Received: 4 August 2016 | Accepted: 28 December 2016

DOI: 10.1002/jcla.22155

RESEARCH ARTICLE

Analytical evaluation of the ADAMS™ A1c HA8180T analyzer

for the measurement of HbA1c

Eloísa Urrechaga

Hospital Galdakao – Usansolo, Galdakao,

Vizcaya, Spain Background: The ADAMS™ A1c HA-­8180T (ARKRAY, Inc., Kyoto, Japan) is a high
pressure liquid chromatography system designed for separation and quantitation of
Correspondence
Eloísa Urrechaga, Core Laboratory, Hospital HbA1c, while detecting hemoglobin (Hb) variants and HbA2, with an analysis time of
Galdakao – Usansolo, Galdakao, Vizcaya, just 3.5 min. The aim was to evaluate the analytical performance of this analyzer for
Spain.
Email: [email protected] routine HbA1c quantitation.
Methods: Trueness, imprecision, carry over, linearity, the effect of anemia and correla-
tion with ADAMS™ A1c HA-­8180V (in Variant mode) were assessed according to
guidelines of the Clinical and Laboratory Standards Institute (CLSI) and manufacturer.
The presence of coexisting interfering chemically modified Hb variants was also
assessed.
Results: The results were found to be accurate (total error of 1.74%) and precise:
within-­run, between-­run, between-­day and total coefficients of variation were 0.71%,
0.77%, 0.36%, and 0.93% at low concentrations, and 0.43%, 0.29%, 0%, and 0.48% at
high concentrations of HbA1c. Carry over, linearity, and correlation of methods were
excellent. HbA1c measurement was independent of total Hb concentration in a range
40-­214 g/L, and shows no interference from Hb chemically modified commonly ob-
served in this type of analysis, while common Hb variants are detected.
Conclusion: The analyzer is a suitable system for glycemic control in patients with
­diabetes and the detection of Hb variants.

KEYWORDS
diabetes, HA-8180T analyzer, HbA1c, HPLC

1 | INTRODUCTION diagnosed when HbA1c ≥6.5% (48 mmol/mol).3 An extension of the

role of this Hb fraction, from monitoring to the screening and diag-
Measurement of glycated hemoglobin (HbA1c), a glycation product nosis of diabetes, leads to the possibility of even greater demand for
(1-­deoxyfructosyl) of the N-­terminal valine in the β globin chain of he- this analysis.
moglobin (Hb) A0, provides the most important medium-­ to long-­term Various methods have been used for the measurement of Hb
marker of time-­averaged glycemic status. Its relationship with likely fractions based on their physical, chemical, or immunologic charac-
clinical outcome in diabetes mellitus has been convincingly demon- teristics. The automation and precision of high performance liquid
strated for both type I and type II patients in major clinical trials.1,2 chromatography (HPLC) methods have favored their use and technical
For over two decades, measurement of HbA1c has been widely improvements have allowed analysis times to be reduced.
used as a means to retrospectively evaluate long-­term glycemic We have evaluated the analytical performance of the ADAMS™
control in diabetic patients. More recently, the International Expert A1c HA-­8180T (ARKRAY, Inc., Kyoto, Japan) analyzer, assessing the
Committee published a report recommending that type 2 diabetes be quality of the measurement of HbA1c.

J Clin Lab Anal. 2018;32:e22155. wileyonlinelibrary.com/journal/jcla © 2017 Wiley Periodicals, Inc. | 1 of 6

https://doi.org/10.1002/jcla.22155
2 of 6 | URRECHAGA

2 | MATERIALS AND METHODS 2.3 | Analytical performance evaluation

To conduct the evaluation, we followed the recommendations of the
2.1 | ADAMS™ A1c HA-­8180T
manufacturer and guidelines of the Clinical Laboratory Standards
This is a fully automated HPLC analyzer. The system has been de- Institute (CLSI).4
signed to provide the maximum amount of information in the shortest Carry over and within-­run precision were evaluated in accordance
possible time. It combines the sensitivity and specificity of HPLC with with the CLSI EP10-­A3 guideline. Specifically, carry over was assessed
automation software. by analyzing a sample with high values three times (H1, H2, and H3)
Working with the β-­thalassemia screening mode, the analysis time followed immediately by a sample with low levels (L1, L2, and L3); the
is 3.5 min. The system quantitates HbF, HbA2, HbA0, and flags abnor- percentage of carry over was calculated using the following formula
mal peaks; and common variants HbS and HbC can be detected as (L1-­L3)/(H3-­L2)*100.
separate peaks on the chromatogram (Figure 1). Trueness was evaluated following the manufacturer’s guidelines,
The analyzer is equipped with an autosampler, barcode reader and analyzing two commercial controls per run for over 20 days. Mean
cap piercing. Continuous measurements can be made on up to 100 value and standard deviation were calculated; bias was recorded as an
samples with bidirectional communication with a laboratory informa- absolute value and as a percentage.
tion system. The measurement principle is based on reversed phase For within-­run precision, three samples one from a healthy subject,
cation exchange chromatography using a stepwise elution system another from a patient with poorly controlled diabetes and another
driven by a single pump; the column is kept at 39°C with thermostatic at the diagnostic threshold of 7.0% (53 mmol/mol) were analyzed 10
control. times each within the same run. The mean and coefficient of variation
The system can handle both whole bloods and manually diluted (CV) were calculated.
samples. Four microlitre of whole blood are automatically sampled, Within-­laboratory precision was tested by analyzing two commer-
hemolyzed and injected into the column. Hb molecules elute with cial controls with different levels of HbA1c; in accordance with CLSI
inorganic phosphate buffers (80A, 80B, and 80CT) from low to high EP5-­A2 guidelines both controls were analyzed twice over a 20-­day
polarity. Upon elution, sample components pass through the spectro- period.
photometric detector where fractions are monitored with dual wave- Linearity was investigated by sample mixing to evaluate the
length detection (420 and 500 nm). ­response to increasing concentrations of the measurand (CLSI guide-
The total area of all peaks is a critical factor in the quality of the line EP6). Specifically, the linearity of the method was evaluated by
results. When the total area of the chromatogram is less than 15000, measuring HbA1c in serial dilutions of two blood samples with normal
the anemia measurement mode can be selected and the anemia rack is and high concentrations (38-­113 mmol/mol; 5.6% and 12.5%, respec-
used: the analyzer automatically changes the dilution ratio (standard, tively), containing the same total Hb concentration; these ­samples
1:100; anemia, 1:50) to provide accurate results even if in the pres- were mixed at the ratios of 9:1, 8:2…1:9 and the eleven samples
ence of severe anemia. ­obtained were analyzed.
Two-­point calibration is employed; raw data are integrated Passing-­Bablok regression was used to compare theoretical ex-
and a chromatogram report is generated. Quality-­control results pected and measured values. The method comparison with the ana-
can be stored; mean and standard deviation are automatically cal- lyzer currently in use in our Laboratory (an ADAMS™ A1c HA-­8180V in
culated for each HbA2 level; and Levey-­Jennings graphs are also Variant mode) was performed according to CLSI EP9, by running 110
plotted. blood samples in parallel without Hb variants; pairs of HbA1c results
were then compared with Passing-­Bablok regression.
The effect of anemia on HbA1c results was evaluated according to
2.2 | Samples and controls
the manufacturer’s guidelines, as follows: First, a blood sample from a
Blood samples were obtained from patients whose level of glycemic patient was centrifuged for 10 min at 11.5 g. Second, the red cells and
control or hemoglobinopathy was being assessed routinely. After plasma obtained were mixed at ratios of 9: 1, 8:2…1:9 and the sam-
the requested tests had been completed, the residual samples were ples obtained were analyzed. Then, Hb was measured on a XN counter
used for the present study. This practice is in accordance with the (Sysmex Corporation, Kobe, Japan).
guidelines established by the Institutional Review Board at Galdakao-­ The effect of a wide range of concentrations of total Hb was evalu-
Usansolo Hospital. ated; the experiment was performed twice with samples from a healthy
Whole blood samples obtained for routine analysis were collected subject and another from a patient with poorly controlled diabetes.
in K2 EDTA anticoagulant tubes (Vacutainer™ Becton-­Dickinson, Potential interference from the labile fraction (LA1c) was also in-
Rutherford, NJ, USA). Quality-­control materials used throughout vestigated; pooled blood from non-­diabetic adults (HbA1c 31 mmol/
the evaluations were provided by Menarini Diagnostics (Firenze, mol, 5.0%) was poured into six tubes and increasing volumes of glucose
Italy), and we considered target values assigned by the International solution was added to these test samples; final glucose concentrations
Federation for Clinical Chemistry (IFCC) Reference Measurement ranged from 11.1 to 55.5 mmol/L. The samples were incubated for 2 h
Procedure. at 37º C, and then analyzed.
URRECHAGA | 3 of 6

A B

A2
S-A1c

F S-A1c
A2
F

0 20 40 80 120 160 (seg.) 0 20 40 80 120 160 (seg.)

C D

S-A1c A2
S-A1c F
F

A2

0 20 40 80 120 160 (seg.) 0 20 40 80 120 160 (seg.)

F I G U R E 1 Typical chromatograms to HbA0, HbA1c, and HbA2. (A) poorly controlled diabetic, (B) β thalassemia carrier, and an additional
fraction identified as HbS (C) or Hb C (D)
4 of 6 | URRECHAGA

The same procedure was employed for carbamylated Hb (cHb) 13

with increasing concentrations of a sodium cyanide solution (up to
12
10 mmol/L), and for acetylated Hb (AcHb) with increasing concentra-
tions of an acetaldehyde solution (2-­10 mmol/L). 11

Common variant Hbs, Heterozygous HbA, HbC, HbD, HbJ, HbE, 10

HbA1c Measured
Hb Lepore, and double heterozygous HbSC, were also analyzed to in-
9
vestigate the ability of the system to detect, interpret and quantitate
them. 8

2.4 | Numerical analysis 6

We used MedCalc (version 16.2; MedCalc Software, Ostend, Belgium; 5

https://www.medcalc.org; 2016) for the analysis. 4 6 8 10 12 14
HbA1c Theoretic

F I G U R E 2 Linearity study by sample mixing was performed to

3 | RESULTS evaluate the response to increasing concentrations of the measurand,
range 5.6-­12.5 % (38-­113 mmol/mol)
Using samples with concentrations of 30 and 104 mmol/mol (4.9 and
11.7%), no carry over was detected. That is, there was no increase in
the value obtained with the low concentration sample. Correlation between results from the HA-­8180T analyzer and the
Trueness and imprecision results were excellent. Concerning true- HA-­8180V in Variant mode was excellent in the range 15-­130 mmol/
ness, the normal control (theoretical value 40 mmol/mol, 5.8%, mean mol (3.5-­14.0%), mean difference −0.1.
5.82%, standard deviation 0.04%) showed an absolute bias of 0.035 y=1.000 x+0.0, R2=.999 (95% CI 0.998-­1.000) (Figure 3).
(0.6%); and the high control (theoretical value 95 mmol/mol, 10.8%, Slope 95% CI 1.000-­1.000; intercept 95% CI −0.00-­0.00.
mean 10.81%, standard deviation, 0.03%) showed an absolute bias The quantitation of HbA1c was not influenced by the presence of
of 0.015 (0.13%). Regarding within-­run imprecision using blood sam- LA1c, cHb, or AcHb within the range tested. 38 mmol/mol (5.0%) Hb
ples, the CV was 0% for HbA1c concentrations of 40 mmol/mol (5.8%) A1c was the result obtained for the six samples analyzed while LA1c
and 102 mmol/mol (11.5%), and the CV was 0.4% for a concentration raised up to 4.4% (Figure 4); the same level in cHb or AcHb produced
of 53 mmol/mol (7.0%), that is, at the diagnostic threshold. Within-­ a decrease of <0.2% in HbA1c levels).
laboratory imprecision (commercial controls) is summarized in Table 1. The presence of the most frequent heterozygous Hb variants was
Linearity was confirmed at the concentration range of 38-­ assessed: HbS, HbC, HbD (identified by the system with a specific flag)
113 mmol/mol (5.6-­12.5%): and HbO were well separated from HbA1c and HbA0 peaks; HbA1c
and the variant percentage are reported. Hb Lepore co-­elutes in the
y = 0.930 x + 0.51 R2 = 0.994
area of HbA0 and HbA2 and HbE also elutes with HbA2, while HbJ co-­
elutes with HbA0, splicing the peak.
Slope 95% CI 0.824-­1.043; intercept 95% CI −0.31-­1.54 (Figure 2).
The measured values correlated with theoretical ones; the mean
proportion recovered was 98.5%. 4 | DISCUSSION
Results were reliable across a wide range of Hb. In particular, an
HbA1c of 5% (38 mmol/mol) was not affected by Hb in the range 40-­ There is a high demand for HbA1c testing in the clinical laboratory and
214 g/L, total chromatogram area 34 386-­64 999, while an HbA1c of therefore efficiency is required. Efforts of manufacturers have contrib-
11.5% was not affected in the range 66-­223 g/L. uted to improvements in the analyzers, creating faster systems. On the
other hand, therapeutic strategies rely on reproducible and unbiased
T A B L E 1 Imprecision study HbA1c. Total precision was
investigated using CLSI EP5-­A2 guideline: over a 20 working days methods. These analytical and clinical requirements become even more
period, two commercial controls with different concentrations are important now that it has been accepted that HbA1c level can be used
run in duplicate twice a day in analytical runs of at least 10 samples for the diagnosis of diabetes. In this context, guidelines and recommen-
dations were published suggesting an intra-­laboratory CV <2.0%.5
HbA1c 40 mmol/ HbA1c 95 mmol/
mol 5.8% CV % mol 10.8% CV % The analyzer achieved this stringent target in the study, with
between-­day CVs of 0-­0.36%, and total CVs of 0.48-­0.93%; the eval-
Within run 0.71 Within run 0.43
uation revealed excellent reproducibility, far beyond the stringent
Between run 0.77 Between run 0.29
requirements.
Between day 0.36 Between day 0
Actually, desirable specifications have been published for total
Total 0.93 Total 0.48
error, imprecision and bias, derived from intra-­ and inter-­individual
URRECHAGA | 5 of 6

A 14 Total analytical error (TAE) includes bias and imprecision (CVa):

12 TAE = %bias + 1.96 ∗ CVa

The total error in the present evaluation was found to be 1.74%

10
(mean bias of 0.365% and mean CVa of 0.705%), and this meets the
HbA1c %

8 quality standards.
Quality requirements can also be calculated based on likelihood
8180 T

6 ratios7 or the reference change value (RCV), which combines the two
sources of variation, and is defined as the critical difference between
4
two consecutive HbA1c measurements, representing a significant
change in health status at a probability of 95%:
2
2 4 6 8 10 12 14 √
2 2
8180 V HbA1c % RCV = 2.77 ∗ (CVa + CVw )

B 0,3 Considering a difference of 0.5% to be statistically significant

at an HbA1c concentration of 7.0% (5.0 mmol/mol at 53 mmol/
0,2 mol), the goal for HbA1c measurement, an appropriate goal in terms
+ 1.96 SD
0,15 of CVa can be calculated. Assuming that the mean CVw is 1.9%6,
0,1
then the maximum allowable CVa is 1.9%;8 in our evaluation, the
8180 V - 8180T

Mean total CV was 0.93% at a normal HbA1c level, and this satisfies the
0,0
-0,00 requirements.
-0,1 Hb concentration had no significant effect on HbA1c levels. The
- 1.96 SD
results were reliable within a wide range of total Hb values, reflect-
- 0,15
-0,2 ing clinical situations: life-­threatening anemia, severe anemia, mild
anemia, normal iron status, and polycythemia. This feature shows the
-0,3
robustness of the system, and hence it is possible to monitor patients
2 4 6 8 10 12 14 16
Mean of 8180V and 8180T with confidence in a wide range of clinical situations.
LA1c, the concentration of which varies with blood glucose con-
F I G U R E 3 Method comparison of ADAMS™ A1c HA-­8180 centrations at the time of blood collection, is an intermediate in the
Variant mode and ADAMS ™ A1c HA 8180 Thalassemia mode,
production of the corresponding stable HbA1c. The isoelectric point
(Passing -­ Bablok and Bland Altman plots), range 3.5-­14.0 % HbA1c
(15-­130 mmol/mol) of the LA1c fraction is close to that of its stable counterpart, which
leads to little or no separation of the two, giving falsely elevated re-
6 sults in some methods that rely on charge separation.9
Our experiments show that there was no significant increase in
5
HbA1c level after incubation with glucose, suggesting that the tested
method is specific for the stable HbA1c fraction. In the same samples,
4
after incubation with glucose, the LA1c fraction increased to fourfold
HbA1c%

3 the basal level (1.1-­4.4%) while the HbA1c level remained stable at
31 mmol/mol (5.0%); the maximum LA1c is around 2% in samples with
2
HbA1c HbA1c over 12% (108 mmol/mol), and the interference from this frac-

1
Labile A1c tion in daily practice is negligible.
Patients with chronic kidney disease suffer uremia, which can pro-
0 vide another potential source of interference in some HbA1c methods,
1 2 3 4 5 6
cHb eluting next to the HbA1c fraction in most commercially avail-
Sample able systems.10 The in vitro experiments with sodium cyanate showed

FIGURE 4 Effect of Labile A1c fraction on Hb A1c that the cHb fraction increased up to 4.0%, while the HbA1c fractions
were stable within ±0.2% of the value observed with the initial sample
(31 mmol/mol, 5.0%), that is, the HbA1c results remain reliable. The
biologic variation (expressed in terms of coefficients of variation CVw
maximum cHb seen in patients with kidney disease is commonly <3%,
and CVg, respectively).6 Derived from 0.9% (imprecision) and 1.5%
and hence, the interference from this fraction in daily practice is negli-
(bias), the total allowable error is set at 3.0%, applying the following
gible. Nevertheless, despite the improved analytical capabilities of the
equation:
√ modern systems,11 the results must be interpreted with caution, due
2 2
TE = 1.65 ∗ CVa + 0.25 ∗ CVw + CVa to the shortened erythrocyte survival in those patients.12
6 of 6 | URRECHAGA

In the same way, we have shown the absence of interference from system for the control of diabetic patients and the detection of Hb
AcHb in the HbA1c measurement. This modified Hb is found in preg- variants in laboratories with high workflow.
nant women, alcoholic patients and even patients being treated with
aspirin; it is not an analytical issue since there is no variation in HbA1c
REFERENCES
quantitation in the presence of high concentrations of AcHb.
The results produced by the system under review were found to 1. The Diabetes Control and Complications Trial Research Group. The
™ effect of intensive treatment of diabetes on the development and
be very well correlated with those obtained on an ADAMS A1c HA-­
progression of long-­term complications in insulin-­dependent diabetes
8180V in Variant mode, making it suitable to use both systems running mellitus. N Engl J Med. 1993;329:977–986.
in parallel. The advantage of this “all in one” mode is that it offers both 2. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood
the high capability for HbA1c analysis of the HA-­8180V Variant mode, glucose control with sulphonylureas or insulin compared with con-
ventional treatment and risk of complications in patients with type 2
and the added value of the screening for β thalassemia carriers; this
diabetes (UKPDS 33. Lancet. 1998;352:837–853.
model improves productivity and the whole laboratory process, allow- 3. The International Expert Committee. International Expert Committee
ing a lean rapid response. Report on the role of the A1c assay in the diagnosis of diabetes.
It is not uncommon to detect an Hb variant in the course of rou- Diabetes Care. 2009;32:1327–1334.
4. www.clsi.org. Accessed April 1, 2016.
tine monitoring of diabetes.13 The additional benefit of the enhanced
5. Sacks DB, Arnold M, Bakris GL, et al. Guidelines and recommenda-
resolution to detect the presence of hemoglobinopathies is useful for tions for laboratory analysis in the diagnosis and management of dia-
two main reasons. First, the presence of Hb variants may interfere betes mellitus. Diabetes Care. 2011;34:e61–e99.
with HbA1c measurement, their influence having been well studied 6. www.westgard.com/database1.htm. Accessed April 1, 2016.
in various systems available on the market.14 On the other hand, the 7. Asberg A, Odsæter IH, Carlsen SM, et al. Using the likelihood ratio to
evaluate allowable total error – an example with glycated hemoglobin
presence of any abnormality is readily detected by the laboratory pro-
(HbA1c). Clin Chem Lab Med. 2015;53:1459–1464.
fessional for further consideration of its implications by the physician, 8. Lenters-Westra E, Schindhelm RK, Bilo HJ, et al. Hemoglobin 1c:
and is relevant for the interpretation of the HbA1c results and for ge- historical overview and current concepts. Diabetes Res Clin Prac.
netic counselling. But in these patients with reduced erythrocyte life 2013;99:75–84.
9. Nathan DM, Avezzano BJ, Palmer JC. A rapid method for eliminating
span, a non-­hemoglobin based method should be used for assessing
labile glycosylated hemoglobin. Clin Chem. 1982;23:512–513.
glycemic control, because the HbA1c level does not accurately reflect 10. Li Q, Ju Y, Jin T, et al. Haemoglobin A1c measurment in patients with
long-­term glycemic control.15,16 chronic kidney disease. Clin Biochem. 2014;47:481–484.
In this study, a number of samples containing common variants 11. Hamouda AERMA, Nabih ES, El Khalek SAA. Novel equation for cor-
rection of glycated hemoglobin and calculation of carbamylated he-
were analyzed for HbA1c. The system adds flags and for HbS, HbC,
moglobin in diabetic uremic patients. Res J Chem Sci. 2013;3:6–11.
and HbD makes a presumptive identification of the variant peak. A 12. Little RR, Rohlfing CL, Tennill AL, et al. Measurement of HbA1c in pa-
drawback of the system is that Hb Lepore co-­elutes in the area of tients with chronic kidney failure. Clin Chim Acta. 2013;418:73–76.
HbA0 and HbA2, disrupting the chromatogram with a characteristic 13. Urrechaga E. Incidental detection of Hb disorders during HbA1c anal-
ysis. J Diabetes Sci Technol. 2015;9:713.
HbA0 peak. HbLepore may elute in both fractions, rendering a chro-
14. Little RR, La′ulu SL, Hanson SE, et al. Effects of 49 different rare Hb
matogram in which no HbA2 peak can be recognized, and making it Variants on HbA1c measurement in eight methods. J Diabetes Sci
impossible to quantitate HbA1c and HbA2. HbE elutes on top of the Technol. 2015;9:849–856.
HbA2 peak. The system will add a flag for variant Hb: the operator will 15. Rhea JM. Impact of hemoglobin variants on HbA1c interpretation: do
we assume too much? Med Lab Obser. 2012;6:8–14.
know that a variant is present because HbA2 is rarely higher than 8%
16. Camargo JL, Gross JL. Conditions associated with very low values
in β-­thalassemia trait. A drawback of the study is that it was not pos- of glycohaemoglobin measured by an HPLC method. J Clin Pathol.
sible to verify the true effect of these variants on HbA1c quantitation 2004;57:346–349.
(when the values are reported), due to the lack of a reference method
to make the comparison.
In conclusion, the study has shown that the ADAMS™ A1c HA How to cite this article: Urrechaga E. Analytical evaluation
8180T system provides a rapid and reliable separation and determi- of the ADAMS™ A1c HA8180T analyzer for the
nation of the relative percentage of different hemoglobin types, par- measurement of HbA1c. J Clin Lab Anal. 2018;32:e22155.
ticularly HbA2. The results obtained are accurate and reproducible, https://doi.org/10.1002/jcla.22155
with CVs lower than the strict quality requirements. This is a suitable