In the present study, we developed and validated a simple nomogram model to predict the risk of developing sarcopenia in MHD patients with a total of six clinical relevant variables including serum creatinine, albumin, C-reactive protein, serum phosphorus, BMI and upper arm muscle circumference. The AUC, internal validation C-statistic, calibration curve, and DCA curve were constructed and verified the reliability as well as the accuracy of this model. The discovery of this model would not only help to identify MHD patients with increased risk of developing sarcopenia, but may also predict the disease course and provide reference for prognosis estimation.
Nomogram can predict the probability of disease by analyzing and integrating identified disease risk factors, therefore providing valuable information for making better clinical decisions. It has been widely used in oncology and chronic diseases worldwide. For instance, Cheng et al. developed a nomogram to predict the risk of initiating renal replacement therapy within 3 years in patients with diabetic nephropathy and Jing et al developed a nomogram including multiple echocardiographic measures to assess 3-year all-cause mortality in hemodialysis patients which both showed good accuracy and reliability. In addition, Ouyang et al. developed and validated an easy-to-use nomogram that can accurately predict 1-year, 5-year, and 10-year survival in hemodialysis patients. However, to the best of our knowledge, no study has been conducted to develop a nomogram that can predict sarcopenia in MHD patients.
In this study, the multivariate logistic regression analysis indicated that serum creatinine was an independent risk factor for sarcopenia in MHD patients, which was consistent with the findings of Lin et al. Creatinine is a metabolite of human skeletal muscle and is mainly excreted through glomerular filtration. In MHD patients with limited renal function, serum creatinine is not only a marker of renal failure, but also a predictor of nutritional status and decreased skeletal muscle mass . Hyperphosphatemia is a common complication in MHD patients which is closely related to increased risk of vascular calcification and cardiovascular mortality . Interestingly, our study indicated that lower serum phosphorus level was associated with the development of sarcopenia in MHD patients, which was consistent with the findings of Ren et al. We hypothesized that a high-protein diet is the main source of phosphorus for uremic patients who are often accompanied by loss of appetite or even anorexia. A decrease in food intake will inevitably lead to a decrease in serum phosphorus, malnutrition, and protein energy expenditure in patients which ultimately leads to the occurrence of sarcopenia . In our clinical practice, renal function and electrolytes are measured every 1-3 months in MHD patients to assess changes in the condition. In this study, when the measured serum creatinine and serum phosphorus levels were significantly decreased, a detailed assessment of the presence of malnutrition and sarcopenia was conducted in MHD patients and nutritional support with dietary guidance and health education was provided. This intervention might help to reduce the occurrence of sarcopenia.
BMI and MAMC are conventional nutritional assessments for MHD patients and previous studies[26–28] indicated that both were independent predictors of survival. The study by Su et al showed that in MHD patients, the decrease in MAC was associated with increased all-cause mortality and cardiac events, especially those with low BMI. Unsurprisingly, data in our study showed that MHD patients with decreased BMI and MAMC were more likely to develop sarcopenia, which was consistent with previous studies[22, 30]. Therefore, more attention should be paid to MHD patients with low BMI and/or low MAC and sufficient nutritional intervention should be applied in a timely manner to reduce the occurrence of sarcopenia.
Consistent with previous studies[12, 26], this study also found out that the level of C-reactive protein was increased while the level of serum albumin was decreased in MHD patients who developed sarcopenia compared to non-sarcopenia patients. C-reactive protein is one of the most commonly used biochemical indicators to examine inflammation. It has been well established in the field that hemodialysis patients are often under microinflammatory state due to multiple reasons. The close association between inflammation and sarcopenia has been well studied. Inflammatory factors can activate many signaling pathways involved in the pathogenesis of sarcopenia, resulting in decreased anabolism and increased catabolism of proteins . Serum albumin is often used to estimate nutrition levels in MHD patients and is closely related to patient prognosis . Low serum albumin may lead to increased protein catabolism and decreased muscle strength. However, the study by Alves et al. showed that the nutritional status of patients with low serum albumin group was not significantly different from that of patients with normal serum albumin level under non-inflammatory conditions but significantly associated with higher mortality under systemic inflammation. Therefore, more attention should be paid to patients’ inflammatory status, especially for those with decreased serum albumin level.
At present, the commonly used scales for sarcopenia are SARC-F score which measures strength, assistance walking, rising from a chair, climbing stairs and falling, and modified SARC-assisted Cal F score which also measures the situation of the calf. A meta-analysis showed that the sensitivity of SARC-F was low to moderate (28.9% – 55.3%) as well as the specificity (68.9% – 88.9%). Although SARC-CalF is associated with a higher specificity (87.7% – 91.3%), its sensitivity was not satisfying (45.9% – 57.2%). The relatively low sensitivity of these two scales renders a higher risk for miss diagnosis. On the contrary, our novel nomogram provided an alternative method with increased clinical efficacy. The AUC of our constructed nomogram model was 0.8806 in the development cohort and 0.8613 in the validation cohort with an internal validation C-statistic of 0.864. The validity of this novel model was further verified by calibration curves and DCA curves. More importantly, all six variables included in this model are laboratory tests and anthropometric measures that are routinely measured in clinical practice and do not require additional examinations as well as costs.
However, this study has some limitations. First, even though the number of enrollments was relatively large, it was conducted at a single center that might not be representable for the general population. In addition, the constructed nomogram model was not validated with external data. Furthermore, this study excluded cases with incomplete data, which may lead to a selection bias. Therefore, this model should be validated through prospective, multicenter clinical studies in the future.