Corrected Fructosamine improves both correlation with HbA_1C and diagnostic performance

There is increasing interest in using fructosamine measurements in screening for or managing diabetes, yet uncertainty remains as to whether these measurements should be corrected for variation in serum protein concentrations.

We considered all sets of simultaneous measurements of fructosamine, albumin, total serum protein (TP), fasting plasma glucose (FPG) and HbA_1C recorded in our laboratory over 10years. The relationships between fructosamine and other variables were studied by multivariate linear regression and other analyses, and receiver operating curves (ROCs) were analysed to compare the diabetes screening performance of uncorrected fructosamine to those of albumin-corrected fructosamine (FA_Alb) and TP-corrected fructosamine (FA_TP).

40,938 sets of measurements were collected from 20,114 patients. Though correlation between fructosamine and serum proteins was strongest among patients with HbA_1C <6.5% (48mmol/mol), it was also significant in the whole sample (r=0.193 for albumin, r=0.213 for TP). With diabetes defined by HbA_1C ≥6.5% (48mmol/mol), the areas under the ROCs of FA_Alb (0.905) and FA_TP (0.895) were both significantly greater (P<0.001) than that of uncorrected fructosamine (0.878). Correction of fructosamine for albumin or TP slightly improved its correlation with HbA_1C. There was no correlation of protein (albumin or TP) with log(fructosamine/protein).

Fructosamine concentration correlates significantly with albumin and total protein concentrations throughout their ranges. Correction of fructosamine improves its correlation with HbA_1C and its performance in detecting diabetes.