If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of General Medicine and Community Health Science, Sasayama Medical Center Hyogo College of Medicine, Sasayama, Hyogo, JapanDepartment of Rehabilitation Medicine, Sasayama Medical Center Hyogo College of Medicine, Sasayama, Hyogo, Japan
Department of Dentistry and Oral Surgery, Hyogo College of Medicine, Nishinomiya, Hyogo, JapanDivision of Comprehensive Prosthodontics, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Niigata, Japan
Division of General Medicine, Department of Internal Medicine, Hyogo College of Medicine, Nishinomiya, Hyogo, JapanDepartment of General Medicine and Community Health Science, Sasayama Medical Center Hyogo College of Medicine, Sasayama, Hyogo, Japan
Sarcopenia is defined as a combination of low skeletal muscle mass index (SMI), weak muscle strength, and reduced physical function. Recently, many studies have reported that the creatinine/cystatin C ratio (Cr/CysC) is useful for evaluating muscle mass. We designed a cross-sectional study with separate model development and validation groups to develop a prediction equation to estimate bioimpedance analysis (BIA)-measured SMI with Cr/CysC.
The current study was a retrospective cross-sectional study.
Setting and Participants
The model development group included 908 subjects (288 men and 620 women) from the Frail Elderly in the Sasayama-Tamba Area (FESTA) study, and the validation group included 263 subjects (112 men and 151 women) from participants in the medical checkup program at the Anti-Aging Center in Ehime Prefecture.
Multivariate regression analysis indicated that age, hemoglobin (Hb), body weight (BW), and Cr/CysC were independently associated with SMI in both men and women. The SMI prediction equation was developed as follows:
The SMI prediction equation was applied to the validation group and strong correlations were observed between the BIA-measured and predicted SMI (pSMI) in men and women. According to the receiver operator characteristic (ROC) analysis, the areas under the curve were 0.93 (specificity 89.0%, sensitivity 87.2%) among men and 0.88 (specificity 83.6%, sensitivity 79.6%) among women for using pSMI to identify low SMI in the model development group. The pSMI also indicated high accuracy in ROC analysis for low SMI in the validation group. The Bland-Altman plot regression showed good agreement between BIA-measured and pSMI.
Conclusions and Implications
Our new prediction equation to estimate SMI is easy to calculate in daily clinical practice and would be useful for diagnosing sarcopenia.
Sarcopenia refers to the age-related loss of muscle mass and power. Sarcopenia is defined as a combination of low skeletal muscle mass, weak muscle strength, and reduced physical function. Loss of muscle mass is essential for the diagnosis of sarcopenia. The skeletal muscle index (SMI) was calculated as skeletal muscle mass (SMM)/height2. In and the Asian Working Group for Sarcopenia (AWGS 2019) criteria, dual X-ray absorptiometry (DXA), and bioimpedance analysis (BIA) were recommended for the evaluation of skeletal muscle mass and SMI calculation. However, these methods require specific devices and are difficult to use in the daily clinical setting.
Recently, many studies have reported that the creatinine/cystatin C ratio (Cr/CysC) is useful for evaluating muscle mass.
However, in our previous study, the correlation coefficients between Cr/CysC and muscle mass and strength parameters were quite low. For example, the correlation coefficients (r) between Cr/CysC and SMI were r = 0.34 (P < .0001) in men and r = 0.08 (P = .0767) in women,
reported that Cr/CysC could not be a biomarker of sarcopenia in Chinese urban community-dwelling older people because of its low area under the curve (AUC) in receiver operator characteristic (ROC) analysis. Abe et al.
found that calf circumference and mid-arm circumference were more closely associated with SMI than with Cr/CysC. Conversely, SMI is dependent on body weight (BW), and correlation coefficients between BW and SMI are quite high.
We hypothesized that SMI prediction models can be generated using Cr/CysC, BW, and other parameters. This study aimed to develop prediction models for sarcopenia in older Japanese individuals.
Model Development Group
The model development group included 908 subjects (288 men and 620 women) recruited from the Frail Elderly in the Sasayama-Tamba Area (FESTA) study. The study population was composed of individuals aged ≥65 years. Healthy community-dwelling older individuals from the Sasayama-Tamba area, a rural area in Hyogo Prefecture, Japan, were recruited between 2015 and 2019. We recruited study participants by posting advertisements in local newspapers and by placing posters in the Hyogo College of Medicine Sasayama Medical Center.
In the muscles of patients with chronic renal failure, the synthesis of muscle proteins was suppressed, and the degradation of muscle protein was increased. To reduce the influence of these catabolized changes in the muscle, we excluded subjects with low renal function (creatinine estimated glomerular filtration rate: eGFR <45 mL/min per 1.73 m2) in the model development group. Physical function assessment, measurement of body composition, and blood sample analysis were performed as described previously.
Body composition was evaluated by BIA using an InBody 770 device (InBody Japan Inc., Tokyo, Japan). All procedures performed involving human participants were in accordance with the ethical standards of our institution and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
The concentration of Cr in serum was measured using an enzymatic method, and that of CysC was measured using a latex agglutination-turbidimetric immunoassay (IATRO Cys-C; LSI Medience Co. Ltd., Tokyo, Japan) using blood specimens stored 82 at −80 °C. Fasting blood specimens were collected on the morning of the medical checkup.
The validation group included 249 participants (105 men and 144 women) from an anti-aging cohort in Japan. This cohort consisted of middle-aged to very old individuals who participated in the medical checkup program at our university hospital Anti-Aging Center. This medical checkup was provided to community residents of Ehime Prefecture to evaluate age-related disorders, such as atherosclerosis, cardiovascular diseases, physical dysfunction, and cognitive impairment. Among 2163 individuals who participated from February 2006 to December 2020, SMI values measured using a bioimpedance device (MC 780A-N; Tanita Co., Tokyo, Japan) were available for 356 individuals who participated in the checkup program after March 2016. After excluding individuals younger than 60 years with severe renal functional decline (eGFR <45 mL/min per 1.73 m2), a total of 249 participants were included in the analysis.
The basic clinical data used in this study were obtained from health records completed as part of the medical checkup program. The concentration of Cr in serum was measured using an enzymatic method and CysC was measured by latex agglutination-turbidimetric immunoassay (IATRO Cys-C; LSI Medience Co. Ltd., Tokyo, Japan) using blood specimens stored at −80 °C. Fasting blood specimens were drawn the morning of the medical checkup. All study procedures were approved by the Ethics Committee of our institution. Written informed consent was obtained from all participants.
Diagnosis of Sarcopenia
Sarcopenia was defined according to the criteria for the AWGS 2019.
Sarcopenia was considered if the participants had low SMI (<7.0 kg/m2 in men; <5.7 kg/m2 in women) and weak handgrip strength (<28 kg in men; <18 kg in women) or low physical performance [normal gait speed <1.0 m/s, 5-time chair stand test (5CS) ≥12 s, or short physical performance battery (SPPB) ≤9] according to the AWGS 2019 criteria.
Low SMI was defined as SMI <7.0 kg/m2 in men and <5.7 kg/m2 in women.
The results are expressed as the mean ± SD or percentages. Pearson's product-moment correlation coefficient was used to assess the associations between BIA-measured and predicted SMI. Multivariate regression analysis was performed for SMI, Cr/CysC, and other individual parameters.
A ROC analysis was performed to confirm the diagnostic efficacy of the BIA-measured and -predicted SMI, and the AUC was calculated. Bland-Altman plots were created for the model development group using BIA-measured and predicted SMI. JMP 13.1 software was used for data analysis. Statistical significance was set at P < .05.
The baseline characteristics, indices of body composition, and physical performance of the model development group participants are presented in Table 1. The model development group consisted of 288 men and 620 women. Among the 908 participants, 71 (23 men and 48 women) had sarcopenia based on the AWGS 2019 criteria. A total of 289 participants (78 men and 211 women) had low SMI.
Table 1Subject Characteristics and Variables Used in This Study for Model Development
As per our previous study, in the simple correlation analysis of the model development group, Cr/CysC, age, height, weight, and hemoglobin (Hb) were positively correlated with BIA-measured SMI in men in the model development group [Table 2 (men), model 1].
Table 2Relationship of BIA-measured SMI to Variables Using Univariate and Multiple Regression Analyses and Individual Parameters For Men and Women
Model 1: Simple correlations between BIA-measured SMI and age, height, Cr/CysC, and Hb levels.
Model 2: Multivariate regression analysis of BIA-measured SMI and age, height, Cr/CysC, and Hb levels.
Model 3: Multivariate regression analysis of BIA-measured SMI and age, Cr/CysC, and Hb levels.
Data normality was assessed using the Shapiro-Wilk test, and normal distribution was confirmed (Supplementary Table 1). We drew a scatter plot and regression lines between the 4 parameters (Cr/CysC, age, weight, and Hb) and SMI (Supplementary Figure 1). Because these relationships seemed to be linear in both sexes, we adopted a linear regression model in the development of the equation.
To check the multicollinearity and interactions between these independent factors, we showed the correlations among these 4 variables. Although there were mild significant correlations between Cr/CysC and age, BW and age, Hb, and age and Hb, there were no moderate to strong correlations between them (r < 0.4).
Therefore, we judged that there were no problems in choosing these parameters to develop the prediction equation (Supplementary Table 2).
Multiple regression analyses of SMI and Cr/CysC, age, height, weight, and Hb were performed. Cr/CysC, age, weight, and Hb were significant and common positive correlating factors for SMI. According to the partial regression coefficients of the independent factors, the following regression equation was considered as the predicted SMI (pSMI):
where Cr is creatinine (mg/dL), CysC is Cystatin C (mg/L), Hb is hemoglobin (g/dL), and BW is body weight (kg).
The baseline characteristics, indices of body composition, and physical performance of the validation group participants are presented in Table 3. The validation group consisted of 249 participants (105 men and 144 women). A total of 49 participants (17 men and 32 women) had low SMI (Table 3). There was no significant difference between the 2 groups except Cr/CysC in the validation group, which was slightly higher than that of the model development group. Similarly, the SMI in the validation group was slightly higher than that in the model development group. The percentage of patients with low SMI was higher in the model development group than in the validation group.
Table 3Subject Characteristics and Variables Used in This Study for Validation
Figure 1 shows the correlation between BIA-measured and predicted SMI in the model development group in men (A) and women (B) and in the validation group in men (C) and women (D). SMI was positively correlated with pSMI in both the model development and validation groups, regardless of sex (P < .001). We have drawn the scatter plot for pSMI or Cr/CysC ratio with handgrip strength in the model developing group. Both pSMI and Cr/CysC were positively correlated with handgrip strength (Supplementary Figure 2).
According to the ROC analysis, the AUCs were 0.88 (specificity 77.7%, sensitivity 82.6%) among men and 0.84 (specificity 69.2%, sensitivity 87.5%) among women using pSMI to identify AWGS 2019 sarcopenia in the model development group (Figure 2A–D). The AUCs were found to be 0.93 (specificity 89.0%, sensitivity 87.2%) among men and 0.88 (specificity 83.6%, sensitivity 79.6%) among women using pSMI to identify low SMI in the model development group (Figure 2E–H). The AUCs were 0.94 (specificity 75.3%, sensitivity 100.0%) among men and 0.83 (specificity 71.1%, sensitivity 90.0%) among women using pSMI to identify low SMI in the validation group (Figure 2I–L).
The AUCs of pSMI were higher than those of Cr/CysC in both men and women to identify AWGS 2019 sarcopenia in the model development group. The AUCs of pSMI were higher than those of Cr/CysC in both men and women for identifying low SMI both in the derivation and validation samples. The pSMI also indicated high accuracy in ROC analysis for low SMI in the validation group.
A Bland-Altman graph was also plotted as the (pSMI−BIA-measured SMI) versus the (pSMI+BIA-measured SMI)/2 in the model development group (Supplementary Figure 3A and B) The correlation between (pSMI−BIA-measured SMI) and (pSMI+BIA-measured SMI)/2 was not significant both in men and women. The Bland-Altman plot regression showed good agreement between the BIA-measured and pSMI.
In this study, we showed that the SMI prediction equation (pSMI) using Cr/CysC was useful for estimating BIA-measured SMI and identifying low SMI and AWGS 2019 sarcopenia in the model development group. pSMI also indicated high accuracy in ROC analysis for low SMI in the separate validation group. Our prediction equations for BIA-measured SMI were derived and externally validated using separate populations.
SMI is dependent on BW, and correlation coefficients between BW and SMI are quite high.
In this study, BW may have the largest impact on our prediction equation for SMI. A previous study showed that a combination of Cr/CysC and BW is useful for the screening of muscle mass. It has been reported that Cr/CysC × BW was well correlated with weight-adjusted SMI and indicated a high accuracy in ROC analysis for low muscle volume.
In this study, the number of participants with low muscle mass was very low [25 men (14.8%) and 4 women (3.1%)]. Our study showed that a combination of Cr/CysC and BW is useful for screening for muscle mass in a larger number of patients with low SMI and sarcopenia based on the AWGS 2019 criteria.
defined sarcopenic index based on serum adiponectin and sialic acid concentration. The sarcopenic index indicated a high accuracy in ROC analysis (AUC 0.892, specificity 69.9%, and sensitivity 94.9%) for sarcopenia in patients with cardiovascular diseases. This sarcopenic index may be useful for diagnosing sarcopenia; however, serum adiponectin and sialic acid levels are not commonly measured in daily clinical practice. Cr and CysC concentrations are commonly tested in clinical practice and routine health check-ups.
developed prediction models based on anthropometric parameters such as skinfold-corrected upper arm, thigh, and calf girth. Magnetic resonance imaging (MRI)-measured SM was used as the reference in this study, and their prediction models were useful for estimating total skeletal muscle mass. Al-Gindan et al.
also developed prediction equations based on anthropometric parameters, such as waist circumference and hip circumference. Whole-body MRI-measured SM was also used as a reference in this study. Their prediction equations for whole-body muscle mass were derived and externally validated using separate populations.
According to these previous studies, simple anthropometric parameters are useful for evaluating whole-body muscle mass; however, it is not practical to perform physical examinations on every patient in daily clinical practice. In contrast, blood sampling can be performed anywhere, and many samples can be handled simultaneously.
Inspection of imaging such as MRI, computed tomography (CT), and DXA are often used for evaluating whole-body muscle mass. Our novel equation would measure body muscle mass with less cost than MRI, CT, and DXA. Moreover, unlike MRI, CT, and DXA, there is no radiation hazard.
In the ROC analysis for low SMI, the AUC of Cr/CycC and pSMI was higher in men than in women in the model development and validation groups. In addition, in the ROC analysis for AWGS 2019 sarcopenia, the AUC of Cr/CycC and pSMI was higher in men than in women in the model development and validation groups.
the AUC of Cr/CysC identifying sarcopenia was higher in men than in women. Generally, total muscle volume is higher in men than in women. The influence of the change in muscle volume was less in CysC than in Cr. Therefore, the change in Cr/CycC due to the decrease in skeletal muscle mass is expected to be larger in men than in women.
In the ROC analysis for low SMI, the cutoff value of pSMI was higher in the validation group than in the model development group. In the model development group, body composition was evaluated by BIA using an InBody 770 device (InBody Japan Inc.), whereas in the validation group, body composition was evaluated by BIA using an MC-780A-N device (Tanita Co.). The difference in measuring equipment might cause this discrepancy in the cutoff value between the model development group and the validation group. Conversely, SMI was higher in the validation group than in the model development group. This difference in muscle volume may influence the discrepancy in the cutoff value between the 2 groups.
Artificial intelligence and machine or deep learning approaches are poised to influence every aspect of the human condition.
Machine learning focuses on algorithmically representing data structures and making predictions or classifications. In particular, deep learning systems have demonstrated excellent success in CT scan analysis. The deep learning system exhibited high performance and accuracy in analyzing abdominal muscle on CT images, as well as in evaluating muscle mass of the whole body.
However, machine or deep learning approaches are difficult to use in extrapolating to different groups and in evaluating the effectiveness of prediction models. Machine or deep learning makes predictions based on input data, and the outcomes may differ according to the characteristics of the input data. Simple equations can be easily calculated and applied to preventive medicine or daily clinical settings. For example, eGFR calculated using creatinine or cystatin C and age is widely used in daily clinical settings, although it requires the assistance of a computer program.
This study has some limitations that must be considered. First, this was a cross-sectional study. Therefore, any cause-and-effect relationship could not be evaluated. The sex distribution of each group was one of the limitations of this study. Ideally, the sex distribution should be similar between the model development and validation groups. Therefore, we performed a multivariate regression analysis and developed a prediction equation separated by sex. A prospective study must be conducted to assess the causal associations between chronic kidney disease and sarcopenia. Second, most of the participants voluntarily participated in the study. Thus, the study participants may have been healthier, and the study population might have had lower rates of sarcopenia than those in the general population. This could account for the inconsistency between our results and those of previous studies. Third, urinary protein levels were not measured. Thus, the association between chronic kidney disease and sarcopenia, which was modified by the presence of subclinical kidney disease, was not examined.
Bioelectrical impedance analysis (BIA) is a very useful method for evaluating muscle volume in clinical practice because it is an affordable and noninvasive test with little training required for operation. This is accepted by the guidelines in Asia and Europe. However, the BIA device cannot precisely measure SMI. It only estimates lean body mass instead of skeletal muscle mass. BIA-based estimates of muscle mass can be influenced by medical conditions, comorbidities, hydration, exercise history, and food intake.
Finally, a small number of participants with sarcopenia were included in the study, which obviously limits the reliability and applicability of the proposed test. Finally, we could not perform a physical examination, such as gait speed, 5CS, and SPPB, for diagnosing sarcopenia based on the AWGS 2019 criteria in the validation group. We could not perform ROC analysis for AWGS 2019 sarcopenia.
Conclusions and Implications
Our new prediction equation using Cr/CysC to estimate SMI is easy to calculate in daily clinical practice, with less cost compared with inspection of imaging, such as MRI, CT, and DXA. In conclusion, our prediction equation to estimate SMI would be useful for evaluating whole-body muscle mass and diagnosing sarcopenia in Japanese community-dwelling older individuals without severe renal function. Further studies with larger populations are needed to validate their utility in other ethnic populations.
We thank all the medical staff of Sasayama Medical Center Hyogo College of Medicine who supported the FESTA study.
Supplementary Table 1Shapiro-Wilk Test for Variables Associated With SMI
This study was supported in part by JSPS KAKENHI (grant number: 16KT0012 (2016–2018) and 19K16995 (2019–2022)), the Medical Research Fund by Hyogo Medical Association (2016), a grant for the support of collaborative investigation between Hyogo College of Medicine and Hyogo University of Health Sciences (2017– 2019) (Dr. Shinmura and Dr. Kusunoki), and the grant for Good Practice to Establish Centers for Fostering Medical Researchers of the Future by Ministry of Education, Culture, Sports, Science and Technology (2015–2017) (Hyogo College of Medicine).The authors declare no conflicts of interest.