Improving Outcomes in Hemodialysis Patients: The Need for Well-Designed Clinical Trials

End-stage renal disease is a catastrophic illness associated with a very high rate of cardiovascular comorbidities and mortality. It is widely known that cardiovascular mortality among those on hemodialysis ranges from 5 to 500 times higher than age-matched normal populations.1 The personal and financial costs of hemodialysis are high and the cost to society is increasing with the rising prevalence and incidence of patients on hemodialysis.2 For these reasons, many investigators have conducted intervention trials designed to reduce cardiovascular morbidity and mortality among dialysis populations. Unfortunately, the interventions employed in such trials have met with little success.

Graded associations between uremia and cardiovascular disease and death suggest that interventions that increase dialysis dose or modulate removal of middle molecules should improve outcome. However, the Hemodialysis (HEMO) study, a large randomized, controlled trial designed to determine if higher dialysis dose or high-flux dialysis membrane use can improve outcomes, failed to do so.3 More recently, in a randomized double-blind, placebo-controlled trial, long-term administration of atorvastatin once daily in a dose of 20 mg failed to improve overall cardiovascular morbidity and mortality among prevalent patients with type 2 diabetes on long-term hemodialysis.4 Further, interventions aimed at full correction of anemia, a common comorbid condition associated with mortality and cardiovascular morbidity among hemodialysis patients, also failed to show an improvement in outcome.5 One promising study, in which 800 IU of vitamin E was compared with placebo in a small population of hemodialysis patients with prior cardiovascular events, demonstrated a benefit of this intervention. However, to date this trial has not been reproduced or confirmed by similar studies.6

Hypertension occurs in more than 80% of patients on hemodialysis and is associated with increased morbidity and mortality, yet no large-scale long-term intervention study has examined the effect of blood pressure–lowering strategies on outcomes in hemodialysis patients. Many studies have demonstrated the benefit of drugs that block the renin-angiotensin system, including in patients with underlying chronic kidney disease.7, 8, 9 However, few studies have examined the potential benefit of these drug classes among hemodialysis patients. The Fosinopril in Dialysis (FOSIDIAL) study examined the effect of fosinopril on cardiovascular morbidity and mortality in an open-label, placebo-controlled trial of approximately 400 patients with left ventricular hypertrophy on long-term hemodialysis. After a 2-year follow-up, there was a trend for a lower rate of cardiovascular events in the fosinopril arm, but the difference was not significant.10 More recently, Takahashi et al reported results from an open-label, randomized, blinded end-point trial in which candesartan 4 to 8 mg once daily dramatically reduced cardiovascular morbidity and mortality risk (absolute risk reduction, 32%). In this study, blood pressure control was not significantly different between treatment groups. Although this was an open-label study, the large effect size provided promising results suggesting that administration of drugs that block the renin-angiotensin system may provide benefit beyond blood pressure–lowering effect.11

In this issue of the American Journal of Kidney Diseases, Suzuki et al report the results of a larger open-label study in hypertensive hemodialysis patients using various angiotensin receptor antagonists.12 In the study, the authors randomly assigned 366 long-term hemodialysis patients to either an angiotensin-receptor blocker (ARB; valsartan, candesartan, or losartan) or no treatment. The dose of the ARB was uptitrated to a specified maximal dose for each drug and patients were followed for 36 months. The primary outcome was fatal and nonfatal cardiovascular events. Event-free survival for the primary outcome was significantly higher in the ARB group, but overall survival (secondary outcome) was not different. After adjustment for baseline factors, including age, sex, diabetes, systolic blood pressure, and center, they found a significant 49% relative risk reduction for the primary outcome and a nonsignificant 36% relative risk reduction for all-cause mortality among ARB-treated participants. No significant differences in blood pressure control, erythropoietin-stimulating agent dose, or hemoglobin level were noted, and adverse events were similar in both groups. If correct, this is an impressive finding.

Several factors about the study design and conclusion raise questions, and as the authors point out, require caution. These include the small sample size, the open-label nature of the study and lack of information on general care of the patient, the apparent lack of an outcome committee adjudicating end points, and some imbalances in the patient populations at baseline. Importantly, how does one reconcile the finding that cardiovascular (CV) morbidity and mortality together were significantly different, but all-cause mortality was not different? It is likely that actual CV mortality and all-cause mortality were not different, although these were not directly compared in a secondary analysis, and that cardiovascular morbidity, a component in the primary outcome, differed between groups. The reasons for this could include baseline differences in the randomized populations, bias in the management of the patients owing to the open-label nature of the study, and misclassification bias in adjudicating end points. How adjudication of events was actually carried out is not described in the paper, and in particular, whether an end-point adjudication committee reviewed each event according to predefined criteria. Whereas an outcome committee does not exclude errors in classification, it reduces the likelihood of such occurrence. This is important because the primary event rates for the primary and secondary outcomes were relatively low in both groups.

As noted in Table 3 of the Suzuki et al paper, the most common cardiovascular event (fatal and nonfatal) was congestive heart failure. Indeed, the event rate for congestive heart failure was double that in the control group as compared to the ARB group. Differences in occurrence of fatal and nonfatal myocardial infarction, stroke, and aortic dissection were minimal or nonexistent. As compared to the ARB group, overall there were 13 more deaths from any cause in the control group: 8 more CV deaths, 5 more non-CV deaths, and 17 more nonfatal CV events. The mean follow-up was not described, but based on the data in Table 2 and known deaths, it would appear that the average follow-up was around 18 to 24 months. Though difficult to calculate, an estimate of the all-cause mortality rate over the 36-month period is 7% per year in the control group and 4.3% per year in the ARB group. These mortality rates are low, even for an average dialysis unit in Japan, where rates are indeed lower than in the United States.13 Also, they occurred in a population having a mean baseline Kt/V below the US standard, in patients with a hemoglobin level of approximately 9 g/dL over 36 months, and with some patients who were on 2 or 3 dialysis treatments per week. Many other design issues in this study could have influenced the results, including lack of placebo control, the open-label nature of the study, lack of information on standard of care (dose of dialysis), inclusion of different ARB drugs, and, most importantly, volume control. For example, it is possible that use of diuretics was greater in the ARB group, and that there was imbalance in the number of participants receiving thrice-weekly treatment and assigned to ARB treatment. These issues are not addressed in the manuscript.

Another important consideration is the absence of information on adjudication of the nonfatal end point of congestive heart failure. Heart failure is often difficult to adjudicate in clinical trials of hemodialysis patients because it is relatively common and because it is difficult to differentiate from volume overload due to excess fluid and salt intake in the setting of end-stage renal disease. Moreover, a higher rate of calcium channel blocker use in the control group could contribute to misclassification of heart failure diagnosis inasmuch as lower-extremity edema is more common in patients using these agents. Since the criteria for diagnosis of heart failure were not stated in the paper, this remains a potential confounder.

What new important information should be taken from this study? First, as mentioned, the results of this study are in accord with the study of Takahashi (see above), giving credence to the notion that ARB therapy may be beneficial in hemodialysis patients. Still, the sample size is small and the event rate for mortality is low, so a type 1 error could explain these results. Second, it provides hope that easily administered interventions can improve cardiovascular outcomes in patients on hemodialysis, including those with diabetes (approximately 50% of participants in both groups had diabetes). The main take-home message is that better designed multicenter trials using an ARB should be undertaken in order to confirm or refute these findings. A meaningful relative risk reduction in cardiovascular morbidity and/or mortality of lower magnitude than observed in this study would be of great benefit to patients with end-stage renal disease. The nephrology community must continue to design and conduct long-term, large-scale intervention trials to improve the lives of those afflicted with end-stage renal disease.