Clinical Biochemistry 48 (2015) 397–400

Contents lists available at ScienceDirect

Clinical Biochemistry
journal homepage: www.elsevier.com/locate/clinbiochem

Clinical

Association of variation range in glycated albumin (GA) with increase but

not decrease in plasma glucose: Implication for the mechanism by which
GA reflects glycemic excursion
Kunihiko Hashimoto a, Kazuko Tanikawa a, Jun Nishikawa a, Yingchao Chen a,
Toshinobu Suzuki a, Masafumi Koga b,⁎
a
Department of Internal Medicine, NTT West Osaka Hospital, Osaka, Japan
b
Department of Internal Medicine, Kawanishi City Hospital, Hyogo, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Objectives: HbA1c mainly reflects mean plasma glucose (PG), whereas glycated albumin (GA) reflects glyce-
Received 12 September 2014 mic excursion in addition to mean PG; the mechanism of the difference between HbA1c and GA is unknown. We
Received in revised form 1 December 2014 hypothesized that a transient increase in PG irreversibly produces stable GA unlike HbA1c. To prove this hypoth-
Accepted 19 December 2014 esis, we investigated diurnal variations in PG, HbA1c, #C fraction (a fraction containing unstable HbA1c and
Available online 3 January 2015
modified hemoglobin on HPLC) and GA in diabetic patients.
Design and Methods: Sixteen diabetic patients with poor glycemic control were enrolled in this study. Blood
Keywords:
HbA1c
sampling was performed before and after each meal, before bedtime, and before breakfast on the following day;
Glycated albumin PG, HbA1c, #C fraction, and GA were measured. The variations of these indicators were compared with those in
Maillard reaction PG.
Glycemic excursion Results: HbA1c showed almost no change regardless of diurnal glycemic variation. Variation range in #C frac-
Postprandial hyperglycemia tion significantly correlated with variation range in PG when PG increased (R = 0.746, p b 0.0001) and decreased
(R = 0.271, p = 0.035). On the other hand, variation range in GA significantly correlated with variation range in
PG when PG increased (R = 0.322, p = 0.021), but not when PG decreased (R = 0.090, p = 0.493).
Conclusions: We observed that variation range in GA significantly correlated with variation range in PG
when PG increased but not when PG decreased for the first time. It is considered that GA reflects glycemic excur-
sion through this phenomenon.
© 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

Introduction Glycated albumin (GA), another glycemic control indicator, is a

HbA1c is widely used as a gold standard of glycemic control in-
dicators [1]. HbA1c is a glycated protein in which glucose is
nonenzymatically bound to valine in the β-chain of hemoglobin, and
is produced by the two step nonenzymatic reaction called the early
Maillard reaction [2]. HbA1c mainly reflects mean plasma glucose
(PG); it does not reflect glycemic excursion well. It is considered that
this is because a transient increase in PG reversibly produces unstable
HbA1c but does not produce stable HbA1c rapidly. HbA1c is observed
when hemoglobin is eluted by high-performance liquid chromatogra-
phy (HPLC). Fractions eluted from unstable HbA1c and modified hemo-
globin by the HPLC of Arkray Inc. and Tosoh Corp. are called #C fraction
and LA1C+, respectively.
⁎ Corresponding author at: Department of Internal Medicine, Kawanishi City Hospital,
5-21-1 Higashi-Uneno, Kawanishi, Hyogo 666-0195, Japan. Fax: +81 72 794 6321.
E-mail address: m-koga@kawanishi-city-hospital.com (M. Koga).
glycated protein similar to HbA1c [3]. It is considered that GA reflects
glycemic excursion and/or postprandial hyperglycemia in addition to
mean PG [3]. Recently, it has been shown that indicators of glycemic ex-
cursion determined by continuous glucose monitoring (CGM) have a
significant correlation with GA or the GA/HbA1c ratio but not HbA1c
[4,5]. Therefore, the GA/HbA1c ratio is high in patients with postprandi-
al hyperglycemia because GA is higher than HbA1c in these patients;
when a drug to improve postprandial PG is administered to these
patients, the GA/HbA1c ratio decreases, reflecting improvement of post-
prandial hyperglycemia [6,7]. However, the mechanism of the differ-
ence between HbA1c and GA in reflecting PG has been unknown until
now, even though they are both glycated proteins.
Regarding the mechanism of the phenomenon in which GA reflects
glycemic excursion and/or postprandial hyperglycemia, we hypothe-
sized that it is because the Amadori reaction of GA progresses rapidly,
unlike HbA1c. To test this hypothesis, we examined diurnal variations
in PG and GA in diabetic patients with poor glycemic control and

http://dx.doi.org/10.1016/j.clinbiochem.2014.12.021
0009-9120/© 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
398 K. Hashimoto et al. / Clinical Biochemistry 48 (2015) 397–400
investigated the relationship between them. We also investigated diur-
nal variations in HbA1c and #C fraction and compared with that in PG.
Materials and methods
Subjects studied
Sixteen diabetic patients who were hospitalized at the NTT West
Osaka Hospital between May 2013 and November 2013 for glycemic
control were enrolled in this study (Table 1). Patients with liver disease,
renal disease, anemia, or administration of glucocorticoid, which influ-
ence the measurement of HbA1c and/or GA, were excluded. Without
change in the treatment of diabetes mellitus immediately after hospital-
ization, blood sampling was performed before each meal, at 2 h after
each meal, before bedtime, and before breakfast on the following day,
and PG, HbA1c, #C fraction, and GA were measured and evaluated.
The specimens obtained by blood sampling were immediately stored
at 4 °C, and measurements were performed within 12 h. This study
was approved by the Ethics Committee of the NTT West Osaka Hospital;
all patients received an explanation of the aims of this study and gave
their written consent.
Laboratory methods
PG was determined by the hexokinase method. HbA1c, expressed as
a National Glycohemoglobin Standardization Program (NGSP) value or
an International Federation of Clinical Chemistry (IFCC) value (%) [8],
and #C fraction were measured by HPLC [9] with an ADAMS-A1c HA-
8170 analyzer (Arkray Inc., Kyoto, Japan). Interassay coefficient varia-
tions were 0.85% and 0.67%, respectively, as determined in representa-
tive blood samples [5.3% (34.4 mmol/mol) and 10.4% (90.2 mmol/mol)
of HbA1c). GA was determined by the enzymatic method using
albumin-specific proteinase, ketoamine oxidase, and albumin assay re-
agent (Lucica GA-L; Asahi Kasei Pharma Co., Tokyo, Japan) [10,11], with
a Hitachi 7600 autoanalyzer (Hitachi Instruments Service Co., Tokyo,
Japan). Interassay coefficient variations were 1.38% and 1.32%, respective-
ly, as determined in representative serum samples (13.3% and 34.9% of
GA). The reference ranges of HbA1c, GA, and #C fraction were between
4.6% and 6.2% (26.7 mmol/mol and 44.2 mmol/mol), between 11.7%
and 16.0%, and between 1.5% and 2.0%, respectively.
Statistical analysis
All data are shown as mean ± SD. Single linear univariate regression
analysis was employed to assess the association between continuous
variables using the StatView computer program (Version 5.0 for
Windows, Abacus Concepts, Berkeley, CA). A p value b 0.05 was consid-
ered statistically significant.
Table 1
Clinical characteristics of study patients.
Number of patients 16
Men (%) 8 (50)
Type 1 DM/Type 2 DM (n) 2/14
Age (years) 63.8 ± 11.6
BMI (kg/m2) 27.4 ± 11.6
Duration of diabetes (years) 16.2 ± 13.1
FPG (mg/dL) 145 ± 48
MPG (mg/dL) 174 ± 44
HbA1c (%) 9.0 ± 1.6
HbA1c (mmol/mol) 74.6 ± 17.3
GA (%) 23.9 ± 6.0
#C fraction (%) 2.4 ± 0.3
Results
The characteristics of the study patients included in this study were
as follows: eight men and eight women; age: 63.8 ± 11.6 years; BMI:
27.4 ± 11.6 kg/m2; duration of diabetes: 16.2 ± 13.1 years; two type
1 diabetic patients and 14 type 2 diabetic patients) (Table 1). Fasting
PG (145 ± 48 mg/dL), HbA1c (9.0 ± 1.6%; 74.6 ± 17.3 mmol/mol),
GA (23.9 ± 6.0%), and #C fraction (2.4 ± 0.3%) in these patients were
all high. Before the hospitalization, the treatments of the patients were
as follows. Five patients received no medication. Six patients used oral
antidiabetic agents. Five patients received insulin therapy, of whom
three patients also received oral antidiabetic agents.
Variations range in PG, HbA1c, #C fraction, and GA among blood
samples taken at different time points are shown in Fig. 1. PG increased
from before meal to after meal and decreased from after meal to before
meal. Although PG varied throughout the day, HbA1c showed almost no
change. On the other hand, #C fraction varied in a similar way to the
variation in PG. Similar to an increase in PG, GA showed an increase
from before breakfast to after breakfast, and from before dinner to
after dinner. GA showed a decrease from after breakfast to before
lunch, from after lunch to before dinner, and from before bedtime to
after breakfast; this pattern was generally similar to the variation in PG.
Correlations between variation ranges in HbA1c, #C fraction, and GA
and variation range in PG were investigated; no significant correlation
was observed between variation range in HbA1c and variation range
in PG (R = 0.012, p = 0.896) (Fig. 2). On the other hand, a significant
correlation was observed between variation range in #C fraction and
variation range in PG (R = 0.782, p b 0.0001) and between variation
range in GA and variation range in PG (R = 0.330, p b 0.001).
Furthermore, regarding the correlation between variation range in
GA and variation range in PG and the correlation between variation
range in #C fraction and variation range in PG, separate analyses were
performed for the cases when PG increased from the previous value
(ΔPG N 0 mg/dL) and the cases when PG decreased from the previous
value (ΔPG ≤ 0 mg/dL). Regarding variation range in #C fraction, a sig-
nificant correlation with variation range in PG was observed in both
cases (R = 0.746, p b 0.0001; R = 0.271, p = 0.035, respectively);
regarding variation range in GA, a significant correlation with variation
range in PG was observed only in the case when PG increased (R =
0.322, p = 0.021), but not when PG decreased (R = 0.090, p = 0.493)
(Fig. 3).
Discussion
In the present study, we examined diurnal variations in glycemic
control indicators and PG in diabetic patients with poor glycemic con-
trol and analyzed the relationship between each indicator and PG. #C
fraction varied in association with glycemic excursion, whereas HbA1c
showed almost no change. It is considered that this was because when
increased PG decreases, unstable HbA1c is reversibly dissociated to
hemoglobin and glucose.
On the other hand, when PG increased, a significant correlation was
observed between variation range in GA and variation range in PG, but
no such correlation was observed when PG decreased. Because GA is
an irreversible Amadori compound, a decrease in GA in association
with a decrease in PG was not observed. It is considered that GA reflects
glycemic excursion through this phenomenon.
Shima et al. [12] investigated diurnal variation in GA in diabetic pa-
tients and reported that GA showed almost no change whereas PG var-
ied. It should be noted that they did not show PG levels; they defined PG
before breakfast as 100% and expressed other PG levels relative to it. Un-
like the results of the present study, they observed no obvious diurnal
variation in GA; this may have been because the patients in their
study only had a small glycemic excursion, which resulted in no obvious
variation in GA. In contrast, the present study included diabetic patients
with poor glycemic control and a marked glycemic excursion.
 K. Hashimoto et al. / Clinical Biochemistry 48 (2015) 397–400 399

A B

C D

Fig. 1. Variation ranges (Δ) in plasma glucose (PG), HbA1c, #C fraction, and GA. Blood sampling was performed before each meal, at 2 h after each meal, before bedtime, and before break-
fast on the following day, and PG (A), HbA1c (B), #C fraction (C), and GA (D) were measured. Variation ranges (Δ) in each parameter among blood samples taken at different time points of
are shown. B0: before breakfast; B2: at 2 h after breakfast; L0: before lunch; L2: at 2 h after lunch; D0: before dinner; D2: at 2 h after dinner; BB: before bedtime.

Kouzuma et al. [11] reported on the measurement of GA by the en-
zyme method that GA after exclusion of unstable GA from the specimen
by dialysis was similar to GA before dialysis. Therefore, it is considered
that if unstable GA produced from albumin and glucose rapidly un-
dergoes the Amadori reaction and stable GA is produced, then their re-
port and the results of the present study refer to the same phenomenon.
Traditionally, it has been reported that the glycation rate of GA is
faster than that of HbA1c. Day et al. [13] reported that the in vitro
glycation reaction of GA progresses about 10 times faster than that of
HbA1c. The difference in glycation rate is considered to be due to the dif-
ference in Amadori reaction. It is not known why there is a difference in
the rate of the Amadori reaction between HbA1c and GA. In HbA1c, va-
line is glycated, whereas lysine is glycated in GA; the difference in amino
acid may lead to different kinetics of the early Amadori reaction. This
issue requires further investigation.
If GA increases in association with an increase in PG, and if GA does
not decrease in spite of a decrease in PG, does GA continue increasing?
Because HbA1c has a long half-life and a slow rate of catabolism, the de-
crease in HbA1c due to catabolism is very small. On the other hand, the
half-life of GA is as short as 2 weeks [14]; therefore, GA decreases in as-
sociation with its catabolism. When the GA level is defined as “a,” the
daily decrease in GA level can be expressed by the following formula:
(a − 15) / 28 (%/day). In the present study, a mean decrease in GA of
0.2% was observed between before bedtime and before breakfast; this
can be explained as a decrease due to catabolism of GA.
The present article has some limitations. First, only a small number
of patients were included in this study. Second, glycemic excursion
was evaluated by blood sampling seven times daily not by CGM. If
CGM is used, it is expected that the continuity of glycemic excursion
could be accurately monitored, and the relationship between PG and

A B C

Fig. 2. Correlations between variation ranges (Δ) in plasma glucose (PG) and HbA1c, #C fraction, and GA. Correlations between ΔPG and ΔHbA1c (A), Δ#C fraction (B), and ΔGA
(C) are shown.
400 K. Hashimoto et al. / Clinical Biochemistry 48 (2015) 397–400
A B
Fig. 3. Correlations between variation ranges (Δ) in plasma glucose (PG) and #C fraction and between variation range in PG and GA according to variation range in PG level. Correlation
between ΔPG and Δ#C fraction (A) and correlation between ΔPG and ΔGA (B) were investigated for separate cases (ΔPG ≤ 0 mg/dL and ΔPG N 0 mg/dL).
glycemic control indicators could be better understood. In future, it is
necessary to examine the findings of the present study by using CGM
in a large number of patients.
In the present study, we proposed a new hypothesis for the
mechanism by which GA reflects postprandial PG/glycemic excur-
sion. The difference between variation in HbA1c and variation in
GA in association with a transient increase in PG was clearly identi-
fied. This difference is expected to identify different characteristics
of HbA1c and GA as predictors of diabetic complications. In the fu-
ture, we hope that new findings about diabetic complications and
glycemic control indicators will be obtained in large-scale studies
which incorporate these viewpoints.
Conflicts of interest
None of the authors have any conflicts of interest to declare.
Acknowledgments
We are grateful to Mr. Shin-ichi Kumei and Mr. Kiminori Sumi
(Department of Clinical Laboratory, NTT West Osaka Hospital) for
their invaluable assistance.
References
[1] American Diabetes Association. Standards of medical care in diabetes—2014.
Diabetes Care 2014;37(S14-80).
[2] Higgins PJ, Bunn HF. Kinetic analysis of the nonenzymatic glycosylation of hemoglobin.
J Biol Chem 1981;256:5204–8.
[3] Koga M. Glycated albumin; clinical usefulness. Clin Chim Acta 2014;433:
96–104.
[4] Ogawa A, Hayashi A, Kishihara E, Yoshino S, Takeuchi A, Shichiri M. New indices for
predicting glycaemic variability. PLoS One 2012;7:e46517.
[5] Tsutsumi C, Imagawa A, Onishi M, Sano H, Nakagawa S, Murase-Mishiba Y, et al.
Glycated albumin as a useful clinical biomarker for glycemic variability in type 1 di-
abetes assessed by continuous glucose monitoring. Diabetol Int 2013;4:156–9.
[6] Sumitani S, Morita S, Deguchi R, Hirai K, Mukai K, Utsu Y, et al. Metformin decreases
glycated albumin to glycated hemoglobin ratio in patients with newly diagnosed
type 2 diabetes. Ann Clin Biochem 2015;52:76–81.
[7] Kurebayashi S, Motomura T, Goya K, Nakao M, Hashimoto K, Morimoto Y, et al. Ef-
fects of sitagliptin on the glycated albumin to HbA1c ratio. J Diabetes Metab 2014;
5:343.
[8] Committee on the Standardization of Diabetes Mellitus-Related Laboratory Testing
of Japan Diabetes Society (JDS). International clinical harmonization of glycated
hemoglobin in Japan: from Japan Diabetes Society to National Glycohemoglobin
Standardization Program values. Diabetol Int 2012;3:8–10.
[9] Davis JE, McDonald JM, Jarett L. A high-performance liquid chromatography method
for hemoglobin A1C. Diabetes 1978;27:102–7.
[10] Kouzuma T, Usami T, Yamakoshi M, Takahashi M, Imamura S. An enzymatic method
for the measurement of glycated albumin in biological samples. Clin Chim Acta
2002;324:61–71.
[11] Kouzuma T, Uematsu Y, Usami T, Imamura S. Study of glycated amino acid elimina-
tion reaction for an improved enzymatic glycated albumin measurement method.
Clin Chim Acta 2004;346:135–43.
[12] Shima K, Ito N, Abe F, Hirota M, Yano M, Yamamoto Y, et al. High-performance liquid
chromatographic assay of serum glycated albumin. Diabetologia 1988;31:627–31.
[13] Day JF, Ingebretsen CG, Ingebretsen Jr WR, Baynes JW, Thorpe SR. Nonenzymatic
glucosylation of serum proteins and hemoglobin: response to changes in blood
glucose levels in diabetic rats. Diabetes 1980;29:524–7.
[14] Koga M, Suzuki S, Matsuo K, Tanahashi Y, Azuma H, Kasayama S. Calculation of
HbA1c and glycated albumin from serially measured self-monitored blood glu-
cose in patients with type 1 diabetes mellitus. Clin Chim Acta 2013;425:
188–91.