Plasma Albumin Concentration Is a Predictor of HbA_1c Among Type 2 Diabetic Patients, Independently of Fasting Plasma Glucose and Fructosamine

The diabetes outpatient clinics of our center are attended by most local diabetic patients requiring insulin or oral antidiabetics. We enrolled 4,158 diabetic patients who, in this complex in the years 1998–2003, were prescribed insulin or oral antidiabetics for type 2 diabetes diagnosed using American Diabetes Association criteria (8) and who, for glucose control monitoring, underwent regular determination of fasting HbA_1c accompanied, for research purposes, by determination of total protein concentration, albumin, globulins, creatinine, hemoglobin, fructosamine, and glucose; age, sex, duration of diabetes, and type of therapy were also recorded. For this report, we considered for each patient the first such profile obtained within the study period. Urinary albumin excretion (normal range 0–25 mg/24 h) was determined within 6 months of the blood profile considered in this report for 1,840 patients (subgroup UAE).

HbA_1c was determined with the high-performance liquid chromatography DCCT (Diabetes Control and Complications Trial)-aligned method, albumin by the bromcresol purple method, and fructosamine by the nitroblue tetrazolium method. The normal ranges in a group of 197 healthy subjects were as follows: HbA_1c 3.8–5.7%, albumin 35–53 g/l, and fructosamine 208–286 μmol/l. All analyses were performed in the clinical biochemistry laboratory of our center.

Subjects were classified according to their albumin concentrations (by quintiles), and the mean log (HbA_1c) values of these “albumin-level groups” were compared by ANOVA. Student’s t tests and Mann-Whitney tests were used when appropriate. Stepwise multiple regression was employed to evaluate the association between HbA_1c and the possible predictors albumin, fructosamine, globulins, creatinine, and fasting plasma glucose (FPG).

Of the total 4,158 patients, 44.6% showed good apparent glucose control (HbA_1c <7%), 18.6% fair control (7% ≤ HbA_1c ≤ 8%), and 36.8% poor control (HbA_1c >8%).

Figure 1 shows the variation of HbA_1c, fructosamine, and FPG with albumin level. HbA_1c differed significantly among the albumin-level groups (P < 0.001), decreasing with increasing albumin level. Among patients with albumin levels lower than the first quintile of the albumin concentration distribution, the proportion apparently showing poor control (HbA_1c >8%) was 51.9%, 2.3 times larger than in the top albumin-level group (22.4%). There were no statistically significant differences among the albumin-level groups with regards to fructosamine concentration or among the top four albumin-level groups with regards to FPG.

When patients were divided in subgroups defined by hemoglobin concentration (> or ≤123 g/l for women, > or ≤140 g/l for men) or creatinine level (> or ≤115 μmol/l), then, within each subgroup, HbA_1c was lower among patients with albumin concentrations higher than the mean for healthy individuals (45 g/l) than among those with lower albumin concentrations (P = 0.042 for the subgroup with creatinine >115 μmol/l, P < 0.001 otherwise). The same difference in HbA_1c between high- and low-albumin groups was observed when subgroup UAE was divided in subgroups with urinary albumin > or ≤25 mg/24 h (P = 0.006 and P < 0.001, respectively).

Stepwise multiple regression showed significant correlation between HbA_1c and albumin, fructosamine, globulins, and FPG (P = 0.007 for globulins, P < 0.001 for the others, R² = 0.602). As was expected, the most influential predictors were FPG and fructosamine, but albumin concentration (r = −0.325) accounted for 16.4% of the variance in HbA_1c among the 4,158 patients (compared with 23.4% for fructosamine).

In this study of type 2 diabetic patients, there was significant negative correlation between HbA_1c and serum albumin after adjustment of HbA_1c for fructosamine, FPG, and globulins (P < 0.001).

The high HbA_1c values of low-albumin patients and low HbA_1c values of high-albumin patients were not essentially due to alteration of erythrocyte life span by iron deficiency or hemolytic anemia, respectively (9), because the negative association between HbA_1c and albumin persisted among both anemic and nonanemic patients when these two groups were examined separately. Nor was the observed association due to poorly controlled diabetic patients having elevated vascular and renal permeability to albumin because it also persisted when patients with serum creatinine >115 μmol/l or urinary albumin >25 mg/24 h were excluded from the analysis. The possibility (10–15) that albumin levels might have been significantly affected by poor glucose control (duly reflected by high HbA_1c) is also unlikely in view of the lack of correlation between albumin and either FPG or fructosamine (which also weighs against the possibility of some unknown underlying variable with opposite effects on average glycemia and albumin). Moreover, the steady fall in HbA_1c concentration over the whole range of albumin concentrations suggests a physiological rather than a pathological relationship.

HbA_1c, of course, increases with FPG and fructosamine. Its falling with increasing albumin therefore implies that attainment of any given HbA_1c level requires a lower glucose concentration in patients with low albumin levels than in patients with higher albumin levels and, hence, that the glucose control of patients with albumin levels significantly above or below average may not be properly reflected by the standard classification in terms of HbA_1c measurements alone. For such patients, there may be a discrepancy between the degree of control suggested by HbA_1c measurements and the evolution of diabetes complications.

This work was supported by the Xunta de Galicia, Spain (Ref. PGIDIT02BTF20303PR), Menarini Diagnostics, and Roche Diagnostics.