Abstract
Background
Measurement of muscle mass is important in the diagnosis of sarcopenia. Current measurement equipment are neither cost‐effective nor standardized and cannot be used in a variety of medical settings. Some simple measurement tools have been proposed that are subjective and unvalidated. We aimed to develop and validate a new estimation equation in a more objective and standardized way, based on current proven variables that accurately reflect muscle mass.
Methods
Cross‐sectional analysis with The National Health and Nutrition Examination Survey database for equation development and validation. Overall, 9875 participants were included for development (6913 participants) and validation (2962 participants), for whom the database included demographic data, physical measurements, and main biochemical indicators. Appendicular skeletal muscle mass (ASM) was estimated by dual‐energy x‐ray absorptiometry (DXA) and low muscle mass was defined by reference to five international diagnostic criteria. Linear regression was used to estimate the logarithm of the actual ASM from demographic data, physical measurements, and biochemical indicators.
Results
This study of 9875 participants comprised 4492 females (49.0%), with a weighted mean (SE) age of 41.83 (0.36) years and range of 12 to 85 years. The estimated ASM equations performed well in the validation data set. The variability in estimated ASM was low compared with the actual ASM (R 2: Equation 1 = 0.91, Equation 4 = 0.89), with low bias (median difference: Equation 1 = −0.64, Equation 4 = 0.07; root mean square error: Equation 1 = 1.70 [1.69–1.70], Equation 4 = 1.85 [1.84–1.86]), high precision (interquartile range of the differences: Equation 1 = 1.87, Equation 4 = 2.17), and high efficacy in diagnosing low muscle mass (area under the curve: Equation 1 = 0.91 to 0.95, Equation 4 = 0.90 to 0.94).
Conclusions
The estimated ASM equations are accurate and simple and can be routinely applied clinically to estimate ASM and thus assess sarcopenia.
Keywords: Appendicular skeletal muscle mass, Dual‐energy X‐ray absorptiometry, Estimated equations, Muscle mass, Sarcopenia
Introduction
Sarcopenia is a progressive and systemic skeletal muscle disorder manifested primarily by loss of muscle mass and function. It is associated with an increased risk of adverse outcomes including frailty, physical disability, and mortality. 1 , 2 In recent years, several meta‐analyses have reported that sarcopenia increases the risk of mortality and functional loss approximately 2‐ to 4‐fold in community populations. 1 , 3 , 4 Therefore, early diagnosis, assessment and intervention of sarcopenia is crucial.
The measurement of muscle mass in the diagnosis and assessment of sarcopenia has been a clinical challenge. Although magnetic resonance imaging (MRI), computed tomography (CT) and dual‐energy x‐ray absorptiometry (DXA) have been used in clinical practice, 5 , 6 it remains impractical to use such large machines to measure muscle mass in community populations to assess the prevalence of sarcopenia. In contrast, bioelectrical impedance analysis (BIA) is sufficiently lightweight, but the low muscle mass cut‐offs based on BIA are considered method‐, device‐, and ethnicity‐dependent, which limits its standardization and accuracy. 7
Therefore, the European Working Group on Sarcopenia Older Persons (EWGSOP) 5 called for a search for a muscle mass measurement that is more cost‐effective, standardized, and repeatable in a variety of clinical settings. Recent studies have successively proposed some simple tools to diagnose sarcopenia in community populations. 8 , 9 , 10 However, in these studies, the choice of variables was subjective and singular and the results were trained on smaller databases and not validated.
Due to the current mixed bag of proxies for muscle mass, existing tools have performed poorly in subsequent validation by academics. 11 We sought to develop and validate a new estimating equation based on current variables proven to correlate with muscle mass using a more objective and rigorous method to accurately reflect muscle mass in community populations. This will help clinicians to decide whether more rigorous equipment is needed to determine the presence of sarcopenia, or even to replace complex mechanical measurements, for early inclusion in evidence‐based primary prevention programmes targeted at improving diet and exercise behaviours.
Methods
Study population
The National Health and Nutrition Examination Survey (NHANES) is a nationally representative health survey designed and administered by the National Center for Health Statistics (NCHS) at the Centers for Disease Control and Prevention (CDC). NHANES was approved by the NCHS Research Ethics Review Board (https://www.cdc.gov/nchs/nhanes/irba98.htm). Written informed consent was obtained from each participant.
Our study included 9875 US community participants who participated in the NHANES from 1999 to 2002 for development (6913 participants, 70% random sampling) and validation (2962 participants, 30% random sampling). The primary exclusion criteria were participants with missing data on relevant demographics, physical measurements, and main biochemical indicators (Figure S1). We used variables that have been confirmed as associated with muscle mass in previous studies to develop the estimated appendicular skeletal muscle mass (ASM, lean mass of the extremities based on DXA) equations.
Laboratory methods
NHANES 1999–2000 serum creatinine measurements used the Jaffe kinetic alkaline picrate method, while NHANES 2001–2002 used a Roche coupled enzymatic analysis, at the Cleveland Clinic Foundation (CCF) laboratory. There were significant differences in results between these two measurements. We referred to the official NHANES analysis notes and corrected the values for 1999–2000 according to the CCF laboratory standards by multiplying by 1.013 and then adding 0.147. Serum cystatin C was only measured in the NHANES 1999–2002. Serum cystatin C was measured in a subset of 4261 participants, including all participants aged 60 years or older with available specimens and a 25% random sample of participants aged 12 to 59 years. All assays of cystatin C were conducted using the Dade Behring N Latex Cystatin C assay. 12 Detailed specimen collections and processing instructions for other indicators are provided in the NHANES Laboratory Procedures Manual, which is available on the NHANES website.
Physical measurements
A review published in The Lancet in 2019 by Cruz‐Jentoft, 13 first author and corresponding author of the EWGSOP2 expert consensus, stated that ‘The most effective procedure to date is the use of DXA, which estimates lean mass’. Whole body DXA scans were acquired using Hologic QDR‐4500A fan‐beam densitometers (Hologic, Inc, Bedford, MA, USA) to collect data on muscle mass in NHANES participants over the age of 8 years. DXA exclusion criteria included pregnancy, weight > 300 pounds (136 kg, weight limit of the scanner), height over 6 feet 5 inches (DXA table limitations), history of radiographic contrast material (barium) use in the past 7 days, or nuclear medicine studies in the past 3 days. The estimate by DXA is lean mass, which is defined as total mass minus the mass represented by pixels containing mineral and lipid. 14 The ASM (kg), skeletal muscle mass index (SMI = ASM/height2, kg/m2) and body mass index (BMI)‐adjusted ASM [ASM/BMI, kg/(kg/m2)] derived from the above data are among the currently accepted indicators for the assessment of muscle mass in sarcopenia. 15 , 16 The estimation of ASM by using relevant variables, and thus the fitting of assessment indicators, seems to be a feasible approach.
Definitions of low muscle mass
In this study, low muscle mass was defined referring to five international diagnostic criteria: (1) EWGSOP 2 consensus 15 ; (2) EWGSOP1 consensus 17 ; (3) Asian Working Group for Sarcopenia (AWGS) consensus 18 ; (4) International Working Group on Sarcopenia (IWGS) consensus 19 ; and (5) Foundation of the National Institute of Health (FNIH) consensus. 16 Table S1 summarizes the operational definition for low muscle mass. Descriptive statistics were examined across sets of criteria (the operational definition for EWGSOP1 and IWGS are the same and described together below).
Statistical analysis
Descriptive statistics
We used the NHANES recommended weights to account for planned oversampling of specific groups. 20 Descriptive weighted statistics were used to describe the study population, with categorical characteristics summarized as counts (percentages) and continuous characteristics summarized as means (SE [standard error]). Continuous variables were compared by the two independent samples t‐test, and categorical variables were compared by the chi‐square test or Fisher's exact test.
Multivariate analysis of appendicular skeletal muscle mass
We used univariate linear regression, multiple linear regression, and stepwise regression to identify a set of variables that jointly estimated ASM. The stepwise regression model was developed from a training sample consisting of a random sample of 6913 of the 9875 participants. Because many estimation variables are not linearly correlated with ASM, we performed regressions of the log‐transformed ASM using natural‐scale and log‐scale estimation variables separately to eliminate nonlinear correlations. To facilitate clinical interpretation, the results were re‐expressed in terms of the original units. Consequently, the estimation equations are multiplicative, and regression coefficients refer to the change in geometric mean ASM associated with unit changes in the independent variable.
The following variables were considered for inclusion in the regression model: age, sex, ethnicity/race, height, weight, serum creatinine, serum cystatin C, serum albumin, serum urea nitrogen, serum calcium, serum phosphorus, urine protein, and urine creatinine. Variables with P‐value <0.01 in the univariate regression were included in the multivariate regression. A P‐value <0.001 in the multivariate regression was used as the criterion for entry of variables into the stepwise regression model. The merit of the equations was judged according to the Akaike information criterion (AIC).
Methods for comparing equations to estimate appendicular skeletal muscle mass
We first developed coefficients for each estimation equation (estimation variables selected by stepwise regression) using the data from the training sample to estimate the log ASM. Each estimation equation also included a multiplicative constant to account for any consistent bias that occurred when the equation was applied in the study. The regression coefficients identified in the training sample were applied to a separate validation sample consisting of the remaining 2962 participants to obtain the estimated ASM. These estimated ASMs were compared with the actual ASM in the validation sample to assess the performance of each estimation equation. In this way, separate data sets were used to construct the equations and their accuracy was evaluated after removing systematic biases. For each equation, we computed the total R 2 (percentage of variability in log ASM explained by the regression model), root mean square error (RMSE), and median and interquartile range of the distribution of absolute differences between the actual and estimated ASM in the validation sample and visualized them using Bland–Altman plots. The median indicated the typical size of error in the estimation of the ASM, and the interquartile range assessed the size of the larger errors occurring in each model. Receiver operator characteristic (ROC) curves were used to demonstrate the ability to make an ASM‐based diagnosis of low muscle mass and quantified by area under the curve (AUC), sensitivity, and specificity.
Development of final estimation equations
The estimated ASM equations and regression coefficients were derived from the training sample and updated based on data from 9875 participants. Therefore, the SEs of the regression coefficients in the final estimation equations of this study are slightly smaller than the SEs of the regression coefficients obtained from the training sample. Therefore, the accuracy of the final estimation equations may be slightly better than the accuracy assessed in the validation sample.
All data analyses were performed using R software (version 4.0.4; R Foundation for Statistical Computing, Vienna, Austria) (Data S1).
Results
Participant characteristics and prevalence
This study of 9875 participants comprised 4492 females (49.0%), with a weighted mean (SE) age of 41.83 (0.36) years and range of 12 to 85 years; 3304 participants (14.5%) were of Hispanic ancestry, 2061 (9.8%) of non‐Hispanic black ancestry, and 4196 (71.4%) of non‐Hispanic white ancestry. Mean height was 168.53 (0.12) cm, mean weight 76.53 (0.32) kg, mean BMI 26.86 (0.11) kg/m2, mean serum creatinine 0.87 (0.00) mg/dL, and mean serum cystatin C 0.93 (0.00) mg/dL. The mean actual ASM was 21.65 (0.10) kg with a range of 7.4 to 45.3 kg. Study population characteristics are listed in Table 1. The various operational definitions and prevalence (weighted) of low muscle mass are presented in Table S1. The prevalence of FNIH‐defined low muscle mass (8.6%) was smaller than that defined by the EWGSOP2 (14.0%), EWGSOP1/IWGS (18.9%), and AWGS (12.8%) in the total population.
Table 1.
Characteristics of the study population (weighted)
| Characteristic | Missing | Full data | Development data | Validation data | P‐value |
|---|---|---|---|---|---|
| n (%) | n = 9875 | n = 6913 (70%) | n = 2962 (30%) | ||
| Age (years), mean (SE) | 0 (0.0) | 41.83 (0.36) | 42.08 (0.33) | 41.28 (0.54) | 0.057 |
| <44, n (%) | 6215 (59.7) | 4300 (58.7) | 1915 (61.9) | 0.063 | |
| 45–59, n (%) | 1536 (23.2) | 1086 (23.6) | 450 (22.2) | ||
| ≥60, n (%) | 2124 (17.1) | 1527 (17.6) | 597 (15.9) | ||
| Female, n (%) | 0 (0.0) | 4492 (49.0) | 3179 (49.5) | 1313 (47.9) | 0.188 |
| Race/ethnicity, n (%) | 0 (0.0) | 0.207 | |||
| Hispanic | 3304 (14.5) | 2315 (14.6) | 989 (14.3) | ||
| Non‐Hispanic White | 4196 (71.4) | 2920 (71.2) | 1276 (71.8) | ||
| Non‐Hispanic Black | 2061 (9.8) | 1469 (10.1) | 592 (9.0) | ||
| Other | 314 (4.4) | 209 (4.2) | 105 (4.8) | ||
| Height (cm), mean (SE) | 0 (0.0) | 168.53 (0.12) | 168.43 (0.12) | 168.76 (0.23) | 0.187 |
| Weight (kg), mean (SE) | 0 (0.0) | 76.53 (0.32) | 76.38 (0.41) | 76.84 (0.49) | 0.481 |
| BMI (kg/m2), mean (SE) | 0 (0.0) | 26.86 (0.11) | 26.84 (0.14) | 26.89 (0.17) | 0.801 |
| ≤18.4, n (%) | 566 (3.9) | 395 (3.8) | 171 (4.0) | 0.972 | |
| 18.5–23.9, n (%) | 3163 (29.6) | 2239 (29.8) | 924 (29.1) | ||
| 24.0–29.9, n (%) | 3827 (41.6) | 2661 (41.4) | 1166 (42.1) | ||
| 30.0–39.9, n (%) | 2094 (22.6) | 1462 (22.7) | 632 (22.5) | ||
| ≥40.0, n (%) | 225 (2.3) | 156 (2.3) | 69 (2.3) | ||
| ASM (kg), mean (SE) | 0 (0.0) | 21.65 (0.10) | 21.59 (0.13) | 21.79 (0.14) | 0.330 |
| SMI (kg/m2), mean (SE) | 0 (0.0) | 7.52 (0.03) | 7.50 (0.04) | 7.54 (0.04) | 0.513 |
| ASM/BMI [kg/(kg/m2)], mean (SE) | 0 (0.0) | 0.82 (0.00) | 0.82 (0.00) | 0.82 (0.01) | 0.328 |
| Serum creatinine (mg/dL), mean (SE) | 0 (0.0) | 0.87 (0.00) | 0.87 (0.00) | 0.87 (0.01) | 0.604 |
| Serum cystatin C (mg/dL), mean (SE) a | 5614 (58.9) | 0.93 (0.01) | 0.94 (0.01) | 0.93 (0.01) | 0.475 |
| Serum albumin (g/dL), mean (SE) | 0 (0.0) | 4.42 (0.01) | 4.41 (0.01) | 4.43 (0.01) | 0.109 |
| Serum urea nitrogen (mg/dL), mean (SE) | 0 (0.0) | 13.56 (0.11) | 13.57 (0.11) | 13.54 (0.14) | 0.811 |
| Serum calcium (mg/dL), mean (SE) | 0 (0.0) | 9.48 (0.02) | 9.48 (0.02) | 9.48 (0.02) | 0.935 |
| Serum phosphorus (mg/dL), mean (SE) | 5560 (56.3) | 3.49 (0.01) | 3.50 (0.02) | 3.47 (0.02) | 0.250 |
| Urine protein (mg/L), mean (SE) | 48 (0.5) | 28.60 (2.21) | 31.18 (3.13) | 22.75 (1.69) | 0.027 |
| Urine creatinine (mg/dL), mean (SE) | 49 (0.5) | 135.06 (1.71) | 135.58 (1.81) | 133.90 (2.74) | 0.551 |
ASM, appendicular skeletal muscle mass; BMI, body mass index; SMI, skeletal muscle mass index.
Serum cystatin C was measured in a subset of 4261 participants, including all participants aged 60 years or older with available specimen and a 25% random sample of participants aged 12 to 59 years.
Description of the estimated appendicular skeletal muscle mass equations
We developed equations to estimate log ASM using univariate linear regression (Table S2), multiple linear regression (Tables S3 and S4), and stepwise multiple regression (Table S5) applied to a randomly selected training sample of 6913 participants. The estimated ASM equations for estimating log ASM included log serum creatinine, log serum cystatin C, and log weight, while age and height were on a natural scale. Tables S6 and S7 show the variables in the models. We describe the final equations based on data from all 9875 participants (Table 2). As expected, estimated ASM did not systematically deviate from actual ASM (Figure S2), although a few values for actual ASM were above estimated values when ASM was normal or high. The coefficient for Black persons was greater than 1.0 in the estimated ASM equations, which resulted in higher ASM estimates for Black persons than for other ethnicities, all other variables being equal. The estimated female to male ratio of estimated ASM ranged from 0.753–0.831, thus the estimated ASM for females was smaller than that for males. The estimated ASM was positively proportional to weight, height, and serum creatinine and inversely proportional to age and cystatin C. Weight and sex were the most important estimation variables.
Table 2.
Estimation equations from full data set
| Full data set | |
|---|---|
| Equation 1 | ASM = 0.595 × 0.998age × 0.825 [female] × 1.079[black] × 1.005height × weight0.677 × Cr0.143 × CysC‐0.104 |
| Equation 2 | ASM = 0.598 × 0.998age × 0.828 [female] × 1.005height × weight0.687 × Cr0.167 × CysC‐0.124 |
| Equation 3 | ASM = 0.578 × 0.997age × 0.831 [female] × 1.005height × weight0.674 × Cr0.129 |
| Equation 4 | ASM = 0.485 × 0.998age × 0.814 [female] × 1.006height × weight0.680 |
| Equation 5 | ASM = 0.939 × 0.997age × 0.764 [female] × weight0.774 |
| Equation 6 | ASM = 1.005 × 0.753 [female] × weight0.736 |
ASM, appendicular skeletal muscle mass; Cr, serum creatinine; CysC, serum cystatin C.
Comparison of performance
We compared the performance of these six equations (Table 2) in estimating ASM in a validation sample (Table 3). The highest R 2 value (0.91) was for Equation 1, which incorporated demographic (age, sex, ethnicity/race), physical (height, weight) and serum biochemical data (creatinine, cystatin C). The multiple regression model (Equation 4) based only on demographic and physical measurements was slightly less accurate (R 2 = 0.89). In addition, Equation 2, built by eliminating ethnicity/race alone, had a higher sensitivity for diagnosing low muscle mass (male: 86–91%; female: 85–90%) and was considered a useful diagnostic equation (Table 4).
Table 3.
Comparison of estimation performance of equations to estimate ASM
| Validation data set | R 2 a | RMSE | Median [IQR] b |
|---|---|---|---|
| Equation 1 | 0.91 | 1.70 (1.69–1.70) | −0.64 [1.87] |
| Equation 2 | 0.85 | 2.28 (2.27–2.29) | −1.37 [2.09] |
| Equation 3 | 0.81 | 2.40 (2.39–2.41) | 1.49 [2.24] |
| Equation 4 | 0.89 | 1.85 (1.84–1.86) | 0.07 [2.17] |
| Equation 5 | 0.85 | 2.08 (2.07–2.09) | −0.53 [2.46] |
| Equation 6 | 0.80 | 2.43 (2.42–2.44) | 0.71 [2.81] |
ASM, appendicular skeletal muscle mass; IQR, interquartile range; RMSE, root mean square error.
Percentage of variability in log ASM explained by the regression model.
The median and interquartile spacing of the distribution of absolute differences between actual and estimation ASM in the validation sample.
Table 4.
Comparison of diagnostic performance of equations to estimate ASM
| Estimated | Equation | Male | Female | ||||||
|---|---|---|---|---|---|---|---|---|---|
| AUC | Cutoff | Specificity | Sensitivity | AUC | Cutoff | Specificity | Sensitivity | ||
| EWGSOP2 | |||||||||
| SMI (kg/m2) | Equation 1 | 0.93 | 7.75 | 0.81 | 0.89 | 0.92 | 6.03 | 0.85 | 0.83 |
| Equation 2 | 0.92 | 8.11 | 0.79 | 0.90 | 0.91 | 6.33 | 0.83 | 0.85 | |
| Equation 4 | 0.92 | 7.34 | 0.87 | 0.81 | 0.90 | 6.04 | 0.76 | 0.89 | |
| EWGSOP1/IWGS | |||||||||
| SMI (kg/m2) | Equation 1 | 0.92 | 7.81 | 0.83 | 0.86 | 0.91 | 6.21 | 0.81 | 0.85 |
| Equation 2 | 0.92 | 8.11 | 0.83 | 0.86 | 0.90 | 6.52 | 0.78 | 0.86 | |
| Equation 4 | 0.92 | 7.62 | 0.80 | 0.87 | 0.90 | 6.07 | 0.77 | 0.86 | |
| AWGS | |||||||||
| SMI (kg/m2) | Equation 1 | 0.93 | 7.75 | 0.81 | 0.89 | 0.93 | 6.02 | 0.84 | 0.86 |
| Equation 2 | 0.92 | 8.11 | 0.79 | 0.90 | 0.92 | 6.29 | 0.83 | 0.86 | |
| Equation 4 | 0.92 | 7.34 | 0.87 | 0.81 | 0.92 | 5.96 | 0.78 | 0.87 | |
| FNIH | |||||||||
| ASM/BMI [kg/(kg/m2)] | Equation 1 | 0.95 | 0.87 | 0.86 | 0.90 | 0.93 | 0.57 | 0.82 | 0.90 |
| Equation 2 | 0.95 | 0.91 | 0.85 | 0.91 | 0.92 | 0.60 | 0.80 | 0.90 | |
| Equation 4 | 0.94 | 0.85 | 0.86 | 0.88 | 0.94 | 0.57 | 0.84 | 0.91 | |
ASM, appendicular skeletal muscle mass; AUC, area under the curve; AWGS, Asian Working Group for Sarcopenia; BMI, body mass index; EWGSOP, European Working Group on Sarcopenia Older Persons; FNIH, Foundation of the National Institute of Health; IWGS, International Working Group on Sarcopenia; SMI, skeletal muscle mass index.
The improvement in performance can also be shown by comparing the RMSE and the median and interquartile range of the differences between the estimated and actual ASM for each equation (Table 3). The smaller the RMSE and the median and interquartile range of the differences, the more accurate the estimated ASMs were in estimating the actual ASM. The results showed that Equation 1 and Equation 4 were the most accurate. It is worth noting that Equation 4 did not require the collection of serum biochemical data. The RMSE of equations 1, 2, and 4 were 1.70 (1.69–1.70), 2.28 (2.27–2.29), and 1.85 (1.84–1.86), respectively, with median (interquartile range) of the differences of −0.64 (1.87), −1.37 (2.09), and 0.07 (2.17), respectively. The Bland–Altman Plot (Figure S3) shows that estimated ASM did not systematically deviate from the actual ASM. When the ASM was small (<20 kg), the difference between the actual ASM and estimated ASM was almost always within the 95% agreement limits.
The AUCs for diagnosing low muscle mass for equations 1, 2, and 4 were 0.91–0.95 (male: 0.92–0.95; female: 0.91–0.93), 0.90–0.95 (male: 0.92–0.95; female: 0.90–0.92), and 0.90–0.94 (male: 0.92–0.94; female: 0.90–0.94), respectively. All formulas showed the highest diagnostic ability (AUC: 0.92–0.95) using the consensus of FNIH (Table 4).
The interval distribution of the estimated ASM calculated by equations 1, 2, and 4 and the actual ASM are shown in Table S8. The percentage of false‐positive participants when applying Equation 2 with the FNIH consensus as the diagnosis was 0.2% (n = 4) for males and 1.0% (n = 19) for females. The highest number of false‐positives was observed when applying Equation 4, with 70 (3.1%) males and 90 (4.5%) females, but this also had the lowest number of false‐negatives, with 172 (7.6%) for males and 140 (7.0%) for females (Figure 1). Similar conclusions were reached when other expert consensuses were used as diagnostic criteria (Figures S4 and S5).
Figure 1.
Estimated ASM/BMI diagnostic status (based on FNIH expert consensus). ASM, appendicular skeletal muscle mass; BMI, body mass index; FNIH, Foundation of the National Institute of Health.
Discussion
We developed new equations, the estimated ASM equations, to estimate ASM from demographic, physical measurement, and serum biochemical data by using the large database from NHANES. Using a randomized separate sample, we validated the estimated ASM equations and concluded that their variability in estimated ASM compared with actual ASM was low (R 2 = 0.91 for Equation 1 and 0.89 for Equation 4; Table 3). The results suggest that these equations can accurately estimate ASM in community populations and can be used as a set of assessment tools for sarcopenia, which has important implications for public health and clinical practice.
Sarcopenia has a high prevalence in community populations, and this public health problem is bound to become more pronounced as the global population ages. 21 The measurement of muscle mass as necessary for the diagnosis of sarcopenia has always been a clinical challenge. Even though imaging modalities such as MRI, CT, and DXA are thought to produce accurate results, these methods are limited in terms of cost, standardization, possible radiation exposure, and accessibility to the primary community. 22 Further, the BIA results seem to be inaccurate, making it difficult to get an accepted cut‐off value across various populations and manufacturers/models. 7 Therefore, the search for a simple and accurate assessment of muscle mass has become the current research direction. 23 , 24 The 2019 EWGSOP2 Expert Consensus 5 stated that ‘The development and validation of a single biomarker might be an easy and cost‐effective way to diagnose and monitor people with sarcopenia’, and called for scholarly research on muscle mass indicators and muscle mass assessment.
In response to this call for research, several recent studies have proposed serological indicators and simple tools for the diagnosis of sarcopenia. Kim et al. proposed a new equation for estimating muscle mass based on serum creatinine and cystatin C in a cohort of 107 individuals (R 2 = 85.6%). 10 Lien hypothesized a new indicator associated with sarcopenia, namely, serum creatinine × eGFR cystatin (estimated glomerular filtration rate [eGFR] based on serum cystatin level), based on previous studies. 9 Kashani et al. proposed creatinine/cystatin C as a simple method to identify critically ill patients with reduced muscle mass. 25 Shafiee et al. developed a model including sex, age, weight, and calf circumference to predict sarcopenia using a cohort of 1499 Iranian older adults. 8 However, in these studies, the selection of indicators was subjective and singular, and no studies have yet combined various types of factors affecting muscle mass (demographic, physical measurement, and biochemical) and fitted the best equations with objective statistical methods. Furthermore, the results of the above studies are difficult to generalize because of the small sample sizes, single ethnicities/races and age groups, and some of the tools have performed poorly in subsequent validation. 11 Therefore, comparing and analysing the current simple and available potential muscle mass replacement indicators in the same study and validating them would help to further advance this research direction.
Our study used the NHANES database of 6913 community participants as a development cohort, and included demographics (sex, ethnicity/race, age), physical measurements (weight), 26 and serum biochemical indicators (serum creatinine) 27 , 28 that were readily available and considered to be associated with muscle mass. At the same time, we included relevant indicators (height, serum cystatin C, 29 serum albumin, 30 urea nitrogen, 31 blood calcium, 32 blood phosphorus, 33 urine protein, 34 , 35 and urine creatinine 36 ) that related to the above indicators for analysis, with the aim of correcting for various possible effects. Finally, we obtained and verified a set of equations. The equations had ASM inversely proportional to age and cystatin C and positively proportional to weight, height, and creatinine, with coefficients <1 for females and >1 for black ancestry, which is consistent with previous studies. Both the Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology Collaboration (CKD‐EPI) equations for eGFR take into account the effects of age, sex, and ethnicity/race on muscle mass‐related creatinine and include these factors in the equations. 37 , 38 Current studies that estimate muscle mass have used cystatin C to offset the effect of renal function on creatinine. 25 , 39 Serum creatinine is a derivative of the skeletal muscle protein creatine phosphate and a routine serum marker used to estimate GFR. It is freely filtered in the glomerulus and secreted in small amounts in the proximal tubule. 40 At the same time, creatinine production varies with muscle mass. 41 Thus, serum creatinine concentration is influenced by both renal function and muscle mass. In contrast, cystatin C is a nonionic protein excreted by all nucleated cells. This protein can also be freely filtered in the glomerulus yet completely reabsorbed and metabolized by proximal tubular cells. Thus, cystatin C is also affected by renal function, but muscle mass has little effect on it. 42 Conceptually, cystatin C corrects for the effect of renal function on serum creatinine.
The estimated ASM equations included factors associated with muscle mass, such as age, sex, and ethnicity/race, that contributed to their accuracy. Equation 2 eliminated the ethnicity/race factor and it was less accurate than Equation 1 and Equation 4, but it had the highest sensitivity and contributed to the clinical confirmation of the diagnosis. Equation 4 did not include serum biochemical data (creatinine, cystatin C), which would simplify the procedure used in the community population. Considered together, Equation 1 (Table S7) seems to be the superior choice. Although the estimated ASM Equation 1 overall overestimated the actual ASM, the degree of overestimation appears to be acceptable. First, the median difference between actual ASM and estimated ASM was −0.64, which is small compared with the order of magnitude of actual ASM. In addition, ASM needs to be adjusted using height and BMI when diagnosing low muscle mass, which would further weaken the bias of the estimates. Third, when ASM was small enough to diagnose low muscle mass (male: <20 kg; female: <15 kg), the difference between actual and estimated ASM was almost always within 95% agreement limits. The estimated ASM equation could help clinicians determine whether large devices are needed to assess the loss of muscle mass and loss of muscle function. In addition, the cutoff value for diagnosing low muscle mass using muscle mass measured by devices such as DXA is controversial. Therefore, the estimated ASM equation even has promise to replace complex mechanical measurements.
The strengths of our study include its large dataset for the development and validation of new equations and the fact that all statistical analysis plans are strictly referenced to the development and validation methods of the CKD‐EPI equation 37 and the MDRD equation 38 for estimating GFR. This dataset has a large sample size, abundant and complete ethnic and age subgroups, and data collection specifications and strict quality control, which ensured the reliability and generalizability of the estimated ASM equations. The accuracy of our equation seems to be better than other assessment tools. In addition, the required demographic data and physical measurements are easily available and cystatin C measurements can be obtained from the same serum samples used to measure creatinine. However, there are limitations in this study. First, the validation sample was from the same overall population as the training sample. Therefore, the estimated ASM equation may perform differently in populations with different patient characteristic distributions than in the validation sample. However, this study has the largest sample size and the most complete set of ethnicities and ages of any study known to develop a muscle mass assessment model. In a follow‐up study, we will validate the estimated ASM equation in many more specific subgroups. Second, we used the ASM estimated by DXA as the actual ASM. DXA estimates of ASM are based only on the molecular level and not on the organ/tissue level. It seems to lack some accuracy compared with organ/tissue level‐based measurements of skeletal muscle mass such as MRI and CT. However, it is most used in the current assessment of muscle mass and is recommended by all current consensuses. Third, the estimated ASM equation is complex, but it can be readily implemented into a clinical laboratory information system through input variables.
In summary, we derived a new equation (Table 3, Equation 1) estimating ASM using demographic data (sex, ethnicity/race, age), physical measurements (weight, height), and serum biochemical indicators (creatinine, cystatin C) that are more accurate than other known assessment methods. We recommend routine use of the estimated ASM equation for estimating ASM and thus assessing low muscle mass. Increased awareness and early diagnosis of sarcopenia is of public health importance.
Acknowledgements
The authors thank all the participants and staff of the National Health and Nutrition Examination Survey and the National Center for Environmental Health for their valuable contributions. The authors of this manuscript certify that they comply with the ethical guidelines for authorship and publishing in the Journal of Cachexia, Sarcopenia and Muscle. 43 We thank International Science Editing (http://www.internationalscienceediting.com) for editing this manuscript. This work was supported by the Natural Science Foundation of Fujian Province (2022J011503).
Conflict of interest statement
All authors declare that they have no conflict of interest.
Supporting information
Figure S1 Flow Chart.
Abbreviations: NHANES, The National Health and Nutrition Examination Survey. * 4261 participants measured serum cystatin C value, including all participants aged 60 years or older with available specimen and a 25% random sample of participants aged 12–59 years.
Table S1 Summary of Operational Definition and Prevalence (Weighted) for Low Muscle Mass by Sex.
Table S2 Univariate Linear Regression a.
Table S3 Multicollinearity Analysis by Multiple Linear Regression (VIF a).
Table S4 Multiple Linear Regression a.
Table S5 Stepwise Regression of Log ASM.
Table S6 Estimation Equations from Development Data Set.
Table S7 Estimation Equation 1 from Full Data Set.
Table S8 Comparison of the Equations in Estimating ASM Stage ab.
Figure S2 The Relationship between the Actual ASM and the Estimated ASM.
Abbreviations: ASM, appendicular skeletal muscle mass;
Figure S3 Bland–Altman Plot.
Abbreviations: ASM, appendicular skeletal muscle mass;
Figure S4 Estimated ASM Diagnostic Status (based on EWGSOP2 expert consensus).
Abbreviations: EWGSOP, European Working Group on Sarcopenia Older Persons; ASM, appendicular skeletal muscle mass.
Figure S5 Estimated SMI Diagnostic Status (based on EWGSOP2 expert consensus) Abbreviations: EWGSOP, European Working Group on Sarcopenia Older Persons; SMI, skeletal muscle mass index.
Data S1. Supporting Information
Shi S, Chen W, Jiang Y, Chen K, Liao Y, Huang K. (2023) A more accurate method to estimate muscle mass: A new estimation equation, Journal of Cachexia, Sarcopenia and Muscle, 14, 1753–1761, 10.1002/jcsm.13254
Shanshan Shi and Weihua Chen have contributed equally to this work and share first authorship.
Contributor Information
Ying Liao, Email: wingjays@163.com.
Kun Huang, Email: k-huang18@mails.tsinghua.edu.cn.
References
- 1. Beaudart C, Zaaria M, Pasleau F, Reginster JY, Bruyère O. Health Outcomes of Sarcopenia: A Systematic Review and Meta‐Analysis. PLoS ONE 2017;12:e0169548. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Filippin LI, Teixeira VN, da Silva MP, Miraglia F, da Silva FS. Sarcopenia: a predictor of mortality and the need for early diagnosis and intervention. Aging Clin Exp Res 2015;27:249–254. [DOI] [PubMed] [Google Scholar]
- 3. Liu P, Hao Q, Hai S, Wang H, Cao L, Dong B. Sarcopenia as a predictor of all‐cause mortality among community‐dwelling older people: A systematic review and meta‐analysis. Maturitas 2017;103:16–22. [DOI] [PubMed] [Google Scholar]
- 4. Kelley GA, Kelley KS. Is sarcopenia associated with an increased risk of all‐cause mortality and functional disability? Exp Gerontol 2017;96:100–103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Cruz‐Jentoft AJ, Bahat G, Bauer J, Boirie Y, Bruyère O, Cederholm T, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing 2019;48:16–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Dupuy C, Lauwers‐Cances V, Guyonnet S, Gentil C, Abellan Van Kan G, Beauchet O, et al. Searching for a relevant definition of sarcopenia: results from the cross‐sectional EPIDOS study. J Cachexia Sarcopenia Muscle 2015;6:144–154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Walowski CO, Braun W, Maisch MJ, Jensen B, Peine S, Norman K, et al. Reference Values for Skeletal Muscle Mass ‐ Current Concepts and Methodological Considerations. Nutrients 2020;12:755. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Shafiee G, Ostovar A, Maleki Birjandi S, Nabipour I, Larijani B, Heshmat R. Development of a Simple and Practical Screening Tool for Detection of Sarcopenia in Older People: The Bushehr Elderly Health Program. Front Med 2021;8:655759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Lien YH. Looking for Sarcopenia Biomarkers. Am J Med 2017;130:502–503. [DOI] [PubMed] [Google Scholar]
- 10. Kim SW, Jung HW, Kim CH, Kim KI, Chin HJ, Lee H. A New Equation to Estimate Muscle Mass from Creatinine and Cystatin C. PLoS ONE 2016;11:e0148495. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Shi S, Jiang Y, Chen W, Chen K, Liao Y, Huang K. Diagnostic and prognostic value of the Creatinine/Cystatin C ratio for low muscle mass evaluation among US adults. Front Nutr 2022;9:897774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Finney H, Newman DJ, Gruber W, Merle P, Price CP. Initial evaluation of cystatin C measurement by particle‐enhanced immunonephelometry on the Behring nephelometer systems (BNA, BN II). Clin Chem 1997;43:1016–1022. [PubMed] [Google Scholar]
- 13. Cruz‐Jentoft AJ, Sayer AA. Sarcopenia. Lancet (London, England) 2019;393:2636–2646. [DOI] [PubMed] [Google Scholar]
- 14. Wang ZM, Pierson RN Jr, Heymsfield SB. The five‐level model: a new approach to organizing body‐composition research. Am J Clin Nutr 1992;56:19–28. [DOI] [PubMed] [Google Scholar]
- 15. Cruz‐Jentoft AJ, Bahat G, Bauer J, Boirie Y, Bruyère O, Cederholm T, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing 2019;48:601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. McLean RR, Shardell MD, Alley DE, Cawthon PM, Fragala MS, Harris TB, et al. Criteria for clinically relevant weakness and low lean mass and their longitudinal association with incident mobility impairment and mortality: the foundation for the National Institutes of Health (FNIH) sarcopenia project. J Gerontol A Biol Sci Med Sci 2014;69:576–583. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Cruz‐Jentoft AJ, Baeyens JP, Bauer JM, Boirie Y, Cederholm T, Landi F, et al. Sarcopenia: European consensus on definition and diagnosis: Report of the European Working Group on Sarcopenia in Older People. Age Ageing 2010;39:412–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Chen LK, Woo J, Assantachai P, Auyeung TW, Chou MY, Iijima K, et al. Asian Working Group for Sarcopenia: 2019 Consensus Update on Sarcopenia Diagnosis and Treatment. J Am Med Dir Assoc 2020;21:300–307.e302. [DOI] [PubMed] [Google Scholar]
- 19. Fielding RA, Vellas B, Evans WJ, Bhasin S, Morley JE, Newman AB, et al. Sarcopenia: an undiagnosed condition in older adults. Current consensus definition: prevalence, etiology, and consequences. International working group on sarcopenia. J Am Med Dir Assoc 2011;12:249–256. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Johnson CL, Paulose‐Ram R, Ogden CL, Carroll MD, Kruszon‐Moran D, Dohrmann SM, et al. National health and nutrition examination survey: analytic guidelines, 1999‐2010. Vital and health statistics Series 2. Data Eval Methods Res 2013:1–24. [PubMed] [Google Scholar]
- 21. Janssen I, Heymsfield SB, Ross R. Low relative skeletal muscle mass (sarcopenia) in older persons is associated with functional impairment and physical disability. J Am Geriatr Soc 2002;50:889–896. [DOI] [PubMed] [Google Scholar]
- 22. Tosato M, Marzetti E, Cesari M, Savera G, Miller RR, Bernabei R, et al. Measurement of muscle mass in sarcopenia: from imaging to biochemical markers. Aging Clin Exp Res 2017;29:19–27. [DOI] [PubMed] [Google Scholar]
- 23. Calvani R, Marini F, Cesari M, Tosato M, Anker SD, von Haehling S, et al. Biomarkers for physical frailty and sarcopenia: state of the science and future developments. J Cachexia Sarcopenia Muscle 2015;6:278–286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Marzetti E. Editorial: Imaging, Functional and Biological Markers for Sarcopenia: The Pursuit of the Golden Ratio. J Frailty Aging 2012;1:97–98. [DOI] [PubMed] [Google Scholar]
- 25. Kashani KB, Frazee EN, Kukrálová L, Sarvottam K, Herasevich V, Young PM, et al. Evaluating Muscle Mass by Using Markers of Kidney Function: Development of the Sarcopenia Index. Crit Care Med 2017;45:e23–e29. [DOI] [PubMed] [Google Scholar]
- 26. Kim KM, Jang HC, Lim S. Differences among skeletal muscle mass indices derived from height‐, weight‐, and body mass index‐adjusted models in assessing sarcopenia. Korean J Intern Med 2016;31:643–650. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Evans WJ, Hellerstein M, Orwoll E, Cummings S, Cawthon PM. D(3) ‐Creatine dilution and the importance of accuracy in the assessment of skeletal muscle mass. J Cachexia Sarcopenia Muscle 2019;10:14–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Keshaviah PR, Nolph KD, Moore HL, Prowant B, Emerson PF, Meyer M, et al. Lean body mass estimation by creatinine kinetics. J Am Soc Nephrol 1994;4:1475–1485. [DOI] [PubMed] [Google Scholar]
- 29. Muntner P, Winston J, Uribarri J, Mann D, Fox CS. Overweight, obesity, and elevated serum cystatin C levels in adults in the United States. Am J Med 2008;121:341–348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Chen Z, Song C, Yao Z, Sun J, Liu W. Associations between albumin, globulin, albumin to globulin ratio and muscle mass in adults: results from the national health and nutrition examination survey 2011‐2014. BMC Geriatr 2022;22:383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Streng KW, Hillege HL, Ter Maaten JM, van Veldhuisen DJ, Dickstein K, Ng LL, et al. Clinical implications of low estimated protein intake in patients with heart failure. J Cachexia Sarcopenia Muscle 2022;13:1762–1770. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Ceran O, Akin M, Aktürk Z, Ozkozaci T. Normal urinary calcium/creatinine ratios in Turkish children. Indian Pediatr 2003;40:884–887. [PubMed] [Google Scholar]
- 33. El Mallah C, Ghattas H, Shatila D, Francis S, Merhi K, Hlais S, et al. Urinary Magnesium, Calcium, and Phosphorus to Creatinine Ratios of Healthy Elementary School Lebanese Children. Biol Trace Elem Res 2016;170:264–270. [DOI] [PubMed] [Google Scholar]
- 34. Watanabe M, Funabiki K, Tsuge T, Maeda K, Horikoshi S, Tomino Y. Using protein/creatinine ratios in random urine. J Clin Lab Anal 2005;19:160–166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Newman DJ, Pugia MJ, Lott JA, Wallace JF, Hiar AM. Urinary protein and albumin excretion corrected by creatinine and specific gravity. Clin Chim Acta 2000;294:139–155. [DOI] [PubMed] [Google Scholar]
- 36. Heymsfield SB, Arteaga C, McManus C, Smith J, Moffitt S. Measurement of muscle mass in humans: validity of the 24‐hour urinary creatinine method. Am J Clin Nutr 1983;37:478–494. [DOI] [PubMed] [Google Scholar]
- 37. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:604–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999;130:461–470. [DOI] [PubMed] [Google Scholar]
- 39. Stevens LA, Coresh J, Schmid CH, Feldman HI, Froissart M, Kusek J, et al. Estimating GFR using serum cystatin C alone and in combination with serum creatinine: a pooled analysis of 3,418 individuals with CKD. Am J Kidney Dis 2008;51:395–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Levey AS, Perrone RD, Madias NE. Serum creatinine and renal function. Annu Rev Med 1988;39:465–490. [DOI] [PubMed] [Google Scholar]
- 41. Caregaro L, Menon F, Angeli P, Amodio P, Merkel C, Bortoluzzi A, et al. Limitations of serum creatinine level and creatinine clearance as filtration markers in cirrhosis. Arch Intern Med 1994;154:201–205. [PubMed] [Google Scholar]
- 42. Randers E, Erlandsen EJ. Serum cystatin C as an endogenous marker of the renal function‐‐a review. Clin Chem Lab Med 1999;37:389–395. [DOI] [PubMed] [Google Scholar]
- 43. von Haehling S, Morley JE, Coats AJS, Anker SD. Ethical guidelines for publishing in the journal of cachexia, sarcopenia and muscle: update 2021. J Cachexia Sarcopenia Muscle 2021;12:2259–2261. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1 Flow Chart.
Abbreviations: NHANES, The National Health and Nutrition Examination Survey. * 4261 participants measured serum cystatin C value, including all participants aged 60 years or older with available specimen and a 25% random sample of participants aged 12–59 years.
Table S1 Summary of Operational Definition and Prevalence (Weighted) for Low Muscle Mass by Sex.
Table S2 Univariate Linear Regression a.
Table S3 Multicollinearity Analysis by Multiple Linear Regression (VIF a).
Table S4 Multiple Linear Regression a.
Table S5 Stepwise Regression of Log ASM.
Table S6 Estimation Equations from Development Data Set.
Table S7 Estimation Equation 1 from Full Data Set.
Table S8 Comparison of the Equations in Estimating ASM Stage ab.
Figure S2 The Relationship between the Actual ASM and the Estimated ASM.
Abbreviations: ASM, appendicular skeletal muscle mass;
Figure S3 Bland–Altman Plot.
Abbreviations: ASM, appendicular skeletal muscle mass;
Figure S4 Estimated ASM Diagnostic Status (based on EWGSOP2 expert consensus).
Abbreviations: EWGSOP, European Working Group on Sarcopenia Older Persons; ASM, appendicular skeletal muscle mass.
Figure S5 Estimated SMI Diagnostic Status (based on EWGSOP2 expert consensus) Abbreviations: EWGSOP, European Working Group on Sarcopenia Older Persons; SMI, skeletal muscle mass index.
Data S1. Supporting Information