Development and validation of a predictive nomogram for sarcopenia among older people in China

To the Editor: Sarcopenia is defined as age-related loss of skeletal muscle mass and strength, with an incidence of 5% to 13% in persons aged >60 years.[1] Sarcopenia is considered a major contributor of frailty in the elderly, and frailty is reported to be associated with poor outcomes of older people. Sarcopenia can also increase the risk of falls and fractures, leading to lower quality of life, other related diseases, and even death. Thus, early screening of sarcopenia is recommended for older people. The universally accepted diagnostic criteria of sarcopenia include the measurement of muscle mass, muscle strength, and muscle function.[2] Among the muscle mass measurement methods, magnetic resonance imaging, computed tomography, and dual-energy X-ray absorptiometry were difficult to popularize because of the high cost or radiation exposure. Bioelectrical impedance analysis (BIA) is a non-invasive, portable, and economical method; however, its results can be affected by various factors such as hydration status, food intake, and fat distribution. The above deficiencies limit its use in the screening of sarcopenia; thus, a multi-parameter screening tool based on easy-to-get and reliable clinical factors is urgently needed. In this study, we collected easily obtainable data from people living in Zhejiang in eastern China. By comparing the characteristics of participants with sarcopenia and healthy participants, disease-related factors were identified and included in a multifaceted prediction model. The multifaceted prediction model aims to improve the detection rate of individuals at a high risk of sarcopenia to achieve early detection and thus curb disease progression. This study was approved by the Ethical Committee of the First Affiliated Hospital, School of Medicine, Zhejiang University (No. 20191276). All participants provided written informed consent. A total of 835 residents aged ≥60 years were recruited in Shaoxing, Zhejiang Province. People with malignancies, chronic liver or renal disease, gout, or amputation were excluded from this research. Besides, people with important information deficits were also excluded. Sarcopenia was diagnosed according to diagnostic criteria published by the Asian Working Group for Sarcopenia (AWGS) in 2019 [Supplementary File, https://links.lww.com/CM9/B323]. Demographic data such as age and gender were collected. The grip strength and 6 m walking speed were measured. The appendicular skeletal muscle mass (ASM) was evaluated using BIA. The appendicular skeletal muscle mass index was derived from ASM (kg) divided by height (m) squared. In order to make the sarcopenia screening more practical, we collected the routine healthy checkup items in China, including liver function (alanine aminotransferase [ALT], aspartate aminotransferase, and cholinesterase); renal function (serum urea nitrogen, and creatinine); proteometabolism (total serum protein, albumin, and prealbumin); purine metabolism (uric acid [UA]); glycometabolism (glycated hemoglobin, fasting blood glucose, and fasting insulin), and lipid metabolism (high-density lipoprotein, low-density lipoprotein, triglyceride, and cholesterol). All statistical analyses were performed using IBM SPSS Statistics for Windows (version 25.0, IBM Corporation, Chicago, IL, USA) and R software (version 4.0.2, R Foundation for Statistical Computing, Vienna, Austria). Continuous data were evaluated using the Student's t-test (normal distribution) or the Mann-Whitney U test (non-normal distribution), and categorical variables were evaluated using the chi-squared test or Fisher's exact test. To identify independent diagnostic factors, we performed multivariate logistic regression analyses with a forward stepwise method. Of all included study subjects, 80% of participants were randomly selected as the training dataset, and the remaining 20% constituted the validation dataset. A nomogram was constructed based on the final selected prognostic factors to predict the probability of getting sarcopenia using the regplot package of R (version 4.0.2). The discrimination capacity of the model was accessed by calculating the area under the receiver operating characteristic (ROC) curve (AUC). The calibration capacity, representing the difference between the predicted value of the model and the actual value, was evaluated by the Hosmer–Lemeshow test. Between May 1, 2019 and September 1, 2019, a total of 835 participants were recruited for the study. After screening, 633 participants were eventually included in the study, including 125 patients with sarcopenia and 508 individuals without sarcopenia. Of all included study subjects, 507 (80%) were randomly selected as the training dataset, and the remaining 126 (20%) constituted the validation dataset [Supplementary Figure 1, https://links.lww.com/CM9/B323]. In the training dataset, the sarcopenia group was significantly older than the non-sarcopenia group (71.0 [66.0–75.0] years vs. 69.0 [64.0–72.0] years, Z = 3.7, P < 0.001). The non-sarcopenia group had a significantly higher body mass index (BMI) than the sarcopenia group (24.3 ± 2.5 kg/m2vs. 19.7 ± 1.5 kg/m2, t = 17.2, P < 0.001). The sarcopenia group had significantly lower appendicular skeletal muscle mass index (AMSI), 6 m walking speed, and grip strength compared with non-sarcopenia group [Supplementary Table 1, https://links.lww.com/CM9/B323]. Regarding laboratory results, the mean or median values of all parameters were within the normal ranges, while there were quantitative differences in several parameters between the two groups in the training dataset. The sarcopenia group had lower hemoglobin levels (135.0 [125.5–143.0] g/L vs. 142.0 [132.0–153.0] g/L, Z = 4.9, P < 0.001) than the non-sarcopenia group. As regards glycometabolism, the sarcopenia group had significantly lower glycated hemoglobin (5.5 [5.2–5.7]% vs. 5.6 [5.3–5.9]%, Z = 2.7, P = 0.006), and fasting insulin (3.1 [2.2–4.5] mmol/L vs. 5.0 [3.1–8.3] mmol/L, Z = 6.2, P < 0.001) levels than the non-sarcopenia group. Regarding liver function, the sarcopenia group had a significantly lower ALT level (14.9 [11.0–20.3] U/L vs. 18.7 [14.9–24.1] U/L, Z = 4.9, P < 0.001) than the non-sarcopenia group. For purine metabolism, the sarcopenia group had a significantly lower UA level (289.0 [241.5–341.5] μmol/L vs. 350.0 [304.0–401.0] μmol/L, Z = 7.5, P < 0.001). Other indicators related to metabolism, including cholinesterase, triglyceride, and prealbumin, were all lower in the sarcopenia group compared with non-sarcopenia group [Supplementary Table 1, https://links.lww.com/CM9/B323]. Based on the primary analyses, the parameters with P < 0.05 were included in the multivariate logistic regression analysis, and we found that BMI, age, UA, ALT, and gender were identified as independent diagnostic factors for sarcopenia [Supplementary Table 2, https://links.lww.com/CM9/B323]. Moreover, these factors were used to formulate a nomogram by R software [Figure 1A]. The discrimination of the model was valued using the ROC analysis, and the AUC was 0.974 in the training dataset (95% confidence interval [CI]: 0.962–0.987; Figure 1B) and 0.968 in the validation dataset (95% CI: 0.941–0.994; Figure 1C), which suggested that the predictive model had great discriminability. The calibration curve of the model demonstrated a good agreement between prediction and observation. The Hosmer–Lemeshow test suggested that there was no significant deviation from perfect fit both in the training dataset (P = 0.813; Figure 1D) and the validation dataset (P = 0.947; Figure 1E).Figure 1: Development and validation of the diagnostic model of sarcopenia. (A) Nomogram; AUC under the ROC for the training dataset (B) and the validating dataset (C); Calibration curve for the training dataset (D) and the validating dataset (E). The orange dotted line represents a perfect prediction by an ideal model. The blue dotted line represents the performance of the model, of which a closer fit to the orange line indicates a better prediction. ALT: Alanine aminotransferase; AUC: Area under the ROC curve; BMI: Body mass index; CI: Confidence interval; ROC: Receiver operating characteristic. UA: Uric acid.Recently, studies on the risk factors of sarcopenia based on data from Asians remained limited. People from different regions exhibited great differences in body size, eating habits, lifestyle, etc. Moreover, the threshold value of sarcopenia definition developed by the European Working Group on Sarcopenia in Older People was different from that of AWGS.[2] Therefore, the research results from European countries may not be applied to Asian countries. Thus, we developed a novel screening tool based on data from Zhejiang Province, China. Muscle mass and strength gradually decreased with age, and the prevalence of sarcopenia was increased with aging.[1] In this study, age was an independent risk factor of sarcopenia, which was easy to understand. A study reported that the annual rate of leg strength decline was higher in men than in women.[3] Our study also found that the male had a higher risk of sarcopenia compared with female, which suggested that men were more likely to develop sarcopenia. BMI, a practical, objective index of health, was commonly used to distinguish obesity from healthy weight. A cohort study of 3.6 million adults in the UK indicated that BMI had J-shaped associations with all-cause mortality, which meant that a high or low BMI all increased the risk of unfavorable outcomes.[4] In our study, individuals with lower BMI were more likely to develop sarcopenia. Possible explanations included the benefits of higher metabolic reserves and better ability of heavier people to resist sarcopenia. UA is the end-product of purine metabolism. It was commonly regarded as a harmful substance, as the increased level of serum UA was related to various diseases, such as diabetes, arthritis, renal disease, and metabolic syndrome. However, various studies have demonstrated that UA has antioxidant and neuroprotective effects, which reduced the risk of neurodegenerative diseases.[5] In our study, the UA level was higher in the non-sarcopenia group than in the sarcopenia group, which suggested that higher UA levels within the control range may be associated with a lower risk of sarcopenia. ALT was reported to be strongly associated with the risk of metabolic syndrome. A lower ALT level always indicated a lower metabolic rate. In our study, the ALT level was higher in the non-sarcopenia group than in the sarcopenia group, which suggested that high ALT levels within the control range may be associated with a lower risk of sarcopenia. Without relying on special measurement equipment, our screening tool could be used in all levels of hospitals, especially in primary hospitals or health care centers. Moreover, the model did not contain subjective items; thus, accuracy would not be influenced by such factors. However, our study has several shortcomings. First, this study was a cross-sectional design, which prevented us from establishing temporal relationships and then confirming cause–effect relationships. Second, data were collected from a population in one area in China. Since people in the same area have many similar characteristics and we used internal verification methods in this study, the general applicability of the proposed model needs to be investigated further. To minimize the sampling error, we plan to conduct relevant research in other areas in China to make the sample more representative and further optimize the model. Despite these limitations, it is indeed a new idea that can be considered for the early detection of sarcopenia. In summary, we developed a novel screening tool for sarcopenia, which contains easily obtained parameters. This simpler and more efficient screening tool might be more favorable for use in clinical practice and could achieve early detection of sarcopenia. In addition, the screening tool based on data from China had a good screening ability for sarcopenia in local areas. Acknowledgment We thank Wen Fang (the First Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China) for providing assistance with language editing. Funding This study was supported by the National Key Research and Development Program (No. 2018YFC2000301) and Key Research & Development Program of Zhejiang (No. 2022C03161). The funders had no role in the design of the study and the collection, analysis, and interpretation of data and in writing the manuscript. Conflicts of interest None.