Diabetes represents a huge global health burden and is a leading cause of morbidity and mortality worldwide . It is estimated that up to 50% of people with diabetes are currently undiagnosed, and this is a particular issue in hard to reach settings such as rural communities. The ability to identify and effectively treat people with diabetes is dependent on accurate and timely diagnostic testing, most commonly provided by hospital clinical laboratories, using a range of methods.
Recently, the World Health Organization advocated the use of HbA1c testing for the diagnosis of type 2 diabetes; however, there must be stringent quality control procedures in place to ensure accurate and precise test results and methods must be aligned to the international reference measurement procedure .
Although the Diabetes Control and Complications Trial (DCCT) and the UK Prospective Diabetes Study were seminal trials of the time, more recently treatment targets for people with diabetes have become more individualized with the needs of the patient at the core of decision making . With a patient-centered approach, it is essential that methods for detecting and monitoring diabetes are both accurate and precise to enable high-quality, consistent care.
The International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) Task Force on Implementation of HbA1c standardization (TF-HbA1c) recently advocated sigma-metrics as the model of choice to set and evaluate quality targets for HbA1c . In the laboratory sigma-metrics is a quality management strategy that provides a universal benchmark for process performances. Sigma-metrics places analytical characteristics (bias and imprecision) within the framework of clinical requirements (total allowable error [TAE]). The risk is defined in σ units: a σ of two implies a 5% risk to fail the TAE. The TF-HbA1c has set default risk levels of 2σ for routine laboratories and 4σ for laboratories performing clinical trials . These targets can be universally applied to commercially available HbA1c methods, with comparison to the IFCC primary reference measurement procedure (PRMP) via the secondary reference measurement procedures (SRMPs) as the correct way to determine bias.
The European Reference Laboratory for Glycohemoglobin (ERL) is responsible for the production of IFCC secondary reference material, which enables manufacturers to be traceable to the IFCC PRMP and thus meet the requirement of the WHO and international consensus for the global standardization of HbA1c. Currently, the ERL consists of seven SRMPs certified by the IFCC and five SRMPs certified by the National Glycohemoglobin Standardization Program (NGSP) for the determination of HbA1c in two laboratories and is therefore able to evaluate any new HbA1c method at the highest level [5, 6].
The aim of this paper was to evaluate the Roche Gen.3 Tina-quant HbA1c method on the Cobas c513 and the Abbott Architect Enzymatic method on the Architect c4000 for adoption by the IFCC and NGSP reference networks as certified SRMPs using international quality targets. Beside these two methods, we also evaluated the Tosoh G11.
Materials and methods
Abbott Enzymatic method on the Architect c4000 (Abbott Enzymatic)
The Architect c4000 is a routine chemistry analyzer with photometric, potentiometric and turbidimetric methods available. The instrument is not specifically dedicated to HbA1c and has a maximum sample throughput of up to 800 tests per hour. The total run time for HbA1c is 10 min. The enzymatic method principle has been described before and consists of two separate steps: measurement of glycated dipeptide, obtained by enzymatic cleavage, and measurement of total hemoglobin [7, 8]. Samples can be run using either the whole blood mode or the hemolysate mode, if there is not sufficient whole blood available, using a manual predilution step. The instrument does not have closed tube sampling.
The reagents are ready for use (350 tests per reagent cartridge) and are stable for 50 days onboard the instrument, which needs calibration every 50 days provided there is no change of reagent lot number. During the precision study, three different reagent lot numbers were used so the instrument was calibrated 3× using a single calibrator lot number.
Roche Tina-quant Gen.3 on Cobas c513 (Roche c513)
The Cobas c513 is the successor of the Integra 800 and is a fully dedicated instrument for HbA1c. The sample throughput is 400 patient results per hour, which doubles the throughput compared with the Integra 800. The ready to use reagent is available in a large kit size (500 test per reagent cartridge) suitable for handling high workloads.
The HbA1c determination is based on a turbidimetric inhibition immunoassay for hemolyzed whole blood. Total hemoglobin is measured bichromatically during the preincubation phase of the immunological reaction. Samples can be run using the whole blood mode with closed tubes sampling and the hemolysate mode for small blood volumes .
Onboard stability of the reagent is 4 weeks. The method needs to be calibrated every 28 days or when there is change of reagent lot number. The calibration is a reagent lot specific, which means that not every calibrator can be used with the same values for every reagent lot. During the precision study, the instrument was calibrated 3× using the same lot number for the calibrator and reagent. During the method comparison study, the instrument was calibrated once using the same lot number for the calibrator and the reagent.
The Tosoh HLC-723G11 variant mode (software version V02.00) uses cation exchange HPLC to separate hemoglobin components by different ionic charge. The various fractions of hemoglobin, including HbA1c, are quickly (30 s per sample) separated into six peaks and assayed. A step gradient of three different salt concentrations is used for peak separation and elution. The Tosoh G11 is the successor of the Tosoh G8 and has a reduced run time of 60 s. As a direct result of the shortened run time, there is no longer a specific Hb variant window. The G11 only has a H-VO window (HbAD, HbAS and HbAC all appear in this window), P-HV3 (HbAE) and POO (for unknown Hb variants).
The reagents are stable for 90 days after opening. The instrument requires calibration every 30 days. During the precision and method comparison study, the instrument was calibrated once. Only one lot number of reagent and calibrator was used.
Two samples with an HbA1c value of approximately 48 mmol/mol (6.5%) and 75 mmol/mol (9.0%) were used, according to the CLSI EP-5 protocol (duplicate measurements twice per day for 20 days), to investigate assay imprecision. Aliquots were made from patient samples and stored at minus 80 °C until analysis .
The CLSI EP-9 protocol was performed with 80 frozen samples with HbA1c values over a clinically relevant range (27 mmol/mol [4.6%] to 86 mmol/mol [10.0%]), and the data were used to investigate the bias between the investigated methods and six IFCC and NGSP SRMPs (n=80, 16 samples per day for 5 days, duplicate measurements) . HbA1c value assignment for the patient samples was performed with six IFCC SRMPs (4 of which are also NGSP SRMPs):
Roche Tina-quant Gen.2 HbA1c on Integra 800, immunoassay, IFCC and NGSP certified (Roche Diagnostics, Rotkreuz, Switzerland)
Premier Hb9210, affinity chromatography HPLC, IFCC and NGSP certified (Trinity Biotech, Bray, Ireland)
Tosoh G8, cation exchange HPLC and IFCC certified (Tosoh Bioscience, Tessenderlo, Belgium).
Queen Beatrix Hospital, Winterswijk
Premier Hb9210, affinity chromatography HPLC and IFCC certified (Trinity Biotech, Bray, Ireland)
Menarini HA8180V, cation exchange HPLC, IFCC and NGSP certified (Menarini Diagnostics, Florence, Italy)
Sebia Capillarys 2 Flex Piercing, IFCC and NGSP certified (Sebia, Paris, France).
To check bias, independently of the chosen SRMP, the results of the investigated instruments in the EP-9 procedure were compared with the mean of the six SRMPs, and medical decision point analysis was performed at 48 mmol/mol (6.5%) and 75 mmol/mol (9.0%). The six SRMPs were calibrated with IFCC secondary reference material placing them one step higher in the traceability chain than when using calibrators supplied by the manufacturer.
IFCC monitoring program:
The Roche c513 and the Abbott Enzymatic were both candidates to become an official SRMP in the IFCC and the NGSP network. To become IFCC certified, the methods must demonstrate traceability to the IFCC Reference System by participation in the IFCC monitoring program. This monitoring program consists of 24 interconnected samples (12 samples in duplicate). One sample is analyzed every 2 weeks, and the results submitted via the website. Values are assigned by all of the approved laboratories of the IFCC Network (n=21) . The 24 samples from the IFCC monitoring program were analyzed in one run by the evaluated methods.
Linearity was assessed using the CLSI EP-6 protocol . After adjustment for Hb concentration, patient samples with a low HbA1c value (27 mmol/mol [4.6%]) and a high HbA1c value (148 mmol/mol [15.7%]) were mixed in incremental amounts to generate a series of samples over a broad HbA1c concentration range (n=11). The theoretical HbA1c value and the measured values were compared.
Hemoglobin variant interferences
Interference from common Hb variants HbAS, HbAC, HbAD and HbAE increased A2 (β-thall), and HbF was investigated. Five samples of each variant with different HbA1c values were analyzed in one run. The specific variants were identified using cation exchange HPLC (Menarini HA8180V, Diabetes Mode) and capillary electrophoresis (Sebia Capillarys 2 Flex Piercing, Hemoglobin program). HbA1c values were assigned using an IFCC calibrated boronate affinity HPLC (Premier Hb9210).
The percentage HbF was determined using the Sebia Capillarys 2 Flex Piercing, and HbA1c values of the samples with HbF were assigned using an IFCC calibrated cation exchange HPLC (Menarini HA8180V, Diabetes Mode). The percentages of HbF in the five HbF samples were 4.6%, 6.2%, 15.0%, 18.0% and 39.0%. The investigated Hb variant can be considered as not causing an interference if the results of the Hb variant fall within the deviation of the nonvariant samples distributed around the regression line. A mean relative difference exceeding ±10% in SI units compared to the assigned value was defined as clinical significant.
Four samples with 12.9%, 9.1%, 5.4% and 3.4% carbamylated hemoglobin were made according to a previously published method .
The plasma of six patient samples with triglyceride concentrations of 5.2, 8.1, 9.3, 10.1, 14.6 and 15.6 mmol/L and the plasma of three samples containing 164, 215 and 409 μmol/L bilirubin were used to resuspend pooled red cells from samples with an HbA1c value of approximately 48 mmol/mol (6.5%). The samples were measured in singleton together with the original pooled sample.
Investigating the effect of using fresh versus frozen samples in both hemolysate and whole blood modes and sedimentation of red blood cells
Aliquots were made from nine samples with HbA1c values ranging from, approximately, 26 mmol/mol (4.5%) to 103 mmol/mol (11.6%) and stored at minus 80 °C for 2 days. After 2 days hemolysates were made from the frozen samples and from the primary samples, which were kept at +4 °C. The same fresh whole blood samples were used to investigate the influence of sedimentation of the red blood cells, whereas samples were on the analyzer awaiting analysis. The whole blood samples were thoroughly mixed before loading and analyzed (T=0). After 30, 60, 90, 120, 150 and 180 min, the samples were analyzed again without mixing and the results were compared with the T=0 sample. All samples were analyzed in a single run (fresh whole blood using the whole blood mode, hemolysates made from both frozen blood and the whole blood samples using the hemolysate mode) and compared with each other. The Student two-tailed t-test for paired samples was used to check for statistically significant difference between the results obtained in hemolysate and the whole blood mode and at different times. A p-value <0.05 was considered significant.
Analytical performance criteria
The total allowable error (TAE) for HbA1c has been set by the TF-HbA1c as a default of 5 mmol/mol (0.46% DCCT) at an HbA1c level of 50 mmol/mol (6.7% DCCT), which corresponds with a relative TAE of 10% [(5/50)*100%] in SI units [6.9% DCCT units (0.46/6.7)*100%] with risk levels of 2σ for routine laboratories and 4σ for laboratories performing clinical trials .
Medical decision point analysis:
When two methods are statistically identical, the 95% CI for each y MDP includes the corresponding x MDP. For example, 48 mmol/mol, the diagnostic cutoff value for the diagnosis of diabetes, falls within 46.5–48.1 mmol/mol, the 95% CI around the calculated y so both methods are statistically identical.
IFCC monitoring criteria:
The analytical performance is considered excellent if the mean deviation from the assigned value is <1.9 mmol/mol, CV<2% and linearity >0.9950.
NGSP manufacturer certification criteria:
Thirty seven of 40 results need to be within 6% (relative) of an individual NGSP SRMP to pass certification .
Calculations were performed using Microsoft® Excel 2010 (Microsoft Corporation). Statistical analyses were performed using Analyse-It® (Analyse-It Software) and EP Evaluator Release 9 (Data Innovations) .
For the duplicates in the IFCC monitoring program, CV was calculated with the following formula:
where CVa is the analytical CV, ∆ is the difference between duplicates, n is the number of duplicates and x̅ is the mean of the duplicates.
In the EP-9 study, the Roche c513 and the Tosoh G11 both had a mean bias of approximately −2 mmol/mol (0.2%), and the Abbott Enzymatic method had a mean bias of −0.5 mmol/mol (−0.05%) compared with the mean of the six SRMPs (Figure 1A–C). Medical decision point analysis for the Abbott Enzymatic method at 48 mmol/mol compared to the mean of the six SRMPs was 47.5 mmol/mol (95% CI: 47.4–47.6) and at 75 and 74.4 mmol/mol (74.2–74.5). For Roche c513, it was 46.3 mmol/mol (46.1–46.5) and 72.1 mmol/mol (71.8–72.4). For the Tosoh G11, it was 46.2 mmol/mol (46.1–46.4) and 72.4 mmol/mol (72.2–72.6). Supplemental Table 1 details the results of individual method comparisons with each of the included SRMPs.
Linearity and interferences
All three methods were linear up to 140 mmol/mol (15%) (Supplemental Figure 1A–C). All three methods showed no clinically significant interference from the common Hb variants (HbAS, HbAC, HbAD, HbAE and elevated A2). HbF >6.2% interfered with the Abbott Enzymatic and the Roche c513 methods but not with the Tosoh G11 (Figure 2A–C and Supplemental Table 2). Carbamylated Hb up to 12.9% showed no clinically significant interference on the Abbott Enzymatic and the Roche c513 method. The Tosoh G11 showed no clinically significant interference with HbCarb of 3.4% but HbCarb of 5.4% showed clinically significant interference, and no results were given at an HbCarb of 9.1% and 12.9% because of a “total plate too low” flag. The three investigated methods showed no clinically significant interference of total bilirubin up to 409 μmol/L and triglycerides up to 15.6 mmol/L (Supplemental Table 3).
The effect of using fresh versus frozen samples in both hemolysate and whole blood modes and sedimentation of red blood cells
There was no statistical difference between frozen samples and whole blood samples or samples analyzed using either the hemolysate mode or the whole blood mode. Results of samples that had been stood for 3 h without mixing showed no statistical difference to those which were mixed just prior to analysis (Supplemental Table 4).
Analytical performance criteria
All three methods had a σ>3 using the precision results, from the EP-5 protocol, at an HbA1c value of 46 mmol/mol (6.4%) and the bias calculated at 48 mmol/mol (6.5%) compared to the mean of the six SRMPs. Sigmas calculated using the results of the IFCC monitoring program were >6 for all three methods (Figure 3 and Table 2). Sigmas for the Abbott Enzymatic method, compared with the six individual SRMPs, ranged from 8.4 to 10.0 for the Roche c513 and from 3.2 to 4.4 for the Tosoh G11 from 7.0 to 8.7 (Supplemental Table 1).
Medical decision point analysis
All three methods showed statistically significant difference at 48 mmol/mol (6.5%) and 75 mmol/mol (9.0%) but clinically seen the differences were very small and therefore acceptable.
IFCC monitoring criteria
When using the criteria and samples of the IFCC monitoring program, the three methods showed excellent performance with a mean deviation from the target value of <1.9 mmol/mol, CV<2.0% in SI units and linearity >0.9950. In addition, using this protocol, the σ values were >6 for each method (Figure 3 and Table 2).
All three methods passed NGSP manufacturer criteria compared with the six individual SRMPs (Supplemental Table 1). Pass/fail calculations were based on passing with 74/80 samples.
The three methods also passed the NGSP Secondary Reference Method Certification Criteria, which includes precision (EP-5) and comparison with all SRMPs in the NGSP network (data not shown) .
Overall, each of the three methods performed well meeting the essential performance criteria detailed by the IFCC Task Force on Implementation of HbA1c Standardization, guidance on sigma metrics targets for routine laboratories (σ>2). The Abbott enzymatic method and the Tosoh G11 also met the more stringent criteria for methods used in clinical trials (σ>4). A small shift in results after some calibrations of the Roche c513 resulted in a higher CV in EP-5 protocol and a small bias (CV was 2.0% at 46 mmol/mol, bias was −2 mmol/mol), which may have contributed to the slightly lower σ value observed. Sigmas calculated from the results of IFCC monitoring program were >4 as these CVs were not influenced by calibrations. CV has a bigger impact on the calculation of σ than bias . The Abbott Enzymatic method showed the most robust performance with minimal bias and a very stable CV even with several different calibrations and different reagent lot numbers. This is in line with results of this method in the College of American Pathologist External Quality Scheme .
Reducing the run time for the Tosoh G11 had no influence on the analytical performance in general. However, the disadvantage of shortening the run time is that it is no longer possible to distinguish the different Hb variants from each other as the retention times are very close to each other. Ion exchange methods in particular have shown tendencies to show variable interferences over time due to software/reagent changes. The recent publications of Rohlfing et al.  and Lenters-Westra  show this very clearly. Shortening the run time of the Tosoh G11, like all cation exchange methods, has the potential to make the instrument vulnerable to interference from Hb variants and other substances such as carbamylated Hb. Carbamylated Hb up to 3.4% did not interfere with the Tosoh G11, but an HbCarb of 5.4% showed an clinically significant interference. This might be a problem with patients with diabetes and advanced kidney disease. However, many patients are not allowed to become that uremic any longer so this may only be an issue in poorer and underdeveloped health systems. In addition to potential analytical interferences, such as carbHb, patients with end-stage renal failure are likely to have multiple clinical factors that may affect the validity of HbA1c such as anemia and the use of Epo, which can only be accounted for with good clinical information on the patient. The Tosoh G11 showed no clinically significant interference with the common Hb variants, but HbAC was borderline. The mean relative difference of the five HbAC samples was 9.6%, but two out of the five HbAC samples had a difference >10%. Historically, HbAE has been a problem with the Tosoh analyzers but interestingly showed no interference with the Tosoh G11. This is remarkable as, unlike the HbD peak, the HbE peak does not separate from the HbA0 peak. This means that the instrument incorporates an adjustment factor, which worked with the samples we investigated but might not work with all samples containing HbAE.
Although all methods passed the NGSP criteria for manufacturer certification (where methods are compared against one NGSP SRMP rather than the mean of all), the Abbott enzymatic method performed very well with no samples more than ±6% of the designated SRMP. The results of the Roche c513 show that it is possible to pass the manufacturer certification criteria while failing to meet the sigma metrics criteria (σ>4) for laboratories engaged in clinical trials.
When an offline calibration, using IFCC secondary reference materials, is applied to the results, the small shifts in results, seen with changes in manufacturer’s calibrants, are negated, resulting in σ values >4. The IFCC secondary reference material is one step higher in the traceability chain than the calibrators of the manufacturer and therefore more accurate. However, this is not practical in a routine laboratory setting.
Each of the methods included in this study are produced by manufacturers that can demonstrate traceability of their calibrators to the IFCC PRMP, which complies with ISO 17511:2003 standards detailing how to assure the metrological traceability of patient sample values. This also complies with the WHO criteria for the diagnosis of type 2 diabetes (T2DM) using HbA1c . In addition, the MDP analysis at 48 mmol/mol offers reassurance to clinicians that the instruments perform well at this level.
The Abbott Enzymatic method on the Architect c4000, the Roche Gen.3 HbA1c on the Cobas c513 and the Tosoh G11 were shown to perform well and are suitable for clinical application in the analysis of HbA1c. In addition, a critical aim of the study was to assess the suitability of the Abbott Enzymatic and the Roche c513 as candidate SRMPs, and based on most of the data shown here, they are now official, certified IFCC and NGSP SRMPs [5, 6].
The authors would like to thank Abbott Diagnostics, Roche Diagnostics and Tosoh Bioscience for the contribution of the instruments and reagents for this evaluation. The authors would also like to thank Agnes den Ouden for her contribution of this evaluation by analysis of the samples.
Federation ID. IDF diabetes Atlas, 7th ed. Brussels: Belgium International Diabetes Federation, 2015. Google Scholar
WHO guidelines Use of glycated hemoglobin A1c in the daignosis of diabetes. Available at: www.who.int/diabetes/publications/diagnosis_diabetes2011/en/. Accessed: January 2017.
Inzucchi SE, Bergenstal RM, Buse JB, Diamant M, Ferrannini E, Nauck M, et al. Management of hyperglycemia in type 2 diabetes, 2015: a patient-centered approach: update to a position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2015;38:140–9. CrossrefWeb of ScienceGoogle Scholar
Weykamp C, John G, Gillery P, English E, Ji L, Lenters-Westra E, et al. Investigation of 2 models to set and evaluate quality targets for HbA1c: biological variation and sigma-metrics. Clin Chem 2015;61:752–9. PubMedCrossrefGoogle Scholar
Approved laboratories of the IFCC network laboratories for HbA1c 2016. Available at: www.ifcchba1c.net/network/approved. Accessed: January 2017.
NGSP Network Laboratory Members 2016. Available at: www.ngsp.org/network.asp. Accessed: January 2017.
Teodoro-Morrison T, Janssen MJ, Mols J, Hendrickx BH, Velmans MH, Lotz J, et al. Evaluation of a next generation direct whole blood enzymatic assay for hemoglobin A1c on the ARCHITECT c8000 chemistry system. Clin Chem Lab Med 2015;53:125–32. PubMedGoogle Scholar
Institute CaLS. Evaluation of Precision Performance of Clinical Chemistry Devices; Approved Guideline. NCCLS Document EP05-A. PA: CLSI Wayne, 1992. Google Scholar
Institute CaLS. Method Comparison and Bias Estimation using Patient Samples; Approved Guideline. CLSI Document EP09-A PA: CLSI Wayne, 1995. Google Scholar
Institute CaLS. Evaluation of the Linearity of Quantitative Measurement Procedures: A Statistical Approach; Approved Guideline. CLSI Document EP06-A (2003). PA: CLSI Wayne, 2003. Google Scholar
Weykamp CW, Penders TJ, Siebelder CW, Muskiet FA, van der Slik W. Interference of carbamylated and acetylated hemoglobins in assays of glycohemoglobin by HPLC, electrophoresis, affinity chromatography, and enzyme immunoassay. Clin Chem 1993;39:138–42. PubMedGoogle Scholar
NGSP Manufacturer Certification Criteria. Available at: www.ngsp.org/critsumm.asp. Accessed: January 2017.
EP Evaluator Data Innovations. Available at: www.datainnovations.com/products/ep-evaluator. Accessed: January 2017.
NGSP Protocol Laboratory Network. Available at: www.ngsp.org/protln.asp. Accessed: January 2017.
College of American Pathology External Quality Scheme for HbA1c 2016. Available at: www.ngsp.org/CAP/CAP16c.pdf. Accessed: January 2017.
Rohlfing C, Hanson S, Weykamp CW, Siebelder C, Higgens T, Molinaro R, et al. Effect of hemoglobin C,D,E and S traits on measurements of hemoglobin A1c by twelve methods. Clin Chim Acta 2016;455:80–3. CrossrefGoogle Scholar
The online version of this article offers supplementary material (DOI: https://doi.org/10.1515/cclm-2017-0109).
About the article
Published Online: 2017-04-22
Published in Print: 2017-08-28
Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.
Research funding: Tosoh Bioscience financially supported the study.
Employment or leadership: None declared.
Honorarium: None declared.
Competing interests: The funding organization(s) played no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.
Citation Information: Clinical Chemistry and Laboratory Medicine (CCLM), Volume 55, Issue 9, Pages 1426–1434, ISSN (Online) 1437-4331, ISSN (Print) 1434-6621, DOI: https://doi.org/10.1515/cclm-2017-0109.
©2017 Erna Lenters-Westra et al., published by De Gruyter, Berlin/Boston. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. BY-NC-ND 4.0