Diagnostic value of computed tomography‐based muscle quality and metabolic signatures in heart failure with sarcopenia

Abstract

Aims

Sarcopenia is a major comorbidity in heart failure (HF) patients, but its comprehensive evaluation remains challenging due to resource limitations. Computed tomography (CT) imaging allows opportunistic assessment of both muscle quantity and quality; however, its diagnostic utility and relationships with physical performance and underlying metabolic alterations remain unclear. We investigated the diagnostic utility of single‐slice abdominal CT for sarcopenia assessment and explored its association with plasma amino acid profiles in HF patients.

Methods and results

We retrospectively analysed 212 HF patients (73 ± 13 years old; 41% females) who underwent dual‐energy X‐ray absorptiometry (DXA), abdominal CT imaging, and amino acid profiling, of whom 186 had complete sarcopenia assessment. Sarcopenia was diagnosed based on the Asian Working Group for Sarcopenia 2019 criteria using DXA‐measured appendicular skeletal muscle mass index and grip strength, with severe sarcopenia additionally requiring low Short Physical Performance Battery scores. CT‐measured skeletal muscle area index (SMI) at the lumbar vertebral level combined with grip strength showed high diagnostic accuracy for sarcopenia [area under the receiver operating characteristic curve (AUC): 0.96 for males, 0.88 for females], with optimal cut‐off values of 54 cm²/m² for males and 40 cm²/m² for females. SMI was similarly reduced in non‐severe and severe sarcopenia patients, whereas mean muscle attenuation (MMA), an index of intramuscular lipid infiltration, was decreased exclusively in patients with severe sarcopenia. In multivariable analysis adjusted for age, sex, HF functional class and diabetes, decreased MMA remained the only CT‐measured index significantly associated with low physical performance. MMA showed higher predictive value than SMI for severe sarcopenia in males (AUC 0.81 vs. 0.54, P = 0.039). Several amino acids (leucine, β‐alanine and 3‐methylhistidine) and the Fischer ratio were significantly associated with MMA, independent of SMI and established MMA determinants including sex, age, and body mass index.

Conclusions

Single‐slice abdominal CT provides accurate sarcopenia diagnosis in HF patients with reduced muscle strength, with muscle quality rather than quantity determining physical performance. Specific amino acid profiles associated with muscle quality changes suggest potential therapeutic targets for preventing sarcopenia progression.

Introduction

Sarcopenia, a major comorbidity in heart failure (HF) patients, is linked to higher mortality and worse outcomes. ¹ The European Working Group on Sarcopenia in Older People 2 (EWGSOP2) proposes a diagnostic process where reduced muscle strength is the primary criterion. ² If muscle strength is low, then appendicular skeletal muscle mass (ASM), often measured by dual‐energy X‐ray absorptiometry (DXA) or bioelectrical impedance analysis (BIA), is assessed to confirm sarcopenia. Severe sarcopenia is diagnosed when reduced muscle strength and low muscle mass are accompanied by low physical performance. The Asian Working Group for Sarcopenia (AWGS) 2019 similarly uses these three components for diagnosing sarcopenia and its severity with a slightly different algorithm: sarcopenia is diagnosed when low ASM is accompanied by either reduced muscle strength or low physical performance, with the presence of all three components indicating severe sarcopenia. ³ However, the comprehensive evaluation of sarcopenia in clinical settings remains challenging due to resource limitations, particularly for measuring ASM and physical performance. ⁴ , ⁵

Routine clinical computed tomography (CT) imaging allows assessment of muscle quantity. EWGSOP2 recommends CT imaging at the lumbar vertebra level as an alternative for muscle mass quantification, though diagnostic cut‐off values are not provided. ² EWGSOP2 also highlights the relevance of muscle quality including muscle architecture and composition. Beyond muscle quantity, CT imaging offers the additional advantage of evaluating muscle quality, for instance, by measuring intermuscular and intramuscular fat infiltration (known as myosteatosis). ⁶ Evidence indicates that these qualitative changes in skeletal muscle are determinants of physical performance independently of muscle mass in older adults. ⁷ , ⁸ Despite the recognition of CT imaging's potential utilities, research using CT‐measured muscle parameters for sarcopenia diagnosis in HF patients is scarce. Additionally, unclear mechanisms of skeletal muscle remodelling hinder the development of specific therapies for sarcopenia. We have recently reported that an adverse plasma branched‐chain amino acid (BCAA) profile reflects fatty muscle degeneration in HF patients with diabetes mellitus. ⁹ Identifying metabolic signatures associated with unfavourable muscle remodelling could provide new insights into the pathophysiology of sarcopenia in HF patients.

In this study, we investigated the diagnostic performance of single‐slice abdominal CT for sarcopenia in HF patients who were evaluated with DXA, grip strength and the Short Physical Performance Battery (SPPB). We hypothesized that CT‐assessed myosteatosis provides independent information for staging sarcopenia severity beyond muscle mass. Furthermore, we explored the relationship between amino acid profiles and muscle quality changes to elucidate metabolic alterations involved in skeletal muscle remodelling.

Methods

Study subjects

This study was a single‐centre, cross‐sectional study. We retrospectively enrolled patients who were hospitalized at Sapporo Medical University Hospital between 15 February 2018 and 31 March 2021 for the diagnosis and management of HF (Figure 1 ). HF was diagnosed according to the 2016 ESC guidelines for the diagnosis and treatment of acute and chronic HF. ¹⁰ Patients with chronic kidney disease at Stage 5 defined as estimated glomerular filtration rate (eGFR) < 15 mL/min/1.73 m² were excluded. Initially, a total of 354 patients who underwent both DXA and plasma amino acid profiling during hospitalization were selected. To analyse the correlation between DXA and CT in body composition, we identified 220 patients from the 354 who underwent abdominal CT imaging for clinical purposes within a 2 month window of plasma amino acid profiling. The indications for abdominal CT imaging were as follows: systemic evaluation for acute decompensated HF (N = 91); pre‐ or post‐procedural assessment for invasive procedures such as catheter interventions, device implantations and cardiac surgeries (N = 92); systemic evaluation for secondary cardiomyopathy (N = 18); and evaluation for non‐cardiac diseases (N = 19). We then excluded 8 of these 220 patients due to inadequate cross‐sectional CT images at the L3/4 level, resulting in a cohort of 212 patients for the DXA–CT correlation analysis. Subsequently, to evaluate sarcopenia diagnostic accuracy and explore the relationship between plasma amino acid profiles and CT measurements, we excluded 26 patients lacking grip strength or SPPB data, leading to a final cohort of 186 patients for these analyses. This study was conducted in strict accordance with the principles of the Declaration of Helsinki and was approved by the Clinical Research Ethics Committee of Sapporo Medical University Hospital (approval number: 302‐243).

Figure 1.

Flow chart of inclusion of patients. CT, computed tomography; DXA, dual‐energy X‐ray absorptiometry; HF, heart failure; MMA, mean muscle attenuation; SPPB, the Short Physical Performance Battery.

Biochemical analyses, echocardiography and DXA scan

Data for blood tests and left ventricular ejection fraction, measured by the modified Simpson method, were retrieved from the patients' medical records. eGFR was calculated using equations developed for Japanese subjects as follows: eGFRcre (mL/min/1.73 m²) = 194 × Scr^‐1.094 × age^‐0.287(×0.739 if female). ¹¹ Chronic kidney disease was defined as eGFRcre of less than 60 mL/min/1.73 m².

Whole‐body and regional fat/lean mass analyses were performed using the Horizon A DXA System (HOLOGIC, Waltham, MA, USA) as we previously reported. ⁹ Lean mass was defined as an indicator of skeletal muscle mass. ASM was calculated as the sum of bone‐free lean mass in the arms and legs, and the ASM index (ASMI) was defined as ASM divided by height squared.

Measurements of plasma amino acid concentration

Plasma concentrations of amino acids were measured in a fasting state after stabilization of HF using ultraperformance liquid chromatography (UPLC, LSI Medience Corporation, Tokyo, Japan), and UPLC analysis was conducted using the Acquity™ UPLC system with a TUV detector and MassTrak™ AAA Solutions Kit (Waters Corporation, Milford, MA, USA) and validated as we previously described. ¹² Missing values were observed for aspartate (N = 7), β‐alanine (N = 5), monoethanolamine (N = 9) and 1‐methylhistidine (N = 2) due to low abundance (i.e., below the detection limit). For imputation, these missing values were replaced with one‐fifth of the minimum observed value for each respective metabolite. We previously validated the appropriateness of this imputation method in our prior study. ¹² The Fischer ratio was calculated as the ratio of total BCAAs to the sum of tyrosine and phenylalanine.

Measurements of muscle strength and physical performance

Muscle strength and physical performance were evaluated by trained physical therapists using handgrip strength and the SPPB score, respectively, as detailed in our prior publication. ¹³ Handgrip strength was measured with a Smedley spring‐type digital dynamometer (TKK 5401 GRIP‐D, TAKEI, Niigata, Japan), and the maximum value from two trials was recorded in kilogrammes. The SPPB consists of a 4 m gait‐speed test, a time‐based test for five repeated chair stands and a balance test, with total scores ranging from 0 (worst) to 12 (best) points. Handgrip strength and physical performance assessments were performed on the same day as the DXA scan.

Diagnosis of sarcopenia

We used a slightly modified version of the AWGS 2019 criteria ³ to diagnose sarcopenia. In our study, sarcopenia was defined by the presence of both low skeletal muscle mass (ASMI <7.0 kg/m² for males and <5.4 kg/m²) and low muscle strength (handgrip strength <28 kg for males and <18 kg for females). Severe sarcopenia was diagnosed when patients with sarcopenia also exhibited low physical performance (SPPB score ≤9). We deviated slightly from the AWGS criteria by not classifying patients with low muscle mass and low SPPB but normal muscle strength as sarcopenic, primarily because (1) the EWGSOP2 guidelines require reduced muscle strength for a sarcopenia diagnosis, ² and (2) this subgroup (low muscle mass, low SPPB, but normal muscle strength) was small in our study cohort (N = 9).

Assessment of body compositions by CT imaging

Abdominal CT scans were performed using two CT scanners: LightSpeed VCT (GE Healthcare, Milwaukee, WI, USA) and Aquilion Prime (Canon Medical Systems, Otawara, Tochigi, Japan) and were retrospectively analysed. To avoid the influence of contrast agents on muscle CT values, all analyses were performed on non‐contrast‐enhanced CT images. The standard CT acquisition protocol in our hospital included a tube voltage of 120 kV, variable mA with dose modulation, a rotation time of 0.5 s, a matrix of 512 × 512, a field of view of 300–360 mm and a reconstructed slice thickness of 5.0 mm. Hounsfield units (HU) were used to quantify CT attenuation, with air defined as −1000 HU and water as 0 HU. SYNAPSE VINCENT software (Fujifilm, Tokyo, Japan) was used for automated segmentation of skeletal muscle, visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) cross‐sectional areas at the level between L3 and L4. Manual correction of segmentation boundaries was performed when necessary to ensure accuracy. The software's initial segmentation algorithm identified adipose tissue as voxels with CT attenuation values ranging from −200 to −50 HU, and skeletal muscle as voxels ranging from −50 to 100 HU. For subsequent muscle analysis, total skeletal muscle area (SMA) was quantified as the cross‐sectional area within −30 to 100 HU, and mean muscle attenuation (MMA), an index of intramuscular lipid infiltration, was calculated as the average CT attenuation value within this area, according to previously published methods. ⁶ , ¹⁴ , ¹⁵ SMA index (SMI) was defined as SMA divided by height squared.

Statistical analysis

Data are presented as medians [interquartile ranges (IQR)]. Categorical variables are expressed as frequencies and percentages. For continuous variables, group comparisons among non‐sarcopenia, sarcopenia and severe sarcopenia groups were initially performed using Kruskal–Wallis tests. When significant differences were observed, post‐hoc analyses were conducted using Mann–Whitney U tests with Bonferroni correction (P < 0.017). For categorical variables, overall differences were first assessed using χ ² tests, followed by pairwise χ ² tests with Bonferroni correction when significant differences were detected.

Linear regression analysis was performed to examine the relationship between ASMI measured by DXA and SMI measured by CT. The regression equation was then used to determine CT‐measured SMI values corresponding to established ASMI cut‐off values for sarcopenia diagnosis. The diagnostic accuracy of CT‐measured muscle area combined with grip strength for identifying sarcopenia was evaluated separately for males and females using the area under the receiver operating characteristic curve (AUC), sensitivity and specificity.

Univariate and multivariable logistic regression analyses were performed to examine associations between body compositions and low SPPB scores. In multivariable analyses, separate models were constructed for each body composition parameter, with adjustment for age, sex, New York Heart Association (NYHA) functional class and diabetes mellitus. Receiver operating characteristic curve analyses were conducted separately for males and females to evaluate the diagnostic performance of different skeletal muscle measurements for detecting severe sarcopenia. AUC was calculated with 95% confidence intervals (CI), and optimal cut‐off values were determined using Youden's index. The DeLong test was used to compare AUCs between different measurements.

Associations between MMA and amino acid profiles were first examined using univariate linear regression analyses. Amino acids showing significant associations in univariate analyses were then evaluated in separate multivariable models, adjusting for age, sex, body mass index (BMI) and SMI. For these linear regression analyses, model assumptions were verified through residual analysis, including normality (Shapiro–Wilk test) and homoscedasticity assessment.

For all regression analyses, the absence of significant multicollinearity among explanatory variables was confirmed using variance inflation factors <10. Statistical analyses were conducted using R version 4.2.3 (R Foundation for Statistical Computing, Vienna, Austria). A two‐tailed P value <0.05 was considered statistically significant, except for multiple comparisons where Bonferroni‐adjusted P values were used.

Results

Baseline clinical characteristics of HF patients according to sarcopenia severity

The baseline characteristics were similar between the total study population (N = 212) and the subgroup that underwent sarcopenia assessment (N = 186) (Table 1 ). Among the 186 patients, 65 patients (35%) met the diagnostic criteria for sarcopenia based on DXA measurements of muscle mass and grip strength evaluations, with 34 (52%) classified as having severe sarcopenia with low SPPB scores. Patients with severe sarcopenia were significantly older compared with non‐sarcopenic patients and non‐severe sarcopenic patients. While BMI was lower in both non‐severe and severe sarcopenic patients, the prevalence of diabetes mellitus was higher and serum albumin levels were lower exclusively in severe sarcopenic patients. ASMI showed no significant difference between non‐severe and severe sarcopenic patients; however, grip strength was significantly lower in severe sarcopenic patients compared with non‐severe sarcopenic patients in males, with a similar trend observed in females.

Table 1.

Baseline characteristics of the study population.

	Total population	Sub‐population with sarcopenia assessments
		All	Non‐Sarcopenia	Sarcopenia
				Non‐severe	Severe	P
	N = 212	N = 186	N = 121	N = 31	N = 34
Age (years)	77 [66, 83]	77 [66, 83]	72 [62, 82]	74 [70, 85]	83 [80, 86] a , b	<0.001
Female sex	87 (41)	80 (43)	50 (41)	11 (36)	19 (56)	0.206
BMI (kg/m2)	22.6 [20.9, 25.2]	22.9 [21.3, 25.3]	23.7 [21.9, 26.5]	21.6 [19.4, 22.4] a	22.4 [20.0, 24.4] a	<0.001
NYHA III/IV	67 (32)	54 (29)	25 (21)	8 (26)	21 (62) a , b	<0.001
LVEF (%)	52.3 [32.4, 63.9]	55.2 [33.7, 64.2]	50.5 [32.8, 63.5]	57.4 [31.3, 65.3]	61.3 [48.5, 65.8]	0.135
LVEF<40%	75 (35)	64 (34)	44 (36)	13 (42)	7 (21)	0.145
Aetiology						0.094
Cardiomyopathy	61 (29)	53 (28)	42 (35)	5 (16)	6 (17)
Ischemic	28 (13)	22 (12)	14 (11)	6 (19)	2 (6)
Valvular heart disease	101 (48)	91 (49)	52 (43)	16 (52)	23 (68)
Others	22 (10)	20 (11)	13 (11)	4 (13)	3 (9)
Hypertension	142 (67)	128 (69)	77 (64)	24 (77)	27 (79)	0.113
Diabetes mellitus	89 (42)	76 (41)	41 (34)	15 (48)	20 (59) a	0.021
Dyslipidaemia	124 (59)	111 (60)	78 (65)	15 (48)	18 (53)	0.180
Chronic kidney disease	125 (59)	109 (59)	67 (55)	19 (61)	23 (68)	0.415
Atrial fibrillation	78 (37)	69 (37)	44 (36)	13 (42)	12 (35)	0.824
Haemoglobin (g/dL)	11.9 [10.7, 13.6]	11.9 [10.7, 13.6]	12.3 [10.8, 14.1]	11.8 [10.6, 13.2]	11.4 [10.4, 12.4]	0.070
NT‐proBNP (pg/mL)	1002 [499, 2192]	967 [460, 2199]	851 [395, 1803]	1653 [755, 3024]	1081 [480, 2770]	0.056
eGFR (mL/min/1.73 m2)	55.3 [39.0, 67.8]	55.3 [39.2, 68.0]	57.3 [40.8, 69.8]	50.2 [34.8, 71.7]	50.9 [40.0, 62.8]	0.303
Albumin (g/dL)	3.6 [3.3, 4.0]	3.6 [3.4, 4.0]	3.7 [3.5, 4.1]	3.6 [3.3, 3.9]	3.5 [3.2, 3.7] a	0.001
ACEI or ARB	122 (58)	106 (57)	70 (58)	20 (65)	16 (47)	0.346
β‐blocker	125 (59)	109 (59)	74 (61)	18 (58)	17 (50)	0.505
MRA	80 (38)	70 (38)	46 (38)	9 (29)	15 (44)	0.451
Loop diuretics	117 (55)	101 (54)	62 (51)	19 (61)	20 (59)	0.510
Device therapy						0.528
Pacemaker	16 (8)	15 (8)	8 (7)	2 (7)	5 (15)
ICD	27 (13)	22 (12)	17 (14)	3 (10)	2 (6)
CRT‐P/CRT‐D	14 (7)	12 (7)	8 (7)	3 (10)	1 (3)
ASMI by DXA (kg/m2) in males	6.7 [6.0, 7.3]	6.9 [6.1, 7.4]	7.1 [6.7, 7.8]	6.0 [5.6, 6.5] a	6.2 [5.4, 6.6] a	<0.001
ASMI by DXA (kg/m2) in females	5.3 [4.8, 5.9]	5.3 [4.7, 5.9]	5.8 [5.4, 6.1]	4.7 [4.1, 5.1] a	4.7 [4.3, 4.9] a	<0.001
Low ASMI by AWGS 2019	123 (58)	103 (55)	38 (31)	31 (100) a	34 (100) a	<0.001
Grip strength (kg) in males		29.4 [24.6, 35.8]	32.4 [29.4, 38.8]	25.4 [22.7, 26.8] a	22.6 [19.0, 24.3] a , b	<0.001
Grip strength (kg) in females		16.2 [12.9, 19.2]	19.0 [14.8, 21.0]	14.8 [13.2, 16.0] a	12.6 [9.6, 14.6] a	<0.001
Low grip strength by AWGS 2019		96 (52)	31 (26)	31 (100) a	34 (100) a	<0.001
SPPB		11.0 [8.0, 12.0]	12 [9, 12]	11 [11, 12]	6 [4, 8] a , b	<0.001
SPPB ≤ 9 (%)		71 (38)	37 (31)	0 (0) a	34 (100) a , b	<0.001

Note: Values are presented as median [interquartile range] or n (%). N refers to the number of patients for whom the parameter was available.

Abbreviations: ACEI, angiotensin‐converting enzyme inhibitor; ARB, angiotensin receptor blocker; ASMI, appendicular skeletal muscle mass index; AWGS, Asian Working Group for Sarcopenia; BMI, body mass index; CRT‐D, cardiac resynchronization therapy defibrillator; CRT‐P, cardiac resynchronization therapy pacemaker; DXA, dual‐energy X‐ray absorptiometry; eGFR, estimated glomerular filtration rate; ICD, implantable cardioverter defibrillator; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NT‐proBNP, N‐terminal pro‐brain natriuretic peptide; NYHA, New York Heart Association; SPPB, Short Physical Performance Battery.

^a The post‐hoc analysis with Bonferroni correction revealed statistical significance compared with the non‐sarcopenia group.

^b The post‐hoc analysis with Bonferroni correction revealed statistical significance compared with the non‐severe sarcopenia group.

Accuracy of CT‐based sarcopenia diagnosis

In the total study population (N = 212), a moderate positive correlation was observed between ASMI and SMI (R = 0.71, P < 0.001, Figure S1). In the sub‐group population with sarcopenia assessment (N = 186), the diagnostic model for sarcopenia using CT‐measured SMI combined with grip strength showed high discriminatory ability, with an AUC of 0.96 (95% CI: 0.93 to 0.99) for males and 0.88 (95% CI 0.81 to 0.96) for females (Table 2 ). To determine optimal cut‐off values for sarcopenia diagnosis, we first determined SMI values corresponding to AWGS 2019 ASMI criteria, which yielded 49.4 cm²/m² for males and 39.9 cm²/m² for females (Figure S1). Based on these values and previously reported cut‐off values in sarcopenic patients with cancer (52.4 cm²/m² for males and 38.5 cm²/m² for females), ¹⁶ we evaluated the diagnostic accuracy using CT‐measured SMI cut‐off values ranging from 49 to 55 cm²/m² for males and 38 to 41 cm²/m² for females at 1 cm²/m² intervals. The optimal cut‐off values determined by the highest Youden index were 54 cm²/m² for males (sensitivity 94%, specificity 89%) and 40 cm²/m² for females (sensitivity 80%, specificity 82%) against the reference standard of DXA‐measured ASMI and grip strength criteria (Table 2 ).

Table 2.

The accuracy of combined assessments of abdominal skeletal muscle area and grip strength for sarcopenia diagnosis.

Model	AUC (95% CI)	SMI cut‐off value	Sensitivity	Specificity	PPV	NPV	Youden index
Male
Grip strength <28 kg plus SMI at L3/4	0.964 (0.933–0.995)	55 cm2/m2	94%	86%	77%	97%	0.802
		54 cm2/m2 c	94%	89%	81%	97%	0.830
		53 cm2/m2 b	91%	90%	82%	96%	0.815
		52 cm2/m2	89%	90%	82%	94%	0.787
		51 cm2/m2	86%	93%	86%	93%	0.787
		50 cm2/m2 a	83%	94%	88%	92%	0.773
		49 cm2/m2	66%	94%	85%	85%	0.601
Female
Grip strength <18 kg plus SMI at L3/4	0.883 (0.809–0.957)	41 cm2/m2	80%	80%	71%	87%	0.600
		40 cm2/m2 a , c	80%	82%	73%	87%	0.620
		39 cm2/m2 b	73%	86%	76%	84%	0.593
		38 cm2/m2	63%	88%	76%	80%	0.513

^a Corresponding to the ASMI cut‐off values defined by AWGS 2019 in the linear regression analysis of the current study.

^b Previously reported cut‐off values in cancer patients. ¹⁶

^c The highest Youden index.

AUC, the area under the receiver operating characteristic curve; CI, confidence intervals; NPV, negative predictive value; PPV, positive predictive value; SMI, skeletal muscle area index.

MMA rather than muscle area was associated with physical performance in HF patients

In the pooled analysis of male and female patients, CT‐measured body compositions showed that SMI was significantly reduced in sarcopenic patients as expected, although there was no significant difference in SMI between non‐severe and severe sarcopenic patients (Figure 2 ). In contrast, MMA was decreased exclusively in severe sarcopenic patients. Females showed lower SMI and MMA values than males (Table S1). Although both males and females showed a greater reduction in MMA than SMI in severe sarcopenic patients compared with non‐severe sarcopenic patients, this trend was more pronounced in males. In univariate logistic regression analyses to examine factors associated with low physical performance (SPPB score ≤9) identified age [odds ratio (OR) 2.71 per IQR increase, 95% CI: 1.67 to 4.40], female sex (OR 2.93, 95% CI: 1.59 to 5.41), MMA (OR 0.37 per IQR increase, 95% CI: 0.23 to 0.59) and NYHA Class III or higher (OR 6.27, 95% CI: 3.13 to 12.56) as significant factors. In multivariable analysis adjusted for age, sex, NYHA class and diabetes, decreased MMA remained significantly associated with low physical performance (adjusted odds ratio 0.52 per IQR increase, 95% CI: 0.29 to 0.91, P = 0.023, Table 3 ).

Figure 2.

Comparison of CT‐measured muscle and fat indices across different severities of sarcopenia. Box plots illustrate differences in body composition parameters across three groups (non‐sarcopenia, non‐severe sarcopenia and severe sarcopenia). Horizontal lines within boxes denote medians, boxes represent interquartile ranges (IQR) and whiskers extend to 1.5 × IQR. The P values obtained for comparisons of the groups at both ends of the line are shown. CT, computed tomography; HU, Hounsfield units; MMA, mean muscle attenuation; SAT, subcutaneous adipose tissue; SMI, skeletal muscle area index; VAT, visceral adipose tissue.

Table 3.

Multivariable logistic regression analyses to examine CT‐measured body compositions associated with low physical performance.

Variables	Adjusted odds ratio (95% CI)	P
SMI a	1.27 (0.74, 2.17)	0.385
MMA a	0.52 (0.29, 0.91)	0.023
SAT area a	1.11 (0.70, 1.74)	0.657
VAT area a	1.13 (0.65, 1.96)	0.669

Note: Adjusted odds ratios per interquartile range increase are shown.

Abbreviations: CI, confidence intervals; MMA, mean muscle attenuation; SAT, subcutaneous adipose tissue; SMI, skeletal muscle area index; VAT, visceral adipose tissue.

^a Each body composition was analysed separately using multivariable logistic regression models adjusted for age, sex, NYHA class and diabetes mellitus.

Among sarcopenic patients, receiver operating characteristic analyses were performed to evaluate the predictive value of MMA for sarcopenia severity (Figure 3 ). In sarcopenic male patients, MMA demonstrated superior predictive value with an AUC of 0.81 (95% CI: 0.67–0.95) compared with SMI (AUC 0.54, 95% CI: 0.33–0.74, P = 0.039), with the optimal cut‐off value being 32.2 HU (sensitivity 87%, specificity 65%). Similar trends were observed in sarcopenic female patients (AUC 0.75, 95% CI: 0.56–0.94 for MMA vs. AUC 0.56, 95% CI: 0.35–0.78 for SMI, P = 0.221), with the optimal cut‐off value being 26.0 HU (sensitivity 89%, specificity 64%). The Venn diagram analysis revealed that severe sarcopenia was enriched in the group that met all three criteria: reduced grip strength, low SMI and low MMA (Figure S2).

Figure 3.

Receiver operating characteristic curves for discriminating severe sarcopenia in patients with sarcopenia. In male patients (left panel), the optimal mean muscle attenuation (MMA) cut‐off value was 32.2 HU, with a sensitivity of 87% and specificity of 65%. The AUC was 0.81 (95% CI: 0.67–0.95), which was significantly higher than that of skeletal muscle area index (SMI) (0.54, 95% CI: 0.33–0.74; P = 0.039). In female patients (right panel), the optimal MMA cut‐off value was 26.0 HU, with a sensitivity of 89% and specificity of 64%. The AUC was 0.75 (95% CI: 0.56–0.94), showing a similar trend but without a significant difference from SMI (0.56, 95% CI: 0.35–0.78; P = 0.221). The red point indicates the optimal cut‐off value of MMA based on the highest Youden index. AUC, area under the receiver operating characteristic curve; CI, confidence interval; HU, Hounsfield units; MMA, mean muscle attenuation; SAT, subcutaneous adipose tissue; SMI, skeletal muscle area index; VAT, visceral adipose tissue.

Distinct amino acid profiles were independently associated with MMA

We explored plasma amino acid profiles contributing to MMA variation. In univariate linear regression analyses, aspartate, glutamate, proline, BCAAs (valine, isoleucine, leucine), β‐alanine, histidine and the Fischer ratio showed positive associations with MMA [regression coefficients (β) ranging from 1.4 to 2.8 HU per IQR increase], while citrulline, cystine, ornithine and 3‐methylhistidine demonstrated negative associations (β ranging from −1.4 to −2.4 HU per IQR increase, Table S2). As previously reported, ¹⁷ , ¹⁸ female sex and age were the strong determinants of MMA, with regression coefficients of −8.7 HU (95% CI: −11.0 to −6.5) and −3.7 HU per IQR increase (95% CI: −5.2 to −2.2), respectively, which were substantially larger than those observed for any amino acid parameters. MMA was also positively correlated with SMI. Nevertheless, in multivariable regression analysis adjusted for SMI and established determinants of MMA (age, sex and BMI), ¹⁷ , ¹⁸ leucine (β = 1.6 per IQR increase, 95% CI: 0.3 to 2.9), β‐alanine (β = 1.5 per IQR increase, 95% CI: 0.2 to 2.9), 3‐methylhistidine (β = −1.1 per IQR increase, 95% CI: −2.0 to −0.2) and the Fischer ratio (β = 1.3 per IQR increase, 95% CI: 0.3 to 2.4) remained independently associated with MMA (Table 4 ).

Table 4.

Multivariable linear regression analyses to examine the association between MMA and plasma amino acids.

Variables		Coefficient (95% CI)	P
Amino acid (nmol/mL) a
Aspartate	3.8 [3.2, 4.5]	0.6 (−0.3, 1.5)	0.204
Glutamate	62.2 [49.0, 78.1]	0.6 (−1.0, 2.1)	0.463
Proline	180.4 [154.5, 215.3]	1.1 (−0.1, 2.3)	0.074
Citrulline	41.0 [32.4, 51.5]	−1.1 (−2.4, 0.2)	0.096
Valine	239.2 [211.8, 265.9]	1.0 (−0.2, 2.2)	0.099
Cystine	65.6 [55.4, 76.0]	−0.6 (−1.9, 0.7)	0.370
Isoleucine	73.8 [65.6, 85.0]	0.5 (−0.7, 1.7)	0.405
Leucine	127.3 [113.3, 151.0]	1.6 (0.3, 2.9)	0.020
β‐Alanine	7.3 [5.7, 9.2]	1.5 (0.2, 2.9)	0.028
Ornithine	112.8 [96.3, 129.5]	−0.9 (−2.0, 0.2)	0.104
Histidine	76.4 [70.5, 85.0]	1.0 (−0.2, 2.3)	0.106
3‐Methylhistidine	7.5 [5.5, 10.4]	−1.1 (−2.0, −0.2)	0.014
Fischer ratio a	3.2 [2.9, 3.5]	1.3 (0.3, 2.4)	0.015

Note: Values are presented as median [interquartile range (IQR)]. Regression coefficients per IQR increase are shown.

^a Each amino acid or Fischer ratio was analysed separately using multivariable linear regression models adjusted for sex, age, BMI and SMI. Abbreviations are explained in Tables 1 and 2.

Discussion

This study reveals two key findings in sarcopenia assessment for HF patients. First, we demonstrated that evaluating muscle area and attenuation through single‐slice abdominal CT enables accurate diagnosis and severity assessment of sarcopenia in HF patients with reduced muscle strength. Second, we identified specific plasma amino acid profiles that correlate with muscle quality alterations. These findings provide both a practical diagnostic tool using standard CT imaging and new insights into the metabolic pathways involved in muscle quality alterations in HF patients.

Agreement between CT and DXA: Implications for sarcopenia diagnosis

In patients with HF and reduced muscle strength, we found that CT‐measured SMI cut‐off values (54 cm²/m² for males and 40 cm²/m² for females) accurately identified sarcopenia, with particularly high accuracy in males. These cut‐off values closely aligned with consensus ranges found in two previous systematic reviews in non‐HF populations (52–55 cm²/m² for males and 39–41 cm²/m² for females), one of which focused primarily on cancer patients. ¹⁹ , ²⁰ However, direct comparison requires careful consideration due to methodological variations, primarily in the HU ranges used to define SMA. ²⁰ Moreover, oncology studies often determine threshold values based on mortality prediction rather than sarcopenia diagnosis.

As previously reported, ²¹ the agreement between CT and DXA for identifying low muscle mass was limited. In our whole study population, the correlation between CT‐measured SMI and DXA‐measured ASMI was moderate (R = 0.71), and the AUC for SMI in discriminating low ASMI (defined by AWGS 2019 criteria) was 0.79 for males and 0.72 for females. However, when CT assessments were applied to patients with reduced muscle strength, diagnostic performance for sarcopenia was significantly improved (Table 2 ). Thus, the sub‐optimal agreement between CT and DXA becomes clinically acceptable for sarcopenia diagnosis in patients pre‐selected by reduced grip strength. The lower agreement between SMI and ASMI in females may be explained by sex‐specific differences in regional muscle distribution and lower absolute muscle mass, which could make single‐slice trunk measurements less representative of appendicular muscle status compared with males. ²²

Although DXA is currently recommended for sarcopenia assessment over CT imaging because of feasibility, acceptable accuracy, safety and low cost, ²³ it has significant practical limitations: its availability in clinical settings is actually lower than expected, and clinicians rarely order DXA specifically for muscle mass measurement. ²⁴ , ²⁵ CT imaging offers a practical advantage as elderly HF patients frequently undergo CT scanning for various comorbidities, allowing opportunistic body composition analysis without additional testing.

Myosteatosis as a determinant of physical performance in HF

In this study, we demonstrated that muscle attenuation (measured as MMA) rather than muscle area determines physical performance, especially in male patients. While the term ‘myosteatosis’ broadly encompasses intramuscular lipid infiltration and/or intermuscular adipose tissue depending on the study, ²⁰ our approach (measuring the attenuation of skeletal muscle defined by −30 to 100 HU range) primarily reflects the former, as validated by Goodpaster et al.'s seminal study investigating the correlation between CT attenuation and muscle biopsy lipid staining. ²⁶ The previously reported strong correlation of CT‐derived muscle attenuation with magnetic resonance imaging (MRI)‐derived proton density fat fraction, another gold standard for assessing myosteatosis, further supports the validity of our methodology. ²⁷ , ²⁸ Although the relative contribution of lipids in the two compartments to muscle dysfunction remains unclear, we focused on intramuscular lipid infiltration because muscle attenuation can be measured simultaneously during SMA measurements without additional imaging analysis.

Previous studies have demonstrated associations between CT‐assessed myosteatosis and physical performance, primarily focusing on lower extremity muscles: quadriceps muscle attenuation correlated with SPPB in geriatric outpatients, ⁸ and thigh intermuscular lipid infiltration predicted exercise capacity in patients with cardiomyopathy. ²⁹ In contrast, the Health ABC Study of community‐dwelling older adults revealed that trunk muscle attenuation contributes more significantly to physical performance than thigh muscle measurements. ⁷ The weaker association we observed in females may be partly attributed to their naturally lower baseline muscle attenuation, ¹⁷ , ¹⁸ which potentially makes myosteatosis‐related changes less detectable. Nevertheless, it is notable that the association of MMA and low physical performance remained significant after adjusting for sex (Table 3 ).

Mechanistically, intermuscular and intramuscular lipid infiltration impairs muscle function independently of muscle mass through distinct pathways. These pathways include inflammation, insulin resistance, anabolic resistance, capillary rarefaction, impaired neuromuscular activation and increased muscle stiffness. ³⁰ Furthermore, myosteatosis progresses more rapidly than muscle mass loss with aging and reduced physical activity levels. ¹⁸ , ³¹ Collectively, these findings suggest that myosteatosis assessment through CT‐measured muscle attenuation provides a sensitive and clinically relevant early marker of physical performance decline in HF patients.

Metabolic signatures of myosteatosis: Insights from amino acid profiling

While several studies have investigated the relationship between plasma amino acids including metabolomics techniques and skeletal muscle mass in older adults or HF patients, ³² , ³³ , ³⁴ , ³⁵ no previous studies have reported their association with qualitative muscle changes, particularly myosteatosis. We have recently reported that low plasma BCAAs might reflect fatty muscle degeneration in diabetic HF patients, based on body composition analysis using DXA. ⁹ The current study expands our previous findings and revealed that reduced muscle attenuation was independently associated with decreased levels of leucine, β‐alanine, and the Fischer ratio, and elevated 3‐methylhistidine, even after adjusting for skeletal muscle mass and established determinants of muscle attenuation.

3‐Methylhistidine is a marker of muscle protein breakdown and leucine is an essential amino acid required for protein synthesis, suggesting that myosteatosis likely progresses through an anabolic‐catabolic imbalance. ⁹ Similar patterns of altered 3‐methylhistidine and BCAA profiles have been reported in elderly patients with multiple comorbidities and frailty, indicating their involvement in sarcopenia progression beyond HF populations. ³⁶ Indeed, plasma leucine was reduced in community‐dwelling older adults who had low muscle function even at the stage of preserved muscle mass (i.e., dynapenia). ³⁷ Reduced plasma β‐alanine levels, a rate‐limiting precursor for carnosine synthesis may also contribute to muscle dysfunction because carnosine enhances skeletal muscle function through several mechanisms including antioxidant effects. ³⁸ Notably, we previously demonstrated that 3‐methylhistidine, valine and β‐alanine independently and additively predicted poor outcomes in HF patients, suggesting that these amino acid alterations also have prognostic implications. ¹²

The Fischer ratio, reflecting the balance between BCAAs and aromatic amino acids, is characteristically decreased in liver cirrhosis, a condition also associated with sarcopenia and myosteatosis. ³⁹ Given that hepatic dysfunction and a reduced Fischer ratio in HF are associated with poor prognosis, ⁴⁰ common mechanisms may underlie sarcopenia progression in both HF and liver cirrhosis.

Implications for sarcopenia screening and potential therapeutic strategies

Our findings have important clinical implications for HF management. First, CT‐measured muscle assessment demonstrated high diagnostic performance for sarcopenia when applied to patients with reduced muscle strength. This supports the incorporation of handgrip strength into routine HF examination as a ‘new vital sign’. ⁴¹ Combining grip strength, a simple and reliable measure, with CT imaging already performed for routine clinical purposes enables an efficient and practical sarcopenia screening strategy, allowing targeted use of DXA and comprehensive physical performance testing for patients requiring further evaluation (Figure 4 ). Notably, we observed that among patients with reduced grip strength, 14% had preserved SMI but reduced MMA, and 31% of them had severe sarcopenia (Figure S2). Because EWGSOP2 permits sarcopenia diagnosis based on reduced muscle strength and low muscle quality without requiring low muscle mass, ² these findings highlight the limitations of relying solely on muscle mass measurements, such as DXA.

Figure 4.

Proposed algorithm for sarcopenia screening using single‐slice abdominal computed tomography imaging in heart failure patients.

Second, the significant association of a specific amino acid profile with MMA, independent of skeletal muscle mass, highlights the critical role of amino acid dysmetabolism in adverse muscle remodelling. Although oral BCAA supplementation has shown mixed results in older adults ⁴² and cirrhosis ⁴³ for sarcopenia treatment and β‐alanine supplementation failed to improve muscle function in chronic obstructive pulmonary disease, ⁴⁴ the observed association between these amino acids and muscle quality in HF suggests their potential therapeutic role in this population. Because myosteatosis progresses more rapidly than muscle mass loss with sarcopenia progression, ¹⁸ , ³¹ MMA may serve as a more sensitive surrogate marker than muscle mass for identifying the optimal timing and approach for interventions, including tailored nutritional and exercise therapies, to improve muscle quality and physical function in HF patients.

Limitations

There are several limitations in this study. First, selection bias may have existed because subjects were retrospectively enrolled at a single centre. Second, the methodology for CT‐based muscle assessment requires further standardization. Definitions of muscle area and attenuation vary across studies ²⁰ and as recent large‐scale population studies highlight, both metrics are influenced by sex and age. ¹⁷ , ⁴⁵ While establishing age‐ and sex‐specific reference values is an important goal for future research, we believe the optimal balance between precision and simplicity must be considered, as overly detailed criteria could compromise practicality. Notably, current international sarcopenia guidelines recommend sex‐specific, but not age‐stratified, cut‐off values for DXA‐measured muscle mass, ² , ³ supporting a similar, simplified approach for CT‐based diagnosis to maintain clinical utility. Third, the diagnostic accuracy of CT for sarcopenia was lower in females compared with males in this study. Whether this is due to biological sex differences or simply to the small number of female participants remains unclear. Further investigation in larger cohorts with sufficient numbers of both men and women is required. Fourth, because we relied solely on opportunistically acquired abdominal CT scans, this design precluded a direct comparison of diagnostic yield with CT measurements at other anatomical sites. Although mid‐thigh CT imaging may be preferable for intentional sarcopenia screening due to lower radiation exposure, our approach reflects the practical reality that abdominal scans are more frequently performed for various clinical indications.

Conclusions

This study demonstrates that single‐slice abdominal CT is a practical and accurate tool for diagnosing sarcopenia and assessing its severity in HF patients with reduced muscle strength. The identification of CT‐measured muscle attenuation as a marker of physical performance and its association with specific amino acid profiles, independent of muscle mass, provides novel insights for tailored cardiac rehabilitation. Although standardization of CT protocols is needed for widespread clinical applicability, this proof‐of‐concept study highlights the potential of incorporating CT‐based muscle assessment into routine workflows to improve sarcopenia detection and management in HF patients.

Conflict of interest statement

None.

Funding

This study was supported by grant 23K07481 (H. K.) from the Japan Society for the Promotion of Science, a grant from MSD Life Science Foundation, Public Interest Incorporated Foundation (H. K.), and Sapporo Medical University grants for programs promoting academic advancements (W. K.).

Supporting information

Table S1. Comparison of CT‐measured muscle and fat indices across different severities of sarcopenia.

Table S2. Univariate linear regression analyses to examine the association between mean muscle attenuation and plasma amino acids.

Figure S1. Correlation between DXA‐measured ASMI and CT‐measured abdominal SMI.

Figure S2. Venn diagram of grip strength, skeletal muscle area index (SMI), and mean muscle attenuation (MMA). The values indicate the number of individuals in each group, with those in parentheses representing individuals with severe sarcopenia.

Kawaharata, W. , Kouzu, H. , Ogura, K. , Yano, T. , Katano, S. , Numazawa, R. , Nagaoka, R. , Aida, H. , Ohori, K. , Nishikawa, R. , Fujito, T. , Nagano, N. , Tanaka, M. , and Furuhashi, M. (2025) Diagnostic value of computed tomography‐based muscle quality and metabolic signatures in heart failure with sarcopenia. ESC Heart Failure, 12: 3589–3600. 10.1002/ehf2.15371.