Study design
The CORPOS study (French acronym for “corps” [body] and “os” [bone]) is a prospective longitudinal study in which candidates for first kidney transplantation were tested at the time of listing (baseline), 12 months before kidney transplantation (waiting period) and on months 1, 3, 6, 12 and 24 post-kidney transplantation (Fig. 1). Patient eligibility criteria were described previously [9]. The main inclusion criteria were patients on dialysis aged 18–65 years waiting for first kidney transplantation from a deceased donor. Enrolment was performed between August 2007 and January 2010. KT occurred between 2008 and December 2011. End of follow-up for the last kidney transplant patient occurred in December 2013. The hospital ethics committee approved the study (Comité de Protection des Personnes de Bordeaux A, number 2006/28, June 2006) and all subjects provided informed consent.
At the time of transplantation, the immunosuppressive regimen included induction therapy by either polyclonal anti-thymocyte globulin (in anti-HLA-sensitised patients) or monoclonal anti-CD25 antibody. The maintenance immunosuppressive regimen included calcineurin inhibitors (tacrolimus or cyclosporine in cases of medical intolerance to tacrolimus) and mycophenolate mofetil. Tacrolimus trough levels were maintained between 8 and 10 ng/mL during the first 12 months after transplantation and later decreased to 6–8 ng/mL. Prednisone was stopped on day 7 in non-sensitised patients, except in whose who had received corticosteroid treatment before transplantation.
Biochemical analysis
All measurements were performed after an overnight fast (before starting dialysis on a midweek dialysis day during the waiting period before KT). Serum total proteins and creatinine were assessed by standard techniques. Albumin was measured using the green bromocresol technique, prealbumin and C-reactive protein using an immunoturbimetric test Dietary recall.
Mean total energy and protein intakes were assessed from the average of 3-day food records. Patients wrote down everything eaten, including food portion sizes, followed by an interview with a specialised dietician to ensure accuracy of reporting. Calculations were performed with a nutrient analysis program (Bilnut 4.0; SCDA Nutrisoft, Le Hallier, Cerelles, France).
Whole-body composition analysis
Anthropometry: patients wearing light clothing and no shoes were weighted twice to the nearest 0.1 kg using a precision scale. Height measurement was done against a vertical wooden traditional height gauge with a horizontal pointer (Robé Medical, Remiremont, Grand-Est, France). To obtain precise measurement, standing straight patients were heighted without shoes and with feet together.
DXA: The fan-beam dual-energy X-ray absorptiometer (DXA) scanner (QDR 4500A; Hologic, Bedford, MA, USA) was used to assess whole body composition. During the waiting time before transplantation, measurements were performed on the day after dialysis using the weight measured on the same day. An empty peritoneal cavity was required for patients on peritoneal dialysis. Patient positioned was standardised. The DXA scans were acquired with the APEX 2 version 8.26 for QDR 4500A software version (classic calibration). Before 2009, the classic calibration method was used for the body composition (fat and lean soft tissue) calculation [10]. After 2009, a new software was developed with an upgrading of the APEX 2 version: the NHANES calibration provides a slightly different approach for body composition calibration [11]. The scans acquired after 2009 were re-analysed using the first version to ensure consistency of data and avoid measurement bias.
The values of whole-body composition are expressed as fat mass (FM) and fat-free mass (FFM) in absolute values, and as fat mass index (FMi, kg/m2) [FM indexed for height squared] and fat-free mass index (FFMi, kg/m2) [FFM and indexed for height squared].
Bioimpedance spectroscopy: Whole-body bioimpedance spectroscopy (BIS) was performed using a multifrequency device (Imp SFB7; ImpediMed, Pinkenba, QLD, Australia). Measurements were performed just after DXA evaluation, with a 10-min rest in the supine position. The electrodes and recording pads were placed on the non-assessment side of the body in patients with an arm access. In those with a central catheter or for patients on peritoneal dialysis, the dominant arm was chosen. The device scans 256 frequencies between 4 kHz and 1000 kHz for estimation of body composition, and it utilises Cole modelling with Hanai mixture theory to determine total body water (TBW), extracellular water (ECW) and intracellular water (ICW) [12].
BCM is defined as follows:
$$ \mathrm{BCM}\ \left(\mathrm{kg}\right)=\mathrm{FFM}\ {\left(\mathrm{kg}\right)}_{\mathrm{DXA}}-\mathrm{ECW}{\left(\mathrm{l}\right)}_{\mathrm{BIS}} $$
where DXA and BIS are performed on the same day.
Physical activity level
Physical activity was estimated using the French version of the Baecke Physical Activity questionnaire [13, 14]. The self-perception of duration and intensity of three components of physical activity (work, leisure time and sport). leads to the calculation of scores. The sum of the scores obtained for these components leads to the global physical activity level for each patient.
Statistical analysis
Statistical analysis was performed using SAS (ver. 9.2; SAS Institute Inc., Cary, NC, USA). The results are expressed as the median [interquartile range] for continuous variables and frequency for categorical variables. Patient characteristics at the time of listing (inclusion) were compared between transplanted and non-transplanted patients by Wilcoxon’s test for continuous variables and the χ2 test or exact Fisher’s test for categorical variables. A linear mixed regression model, with a random intercept and random slopes before and 1 month after KT, was used to examine the factors related to changes in BCM. The effects of each variable of interest were tested on the intercept (difference at KT) and each slope. Univariable analyses took into account the effects of only one variable; while multivariable analyses took into account all variables in the same model.