Mosca et al.: Commutability of controls
end (arrow). This pattern was common in the Bayer (lanes 6 and 7) and Roche materials (lanes 26 and 27), which also shared a heavier background developed during migra- tion. The Bio-Rad controls (lanes 10 and 11) showed grossly diffuse Hb bands, with a lateral width almost double with respect to that of normal samples. On the contrary, band width was markedly reduced in the Mena-
a Codes as explained under Materials and Methods. Mean Hb A1c not declared. Adults, men and women. Mean SD. HPLC, Diabetes Control and Complications Trial traceable. Means or ranges of the Hb A1c values are reported from the technical information provided by the manufacturers. ND, not done. b c d e
Table 1. Some properties of the control materials for glyHb analysis.
MetHb, % of
Hb A1c, % of
present in the
Main Hb derivatives
Normal blood reference changes
rini controls (lanes 14 and 15). Finally, a totally unusual migration pattern was observed in the Boehringer mate- rials (lanes 30 and 31) where no Hb A, Hb A2, and carbonic anhydrase bands could be detected. The electro- phoretic pattern was normal for the Cal and Sam controls (lanes 18–20 and 22–24, respectively).
The regression analyses of results obtained by different methods for glyHb determination are reported in Table 2. The regression and correlation analysis calculated from the data obtained on blood samples (Table 2, upper part) proved that method–instrument pairs gave quite often different results. The intercept values were always signif- icantly different from 0 and slope values ranged from 0.964 to 1.053. In the case of comparisons among HPLC methods (the four most right columns in Table 2) the correlation coefficients (r) were constantly higher and sample residuals (Sy x) constantly smaller than those of the affinity chromatography (Abbott) and immunochemical methods (Boehringer, Roche). The imprecision of the methods (intraassay CV) was as follows: Abbott IMx, 2.7%; Bio-Rad Variant, 0.6%; Boehringer Tina-quant, 1.4%; Kontron s400, 1.2%; Menarini H8140, 1.1%; Merck-Hitachi L-9100, 0.9%; Roche Unimate A1c, 1.1%; Tosoh A1c 2.2, 0.9%.
In the lower part of Table 2 the results obtained on control materials, measured by the different techniques in the same runs with patient samples, are presented in terms of normalized residuals. The results from the Ab- bott controls were not reported, since these materials analyzed by methods other than Abbott IMx gave very low Hb A1c values, always below the physiopathological interval. Boehringer controls were excluded because of their very low total Hb content. In two cases (Lyp II, Abbott IMx; Roc P, Menarini H8140) normalized residuals exceeding the 3 interval were found when comparing the control materials to patient samples.
Among the different control materials that we tested, by taking into account the results obtained from Hb analysis (content of total Hb and derivatives, cellulose
Fig. 1. Electrophoretic Hb migration patterns obtained by alignment of eight cellulose acetate plates (four lanes per plate).
The reproduction of the last plate (lanes 29–31) was com- pressed to keep the abnormal Hb bands (arrows) in frame. The anodic compartment was at the top of each plate. c, normal blood sample; Abb, Abbott control materials; Bay, Bayer; Lyp, Bio-Rad Labs.; Gly, Menarini; Cal, Sam, home-prepared con- trols; Roc, Roche materials; Pre, Boehringer; CA, carbonic anhydrase.