Abstract
Background/Aim: The Sarco-Detect study aim was to test the suitability of bioimpedance analysis (BIA) as a cost-efficient and practical alternative to computed tomography (CT) for diagnosing sarcopenia and to assess the agreement between CT and BIA.
Patients and Methods: In this study, the skeletal muscle cross-sectional area (SMA) was measured at the third lumbar vertebra (L3) on CT images, and the skeletal muscle index (SMI) was calculated. BIA skeletal muscle mass (SMM) and appendicular SMM (ASMM) were determined using the manufacturer’s software (SMM-Seca, ASMM-Seca) and using the Sergi and Kyle equations. All calculated masses were converted to height-normalized index values and compared with the European Working Group on Sarcopenia in Older People cut-offs.
Results: A total of 70 patients were included, with mean (± standard deviation) age of 66±11 years, body mass index of 25±5 kg/m2, CT-SMA of 135.55±32.01 cm2, and CT-SMI of 44.28±8.14 cm2. BIA results were 13.85±3.85 kg according to ASMM-Seca, 23.45±5.64 kg using ASMM-Sergi, and 21.86±5.74 kg using ASMM-Kyle. The highest Pearson correlation was determined between SMI-Seca and CT-SMI (r=0.839, p<0.01), followed by ASMI-Sergi (r=0.831, p<0.01). Comparing absolute SMM values, SMM-Seca showed the highest correlation with CT-SMA (r=0.917, p<0.01) followed by ASMM-Kyle and ASMM-Sergi (r=0.905 and r=0.904, respectively; p<0.01). When diagnosing sarcopenia, SMI-Seca showed the highest sensitivity of 50% (specificity 73%) followed by ASMI-Kyle at 38% (specificity 81%) and ASMI-Sergi at 13% (specificity 91%). Compared to the reference method (CT-SMI), 17 patients were falsely identified as having sarcopenia via the calculation of SMI-Seca and four patients with sarcopenia were not detected.
Conclusion: The results indicate limited congruence of BIA and CT in the diagnosis of sarcopenia.
Introduction
Sarcopenia is a muscle disease characterized by low skeletal muscle strength accompanied by low skeletal muscle mass (SMM) or low skeletal muscle quality (1). It is associated with numerous negative consequences for affected patients (2-5). Oncological patients form a particularly vulnerable group (6). Studies have shown an increased risk of postoperative complications in patients with cancer with sarcopenia and a negative effect on their overall and progression-free life (6-9).
Various methods exist in the diagnosis of sarcopenia, each presenting a unique set of advantages and disadvantages. Currently computed tomography (CT) and magnetic resonance imaging are considered to be the gold standard in the diagnosis of sarcopenia (1). The European Working Group on Sarcopenia in Older People 2 (EWGSOP2) recommends, in addition to dual X-ray absorptiometry (DXA), and bioelectrical impedance analysis (BIA) for the evaluation of muscle mass in clinical routine (1). BIA is simple and quick to perform, using easily portable instruments, it is inexpensive and does not require specialist staff to perform the measurement (1, 10, 11). Other methods are associated with significantly higher costs and personnel expenditure, such as CT, however, in patients with cancer, this is often routinely performed. Currently, the accuracy and the suitability of these methods for diagnosing sarcopenia are controversially discussed in current research (3, 12).
The aim of this study, the Sarco-Detect Study, was to identify the suitability of BIA using EWGSOP2 (1) criteria for diagnosing sarcopenia in an oncological cohort in Northern Germany. Oncological patients of the Interdisciplinary Outpatient Clinic of the University Cancer Center Schleswig-Holstein (UCCSH) in Kiel were included after obtaining their written consent.
Handgrip strength measurements (HGS) and BIA were performed. The reference method for SMM and sarcopenia diagnosis was CT. The CT reference values were CT skeletal muscle index (CT-SMI) and CT skeletal muscle area (CT-SMA). EWGSOP2 (1) refers to the work of van der Werf et al. (13) for their proposed reference values. Pearson correlation between the parameters of SMM determined by BIA and the reference method (CT) was measured, followed by a comparison of these two for agreement in diagnosing sarcopenia.
Patients and Methods
This cross-sectional study was conducted at the Interdisciplinary Outpatient Clinic of the UCCSH in Kiel from January until February 2023.
Patients 18 years and older, who had undergone a CT scan within the previous 28 days at the level of the third lumbar vertebra (L3) were able to participate. Exclusion criteria were non-interpretable CT imaging, pregnancy, connection to life-sustaining electronic systems (e.g., heart-lung machine), to portable electronic medical devices (e.g., pacemaker devices or infusion pumps) and the presence of electronic implants (e.g., pacemakers) and active prostheses.
CT. Threshold values for CT-based sarcopenia diagnosis varies among studies. Cruz-Jentoft et al. (1) refer to the study by van der Werf et al. (13) in the revised EWGSOP2. For most accurate application of the thresholds, the collection of CT-SMA and CT-SMI were methodologically adapted to the reference study as far as possible. The CT-scans were analyzed using the free version of Image J (National Institutes of Health, Bethesda, MD, USA). Image J shows excellent agreement in SMA measurement with SliceOmatic V5.0 software from TomoVision (Magog, QC, Canada) used by van der Werf et al. (13), as well as excellent inter- and intraobserver agreement (14). To determine CT-SMA, the following muscles (musculus obliquus externus abdominis, obliquus internus abdominis, transversus abdominis, quadratus lumborum, latissimus dorsi, erector spinae and the psoas major) present on an axial CT image at the level of L3 were measured. The Hounsfield Unit range was set to −29 to +150 HU. The slice that most clearly depicted both vertebral transverse processes of the L3 was chosen (13). A high correlation between the SMA on a single axial slice and the total SMM has been proven (15). CT-SMA varies depending on the selected height of the abdominal axial slice and has its maximum at the level of L3. This level has been established for determination of the SMM (16). Determination of CT-SMI followed. As muscle mass is associated with body height, the use of a height-normalized index value has been considered (17).The CT-SMI values were then compared with the cut-off values presented by van der Werf et al. (13) considering sex, age and body mass index (BMI) (Table I).
Computed tomography-determined skeletal muscle index (CT-SMI) cut-off values for determining a loss of muscle mass with regards to sex, age and body mass index (BMI) according to van der Werf et al. (13).
BIA. BIA was carried out with a medical Body Composition Analyzer 525 from Seca (Hamburg, Germany). The 8-point BIA was performed at a frequency of 50 kHz and a current of 100 μA (tolerance specifications −50%/+20%). Basic parameters such as weight, height and waist circumference were also measured. The last full meal had been eaten at least 4 h before the measurement, the last physical exercise was to have been performed 12 h before, and the last alcohol consumption 24 h before BIA. In addition, care was taken to ensure that the extremities had a normal temperature and were at the same level as the core of the body during the measurement. All measurements were taken in supine position on a non-conductive surface, after a resting time of at least 10 min. The arms and legs were not touching the core of the body for the measurement. The software of the Seca Body Composition Analyzer 525 shows values for the SMM (SMM-Seca) in addition to the raw data reactance (Xc) and resistance (R). The appendicular SMM was then calculated by adding the SMM of the extremities (ASMM-Seca). Following the EWGSOP2 (1) recommendation to calculate the ASMM from the raw data of the BIA using the Sergi equation, the ASMM-Sergi was additionally determined for each measurement (18). For comparison, the ASMM-Kyle was also determined from the raw data using the Kyle equation (19). The Sergi and Kyle equations for calculating the ASMM are:
Kyle: ASMM (kg)=[0.227×(Ht2/R)]+(0.095×weight)+ (1.384×sex)+(0.064×Xc)−3.964
Sergi: ASMM (kg)=(0.267×(Ht2/R)]+(0.095×weight)+ (1.909×sex)−(0.012×age)+(0.058×Xc)−4.211
where height (Ht) is given in centimetres and weight is given in kilogrammes; sex: men=1 and women=0.
All calculated muscle masses were also converted to height-normalized index values (e.g., ASMI-Seca) and compared with the EWGSOP2 (1) cut-off value recommendations. Cruz-Jentoft et al. (1) refer to Gould et al. (20) stating the cut-off values for height-normalized ASMM (ASMI). The ASMI cut-off values in the revised EWGSOP2 consensus are <7.0 kg/m2 for men and <5.5 kg/m2 for women. Gould et al. (20) determined their cut-off values using DXA.
HGS. The HGS was obtained with a hand dynamometer SH5002 from SAEHAN (Gyeongsangnam-do, Republic of Korea). The manufacturer’s instructions were followed. Three measurements of the dominant hand were taken for each person, interrupted by a break of at least 30 s. Only the maximum value of the three measurements was recorded (1, 21).
Statistics. The data were statistically analyzed using IBM SPSS Statistics version 29.0. (IBM Corp., Armonk, NY, USA). Descriptive statistics were used to calculate means and standard deviations, and in some cases the range, to describe the overall group and various subgroups. The data were additionally analyzed by means of bivariate Bravais-Pearson correlation CT and BIA.
Results
From January to February 2023, 74 patients were included. Four people were subsequently excluded due to missing data or CT imaging that could not be evaluated. In total 70 people, 28 women and 42 men, aged between 24 and 86 years were considered. Table II shows the patient characteristics for the total group as well as by sex. The average age of the total group was 66±11 years and was homogeneously distributed across both sexes. As expected, the average height of men (1.80±0.07 cm) was higher than that of women (1.66±0.06 cm). The mean BMI for the entire group was 25±5 kg/m2. On average, men (26.34±4.72 kg/m2) had a BMI of about 2 kg/m2 higher compared to women (24.26±4.04 kg/m2) (Table II). The time between CT imaging and BIA was on average 9 days with a standard deviation of 6.27 days, and a range of 0 to 28 days.
Average, minimum and maximum age, height, weight and body mass index (BMI) of the entire cohort and female and male subgroups in the Sarco-Detect study.
HGS. HGS served as the primary diagnostic criterion for sarcopenia. The mean value of HGS was 31.7±10.2 kg for the total group. It ranged 11 to 58 kg. The mean value for the male group (37.98±7.86 kg) was significantly higher compared to that of females (22.2±4.6 kg). Three men and three women fell below the reference value established according to their age group when comparing to the values used by the EWGSOP2 (1).
CT. The mean SMA for the total group was 135.55±32.01 cm2. The SMA was on average 43% larger in men (153.97±24.29 cm2) than in women (107.91±20.06 cm2). When normalized for height, males have a higher SMI of 47.61±6.92 cm2/m2 compared to the SMI of females 39.20±7.32 cm2/m2 (Table III). Using the SMI thresholds of Van der Werf et al. (13) two women and eight men were identified as having an SMI below their respective reference value.
Average, minimum and maximum of the computed tomography (CT)-determined skeletal muscle cross-sectional area (SMA) and skeletal muscle index (SMI) results for the entire cohort, and the female and male subgroups in the Sarco-Detect study.
BIA. Table IV shows the results of the descriptive statistics for relevant measurement parameters of the BIA. The Xc, R and resistance index served as raw data for the calculation of SMM-Seca and ASMM-Seca, as well as ASMM-Sergi and ASMM-Kyle. The mean values for the whole cohort differed by 8.01 kg between ASMM-Seca and ASMM-Kyle results. The difference in mean values between ASMM-Seca and ASMM-Kyle for the whole cohort was even greater at 9.6 kg.
Average, minimum and maximum of the relevant bioimpedance analysis (BIA) results for the entire cohort, and the female and male subgroups in the Sarco Detect study.
Comparing the results of ASMI-Seca with the cut-off values recommended by EWGSOP2 (1), all patients would be classified as pre-sarcopenic. Considering ASMM using the Sergi equation, three men and three women would be identified as pre-sarcopenic. Whilst with ASMM calculated with the Kyle equation, five men and 10 women would be identified as pre-sarcopenic.
As ASMI-Seca does not provide meaningful results for the assessment of pre-sarcopenia, SMI-Seca was analyzed for congruence of sarcopenia diagnosis. The cut-off values for height-normalized SMM determined via BIA were taken from the first European consensus on sarcopenia in the elderly population (22). They are therefore applied here with reservation. Cruz-Jentoft et al. (22) with reference to Janssen et al. (23) defined the limits for severe sarcopenia as <5.75 kg/m2 for women and <8.5 kg/m2 for men (24). Seven women and 14 men with values below the cut-off values were identified.
Correlation of the CT and BIA results. Good correlation between CT-SMI as the central diagnostic criterion of CT-based sarcopenia diagnostics and BIA results was found for all diagnostic criteria for the whole cohort, as well as for all subcategories, with correlation coefficients of 0.642-0.839 to (p<0.01). The effect strength can be judged as strong for all combinations according to Cohen (25). In the group overall, the greatest congruence was found between CT-SMI and SMI-Seca (r=0.839, p<0.01). For ASMI, the procedure via the Sergi equation showed the greatest congruence at r=0.831 (p<0.01). Additionally, a higher correlation was found between SMI and all BIA measures in the female group, as well as in the group with an overall lower BMI compared to their peers with similar height and age.
For non-height-normalized measures of the CT- and BIA-based sarcopenia diagnostics, higher Pearson correlations were found between CT-SMA and the respective BIA measurements. The correlation coefficients were between r=0.752 and r=0.917 (p<0.01). The absolute values showed a closer relationship to each other compared to the height-normalized values in Table IV and Table V. The effect strength according to Cohen (25) can be considered as strong for all combinations. Again, the correlation between CT-SMA and the SMM-Seca showed the strongest correlation in the whole group (r=0.917, p<0.01). The correlations between CT-SMA and ASMM-Seca, ASMM-Kyle and ASMM-Sergi were very similar (r=0.899-0.905, p<0.01). The correlation between SMA and all BIA measures were stronger for men and for patients with a lower BMI.
Congruence of bioimpedance analysis and computed tomography (CT) in diagnosis of sarcopenia in the Sarco Detect study. The skeletal muscle index (SMI) by CT, according to the manufacturer’s software (SMI-Seca), and using the Sergi (18) (SMI-Sergi) and Kyle (19) equations are given, as well as Sergi and Kyle indices for appendicular SMI (ASMI), with corresponding prevalence, sensitivity and specificity.
Agreement of CT and BIA in relation to (pre-)sarcopenia diagnosis. According to CT and appliance of the CT-SMI cut-off values, eight patients were classified as pre-sarcopenic. One person had a HGS that was below the cut-off and could therefore be diagnosed with sarcopenia. The results of the CT measurements serve as a reference for the assessment of the BIA measurement data in the present work. Values for sensitivity, specificity, as well as false- and true-positive and -negative diagnoses, refer accordingly to the results for CT-SMI (Table V).
SMI-Seca showed the highest sensitivity at 50%. However, compared to the reference method, this method falsely identified 17 patients as being pre-sarcopenic, which is reflected in a specificity of 73%. ASMI according to Kyle and SMI-Seca are clearly less sensitive in comparison.
Discussion
The aim of the present study was to examine BIA regarding its suitability for sarcopenia diagnostics and SMM determination. CT-SMI and CT-SMA determined via CT-L3-SMA served as reference values.
The measurements in this work were based on 61 scans performed during the venous phase and three during the arterial phase. Six scans were non-contrast-enhanced. It has been demonstrated that there is no statistically significant difference for the measurement of SMA within a HU range of −29 to +150 HU in relation to the contrast agent phase (26).
Classification of correlation coefficients according to Pearson. SMI-Seca and SMM-Seca showed the closest correlation (r=0.839 and r=0.917, p<0.01) with the CT measurement. The high correlation between CT and BIA-SMI is consistent with the results of other studies (27, 28). Considering their whole cohort, Mueller et al. (29) found strong Pearson correlation (r=0.794, p<0.01) between SMI-BIA and CT-SMI. Reasons for the stronger correlation between BIA and CT measurements in our study might be a shorter time span between CT imaging and BIA, and the choice of the BIA device used. In the present study, the mean time span was 9±6.27 days. The sensitivity of the bioimpedance measurement as a result of the correct combination of BIA device, study group and BIA equation has been demonstrated by several studies (28, 30, 31).
The Sergi and Kyle equations show promising results but are subject to the device-specific BIA equation. The higher accuracy of device-specific BIA equations was confirmed by Looijaard et al. (27), who compared values for SMM based on CT-L3-SMA and BIA. Their study showed that the manufacturer’s own software and BIA-equation led to muscle measurements with higher correlation with those of CT measurements (r=0.834, p<0.001) than using BIA equations developed by other manufacturers. These include the BIA equations of Janssen et al. (23) and Kyle et al. (19), which gave Pearson correlations of (r=0.635 and r=0.714, p<0.001), respectively.
In our study, the highest correlation with the reference method resulted from the use of the manufacturer’s own BIA equations. This would make the call to introduce device-specific limits plausible (30), which would go against the EWGSOP2 (1) goal of standardizing sarcopenia diagnostics.
However, ASMM-Sergi and the ASMM-Kyle values showed higher correlation coefficients than the manufacturer’s ASMM-Seca for all except one calculated correlations. The Seca Body Composition Analyzer 525 does not indicate a value for the total ASMM, but for the SMM of all four limbs individually. ASMM-Seca was therefore calculated retrospectively and manually by adding the values of the four limbs. In addition to the weaker correlation with the reference method, it was found that the determination of ASMM this way does not yield any meaningful results for the assessment of the presence of pre-sarcopenia. Applying the EWGSOP2 (1) cut-off values to the ASMI-Seca values, all individuals showed underscoring. Comparing this result to the prevalence of sarcopenia stated by different meta-analyses (32, 33), ranging from 1% to 33%, this method appears to be subjected to error. This suspicion is intensified in view of the previously described large deviations between the mean values of ASMM-Seca and ASMM-Sergi (6.67 kg) and ASMM-Seca and ASMM-Kyle (8.01 kg) (Table III).
Looking at the subgroup results it is apparent that the correlations between CT-SMA and all absolute BIA measures are stronger for men and for people with lower BMI. The table of height-normalized values also shows a higher correlation among those with lower BMI. This result is consistent with several other studies that found decreasing accuracy of BIA measurement with increasing BMI, especially for the single-frequency technique and severe obesity (34-36).
Classification of the concordance of (pre-)sarcopenia diagnosis. Although high correlation coefficients were found for the height-normalized SMM determined by SMI-Seca, ASMI-Sergi and ASMI-Kyle, the identification of pre-sarcopenia showed wide variations in the accuracy by the different methods. These results are in line with those of Gort-van Dijk et al. (37), who compared BIA measurements with the CT-based psoas muscle index. Sensitivity of SMI-BIA (23%) and ASMI-BIA (38%) were low. Other studies found very high sensitivities when comparing CT-L3-SMI and SMI-BIA (94% and 80%, respectively) with moderate specificities (54% and 52%, respectively) (38). Jones et al. (38) confirmed strong differences in prevalence found depending on the diagnostic method chosen. In our study, the reference method gave values that were below the reference values in 11% of patients, therefore showing a reduced skeletal muscle mass or presarcopenia. The BIA overestimated this prevalence by a factor of two to three using ASMI-Kyle (21%) and SMI-Seca (30%). For the ASMI-Sergi, the prevalence of reduced muscle mass or presarcopenia was 9%. The agreement of prevalence values is similar in Gort-van Dijk et al. (37). The reference method determined a prevalence of 26.5%, while BIA-SMM yielded 10.2% and BIA-ASMM 20.4% (37).
The reason for the low congruence in the diagnosis of sarcopenia by CT measurement and BIA is suspected to be due to the ASMI-cut-off values used. The reference values for ASMM recommended by EWGSOP2 (1) are based on DXA scans. Comparison with reference values based on measurements from another measurement method (CT vs. DXA) presents potential for inaccuracies in sarcopenia diagnosis. Buckinx et al. (39) found an intraclass correlation coefficient of 0.37 (95% confidence interval=0.25-0.48) when comparing ASMM-BIA to ASMM-DXA. Grover et al. (40) determined a correlation of r=0.86 (p<0.001) when comparing fat-free mass. Considering the results of Hansen et al. (41) and Ballesteros-Pomar et al. (42), who found an overestimation of SMM determined via BIA compared to DXA, underestimation of sarcopenia cases is likely. This influence should be considered when interpreting the results.
Study limitations. The Sarco-Detect study was conducted in a monocentric setting without randomization. Regarding the sensitivity of the BIA measurement, influences such as the composition of the study group, the BIA equation selected, the ethnicity and the collective of persons with oncological disease should be considered. A transfer of the results to other regions is possible only to a limited extent.
Only people who voluntarily agreed to participate were included. This may have influenced the results. The survey showed that people who suffered from a more severe course of an oncological disease were most likely less interested in participating or were less able to participate due to their state of health. The sample is therefore not representative of the basic population of oncological patients at the UCCSH in Kiel.
The sample size of 70 persons can be considered moderate. It should be noted that subgroup analyses are only meaningful to a limited extent due to unequal group sizes and, in this case, a small sample size (43). This applies especially when looking at the male and female subgroups.
Assessment of the accuracy of the BIA by determining the congruence between CT and BIA measurements, as done by Coeffier et al. (28), was not possible based on the available data. Coeffier et al. (28) determined the percentage of bioimpedance measurements that fell within 5% of the associated reference measurements to establish a value for the accuracy of BIA. The reason for this is the different physical dimension of the measurement data: while the BIA determines the muscle mass, the CT measures a cross-sectional area (28).
Conclusion
BIA showed high correlations with the CT reference method in the measurement of SMM. While the manufacturer’s own calculation formula showed the highest correlations with the reference method, which speaks in favor of the device-specific cut-off values required in some studies, the Sergi and the Kyle equations showed good performance. The determination of the SMI using the Seca equation resulted in the highest sensitivity. Furthermore, the high correlation indicates that the Seca Body Composition Analyzer 525 was a suitable choice for the determination of SMM in the examined patient collective. Although the applied cut-off values resulted in moderate congruence between CT and BIA in identifying patients at risk for sarcopenia, both methods diagnosed sarcopenia in different patients, resulting in false-positive and false-negative findings. Therefore, the cut-off values appear unsuitable for reliable use in the diagnosis of sarcopenia and question the interchangeability of both diagnostic methods.
Footnotes
Authors’ Contributions
L. Jochem: Data curation, formal analysis, writing. F. Hilpert: Data curation, formal analysis, writing. T. Wulff: Data curation, formal analysis, software. F. Stölzel: Project administration. I. Ratjen: Project administration. A. Letsch: Conceptualization, methodology, review and editing. T. Schmidt: Conceptualization, methodology, project administration, supervision, writing − review and editing
Conflicts of Interest
Anne Letsch declares honoraria from Astra Zenenca, Bayer, BMS, Böhringer, Ingelheim, Grünenthal, Janssen, Lilly, MSD, Novartis, Roche, Servier and Tesaro and has received research grants from Böhringer, Ingelheim and Amgen. All other Authors declare they have no conflict of interest.
Funding
The study was self-financed.
Artificial Intelligence (AI) Disclosure
No artificial intelligence (AI) tools, including large language models or machine learning software, were used in the preparation, analysis, or presentation of this manuscript.
- Received December 22, 2025.
- Revision received March 10, 2026.
- Accepted March 16, 2026.
- Copyright © 2026 The Author(s). Published by the International Institute of Anticancer Research.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
References
Jump to section
Related Articles
Cited By...
- No citing articles found.