Background: Renal anaemia is caused by a relative erythropoietin (EPO) deficiency. Due to difficult interpretation of serum EPO concentrations adapted to anaemia and renal function, the diagnostic value of measuring serum EPO concentrations is limited.
Methods: We retrospectively analysed the relationship between haemoglobin and serum EPO concentrations routinely measured in in- and out-patients of our university hospital from 2001-04. Patients under EPO substitution or those with acute renal failure, polycystic kidney disease, renal carcinoma or polycythaemia due to pulmonary disease were excluded. The study population (n = 500) was then stratified according to the presence or absence of chronic kidney disease (CKD) and to the stage if CKD was present. EPO concentrations were expressed in percentiles corrected for the severity of anaemia and based on the EPO response in patients without CKD.
Results: In patients without CKD (n = 167) there was a strong parametric correlation between severity of anaemia and increase in EPO (r = -0.81). Linear regression of the log-transformed EPO values revealed the equation log EPO (mIU/ml) = -0.135 x Hb (g/dl) + 2.821 (r(2) = 0.65). With increasing stages of CKD the correlation between haemoglobin and EPO concentrations was gradually attenuated and was completely lost in CKD stage four and five. In anaemic patients with Hb < 11 g/dl, relative EPO deficiency defined as EPO concentrations below the 25th percentile was present in 38%, 67%, 93% and 100% of the patients with CKD stages 1-5, respectively.
Conclusions: Expression of EPO concentrations in percentiles improves the diagnostic value of measuring EPO concentrations for diagnosing relative EPO deficiency and renal anaemia.