Using 123 specimens, we compared the concordance of three different me
thods for determining glycohemoglobin (GHb): the Diamat(TM) (Bio-Rad L
aboratories), an automated analyzer measuring HbA(1c) by cation-exchan
ge chromatography; an assay with the IMx analyzer (Abbott Laboratories
), based on boronate affinity binding; and an HPLC method measuring Hb
A(1c) by cation-exchange chromatography on a PolyCAT A column (PolyLC
Inc.). The Pearson's correlation coefficient between PolyCAT A and Dia
mat was 0.900 +/- 0.038 (mean +/- 2 SD) and between PolyCAT A and IMx,
0.857 +/- 0.042. However, up to twofold differences were seen in some
samples. The proportion of GHb was consistently lower with the PolyCA
T A method than with the other two assays, apparently because of bette
r separation of HbA(1c) from nonglycated coeluting forms of Hb. The, d
ifference in glycation percentage between the PolyCAT A and Diamat met
hods is 2-3% over the whole concentration range. These results point t
o the limitations of Diamat as a reference method to be used to calibr
ate other methods for determining HbA(1c). Further, a switch from one
method to another is likely to cause considerable problems in the clin
ical follow-up of certain patients.