Introduction
Sarcopenia, characterized by progressive loss of skeletal muscle mass, strength, and physical performance [1], elevates risks of disability, reduced quality of life, and mortality in older adults [2], [3]. With aging populations escalating its public health burden, objective diagnostic tools for muscle mass and function are urgently needed. Despite its clinical significance, sarcopenia remains underrecognized, impeding effective intervention [4].
Muscle strength serves as a significant predictor of fall risk and is known to predict both initial and repeated falls among the elderly [5]. Specifically, the loss of lower limb muscle strength (LLMS) that occurs with age is linked to functional impairments, which compromise independence in performing daily tasks [6], [7]. As a crucial factor influencing functional mobility [8], assessing quadriceps strength is highly valuable in geriatric care due to its wide-ranging utility—such as screening for sarcopenia and dynapenia, evaluating disability levels, and monitoring treatment outcomes.
Consequently, isokinetic dynamometry has emerged as the gold standard method for measuring muscle strength in clinical geriatric physical therapy and research. Muscle mass quantification employs modalities including dual-energy X-ray absorptiometry (DXA), computed tomography (CT), magnetic resonance imaging (MRI), and bioelectrical impedance analysis (BIA) [9], [10], [11], [12], [13]. DXA, a three-compartment model (fat mass, lean mass, bone mineral content) [14], [15], offers validated whole-body assessment with minimal radiation exposure, serving as a reference method for muscle mass evaluation [9], [11], [15], [16], [17].
Furthermore, BIA provides a portable, cost-effective estimate of fat-free mass and skeletal muscle mass via tissue impedance measurement [18], [19], [20], though variability in hydration status can affect its accuracy [8], [19], [21].
Despite the clinical utility of DXA and BIA for assessing muscle mass, their relationship with lower limb muscle strength (LLMS)—a critical functional indicator—remains inadequately characterized. BIA is widely used in community screening due to its convenience. However, European guidelines do not recommend this method.
Although some Asian guidelines endorse it, the evidence base remains weak, particularly lacking detailed research on the association between BIA and physical function, such as lower limb isokinetic muscle strength. Therefore, this study aimed to: (1) determine the correlations between LLMS and muscle mass as assessed by DXA and BIA; and (2) evaluate the association of these measures for functional muscle status. We hypothesize that BIA and DXA will show differential associations with upper and lower limb muscle strength, with BIA being more closely linked to handgrip strength and DXA-derived lean mass indices being more strongly associated with lower limb strength.
Materials and methods
Participants
This cross-sectional trial initially assessed 116 individuals from the community surrounding our hospital. After applying inclusion and exclusion criteria (see Fig. 1), a total of 94 ambulatory, cognitively intact, community-dwelling adults (59 women, 35 men) aged ≥40 years were enrolled.
All enrolled participants underwent both DXA and BIA body-composition analyses, along with lower extremity muscle-strength and hand-grip assessments. All were able to stand without assistance and had no recent history (i.e., within 6 months) of injury or major medical concerns. The study was conducted from March 2023 to November 2024.
The exclusion criteria included individuals with musculoskeletal conditions impairing muscle function (daily pain, hip or knee prostheses, recent fractures or sequelae, and other symptomatic functional impairments of the trunk muscle), neurological disorders, decompensated cardiovascular diseases, as well as pregnant women, active military personnel, and athletes. All participants completed muscle strength tests (including handgrip strength, knee extensor and flexor strength), BIA measurements, and a whole body DXA scan. Written informed consent was obtained prior to the conduct of any study procedure.
The study was approved by the ethics committee of our hospital.Fig. 1Participant selection for final analyses.
Demographic and anthropomorphic measures
Demographic data including age (years), sex, height (cm), weight (kg), and body mass index (BMI, kg/m2) were collected. Height was measured using a stadiometer (Harpenden stadiometer, Holtain Limited, Crosswell, UK) and weight was measured using an electronic scale (Yuanyan, Dahaibian Tech., Danyang, China). BMI was calculated as weight in kilograms divided by the square of height in meters.
The knee flexors and extensors strength measures
The knee flexors and extensors were measured using isokinetic dynamometry (Biodex system 3, Biodex Medical Systems, New York). Participants were positioned in the dynamometer with adjustable straps placed around their shoulders, waist, and right thigh to restrain movement. The center of the right knee joint was aligned with the axis of rotation of the dynamometer and the shank restrained by a resistance cuff.
The lever arm was fixed 3 cm above the lateral malleolus, and gravity correction was performed according to the manufacturer's procedure. The range of motion was set from 0° (full knee extension) to 90° (knee flexion), and all tests were performed in the concentric mode at an angular velocity of 60°/s. All tests were performed unilaterally on the right leg.
Subjects were instructed to extend and flex their leg 5 times as hard as possible. All subjects were verbally encouraged to give maximal effort during all tests by the same tester. The peak torque (Tpeak), peak torque to weight ratio (Tpeak/weight), mean torque (Tmean), mean power (Pmean) for both knee extension and flexion were recorded.
Handgrip strength measures
Handgrip strength was assessed with a calibrated Jamar dynamometer (Model 1, Los Angeles, CA, USA) on the dominant side following standardized protocols. The mean value of two consecutive trials, separated by a 1-min rest interval, was recorded as the final measurement.
Lean mass measures
Parameters related to lean tissue/fat-free mass were assessed using DXA and BIA. DXA-measured whole body composition was assessed using a fan beam whole body dual energy X-ray absorptiometry device (Hologic, Bedford, Massachusetts, USA). All scans were acquired and analyzed by a single technologist certified by the International Society for Clinical Densitometry.
DXA-measured total body lean mass, total body lean mass excluding the head, appendicular lean measurements, lower extremity lean measurements and upper extremity lean measurements were utilized as surrogates of muscle mass. The primary outputs provided from this analysis included estimates for total body lean soft tissue mass (LST, kg), appendicular lean mass (ALM, kg), lower extremity lean mass (LELM, kg), and upper extremity lean mass (UELM, kg), which were further normalized for height to yield the indices LST/height2 (kg/m2), ALM/height2 (kg/m2), LELM/height2 (kg/m2), and UELM/height2 (kg/m2).
BIA measurement (Tanita, Arlington Heights, Illinois, USA) was performed with the subject standing barefoot on the analyzer's electrodes and gripping the hand electrodes with arms extended parallel to the floor. For accurate results, key prerequisites must be met: the test should not be performed within 4 h of a large meal or alcohol consumption or within 12 h of strenuous exercise or dehydration, and the bladder should be emptied beforehand. The operator must input the subject's age, sex, and height into the device prior to initiating the measurement, during which the subject must remain still.
The primary outputs provided from this analysis include estimates for fat-free mass (FFM, kg), appendicular fat-free mass (AFFM, kg), lower extremity fat-free mass (LEFFM, kg), and upper extremity fat-free mass (UEFFM, kg). These absolute values are also normalized for height into indices: FFM/height2 (kg/m2), AFFM/height2 (kg/m2), LEFFM/height2 (kg/m2), and UEFFM/height2 (kg/m2).
Statistical analysis
Statistical analyses were performed using SPSS software (version 26; IBM Corp, Armonk, NY, USA). Continuous variables are presented as mean ± standard deviation for normally distributed data or median (interquartile range) for non-normally distributed data, as appropriate. Categorical variables are presented as numbers and percentages.
Based characteristics and body composition parameters were compared between male and female participants using independent samples t-tests for continuous variables. The relationships between variables were assessed using Pearson correlation coefficients (r). The strength of correlation was interpreted as follows: weak (0.00-0.30), moderate (0.30-0.70), or strong (>0.70).
Correlation analyses were conducted for the entire cohort and were also stratified by sex to examine sex-specific associations. To evaluate the predictive capacity of body composition parameters for muscle strength (handgrip strength, peak torque of knee extensors and flexors), multivariable linear regression analyses were performed. Three regression models were constructed for each predictor and outcome variable: Model 0 (Univariate): Unadjusted model.
Model 1: Adjusted for age. Model 2: Adjusted for age and body mass index (BMI). Model 3: Adjusted for age, BMI and sex.
The results are presented as standardized beta coefficients (β). A two-tailed p-value of < 0.05 was considered statistically significant.
Results
Participant characteristics
A total of 94 community-dwelling adults (59 women, 35 men) with a mean age of 69.2 ± 9.3 years were included. Men exhibited significantly greater height, weight, absolute muscle mass (by DXA and BIA), handgrip strength (34.3 ± 10.5 kg vs. 20.4 ± 7.8 kg, p < 0.001), and lower limb isokinetic strength (e.g., E_Tpeak: 63.9 ± 30.9 Nm vs.
46.5 ± 16.7 Nm, p < 0.001) compared to women (all p < 0.05), except for the ratio of peak extensor torque to weight (p = 0.319) (Table 1).Table 1Baseline subject characteristics and body composition by modality.Modality.VariablesTotalFemaleMalepSample size (n)945935Age (years)69.2 ± 9.368.8 ± 9.069.8 ± 9.90.627Weight (kg)65.2 ± 12.160.1 ± 8.873.9 ± 12.0< 0.001Height (m)1.6 ± 0.11.6 ± 0.11.7 ± 0.1< 0.001BMI (kg/m2)24.7 ± 3.324.3 ± 3.325.2 ± 3.30.215DXA, HologicLST (kg)40.1 ± 8.535.0 ± 4.048.7 ± 7.1< 0.001ALM (kg)16.7 ± 4.214.3 ± 2.320.8 ± 3.4< 0.001LELM (kg)12.6 ± 3.110.9 ± 2.015.5 ± 2.6< 0.001UELM (kg)4.1 ± 1.23.4 ± 0.65.3 ± 0.9< 0.001LST/height2 (kg/m2)15.1 ± 2.114.2 ± 1.516.6 ± 1.9< 0.001ALM/height2 (kg/m2)6.3 ± 1.15.8 ± 0.97.1 ± 0.9< 0.001LELM/height2 (kg/m2)4.7 ± 0.94.4 ± 0.85.3 ± 0.7< 0.001UELM/height2 (kg/m2)1.5 ± 0.31.4 ± 0.21.8 ± 0.3< 0.001BIA, TanitaFFM (kg)42.2 ± 8.736.4 ± 3.151.8 ± 6.1< 0.001AFFM (kg)18.6 ± 5.015.3 ± 1.724.0 ± 3.7< 0.001LEFFM (kg)14.2 ± 3.911.7 ± 1.418.5 ± 3.0< 0.001UEFFM (kg)4.3 ± 1.13.6 ± 0.55.5 ± 0.8< 0.001FFM/height2 (kg/m2)15.8 ± 1.814.7 ± 0.917.7 ± 1.5< 0.001AFFM/height2 (kg/m2)6.9 ± 1.26.2 ± 0.68.2 ± 1.0< 0.001LEFFM/height2 (kg/m2)5.3 ± 1.04.7 ± 0.56.3 ± 0.8< 0.001UEFFM/height2 (kg/m2)1.6 ± 0.31.5 ± 0.21.9 ± 0.2< 0.001Jamar DynamometerHandgrip strength (kg)25.5 ± 11.120.4 ± 7.834.3 ± 10.5< 0.0013D Biodex Isokinetic DynamometerE_Tpeak (Nm)53.0 ± 24.446.5 ± 16.763.9 ± 30.9< 0.001E_Tpeak/Wei (Nm/kg)0.8 ± 0.30.8 ± 0.30.9 ± 0.40.319E_Tmean (Nm)44.9 ± 21.838.9 ± 15.155.0 ± 27.4< 0.001E_Pmean (watt)21.5 ± 13.518.2 ± 9.527.0 ± 17.20.002F_Tpeak (Nm)30.3 ± 17.625.4 ± 8.838.5 ± 24.6< 0.001F_Tpeak/Wei (Nm/kg)0.5 ± 0.20.4 ± 0.10.5 ± 0.30.07F_Tmean (Nm)24.2 ± 13.021.3 ± 8.329.0 ± 17.50.005F_Pmean (watt)10.7 (6.5, 17.1)10.1 (6.6, 13.7)15.2 (6.5, 23.0)0.031F_ Pmean data is non-normally distributed and presented as median (interquartile range); other continuous variables are presented as mean ± standard deviation; BMI, body mass index; LST, lean soft tissue; ALM, appendicular lean tissue; LELM, the lower extremity lean tissue; UELM, the upper extremity lean tissue; FFM, fat-free mass; AFFM, appendicular fat-free mass; LEFFM, the lower extremity fat-free mass; UEFFM, the upper extremity fat-free mass; E_Tpeak, maximum knee extensor torque; E_Tpeak/Wei, the ratio between maximum knee extensor torque and weight; E_Tmean, mean knee extensor torque; E_Pmean, mean knee extensor power; F_Tpeak, maximum knee flexor torque; F_Tpeak/Wei, the ratio between F_Tpeak and weight; F_Tmean, mean knee flexor torque; F_Pmean, mean knee flexor power.
Correlations with age
In women, age showed significant negative correlations with several BIA-derived metrics (e.g., FFM: r = −0.264; AFFM: r = −0.382; LEFFM: r = −0.493; all p < 0.05), whereas no such correlations were observed in men (e.g., FFM: r = −0.183, p > 0.05) (Table 2).Table 2Correlations of DXA and BIA body composition parameters with age, stratified by sex.VariablesAgeAllFemaleMalerrrBMI0.1570.2040.072LST0.038−0.0700.057ALM0.018−0.0720.012LELM−0.001−0.091−0.009UELM0.0690.0240.071LST/height20.1620.1710.154ALM/height20.1000.0860.087LELM/height20.0720.0490.063UELM/height20.1570.1890.141FFM−0.064−0.264 b−0.183AFFM−0.085−0.382 a−0.165LEFFM−0.112−0.493 a−0.179UEFFM0.012−0.001−0.099FFM/height20.0170.046−0.115AFFM/height2−0.053−0.180−0.127LEFFM/height2−0.096−0.309 b−0.148UEFFM/height20.1000.202−0.036LST, lean soft tissue; ALM, appendicular lean tissue; LELM, the lower extremity lean tissue; UELM, the upper extremity lean tissue; FFM, fat-free mass; AFFM, appendicular fat-free mass; LEFFM, the lower extremity fat-free mass; UEFFM, the upper extremity fat-free mass.r, Pearson correlation coefficient.ap < 0.001.bp < 0.05.
Correlations between grip strength and knee strength
Significant positive correlations were observed between handgrip strength and all isokinetic knee strength parameters for both extensors and flexors (all P < 0.05) (Fig. 2). Handgrip strength was strongly correlated with E_Tpeak (r = 0.740, P < 0.0001) and F_Tpeak (r = 0.678, P < 0.0001).
Similarly, handgrip strength showed strong associations with E_Tmean (r = 0.701, P < 0.0001) and F_Tmean (r = 0.532, P = 0.0026). Handgrip strength was also correlated with E_Pmean (r = 0.652, P < 0.0001) and mean knee flexor power F_Pmean (r = 0.471, P = 0.0002). When normalized to body weight, the correlations remained significant though slightly attenuated (all P between 0.05 and 0.001) (Fig.
2).Fig. 2Relationships between grip strength and isokinetic knee strength parameters for extensor (E) and flexor (F) muscles. E_Tpeak, maximum knee extensor torque; E_Tpeak/Wei, the ratio between maximum knee extensor torque and weight; E_Tmean, mean knee extensor torque; E_Pmean, mean knee extensor power; F_Tpeak, maximum knee flexor torque; F_Tpeak/Wei, the ratio between F_Tpeak and weight; F_Tmean, mean knee flexor torque; F_Pmean, mean knee flexor power.
Correlations between muscle strength and body composition
Handgrip Strength was strongly correlated with BIA-derived parameters (e.g., FFM: r = 0.653; AFFM: r = 0.640, all p < 0.001) in the entire cohort, and these correlations remained significant in both sexes (females: FFM r = 0.315, p < 0.05; males: FFM r = 0.351, p < 0.05). In contrast, DXA-derived parameters showed weaker or non-significant correlations in sex-stratified analyses (Table 3).Table 3Correlation between handgrip strength and body composition, stratified by sex.VariablesHandgrip StrengthAllFemaleMalerrrAge−0.290a−0.370 a−0.464 aBMI0.1230.0170.096LST0.567 a0.1920.202ALM0.516 a0.0260.212LELM0.476a−0.0210.198UELM0.579 a0.1790.239LST/height20.400a0.0680.092ALM/height20.355a−0.0680.127LELM/height20.284a−0.1050.114UELM/height20.476a0.0980.150FFM0.653 a0.315 b0.351 bAFFM0.640 a0.382 a0.268LEFFM0.632a0.403 a0.232UEFFM0.640 a0.2580.367 bFFM/height20.594a0.2300.261AFFM/height20.591a0.302b0.190LEFFM/height20.587a0.326b0.152UEFFM/height20.555a0.1710.284LST, lean soft tissue; ALM, appendicular lean tissue; LELM, the lower extremity lean tissue; UELM, the upper extremity lean tissue; FFM, fat-free mass; AFFM, appendicular fat-free mass; LEFFM, the lower extremity fat-free mass; UEFFM, the upper extremity fat-free mass.r, Pearson correlation coefficient.ap < 0.001.bp < 0.05.
Lower Limb Strength (extensors and flexors) was moderately correlated with both DXA and BIA measures in the overall cohort (e.g., ALM with E_Tpeak: r = 0.456; AFFM with F_Tpeak: r = 0.440; both p < 0.001). However, these associations were primarily driven by strong correlations in men (e.g., ALM with E_Tpeak: r = 0.445, p < 0.001), while most correlations were non-significant in women (e.g., ALM with E_Tpeak: r = 0.112, p > 0.05) (Table 4, Table 5).Table 4Correlation between knee extensor indices and body composition.VariablesAllFemaleMaleE_TpeakE_Tpeak/WeiE_TmeanE_PmeanE_TpeakE_Tpeak/WeiE_TmeanE_PmeanE_TpeakE_Tpeak/WeiE_TmeanE_PmeanrrrrrrrrrrrrAge−0.231b−0.235b−0.196−0.221b−0.191−0.164−0.167−0.228−0.344b−0.336b−0.297−0.284BMI0.152−0.1680.1370.070−0.083−0.389a−0.116−0.1290.3110.0740.3100.186LST0.415a0.0460.417a0.355a0.018−0.299b−0.011−0.0250.363b0.1130.362b0.293ALM0.456 a0.0940.459 a0.401 a0.112−0.1910.0880.0750.445 a0.2090.450 a0.382 bLELM0.454 a0.0980.454 a0.400 a0.140−0.1450.1120.0990.445 a0.2070.447 a0.388 bUELM0.420a0.0770.432a0.368a−0.029−0.280b−0.031−0.0360.415b0.2000.427b0.342bLST/height20.308a−0.0410.311a0.256b−0.095−0.343 a−0.114−0.0660.3150.0930.3240.217ALM/height20.384a0.0430.389a0.337a0.042−0.2190.0230.0510.442 a0.2260.457 a0.356bLELM/height20.370a0.0470.370a0.326a0.077−0.1670.0550.0770.454 a0.2300.465 a0.374bUELM/height20.349a0.0240.365a0.303a−0.108−0.320b−0.103−0.0670.374b0.1970.395b0.280FFM0.457 a0.1210.458 a0.390 a0.140−0.1320.1190.0610.421 b0.1880.409 b0.336bAFFM0.445 a0.1060.444 a0.384 a0.154−0.1080.1380.1470.371b0.1210.350b0.272LEFFM0.450a0.1200.449 a0.392 a0.181−0.0490.1710.1910.371b0.1260.349b0.276UEFFM0.410a0.0550.409a0.339a0.050−0.2530.022−0.0010.335b0.0930.3210.230FFM/height20.419a0.0800.423a0.348a−0.006−0.250−0.0180.0130.391b0.1810.388b0.258AFFM/height20.405a0.0710.405a0.344a0.048−0.1650.0420.1180.3330.0990.3150.194LEFFM/height20.415a0.0930.416a0.360a0.079−0.0990.0790.1660.340b0.1090.3200.207UEFFM/height20.333a−0.0120.332a0.260b−0.043−0.304b−0.065−0.0350.2600.0540.2510.122BMI, body mass index; LST, lean soft tissue; ALM, appendicular lean tissue; LELM, the lower extremity lean tissue; UELM, the upper extremity lean tissue; FFM, fat-free mass; AFFM, appendicular fat-free mass; LEFFM, the lower extremity fat-free mass; UEFFM, the upper extremity fat-free mass; E_Tpeak, maximum knee extensor torque; E_Tpeak/Wei, the ratio between maximum knee extensor torque and weight; E_Tmean, mean knee extensor torque; E_Pmean, mean knee extensor power; F_Tpeak, maximum knee flexor torque; F_Tpeak/Wei, the ratio between F_Tpeak and weight; F_Tmean, mean knee flexor torque; F_Pmean, mean knee flexor power.r, Pearson correlation coefficient.ap < 0.001.bp < 0.05.Table 5Correlation between knee flexor indices and body composition.VariablesAllFemaleMaleF_TpeakF_Tpeak/WeiF_TmeanF_PmeanF_TpeakF_Tpeak/WeiF_TmeanF_PmeanF_TpeakF_Tpeak/WeiF_TmeanF_PmeanrrrrrrrrrrrrAge−0.192−0.212 b−0.295a−0.258b−0.131−0.112−0.172−0.224−0.315−0.330−0.465a−0.363bBMI0.165−0.0740.1070.0350.026−0.297b−0.063−0.1090.2180.0490.1940.087LST0.443a0.1670.363a0.332a0.089−0.2550.0740.0120.347b0.1650.3010.212ALM0.462 a0.1950.396a0.383 a0.150−0.1690.1450.1060.395 b0.2230.365 b0.313LELM0.455 a0.1910.392a0.383 a0.164−0.1360.1570.1200.395 b0.2230.364 b0.325UELM0.442a0.1880.373a0.351a0.044−0.2200.0510.0180.371b0.2090.347b0.261LST/height20.342a0.0810.264b0.221b0.018−0.248−0.045−0.0360.2690.1030.2390.113ALM/height20.386a0.1340.328a0.313a0.107−0.1640.0700.0780.355b0.1940.341b0.265LELM/height20.363a0.1240.311a0.301a0.123−0.1310.0870.0930.365b0.2000.347b0.288UELM/height20.374a0.1350.312a0.284a0.001−0.218−0.024−0.0090.3010.1630.2970.186FFM0.484 a0.238b0.400 a0.383 a0.149−0.1420.1490.0780.438 a0.2760.377 b0.300AFFM0.440a0.1930.357a0.360a0.142−0.1290.1240.1230.3110.1390.2590.212LEFFM0.436a0.1970.359a0.371a0.147−0.0880.1400.1570.2990.1320.2570.226UEFFM0.434a0.1700.335a0.309a0.102−0.2210.0540.0050.3220.1490.2410.139FFM/height20.454a0.209b0.358a0.330a0.091−0.1620.0010.0310.377b0.2280.3270.198AFFM/height20.391a0.1490.302a0.307a0.084−0.1300.0100.0860.2330.0700.1880.111LEFFM/height20.390a0.1580.308a0.325a0.089−0.0850.0290.1230.2260.0670.1900.136UEFFM/height20.359a0.1050.254b0.218b0.054−0.220−0.043−0.0310.2210.0700.1500.011BMI, body mass index; LST, lean soft tissue; ALM, appendicular lean tissue; LELM, the lower extremity lean tissue; UELM, the upper extremity lean tissue; FFM, fat-free mass; AFFM, appendicular fat-free mass; LEFFM, the lower extremity fat-free mass; UEFFM, the upper extremity fat-free mass; E_Tpeak, maximum knee extensor torque; E_Tpeak/Wei, the ratio between maximum knee extensor torque and weight; E_Tmean, mean knee extensor torque; E_Pmean, mean knee extensor power; F_Tpeak, maximum knee flexor torque; F_Tpeak/Wei, the ratio between F_Tpeak and weight; F_Tmean, mean knee flexor torque; F_Pmean, mean knee flexor power.r, Pearson correlation coefficient.ap < 0.001.bp < 0.05.
Notably, appendicular and leg-specific measures (ALM, LELM, AFFM, LEFFM) showed stronger associations with lower limb strength, while handgrip strength correlated more strongly with arm-specific measures (UELM: r = 0.579; UEFFM: r = 0.640; both p < 0.001) (Table 4, Table 5).
The associations between body composition parameters and muscle strength: linear regression analysis
In the fully adjusted model (Model 3: age, BMI, and sex), the associations between body composition parameters and muscle strength were modality-specific. (Table 6 and S1).Table 6Standardized β from linear regression of muscle strength on body composition parameters (n = 94).VariablesHandgrip strengthPeak torque, knee extensorsPeak torque, knee flexorsβββModel 0Model 1Model 2Model 3Model 0Model 1Model 2Model 3Model 0Model 1Model 2Model 3DXA-derived parametersLST0.577a0.591a0.705a0.2900.415a0.424a0.464a0.443e0.443a0.451a0.497a0.529eALM0.551a0.557a0.656a0.1540.456a0.460a0.517a0.579c0.462a0.466a0.520a0.551cLELM0.519a0.519a0.610a0.0610.454a0.454a0.516a0.510c0.455a0.455a0.512a0.466eUELM0.588a0.615a0.675a0.332e0.420a0.438a0.450a0.431e0.442a0.458a0.470a0.463eLST/height20.412a0.479a0.838a0.1770.308a0.354a0.524a0.2420.342a0.383a0.584a0.360ALM/height20.404a0.443a0.645a0.0180.384a0.412a0.585a0.468e0.386a0.409a0.568a0.409eLELM/height20.346a0.372a0.491a−0.0620.370a0.389a0.532a0.368e0.363a0.379a0.500a0.302UELM/height20.485a0.552a0.675a0.2370.349a0.395a0.436a0.2640.374a0.414a0.458a0.291BIA-derived parametersFFM0.664a0.645a0.674a0.428e0.457a0.444a0.437a0.542 e0.484a0.474a0.469a0.650 cAFFM0.656a0.632a0.682a0.383e0.445a0.428a0.429a0.4750.440a0.427a0.423a0.374LEFFM0.650a0.620a0.655a0.2740.450a0.429a0.424a0.4530.436a0.420a0.410a0.316UEFFM0.653a0.657a0.759a0.646a0.410a0.413a0.429a0.3800.434a0.436a0.456a0.451FFM/height20.608a0.614a0.762a0.479e0.419a0.422a0.474a0.4880.454a0.457a0.520a0.648 cAFFM/height20.611a0.595a0.733a0.3360.405a0.394a0.430a0.2800.391a0.382a0.405a0.095LEFFM/height20.608a0.581a0.676a0.1820.415a0.397a0.416a0.2650.390a0.375a0.379c0.050UEFFM/height20.568a0.608a0.886a0.660a0.333a0.360a0.429a0.1220.359a0.383a0.461a0.192Notes. Values are standardized β coefficients (95% confidence intervals).
Model 0, unadjusted; Model 1, adjusted for age; Model 2, adjusted for age and BMI; Model 3, adjusted for age, BMI and sex. BMI, body mass index; LST, lean soft tissue; ALM, appendicular lean mass; LELM, lower extremity lean mass; UELM, upper extremity lean mass; FFM, fat-free mass; AFFM, appendicular fat-free mass; LEFFM, lower extremity fat-free mass; UEFFM, upper extremity fat-free mass. Sex was coded 0 = female, 1 = male. a p < 0.001, b p = 0.001, c p < 0.01, d p = 0.01, e p < 0.05.
Handgrip strength. After adjustment for age, BMI, and sex, BIA-derived parameters remained significantly associated with handgrip strength, including UEFFM (β = 0.646, p < 0.001), UEFFM/height2 (β = 0.660, p < 0.001), FFM (β = 0.428, p < 0.05), AFFM (β = 0.383, p < 0.05), and FFM/height2 (β = 0.479, p < 0.05). In contrast, most DXA-derived parameters were no longer significant after full adjustment (e.g., LST: β = 0.290, p > 0.05; ALM: β = 0.154, p > 0.05), except for UELM (β = 0.332, p < 0.05).
These results indicate that BIA-derived parameters were more consistently associated with handgrip strength than DXA-derived parameters in the fully adjusted model.
Knee extensor strength. For peak torque of the knee extensors, DXA-derived appendicular and leg-specific lean mass indices remained significantly associated with strength after full adjustment (ALM: β = 0.579, p < 0.01; LELM: β = 0.510, p < 0.01; ALM/height2: β = 0.468, p < 0.05; LELM/height2: β = 0.368, p < 0.05). Among BIA-derived parameters, only FFM remained significant (β = 0.542, p < 0.05), while other BIA indices were attenuated.
These findings suggest that DXA-derived height-adjusted indices were more strongly associated with knee extensor strength.
Knee flexor strength. For peak torque of the knee flexors, both DXA- and BIA-derived parameters showed significant associations in Model 3. DXA-derived ALM (β = 0.551, p < 0.01), LELM (β = 0.466, p < 0.05), UELM (β = 0.463, p < 0.05), and ALM/height2 (β = 0.409, p < 0.05) were significant, as were BIA-derived FFM (β = 0.650, p < 0.01) and FFM/height2 (β = 0.648, p < 0.01).
Collinearity. VIF values for all models are reported in Supplementary Table S2. In Model 3, VIF for the body-composition variable ranged from 3.4 to 7.5, and for BMI from 1.6 to 4.6, indicating moderate-to-high collinearity.
Therefore, the corresponding regression coefficients in Model 3 should be interpreted with appropriate caution.
Comparison of DXA- and BIA-derived parameters in relation to muscle strength
In the fully adjusted models, BIA-derived parameters showed stronger and more consistent associations with handgrip strength, whereas DXA-derived height-adjusted indices were more strongly associated with lower limb strength. Specifically, for handgrip strength, BIA parameters such as UEFFM and UEFFM/height2 remained significant after adjusting for age, BMI, and sex, while most DXA parameters became non-significant. For knee extensor strength, DXA-derived ALM and LELM and their height-adjusted indices remained significant, whereas only FFM among BIA parameters was significant.
For knee flexor strength, both modalities showed significant associations, with BIA-derived FFM and FFM/height2 and DXA-derived ALM, LELM, UELM, and ALM/height2 being significant.
Discussion
In this study, we enrolled 94 community-dwelling adults aged ≥40 years and, within the same cohort, simultaneously assessed body composition by both DXA and BIA to examine its relationship with muscle strength, while directly comparing the associations of DXA-versus BIA-derived parameters with strength. Our study demonstrates that while both BIA and DXA are valuable tools for body composition assessment, their associations with muscle strength are modality-specific. After adjustment for age, BMI, and sex, BIA-derived parameters remained more strongly associated with handgrip strength, whereas DXA-derived height-adjusted appendicular and leg-specific lean mass indices remained more strongly associated with knee extensor strength.
For knee flexor strength, both modalities showed significant associations. These findings suggest that BIA may be more informative for upper-body strength, while DXA height-adjusted indices may be more informative for lower-body strength. To our knowledge, few studies have simultaneously examined lower-limb isokinetic strength in relation to both DXA- and BIA-derived body composition parameters in the same cohort.
Knee flexion and extension are crucial for maintaining upright posture, preventing falls, and preserving mobility and agility [22], [23]. We observed a segment-specific pattern: lower limb strength was more strongly associated with lower extremity lean mass (LELM: r = 0.454–0.455), while handgrip strength was more closely related to upper extremity lean mass (UELM: r = 0.579; UEFFM: r = 0.640), supporting the functional specificity of muscle mass distribution. Our results are consistent with the findings reported by Rush [24], who observed that lower limb-based physical performance was more strongly associated with lower limb muscle mass adjusted for intracellular and extracellular water distribution.
Our findings further support this earlier work by demonstrating a clear relationship between lower limb muscle strength and body composition.
Muscle mass and strength are two key references for diagnosing sarcopenia, and hand-grip strength has traditionally served as the diagnostic and screening tool [3]. Although handgrip strength is less affected by sex and age [25], its representativeness of whole-body strength has been questioned [26]. In 2022, Chan et al. [27] reported that in men, handgrip strength (HGS) was significantly correlated with upper extremity muscle mass (r = 0.576, p < 0.001), whereas in women, HGS showed a significant correlation with lower extremity muscle mass (r = 0.364, p < 0.001).
The main findings of that study are partially consistent with our results, though some discrepancies remain. These differences may be attributable to variations in the mean age of the study populations between the two cohorts.
Our findings are consistent with the results reported by Pisciottano et al. [6] in 2014, who observed that age was not significantly correlated with appendicular lean mass (ALM) measured by DXA in elderly women. Furthermore, our data complement their study by including male participants and multiple body composition variables, thereby addressing previous limitations of a single-sex focus and a narrow set of measures. The studied by Rush et al.
(24) and Kuchnia et al. (28), which utilized BIA to differentiate the distribution of body water, demonstrated that the combination of DXA and bioelectrical impedance spectroscopy (BIS) predicts jump power more effectively than traditional measures of sarcopenia [24], [28]. Our results indicate slight differences in the correlations between age and body composition as assessed by DXA versus BIA, which may be attributed to the fundamental differences in the underlying principles of the two techniques.
An important methodological consideration in this study was the inclusion of sex in the fully adjusted model. After adding sex to the model, the associations between DXA-derived parameters and handgrip strength were largely attenuated, whereas BIA-derived parameters remained significant. This suggests that sex is an important confounder in the relationship between body composition and upper-body strength.
In contrast, DXA-derived height-adjusted indices remained significantly associated with knee extensor strength after full adjustment, supporting their specificity for lower-limb strength. We also observed moderate-to-high collinearity between BMI and lean mass/height2 indices (VIF 3.4–7.5), which may partly explain the suppression effect observed in some models. Consequently, the corresponding regression coefficients should be interpreted with caution.
Future studies should consider using fat mass index or waist circumference instead of BMI in similar models.
This study has several limitations that should be considered when interpreting the results. (1) The cross-sectional design prevents causal inference regarding the relationships between body composition parameters and muscle strength. (2) The modest sample size (n = 94)—particularly in sex-stratified analyses (e.g., only 35 men)—may reduce statistical power and limit the generalizability of the findings.
(3) All strength measurements were conducted unilaterally (on the right side), which may not fully capture bilateral muscle function or potential limb asymmetries. (4) Although the analysis adjusted for age, BMI, and sex, other potential confounders—such as physical activity levels, nutritional status, and specific comorbidities affecting muscle mass and function—were not accounted for. (5) The study included only relatively healthy, community-dwelling adults, which may restrict the applicability of the results to clinical populations, institutionalized older adults, or those with more advanced sarcopenia or functional impairments.
(6) BMI and lean mass/height2 showed moderate-to-high collinearity in the fully adjusted models, which may have influenced the β estimates, and thus these results should be interpreted with caution; VIF values are reported to allow readers to interpret this limitation. (7) DXA and BIA measure lean soft tissue rather than true skeletal muscle mass. Recent studies using D3-creatine dilution have shown stronger relationships between actual muscle mass and strength than those observed with DXA or BIA.
Therefore, our findings should be interpreted as a comparison of two widely used surrogate measures, not as validation of their accuracy for muscle mass quantification.
In conclusion, after adjustment for age, BMI, and sex, BIA and DXA showed modality-specific associations with muscle strength. BIA-derived parameters were more strongly associated with handgrip strength, whereas DXA-derived height-adjusted appendicular lean mass indices were more strongly associated with lower limb strength. These findings support the tailored use of BIA for upper-body and DXA for lower-body functional assessment, although neither method directly measures skeletal muscle mass.
Future studies using accurate muscle mass measures and formal statistical comparison between modalities are warranted.
CRediT authorship contribution statement
Jian Geng: Writing – original draft, Methodology, Investigation, Conceptualization. Fayan Liu: Writing – original draft, Methodology, Investigation. Kexin Wang: Validation, Investigation.
Chaoying Liu: Validation, Investigation. Zitong Cheng: Resources, Data curation. Yi Yuan: Validation, Project administration.
Bo Hu: Validation, Project administration. Jing Zheng: Validation. Qingyu Zhang: Investigation, Data curation.
Jiabao Mo: Writing – review & editing, Formal analysis. Xiaoguang Cheng: Writing – review & editing, Supervision, Project administration, Funding acquisition. Ling Wang: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.
Ethics declaration
Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.
This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was approved by the Institutional Review Board of Beijing Jishuitan Hospital. (Approval No.
201210_08)
Ethical approval and informed consent
This cross-sectional observational study was approved by the Ethics Committee of Beijing Jishuitan Hospital, Capital Medical University (Approval No.: JST-[K2024] No. [274]-01; Approval Date: May 19, 2024; Expiration Date: May 18, 2026).
Written informed consent was obtained from all 94 community-dwelling participants aged ≥40 years before any study procedures including DXA, BIA, handgrip strength and isokinetic knee strength measurement. All participants were fully informed of study procedures, potential risks and their right to withdraw voluntarily at any time without penalty.
All research procedures strictly comply with the Declaration of Helsinki (2022 revision), CIOMS international biomedical research ethical guidelines and China's Measures for the Ethical Review of Biomedical Research Involving Humans. All participant information was anonymized with unique identification codes to protect personal privacy. The original testing data are stored in a password-locked hospital server and will be retained for 5 years after publication in accordance with local ethical regulations.
This study does not involve collection or detection of human genetic samples, so additional genetic resource approval is not required. No protocol deviations, adverse events or safety incidents occurred during the entire research period (March 2023–November 2024).
Author eligibility & authorship statement
All listed authors fully satisfy the ICMJE authorship criteria, with no qualified contributors omitted and no inappropriate guest authorship assigned. Each author completed the following requirements:1.Made substantial intellectual contributions to study conception, experimental design, data acquisition, statistical analysis or result interpretation;2.Participated in drafting the manuscript or critical revision for key academic content;3.Reviewed and approved the final submitted version;4.Accept public accountability for the accuracy and integrity of their respective research portions.
Personnel only providing routine administrative support, simple equipment operation without methodological optimization, or funding provision without intellectual input are not included as authors.
Data availability statement
The raw datasets supporting the findings of this study are available from the corresponding authors upon reasonable request, subject to ethical restrictions and participant privacy protection. All processed statistical data, correlation coefficients and regression analysis results are fully presented within the manuscript and supplementary materials.
Publication ethics statement
This manuscript is original research that has not been previously published in any form (including preprints), and is not under editorial consideration by any other journal. We will not submit this manuscript to other journals until Metabolism – Open issues a final editorial decision. All authors confirm that no eligible authors have been omitted, and all authors take full responsibility for the integrity and authenticity of all content in this paper.
Funding statement
This work was supported by the following funding projects:1.Beijing Municipal Public Welfare Development and Reform Pilot Project for Medical Research Institutes (Grant No.: JYY2023-8)2.10.13039/501100001809National Natural Science Foundation of China (Grant Nos.: 82371956, 82371957)3.Beijing Natural Science Foundation (Grant No.: L252006)4.Beijing Physician Scientist Training Project (Grant No.: BJPSTP-2024-08)
The funders had no role in study design, data collection, statistical analysis, manuscript drafting, or the decision to submit this manuscript for publication.
Declaration of competing interests
All authors declare that there are no financial or non-financial competing interests that could affect the design, execution, analysis, interpretation and writing of this manuscript.1.Financial conflicts: No author has received any financial benefits including research grants, honoraria, consulting fees or equipment donations from commercial manufacturers of DXA, BIA, isokinetic dynamometers or handgrip dynamometers. This study did not receive any industrial funding. None of the authors obtained travel subsidies or lecture fees from enterprises related to body composition testing and sarcopenia research within the past three years.2.Non-financial conflicts: No author holds relevant intellectual property rights, enterprise advisory positions or institutional affiliations that may introduce research bias.
No authors serve as consultants or board members for body composition testing equipment manufacturers.3.Study independence: The whole research was independently designed, conducted and analyzed by the research team of Beijing Jishuitan Hospital without any intervention from external commercial or non-commercial organizations. All statistical analyses were completed using SPSS version 26 under independent control of the authors.