Anemia of chronic disease: A harmful disorder or an adaptive, beneficial response?

Anemia and oxygen delivery

There is little debate that severe anemia is associated with impaired oxygen delivery. However, in patients with anemia of chronic disease, the anemia is generally mild to moderate. Under normal circumstances, oxygen consumption in the resting human body is approximately 4 times less than oxygen delivery to the tissues.4 Increased extraction of oxygen from the tissues, combined with a rightward shift in the oxygen–hemoglobin dissociation curve, is sufficient to maintain normal oxygen delivery in the presence of mild to moderate anemia.5^,6

Experiments designed to quantify the cardiovascular response to anemia have generally done so by the acute induction of severe anemia below 80 g/L.7^–9 Induced acute anemia is associated with a progressive increase in heart rate and cardiac output;10^,11 however, these changes are not found in patients with stable mild or moderate anemia. Studies involving adults with chronic anemia showed no consistent changes in cardiac output until hemoglobin levels fell below 70 g/L or 50% of normal.12^–14 Thus, physiologic compensatory mechanisms in adults with non-acute mild anemia, although incompletely characterized, appear to differ from the changes seen in response to acute severe anemia. One obvious limitation of these studies is that testing was performed on resting participants. In people who are exercising, reductions in hemoglobin levels have been shown to reduce maximal oxygen consumption and endurance performance, but this is of uncertain relevance to chronically ill patients.15

Anemia as a marker of poor prognosis

Anemia is known to be associated with mortality in many disorders. Meta-analyses of large numbers of studies have established this association in renal failure,16 congestive heart failure17 and cancer.18

The observed association of anemia with increased mortality is not evidence of causation and should not be interpreted as such. Instead, both the degree of anemia and the direness of the prognosis likely reflect the severity of the underlying disease.18 Moreover, few of the studies adjusted for parameters such as cytokine levels or C-reactive protein that would help to distinguish between death from the anemia and death from the underlying inflammatory state. In a single study of heart failure that adjusted for cytokine levels, anemia lost significance as a predictor of death.19 In another study, anemia also lost prognostic significance after adjustment for clinical variables reflecting disease severity.20 A systematic review of studies involving patients with lung cancer showed that anemia was not an independent predictor of survival in the majority of studies that adjusted for other clinical factors.21

Although many other studies identify anemia as an independent predictor of poor prognosis, interpreting those studies requires careful consideration. Routinely measured clinical variables are unlikely to fully register the severity of underlying inflammatory processes and stress responses,18 which makes it hard to adjust for these effects completely. Excess deaths among chronically ill patients with anemia are generally not due directly to the anemia or to processes on which the anemia would be expected to have a major impact. Treatment of the anemia would not be expected to improve a patient's overall prognosis if the anemia is simply a marker of a serious, and possibly undiagnosed, underlying condition. Indeed, as discussed later in this article, treatment of anemia has repeatedly been found to worsen rather than improve clinical outcomes.

Critical care

Among critically ill patients, 2 observational studies showed red blood cell transfusion to be an independent predictor of death.29^,30 More substantial evidence is provided by a prospective, randomized controlled trial that reported increased mortality among patients randomly assigned to receive a liberal transfusion strategy (transfusion if hemoglobin levels fell below 100 g/L) compared with those assigned to receive a conservative transfusion strategy (transfusion if hemoglobin levels fell below 70 g/L).31 Mortality was significantly increased among less critically ill patients (16.1% in liberal-strategy group v. 8.7% in conservative-strategy group); the number needed to harm was 13. Critically ill patients have many reasons for anemia, but because systemic inflammation typically underlies critical illness, almost all can be expected to have some degree of anemia of chronic disease.

Transfusion in patients with anemia was also associated with increased mortality in 3 observational studies involving patients admitted with acute coronary syndrome and myocardial infarction.32^–34 Among patients who received red blood cell transfusion, the predicted probability of death was higher among those with a pretransfusion hematocrit above 0.25–0.30 than among those with a lower hematocrit. In other words, transfusion appeared to be harmful in patients with mild to moderate anemia but seemed to be beneficial to patients with more severe anemia.

One possible inference from these studies would be that red blood cell transfusion is harmful. Although there are a number of well-established risks associated with red blood cell transfusion, none is estimated to occur with a frequency that could explain more than a small fraction of the mortality risk described above. Other hypotheses (e.g., immunosuppressive and rheologic effects) are plausible but not well established. It is impossible to ascertain the extent to which anemia of chronic disease contributed to the anemia in the patients studied. Other causes of anemia would have also existed. Although any conclusion about anemia of chronic disease drawn from these studies is necessarily tentative, treatment of anemia by transfusion seems not to be beneficial, even in critically ill patients, who are least able to tolerate anemia and presumed to benefit most from its correction.

If transfusion in itself is harmful, then the use of erythropoiesis-stimulating agents could be expected to circumvent the increased risks associated with transfusion. In a large randomized controlled trial involving critically ill patients that attempted to address this question, administration of erythropoietin increased hemoglobin levels but did not reduce the frequency of transfusion or overall mortality.35 A meta-analysis of erythropoietin use in critical care was similarly unable to detect a mortality benefit despite erythropoietin's ability to reduce the frequency of transfusion in the overall study population.36 Thus, it may not be the red blood cells transfused but, rather, the achieved hemoglobin level that confers risk in critically ill patients with anemia of chronic disease. One further possibility exists: that anemia of chronic disease is harmful but that available interventions that increase hemoglobin levels confer even greater harm.

Renal failure

The treatment of severe anemia with erythropoietin has been found to improve well-being and decrease the frequency of transfusion in patients with renal failure. However, a recent meta-analysis of randomized controlled trials found that erythropoietin at a dose designed to achieve normal hemoglobin concentrations was associated with significantly higher mortality and increased thrombosis compared with regimens designed to achieve lower target hemoglobin levels.37 One of the trials involved 1233 patients with cardiovascular disease and end-stage renal failure; the number needed to harm (death or myocardial infarction) was 18.38 Another trial included in the meta-analysis involved 1432 patients with non-dialysis–dependent chronic kidney disease; the number needed to harm was 26.39

When interpreting these studies, one needs to consider the high prevalence of other illnesses and inflammatory conditions. Anemia in many of these patients will be due to at least 2 components: erythropoietin deficiency and anemia of chronic disease. It is plausible that physiologic correction of the erythropoietin deficiency may be beneficial; however, pharmacologic override of anemia of chronic disease may confer harm. This conceptual model could explain the paradox between observational studies of erythropoietin treatment that have shown improved outcomes associated with increased hemoglobin levels among patients with renal failure40^,41 and randomized trials that illustrated harm associated with higher hemoglobin levels.38^,39^,42 The patients in the observational studies who had higher hemoglobin levels may have had less systemic illness and therefore less severe and less prevalent anemia of chronic disease.

Cancer

Anemia is pervasive among patients with cancer, with multiple contributing mechanisms in different patients. Most intervention studies have evaluated the use of erythropoiesis-stimulating agents to support hemoglobin levels during chemotherapy or radiation therapy, but goals of treatment have varied.

In a meta-analysis of 51 randomized trials investigating the use of erythropoiesis-stimulating agents among 13 611 cancer patients, mortality was increased in the intervention arms (hazard ratio 1.10, 95% confidence interval 1.01–1.20).43 Those trials in which the intervention was designed to achieve a normal hemoglobin level were more likely to demonstrate harm than the trials that had a lower target hemoglobin level. Among patients with head and neck cancer who received radiotherapy, those given erythropoietin to maintain a normal hematocrit had a higher mortality than those in the control group.44 A randomized controlled trial of erythropoietin given to patients with breast cancer was stopped early because of excess mortality in the erythropoietin-treated group.45

Erythropoiesis-stimulating agents have also been used to ameliorate symptoms of anemia in cancer patients not receiving chemotherapy or radiation therapy. Although anemia in such patients is likely to be multifactorial, in the absence of myelosuppressive treatment, the anemia seen is more likely secondary to anemia of chronic disease. In randomized controlled trials involving patients with head and neck cancer46 or non-small-cell lung cancer,47 darbepoetin therapy was found to increase mortality. The US Food and Drug Administration has subsequently issued a warning against the use of erythropoiesis-stimulating agents in cancer patients not receiving chemotherapy or radiation therapy.48

In addition to critical care, renal failure and cancer, erythropoiesis-stimulating agents have been used to treat anemia of chronic disease in patients with rheumatoid arthritis49 and inflammatory bowel disease.50^,51 These patient populations may be more informative to our hypothesis, since the anemia in such patients may be more consistently attributed to anemia of chronic disease. However, no trials involving these patient populations have reported mortality as a prespecified outcome.