The balANZ Study—Strengthening the Evidence for Neutral-pH Solutions Low in Glucose Degradation Products

In the area of dialysis, the track record of positive clinical trials is not strong, presenting considerable difficulty when it come to providing Grade A evidence for practice-based decisions. Still, well-conducted but negative studies make an important contribution to advancing clinical care (1). Of course, one of the difficulties in trial design and delivery is that clinical practice patterns vary between centers in an unmeasured way. In peritoneal dialysis (PD), those variations might be the approach to patient training or the use of hypertonic exchanges.

Against that background, the impact of any new intervention has to be sufficiently powerful to avoid being lost in the variability of practice. A key example is the marked between-center variation in peritonitis rates (2-4). Equally, it is important that the renal community have a good understanding of the outcomes that matter to patients, and to that end, the role of patient partners in trial design is being increasingly recognized (5) and can identify potential barriers to recruitment. The study of biomarkers might provide insights into the mechanisms of disease, but robust clinical endpoints matter to patients, their health care teams, and planners of health services. At the same time, cost-effective delivery of health care is very much in the spotlight, and thus it is important that evaluations include economic analyses, which will vary according to local factors.

Peritoneal dialysis makes an important contribution to renal replacement therapy, but its use is associated with well-recognized complications and relatively short technique retention—the median time on PD being 2 years in the 2010 UK Renal Registry Report (6). The commonest causes of technique failure include peritonitis and an inability to achieve adequate volume and solute clearance (which can be compounded by loss of residual renal function and changes to the peritoneal membrane). Several components of traditional PD solutions—including acidic pH, high concentrations of glucose and glucose degradation products (GDPs), raised osmolality, and lactate buffer (7,8)—have toxic effects both in vitro and in vivo and may potentially be responsible for some of the membrane changes seen in long-term PD (9,10).

In response, several neutral-pH, low-GDP (“biocompatible”) solutions have been developed, and the impacts of those solutions on a range of dialysate biomarkers that may indicate enhanced peritoneal integrity or reduced inflammation have been explored (8,11). Proving that robust clinical benefits are associated with the use of the new solutions has been difficult (12), and a recent review commented that studies in the area have often been underpowered, of short duration, and focused on intermediate endpoints (13). Complicating matters further, the modern solutions are not identical with respect to their concentrations of GDPs (11). The 2010 UK Renal Clinical Guideline concluded that, although available evidence [GRADE rating 2B (14)] supported the use of “biocompatible” PD solutions for patients experiencing infusion pain, information was insufficient to recommend that all PD patients be treated with such solutions (15). Has that situation now changed?

THE BALANZ STUDY

The balANZ trial was conducted across 16 centers in Australia, New Zealand, and Singapore between 2004 and 2010. To date, it is the largest multicenter randomized controlled trial to have compared a low-GPD, neutral-pH peritoneal dialysate with a standard dialysate. Three publications from the trial describe (a) the study design and key endpoints (16), (b) changes in peritoneal membrane function (17), and in this issue of Peritoneal Dialysis International, (c) the impact on peritonitis rates and microbiology.

Incident patients who had a residual glomerular filtration rate of 5 mL or more per minute and 24-hour urine output of 400 mL or more were randomly allocated to receive either neutral-pH, lactate-buffered, low-GDP Balance solution (Fresenius Medical Care, Bad Homburg, Germany) or conventional, pH 5.5, lactate-buffered Stay·Safe solution (Fresenius Medical Care). They were then followed for 2 years.

The aim was to recruit a total of 336 patients (168 in each group) with a calculated 80% power to detect a clinically important difference in the slope of residual renal function over time (0.067 mL/min per 1.73 m² per month). Because of recruitment difficulties, only 55% of the target patients (185) were randomized into the study (92 to Balance and 93 to Stay·Safe), with the groups being well matched for baseline characteristics. Although the primary outcome measure was not met, a significant difference was observed between the groups in time to anuria (p = 0.009), time to first peritonitis episode (p = 0.01), and slope of the change in 4-hour dialysate-to-plasma ratio of creatinine (4-h D/P Cr) over the course of the study.

In the Balance group, 6 patients (7%) became anuric during the course of the study, compared with 18 (20%) in the control group. After adjustment for baseline glomerular filtration rate, diabetic nephropathy, and use of automated PD, Balance was associated with a hazard ratio of 0.36 (95% confidence interval: 0.13 to 0.96) of developing anuria. Peritoneal small-solute clearance and ultrafiltration (UF) were measured at 3, 6, 9, 12, 18, and 24 months, and 90% of patients had undergone peritoneal equilibration tests (17). The 24-hour UF volumes were significantly lower at 3 and 6 months in the Balance group than in the Stay·Safe group, with the difference being approximately 300 mL. In addition, 24-hour urine volumes differed by a similar amount, being greater in the Balance group than in the Stay·Safe group. Over the 2-year study period, peritoneal UF increased significantly in the Balance group, but remained stable in control patients (difference: 24 mL daily per month; 95% confidence interval: 9 mL to 39 mL; p = 0.002), and by the end of the study, 24-hour UF did not differ between the groups. Equally, no differences were observed between the groups in peritoneal small-solute clearances, prescribed dialysate fill volumes, or peritoneal glucose exposure. Hydration status was not directly measured, but importantly, the groups showed no differences in terms of weight, blood pressure, hemoglobin, serum sodium, or albumin.

Mean 4-h D/P Cr at 1 month was significantly higher in the Balance group than in control patients (0.67 ± 0.10 vs 0.62 ± 0.10, p = 0.002). Over the 2-year study period, mean 4-h D/P Cr measurements remained stable in the Balance group, but increased significantly in control patients. By the end of the study, those measurements were 0.67 ± 0.09 in the Balance group and 0.70 ± 0.08 in the control group. The stability of the 4-h D/P Cr measurements in the Balance group can be taken as evidence of relative preservation of the peritoneal membrane in that group. However, residual renal function was well preserved in the study, with the median urine volume at study end being 814 mL in the Balance group and 699 mL in the control group. Glucose exposure was modest. It would clearly be important to study the impact of neutral-pH low-GDP solutions on membrane function in anuric patients. For example, at the start of the European Automated Peritoneal Dialysis Outcomes Study, anuric patients had a much higher 4-h D/P Cr ratio (0.75) than the ratios seen at the end of balANZ (18).

The peritonitis rate was 0.30 episodes per year in the biocompatible group compared with 0.49 episodes per year in control patients (38 events vs 67 events, p = 0.01), and in terms of causative organisms, statistical significance was seen for non-pseudomonal gram-negative peritonitis. In addition to the reduction in the overall rate of peritonitis, a reduction in physician-scored peritonitis severity was also observed, with 37% of episodes in the biocompatible group being rated “mild” compared with 10% of episodes in the control group (p = 0.001). The median duration of peritonitis-associated hospitalization was also shorter in the biocompatible group (6 days vs 11 days, p = 0.05). There were no differences between the groups with respect to hospitalization for peritonitis, transfer to hemodialysis, or patient death. Intriguingly, a significant reduction in overall infections was observed in the biocompatible group (4 non-PD infections occurred in 91 patients in the intervention group vs 20 non-PD infections in 91 patients in the control group).

CAVEATS

As usual with studies of this kind, there are several caveats. The main concern was that the recruitment target was not met, and because the study design was not based around the secondary endpoints, such analyses need to be viewed with caution and not over-interpreted. Barriers to recruitment for balANZ were summarized as clinician bias either for or against biocompatible dialysate, contractural commitments to alternative suppliers, and insufficient research staff. In the United Kingdom in recent years, considerable emphasis has been applied by the National Institute for Health Research on the importance of recruitment, with active management being directed at portfolio studies that are falling behind. Over the last year, the consequence has been an increase to 48% from 29% in renal studies that have recruited to time and target. Ultimately, important clinical questions concerning optimal treatment will not be able to be resolved for patients unless well-designed and sufficiently powered studies are able to recruit effectively. The strong collaboration, clear prioritization of objectives, and adequate funding that will be required are a responsibility for the whole renal community.

By its nature, balANZ was an open-label study, which increased the risk for observer bias. The drop-out rate from the study was 45% by the end of 2 years, and although that drop-out was equally distributed between the groups, it may have resulted in informative censoring. Complex statistical techniques were used to minimize that risk. An unusual aspect of the clinical practice was that fewer than 10% of patients were using automated PD and icodextrin use was also very low. Icodextrin use is associated with clear clinical benefits with respect to UF (19) and metabolic parameters (20), but its signal with respect to inflammatory markers may not be entirely benign (21). It is not clear that the results of balANZ would apply equally to neutral-pH solutions in which the concentration of GDPs may be higher, and that argument has been used to explain the lack of an impact on either residual renal function or peritonitis rates from the recently reported single-center UK study (12). It is also difficult to evaluate the relative benefit of using Balance in comparison with other practice patterns associated with reduced peritonitis rates.

IMPLICATIONS FOR PRACTICE

BalANZ has to be considered in light of the studies that have gone before, in which the value of biocompatible solutions has not clearly been demonstrated. It gives a signal of benefit for the low-GDP, pH-neutral solution with respect to important outcomes in patients on PD— time to anuria, peritonitis rates, severity of peritonitis when it does occur, and stability of peritoneal membrane function. It is not the first study to report preservation of residual renal function with a biocompatible solution, and although the mechanism has not been worked out, possibilities could include a direct nephrotoxic effect of GDPs or a difference in volume status related to an early reduction in UF or, alternatively, to a reduced peritonitis rate.

Progress in medicine is incremental, and although balANZ has not answered all the outstanding questions on the impact of solution design on patient outcomes in PD, it has provided information that is relevant to patients and their clinical teams. Here, it is tempting to draw an analogy from the British track cycling team: The team’s director, David Brailsford, recently emphasized the value of combining marginal gains as the key to success (22). Going back to the GRADE evidence system, it could be argued that balANZ provides 1B evidence for the use of low-GDP, neutral-pH solutions. The clinical implication would be that this is a strong recommendation that applies to most patients (14).

DISCLOSURES

MW has received speaker’s honoraria from Gambro Lundia, Baxter Healthcare, and Fresenius Medical Care, and has participated in clinical trials with Baxter Healthcare and Fresenius Medical Care.