The cohort studied in this paper was enrolled in a randomized trial of correction of anemia with erythropoietin (
[9]). The design, methods, randomization, and epoetin dosing protocols have been reported previously (
[9]). Patients were randomly assigned to one of the following hemoglobin (Hb) targets: 9.5 to 11.5 (low target) or 13.5 to 14.5 g/dl (high target). Patients were masked to treatment assignment, as were their doctors. Mean Hb levels at the end of the initial 24-week titration phase were 13.3 and 10.9 g/dl, respectively. During the maintenance phase, from weeks 24 through 96, corresponding mean Hb levels were 13.1 and 10.8 g/dl. Cardiac structure constituted the primary study outcome, and no difference was observed between the two groups (
[9]). Consequently, for the purpose of this article, all patients were included as one cohort. The study was centrally coordinated from St. John’s, Canada, for Canadian patients and Manchester, England, For European patients.
Local independent ethics committees or institutional review boards approved the study protocol form before study initiation. The study was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. All patients gave informed consent before study enrollment.
Study population
Inclusion criteria were as follows: Age ≥ 18 years; inception of maintenance HD within the previous 3 to 18 months; predialysis Hb level between 8 and 12 g/dl; LV volume index (LVVI) < 100 ml/m2, and predialysis diastolic BP (DBP) < 100 mmHg. Exclusion criteria were as follows: Clinical evidence or history of symptomatic cardiac failure or ischemic heart disease; daily prednisone dosages ≥ 10 mg; medical conditions that are likely to reduce epoetin responsiveness, including uncorrected iron deficiency; concurrent malignancy; blood transfusion in the preceding month; therapy with cytotoxic agents; seizure in the preceding year; hypersensitivity to intravenous iron; and current pregnancy or breastfeeding.
Description of study procedures
Laboratory tests were measured centrally by Quest Diagnostics (Van Nuys, CA, and Heston, UK). With the high target, the treatment goal was increments of 0.5 to 1.0 g/dl every 2 weeks, until achieving stability between 13.5 and 14.5 g/dl. Other treatment goals included predialysis diastolic blood pressure between 70 and 90 mmHg, urea reduction ratio > 67%, and transferrin saturation ≥ 20%. On a weekly basis, midweek predialysis blood pressure levels were communicated weekly to and treatment recommendations sent from the coordinating center. These were determined by the protocols in use at the individual treating centers, and a single, formal, standardized method was not imposed. The last patient completed the study in May 2003.
Quality of life was assessed using the KDQOL and FACIT fatigue questionnaires (
[10, 11]). We pre-specified the domains of HRQOL for this analysis: physical function and vitality measured using the SF-36 questionnaire and fatigue using the FACIT fatigue questionnaire. Higher scores reflect better health. These were also the pre-specified domains for HRQOL outcomes in the trial (
[5]). A-priori we defined short-term change in HRQOL as the change from baseline to the assessment at 24 weeks, and long-term change as the change from baseline to the assessments made at 48 and 96 weeks.
There was no difference in the clinical or demographic characteristics or in treatment assignment of those who had only one KDQOL assessment n = (
[12]) and those who had serial assessments (n = 484). At each assessment, > 90% of patients remaining in the study completed the questionnaire.
Sample size estimate
The sample size needed to detect a 15% difference between treatment groups in the primary outcome (left ventricular cavity volume index) was calculated as 166 per treatment group, given a two-tailed significance of 0.05, a power of 0.90, and an SD of the percentage change in left ventricular cavity volume index of 42% (
[9]). With an expected dropout rate of 40%, primarily as a result of transplantation, 277 patients were required for each treatment group.
At baseline SF-36 was completed in 457 patients and FACIT fatigue in 572 patients. The comparable numbers at 24 weeks were 443 and 499, at 48 weeks 435 and 432, and at 96 weeks 335 and 325.
Risk factors
In addition to demographic factors (age, gender, and race), baseline clinical characteristics (diabetes, primary renal disease, dialysis vintage, blood pressure, body mass index, type of vascular access) were recorded, as were conventional laboratory tests results (Hb, white cell count, percentage of iron saturation, urea reduction ratio, and serum albumin) and other serologic tests at baseline. Serum concentrations of N terminal pro-B type natriuretic peptide (NT-proBNP) and cardiac troponin T were measured in 2009 using diagnostic kits and performed on an Elecsys 2010 immunochemistry analyzer (Roche Diagnostics, Montreal, Quebec, Canada). Serum high-sensitivity C-reactive protein was measured using CRP reagent (Beckman Coulter, Fullerton, CA) and performed on an IMMAGE Immunochemistry system (Beckman Coulter). Serum IL-6 was measured using IL-6 kits and performed on a Unicel DxI 800 Access Immunoassay system (Beckman Coulter). Serum leptin was measured by ELISA technique using kits purchased from Diagnostic Systems Laboratories, Inc (Webster, Tx).
Statistical analysis
Baseline characteristics were described by number (%) or by median with interquartile range. For the cardiac and inflammatory biomarkers high levels were pre-specified as above the 75th percentile. Univariate analysis of the association between baseline variables and baseline HRQOL scores and change in scores over time was undertaken using linear regression. Multivariate models were created using multiple linear regression, and it was pre-specified that these include the variables which were associated with the outcome at a p < 0.05 in the univariate regression. In addition multivariate models were created to determine the effect of troponin T levels on change in HRQOL independent of age, sex, diabetes and baseline HRQOL score.