Trajectory of mid-arm subcutaneous fat, muscle mass predicts mortality in hemodialysis patients independent of body mass index

Although decreasing body mass index (BMI) is associated with higher mortality risk in patients undergoing hemodialysis (HD), BMI neither differentiates muscle and fat mass nor provides information about the variations of fat distribution. It remains unclear whether changes over time in fat and muscle mass are associated with mortality. We examined the prognostic significance of trajectory in the triceps skinfold (TSF) thickness and mid-upper arm circumference (MUAC). In this multicenter prospective cohort study, 972 outpatients (mean age, 54.5 years; 55.3% men) undergoing maintenance HD at 22 treatment centers were included. We calculated the relative change in TSF and MUAC over a 1-year period. The outcome was all-cause mortality. Kaplan–Meier, Cox proportional hazard analyses, restricted cubic splines, and Fine and Gray sub-distribution hazards models were performed to examine whether TSF and MUAC trajectories were associated with all-cause mortality. During follow-up (median, 48.0 months), 206 (21.2%) HD patients died. Compared with the lowest trajectory group, the highest trajectories of TSF and MUAC were independently associated with lower risk for all-cause mortality (HR = 0.405, 95% CI 0.257–0.640; HR = 0.537; 95% CI 0.345–0.837; respectively), even adjusting for BMI trajectory. Increasing TSF and MUAC over time, measured as continuous variables and expressed per 1-standard deviation decrease, were associated with a 55.7% (HR = 0.443, 95% CI 0.302–0.649), and 97.8% (HR = 0.022, 95% CI 0.005–0.102) decreased risk of all-cause mortality. Reduction of TSF and MUAC are independently associated with lower all-cause mortality, independent of change in BMI. Our study revealed that the trajectory of TSF thickness and MUAC provides additional prognostic information to the BMI trajectory in HD patients.

End-stage kidney disease (ESKD) affects approximately 2 million people globally 1 , and approximately 69% of patients with ESKD are treated with hemodialysis (HD), which is one of the most widely used renal replacement therapies 2 .Despite advances in dialysis technology, these patients face a high risk of mortality, and the annual report on kidney disease from China shows that the mortality rate of HD patients reaches 12.5%, imposing a substantial burden on patients and the healthcare system 3 .Cardiovascular disease (CVD) is the most common cause of death, accounting for approximately 50% 3,4 .
The "Obesity paradox" phenomenon has been investigated in HD patients from several large cohort studies in the past decades 5 .Obesity is a traditional risk factor for mortality in the general population 6,7 .By contrast, obesity is inversely associated with better survival among HD patients 8 .Although obesity is generally defined by body mass index (BMI), some studies have stated that BMI is an imperfect predictor of mortality 9,10 , considering that BMI could not give a reliable assessment of body composition, especially could not differentiate muscle mass from fat mass 11 .Anthropometric measurements are simple, noninvasive, and cost-effective techniques, and have been widely used to evaluate body composition.As indicators of mid-arm measurements, triceps skinfold (TSF) thickness determines subcutaneous adipose tissue 12 , and mid-upper arm circumference (MUAC) reflects the amount of muscle mass of the mid-arm 13 , which is superior to BMI in predicting nutritional status.Previous

Anthropometric measurements
Mid-arm measurements were performed on the non-fistula arm by trained workers following standardized protocols for anthropometric measurements, as recommended by the World Health Organization.TSF thickness was measured at the mid-point of the posterior line between the olecranon and the tip of the acromion using skinfold calipers and recorded to the nearest 0.5 mm.MUAC was measured at the mid-point of the mid-upper arm with the elbow fully extended, and results were recorded to the nearest 0.1 cm.The mid-point of the midupper arm was defined as the midway between the olecranon process of the ulna and the acromion process of the scapula, which was located after bending the right arm to a 90° angle at the elbow.
Mid-arm muscle circumference (MAMC) was derived from the equation 22 : Other anthropometric measurements.Weight was measured using a calibrated beam scale with the participant wearing lightweight clothing, and height was measured without shoes using a portable stadiometer.BMI was calculated as weight (kilograms) divided by height (meters) squared (overweight ≥ 24 kg/ m 2 ).Waist circumference (WC) was measured at a point midway between the lowest rib and the iliac crest in a horizontal plane, and hip circumference was measured using non-plastic tape (central obesity ≥ 90 cm for males, and ≥ 85 cm for females).Waist-height ratio (WHtR) was calculated as WC (centimeters) divided by height (centimeters).
We calculated TSF trajectory from the difference in TSF over the first 1 year after randomization as (TSF at 1 year visit-baseline TSF)/baseline TSF, expressed as a percentage.MUAC, BMI, and WHtR trajectories were measured similarly.The values for trajectories in mid-arm measurements (TSF and MUAC) were divided into four categories according to quartiles of changes in TSF and MUAC, respectively.
The cutoff of TSF and MUAC trajectories were determined by the median values, respectively.The selected thresholds were then used to define the C1 (≤ cutoff) and C2 (> cutoff) groups.The survival analysis was further performed in the groups cross-classified with trajectories of TSF and MUAC.According the cutoff group, patients were categorized into four group: T1 (C1 of both TSF and MUAC), T2 (C2 of TSF and C1 of MUAC), T3 (C1 of TSF and C2 of MUAC), and T4 (C2 of both TSF and MUAC).

Clinical covariate measurements
Information on demographic characteristics, obtained via a questionnaire, included age, sex, and educational level (high: ≥ 12th or low: < 12th).Lifestyle factors were collected via the questionnaire and included smoking status (current/ever smoker or never), alcohol consumption (yes or no), and living status (living alone or with family).Blood pressure was measured after rest for 10 min in the seated position using standard sphygmomanometers before the HD treatment.Comorbidities were scored the modified Charlson comorbidity index (mCCI) for each HD patient, based on ICD-10 diagnostic codes 23 .All patients were evaluated with the malnutritioninflammation score (MIS), a specifical nutritional scoring system of evaluating malnutrition and inflammation for

Follow-up and outcomes
The primary endpoint was all-cause mortality.Participant death, the exact time of death, and the cause of death were identified from reports from each HD unit in each survey.The baseline for each participant was set as the survey day of the first entry into the survey with complete mid-arm measurements.The observation period was basically until either death, transfer to kidney transplantation, peritoneal dialysis, loss to follow-up, or the end of the study on September 30, 2022.

Statistical analysis
Participant baseline characteristics were described as a number (percentage) for categorical variables and means (SD) or as a median (interquartile range) for continuous variables.Normally distributed variables were expressed using mean ± standard deviation, whereas non-normally distributed variables were expressed as median and interquartile range.The differences among quartile groups divided by TSF, and MUAC trajectories were compared using one-way analysis of variance or the Kruskal-Wallis test for continuous variables and the chi-squared test for categorical variables.
The associations between baseline and trajectory of BMI, TSF, and MUAC were analyzed with linear regression.Kaplan-Meier survival curves were used to assess the associations between trajectories of mid-arm measurements and mortality, with the log-rank test examining the significant differences between different groups.Cox proportional hazards models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause mortality, and multiple covariates were included in multivariable-adjusted models.TSF and MUAC trajectories were analyzed as continuous variables (per SD increment), quartile variables, and threshold values.Model 1 was adjusted for age, sex, educational level, living status, smoking status, and alcohol consumption.Model 2 was adjusted for the variables in Model 1 as well as systolic blood pressure, diastolic blood pressure, and diabetes, hypertension, cardiovascular disease, dialysis frequency, hemoglobin level, albumin level, creatinine level, MIS, mCCI, and BMI trajectory.Model 3 was adjusted for the variables in Model 2 as well as baseline BMI, TSF, and MUAC.To examine the independent association between TSF and MUAC trajectories, a mutually adjusted model was created by including both MUAC and TSF change as well as the covariates in Model 3. In order to model the shape of associations with flexibility and to test for linearity, restricted cubic splines (RCS) with three knots at the 10th, 50th, and 90th centiles were used for TSF and MUAC trajectories in multivariable-adjusted models.
Possible modifiers of the association between TSF, MUAC trajectories with the risk of all-cause mortality were assessed for variables including age (< 65 years or ≥ 65 years), sex (male or female), diabetes (yes or no), BMI (Overweight or no), WC (central obesity or no) at baseline, and MUAC/TSF trajectory, respectively.The heterogeneity between groups was determined using the P value for heterogeneity, which was calculated with multiplicative terms by multiplying mid-arm measurements by continuous variables used in the multivariable model.
Additionally, as kidney transplantation was a competing risk event against death, the cumulative incidence considering the competing risk was compared using Gray's test, and the Fine and Gray sub-distribution hazards model was used in the multivariate model as a sensitivity analysis method for the outcome, together with the standard Cox regression model for cause-specific hazards.
All statistical analyses were performed with R version 4.2.2 (www.r-project.org/), and P < 0.05 (two-sided) was considered statistically significant.

Ethical standards
Ethical approvals were obtained from Guizhou Provincial People's Hospital-Research Ethics Committees (Approval number: (2020)208).All participants provided written informed consent and all research procedures were conducted in accordance with relevant guidelines and regulations.
During the follow-up for a median of 48.0 (37.0, 59.0) months, 206 (21.2%) all-cause mortality events were recorded.Compared with the lowest quartile of TSF trajectory, HD patients with a higher quartile had a lower risk of all-cause mortality (P < 0.05).Patients stratified by MUAC trajectory had similar results (Table 1).

Association between all-cause mortality and TSF, MUAC trajectories
For both TSF and MUAC trajectories, Kaplan-Meier analyses revealed a lower risk of all-cause mortality in the highest quartile group (Fig. 2), and the log-rank statistics showed significant differences in survival time among groups (P < 0.001).
Mid-arm measurements were treated as continuous variables, TSF, and MUAC were consistently and inversely associated with all-cause mortality (Table 2).Representing HRs per 1-SD increase (standardized HR), after full adjustments, a 1-SD increase in relative TSF, and MUAC trajectories were associated with a 44.5%, and 96.2% decreased risk of all-cause mortality, respectively (mutual model).
The median stratification method showed that the cutoff values of the TSF trajectory was 0%, and the MUAC trajectory was − 4.0%.Compared with the C1 group, HD patients in the C2 group of TSF, and MUAC trajectory both had a lower risk of all-cause mortality (HR = 0.661, 0.654, both P < 0.005).

Cross-classified analyses
The associations in the cross-classified analyses are shown in Table 3.Compared with T1 group (both lower TSF and MUAC trajectory), T2, T3 (lower TSF or MUAC trajectory), and T4 (both higher TSF and MUAC trajectory)  www.nature.com/scientificreports/all had lower risk of all-cause mortality (P < 0.05) in the multivariate model.HD patients with higher TSF and MUAC trajectories had a 46.9% decreased risk of all-cause mortality (Table 3).

RCS analyses
In multivariable-adjusted cubic spline analyses showed that all-cause mortality decreased consistently with increasing MUAC trajectory (Fig. 3B), and the spline variable confirmed no departure from the linear relationship www.nature.com/scientificreports/(nonlinear P = 0.4531) of MUAC trajectory and all-cause mortality.However, the TSF trajectory showed an inverted L-shaped association with all-cause mortality (nonlinear P = 0.0230; Fig. 3A).

Subgroup analyses
This study also performed subgroup analyses to explore potential heterogeneity between TSF, MUAC trajectory, and mortality, stratified by age (< 65 or ≥ 65 years), sex (female or male), diabetes (no or yes), BMI (normal or overweight), WC (normal or central obesity), and TSF or MUAC at baseline (low or high as the median), respectively (Table S1), which are shown in the forest plots (Fig. 4).In any subgroups, TSF and MUAC trajectory were considered as continuous variables with 1-SD.There were significant interactions between MUAC trajectory and age (P = 0.013), baseline MUAC (P = 0.039), and all-cause mortality (Table S1; Fig. 4).Robust results

Figure 3.
The dose-response relationship between TSF, MUAC trajectory and all-cause mortality in hemodialysis patients.Point estimates (solid line) and 95% confidence intervals (dashed lines) were estimated by restricted cubic splines analysis with knots placed at the 10th, 50th, and 90th percentile.Model was adjusted for age, sex, educational level, living status, smoking status, alcohol consumption, systolic blood pressure, diastolic blood pressure, diabetes, hypertension, cardiovascular disease, dialysis frequency, body mass index trajectory, and MUAC or TSF trajectory.TSF, triceps skinfold; MUAC, mid-upper arm circumference.www.nature.com/scientificreports/were found in associations between all-cause mortality and TSF and MUAC trajectory in subgroups, including age < 65, diabetes, central obesity at baseline, and low TSF or MUAC at baseline (Fig. 4).

Discussion
In this multicenter, prospective cohort study, we investigated the association between the trajectory of mid-arm measurement and mortality in a cohort of patients receiving HD.Results demonstrated that reduction in TSF, and MUAC over a 1-year period is strongly associated with the higher risk of all-cause mortality among HD patients, independent of demographic characteristics, and comorbid conditions.Of note, the reduction in TSF and MUAC were still negatively associated with survival, even after adjusting baseline BMI, TSF, MUAC, and BMI trajectories over a 1-year period.Therefore, these findings underscore the importance of routine monitoring of TSF and MUAC, in order to provide significant prognostic information and will guide interventions to optimize the survival outcomes in HD patients.
A number of studies have proven that free fat is beneficial to the health of humans 25 , and its depletion is common in HD patients 26 .The skin is one of the largest organs that store adipose tissue, TSF-reflected midarm subcutaneous fat could represent the distribution of peripheral fat well, and its loss is an effective predictor for both inflammation and malnutrition 27 .TSF thickness has been considered a promising tool for predicting cardiovascular events and mortality risk 14,15 .For HD patients, Huang et al. also found a significant association between lower quartiles of TSF thickness and higher all-cause mortality in 1709 patients 16 .Inadequate subcutaneous fat can disturb normal glucose and lipid metabolism and immune response by inhibiting the production of leptin 28 .The deficiency of subcutaneous tissue adipose would further cause ectopic fatty deposition, inducing chronic inflammation, and insulin resistance 29 .In addition, insufficient subcutaneous fat can also cause atherosclerosis and non-adipose tissue lipotoxicity by inhibiting the separation of non-esterified fatty acids from food 30 , potentially increasing all-cause death risk 31,32 .
This current study found that TSF decrease over a 1-year period was associated with a higher risk of all-cause mortality for the first time.Several studies showed similar results 21,33 .The study of Kamyar et al. found that fat loss over time, defined as TSF-estimated body fat fraction, was independently associated with higher mortality in HD patients.This study showed that a fat loss (< 1%) was associated with a death risk 2 times that of patients who gained fat (> 1%) (HR:2.06;95% CI 1.05-4.05;P = 0.04) 33 .However, Hollander et al. showed decreases in TSF had no association with all-cause mortality among the European elderly population of 70-to 77-year-old individuals after adjusting for characteristics 20 .The different results may depend on that this study analyzed the TSF divided into quintiles and used the smallest change as the reference category, they did not examine the TSF as a continuous variable.
MUAC is a reliable substitution of body mass or muscle mass and is readily a clinically useful indicator of nutritional status 13 .A decrease in MUAC may primarily reflect loss of muscle mass, leading to protein-energy wasting, sarcopenia, and malnutrition inflammation syndrome.These processes have been proven to be the determinants of CVD and mortality risk 19,20 .The present result of a decrease in MUAC being associated with increased mortality risk is in line with the previous finding by Schaap et al., in which decreases in MUAC have the strongest association with all-cause mortality 19 .Moreover, this study focused on the specific population with HD as kidney replacement therapy, which extends the existing conclusions.The current study observed that 1-SD increase in relative MUAC change was inversely associated with risk reductions for all-cause mortality in a consistent linear trend without significant departure, which is not consistent with the findings by De Hollander et al. 20 .In the study, a decrease, as well as an increase in MUAC, were significantly associated with increased all-cause mortality risk.Our current study did not observe a similar pattern.However, to date, no other studies have reported an association between increasing MUAC and higher mortality risk among HD patients.
Of note, in the current study, after adjusting for BMI trajectory in the Cox hazard analyses, the relationships between TSF, MUAC trajectories, and all-cause mortality were still observed.Although previous studies have proven that a declining body weight, commonly defined with BMI or WC, is associated with higher mortality 34,35 , some studies demonstrated that BMI and its change were not associated with mortality 36,37 .Taken the accuracy of BMI trajectory could be influenced by a change in body water balance, BMI might diminish the reproducibility of this measure, weakening the association with mortality.Its interpretation in the assessment of nutritional status could be compromised by the inability to assess the distribution of body composition, which implies that BMI is not a reliable marker of nutritional status in HD patients 38 .What's more, the ratio of fat and muscle mass loss in BMI decrease may also vary depending on age and sex 36 .Recent studies have suggested that TSF and MUAC can be more feasible and more valid measures of free fat and thinness than BMI, respectively 37,39 .
The current study implies that subcutaneous fat mass and muscle mass trajectories provide additional prognostic information to the BMI trajectory.Additionally, the lack of interaction between baseline and 1-year change in fat mass in terms of mortality suggests that the relative benefit of increasing fat mass, or the deleterious effect of decreasing fat mass, remains consistent irrespective of starting level.Thus, this study provides further justification for evaluating the fat mass trajectory in addition to the BMI trajectory in patients receiving HD.However, we found an interaction between baseline and 1-year change in muscle body in terms of mortality.When initial MUAC is low, a reduction in MUAC is more prominently associated with all-cause mortality.A possible explanation for the result is that decreasing MUAC reflects muscle mass loss and protein energy wasting, and is related to an increased risk of mortality 40 .Anyway, we found that there were higher prognostic values for the TSF and MUAC changes than for BMI change, which further emphasizes the importance of performing subcutaneous fat and muscle mass assessment for managing the HD population.
We also calculated thresholds of TSF and MUAC trajectories over a 1-year period to facilitate the identification of survival-related low fat or low muscle mass for HD patients.Although a number of studies have emphasized the integrality of the assessments of fat and muscle mass, which can be reflected by the TSF and MUAC respectively, any current nutritional risk screening tools or diagnosing criteria for HD patients have included the two assessments as the main components.Furthermore, the importance of these regular measurements is still seriously underestimated among the HD population.Therefore, we provided the thresholds of TSF and MUAC trajectories for the first time, which may offer reference values for future studies addressing the assessment of the body composition change in clinical settings.This is the first study to explore the prognostic potential of mid-arm measurement trajectory in mortality for HD patients.These findings are valuable for improving regular nutritional status evaluations and risk stratification.In addition, the long follow-up time of a median of 48 months and the fact that data were obtained from 22 HD units in Southwestern China make it possible to generalize the results to the general HD population in this area.Due to their simplicity and cost performance, regular and reduplicative measurements of TSF and MUAC over time may have critical values as surrogate indicators of changes in fat mass and body mass among HD patients.These findings emphasize the importance of subcutaneous fat and muscle mass assessment to guide strategies to optimize long-term outcomes in HD patients.Our study also suggests that the measurements of TSF and MUAC trajectories may be better methods for assessing nutritional status to predict poor prognosis independent of body weight trajectory.Therefore, in addition to weight loss, clinicians should also consider interventions to improve the fat and muscle mass in HD patients, such as more individualized nutritional supplementation.These may be considered strengths of the current study.
There are several limitations to declare.First, the results of our study were based on the data from the HD population in Southwestern China, and the findings need to be confirmed in other populations.Second, despite a great quantity of potentially confounding factors having been adjusted, and the nature of all observational studies, some undetected and unmeasured confounders still cannot be excluded, such C-reactive protein levels, and oral medications.Third, although muscle mass and soft tissue volume are critical prognostic factors, it was essential to consider muscle mobility and motility as risk factors.However, the quantity and quality of exercise were not evaluated, due to the inconvenience among HD outpatients.Further analyses of muscle parameters are still needed to comprehensively elaborate the effect of muscle on prognosis among HD patients.Finally, TSF and MUAC measurements are commonly considered to have a low reproducibility.To overcome this barrier, this study dispatched the same group of trained research doctors to perform clinical assessments through faceto-face questionnaire interviews and physical measurements.Moreover, baseline and follow-up measurements were performed following standardized protocols for anthropometric measurements, as recommended by the World Health Organization.Therefore, this ensures the reproducibility of TSF and MUAC measurements in a certain extent.Currently, some precise measurements of changes in body composition have been applied in clinical practice, such as dual energy x-ray absorptiometry, CT, MRI, or bioelectrical impedance analysis 41 , which are more sophisticated than mid-arm measurements.However, the use of these measures is not feasible to popularize for clinical application, due to the complicated operation, expensive, and dependent on facility instruments.Besides, TSF and MUAC can be conveniently measured at smaller institutions and in community settings.Therefore, measurements of body composition represent a practical approach that can balance clinical needs and cost-effectiveness in a wide range of scenarios.Nevertheless, future studies on body composition trajectory are needed to confirm our findings.
In conclusion, we found that decreasing TSF thickness and MUAC over time, are independently associated with higher risk of all-cause mortality, independent of BMI trajectory.The TSF thickness and MUAC trajectories are recommended to be convenient and credible indicators to predict mortality in clinical practice.Further high-quality randomized controlled trials of early intervention of TSF thickness and MUAC are required in patients receiving HD.

Figure 2 .
Figure 2. Kaplan-Meier analyses for the incidence of all-cause mortality in patients receiving hemodialysis according to the TSF and MUAC trajectory.Patients were divided into 4 groups according to percent change quartiles of TSF and MUAC.TSF, triceps skinfold; MUAC, mid-upper arm circumference.

Figure 4 .
Figure 4. Subgroup analyses of association between TSF, MUAC trajectory and all-cause mortality in hemodialysis patients.Model was adjusted for age, sex, educational level, living status, smoking status, alcohol consumption, systolic blood pressure, diastolic blood pressure, diabetes, hypertension, cardiovascular disease, dialysis frequency, body mass index trajectory, and MUAC or TSF trajectory.TSF, triceps skinfold; MUAC, midupper arm circumference; BMI, body mass index; WC, waist circumference; HR, hazard ratio; CI: confidence interval.

Table 2 .
Cox regression analysis of all-cause mortality with TSF and MUAC trajectories in HD patients.
a Model 1, adjusted for age, sex (male or female), educational level (low or high), living alone (yes or not), smoking status (ever/current or never smoker), alcohol consumption (yes or no).b Model 2, adjusted for Model 1, systolic blood pressure, diastolic blood pressure, and diabetes, hypertension, cardiovascular disease, dialysis frequency (less than thrice one week or thrice one week and more), hemoglobin level, albumin level, creatinine level, and malnutrition inflammatory score, modified Charlson comorbidity index, and BMI trajectory.c Model 3, adjusted for Model 2, baseline body mass index, TSF, and MUAC.d Mutual model, adjusted for Model 2, and MUAC trajectory or TSF trajectory.*The cutoff value was calculated as the median of TSF or MUAC trajectory, and C1 were regard as the reference.TSF triceps skinfold, MUAC mid-upper arm circumference, HR hazard ratio, CI confidence interval.

Table 3 .
Cox regression analysis of mortality according to the cutoff values of TSF and MUAC trajectories in HD patients.a Model 1, adjusted for age, sex (male or female), educational level (low or high), living alone (yes or not), smoking status (ever/current or never smoker), alcohol consumption (yes or no).b Model 2, adjusted for Model 1, systolic blood pressure, diastolic blood pressure, and diabetes, hypertension, cardiovascular disease, dialysis frequency (less than thrice one week or thrice one week and more), hemoglobin level, albumin level, creatinine level, and malnutrition inflammatory score, modified Charlson comorbidity index, and BMI trajectory.c Model 3, adjusted for Model 2, baseline body mass index, TSF, and MUAC.TSF triceps skinfold, MUAC mid-upper arm circumference, HR hazard ratio, CI confidence interval.