Invited commentary by Dr. Steven Fishbane and Azzour Hazzan

In the Jan 2012 issue of the AJKD, Fukuma et al conducted an elegant analysis of ESA dose and hemoglobin (Hgb) concentration as predictors of patient outcomes, even after adjusting for other potential confounding factors, such as age, sex, time on dialysis therapy, postdialysis body weight, diabetes, history of CVD, serum albumin level, and transferrin saturation. They found that both were predictors of all-cause mortality, and there was an interaction between the variables. This is not a novel finding as both relationships have been demonstrated in a variety of patient populations. The strength of the current analysis was that it was conducted in Japan where both ESA doses and Hgb targets tend to be lower than in the United States. For example 9,000 U/wk is considered a high ESA dose in Japan, but not in the United States. The observed relationships despite a mean Hgb of 10.1 g/dL in their cohort is a valuable observation.

Nevertheless, the controversy of whether low Hgb levels and high ESA doses are harmful to patients or just a marker for worse health status remains. It is increasingly clear that sicker patients, those with greater risk for adverse outcomes, tend to have lower Hgb levels and therefore require higher ESA doses. It is also evident that 1) ESA hyporesponsiveness is an important marker of increased risk, probably by identifying sicker patients, and 2) the ability to demonstrate causality–that either low Hgb or high ESA dose requirements are truly harmful–is very difficult. While ESA hyporesponsiveness is at least partially a reflection of sicker patients, this does not exclude the possibility of some degree of harm. Indeed, low Hgb causes reduced oxygen carriage to tissues, and high, nonphysiologic ESA doses may cause injurious off-target effects.

Clinical practice is poorly informed by observations that are difficult to test in interventional studies. While at least 1 randomized controlled trial is currently underway to isolate the effect of higher or lower ESA doses, it is unclear whether conclusive results will be achievable. The clinician can only be advised that high ESA doses should be avoided and that the patient with low Hgb and higher ESA dose requirements (ESA hyporesponsiveness) should be evaluated carefully due to the high risk for adverse outcomes. This is especially true in centers where the dosing is performed by a physician assistant according to a protocol. A patient with high ESA requirement therefore should be flagged and the physician should examine this patient carefully for any sign of acute or worsening clinical condition.