A prospective, one-year study was conducted in which 18 patients undergoing HD were included. The population was selected among patients treated in the Nephrology Department of Hospital Universitario La Paz. The inclusion criteria were as follows: patients over 18 years of age; on renal replacement therapy for a period of more than 3 months and less than 5 years, with no deterioration in cognitive abilities and who will sign the informed consent. The exclusion criteria were: patients with active cancer, or limited life expectancy. Advanced age was not an exclusion criterion. Demographic data and data on renal replacement therapy, as well as comorbidities or intercurrent processes, were recorded. The recruitment period for patients lasted from May 2011 to March 2013. An informed consent template was designed and it was signed by each participant. This study was approved by the Independent Ethics Committee of Hospital Universitario La Paz.

The measurements of anthropometric and body composition, and the collection of blood samples, were obtained under fasting conditions at baseline (between 2011 and 2013) and 12 months later.

The biochemical analysis was performed in the Clinical Analysis Department of Hospital Universitario La Paz. In both samples the following parameters were measured: glycaemia, lipid profile (cholesterol and triglycerides), albumin, insulin, HOMA-IR index, free fatty acids, C-reactive protein (CRP), leptin, adiponectin, interleukin-6 (IL-6), tumour necrosis factor alpha (TNF-α) and FGF-21. All the laboratory tests were done under fasting conditions: pre-dialysis during the morning shift in HD patients, coinciding with peritoneal kinetics in peritoneal dialysis (PD) patients.

Subsequently, we performed a second part of the study; a control group of 17 patients on PD, from the Nephrology Department, who fulfilled the same criteria as the HD patients were compared with the 29 patients on HD (cross-sectional baseline). Thus, this study compares data on the body composition of both groups of patients to perform concordance studies among methods for analysing body composition.

The anthropometric measurements of the patients were obtained in accordance with the standard technique and the current international recommendations (WHO, 1976). These measurements were obtained in subjects barefoot and in their underwear. The measurements were always taken pre-dialysis in HD patients, and with the abdomen empty in PD patients. A body-composition analyser (TANITA BC-420MA, Biológica Tecnología Médica S.L., Barcelona, Spain) was used to measure weight. Height was measured by a millimetre precision rod (range: 80–200cm). To measure mid-upper arm muscle area (MUAMA), an inextensible millimetre precision metric tape (0.1cm) was used. Triceps skin fold (TSF) was obtained by means of a Holtain skinfold calliper with a range of 20cm and a sensitivity of 0.2mm. BMI (weight [kg]/height [m2]) was calculated from the anthropometric weight and height measurements. Waist circumference (WC) was measured in cm, using a millimetre precision tape measure with the waist in a horizontal position (narrower torso level, midline between the iliac crest and the last rib). This was measured at the end of a normal exhalation.

Body composition was quantified using both BIA and DXA. To determine body composition by BIA, a four-pole vector device (50kHz), with an intensity of 0.8mA, was used (model BIA 101 by Akern Systems, Florence, Italy). The measurement was performed according to the criteria established by the National Institutes of Health Technology Assessment Conference Statement.16 For patients who presented with vascular access for dialysis, all measurements were performed on the contralateral side.17 The BCM (Body Composition Monitor, Fresenius Medical Care, Germany) was used as multifrequency BIA, with a frequency range between 5 and 1000kHz. DXA was performed in the Nuclear Medicine Department of Hospital Universitario La Paz using the region of interest (ROI) recommended by the bone radiodensitometer manufacturer with enCore LUNAR 43616 ES software (GE Healthcare, Buckinghamshire, United Kingdom) for the regional measurements of BFM (trunk, gynoid and android).

The blood samples used to determine FGF-21 were collected in heparin-lithium plasma tubes and centrifuged at 3500rpm for 10min prior to storage at −40°C or at −80°C, until analysis. FGF-21 was quantified using an enzyme-linked immunosorbent sandwich assay (ELISA), according to the manufacturer's instructions (Merck-Millipore, Darmstadt, Germany). The assay sensitivity was 1.7pg/ml, while the intra- and inter-assay precisions were 5.7% and 6.9%, respectively. The expected normal range was less than 200pg/ml. Leptin, IL-6, TNF-α and adiponectin were quantified by multiplex immunoassay (Milliplex®, Merck-Millipore, Darmstadt, Germany). IL-6 sensitivity was 1.4pg/ml, while the intra- and inter-assay precisions were 7% and 13%, respectively. Leptin sensitivity was 7.3pg/ml, while the intra- and inter-assay precisions were 5% and 11%, respectively. TNF-α sensitivity was 0.4pg/ml, while the intra- and inter-assay precisions were 8% and 7%, respectively. Free fatty acids were measured based on an in vitro enzymatic colorimetric method (A25, Biosystems, Barcelona, Spain). The expected values were 2.8–16.9mg/dl for men and 2.8–12.7mg/dl for women. Precision was 1.5%. Insulin was measured by direct chemiluminescence-based immunoassay (Liaison, DiaSorin, Saluggia, Italy). Insulin sensitivity was 0.5μIU/ml, while the intra- and inter-assay precisions were 3.9% and 4.3%, respectively. The HOMA-IR index was calculated for non-diabetic patients according to the Matthews formula18: HOMA-IR=glucose (mmol/l)×insulin (μU/ml)/22.5.

The data are expressed as the mean±the standard deviation (SD) for continuous variables of normal distribution or as the median and the 25th and 75th percentiles for non-parametric distribution variables, or as absolute frequencies and percentages for qualitative variables. Considering the baseline analysis of the data, qualitative variables among 2 or more groups were compared by using a chi-squared test or Fisher's exact test, depending on the data distribution. Quantitative variables between 2 groups were compared by using the Mann–Whitney U test or a Student's t-test, depending on the data distribution. Levene's test was used to compare response variability by group. Spearman's rank correlation coefficient was used to analyse the baseline linear association of 2 continuous variables. To evaluate the degree of agreement between 2 variables that evaluated the same concept, Pearson's bivariate correlations were used.

For the longitudinal analysis of the laboratory parameters studied based on 2 time points (baseline and yearly), a stratified analysis by modality and incidence of the response variable was performed first. The effect of the method, the incidence or prevalence and the interaction between them was estimated, and the adjustment of the model was considered by using generalised estimating equations (GEE) for longitudinal data. In order to analyse any association among the different laboratory parameters studied and the independent variables analysed (anthropometric and body-composition parameters estimated by BIA and DXA), the effect was estimated by using the Score Statistics For Type 3 GEE Analysis. Spearman's correlations were used to analyse the association of stratified variables, for which the sample size was small. The same study was performed for the analysis of the increase (delta) or the difference between the final and baseline values, although in this case the model was adjusted with a general linear model.

All statistical tests were considered two-sided, while significant values were considered those where p was less than 0.05. The data were analysed using SAS 9.3 (SAS Institute Inc., Cary, NC, USA).

The main clinical, laboratory and anthropometric characteristics of both groups are presented in Table 1. HD patients had a higher TSF and a lower MUAMA than PD patients. In terms of lab parameters, there were no differences between the groups, except for higher albumin and lower cholesterol in HD patients, despite the fact that these patients had been on dialysis for more time than PD patients.

HD patients had a lower phase angle and higher resistance than PD patients. In addition, they had a higher percentage of BFM, and less muscle mass (MM) and LBM. HD patients showed a lower percent of body-water: the differences in total body water (TBW), extracellular fluid (ECF) and intracellular fluid (ICF) were statistically significant. Regarding DXA parameters, HD patients had a higher percentage of fat in the trunk as compared with PD patients.

In the longitudinal study performed in 18 HD patients, there was an increase in mid-upper arm circumference (MUAC) and MUAMA (Table 2). No significant differences were found in the laboratory parameters, except for cholesterol. An increase in phase angle and ICF, as well as the ratio between the percentage of android/gynoid (A/G) distribution fat measured by DXA (cardiovascular risk indicator), was observed. There was also a decrease in the percentages of MM and ECF.

Given that gender affects body composition, Table 3 shows the same data, but separated by gender. In men, there was an increase in MUAMA per year. No statistically significant changes were seen in body composition as estimated by DXA, except for the increase in the A/G fat ratio. In women, no significant differences were found in the anthropometric, laboratory or body-composition parameters measured by BIA. Although there was a trend towards an increase in fat in all the parameters studied (Fig. 1), only the increase in the A/G fat ratio was statistically significant.

Fig. 1.

Evolution of the percentage of fat mass, estimated by DXA, over one year in the 18 HD patients. The graph below shows the difference of this change (the increase in BFM is represented above the y-axis and the decrease, below).

(0.14MB).

Baseline anthropometric or laboratory parameters in men and women were not significantly different, except for albumin and adiponectin values (Table 3). Significant differences were found in almost all the body-composition parameters measured by BIA and in most measured by DXA. Contrary to what it was expected, men had a higher percentage of BFM as compared with women at baseline; this difference was no longer present after one year of follow up because of a trend towards a gain in BFM (especially in women), (Fig. 2). DEXA showed that women had a statistically significant increase in the percentage of BFM at all levels (trunk, android and gynoid); however the A/G ratio was higher in men and it continued to increase during the year of follow-up (Fig. 2).

Fig. 2.

Evolution of the A/G body-fat mass ratio in abdominal tissue, estimated by DXA, over one year in the 18 HD patients. The graph below shows the difference of this change (the increase in A/G ratio is represented above the y-axis and the decrease, below).

(0.11MB).

Men and women showed no significant differences in baseline leptin and adiponectin values, nor at one year of follow up. This is despite the trend towards an increase in leptin in women and in adiponectin in men. A statistically significant relationship was found between leptin and adiponectin concentrations, with both the percentage of fat mass measured by BIA and abdominal fat estimated by DXA (Table 4). In the models estimated, a direct relationship was found between the percent of BFM (measured by BIA or DXA) and leptin levels, as well as an inverse relationship between the percent of BFM and adiponectin levels. In both cases the relationships were statistically significant.

In addition, increment in these adipocytokines correlated with the percentage change in BFM (Table 5). There was a statistically significant correlation between the baseline A/G fat ratio and leptin levels, as well as an inverse relationship – also statistically significant – between the baseline A/G fat ratio and adiponectin levels corrected for BMI. This later correlation was also observed after a year of follow-up. However such a correlation was not observed with leptin. There were no significant differences in the longitudinal study in serum FGF-21 concentrations in the overall population or in the subgroups by gender.

Regarding the second objective of our study, a significant correlation was found between the fat-mass parameters quantified by BIA and DXA (Table 6), thus indicating a good concordance between both methods of body-composition measurement.