectively. DXA‐derived SMI correlated significantly with BIA and BMI in ulcerative colitis (r = 0.96 and r = 0.83, respectively) and in Crohn's disease (r = 0.84 and r = 0.53, respectively).

For participants with IBD, Bland–Altman analyses found a mean difference of 0 kg/m² for ultrasound of the arm_, ultrasound of the arm and legs, calf circumference, mid‐arm circumference and BMI when compared with DXA‐derived SMI, see Figure 3. Ultrasound of the arm and legs had the smallest 95% limits of agreement (−1.3 to 1.3 kg/m²) followed by ultrasound of the arm (−1.44 to 1.44 kg/m²). BIA‐predicted SMI overestimated DXA‐derived SMI by 1.07 kg/m² with 95% limits of agreement −0.16 to 2.30 kg/m². When total BIA fat‐free mass was compared with DXA fat‐free mass, the mean difference was 3.95 kg with wide 95% limits of agreement −2.30 to 10.19 kg.

FIGURE 3.

Bland–Altman plots showing the agreement between different muscle assessment methods and dual energy X‐ray absorptiometry (DXA)‐derived skeletal muscle index (SMI) in patients with inflammatory bowel disease (IBD). BIA, bioelectrical impedance analysis; BMI, body mass index.

In control participants, BIA also overestimated DXA‐derived SMI (mean difference 0.55, 95% limits of agreement −0.53 to 1.62 kg/m²), see Table 2. The mean difference for all other muscle assessments was 0 with the smallest 95% limits of agreement again found for ultrasound of the arm and legs (−1.2 to 1.2 kg/m²) and ultrasound of the arm (−1.23 to 1.23 kg/m²).

There were no differences in all muscle assessments between patients with Crohn's disease compared to ulcerative colitis. Similarly, there were no differences in patients receiving steroids compared to those not receiving steroids (data not shown).

3.4. Receiver‐operating characteristic curves for determining low DXA‐SMI

In men with IBD, the muscle assessment technique with the highest area under the curve (AUC) for detecting low SMI was ultrasound of the arm and legs (AUC 0.90), see Figure 4. A cut‐off of <302 cm² (sensitivity 100% and specificity 76%), predicted low SMI (Table 3). Ultrasound of the arm also differentiated low from normal SMI with an AUC 0.88 and cut‐off <118 cm² (sensitivity 100% and specificity 81%).

FIGURE 4.

Receiver‐operating characteristic (ROC) curves for predicting low skeletal muscle index (SMI) in patients with inflammatory bowel disease. The left column represents ROC curves for men undergoing ultrasound of the arm and ultrasound of the arm and legs. The right column represents the same ultrasound assessment of women.

TABLE 3.

Receiver‐operating characteristic curve analysis and performance characteristics in men and women with inflammatory bowel disease (IBD).

	Area under the curve	p‐value	Cut‐off value	Sensitivity (%)	Specificity (%)	Positive predictive value (%)	Negative predictive value (%)	Accuracy (%)
Men with IBD
Ultrasound (arm) (cm2)	0.88	0.002	<118	100	81	58	100	85
Ultrasound (arm and legs) (cm2)	0.90	0.001	<302	100	76	53	100	81
BIA a predicted SMI b (kg/m2)	0.83	0.008	<9	86	77	50	95	79
Calf circumference (cm)	0.75	0.043	<33	57	87	54	88	81
Mid‐arm circumference (cm)	0.88	0.003	<32.5	100	69	46	100	76
Body mass index (kg/m2)	0.79	0.020	<25.4	86	69	42	95	73
Women with IBD
Ultrasound (arm) (cm2)	0.93	0.021	<73	100	89	60	100	91
Ultrasound (arm and legs), (cm2)	0.92	0.023	<220	100	76	40	100	79
BIA a predicted SMI b (kg/m2)	0.98	0.009	<6.6	100	95	77	100	96
Calf circumference (cm)	0.98	0.009	<33.3	100	94	73	100	95
Mid‐arm circumference (cm)	0.92	0.024	<26.3	100	83	49	100	85
Body mass index (kg/m2)	1.00	0.007	<19.9	100	100	100	100	100

^a Bioelectrical impedance analysis.

^b Skeletal Muscle Index.

In women with IBD, a cut‐off of <19.9 kg/m² for BMI provided an AUC of 1.00 with 100% sensitivity and specificity. BIA and calf circumference (both AUC 0.98) were also able to identify low SMI. No receiver‐operating characteristic analysis was performed for controls, as only one male participant in the entire cohort had low SMI.

3.5. Longitudinal analysis

In 17 patients who underwent biologic therapy for active disease, the % change in SMI measured by DXA, correlated significantly with the % change of all other muscle assessment techniques (Figure 5). Of responders, DXA‐derived SMI increased by 9.6% (95% CI 3.2–16.2) (p = 0.004) from baseline to follow‐up. Low SMI was found in 3/9 responders at baseline, and all increased at the time of follow‐up, with 2/9 normalising their SMI. As seen in Figure 6, other muscle assessments also increased, ultrasound of the arm by 11.9% (95% CI 1.4–22.3) (p = 0.034), ultrasound of the arm and legs by 13.7% (95% CI −0.5 to 27.9; p = 0.021), BIA by 4.3% (95% CI 1–7.6; p = 0.011), calf circumference by 2.8% (95% CI 0.1–5.4; p = 0.029) and BMI by 6.4% (95% CI 1.3–11.5; p = 0.012). There was no significant change in mid‐arm circumference (increase of 1.3% [95% CI −4.1 to 6.7], p = 0.703). Weight increased in responders by 4.3 kg (95% CI 1–7.6), consisting of 69% muscle mass. In non‐responders, there was no change in DXA‐derived SMI (increase of 2% [95% CI −2.1 to 6.1], p = 0.277) or any other muscle assessments.

FIGURE 5.

Correlation graphs comparing skeletal muscle index (SMI) % change, determined by dual energy X‐ray absorptiometry (DXA), to the % change of other muscle assessment techniques in patients (n = 17) before and after induction biologic therapy. Percentage change calculated from the (difference of the follow‐up and baseline values) divided by (baseline value) and multiplied by 100. Pearson correlation coefficients were used for all graphs. In the SMI % change versus ultrasound (arms and legs) % change graph, three overlapping data points (at approximately x‐axis = 11 and y‐axis = 5) have been minimally changed for graphical representation; however, the original values were used in the statistical analysis. BIA, bioelectrical impedance analysis; BMI, body mass index.

FIGURE 6.

Baseline and follow‐up muscle assessment measurements in patients who responded and who did not respond to biologic induction therapy. For responders, change in means from baseline to follow‐up were for SMI 7.8 to 8.5 kg/m², ultrasound (arm) 124.2 to 139.2 cm², ultrasound (arm and legs) 302.3 to 341.3 cm², bioelectrical impedance analysis (BIA) 9.2 to 9.6 kg/m², calf circumference 35.8–36.8 cm, mid‐arm circumference 30.9 to 31.1 cm and body mass index (BMI) 24.1 to 25.6 kg/m². Paired t‐tests were used for all comparisons. The zero change of the mean difference was highlighted with a red line.

4. DISCUSSION

Low skeletal muscle mass in patients with IBD is not only associated with reduced quality of life and poor outcomes such as surgical complications and reduced response to therapy, but may also indicate adequacy of therapy. Yet its measurement is not routine in clinical practice nor in clinical therapeutic trials. The reference standard in assessing muscle mass is DXA, but this cannot be performed at the point‐of‐care, is expensive and inconvenient for patients, and the need for repeated evaluation increases concern regarding radiation exposure. The current study, therefore, evaluated alternative simple bedside methods of assessing muscle mass and compared it to DXA as a reference method. While all muscle assessment techniques correlated significantly with DXA‐derived SMI in IBD patients, BIA overestimated SMI by 1.07 kg/m², while all other techniques did not. Ultrasound of the leg and arms combined provided the most accurate measure with no bias and small limits of agreement (−1.3 to 1.3 kg/m²). On longitudinal analysis, all muscle assessments, except mid‐arm circumference, were responsive to change induced by induction therapy in line with those defined by repeated DXA.

The novel technique evaluated was muscle thickness ultrasound, which has not been applied to the IBD population previously. Ultrasound not only correlated well with DXA‐derived SMI, but changed positively with disease control. Muscle ultrasound has been shown in other settings to correlate well with reference methods. For example, ultrasound of the arm and leg, along with covariates of age and sex, accurately estimated appendicular lean muscle measured by DXA in 96 healthy volunteers (R ² = 0.91) ⁸ and correlated significantly with computed tomography muscle cross‐sectional area (r = 0.85) in patients admitted into intensive care. ¹⁰ Due to the time required to landmark and scan five positions for the dual‐limb protocol (approximately 15 min), our study also assessed just one ultrasound measurement of the dominant arm as an alternative. This can be performed under 5 min, but its prediction of DXA‐derived SMI was inferior to that of the dual‐limb approach in that there were greater limits of agreement, although no bias was observed and it was responsive to change. Hence, dual‐limb assessment provides more precision and is the preferred method. ⁸ , ¹⁰

In our study, dual‐limb ultrasound demonstrated the most utility in men with IBD. Dual‐limb ultrasound had the highest AUC for predicting DXA‐derived SMI of all muscle assessments and was most accurate in predicting change of DXA‐derived SMI after treatment. This finding suggests that in men, dual‐limb ultrasound is an accurate baseline and follow‐up measure for estimating SMI. The utility of ultrasound in women with IBD, however, remains uncertain. Alternative, quicker and simpler muscle measurements, such as BMI and calf circumference had superior AUC than ultrasound for predicting DXA‐derived SMI, and response to therapy was limited by small samples of female patients, with none included in the responder group. Thus, muscle thickness ultrasound in women with IBD requires further investigation before implementation can be recommended into clinical practice.

Ultrasound has the merit of convenience over whole‐body DXA, where it can be performed at the bedside without radiation or excessive cost. ⁸ Ultrasound equipment is becoming more commonly available in clinics due to the impact of point‐of‐care intestinal ultrasound on clinical management. ²⁵ Additionally, ultrasound can also measure muscle quality, which is not possible with DXA. High ultrasound echogenicity values, reflecting extensive fat infiltration in the muscle, has been correlated with lower muscle quality and grip strength. ²⁰ , ²⁶ , ²⁷ This was, however, not included in the current study. Muscle thickness ultrasound requires accurate landmarking, probe placement and identification of fascia borders and hence needs skilled operators. However, training is not arduous and can be performed within two to three training sessions. We showed good inter‐rater correlation coefficients, similar to other studies, suggesting that the ultrasound protocol can be performed by multiple individuals with adequate training. ⁸ , ¹⁰ Obesity, particularly at a BMI >40 kg/m², however, limits ultrasound assessment, as the quadriceps was unable to be imaged properly due to insufficient depth of the ultrasound probe to the reach the femur. The psoas muscle, commonly visualised on intestinal ultrasound, may be an alternative muscle that could be considered for measurement in future studies.

Bioimpedance assessment has the distinct advantage of being very quick to perform, but its accuracy for estimating DXA‐derived SMI is less certain. Previous studies have consistently shown that BIA overestimates muscle mass, using DXA as a reference method to a variable degree. For example, two studies found a bias for estimating SMI between 1.5 and 1.75 kg, ¹³ , ¹⁴ while another study of 40 patients with IBD overestimated fat‐free mass by a mean of 1.3 kg. ¹⁵ In the current study, BIA overestimated fat‐free mass of BIA by threefold more (3.95 kg). The variation between studies may be explained by the BIA devices used, with numerous setups and engineering, single versus multi‐frequency, different manufacturer equations utilised, as well as different populations studied, varying in age, ethnicity, and body weights. Additionally, DXA scanning techniques may also vary between studies as its reference method. Despite overestimation of BIA‐predicted SMI, in our study the bias remained constant over time, and was similarly responsive to change after induction therapy as demonstrated by changes in DXA‐derived SMI. Therefore, it may serve as a monitoring tool between clinic visits provided the same instrument and protocol is being followed. However, in research where precise body composition muscle parameters are used for phenotyping and prognosis, BIA should not replace DXA.

As a point‐of‐care tool in the clinic, BIA is an attractive option due to the speed of performing it (under a minute), albeit if the BIA device is available in the clinic to be used routinely. However, BIA requires more attention to confounders. It makes the assumption that hydration is fixed, but this may be altered in patients with obesity ²⁸ or other pathologies such as renal disease or indeed dehydration related to diarrhoea. ²⁹ We attempted to minimise hydration variations by asking all participants to fast for at least 2 h and empty their bladder prior to BIA assessment and indeed total body water was found to be similar in both IBD and control patients in our study. Patients with IBD had lower albumin levels than controls, although the potential effect of variations of albumin in total body water assessment is not well defined. Finally, BIA cannot be performed in patients with amputations, implantable defibrillators or pacemakers. ²⁹ Although our IBD patient cohort was relatively young (median age 30 years), with increasing abundance of effective medical treatments available for IBD, older patients may face this as a barrier to using BIA.

Simple anthropometric measurements require minimal technology and less expertise in interpretation. ⁷ While our study showed that anthropometric measurements, including BMI, correlated significantly with DXA‐derived SMI, other studies have demonstrated poor prediction of muscle mass in patients with IBD using BMI. ¹ , ³⁰ It is of note that, in the current study, BMI perfectly predicted women with low SMI, but only three had a BMI <19.9 kg/m². If these results are reproducible with larger sample numbers, BMI may be a useful screening tool towards classifying which patients should undergo further DXA scanning. Both calf circumference and BMI correctly predicted changes in DXA‐derived SMI after biologic induction therapy, suggesting a role for these techniques in monitoring individual patients. Mid‐arm circumference did not change significantly, but our assessment did not measure the triceps skinfold thickness, a layer of concentric fat that varies with accumulating adiposity. ⁷ , ³¹ While anthropometry in clinical settings is certainly quicker and cheaper than an ultrasound assessment, it does pose some challenges. For calf‐ and mid‐arm circumference, appropriate methodological training is required. ⁷ Cut‐off points are not clearly defined and may be age, gender and ethnic‐specific, making this somewhat impractical for a busy clinician. ⁷ , ¹⁶ , ³²

The strengths of the current study include the prospective design that limited recall bias. A comparison was performed between multiple muscle assessments in the same IBD and control populations with the same reference method (DXA). Longitudinal analysis in our study was able to demonstrate the potential of all methods as a monitoring tool. Additionally, this is the first study to look at ultrasound muscle thickness as a measure of low SMI in an IBD cohort. Limitations in our study included its small sample size, particularly evaluating response to therapy. Larger populations are needed to ensure the accuracy of repeated measurements. Additionally, during follow‐up assessments after treatment, investigators were not actively blinded to objective response if it was available (in 7 of 17 patients) and also documented clinical symptoms at the time. This may have introduced potential bias in measurements recorded.

In summary, the point‐of‐care muscle assessments examined in this study were simple, quick and portable tools, which can be used to predict muscularity and identify low SMI in patients with IBD. Sonographic measurement of the dominant arm ± bilateral thigh muscle thickness is a promising new technique that could be easily integrated into clinical practice and performed by both gastroenterologists and allied health staff with appropriate training. A monitoring role was identified for all techniques except mid‐arm circumference, but further prospective studies with larger sample sizes are now required.

AUTHOR CONTRIBUTIONS

Anke L. Nguyen: Conceptualization (equal); formal analysis (lead); investigation (lead); methodology (equal); writing – original draft (lead); writing – review and editing (equal). Megan Burns: Investigation (supporting); writing – review and editing (equal). Madhuni Herath: Conceptualization (equal); methodology (equal); writing – review and editing (equal). Kate Lambell: Conceptualization (equal); methodology (equal); writing – review and editing (equal). Darcy Holt: Conceptualization (equal); methodology (equal); resources (supporting); writing – review and editing (equal). Jessica Fitzpatrick: Conceptualization (equal); methodology (equal); writing – review and editing (equal). Frances Milat: Resources (supporting); writing – review and editing (equal). Peter R. Ebeling: Resources (supporting); writing – review and editing (equal). Peter Gibson: Conceptualization (equal); formal analysis (supporting); supervision (equal); writing – original draft (supporting); writing – review and editing (equal). Gregory Moore: Conceptualization (equal); formal analysis (supporting); methodology (equal); resources (lead); supervision (equal); writing – original draft (supporting); writing – review and editing (equal).

FUNDING INFORMATION

A.L.N. was supported by the Research Training Program Stipend (Monash University). M.H. was supported by the National Health and Medical Research Council. J.F. is a recipient of the Crohn's Colitis Australia PhD Scholarship.

AUTHORSHIP

Guarantor of this article: Anke L Nguyen and Gregory T Moore.

Nguyen AL, Burns M, Herath M, Lambell K, Holt D, Fitzpatrick J, et al. Accuracy of ultrasound, bioelectrical impedance analysis and anthropometry as point‐of‐care measurements of skeletal muscle mass in patients with inflammatory bowel disease. Aliment Pharmacol Ther. 2023;58:309–321. 10.1111/apt.17607