Correlation of fasting blood glucose and haemoglobin A1c measured with an automated analyser

A subtype of glycohaemoglobin, haemoglobin (Hb) A1c, in specimens of whole blood was assayed on a new automated analyser that makes use of high-pressure liquid chromatography. The analyser provided precise and reproducible values. The mean of the HbA1c values was lower than that with an older instrument. The mean tended to increase with the age of the subjects, who were undergoing routine health examinations. No sex difference was found. When measurement was made 1 h after the subjects drank 50g of glucose, the value of HbA1c was unaffected. Correlation was strong between the HbA1c value and the fasting blood glucose value, which suggested that fasting blood glucose could be estimated from the HbA1c value.


Introduction
There are several methods for the measurement of haemoglobin Ale (HbAlc) in red blood cells, involving electrophoresis, colorimetry, minicolumns and highpressure liquid chromatography (HPLC). Each method has different advantages but a problem shared by all of these is the long testing time.
Here, we evaluate the performance of an automated apparatus for the assay of HbA1 and HbAI (Hi-Auto AI, model 8121, Kyoto Daiichi Kagaku Co., Ltd) first marketed in 1987. In this model, HbAla, HbAlb, HbAI, and HbF are separated more completely than in the earlier analyser (model 8120) tested here for comparison,  [2]) was used.

Materials
Everyone who comes to our health care centre is a company employee with no known current health problems who is to undergo routine health examinations.  Tests of urine, stools, blood chemistry and pulmonary function are carried out, and a complete blood cell count, X-ray films of the chest and the upper gastrointestinal tract, and electrocardiographs are taken. Glucose metabolism is examined by tests of urinary glucose and fasting blood glucose (FBG), and by a 50 g oral glucose tolerance test (OGTT), from which the blood glucose concentration h after the intake of glucose (postprandial blood glucose, or PBG) was used here. In the OGTT, blood was collected into bottles containing NaF when the subject was fasting and again at h.
Venous blood was withdrawn from the examinees with a Venoject syringe (Terumo Co., Osaka) containing NaF (1"25 mg/ml blood), sodium citrate (0"5 ml of a 3"85% solution per 4"5 ml of blood), or a mixture of EDTA-2Na (3'7 mg/ml blood) and heparin (12"5 U/ml blood). The glycohaemoglobins in the blood were assayed within h of sampling. Blood glucose was assayed by the neocuproine method [3].  than the older model in assays of both of these haemoglobins. With NaF, the mean values for HbAc for the two glycohaemoglobins were significantly different. The differences probably arose from differences in the column fillers and in the elution method, which cause differences in the efficiency of the removal of unstable glycohaemoglobin. With NaF, the blood glucose value of the specimens was not lowered as much as with the other anticoagulants, so that nearly all of the unstable form remained unbound in the specimens. For these reasons, the HbAlc values obtained by the older model, in which removal of the unstable glycohaemoglobin was less efficient, were higher. When samples were stored at 4 C or room temperature in one of the three anticoagulants for up to 5 days (figure 1), results for HbAlc stayed about the same. However, at room temperature, values for HbA increased during storage. Table 2 shows the day-to-day reproducibility of HbAlc and HbA in control material assayed by the newer model with use of the same column for 20 days. The number of other specimens measured daily was about 90, each from a different subject; in all, 1786 specimens were assayed during these 20 days. Retention time with the HPLC column did not change. Each day, after all test specimens were assayed, control material was tested. Correlation was satisfactory for both glycohaemoglobins. Thus, repeated use such as is described here did not affect the results obtained with the column.

Relationship betweeen glycohaemoglobin values and blood glucose levels
In diabetic subjects, the unstable form of HbAlc increases as the glucose level increases but the stable form is unchanged [4]; the same changes occur in healthy subjects, as in the OGTT. Thus, total HbAc (stable plus unstable forms) rises with increasing glucose. In the 69 subjects chosen at random during the test period to illustrate typical results here, the mean FBG level was 106 + 17 mg/dl and the PBG level was 161 + 45 mg/dl. Figure 2 shows the values for each of the three forms of haemoglobin assayed by the new model, HbA1, stable HbAlc, and HbF, when the subjects were fasting (x-axis) and at h after glucose intake (y-axis). Correlation was satisfactory for all these three forms. Figure 3 shows the distribution of HbAlc values in 7260 examinees screened over a 4 month period. The coefficient of variation (CV) was 5"22% .and the standard deviation (SD) was 0"70%. The reference interval was calculated by the method of Hoffman [5]. The mean reference value and the SD were 5"13% and 0"43%, respectively.    and one SD is shown. The mean values for this one large group measured with the older model were generally higher than those of the other large group measured with the newer model. HbA]c tended to be higher in men than in women. It also tended to increase with age. Figure 5 shows the correlation between HbAlc and FBG in the 7260 specimens of figure 3. The correlation was high.

Discussion
The clinical significance of glycohaemoglobins has been established and their assay is becoming routine.
However, the presence of HbF and the instability of HbA]c [4]  Because HbA is unstable when blood is stored, and because correlation is good between HbA and HbAlc levels in blood assayed soon after being sampled, in general, assay of HbAlc is more likely to give reliable results. Results for the assay of control material were highly reproducible (table 2).
The anticoagulants did not affect the level of HbA1 and HbAc, so any of them can be used when HbAlc is to be assayed. If the glucose level of the specimen is also to be measured, NaF can and should be used.
The mean difference in the glucose level when 69 of the subjects were fasting and h after the intake of 50 g of glucose was about 50 mg/dl but the levels of HbA1 and HbAlc were unaffected by this difference.
The reference values for HbA and HbA obtained by groups using cation-exchange chromatography [6] are wider than ours, probably because of the removal of unstable HbAlc by the machine we used. We found that HbAI tended to increase with the age of our subjects (aged 22 to 82 years). The same tendency has been reported for subjects divided into those aged 21 to their 45th birthday and those older. Because both the blood glucose level and glucose intolerance increase with age [8], the glycohaemoglobin level can be expected to increase, so our findings are reasonable. HbAlc is ofuse in the monitoring of the glucose levels of a subject in the preceding few months. Because correlation between HbAlc and FBG is high, the value of HbAc should also be of use for prediction of the likeliness that a subject will develop diabetes mellitus.