Carbamylated Hemoglobin Interference in Glycohemoglobin Assays

Glycohemoglobin is widely accepted as a valuable indicator for long-term diabetic control (1). The in vivo reaction of hemoglobin with urea-derived isocyanate and the possible interference of the resulting carbamylated hemoglobin in uremic patients have been described (2). Since our earlier studies (3)(4), conflicting results have been reported (5)(6)(7) and methods have been modified or newly introduced. We recently revisited carbamylated hemoglobin as part of the educational program of the European Reference Laboratory for Glycohemoglobin (ERL). We investigated the interference of carbamylated hemoglobin by comparing HbA1c results in lyophilized specimens from a uremic patient and a nonuremic volunteer, both being nondiabetic, in the 1998 ERL program. The first sample was prepared from the blood of a 50-year-old man with a mean urea during the preceding month of 27.9 mmol/L. According to a previous study (3), which concluded that 0.063% carbamylated hemoglobin is associated with each mmol/L of urea, a carbamylated Hb percentage of 27.9 × 0.063% = 1.8% was expected. Using capillary electrophoresis, our group separated carbamylated hemoglobin from HbA1c (8), and a value of 2.0% was found in this specimen. The second sample was chosen on the basis of having an equal HbA1c percentage as the first sample and a carbamylated hemoglobin concentration of 0.3%, which is within the health-related reference range (with both HbA1c and carbamylated hemoglobin measured with National Glycohemoglobin Standardization Program-certified capillary electrophoresis). This specimen was analyzed by 24 laboratories of the ERL program (in 17 countries) using 14 methods, and the results were compared with those for the sample from the nonuremic volunteer (Table 1). The results on the two samples agreed within 0.5% by immunologic, affinity chromatographic, capillary electrophoresis, and IMx methods. All HPLC methods based on ion-exchange chromatography showed higher HbA1c percentages in the sample of the uremic patient. The mean difference for all 21 HPLC users was 1.6%, quite near the calculated value of 1.8% and the observed value of 2.0% for carbamylated hemoglobin with capillary electrophoresis. This is not entirely unexpected because urea reacts with the hemoglobin molecule at the same site as does glucose, and the isoelectric point of carbamylated hemoglobin is thus similar to that of HbA1c. As a consequence carbamylated hemoglobin might be assayed as HbA1c in methods based on differences in electrical charge. In the immunoassays (+0.1%) and capillary electrophoresis (−0.1%), hardly any difference was seen. Interestingly, HPLC based on affinity chromatography (+0.5%) and Abbott IMx (+0.4%) showed a small but significant difference. Conclusions, based on results from a single sample, should be made with caution: even between two nonbiased methods differences up to 1% might occur because of scatter in individual patients. The mean difference of 1.6% as seen with ion-exchange methods seems too large to be explained by individual scatter, but the 0.5% seen with affinity methods might be attributed to this phenomenon. The persistence of carbamylated hemoglobin interference will be investigated in the next ERL program with a specimen from another uremic patient. An additional problem, although not directly related to carbamylated hemoglobin, is the frequent occurrence of shortened erythrocyte life-span in renal failure, which might obscure correct interpretation of HbA1c outcome. Difference in HbA1c outcome between samples with different carbamylated hemoglobin concentrations but equal HbA1c percentages as measured with 14 methods. a–c Total number of laboratories for each method: an = 21; bn = 8; cn = 4. Difference in HbA1c outcome between samples with different carbamylated hemoglobin concentrations but equal HbA1c percentages as measured with 14 methods. a–c Total number of laboratories for each method: an = 21; bn = 8; cn = 4. We believe that carbamylated hemoglobin gives rise to falsely high HbA1c in all HPLC methods based on ion-exchange chromatography.