Caution: interpretation of results of HPLC assay for serum glycated albuminDear Sir,The exceedingly high values for glycated albumin (normal, 20.2 _+ 1.6 %, diabetic, 39.6 _+ 5.4 %) reported with the high performance liquid chromatographic (HPLC) assay described by Shima et al. [1] are far in excess of values established with any other method, including boronate affinity chromatography (normal range, 1.5-2.6 % [2]), densitometric scanning of agarose gel electrophoresis (normal range, 0.4-2.0% [3]), or monoclonal antibody based enzyme linked immunosorbent assay (normal, 2.3 + 0.29 % [4, 5]). Careful examination of the principles underlying the HPLC assay and the interpretation of its results are therefore necessary in order to avoid erroneous assignment of the acceptable range of values for non-diabetic populations and for assessment of glycaemic control.The principal determinants of the extent of nonenzymatic glycation of any protein are the glucose concentration to which it is exposed, the half-life of the involved protein, and the accessibility of free amino groups within the protein for condensation with glucose, the latter in turn dependent on local factors such as pKa and structural considerations. Thus, not every epsilon amino group of lysine residues becomes glycated; the principal site of albumin glycation in vivo is lysine-525, which has been reported to account for about 33 % of the overall glycation of this protein [6]. Glucose adducts continue to form as a function of time only until equilibrium is reached and, in general, there is a positive relation between protein half-life and extent of in vivo glycation; the dominant factor at low to moderate glucose concentrations appears to be half-life, those proteins with lower half-lives showing greater glycation [7,8].The number of glucose residues in a molecule of albumin is not the same thing as the percent of total number of albumin molecules that become glycated. In other words, glycation of 20 % of the lysine residues in a single molecule does not mean that 20% of the total number of albumin molecules are glycated (at one or more sites), and this point is apparently one of confusion in the report by Shima et al. The authors interpret the amount of albumin glycation of 0.37 mol/mol protein reported by other investigators [9] as equivalent to glycation of 37 % of the molecules and thereby conclude that other methods, which find considerably lower percentages, underestimate glycated albumin levels. They also interpret their value of 12.5 nmol/mg of protein as equivalent to glycation of 87 % of the albumin molecules, which is clearly not the case. One mg of albumin, at a molecular weight of 69, 600, is about 14.4 nmol; this means that they found 0.87 mol of glucose per mol of albumin, and not glycation of 87 % of the total number of albumin molecules.Since HPLC estimation of glycated albumin uses a boronate column to separate glycated from nonglycated species, it is important to know whether only glycated species bind to boronate. Austin et al.[3] point o...