Icodextrin Improves the Fluid Status of Peritoneal Dialysis Patients: Results of a Double-Blind Randomized Controlled Trial

Study Design

The study had a prospective, multicenter, controlled, randomized, double-blind design. The goal was to recruit 50 patients from centers in Germany, Sweden, and the United Kingdom, with the aim of 40 patients completing at least 3 mo of the study. After giving consent to participate, patients not receiving 1.5 to 2.5 L of 2.27% glucose (Dianeal; Baxter) for the long dwell entered a 1-mo screening period, during which the long-dwell solution was 2.27% glucose (Dianeal). Those already receiving 2.27% glucose for the long dwell entered a 2-wk baseline period, during which the long-dwell prescription was unchanged.

At the end of the baseline period, patients were randomized 1:1, with stratification for center/country, dialysis modality (CAPD or APD), and presence of cardiovascular disease, defined as previous myocardial infarction or cerebral stroke, angina pectoris, or LVH. Randomization was to either 7.5% icodextrin or 2.27% glucose for the long-dwell exchange, with the length and fill volume of the long exchange during the treatment phase being the same as that in the baseline period. After randomization, further study assessment visits were at 1, 3, and 6 mo, with statistical analysis of variables primarily based on change of value from baseline after 3 mo of treatment. During the treatment period, patients were assessed by body weight, multifrequency bioelectrical impedance analysis, and deuterium oxide dilution (3 and 6 mo only). In addition, BP was assessed by 24-h ambulatory monitoring. Blood, urine, and dialysate samples were collected for assessment of electrolytes, high-resolution C-reactive protein, total cholesterol and triglycerides, dialysis adequacy, peritoneal membrane transport characteristics, and residual renal function.

After commencement of the study product, patient medications, including antihypertensive drugs, could be altered according to clinical need, other than changes in diuretics. Changes could also be made as necessary to short-dwell fill volume glucose concentrations.

Inclusion and Exclusion Criteria

The principal inclusion criteria for the study were (1) either untreated hypertension (BP > 140/90 mmHg), treated hypertension, or a dialysis prescription with a daily average glucose concentration of 2.27% or greater, (b) high or high-average peritoneal solute transport (corrected 4 h D/P creatinine ratio ≥0.65, and (3) urine output ≤750 ml/d. Patients had to tolerate a dialysis regime that included a long dwell of ≥6 h with 2.27% glucose with fill volume of 1.5 to 2.5 L, as demonstrated in the screening period. All patients were over 18 yr of age, were able to give written informed consent, and had to have been on PD for at least 90 d.

Patients were excluded if they had received icodextrin in the 30 d before randomization, used other nonglucose solutions in the 30 d before randomization, had been treated for peritonitis in the 30 d before randomization, were considered noncompliant, or were considered to have hypertension despite being clinically volume depleted (although in practical terms, while this was in the protocol, it did not lead to exclusions), used a 1.36% glucose concentration for all exchanges, were allergic to starch, had a glycogen storage disease, had a life expectancy less than 12 mo, had a serious illness or injury in the 30 d before randomization that would invalidate study entry, were participating in another interventional study, were pregnant or lactating, or had a significant psychiatric disorder that would interfere with their ability to provide informed consent and/or comply with the study procedures.

The identity of the long-dwell solution was blinded to patients, investigators, and clinical monitors; specially created packaging was used to conceal which solution was which. The treatment codes were supplied to study sites in sealed envelopes, which were checked at the end of the study. Approval for the study was granted by the local research ethics committees of all centers, and all subjects provided written informed consent.

Statistical Analyses

Analysis of outcome variables was by ANCOVA for changes from baseline with the patient’s baseline value as covariate. Analysis of baseline variables for differences between treatment groups was by ANOVA for continuous variables, with χ² test or Fisher’s exact test for categorical variables.

Changes with time were analyzed in two ways: a between-group comparison of the change from baseline by unpaired t tests (or Mann-Whitney U test for nonparametric variables), and a within-group comparison of change from baseline using paired statistics allowing for repeated measures. The intent-to-treat analysis is presented, although an analysis of assessable patients was also undertaken. The assessable population was the subset of intent-to-treat group who completed the baseline period and at least 3 mo of the treatment period, with exclusions of patients who were off the investigational product for a cumulative time of more than 30 d, did not receive PD for more than 30 cumulative days, changed diuretic medication (except withdrawal of diuretics in those with urine output <300 ml/24 h), and those changing dialysis category. The outcome of these analyses did not differ.

The sample size was based on previous experience of trials comparing icodextrin with 2.27% glucose. These calculations indicated that with a type 1 error of 0.05 and power of 0.80, a minimum of 19 patients per treatment group would have to complete the study to detect a treatment weight difference of 1.6 kg or greater change from baseline.

Clinical and Laboratory Procedures

The simplified Standardized Permeability Analysis test, which used a 3.86% glucose dwell, was used to assess peritoneal membrane transport characteristics, with collection of dialysate samples at 0, 1, 2, and 4 h and a plasma sample within 1 h. Twenty-four-hour ambulatory BP was measured with a Spacelabs 90207 monitor; readings were performed at hourly intervals. Mean 24-h systolic and diastolic BP were determined, as were mean systolic and diastolic BP for night (23:00 to 07:00 h) and day (07:00 to 23:00 h).

Multiple-frequency bioelectrical impedance analysis was performed with the Hydra analyzer (Xitron Technologies, San Diego, CA). Measurements were performed by means of the standard tetrapolar technique, with electrodes placed on the dorsum of wrist and anterior aspect of the ankle on the left side of the body. The patient was supine for at least 10 min before measurements with dialysis fluid present were performed. Three consecutive measurements were performed over a 2-min period, with recording of values for total body water (TBW), ECF, and intracellular fluid volume, which are determined by the analyzer by a bioelectrical impedance spectroscopy method (19). The coefficient of variation, determined from readings taken a month apart in the screening phase of the study, was 5.3% for TBW and 5.8% for ECF.

TBW was also estimated by deuterium oxide dilution. Patients drank 4 g of deuterium oxide (²H₂O) with 100 ml of tap water. Blood samples were drawn before ingestion and after 2 and 4 h. TBW was determined from isotope enrichment in plasma via isotope ratio mass spectrometry (20), with a coefficient of variation of 5.4% in the control limb of the study between baseline and 3 mo.

Demographics

Of 57 patients screened for inclusion in the study, 50 (88%) met the criteria for randomization. All were using lactate-buffered glucose-based dialysis solutions (Dianeal). Exclusions were the result of lower-than-required solute transport in four patients and excess residual urine volume in three. There was some inequality in the number of patients randomized to each group as a result of the block randomization, but the baseline characteristics of the groups were not significantly different from each other (Table 1). The number of patients remaining in the study at each time point is summarized in Figure 1. Three withdrawals from each group were associated with clinical adverse events: peritonitis and catheter leak (one from each group, respectively), perianal abscess (icodextrin), and pancreatitis (Dianeal). Additional withdrawals from the 2.27% glucose group were for ultrafiltration failure and patient preference. Peritonitis rates and vital signs (pulse, respiration, and body temperature) did not differ between groups throughout the study. No adverse events attributable to icodextrin (e.g., skin rash) were experienced.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Summary of patient flow through the study, with numbers at each time point.

Changes in Achieved Ultrafiltration, Sodium Removal, and Residual Urine Volume

Achieved ultrafiltration, sodium removal, and residual urine volume are summarized in Table 2 as change from baseline according to treatment group. Statistically significant between-group differences were observed at 3 mo in the achieved ultrafiltration and total fluid losses as a result of maintained ultrafiltration in the icodextrin group and a decline in ultrafiltration in the 2.27% glucose controls. Urine volume was relatively better maintained in the icodextrin group, especially at 6 mo, although this did not quite reach statistical significance. If all patients were included in the analysis, when nonparametric statistics are used, the between-group difference at 6 mo was 89 ml (P = 0.039). Absolute sodium losses were consistently and significantly less in the glucose-treated group throughout the study, again the result of a decline in sodium losses in the control group. Plasma albumin, often considered a marker of hydration status in PD patients, diverged during the study, but the differences did not quite reach statistical significance. There was a negative relationship between the change in albumin and ECF-BIA at 3 mo from baseline (r = −0.44, P = 0.004), which was absent at 6 mo.

Changes in Body Composition

Significant between-group differences in the longitudinal change in the primary end point, drained body weight, were observed, such that patients randomized to icodextrin lost weight early, whereas those using 2.27% glucose steadily gained weight throughout the study (Figure 2a). The between-group differences were 1.45 kg (P = 0.015) at 1 mo, 1.67 kg (P = 0.026) at 3 mo, and 2.3 kg (P = 0.036) at 6 mo. These differences were reflected in changes in the TBW as estimated by multifrequency bioimpedance following a similar, but not identical, pattern. The continued increase in weight in the 2.27% glucose group was not reflected in a further increase of body water at 6 mo (Figure 3a). Between-group differences were 1.7 kg (P = 0.006), 1.53 kg (P = 0.003), and 1.39 kg (P = 0.036) at 1, 3, and 6 mo, respectively. These differences were largely accounted for by changes in the extracellular fluid component as estimated by bioimpedance, being 1.06 kg (P = 0.008), 0.85 kg (P = 0.035), and 0.82 kg (P = 0.1), respectively (Figure 3b). In each case, there were significant changes from baseline values observed in the icodextrin group, with the exception of drained weight, where both groups changed longitudinally from baseline (Figures 2 and 3⇓).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Changes in (a) drained body weight and (b) total body water determined from deuterium dilution. At each time point, values represent mean ± SEM change from baseline for patients randomized to icodextrin (□) or 2.27% glucose (▪). Between-group differences, † P < 0.05, ‡ P < 0.04. Longitudinal differences from baseline, * P < 0.04, ** P < 0.001.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Changes in (a) total body water and (b) extracellular fluid determined from bioelectrical impedance. At each time point, values represent mean ± SEM change from baseline for patients randomized to icodextrin (□) or 2.27% glucose (▪). Between-group differences, † P < 0.04, ‡ P < 0.008. Longitudinal differences from baseline, * P < 0.002, ** P < 0.001.

Significant changes in TBW, measured independently by deuterium dilution, also occurred (Figure 2b). The change in ECF estimated from bioimpedance and the change in TBW using deuterium dilution correlated at 3 mo (r = 0.37, P = 0.02) and at 6 mo (r = 0.58, P < 0.001).

BP Control, Antihypertensive Medication, Lipids, and C-Reactive Protein

There were no significant differences in the mean and median 24-h BP readings between the treatment groups at any time point. The number of patients receiving antihypertensive drugs was similar in both groups at the start of the study (icodextrin 87%; Dianeal 81%). Patients randomized to icodextrin were more likely to reduce their hypertensive medication (n = 9, 33%) compared with the dianeal group (n = 3, 14%), but they were also more likely to increase their medication (n = 5, 19%, versus n = 2, 10%), and the multiple changes in some patients made this difficult to analyze. There was no correlation between longitudinal changes in BP and changes in weight, achieved ultrafiltration, or extracellular fluid as estimated from BIA. There were no between-group changes from baseline in the total cholesterol, triglycerides, or C-reactive protein measurements.