European Journal of Cardiovascular Medicine

Sarcopenia, characterized by progressive loss of skeletal muscle mass and function, has emerged as a major determinant of frailty, falls, and mortality in the elderly population. The European Working Group on Sarcopenia in Older People (EWGSOP2) defines sarcopenia as a muscle disease rooted in adverse muscle changes that accumulate across a lifetime, with low muscle strength and low muscle quantity or quality as key diagnostic criteria. Quantifying skeletal muscle mass accurately is critical for both research and clinical decision-making, particularly for diagnosing sarcopenia and assessing its relationship with comorbidities and outcomes in aging populations.^[1]

Dual-energy X-ray absorptiometry (DXA) is widely used for estimating appendicular lean mass (ALM) and deriving indices such as the Appendicular Skeletal Muscle Index (ASMI). DXA’s advantages include low radiation exposure, rapid scan time, and standardized reference values for sarcopenia thresholds. However, DXA measurements are indirect, influenced by hydration status, fat infiltration, and positioning errors, and may not reflect true muscle cross-sectional morphology. In contrast, computed tomography (CT) at the level of the third lumbar vertebra (L3) provides a precise quantitative measure of muscle cross-sectional area and radiodensity, serving as a gold standard for estimating whole-body muscle mass. L3 muscle area has shown strong correlations with total body skeletal muscle and has been validated for both sarcopenia assessment and prognostication in various clinical populations, including oncology and geriatric cohorts.^[2][3]

Despite its accuracy, CT’s higher radiation exposure and cost restrict its use for routine screening. Therefore, establishing the degree of agreement between DXA-derived indices and CT-based measurements can optimize diagnostic protocols-allowing clinicians to rely on DXA when CT is unavailable or impractical while ensuring accuracy in estimating muscle mass. Previous studies have demonstrated moderate to strong correlations between ALM and L3 muscle area, but the degree of agreement and bias between these modalities in older Indian adults remains underexplored. Differences in body composition, ethnicity-specific cut-offs, and varying adiposity patterns may further influence the relationship between DXA and CT muscle estimates.^[4][5]

This cross-sectional agreement study thus aims to evaluate the concordance between DXA-derived sarcopenia indices and CT muscle measurements at L3 in older adults. By comparing quantitative indices from both modalities and determining their diagnostic thresholds for sarcopenia, the study seeks to validate the use of DXA as a reliable surrogate for CT-based muscle assessment in geriatric populations, enabling better screening and intervention strategies to prevent disability and frailty.

Aim

To assess the agreement between DXA-derived sarcopenia indices and CT-based skeletal muscle measurements at the L3 vertebral level in older adults.

Objectives

1. To quantify skeletal muscle mass using dual-energy X-ray absorptiometry (DXA) and computed tomography (CT) at L3 in older adults.
2. To determine the correlation and level of agreement between DXA-derived indices and CT-measured muscle cross-sectional area.
3. To identify diagnostic thresholds for sarcopenia on DXA corresponding to CT-based reference values.

Source of Data

Data were obtained from older adult patients (≥60 years) attending the Radiology and Geriatric Medicine departments of the tertiary care hospital, who underwent both DXA and abdominal CT scans as part of clinical evaluation.

Study Design

A hospital-based cross-sectional analytical study.

Study Location

Department of Radiodiagnosis and Imaging, in collaboration with the Department of Geriatric Medicine, at a tertiary care teaching hospital.

Study Duration

The study was conducted over 18 months, from January 2023 to June 2024.

Sample Size

A total of 120 older adults fulfilling inclusion and exclusion criteria were enrolled.

Inclusion Criteria

• Age ≥ 60 years.
• Availability of both abdominal CT (including L3 level) and whole-body DXA scans within three months.
• Informed consent obtained for participation.

Exclusion Criteria

• History of malignancy with cachexia or significant weight loss in the preceding six months.
• Presence of metallic implants or artifacts interfering with DXA or CT measurements.
• Chronic kidney disease, ascites, or severe edema affecting body composition accuracy.
• Unstable clinical condition or inability to undergo scanning safely.

Procedure and Methodology

Participants underwent DXA scanning using a standardized Hologic/GE Lunar system, positioned supine with arms at the sides. Appendicular lean mass (ALM) and total lean mass were recorded, and Appendicular Skeletal Muscle Index (ASMI = ALM/height²) was calculated.

CT images of the abdomen were retrospectively retrieved. A single axial slice at the mid-L3 vertebral level was selected. Skeletal muscle cross-sectional area (cm²) was measured using semi-automated segmentation software with standard Hounsfield Unit thresholds (-29 to +150 HU). The Skeletal Muscle Index (SMI = L3 muscle area/height²) was computed.

All measurements were performed by two blinded observers to minimize inter-observer bias.

Sample Processing

CT data were processed using validated image-analysis software, and DXA results were standardized using manufacturer-specific algorithms. Quality control and calibration were maintained daily for both modalities.

Statistical Methods

Data were analyzed using SPSS version 26.0. Continuous variables were summarized as mean ± SD. Correlation between DXA- and CT-derived indices was assessed using Pearson’s correlation coefficient. Agreement was analyzed through Bland–Altman plots and intraclass correlation coefficients (ICC). Receiver operating characteristic (ROC) analysis was conducted to identify optimal DXA cut-offs for CT-defined sarcopenia. A p-value <0.05 was considered statistically significant.

Data Collection

Demographic details, anthropometric parameters (age, sex, height, weight, BMI), DXA-derived lean mass, and CT-based muscle area were recorded in a structured proforma. All data were anonymized and stored securely for analysis

Table 1: Baseline profile by CT-defined sarcopenia status (N = 120)

Table 1 presents the baseline demographic and clinical characteristics of the study population (N = 120), stratified by CT-defined sarcopenia based on sex-specific L3 Skeletal Muscle Index (SMI) cut-offs. Participants with CT-defined sarcopenia were significantly older than those without sarcopenia (mean age 71.0 ± 6.3 vs. 68.1 ± 6.6 years; p = 0.021). The proportion of males was lower among the sarcopenic group (39.5%) compared to the non-sarcopenic group (58.4%), showing a statistically significant association (χ² = 4.24, p = 0.039). Although height differences were not statistically significant (p = 0.095), body weight and BMI were notably reduced in the sarcopenic cohort (mean weight 64.1 ± 9.6 kg vs. 72.7 ± 10.4 kg; BMI 25.4 ± 3.5 kg/m² vs. 27.7 ± 3.7 kg/m²; both p ≤ 0.001), emphasizing the leaner phenotype among those with muscle loss. Type 2 diabetes was more prevalent in sarcopenic individuals (44.2%) than non-sarcopenic (28.6%), though the difference did not reach statistical significance (p = 0.077). Hypertension and interval between CT and DXA scans were similar across groups (p > 0.05).

Table 2: Quantification of skeletal muscle by DXA and CT, overall and by sex (N = 120; Men = 62, Women = 58)

Table 2 summarizes the quantification of skeletal muscle parameters obtained by dual-energy X-ray absorptiometry (DXA) and computed tomography (CT) across all participants and by sex. Mean appendicular lean mass (ALM) and appendicular skeletal muscle index (ASMI) by DXA were significantly higher in men (21.6 ± 3.9 kg and 7.26 ± 0.86 kg/m²) compared to women (16.0 ± 3.0 kg and 6.26 ± 0.81 kg/m²), with p < 0.001 for both. Similarly, CT-derived L3 muscle cross-sectional area (CSA) and skeletal muscle index (SMI) were higher in males (147.5 ± 26.4 cm² and 47.8 ± 7.1 cm²/m²) than females (118.6 ± 22.3 cm² and 40.1 ± 6.2 cm²/m²; both p < 0.001). CT muscle radiodensity showed no significant sex difference (p = 0.317), suggesting comparable intramuscular fat infiltration between men and women.

Table 3: Correlation and agreement between DXA-derived indices (ASMI) and CT-measured muscle (SMI) (N = 120)

Table 3 explores the strength of correlation and agreement between DXA-derived ASMI and CT-measured SMI among the study participants. Pearson’s correlation revealed a strong positive linear relationship (r = 0.82; 95% CI 0.75–0.87; p < 0.001), corroborated by Spearman’s rho (ρ = 0.80; p < 0.001), indicating consistency across parametric and non-parametric analyses. Agreement analysis demonstrated high concordance between modalities, with Lin’s concordance correlation coefficient (CCC = 0.84; 95% CI 0.78–0.89) and intraclass correlation coefficient (ICC = 0.86; 95% CI 0.80–0.90), both statistically significant (p < 0.001). Bland–Altman analysis showed minimal mean bias (-0.06 SD) with limits of agreement from -1.02 to 0.90 SD, indicating negligible systematic deviation between DXA and CT estimates. No significant proportional bias was observed (β = -0.08; p = 0.213). Passing–Bablok regression yielded a near-unity slope (0.94; 95% CI 0.86–1.03) and non-significant deviation from linearity (p = 0.28), confirming that DXA-derived ASMI closely mirrors CT-based SMI across the measurement range.

Table 4: Diagnostic performance of DXA ASMI to detect CT-defined sarcopenia (reference = CT SMI cut-offs)

Table 4 evaluates the diagnostic performance of DXA-derived ASMI for detecting CT-defined sarcopenia. The area under the ROC curve (AUC) was 0.88 (95% CI 0.81–0.93; p < 0.001), indicating excellent discriminatory ability. The optimal ASMI threshold determined by Youden’s index was 6.70 kg/m², providing balanced sensitivity (83.7%) and specificity (83.1%). Positive and negative predictive values were 73.5% and 90.1%, respectively, while likelihood ratios (LR⁺ = 4.96, LR⁻ = 0.20) further support good diagnostic performance. Sex-stratified analysis revealed slightly higher diagnostic accuracy in women (AUC = 0.90; optimal cut-off = 5.52 kg/m²) compared to men (AUC = 0.86; cut-off = 7.02 kg/m²). Calibration analysis using the Hosmer–Lemeshow test demonstrated good model fit (χ² = 6.72, p = 0.568).

Figure 1: ROC curve with AUC

In our cohort, CT-defined sarcopenia clustered with higher age and a leaner anthropometric profile. Participants with sarcopenia were 3 years older (mean difference 2.9 years; p=0.021) and showed markedly lower weight and BMI (-8.6 kg and -2.3 kg/m², respectively; both p≤0.001). This mirrors the age-graded decline in muscle quantity emphasized by EWGSOP2 and the Asian consensus, which both describe accelerating losses beyond the seventh decade and lower BMI among sarcopenic older adults Li L et al..(2024)^[6]. The lower proportion of males in the sarcopenic group (39.5% vs 58.4%; p=0.039) aligns with population-based observations that sex differences in body size and adiposity interact with diagnostic cut-offs and may yield different apparent prevalence patterns when mass-based criteria are used Zhao R et al..(2022)^[7]. Height did not differ significantly, consistent with reports that stature per se contributes less than age and adiposity to sarcopenia classification once indices are height-adjusted. Trends toward higher diabetes prevalence in sarcopenia (44.2% vs 28.6%; p=0.077) are directionally concordant with literature linking insulin resistance to intramuscular fat and functional decline, though not always statistically definitive in moderate samples Chianca V et al..(2022)^[8].

Sex-stratified quantification (Table 2) reproduced well-known modality-agnostic gaps: men had higher DXA ALM/ASMI and higher CT L3 CSA/SMI (all p<0.001). These differences are consistent with foundational validation work showing L3 muscle area as a robust surrogate for whole-body muscle, with strong sex effects across modalities Lenchik L et al..(2021)^[9]. Radiodensity (HU), a proxy for myosteatosis, did not differ by sex (p=0.317), echoing studies where fat infiltration varies more with metabolic status and age than sex alone Eriksen CS et al..(2021)^[10].

The core validity signal of this study is the strong linkage between DXA-derived ASMI and CT-derived SMI (Table 3). We observed high linear associations (Pearson r=0.82; Spearman ρ=0.80) and substantial agreement (Lin’s CCC=0.84; ICC=0.86). Bland–Altman analysis showed minimal mean bias (-0.06 SD) with symmetric limits of agreement and no proportional bias. These findings closely track prior cross-modal comparisons where single-slice L3 muscle strongly predicts DXA lean mass, typically with r≈0.75–0.90 and small systematic offsets attributable to hydration and fat infiltration effects on DXA Lee CM et al..(2021)^[11]. The Passing–Bablok slope 0.94 with a non-significant linearity test (p=0.28) supports interchangeability for ranking individuals-an inference consistent with method-comparison papers that advocate CT when precise morphology is needed and DXA for scalable screening.

Diagnostic accuracy of ASMI for CT-defined sarcopenia was excellent (AUC 0.88; Table 4). The Youden-derived threshold of 6.70 kg/m² yielded balanced sensitivity/specificity (84% each) and favorable likelihood ratios (LR+≈5; LR-≈0.20). Sex-specific performance was also strong (AUC 0.86 in men with cut-off 7.02 kg/m²; AUC 0.90 in women with cut-off 5.52 kg/m²). These thresholds sit plausibly within ranges reported by Asian and oncology cohorts when mapping DXA-based indices to CT L3 SMI references-acknowledging that absolute cut-points vary with ethnicity, device, and reference standard Choudhary S et al..(2022)^[12]. Importantly, our calibration statistics (HL-χ² p=0.568) indicate that, at the chosen cut-off, predicted vs observed sarcopenia probabilities were well aligned, echoing the good calibration reported when harmonizing mass-based criteria across modalities in mixed geriatric samples

The present cross-sectional agreement study demonstrates a strong concordance between dual-energy X-ray absorptiometry (DXA)–derived sarcopenia indices and computed tomography (CT) muscle measurements at the L3 vertebral level among older adults. CT-defined sarcopenia was associated with advanced age, lower BMI, and reduced muscle mass, highlighting its clinical relevance in the geriatric population. The correlation between DXA-derived Appendicular Skeletal Muscle Index (ASMI) and CT-derived Skeletal Muscle Index (SMI) was excellent (r = 0.82, ICC = 0.86), with minimal systematic bias and narrow limits of agreement. ROC analysis confirmed that DXA effectively discriminated sarcopenia with an AUC of 0.88, indicating that DXA-derived indices can reliably approximate CT-based muscle estimates. Sex-specific thresholds (7.02 kg/m² for men; 5.52 kg/m² for women) provided optimal sensitivity and specificity, underscoring the need for population-tailored cut-offs. Overall, DXA offers a valid, practical, and low-radiation alternative to CT for screening sarcopenia in clinical and research settings, while CT remains the reference method when precise morphological assessment is required.

LIMITATIONS

1. The study design was cross-sectional; thus, causal or temporal relationships between muscle loss and clinical outcomes could not be inferred.
2. The sample size, though adequate for correlation analysis, was limited to a single-center geriatric cohort, potentially restricting generalizability across ethnicities and clinical settings.
3. CT and DXA scans were performed within a three-month window, which may introduce minor physiological variations in body composition.
4. Functional parameters such as handgrip strength or gait speed were not evaluated; hence, the study focused solely on structural muscle indices.
5. Hydration status and fat infiltration, known to influence DXA estimates, were not independently controlled.
6. CT-derived muscle radiodensity was analyzed in a single slice at L3, which may not represent global muscle quality variations.
7. The study did not compare results across different DXA or CT manufacturers, and device-specific calibration differences may affect reproducibility.
8. Finally, the analysis did not assess the impact of comorbidities such as chronic inflammatory or endocrine disorders that may confound muscle loss measurements.

1. Schluessel S, Mueller B, Drey M, Tausendfreund O, Rippl M, Deissler L, Martini S, Schmidmaier R, Stoecklein S, Ingrisch M. 3D deep learning-based muscle volume quantification from thoracic CT as a surrogate for DXA-Derived appendicular muscle mass in older adults. Aging Clinical and Experimental Research. 2025 Dec;37(1):1-1.
2. Lian R, Tang T, Jiang W, Jiang J, Yang M. Beyond the third lumbar vertebra (L3): Thoracic computed tomography-derived muscle mass and quality assessment as a practical alternative for body composition analysis. Nutrition. 2025 Dec 1;140:112894.
3. Muraki I. Muscle mass assessment in sarcopenia: a narrative review. JMA journal. 2023 Oct 16;6(4):381-6.
4. Wu W, Liu M, Zeng Q, Tang C, Huo J. Research progress on evaluation methods for skeletal muscle mass assessment in sarcopenia. Oncology Letters. 2025 Sep 1;30(3):1-3.
5. Nikodinovska VV, Ivanoski S. Sarcopenia, more than just muscle atrophy: imaging methods for the assessment of muscle quantity and quality. InRöFo-Fortschritte auf dem Gebiet der Röntgenstrahlen und der bildgebenden Verfahren 2023 Sep (Vol. 195, No. 09, pp. 777-789). Georg Thieme Verlag KG.
6. Li L, Xia Z, Zeng X, Tang A, Wang L, Su Y. The agreement of different techniques for muscle measurement in diagnosing sarcopenia: a systematic review and meta-analysis. Quantitative Imaging in Medicine and Surgery. 2024 Mar 7;14(3):2177.
7. Zhao R, Li X, Jiang Y, Su N, Li J, Kang L, Zhang Y, Yang M. Evaluation of appendicular muscle mass in sarcopenia in older adults using ultrasonography: a systematic review and meta-analysis. Gerontology. 2022 Jul 25;68(10):1174-98.
8. Chianca V, Albano D, Messina C, Gitto S, Ruffo G, Guarino S, Del Grande F, Sconfienza LM. Sarcopenia: imaging assessment and clinical application. Abdominal Radiology. 2022 Sep;47(9):3205-16.
9. Lenchik L, Barnard R, Boutin RD, Kritchevsky SB, Chen H, Tan J, Cawthon PM, Weaver AA, Hsu FC. Automated muscle measurement on chest CT predicts all-cause mortality in older adults from the National Lung Screening Trial. The Journals of Gerontology: Series A. 2021 Feb 1;76(2):277-85.
10. Eriksen CS, Kimer N, Suetta C, Møller S. Arm lean mass determined by dual-energy X-ray absorptiometry is superior to characterize skeletal muscle and predict sarcopenia-related mortality in cirrhosis. American Journal of Physiology-Gastrointestinal and Liver Physiology. 2021 May 1;320(5):G729-40.
11. Lee CM, Kang BK, Kim M. Radiologic definition of sarcopenia in chronic liver disease. Life. 2021 Jan 25;11(2):86.
12. Choudhary S, Wadhawan M, Dhawan S, Ganesan PK, Mittal P, Sahney A, Kumar A. Normative values of skeletal muscle indices for nutritional assessment and implications on definition of sarcopenia in Indian adult population. Indian Journal of Gastroenterology. 2022 Feb;41(1):69-76.