D:\Biomedica Vol.27, Jul. – Dec. 2011\Bio-1.Doc P.

161 – 165 (KC) IV

COMPARISON OF CATION EXCHAGE HPLC AND IMMUNOTURBIDIMETRIC

METHOD FOR DETERMINATION OF HbA1c

FARZANA YASMEEN, ASIM MUMTAZ, SALEEM-UZ-ZAMAN ADHAMI AND SAJID A. QURESHI

Department of Pathology, University of Health Sciences
Shalamar Hospital and Shalamar Medical and Dental College, Lahore

ABSTRACT
Objective: The objective of this study was to compare and correlate the analytical performance of
D10 Hemoglobin Testing System based on Cation Exchange HPLC and Roche Hitachi 902 Immu-
noturbidimetric method.
Subjects and Methods: A total of 110 patients of Type 2 Diabetes Mellitus were included in the stu-
dy. HbA1c was determined using D-10 Hemoglobin Testing system and Roche Hitachi 902 Analy-
zer.
Results: Both methods showed good correlation with the correlation coefficient (r) of 0.95. Bet-
ween run Coefficient of variance was found to be lower for HPLC system compared to immunotur-
bidimtric method. HPLC method also produces a chromatogram that shows the different hemoglo-
bin fractions, allowing identification of different hemoglobin variants.
Conclusion: In the present study both methods showed good correlation but the D10 HPLC system
provided adequate throughput and improved precision as compared to immunoturbidimetric me-
thod.
Key words: HbA1c, Diabetes mellitus, Cation exchange HPLC, Immunoturbidimetry.

INTRODUCTION ntrol. For the long term assessment of glycaemic st-

Diabetes mellitus is a metabolic disorder which is atus measurement of glycated hemoglobin or HbA1c
characterized by hyperglycaemia with disturbances is now routinely and widely used in clinical practice.
of carbohydrate, fat and protein metabolism. It re- HbA1c is a reaction product of glucose and the N-
sults from defects in insulin secretion, insulin act- terminal valine of the beta chain of haemoglobin.
ion, or both.1-3 Diabetes mellitus has become a ma- The reaction is irreversible and enzyme indepen-
jor health problem worldwide. The number of cases dent. As red cells are completely permeable to glu-
of diabetes among adults was estimated to be ~ 171 cose, the quantity of HbA1c formed is directly pro-
millions in 2000. This number is expected to rise portional to the concentration of glucose. Tradition-
from 171 million to 366 million in 2030.4 ally glycated haemoglobin reflects mean plasma blo-
Diabetes mellitus is associated with complica- od glucose concentration over the previous two –
tions of eyes, kidneys, heart, blood vessels and other three months.11-13
organ systems. Long – term complications of diabe- After the results of diabetes control and comp-
tes include retinopathy with potential loss of vision, lications trial (DCCT) and The U.K. Prospective Dia-
nephropathy leading to renal failure, peripheral ne- betes Study the HbA1c assay has become the gold
uropathy with risk of foot ulcers, amputations and standard measurement of hyperglycaemia.14 Moni-
autonomic neuropathy causing gastrointestinal, ge- toring of HbA1c is recommended by American Dia-
nitourinary, cardiovascular symptoms and sexual betic Association, American Diabetic Federation,
dysfunction.5,6 The results of the major clinical trai- European Association for the study of Diabetes, In-
ls, The Diabetes Control and Complications Trial ternational Federation of Clinical Chemistry and Si-
(DCCT) and The U.K. Prospective Diabetes Study ngapore MOH Clinical Practice Guidelines.15 HbA1c
(UKPDS) showed that the development and progre- is now generally accepted as the single, most promi-
ssion of diabetic complications can be delayed by nent and independent parameter of metabolic con-
monitoring the glycaemic status of patients.7,8 The trol, a risk factor for the development of diabetic co-
tests most widely used in monitoring the glycaemic mplications and is widely used as treatment goal in
status are blood glucose and glycated haemoglo- disease management.16-18
bin.9,10 A number of methods for determination of Hb-
The measurement of blood glucose is of limited A1c have been developed. These methods are based
value for the long term assessment of glycaemic co- on different physical, chemical or immunological

Biomedica Vol. 27 (Jul. – Dec. 2011) 161

FARZANA YASMEEN, ASIM MUMTAZ, SALEEM-U-ZAMA et al

characteristics of the Glycated haemoglobin.2, 19 ntrol and calibration. This technique requires no
As HbA1c is used for patient management, mo- predilution or manual handling of patient’s samp-
nitoring, education and for patient motivation to les. The samples are directly introduced in their pri-
control diabetes so its measurement should be opti- mary tubes following calibrators and control sam-
mally accurate and precise. After the introduction of ples. The instrument draws sample directly from the
HbA1c assays into routine use, it quickly became ap- EDTA vacutainer and all processing of the sample is
parent that different methods produced inconsis- performed internally. Samples are automatically
tent results.20,21 mixed, diluted and injected into the cartridge. The
There are numerous analytical problems associ- analyser delivers a programmed buffer gradient of
ated with glycated haemoglobin measurement. Such increasing ionic strength to the cartridge, where the
as the lack of assay standardization, interference by haemoglobins are separated on the bases of their
Schiff base and the problems related to its measure- ionic interactions with the cartridge material. The
ment in patients with hemoglobinpathies, fetal hae- separated haemoglobins are then passed through
moglobin, renal failure and haemolytic disease. So- the flow cell of the filter photometer, where changes
me drugs that possess strong ionic charges like aspi- in the absorbance at 415 nm are measured. The run
rin can also alter HbA1c results.22,23 time is approximately 3 minutes per sample with a
As different methods for its measurement yield throughput of 20 samples per hour. A sample report
results with undesirable differences, it has become and a chromatogram are generated for each sample.
important to compare the results of various metho- The immunoturbidimetric assay was performed
ds used by different laboratories. In the present on Hitachi 902 analyser under the manufacturer’s
study we compared analytical performance of D10 instructions. First of all, the sample is diluted with
Hemoglobin Testing System (BIO RAD Laborato- haemolysing agent and kept at room temperature
ries) which is based on cation exchange HPLC with for at least 10 minutes. After that the whole blood
Roche Immunoturbidimetric method (Performed processing is performed automatically. The HbA1c
on Hitachi 902). The aim of this study was to evalu- determination is based on turbidimetric inhibition
ate a method which is extremely accurate, precise, immunoassay. In the first step the Glycohaemoglo-
cost – effective and practical that is suitable for rou- bin in the sample reacts with anti-HbA1c antibody
tine use in the clinical chemistry laboratory. (present in the R1) to form soluble antigen – anti-
body complexes. Then R2 (polyhapten) is added, po-
MATERIALS AND METHODS lyheptens react with excess anti-HbA1c antibodies to
This cross sectional study was conducted at Univer- form an insoluble antibody – polyhapten complex
sity of Health Sciences, Lahore. We evaluated Type which can be determined turbidimetrically. The
2 Diabetic patients after approval from institutional Roche Immunoturbidimetric method is standardi-
ethical review committee and informed consent of sed via IFCC reference system and the results obtai-
the patients. A total of 110 patients of diabetes mel- ned by this method are expressed in mmol/mol. For
litus were included in the study. These patients were the conversion of results to % HbA1c a conversion
selected from Shalamar Hospital, Lahore and Ham- factor is installed in the analyser by the manufactu-
za Foundation, Diabetes Centre, Samanabad, Laho- rers.
re. The age of the patients was from 25 to 80 years.
A sample of 5 ml of blood was collected in EDTA va- Control Material
cutainer tube (BD vacutainer K2 EDTA 5.4 mg). The Total imprecision was determined with commer-
samples were kept stored at 2 – 8°C till they were cially available control blood of high and low HbA1c
analyzed within a period of two days. concentration by repeated analysis. Lyphocheck
Diabetic controls from Bio Rad Laboratories were
Techniques used.
HbA1c was measured by D10 Haemoglobin testing
system (BIO RAD Laboratories) which is based on Data Analysis
cation exchange HPLC and Roche Immunoturbi- The data was entered and analysed by using stan-
dimetric method (Roche Hitachi (902) Cobas Sys- dard SPSS software version - 16 (SPSS Inc, Chicago)
tem). for statistical analysis. Mean (± SD) is given for qua-
The Bio Rad D-10 Haemoglobin Testing System ntitative variables. Frequency and percentages are
is the newly introduced fully automated analyser given for qualitative variables. Pearson correlation
based on Cation Exchange HPLC. The dual kit reor- (r) was utilised for determining the strength of lin-
der pack contains whole blood primer, Calibrator 1 ear association between HbA1c measurements by
and 2, calibrator diluent, wash reagent, elusion buf- two methods. The threshold for significance was
fer 1 and 2 and analytical cartridge. The manufac- 0.01 for two – tailed test. Bland and Altman plots
turer’s instructions were followed for the quality co- were used to calculate mean difference (Bias) and

162 Biomedica Vol. 27 (Jul. – Dec. 2011)

 COMPARISON OF CATION EXCHAGE HPLC AND IMMUNOTURBIDIMETRIC METHOD FOR DETERMINATION OF HbA1c

agreement between two methodologies. It was con-

sidered that 95% of all values lying within ± 2 SD in-
dicate good agreement.

RESULTS
A total of 110 samples were analysed for HbA1c esti-
mation. The mean age of patients included in the
study was 51.48 ± 11.4 years (range 28 – 80 years).
There were 34 (30%) males and 76 (69%) females.
The descriptive statistics for both the techni-
ques are shown in Table 1. The mean HbA1c was sli-
ghtly lower for immunoturbidimetric (8.2%) me-
thod than HPLC (8.6%). Low and High level cont-
rols were tested on consecutive five days in dupli-
cate for the between run precision. Each individual
result (N = 15 per level) was taken for the calcula-
tion. Lyphocheck Diabetic controls from Bio Rad
Laboratories were used (Table 2). Figure 1: Bland Altman plot showing good agreement
between D-10 HPLC and Immunoturbidimet-
Table 1: Values obtained by D-10 HPLC and im- ric method.
munoturbidimetric Method.

Mean ±
HbA1c (%) Range
(SD)

D-10 HPLC 8.6 ± 2.07 5.3 – 14.6

Immunoturbidimetry 8.2 ± 3.07 5.2 – 14.2

Table 2: Mean ± SD and between run CVs for

HbA1c determined by D-10 HPLC and Im-
munoturbidimetric method.

Between run
sControl (Level) Mean (± SD)
CVs

D-10 HPLC (Level 1) 5.3 ± 0.11 1.8%

D-10 HPLC (Level 2) 10.1 ± 0.21 1.9%

Hitachi 902 (Level 1) 5.2 ± 1.12 2.9%

Figure 2: Correlation Curve showing good correlation
between D-10 HPLC and Immunoturbidimet-
Hitachi 902 (Level 2) 9.7 ± 1.25 3.1%
ric method.

Bland and Altman plot was used to calculate cient of determination (r2) was 0.90 (Figure 2). Re-
mean difference (bias) and agreement between two gression analysis between D-10 HPLC and Immuno-
methodologies. The plot shows the presence of good turbidimetric method yielded a slope of 0.95 ± 0.30,
agreement between the two methods, 95% of values an intercept of 0.95 X + 0.8 and an S y/ x of 0.43. A
are lying within the ± 2 SD range from the mean significant value of < 0.001 was achieved.
(Figure 1).
The correlation analysis was also done between DISCUSSION
the results obtained by D-10 cation exchange HPLC The availability of the hemoglobin A1c test has been
and immunoturbidimetric method. The immuno- a major advance in diabetic care and its measure-
assay results are plotted on the x-axis and HPLC re- ment has become an integral part for the manage-
sults are plotted on the y-axis. The results showed a ment of diabetes. A relationship has been establi-
good positive correlation between both the methods shed between the glycaemic control and risk of dia-
tested. A significant value of < 0.001 was obtained. betic complications.7,8 However different methods
The correlation coefficient (r) is 0.95 and the coeffi- for its measurement tend to yield results with un-

Biomedica Vol. 27 (Jul. – Dec. 2011) 163

FARZANA YASMEEN, ASIM MUMTAZ, SALEEM-U-ZAMA et al

desirable differences. It has become very important DCA 2000 immunoassay.18 Aside from these issues
to compare the results of various methods used by concerning precision there has been much discus-
different laboratories.11 sion about how the method is standardized and the
In this study, HbA1c results obtained with D10 effects of common interferences from unstable and/
Haemoglobin Testing System (BIO RAD Laborato- or abnormal haemoglobin. The D-10 HPLC system
ries) which is based on Cation Exchange HPLC were participated in the National Glycohaemoglobin Sta-
compared with Roche Immunoturbidimetric me- ndardisation Program and is traceable to the DCCT
thod (Performed on Hitachi 902). The motivation reference method. The National Glycohemoglobin
for this study was to evaluate a method with a gre- Standardization Program (NGSP) is responsible for
ater practicability and improved precision. The Bio the calibration of the HbA1c methods in many parts
Rad D-10 Hemoglobin Testing System is the newly of the world enabling direct comparison to DCCT ta-
introduced fully automated analyser based on Ca- rgets.20,26
tion Exchange HPLC. It consists of a single module The Roche Immunoturbidimetric method is sta-
that provides an integrated method for sample pre- ndardised via IFCC reference system and the results
paration, separation and determination of specific obtained by this method are lower than the HPLC
haemoglobins in the whole blood. Marzullo et al and are expressed in mmol / mol. Therefore a con-
have critically evaluated the analytical performance version factor is incorporated into the software of
of D-10 hemoglobin testing system. They concluded the analyser to convert the results to % unit and eq-
that D-10 is an easy to use automated procedure ualise the results of immunoassay to HPLC. The
which offers accurate Hb1Ac quantification.24 recent recommendation by International organiza-
Both methods have been compared in terms of tions is that the methods for measurement of HbA1c
precision and their overall correlation was also asse- should be standardised according to NGSP and me-
ssed with the help of correlation analysis. In the thod should be traceable to DCCT method. So the
present study we found an interassay CV of 1.8% D-10 cation exchange HPLC method meets this cri-
(HPLC) and 2.3% (immunoassay) for Level 1 control terion.
and 1.9% (HPLC) and 2.5% (immunoassay) for Le- One of the major concerns with various metho-
vel 2 controls. Previously the Hiatchi 911 immuno- ds is that unstable haemoglobin variants may inter-
assy was compared with Diamat HPLC and found fere with the HbA1c measurement. A major disadva-
that the interassay CV for Hitachi 911 immunoassay ntage of these HbA1c immunoassays is that they do
was 3.4% while it was 3.3% on Diamat HPLC. There not consistently detect the presence of abnormal
was an excellent correlation between Bio Rad D-10 haemoglobin variants. Since red blood cells with ab-
HPLC Hemoglobin testing system and Architect Im- normal haemoglobin variants have shortened life
munoassay by Abbott Diagnostic. They found an in- spans, the reported HbA1c value may not reflect the
terassay CV of 1.6% for D-10 HPLC and 2.1% on Im- preceding 2 – 3 months blood glucose control. The
munoassay.25 previous studies also mentioned that the immuno-
Our results are concordant with the previous assay methods do not identify the haemoglobin var-
studies that Immunoassay has higher variation (CV) iants.21,27 On the other hand D-10 cation exchange
than HPLC. The precision is good on both methods HPLC method produces a chromatogram for each
within the medically allowable CV (< 5% recom- patient sample in which the presence of haemoglo-
mended by National Academy of Clinical Biochemi- bin variants can be easily detected by careful exami-
stry and International Federation of Clinical Chemi- nation of chromatogram. The D-10 HPLC not only
stry). While the improved precision which is obser- correctly identifies the haemoglobin variants but
ved in the study of Beaune et al and in the present also withholds reporting the results in the presence
study can be attributed to the recent advances in of markedly decreased amount of haemoglobin A.
HbA1c assay standardization, lower total CVs make Regarding the pre-treatment of sample, the Ro-
it easier to detect significant trends or shifts in a che Immunoassay requires a pre-dilution step while
diabetic patient’s blood glucose control.25 D-10 HPLC method requires no pre-dilution or ma-
In the present study, comparison between Hb- nual handling of the sample and vacutainer tubes
A1c by HPLC and HbA1c by immunoassay was done are directly introduced in the analyser. The assay
on 110 patients ranging from 5% to 14%. Results time for HPLC is 3 minute per sample while it is ten
showed a correlation coefficient of 0.95. Beaune et minutes per sample for Roche Immunoassay. The
al conducted a comparative study on D-10 HPLC Hitachi 902 Immunoassay has higher throughput
and Arcitect immunoassay on 161 samples. They rate but each sample has to be pre-diluted with hae-
found the correlation coefficient of 0.98.26 Similarly molysing reagent and kept at room temperature for
Hawkins RC, found a correlation coefficient of 0.98. at least 10 minutes before measurement. The mini-
They compared the HbA1c results of 110 patients mum sample requirement for Roche Immunoassay
performed on Bio Rad Diastat HPLC and Bayer is 10 µl and for D-10 HPLC system it is 5 µl. For D-

164 Biomedica Vol. 27 (Jul. – Dec. 2011)

 COMPARISON OF CATION EXCHAGE HPLC AND IMMUNOTURBIDIMETRIC METHOD FOR DETERMINATION OF HbA1c

10 HPLC finger stick samples can be collected in 10. Goldstein D, Nathan D, Little R, Peterson C, Lorenz
capillary tube for analysis. No specimen related car- R, Sacks D et al. Tests of Glycemia in Diabetes. Dia-
ryover is present in both methods. The total direct betes care. 2004; 27 (7): 1761-1773.
per test cost of the D-10 HPLC is higher than the Hi- 11. Shiba T, Yano M, Maehata E, Kiyose H, Kotaki Y,
tachi 902 Immunoassay. Fukuzawa N, Hagura R. Standardization of HbAlc
value and its comparison to immunoassay - two years
In conclusion, although there are a number of of experience. Diabetes Res Clin Pract. 1996: 175 –
different methods for measuring HbA1c, there re- 182.
mains a need for an extremely accurate, precise and 12. Miedema K. Towards Worldwide standardization of
practical method that is suitable for routine use in HbA1c determination. Diabetologia. 2004; 47: 1143-
the clinical chemistry laboratory. The D-10 HPLC 1148.
method appears to satisfy this need. It is fully auto- 13. Kilpatrick ES .Haemoglobin A1c in the diagnosis and
mated system that requires no sample preparation monitoring of diabetes mellitus. J Clin Pathol. 2008;
and has a run time of 3 minutes. It is found to be 61: 977-982.
linear in the HbA1c range of 3 – 18%. In addition the 14. Consensus committee. Consensus statement on the
instrument also demonstrates excellent within run Worldwide Standardization of the Hemoglobin A1c
and total CVs. In the present study both methods measurement. Diabetes Care. 2007; 30 (9): 2399-
2400.
correlated very well, however the HPLC method had
15. Liu L, Hood S, Wang Y, Dou C, Datta A, YAun C.
significantly improved precision compared with the
Direct enzymatic assay for %HbA1c in human whole
Immunoassay and has the additional advantage of blood samples. Clin Biochem. 2008; 41: 576-583.
indicating presence of abnormal hemoglobins. 16. Miedema K. Standardization of HbA1c and Optimal
Range of Monitoring. Scand J Clin Lab Invest. 2005;
REFERENCES 65 (suppl 240): 61-72.
1. Alberti KG, Ziammet PZ. Definition, diagnosis and 17. Moridani MY, Verjee Z, Allen LC. Analytical evalu-
classification of diabetes mellitus and its complica- ation of hemoglobin A1c dual kit assay on Bio – Rad
tions. Part 1: Diagnosis and classification of diabetes Variant II: an automated HPLC hemoglobin analyzer
mellitus provisional report of WHO consultation. for the management of diabetic patients. Clin Bio-
Diabet Med. 1998; 15 (7): 539-53. chem; 2003: 317-320.
2. Sacks D. Carbohydrates. Teitz textbook of Clinical 18. Hawkins RC. Comparison of Four Point – of – Care
Chemistry and Molecular Diagnostics.2003; Fourth HbA1c Analytical system against Central Laboratory
Edition: p.837-901. Analysis. Singapore Med J. 2003; 44: 8-11.
3. Jager AV, Travers M. Novel approach for the analysis 19. Nuttal FQ. Comparison of Percent Total GHb with
of glycated hemoglobin using capillary isoelectric fo- Percent HbA1c in Patients With and Without Known
cusing with chemical mobilization. J. Chromatogr B. Diabetes. Diabetes Care. 1998; 21 (9): 1475-1480.
2003: 285-292. 20. John WG, Mosca A, Weykamp C, Goodall I. HbA1c
4. Shera AS, Jawad F, Maqsood A. Prevalence of diabe- Standardisation: History, Science and Politics. Clin
tes in Pakistan. Diabetes Res Clin Prac. 2007; 76: Biochem Rev. 2007; 28: 163-168.
219-222. 21. Gabriele H, Katzensteiner S, Schnedl W, Purstner P,
5. American Diabetes Association. Diagnosis and Clas- Pieber T. Comparative evaluation of three assay sys-
sification of Diabetes Mellitus. Diabetes Care. 2004; tems for automated determination of hemoglobin
24 (1): S5-S10. A1c. Clin Chem. 1997; 43 (3): 511-517.
6. Rehma S, Zia I. Prevalence of microvascular compli-
22. Kilpatrick ES. Glycated haemoglobin in the year
cations among diabetic patients. Pakistan J. Med.
2000. J Clin Pathol. 2000; 53: 355-339.
Res. 2004: 43 (4).
7. Hoelzel W, Weykamp C, Jeppsson J, Miedema K, 23. Chandalia HB, Krishnaswamy PR. Glycated hemoglo-
Barr J, Goodall I et al. IFCC Reference System for bin. Curr Sci. 2002; 83 (12): 1522-1531.
Measurement of Hemoglobin A1c in Human Blood 24. Marzullo C, Minery M. Evaluation of D10 hemoglobin
and the National Standardization Schemes in the testing system for hemoglobin A1C assay. Ann Biol
United States, Japan, and Sweden: A Method – Com- Clin. 2008; 66 (1): 95-99.
parison Study. Clin Chem. 2004; 50 (1): 166-174. 25. Beaune G, Ducrnet J, Ducruet J, Jund J, Favre S.
8. Steffes M, Cleary P, Goldstein D, Little R, Wiedmeyer Evaluation of HBA1C measurement on Architect
H, Rohlfing C et al. Hemoglobin A1c Measurements CI8200 (Abbott Diagnostic). Comparison with HPLC
over Nearly Two Decades: Sustaining Comparable D-10 Bio-Rad assay. Ann Biol Clin. 2009; 67 (1): 101-
Values throughout the Diabetes Control and Compli- 107.
cations Trial and the Epidemiology of Diabetes Inter- 26. Phillipov G, Patrick J. Components of Total Measure-
ventions and Complications Study. Clin Chem. 2005; ment Error for Hemoglobin A1c Determination. Clin
50 (4): 753-758. Chem. 2001; 47: 1851-1853.
9. Jeppsson J, Kobold U, Barr J, Finke A, Hoelzel W, 27. Khuu H, Robinson A, Goolsby K et al. Evaluation of a
Hoshino T, Miedema K et al. Approved IFCC Refe- Fully Automated High – Performance Liquid Chro-
rence Method for the Measurement of HbA1c in Hu- matography Assay for Hemoglobin A1c. Arch Pathol
man Blood. Clin Chem Lab Med. 2004; 40 (1): 78-89. Lab Med. 1999; 123: 763-767.

Biomedica Vol. 27 (Jul. – Dec. 2011) 165