Impact of Percent Body Fat on All-Cause Mortality among Adequate Dialysis Patients with and without Insulin Resistance: A Multi-Center Prospective Cohort Study

The association between body fat and mortality in hemodialysis patients remains controversial. We examined the effect of percent body fat (PBF) on all-cause mortality among adequate hemodialysis patients with and without insulin resistance (IR). A prospective cohort study was conducted on 365 adequate hemodialysis patients (equilibrated Kt/V ≥ 1.2) from seven hospitals. Patients’ characteristics and clinical and biochemical parameters were assessed at baseline between September 2013 and April 2017. Patients were followed up for all-cause mortality until April 2018. The median value of homeostatic model assessment (HOMA-IR) was used to classify IR. Cox proportional hazard models were utilized to examine predictors of all-cause mortality. During 1.4 (1.0–3.2) years of follow-up, 46 patients died. In patients with IR (HOMA-IR ≥ 5.18), PBF was significantly higher in the survival group than in the death group (31.3 ± 9.0 vs. 25.4 ± 8.2, p = 0.005). After controlling for confounding factors, PBF was significantly associated with lower risk for all-cause mortality in patients with IR (hazard ratio, 0.94; 95% confidence interval, 0.89–1.00; p = 0.033). The association was not observed in patients without IR. In conclusion, percent body fat shows a protective effect on survival in hemodialysis patients with IR.


Introduction
Obesity is increasing across the globe. It is a strong predictor of metabolic syndrome [1], cardiovascular diseases (CVD), and mortality in different populations [2][3][4]. Conversely, in end-stage renal disease (ESRD) patients undergoing hemodialysis, the "obesity paradox" is well-reported [5][6][7]. It appears that higher body fat, such as higher body mass index [7], and higher level of abdominal fat (e.g., visceral fat, and subcutaneous fat area) are associated with lower mortality in hemodialysis patients [8].
Insulin resistance (IR) is highly prevalent and well-known as a predictor of cardiovascular disease [9], and mortality in ESRD patients [10]. In addition, poor glycemic control is significantly associated with high risk of cardiovascular and all-cause mortality in maintenance hemodialysis patients [11]. It becomes more complicated if hemodialysis patients have abnormalities in glucose metabolism [12]. Therefore, novel approaches in insulin resistance treatment and management are set as important targets for improving clinical outcomes and preventing excess death in CKD patients [13], and ESRD patients treated with hemodialysis [12,14].
Hemodialysis patients with obesity have a survival advantage, which has been reported in empirical studies [15][16][17] and in the meta-analysis [18]. However, most of the previous studies have looked at body mass index as a measure of body size, which does not distinguish between lean body mass and fat mass. A direct measure of body fat mass has been used in a limited number of studies. In addition, the protective effect of body fat among hemodialysis patients with insulin resistance remains to be explored. Therefore, we aimed to examine the effect of percent body fat on all-cause mortality in hemodialysis patients with and without insulin resistance. We hypothesized that higher percent body fat is associated with lower risk of death in patients with insulin resistance.

Study Design and Settings
We conducted a prospective cohort study in seven dialysis centers of Taipei Medical University Hospital, Wan Fang Hospital, Shuang Ho Hospital, Cathay General Hospital, Taipei Tzu-Chi Hospital, Wei-Gong Memorial Hospital, and Lotung Poh-Ai Hospital.

Patient Recruitments
Patients aged 20 years and above, receiving three hemodialysis sessions a week for at least three months were included. Patients were excluded from the study if they were diagnosed with edema, pregnancy, amputation, hyperthyroidism, hypothyroidism, malignancy, received tube feeding, exhibited hepatic failure or cancer, had been hospitalized within one month prior to the recruitment, or had scheduled surgery. Patients with inadequate dialysis quality (equilibrated Kt/V < 1.2 g/kg/day) at the recruitment or during the following-up were also excluded. Patients' recruitment criteria was also described previously [19]. At baseline (September 2013-April 2017), patients' characteristics and clinical and biochemical parameters were examined. Patients were then followed up for all-cause mortality until April 2018. A final sample of 365 patients were followed up and analyzed, which is illustrated in Figure 1.

Patients' Characteristics
Patients' age, gender, hemodialysis vintage, and Charlson comorbidity index [20] were assessed using chart review. The short version of the International Physical Activity Questionnaire was used to assess physical activity level [21]. Patients were asked about their time spent on vigorous, moderate, walking, and sitting activities over the last seven days. Physical activity was estimated using the metabolic equivalent task scored in minutes per week (MET-min/week) [22]. The number of MET-min/week was the sum of minutes that patient spent at each level, multiplied by 8.0, 4.0, 3.3, and 1.0, respectively [23].
The homeostatic model assessment index (HOMA-IR) was used to assess insulin resistance. The index was calculated using the formula developed by Matthews et al. [24]: HOMA-IR = fasting plasma insulin (μU/mL) × fasting plasma glucose (mg/dL)/405. The median value of HOMA-IR was used to classify patients into groups with and without insulin resistance. This classification was also applied in previous studies [10,25].

Patients' Characteristics
Patients' age, gender, hemodialysis vintage, and Charlson comorbidity index [20] were assessed using chart review. The short version of the International Physical Activity Questionnaire was used to assess physical activity level [21]. Patients were asked about their time spent on vigorous, moderate, walking, and sitting activities over the last seven days. Physical activity was estimated using the metabolic equivalent task scored in minutes per week (MET-min/week) [22]. The number of MET-min/week was the sum of minutes that patient spent at each level, multiplied by 8.0, 4.0, 3.3, and 1.0, respectively [23].
The homeostatic model assessment index (HOMA-IR) was used to assess insulin resistance. The index was calculated using the formula developed by Matthews et al. [24]: HOMA-IR = fasting plasma insulin (µU/mL) × fasting plasma glucose (mg/dL)/405. The median value of HOMA-IR was used to classify patients into groups with and without insulin resistance. This classification was also applied in previous studies [10,25].

Ethical Consideration
The study protocol was approved by the Ethics Committees (JIRB No. 201302024; OP104001; 04-M11-090). Patients signed informed consent forms before the study conducted.

Statistical Analysis
The distribution of continuous variables was tested using a Shapiro-Wilk test (normal if p-value > 0.05) [36,37], and a visual inspection of their histograms, normal Q-Q plots and box plots was made. The variables with approximately normal distributions were reported as mean ± SD, otherwise median (interquartile range) is reported. The categorical variables were presented as frequency and percentages. The t-test, Mann-Whitney U test, or the Chi-Square test were used to compare the distribution of characteristics and clinical and biochemical parameters between the survival and death groups, respectively.
The associations of patients' characteristics and clinical and biochemical parameters with all-cause mortality in the overall sample and subgroups of HOMA-IR were analyzed using bivariate Cox proportional hazard models. Furthermore, the protective effect of percent body fat on survival in patients with and without insulin resistance was examined using multivariate Cox proportional hazard models. The adjusted variables were those with significant effects on all-cause mortality in the bivariate model, appropriately. Hazard ratios and 95% confidence intervals were reported.
Data were analyzed using the IBM SPSS software version 20.0 for Windows (IBM Corp., New York, NY, USA). The significant level was set at p-value < 0.05.

Discussion
The most important finding of the current study is that a higher percent body fat was significantly associated with lower risk for all-cause mortality in the overall sample, and in patients with elevated HOMA-IR. In previous studies, body fat (as measured by percent body fat, or fat tissue, or body mass index) has been found to be an independent predictor for longer survival in hemodialysis patients [8,[38][39][40][41][42]. In addition, a low baseline PBF and fat loss over time significantly increases mortality in maintenance hemodialysis patients, as found in a previous study [43]. In the current study, the percent body fat, but not body mass index, was significantly associated with all-cause mortality. This might reflect that percent body fat (as a direct measurement of body fat) is more accurate than body mass index (which cannot distinguish lean body mass and fat mass) in predicting all-cause mortality. One previous study in Japan has also reported that fat tissue index is more accurate than body mass index in determining all-cause mortality in hemodialysis patients [41].
In the present study, in the overall sample, higher levels of serum albumin, creatinine, and uric acid were significantly associated with reduced all-cause mortality. Higher Charlson comorbidity index, cardiomegaly (elevated CTR), inflammation (high hs-CRP), and impaired fasting glucose were significantly associated with increased all-cause mortality in the bivariate analysis. However, in the multivariate analysis, among above factors, elevated hs-CRP remained the significant association with mortality. Lower levels of serum albumin and elevated CRP were well-reported predictors of all-cause mortality in a previous study [44].
Elevated levels of intact parathyroid hormone (iPTH) were significantly associated with lower risk of all-cause mortality. This finding is consistent with previous studies' results that decreased iPTH is associated with increased cardiovascular mortality [45] and all-cause mortality in hemodialysis patients [45,46].
In addition, serum uric acid was significantly inversely associated with all-cause mortality in the bivariate analysis and attenuated in the multivariate analysis in the current study. This is contradictory to the finding in the general population that a high level of serum uric acid is a risk for cardiac death [47]. In hemodialysis patients, higher serum uric acid has been reported as a protective factor against all-cause and cardiovascular mortality [48][49][50][51]. In addition, serum uric acid predicts mortality in hemodialysis patients with diabetes [52]. In our study, higher serum uric acid was associated with a lower risk of all-cause mortality in patients with insulin resistance. Serum uric acid could then play a mortality predictive role in hemodialysis patients.
The study has a certain number of limitations. Firstly, observational bias may have existed during the follow-up. However, patients were recruited from seven hospitals which might increase the external validity of findings. Secondly, the medications being used were not assessed, which may affect the associations of studied variables. In addition, estimates made on the basis of so few events are not stable, and thus not generalizable. Future studies with a comprehensive assessment are suggested to assure the current findings, and deeply examine the mechanism. The future research should be conducted in larger population with different races and ethnicities as obesity paradox varies across race-ethnic populations [53]. Regardless of the limitations, the current study demonstrates strength with its objective measures of body composition, and clinical and biochemical values.

Conclusions
The current study presents an association between percent body fat and all-cause mortality in adequate dialysis patients with and without insulin resistance. The results show that higher percent body fat is associated with lower all-cause mortality, especially in patients with insulin resistance. Assessments of body composition and insulin resistance are important for identifying appropriate therapies which can further reduce mortality in hemodialysis patients.