Racial Differences in Erythropoietin Responsiveness

Since their introduction for clinical use in 1989, recombinant human erythropoietin (epoetin alfa) and other erythropoiesis-stimulating agents (ESAs) have become the mainstay of treatment for the anemia of chronic kidney disease (CKD). A number of factors have been shown to impair the responsiveness to ESAs including iron deficiency, inflammation, infection, and hyperparathyroidism. Several studies have suggested that African Americans on hemodialysis require higher epoetin doses than whites.1, 2 Whether this finding is due to the presence of hemoglobinopathies such as sickle cell disease, which occur more frequently in African Americans and impair ESA responsiveness,3 or to other factors is unknown.

In this issue of the American Journal of Kidney Diseases, Lacson et al report the results of a cross-sectional analysis of ESA use in African American and white hemodialysis patients treated with epoetin alfa at Fresenius Medical Care facilities in the United States in January, 2004.4 They observed that African Americans receive weekly epoetin doses which are on average 12.6% (2,575 U/wk) greater than whites. The increased epoetin dose persisted after adjustment for hemoglobin values in the prior 3 months. There was also a disproportionately higher prevalence of African American patients at higher quintiles of epoetin dose. As the authors acknowledge, the relative hyporesponsiveness of African American patients to ESAs has been previously described.1, 2 However, using the rich database of comorbidity and laboratory data available to them, they were able to assess comorbid conditions that are associated with the higher ESA doses in African Americans. Several variables associated with higher epoetin doses were more common in African American patients including younger age, lower past hemoglobin, longer dialysis vintage, higher body surface area, higher postdialysis weight, higher parathyroid hormone levels, lower estimated Kt/V, and lower white blood cell count. In contrast, several variables associated with lower epoetin doses were also associated with African American race including fewer prior hospitalizations, less dialysis catheter use, higher albumin, higher creatinine, and higher transferrin saturation and ferritin. Although in the unadjusted model African Americans receive a 12.6% greater ESA dose, in the adjusted model, which includes all the aforementioned variables, there is no racial difference in ESA dose, suggesting that these variables may be causally related to the increased ESA dose in African Americans.

Although the authors classify these variables as confounders, I would caution against applying this term broadly. From an epidemiologist's perspective, a confounder is a variable associated with the exposure, in this case race, and the outcome, in this case epoetin dose, but that is not in the causal pathway. A classic example would be the relationship between the presence of CKD and age and mortality. CKD and age are both associated with increased mortality and CKD is more common in the elderly, but CKD does not result in mortality by causing older age, so age is a true confounder of the relationship between CKD and mortality. In contrast, obesity is associated with diabetes and mortality, but in fact may result in increased mortality by causing diabetes. Therefore, diabetes should not be considered a confounder, rather it is on the causal pathway between obesity and mortality.

The conceptual relevance to Lacson et al's study of the distinction between confounding and causation is that some of the variables considered as confounders may be on the causal pathway between race and higher epoetin dose. The investigators note that average achieved hemoglobin in the 3 months prior to the epoetin dose used for the analysis was lower in African American patients. Prior hemoglobin is one of the strongest predictors of ESA dose, since most ESA dosing algorithms adjust dose primarily based on the preceding hemoglobin values without consideration of other variables such as age, sex, race, or dialysis adequacy. If African Americans in the dialysis population have lower hemoglobin levels at baseline, they might be expected to require higher ESA doses to achieve similar target hemoglobin levels. Accordingly, hemoglobin level would be in the causal pathway between African American race and higher epoetin dose. In the general US population, African Americans have on average a hemoglobin level that is 0.7 g/dL lower than whites, even after adjusting for possible iron deficiency or hemoglobinopathies which may be more common in African Americans.5 This racial difference also is observed in the CKD population.6, 7 Therefore, one might expect that African Americans, starting at lower hemoglobin levels, would require higher ESA doses to achieve similar hemoglobin targets compared with whites.

What are the clinical implications of Lacson et al's findings? Observational studies have shown an association between greater achieved hemoglobin levels and better outcomes including improved survival.8, 9, 10 These studies led to randomized controlled trials comparing outcomes in patients assigned to conventional hemoglobin targets of 10 to 12 g/dL with those assigned to normalization of hemoglobin (13 to 15 g/dL). Unfortunately, these randomized trials failed to show a benefit of hemoglobin normalization on cardiovascular function and some have indicated the possibility of increased mortality.11, 12, 13, 14, 15, 16

Several years ago, it was observed that in the US hemodialysis population, for any given level of hemoglobin, those patients receiving higher ESA doses had worse survival.17 Although this observation may simply indicate that dialysis patients who are sicker require higher ESA doses, reflecting ESA hyporesponsiveness with inflammation, it may also indicate an independent adverse association of ESA dose with death. There is a critical difference between the observational studies and the randomized trials comparing higher hemoglobin targets. In the observational studies, lower achieved hemoglobins are associated with higher ESA doses, while in the randomized trials, the patients assigned to the higher hemoglobin targets receive higher ESA doses. A recent post hoc analysis of the CHOIR (Correction of Hemoglobin and Outcomes in Renal Insufficiency) study, a randomized trial of hemoglobin normalization in patients with CKD not on dialysis, suggested that in patients who are ESA hyporesponsive, increasing ESA doses may be associated with increased mortality.18 Reflecting the results of these clinical trials and the possible adverse consequences of high ESA doses, the US Food and Drug Administration required changes in ESA product information, emphasizing that the target hemoglobin level in patients with CKD should be 10 to 12 g/dL, and that prescribers should use the minimum dose to avoid the need for recurrent transfusion requirements in patients who do not achieve this target.

Might the higher ESA dosing requirements in African Americans be associated with higher mortality? In general, African Americans in the United States have worse health outcomes than whites, which has been attributed to socioeconomic and healthcare access issues as well as biologic factors. However, African Americans on dialysis have better survival than whites, even adjusting for the younger age of end-stage renal disease onset in African Americans.19 This survival advantage for African Americans suggests that perhaps higher ESA doses may not have adverse consequences. However, it is possible that if African Americans received lower doses, their survival would be further enhanced. None of the randomized trials comparing different hemoglobin targets provided data on outcomes stratified by race, so the benefit of using lower epoetin doses with resultant lower achieved hemoglobins in African Americans remains speculative.

Appropriately, the authors discuss which of their associated variables might be modified to potentially reduce ESA dosing requirements in African Americans. They comment that age, dialysis vintage, and body size are not directly actionable. White blood cell count as a marker of inflammation, dialysis adequacy, and parathyroid hormone levels are potentially actionable, although they note that these variables were within recommended guideline targets. Whether increasing dialysis dose or further reducing parathyroid hormone levels in African American patients would result in significant reductions in ESA dose seems unlikely.

Because of concerns that financial incentives might promote the use of higher doses of ESA and higher hemoglobin targets, Congress has advocated the bundling of ESA reimbursement into the dialysis composite rate. Lacson et al raise the concern that the higher ESA dose requirements in African Americans may have an impact on facility costs under a bundled reimbursement policy. Such concerns may be legitimate if the hemoglobin targets are not race specific. As previously discussed, African Americans have lower hemoglobin levels than whites and it may be that attempting to abrogate these racial differences in the dialysis population may not lead to improved outcomes, particularly if there are adverse consequences of higher ESA doses. Unfortunately, we have a meager understanding of the relationship between target hemoglobin levels, ESA dose, and outcomes, making the design of an evidence-based reimbursement policy fraught with difficulty. Whether a bundled reimbursement should provide a higher rate to African Americans to allow for the additional ESA to achieve similar hemoglobin levels to those achieved in whites is uncertain in the absence of evidence regarding the relative benefits and harms of lower hemoglobin levels versus higher ESA doses on outcomes such as mortality, hospitalization and quality of life.

Finally, understanding the causal pathway between race and higher ESA dose may have application in the general CKD population. The lower hemoglobin values for African Americans in the general population and the relative ESA hyporesponsiveness in the hemodialysis population suggest the possibility of race-related genetic influences on hematopoiesis. Understanding the mechanism of such factors may help identify interventions that enhance the hematopoietic effects of ESA.