Giovanni FM Strippoli
MD MPH(Hons) MM(Epi)
Cochrane Renal Group
Department of Emergency and Organ Transplantation
University of Bari
The relationship between anaemia and the pathogenesis of cardiac abnormalities in patients with chronic kidney disease (CKD) has been the focus of much clinical and basic science research over the past two decades. CKD is best defined as an estimated/measured glomerular filtration rate (GFR) of less than 60ml/min (including any lower level of GFR up to endstage renal disease requiring dialysis). Whenever an individual has CKD, renal and extrarenal complications (including anaemia) become evident. Anaemia develops with the decrease in renal function, due to deficient endogenous erythropoietin (EPO) production by the kidney, decreased red cell survival and gastrointestinal blood losses.(1,2) It triggers adaptive cardiovascular mechanisms, which aim to maintain tissue oxygen supply, resulting in left ventricular hypertrophy and dilation, with potential risk for myocardial ischaemia and cardiovascular events. Anaemia is also recognised to impact on cognitive functions, quality of life, exercise capacity and sexual function in patients with CKD. According to the WHO, anaemia is defined as a level of haemoglobin (Hb) below age- and gender-specific normal ranges (ie, Hb<13g/dl [130g/l] for males and nonmenstruating females and Hb<12g/dl [120g/l] for premenopausal women). Under these definitions, a large proportion of patients with CKD are anaemic. The prevalence of anaemia in patients with GFR<25ml/min, but not yet on dialysis or transplant may be as high as 87%, whereas up to 85% of subjects starting any renal replacement therapy have Hb levels below 10g/dl.(2,3)
Observational data and experimental trials on the management of renal anaemia
Observational data from registry and cohort studies in CKD patients have shown consistent associations between lower Hb values, adverse changes in cardiac structure and function, and mortality.(4,5) Other observed adverse effects include impaired cognition, reduced quality of life, sleep disturbances and decreased exercise tolerance. Since cardiovascular diseases are the leading cause of death in CKD people and Hb levels have been shown to be associated with cardiac events both before and after dialysis commencement, reversing anaemia has been expected to reduce the risk of cardiovascular disease and death. Most patients with CKD before or undergoing dialysis have been successfully treated with recombinant EPO to stimulate erythropoiesis and partly correct anaemia. High Hb targets have been widely advocated in an attempt to reduce the unacceptably high cardiovascular mortality rate of CKD patients. Unexpectedly, however, randomised clinical trials have failed to demonstrate a survival advantage of further increasing Hb levels to normal range.(6) These findings, which contrasted with observational data, have been recently confirmed by a systematic review evaluating benefits and harms of different Hb targets in CKD on the basis of randomised trial evidence.(7) Of the 19 trials (2,696 patients) of at least two-month duration identified, 12 (638 patients) compared EPO vs no EPO, and seven (2,058 patients) compared the effects of high versus low Hb levels (about 10 vs 14g/dl). All-cause mortality was not different by intervention type, but was lower among patients with lower Hb levels (relative risk [RR] for death: 0.84; 95% confidence interval for RR [95% CI]: 0.71 to 1.00) with absence of any heterogeneity among trials. The risk of seizures was found to be higher among subjects treated without EPO (RR: 5.25; 95% CI: 1.13 to 24.34), and not significantly different between Hb levels. However, the risk of hypertension needing additional anti‑hypertensive interventions was lower among subjects not receiving EPO (RR: 0.5, 95% CI: 0.33–0.76), and again not different between Hb levels. Other outcomes available in the published randomised trials literature, such as quality of life, are scant. There is substantial variability of assessment and uncertainties regarding the validity of the instruments used, whereas summary data for angina, infarction, arrhythmia, stroke, vascular access thrombosis, transfusions and hospitalisation rates are incomplete, such that no major conclusion on the effects of EPO administration on these outcomes may be drawn at present.
The analysis of the findings from these randomised trials was mainly influenced by the findings of the trial carried out by Besarab et al,(6) the largest published so far. Other small trials are inadequately powered to show any benefit or harm of EPO treatment and of different Hb targets. It must be noted that the quality of all the trials published is suboptimal according to current methods standards, due to unclear allocation concealment and minimum use of blinding of the participants, investigators and outcome assessors (internal validity issues). The Besarab trial enrolled haemodialysis patients with clinical cardiovascular disease, a high prevalence of diabetes and prosthetic graft access, these conditions being associated with worse survival.(6) For these reasons, it has been argued that any inference from the data should refer only to similar populations for whom Hb targets of <12g/dl proved to be safer than normal range values. Higher Hb targets have been recommended in the CKD population before dialysis or in subjects without overt cardiovascular disease, despite the lack of randomised controlled clinical trials to demonstrate this. If normal Hb targets reduce mortality, then this should be even more evident in studies of high-risk patients such as those enrolled in the Besarab trial.(6) In addition, the higher mortality rates found in higher Hb target groups in recent cancer trials also make any survival benefit in CKD populations at least unlikely.(8) External validity (generalisability) of the Besarab trial is supported by the annualised mortality rate of the patients included in the trial, which is similar to that of the US Renal Data System population.(9) Finally, the biological and clinical nature of the relationship between Hb levels and cardiovascular disease in CKD patients has yet to be established. Despite a growing understanding of physiological and pathological processes, important methodological flaws limit the clinical meaning and impact of accumulating data. To date, the majority of studies have been observational in nature. Observational studies are appropriate to evaluate the association between exposure to a given factor (eg, EPO, high Hb target) and an outcome (eg, mortality), but not a causal relationship between the two. Thus, to prove that higher Hb targets achieved with EPO improve survival, randomised trials are needed. Observational studies are not appropriate to answer an intervention question. Hence, the trend towards higher Hb targets across all races and genders in CKD appears to be unjustified according to epidemiological criteria and availability of evidence.
Several nontraditional or emerging cardiovascular risk factors may also interfere in the apparent relationship with anaemia. In addition to residual confounding, other methodological problems are typical of observational studies, including possible insufficient duration of follow-up to detect clinically meaningful events, use of surrogate markers of disease such as cardiac hypertrophy instead of hard binary morbidity/mortality endpoints and, most importantly, selection bias (healthier subjects who have both clear survival advantage and more favourable exposure levels [higher Hb levels]). For all these reasons, randomised controlled trials, not observational studies, are the most appropriate study design to answer intervention questions. All that can currently be inferred from these studies is that higher Hb targets are associated with no difference (at best) or potential increase in the risk (at worst) of death in patients with endstage renal disease and overt cardiovascular comorbidities.
From available experimental trial evidence, the lower risk of seizures associated with EPO treatment achieving higher Hb levels is outweighed by the higher risk of hypertension and mortality. Data from patients with CKD and cardiovascular disease indicate that the preferred Hb target should be <12g/dl, whereas data related to other CKD populations (predialysis or dialysis patients without overt cardiovascular disease) and outcomes such as quality of life improvement are still unclear. Until additional and well-designed trials can provide answers to these questions, clinicians need to consider potential harms and costs whenever prescribing interventions of unproved efficacy.
- Hsu CY, McCulloch CE, Curhan GC. J Am Soc Nephrol 2002;13:504-10.
- Obrador GT, Roberts T, St Peter WL, et al. Kidney Int 2001;60:1875-84.
- Levin A, Thompson C, Either J, et al. Am J Kidney Dis 1999;34:125-34.
- Levin A. Nephrol Dial Transplant 2002;17:207-10.
- Weiner DE, Tighiouart H, Vlagopoulos PT, et al. J Am Soc Nephrol 2005;16:1803-10.
- Besarab A, Bolton WK, Browne JK, et al. N Engl J Med 1998;339:584-90.
- Strippoli GFM, Craig JC, Manno C, et al. J Am Soc Nephrol 2004;15:3154-65.
- Lancet Oncol 2004;5:1.
- United States Renal Data System. Available from: http://www.usrds.org