Target haemoglobin levels for the treatment of the anaemia of chronic renal failure

Without compelling evidence for a specific level, the appropriate target haemoglobin concentration for patients with chronic renal failure has been debated virtually since recombinant human erythropoietin (epoetin) first became available. Controversy has arisen particularly concerning the relative benefits and risks of maintaining haemoglobin concentrations ([Hb]) at physiological levels. A number of studies have found significant improvements in various parameters as the [Hb] is increased in anaemic subjects with and without associated chronic renal failure. 1–5 Reports however, of an increase in arterio-venous access thrombosis in dialysis patients, 6,7 the exacerbation or development of hypertension 8–10 and the potential for an increase in the number of adverse cardiovascular events at higher target [Hb] either in the short or long term 11 have prompted valid concerns. From a review of the literature in the NKF-DOQI guidelines, 12 785 of 3428 patients (23%) receiving epoetin who also had chronic renal failure either developed hypertension or required an increase in antihypertensive medication during treatment. In addition, it is evident that a substantially higher dose of epoetin at a proportionately increased cost is required to maintain higher target haemoglobin levels. 11,13 Other factors further complicate the dilemma. The [Hb] for haemodialysis patients is conventionally measured before dialysis and some have expressed fears that, if too high, excessive ultrafiltration may produce dangerously high haematocrit levels with associated intravascular red-cell slowing or thrombosis. 14–16 Changes in haematocrit from the start to end of dialysis can vary by as much as 3–6%, 17,18 depending on the volume of ultrafiltration and the timing of the post-dialysis sample. It would seem prudent to measure the post-dialysis [Hb] in patients with a predialysis level > 13 g/dL in order to ensure that haemoconcentration during dialysis does not exceed physiological limits. Recently also, it has become apparent that various patient subgroups might warrant special consideration. The elderly, the very young, diabetics and those with poorly controlled hypertension, or concurrent cardiac or pulmonary disease may conceivably either benefit or suffer particularly if [Hb] is normalized. Similarly, haematological conditions such as sickle cell disease or thalassaemia may require a specific and different target [Hb] as might patients with progressive renal insufficiency. The available evidence pertaining to these issues is limited. 5,19 Studies seldom focus on specific target [Hb] or specific subgroups, and interpretation is often restricted also from poorly designed or executed trials. 1,20 Recommendations therefore for an appropriate [Hb] for patients with renal disease should be regarded as an attempt to provide practical guidance within an incomplete paradigm. Currently, the best that can be hoped for is to conjure a balance between the proven benefits of the alleviation of anaemia in renal failure and the valid concerns (both empirical and theoretical) of adverse events as the [Hb] is increased. Early studies suggested a significant improvement in quality of life and exercise capacity when the [Hb] was increased from ≤ 70 g/L up to 130 g/L. 1,21–23 The largest of these, 21 a double-blinded, randomized, placebo controlled study was marred by a marked overlap in actual Hb levels between the planned higher (115–130 g/L) and lower (95–110) targets. This could well have affected the finding that no difference in quality of life or exercise capacity between the target Hb levels was detected. Although other studies 1,23 demonstrated similar findings, numbers were small and conclusions tentative. More recently, studies have examined quality of life, echocardiographic changes and exercise capacity in dialysis and/or predialysis patients at higher target [Hb], up to 160 g/L. 2–4 Few have been published in full, but most excluded high-risk patients (the elderly, diabetics, those with known cardiac disease or marked hypertension). In general, results appear to indicate a normal [Hb] is associated with an improved quality of life, a reduced left ventricular (LV) mass and an enhanced, but still subnormal, exercise performance. In these selected populations, a higher target [Hb] was found to be safe with little or no effect on blood pressure and no evidence of increased risk of cardiovascular events or arterio-venous access complications. Changes in cognitive and neurophysiological function have also been studied at normalized [Hb] in uraemic patients. Improved cognition and home management skills were identified in predialysis patients at a haematocrit of > 36% compared with a baseline of 28% over 48 weeks. 24 Similarly, improved quality and duration of sleep and enhanced daytime alertness was found in 10 haemodialysis patients after normalization of [Hb]. 25 Caution, however, again seems warranted in the light of another recent publication, 26 which investigated cerebral circulation and metabolism in seven haemodialysis patients using positron emission tomography. It found an increased oxygen extraction rate (OER) by brain tissue at normal Hb levels (Hct 42.4%) compared with the baseline (Hct 29.3%). The authors postulated that this was related to a decrease in erythrocyte velocity in cerebral capillaries due to decreased blood deformability and increased plasma viscosity. They were, however, unable to reconcile the increased OER with an overall increase in cerebral oxygen delivery or a paradoxical rise in cerebral blood volume when blood flow was reduced to normal levels. Cohort studies from Canada 27–29 have demonstrated that most patients have developed left ventricular hypertrophy by the time end-stage renal failure (ESRF) is reached. Preliminary data in another cohort of predialysis patients suggest also that LV hypertrophy is strongly and inversely correlated with [Hb], particularly as levels fall below 120 g/L (A. Levin, personal communication). An additional, controlled, randomized study over 40 weeks 3 has shown that, compared with partial anaemia correction ([Hb] 95–105 g/L), haemoglobin normalization (130–140 g/L) can prevent progressive LV dilatation in haemodialysis patients who have a normal LV volume. Regression of LV mass and dilatation were not achieved however, once established. Although a reduction in LV mass is well described in response to epoetin, 1,30,31 both irreversible structural change and incomplete correction of anaemia could well explain the observed limits in LV mass reduction. Large-scale population studies, examining both mortality and hospitalization, have received attention in recent years. Data from the United States in particular 32–34 have suggested an improved mortality and reduced rate of hospitalization at haematocrit levels of 33–36% compared with lower levels. The number of patients involved (over 70 000), together with an impressive consistency with other studies 35,36 make the data persuasive although formal analysis of data when Hct levels exceeded 36% ([Hb] > 120 g/L) was not possible. This was partly because there were too few patients at these levels to enable appropriate statistical analysis (< 700 patients had a Hct over 36%). Additionally however, since the Medicare system in the United States precludes funding for epoetin when the [Hb] exceeds 120 g/L unless medically indicated, results will potentially be biased owing to the unavoidable inclusion of sicker patients. (Of interest, no such trend was in fact identified.) Finally, the inclusion of a high percentage of blacks (41–54%) and diabetic patients (31–39%) and the lack of data regarding dialysis efficacy in these studies might also contribute to some interpretative uncertainty for other dialysis communities. Uraemia is known to induce high fibrinogen concentrations, poorly deformable erythrocytes and altered platelet and endothelial physiology. 37–40 Although concerns that such changes could interfere with blood flow in vascular access devices, particularly at higher [Hb], studies over the last 10 years have failed to demonstrate clearly that this is the case. 41–43 Recent data also suggest that blood rheology improves with epoetin treatment which, together with beneficial changes in the red-cell envelope and metabolism of the younger circulating red cell population, results in enhanced red cell survival. 44,45 The largest study to examine the effects of normalizing [Hb] was conducted in the USA and sponsored by Amgen: 11 1233 patients with clinical evidence of congestive heart failure or ischaemia were randomized to target haematocrits of 42% and 30%. Access thrombosis was more common both for synthetic grafts (48% thrombosis) and native arterio-venous fistulae (26% thrombosis) in patients whose target haematocrit was 42% compared with those with a target haematocrit of 30% (37% and 11% respectively). The findings did not relate statistically either to the haematocrit or to epoetin dosage and indicate a very high rate of graft dysfunction. It is possible that the age and co-morbidity (in particular poor cardiac function) of these patients was partly responsible for the results. Because of the large number of factors affecting fistula function, both patient and non-patient related, there remains doubt on the validity of individual study findings and whether in particular the use of epoetin at high [Hb] predisposes to access thrombosis. 13,40,46,47 Data from more than 4000 patients on haemodialysis treated with epoetin to obtain a [Hb] of 10–12 g/dL reviewed in the NKF-DOQITM9 guidelines show an overall thrombosis rate of access of all types of 7.5% although controlled data are lacking. There is some evidence however, that polytetrafluorethylene (PTFE) and related synthetic grafts maybe affected by higher [Hb], particularly in the presence of concurrent local stenoses. 7,11,48 The Amgen study was terminated early because of a possible significant increase in deaths in the ‘normal haematocrit’ group (183 vs 150, risk ratio 1.3; 95% CI, 0.9–1.9). It is noteworthy, however, that there was not an increase in cardiovascular deaths (125 vs 112) in the ‘normal haematocrit’ group nor in the incidence of angina, coronary interventions, myocardial infarctions or congestive cardiac failure over a period of 30 months (mean 14) until the trial was terminated. It was also apparent that mortality decreased with increasing haematocrit in both ‘normal haematocrit’ and ‘low haematocrit’ groups. Worsening or de novo development of hypertension after the use of epoetin is well recognized. The rate of increase of [Hb] (and quantity of epoetin used) as described in earlier studies 49,50 is probably important in most cases, however, as recent studies have not found a substantial increase in blood pressure levels at higher compared with lower target [Hb]. 4,51 Earlier fears also of a reduction in dialysis adequacy as [Hb] increased have not been substantiated. 52 The cost of maintaining [Hb] at physiological levels is significantly more than partial anaemia correction, with reports of between a 30% and 300% increase in the amount of epoetin required. 11,12,51 It has become clear that other factors such as inadequate iron replacement particularly and insufficient dialysis can significantly increase the amount of epoetin required in such circumstances. 53 Other issues yet to be adequately assessed include a possible reduction in hospitalization rates. Evidence from patient subgroups is still extremely limited. Age itself does not seem to alter the symptomatic and possibly objective benefits from higher [Hb]. 4,18 Data relating to predialysis patients is still relegated largely to the Canadian studies 27–29 and longer-term morbidity and mortality statistics are required. The use of epoetin in predialysis patients has not been shown to increase the progression of renal failure. 54 There maybe a special case for diabetic patients with or without documented ischaemic heart disease in view of their known generalized vasculopathy, retinopathy and altered blood rheology. 55 Preliminary data suggest also a possible increase in vascular disease at higher [Hb] in diabetic patients receiving peritoneal dialysis. 56 As opposed to patients with thalassaemia, patients with sickle cell disease should probably have a much lower [Hb] to avoid increased sickle crises in association with an increase in HbS. Concurrent treatment with hydroxyurea and exchange transfusions in addition to (often high dose) epoetin therapy may help to increase the proportion of HbF and reduce the sickling tendency. 57 The data would suggest therefore that, together with the development of detrimental symptomatic and cardiovascular changes, morbidity and mortality increase progressively in patients with ESRF as [Hb] falls. Caution in regards to an upper limit is required however, particularly for patients with pre-existing significant cardiac disease, a history of synthetic arterio-venous access thrombosis or possibly diabetes. A relatively arbitrary range of between 110 and 120 g/L would appear reasonable for these patients. A [Hb] of 120 g/L provides also a balanced lower limit for patients who are otherwise well. Available (if limited) data suggest that this latter group do benefit from a higher [Hb] without evidence of compromise. In the absence of sufficient studies it is not possible to advise an upper limit, although the majority of reported works have used 140 g/L as a target level. Together with justifiable theoretical concerns it would seem prudent to ensure that the predialysis [Hb] does not rise above this level. There is no evidence currently that the target [Hb] should differ for patients with progressive renal insufficiency or the elderly, and target levels for patients with chronic airflow obstruction remain unknown. Modification or consolidation of these proposed targets now depends upon the familiar platitude of further adequately designed and executed studies. It can, however, be expected that target levels will change if only to reflect greater certainty within patient subgroups as more data become available.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.