Article Text

Download PDFPDF

Haemoglobin A1c analysis in the management of patients with diabetes: from chaos to harmony
  1. A H Berg,
  2. D B Sacks
  1. Department of Pathology, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, USA
  1. David B Sacks, Brigham and Women’s Hospital, 75 Francis Street, Thorn 530, Boston, MA 02115, USA; dsacks{at}


Effective management of patients with diabetes mellitus requires accurate assessments of blood glucose control. The best characterised marker of long term glycaemic control is whole blood haemoglobin A1c (HbA1c). Published clinical trials have identified quantitative and direct relationships between the HbA1c concentration and risks of diabetic microvascular complications. However, in order to practice evidence-based medicine, assays used to measure patient samples should ideally produce values comparable to the assays used in these trials. Numerous assays using chromatographic and immunological detection methods are used around the world. This paper briefly reviews the scientific evolution of HbA1c and its analysis, discusses the reasons why HbA1c assay standardisation is a challenge, describes the approaches that have been adopted to harmonise HbA1c assays, and addresses the current initiatives to standardise HbA1c globally. These efforts have established HbA1c as an essential component in the management of patients with diabetes mellitus and are likely to lead to the use of HbA1c in the screening/diagnosis of diabetes.

View Full Text

Statistics from

Take-home messages

  • Haemoglobin A1c (HbA1c) is the best-characterised marker of glycaemic control in patients with diabetes mellitus.

  • HbA1c is used to adjust therapy in patients with diabetes and to predict risk for the development of microvascular complications.

  • Global harmonisation of HbA1c assays has made the analyte an essential tool in the management of diabetes.

  • Further improvements may enable HbA1c assays to be used for screening and diagnosis of diabetes.

The global diabetes mellitus epidemic affects over 180 million people worldwide. This number is expected to at least double by 2030, and diabetes-associated deaths are projected to increase by 80% within the next 8 years.1 Preventing the complications of diabetes requires that patients maintain blood glucose concentrations as close to “normal” as possible, and this entails vigilant monitoring of glycaemic control. Patients treated with insulin should monitor their blood glucose concentrations by self-administered “finger stick” blood glucose measurements with hand-held meters for the purpose of titrating their insulin doses. However, these periodic glucose measurements do not provide accurate measures of long-term average blood glucose concentrations.2 The best method of assessing long term glycaemic control is measurement of haemoglobin A1c (HbA1c).2 HbA1c values are used by physicians to counsel patients regarding their glycaemic control, to adjust the dose of medications when necessary, and to assess the risk for the development of diabetic microvascular complications.3 Evidence supporting the translation of HbA1c into glycaemic control and long-term risk of microvascular complications is provided by the Diabetes Control and Complications Trial (DCCT)4 and the United Kingdom Prospective Diabetes Study (UKPDS).5 More recent evidence reveals that tight glycaemic control also significantly reduces cardiovascular complications in patients with type 1 diabetes.6 However, for translation of an individual patient’s HbA1c result to risk of complications, the assay should ideally produce values comparable with those obtained in the DCCT/UKPDS. Assay harmonisation is the process of bringing results produced by disparate assay platforms into agreement, and in this paper we briefly review some of the major events that have contributed to the process of bringing this important clinical analyte from chaos to harmony.


In 1955 Kunkel and Wallenious reported the separation of several minor fractions from human haemoglobin by starch block electrophoresis.7 Subsequent analysis by cation-exchange chromatography revealed five early eluting minor peaks that were designated haemoglobins A1a–A1e.8 These fractions correspond to chemically modified forms of haemoglobin A. It was later documented that the HbA1c fraction is increased in people with diabetes mellitus.911 HbA1c is haemoglobin modified by the covalent attachment of glucose to the amino terminus of the β-globin chain.12 This process, termed glycation, is a spontaneous, non-enzymatic chemical reaction. The concentration of glycated haemoglobin in the blood depends on the lifespan of erythrocytes and the concentration of blood glucose.2 Because the lifespan of erythrocytes is ∼120 days, the glycated haemoglobin concentration provides an index of the glycaemic history over the preceding 6–12 weeks.2 3 A proliferation of clinical assays soon emerged. However, these assays initially faced the problems that all new non-standardised assays do, namely lack of inter-assay comparability and limited evidence validating their clinical significance.


The DCCT4 (published in 1993) and the UKPDS5 (published in 1998) are landmark clinical trials that compared the effects of intensive glucose-lowering therapies with conventional blood glucose control on the long-term risks of complications in patients with type 1 (DCCT) and type 2 (UKPDS) diabetes mellitus. Both of the trials documented that better glycaemic control was associated with improved clinical outcome.4 5 Furthermore, a linear correlation between patients’ average HbA1c values and risk of diabetic microvascular complications was observed. In the DCCT, the relative risk of retinopathy or nephropathy was reduced in patients with type 1 diabetes by 39% for each 10% relative decrease in HbA1c (eg, from 8% to 7.2%).4 Virtually the identical quantitative relationship was observed in the UKPDS.13 The demonstration of significant alterations in diabetic risk with only small differences in HbA1c concentrations makes it essential that HbA1c testing is reliable and accurate. Achieving these requirements among different methods and laboratories, however, is not trivial.


Assays to measure glycated haemoglobin became commercially available in 1978.1416 The initial assays were non-standardised, uncalibrated, used diverse analytical approaches, and measured different glycated haemoglobin species. Numerous commercial methods were introduced, but large differences were noted among assays. For example, a study published in 1983 compared eight different assay platforms, and found inter-assay coefficients of variation (CVs) up to 15%, proportional bias as great as −381%, and constant bias as large as +3.6% (in absolute percentage HbA1c).17

The challenges presented by HbA1c analysis begin with the chemistry of this haemoglobin glycation and the methods used to detect modified haemoglobin. In addition to the attachment of glucose to the N-terminal valine (to form HbA1c), haemoglobin may be modified at other sites and by carbohydrates other than glucose.18 Early assays measured HbA1c, HbA1 (which consists of HbA1a, A1b and A1c) or total glycated haemoglobin (which comprises HbA1 plus other glycated haemoglobins). HbA1c is the major component of HbA1, accounting for ∼80% of HbA1.2 Therefore, the value reported is influenced by the specific component of glycated haemoglobin that is measured, resulting in substantially different results from a single sample.

Approaches that have been used to measure glycated haemoglobin include electroendosmosis, enzymatic methods, and isoelectric focusing (all now obsolete), cation-exchange chromatography, boronate affinity chromatography and immunoassays.18 19 These methods separate the glycated haemoglobins from the non-glycated forms based on differences in charge (cation exchange) or by direct selectivity for the carbohydrate modification (boronate affinity chromatography and immunoassays). Each of these families of assays has its own analytical characteristics and challenges. Cation-exchange high-performance liquid chromatography (HPLC) is usually very precise, but is subject to interference from relatively common genetic haemoglobin variants such as S and E traits (HbAS and HbAE).20 (Recent modifications to the assays have substantially reduced interference from HbAS in most methods.) Modified haemoglobins that co-elute, such as carbamylated hemoglobin, may also interfere with assay results.20 21 Boronate affinity assays, which use resins that bind carbohydrate 1,2-cis-diol groups, quantify total glycated haemoglobin, not only the HbA1c fraction. As a result, glycated haemoglobin values can be substantially greater than HbA1c detected by cation exchange.22 Immunoassays use antibodies raised against the glycated N-terminus of the β-globin chain. These assays benefit from the specificity afforded them by antibodies, but their results are only as accurate as their calibrators, and they may be prone to higher imprecision than HPLC methods.23 Haemoglobin variants may also interfere, depending in part on the specificity of the antibodies.

For many years HbA1c assays were not adequately calibrated. This practice produced both significant within and between method variability.17 2426 When manufacturers began to include calibrators with their assays, within-method reproducibility improved, but the use of different non-standardised calibrators introduced a new source of systematic bias between methods.25 Furthermore, many commercial assays used lyophilised calibrators instead of fresh whole blood. Due to precipitation and denaturation, lyophilised materials could be unstable and sometimes these calibrators actually increased assay imprecision.26 27

At the time of publication of the DCCT in 1993, glycated haemoglobin analysis had been in use for approximately 15 years and routine biannual testing in all patients with diabetes had been recommended by the American Diabetes Association (ADA) since 1986.28 Although problems with standardisation of glycated haemoglobin assays had been recognised for over a decade, some commercial assays were still uncalibrated and bias among methods was unacceptably large (College of American Pathologists (CAP) proficiency testing survey data in 1990 show mean assay values ranging from 4.1 to 6.5% and 10.4 to 15.6% at the two HbA1c concentrations tested).29 The state of commercial HbA1c assay standardisation was in disarray. In order to harmonise assays, separate standardisation programmes were implemented in Japan, Sweden, the USA, and later by the International Federation of Clinical Chemistry.


Standardisation by calibration

The simplest approach to harmonise assays is standardisation of calibrator materials. This strategy was introduced by the Japanese Diabetes Society in 1995.30 31 The Japanese Diabetes Society has produced two lots of calibrators. These calibrators are used by manufacturers to calibrate their instruments, and these are then used to measure the secondary calibrators and control materials that are distributed to customers. After introduction of these calibrators, quality control surveys showed that calibrated inter-assay CV across all platforms nationwide was 4.6%.32 The success of this strategy in Japan was due, in part, to the use of only two ion-exchange HPLC assay methods by Japanese clinical laboratories at that time. As a result, not only were Japanese laboratories using a standardised calibration system, but their assay methods were also practically standardised. However, the strategy of “standardisation by calibration” is not effective when many diverse methods are used, and it is not feasible for international standardisation. This is due in part to the differences among calibration formats and materials for the different assay methods. To achieve harmonisation among methods, another strategy is the use of fresh blood comparisons to a designated comparison method.

Harmonisation to a designated comparison method

In 1993, the American Association for Clinical Chemistry established a Glycohaemoglobin Standardisation subcommittee to formulate a strategy to harmonise glycated haemoglobin measurements.33 The National Glycohaemoglobin Standardization Program (NGSP) was organised in 1996 to implement the protocol developed by the American Association for Clinical Chemistry subcommittee. The NGSP consists of a steering committee and a network of laboratories.33 34 The goal of the NGSP is to standardise glycated haemoglobin test results so that values reported by clinical laboratories are comparable with those reported in the DCCT and UKPDS.

The Central Primary Reference Laboratory (CPRL) of the NGSP is a physical and methodological continuation of the assay method used during the DCCT. The CPRL analyses HbA1c with a cation-exchange HPLC assay using Bio-Rex 70 resin.33 The Bio-Rex 70 method was selected for its long-term stability.35 Three primary reference laboratories (PRLs) (two in the USA and one in Europe) serve as back-up to the CPRL and are certified as traceable to the CPRL.33 There are eight secondary reference laboratories (SRLs) (four in the USA and four in Europe), which certify commercially available methods. SRLs work directly with manufacturers to assist them in standardising their methods and in providing comparison data for certification of traceability. Most SRLs use commercially available methods, which are calibrated to match the CPRL. PRLs and SRLs participate in monthly network comparisons. Each network laboratory must meet strict precision and bias criteria to maintain certification. 34

The process by which the NGSP achieves harmonisation comprises three essential components, all of which are performed with whole blood. Two of these, calibration and certification, are conducted with manufacturers of glycated haemoglobin assays who work directly with SRLs. The third is proficiency testing, which is performed by the clinical laboratories that analyse patient samples. The process is as follows: manufacturers calibrate their methods so that results are comparable with those of the NGSP (and therefore the DCCT and UKPDS). To obtain certification, manufacturers analyse 40 patient samples and compare their results with those of their partner SRL. If the 95% confidence interval of differences between manufacturer and SRL are within ±0.85% HbA1c, the method is awarded a certificate of traceability to the DCCT. The certificate, which is restricted to the instrument and reagents used in the certification process, is valid for 1 year. The final component of the process is proficiency testing (also known as external quality assessment). This process evaluates the effects of standardisation and/or harmonisation on assays in clinical laboratories. For example, the CAP GH2 survey, which is conducted twice a year, provides fresh whole blood at three different HbA1c concentrations. The use of fresh whole blood eliminates possible matrix effects. Since 2007, grading of laboratories by CAP is based on accuracy, with NGSP network laboratories assigning the reference value for each CAP sample. The initial limit of ±15% of the target value, which is set by the NGSP, will be narrowed to reduce variability among clinical laboratories.

The NGSP process has significantly improved the performance of glycated haemoglobin analysis.33 This progress can be assessed by comparing 1993 CAP data with 2007 CAP data.29 36 In 1993, the year the DCCT was published, 50% of laboratories reported results as HbA1c; this fraction has increased to 99% in 2007. Moreover, in the 2007 survey, 74% of NGSP-certified methods had CVs ⩽5% for all three samples. NGSP-certified methods are used in many parts of the world, ranging from North and South America to Europe and parts of Asia.

Harmonisation by the designated comparison method was also implemented in Sweden using the Pharmacia Mono S cation exchange resin as the basis for their comparison method.37 This scheme is used in Sweden only.

Standardisation by reference system

In 1995, the IFCC formed a Working Group on HbA1c standardisation. A primary objective of this committee was to develop a true reference method for international standardisation of HbA1c testing.38 Methods were developed to separately purify HbA1c and HbA0 so that they could be mixed in well-defined proportions for use as certified primary reference calibrators. Two reference assays, which combine reverse-phase HPLC with mass spectrometry or capillary electrophoresis, were developed.39 The principle of the assays is as follows: the enzyme endoproteinase Glu-C cleaves an N-terminal hexapeptide off the β chain of haemoglobin. The glycated hexapeptides are separated from non-glycated hexapeptides, and the two are quantified by mass spectrometry or capillary electrophoresis. The IFCC reference system provides metrologic traceability, which is an essential requirement of the European Union directive on in vitro diagnostic medical devices. It is important to emphasise that the IFCC method of HbA1c analysis is labour intensive, time consuming and very expensive. Therefore, the IFCC reference assays are not suitable for routine analysis of patient samples.40 Note that adoption of the IFCC reference system by manufacturers for calibrating instruments will not require clinical laboratories to change the instruments used to measure HbA1c. Analogous to the NGSP, the IFCC Working Group on HbA1c has established a network of laboratories that implement and maintain the reference system. Comparison of multiple samples has revealed that the IFCC reference system produces results that are 1.5–2.0% lower than those measured by the NGSP-, Swedish- and Japanese-designated comparison methods.41

The differences between the IFCC reference system and the NGSP-, Swedish- and Japanese-designated comparison methods introduces a conundrum. As a result of their improved specificity and greater metrologic traceability, the IFCC assays fit the criteria for reference methods better than other HbA1c assays or reference systems. However, introduction of IFCC-harmonised values would substantially change the HbA1c reference interval and target values for treatment (table 1). If this change were implemented, an extensive, prolonged and expensive education campaign would be required for patients and healthcare providers to avoid possible deleterious effects on patient care. Proposals to reconcile these differences include converting IFCC-harmonised clinical assay values to NGSP-aligned values using the IFCC-NGSP “master equation” or simply altering clinical thresholds to align with IFCC values (ie, using HbA1c <5%IFCC instead of <7%NGSP as clinically acceptable HbA1c values).42 However, mathematical extrapolations which relate clinical assays to the DCCT indirectly through the IFCC reference system (instead of direct harmonisation by the NGSP) may not be statistically valid, and may produce clinically significant inaccuracies.43

Table 1 Target haemoglobin A1c values*

In order to resolve this dilemma, a working group termed the American Diabetes Association (ADA)/European Association for the Study of Diabetes (EASD)/International Diabetes Foundation (IDF) Working Group of the HbA1c Assay, was established in 2004 to harmonise HbA1c reporting.44 The Working Group recommended that the IFCC reference method be adopted as the new global standard for calibration of HbA1c assays by manufacturers. Subsequently, the ADA, IDF, EASD and the IFCC published a consensus statement that recommends a compromise which provides the benefits of both the IFCC and NGSP standardisation programmes.45 The proposal is that both NGSP-harmonised results and IFCC-standardised values should be reported simultaneously. NGSP (DCCT-aligned) values should continue to be reported as previously. IFCC results will be reported in International System (SI) units, namely mmoles of HbA1c per mole of HbA0. For example, an NGSP-certified HbA1c concentration of 7% corresponds to 5.3% in original IFCC units, but will now instead be reported as 53 mmol/mol. This difference in units will facilitate distinction between NGSP and IFCC values, thereby avoiding potential confusion that might have resulted from different reference intervals with a single unit.


Harmonisation has had demonstrable effects on HbA1c testing around the world. Implementation of NGSP harmonisation has dramatically reduced bias and imprecision among the most popular commercial assays.46 The number of methods and laboratories that are NGSP-certified continues to increase steadily each year, and there are now 59 different NGSP-certified commercial assays. The ADA recommends that clinical laboratories use only NGSP-certified assays and that they participate in proficiency testing programmes, such as the CAP survey, which use fresh, whole blood.3 The benefits of harmonisation have resulted in widespread adoption of NGSP-certified assays: CAP surveys show that 99% of participating laboratories now use NGSP-certified HbA1c methods.36 Furthermore, the impact of NGSP harmonisation extends beyond US borders: 60% of NGSP-certified laboratories are outside the USA, and the most popular assays used around the world are NGSP certified.

Improved harmonisation of HbA1c assays has transformed this analyte into an essential clinical tool at the centre of the management of patients with diabetes. In order to extend these benefits worldwide, the ADA/IDF/EASD/IFCC consensus statement contends that the time has come for international standardisation of HbA1c testing, and recommends that the IFCC reference assay be used to form the basis of this standardisation. Another development with considerable implications for interpretation of HbA1c results is a large, multicentre, prospective study undertaken by the ADA/IDF/EASD. The study is designed to determine the quantitative relationship between HbA1c concentrations (measured with an NGSP-certified assay) and average plasma glucose concentrations.44 45 If predefined statistical criteria are met by this study, the results of this trial may enable laboratories to report estimated HbA1c-derived average glucose concentrations along with HbA1c values. (This is analogous to estimated GFR calculated from creatinine measurement.) (Publication of the final results of this study are anticipated in the first half of 2008.47) It remains to be determined how HbA1c results will be reported. Reporting three separate numbers (ie, NGSP-aligned HbA1c, IFCC-aligned HbA1c, and estimated HbA1c-derived average glucose) as suggested45 is likely to enhance, rather than reduce, confusion. It is probable that reporting will vary from place to place. Ideally the decision of what to report will be based on discussions between clinicians and clinical laboratorians. Finally, it has been demonstrated in certain well-controlled clinical trials that HbA1c measurement has performed as well as or perhaps even better than measurement of fasting plasma glucose for screening individuals for diabetes mellitus.48 If global harmonisation and low imprecision can be achieved, it may enable the application of HbA1c to be used for screening and primary diagnosis of diabetes mellitus.4850


The authors thank Randie Little (University of Missouri, Columbia, Missouri, USA) for her careful reading of our manuscript and insightful comments, and Rob Krikorian for expert help in preparation of the manuscript.


View Abstract


  • See ACP best practice, p 977

  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles