Basically, Dialysis Is Not Good Enough

In CKD patients, the daily imbalance between endogenous acid production and renal acid excretion gradually leads to worsening metabolic acidosis. Although serum bicarbonate typically reaches a stable nadir, there is accumulating evidence that mild metabolic acidosis in late-stage CKD may cause adverse clinical outcomes. Specifically, several recent trials have focused on the relationship between metabolic acidosis and the rate of progression of CKD. The randomized controlled trial conducted by de Brito-Ashurst et al. demonstrated that sodium bicarbonate therapy slowed renal deterioration and improved nutritional outcomes (1). Another randomized trial comparing sodium bicarbonate to either sodium chloride or placebo showed that sodium bicarbonate slowed the decline in kidney function (2). Along with other supportive animal data indicating that alkalinization is renoprotective (3), there is now reasonable evidence that correction of metabolic acidosis may preserve residual kidney function.

Because residual kidney function is an important predictor of survival in both hemodialysis (HD) and peritoneal dialysis (PD) patients, these data may also be germane to the ESRD population. Indeed, several small randomized controlled trials have supported the hypothesis that metabolic acidosis is deleterious to patients on dialysis as well as to those with earlier stages of CKD. In one study, Szeto et al. randomized PD patients with metabolic acidosis to treatment with either sodium bicarbonate tablets or placebo (4). Treatment was well tolerated and was associated with improvement in the subjective global assessment score and normalized protein nitrogen appearance rate. The treatment group also had fewer days of hospitalization. In another study, PD patients were randomized to two different lactate concentrations (35 mmol/L versus 40 mmol/L) in their dialysate. The patients receiving the high lactate dialysate had improved weight gain and arm circumference, as well as fewer hospitalization days (5).

Patients’ choices regarding dialysis modality are dictated by a number of factors, not least among which is the recommendation of their nephrologist. As nephrologists, educating patients about the risks and benefits of different dialysis modalities can be challenging. The major impediment to accurately comparing outcomes of PD and HD is the lack of randomized controlled trials. A large trial in which patients were randomly allocated to either HD or PD would answer many important questions. First, it would demonstrate whether either modality is associated with decreased mortality relative to the other. Importantly, it would also measure other outcomes such as infectious episodes and control of metabolic acidosis. However, because most patients have a firm preference regarding dialysis modality, performing a study of this nature has been extremely challenging due to difficulties with recruitment (6). To date, therefore, comparisons of outcomes obtained with the two dialysis modalities rely upon retrospective analysis of large databases. Whereas a large US cohort from the 1990s demonstrated a modest improvement in mortality during the first 2 years of PD and a late mortality benefit with HD (7), a more recent analysis of patients from the last decade showed equivalent mortality among HD and PD patients in the United States (8).

In this issue of CJASN, Vashistha et al. report interesting findings obtained from a retrospective analysis of patients who dialyzed under the auspices of one the large dialysis organizations in the United States between 2001 and 2006 (9). They examined outcomes of HD and PD patients with particular emphasis on the correction of metabolic acidosis. The large number of patients studied—over 10,000 on PD and 100,000 on HD—lends great strength and external validity to the data.

Unfortunately, the first major finding of this study is disheartening. Whereas patients on PD fared better than those on HD, metabolic acidosis was not well controlled in either patient population. Only 60% of HD patients and 75% of PD patients had time-averaged serum bicarbonate levels of 22 mEq/L or higher. In addition, the failure to correct acidosis is harmful: the authors then demonstrate that in both the HD and PD patient populations, there is an inverse relationship between serum bicarbonate levels and mortality, although it is more robust in HD patients than in PD patients. Mild metabolic acidosis (HCO₃ between 20 and 24 mEq/L) was not associated with increased all-cause mortality in PD patients, whereas it was associated with approximately a 10% risk increase in HD patients. Severe metabolic acidosis (HCO₃ <19 mEq/L) was associated with significant increases in mortality in both patient populations.

It is unclear why inadequate control of metabolic acidosis is so common in dialysis patients. The authors note the lack of clarity in international guidelines regarding serum bicarbonate. In the bone metabolism guidelines, the Kidney Disease Outcomes Quality Initiative does recommend bicarbonate levels >22 mEq/L for patients with CKD stages 3–5 (10). In the United States, the Centers for Medicare and Medicaid Services Quality Incentive Program has begun to partially link reimbursement with quality measures. Although delivered dialysis dose is one measure that will be rewarded, control of metabolic acidosis is not mentioned (11). Given the lack of emphasis on metabolic acidosis, it may not be surprising that a large number of both HD and PD patients do not achieve normal serum bicarbonate levels.

The other central finding by Vashistha et al. concerns the relatively worse outcomes seen on HD compared with PD (9). The authors propose several hypotheses to explain the higher prevalence of metabolic acidosis among HD patients. The higher base concentration of peritoneal dialysate may partially explain the difference. Furthermore, as with most studies comparing PD with HD, the possibility of bias exists. Despite controlling for comorbidities, the patients on PD may have been healthier or less likely to develop metabolic acidosis. Another limitation concerns the difference between the intermittent nature of HD and the more continuous nature of PD. Consequently, the authors compared true steady-state serum bicarbonate levels in PD patients with mean predialysis levels, the lowest obtained during the week, in HD patients. To minimize that discrepancy, and to more closely simulate mean steady-state serum bicarbonate values, the authors performed an analysis in which they added 3 mEq/L to the pre-HD bicarbonate concentration and obtained similar results. Although there are limitations regarding this comparison, this is a potentially novel finding.

In summary, Vashistha et al. highlight a potential deficiency in modern dialytic therapy by documenting inadequate control of metabolic acidosis (9). At present, assessment of dialysis adequacy continues to rely heavily on measurements of small solute clearance. This remains the case despite the fact that in both HD and PD, studies have shown no mortality benefit with increases in small solute clearance beyond those commonly obtained (12,13). In contrast, it appears from this study that improved control of metabolic acidosis does decrease mortality of patients on dialysis. Perhaps it is time to broaden the assessment of dialysis adequacy to include this important parameter; more study is needed.

• Copyright © 2013 by the American Society of Nephrology