Implications of using HBA1_C as a diagnostic marker for diabetes

Abstract

Diagnosing diabetes now includes a new criterion; hemoglobin A1_C ≥6.5 % which can have significant implications. This review compares the advantages and disadvantages of using HbA1_C as the main diabetic diagnostic test. HbA1_C has greater stability and less variability than plasma glucose measurements but may not always reflect glycemic levels of glycaemia. The present cut off value identifies fewer diabetics than glucose-based criteria. HbA1_C being more convenient could diagnose more patients but this is not yet proven. When choosing a diagnostic test, the limitations of each test must be clearly understood to use appropriate clinical judgment and consider patient preference.

The burden of diabetes

The worldwide epidemic of diabetes is on the increase and affects 387 million people worldwide who have diabetes [1, 2]. In 2004, approximately 3.4 million people died from consequences of fasting high blood glucose [3] and most were from low and middle income countries [4]. The WHO estimates diabetes to be the seventh leading cause of death [4]. By 2025 an additional 205 million people is estimated to have developed diabetes [5] which means that 7.7 % of the world’s population are expected to be diabetic by 2030 [6]. 12 % of global health expenditure has been estimated to have been spent on diabetes in 2010 [7].

Classification of diabetes [8, 9]

Type 1 diabetes is characterized by deficiency in insulin production. The cause is under investigation and to date is not preventable. The patient experiences some or all of symptoms such as polyuria, polydipsia, polyphagia, weight loss, altered vision and exhaustion.

Type 2 diabetes results from inefficient insulin use. It affects worldwide 90 % of people with diabetes and is linked to obesity and a sedentary lifestyle. The symptoms can be similar to Type 1 but are not as prominent. Diagnosis can be several years after onset.

Gestational diabetes has been defined by the American diabetes association [9] as any degree of glucose intolerance with onset or first recognition during pregnancy. It often manifests as Type 2 diabetes and is diagnosed through prenatal screening.

Impaired glucose tolerance (IGT) and impaired fasting glycaemia (IFG) are intermediate conditions before diabetes set in and is a high risk of developing Type 2 diabetes.

Diagnosis of diabetes

Blood glucose has been used to diagnose diabetes from the twentieth century. Fasting plasma glucose (FPG) or the 2-h plasma glucose (2-h PG) value after a 75-g oral glucose tolerance test (OGTT) are used. The cut off values of glucose values differentiate between the distributions of glucose concentrations of nondiabetics.

In 2008, the American Diabetes Association organized the International Expert Committee, the European Association for the Study of Diabetes, and the International Diabetes Federation. One of the main objectives was to diagnose diabetes using HbA1_C in place of the current diagnostic methods. It was added as a diagnostic tool but whether HbA1_C (threshold 6.5 %) was superior to the other tools was left to the clinician’s discretion.

Criteria [10]

• HBA1_C ≥6.5 % OR.

• FPG ≥126 mg/dl [7.0 mmol/L] OR.

• 2-h PG ≥200 mg/dl [11.1 mmol/L] during and OGTT OR.

• Random plasma glucose ≥200 mg/dl (11.1 mmol/L) with symptoms of hyperglycemia.

Chronic hyperglycemia has been linked to diabetic complications [11]. Long term exposure to glycaemia and laboratory measurements of it provides a better clue about the presence and severity of the disease than snapshot glucose concentration values. Long term glycaemia is measured with HbA1_C and has been linked to complications such as retinopathy and the range between 6.6 and 7 % is important [12]. In a study comparing fasting plasma glucose and HbA1_C, the link between HbA1_C and retinopathy was stronger than fasting glucose levels [13].

HbA1_C levels and complications have been correlated in clinical studies of type 1 and type 2 diabetes.

This has led to the use of HbA1_C goals as targets for glycaemic control. Using FPG and 2-h PG to diagnose diabetes has been debated since both measure different physiological measures of acute glucose metabolism [14].

Hemoglobin HbA1_C

Red blood cells contain haemoglobin [Hb] to carry oxygen around the circulatory system. Haemoglobin A1_C was discovered in the late 1950s [15] and Rahbar linked it to diabetes a decade later [16, 17]. Studies connected HbA1_C to blood glucose concentrations [18, 19]. This helped to better understand HbA1_C [20] and for it to be used in judging glycaemic control.

Glycation is a non-enzymatic process by which glucose binds to proteins, it occurs as a two-step Maillard formation that forms an Amadori product; HbA1_C [21]. Glucose rapidly diffuses into the red blood cells by the GLUT1 transporter [22]. The rate of glycation depends on how much is the blood glucose content and HbA1_C levels indicate glycaemia over the last few months.

Other studies have shown the link between HbA1_C and a chronic state of hyperglycaemia. The HbA1_C-Derived Average Glucose (ADAG) study examined this with the help of continuous glucose monitoring and frequent capillary glucose measurements. HbA1_C was found to correlate with the average level of glucose in the blood over the preceding three month duration [23]. Since HbA1_C is linked to red blood cells, factors that affect them also affect HbA1_C. The longer a red blood cell is in circulation, the more chance it can be be glycated. The average half-life of a red blood cell is approximately 30 days [24]. The last three to four months prior to HbA1_C testing contributes to ten per cent of the final result and the average blood glucose in the prior one month contributes to 50 % [25].

Advantages of HbA1c for diagnosis

There are several advantages of using HbA1_C for diagnosis.

Plasma glucose levels vary throughout the day, most prominently after meals. Factors such as exercise, sleep, stress, mental activity all affect glucose. This variability is more prominent in the 2-h PG and postprandial glucose testing values. The current diagnostic criteria depends on the assumption that fasting blood glucose is reliable and relatively reproducible. However, this value is also affected by variation. In particular, exercise (late evening or early morning), a fast of less than 8 h, apprehension about testing and medications affect this value. Many patients do not consume the recommended 200 g of carbohydrate in the days before glucose is tested. The lack of appropriate preparation for testing influences the glucose levels and could reduce the reliability of using FPG for diagnosis. HbA1_C is not affected by sudden glucose level fluctuations or compromised fasting durations. It can be measured regardless of the time of day and in any state of fasting or absorption.

The oral glucose tolerance test is often considered the gold standard for diabetes diagnosis. However this is time-consuming, requires fasting and uses multiple samples. Considering the large number of undiagnosed diabetes cases, a simpler diagnostic test as HbA1_C may improve detection [25].

Fasting blood glucose levels can show more biological variability in two readings but HbA1_C is not as affected [26]. Even if the preparation for glucose testing is optimal, the glucose levels can be incorrect because of pre-analysis instsbility. Tubes used to collect blood do not always contain antiglycolytic substances and if present, there is significant glucose consumption in red blood cells in the first 2 h of sampling because glycolysis is inhibited in its more distal steps by NaF or other preservatives. For as long as the sample is not processed or centrifuged, there is significant glucose loss. Samples reach the laboratory and are processed after some time. Glucose concentrations can show lower than actual values, a process accelerated in high temperatures [27, 28]. HbA1_C does not undergo such significant variability.

Clinical trial data from the Diabetes Control and Complications Trial and the United Kingdom Prospective Diabetes Study show the importance of reducing HbA1_C to reduce microvascular complications in type 1 and type 2 diabetes [23]. HbA1_C has also been linked to cardiovascular complication risks [23].

Observational studies such as DETECT-2 (evaluation of screening and early detection strategies for type 2 diabetes and impaired glucose tolerance) pooled data from over 44,000 subjects in five countries. It showed HbA1_C has a narrow threshold range in which complications begin to increasing alarmingly. This suggests HbA1_C is linked to diabetic specific complications at least as much as FPG or 2-h PG [29, 30].

Standardization of the HbA1_C has always been a main concern but only when it is considered for diagnosis, not for monitoring diabetes as it is widely used today. Standardization helps reduce laboratory biases and is vital for HbA1_C—whether as a monitoring or diagnostic tool or for both. Glucose assays are considered highly reproducible across a variety of laboratories but up to 12 % of subjects have been misclassified in terms of glucose tolerance [31]. This indicates HbA1_C standardization need not hinder its use as a diagnostic tool. A program for worldwide implementation to equip physicians with more reliable information to monitor diabetic patients is underway [32]. Better standardization of HbA1_C can also be developed when HbA1_C is used for diagnosis.

HbA1_C is already used to observe metabolic control of the individual. Treatment and lifestyle is adjusted according to its value. The use of HbA1_C as diagnosis, ≥6.5 %, also allows immediate deviation detection. When measured again, after the initial diagnosis, it indicates if the metabolic control is adequate or not. This makes it easier for the physician to introduce early therapeutic interventions and lifestyle alterations. In individuals with HbA1_C of 5.50–5.99 % and other risk factors such as hypertension, dyslipidemia and obesity, advice can be offered because of the high risk of diabetes and a single HbA1_C value is more valuable than a fasting plasma glucose reading to understand the chronic values of a borderline glucose level that is borderline.

Considering cost, in terms of reagent and equipment HbA1_C is not cheaper than FPG. But there are other not as obvious costs. From the patient perspective, FPG requires an overnight fast and many patients ask a friend or relative to drive to the laboratory. This can sometimes occur during work hours. HbA1_C is not time bound and can be measured after work hours. HbA1_C is also resorted to when FPG is ≥126 mg/dL or ≥7 mmol/L, requiring an additional lab. When HbA1_C is used initially and values are ≥6.5 %, the next stop to monitor the diabetes after diagnosis is already complete and saves analytical and non-analytical costs. When FPG is used to screen for diabetes and a value of 5.6–6.9 mmol/L or 100–125 mg/dl [impaired glucose tolerance] is obtained, the next step is an OGTT to elucidate glucose tolerance. This needs several laboratory and adds to all costs. In these cases, where a large population is at risk or already has diabetes, HbA1_C instead of FPG would provide a cost effective diabetic diagnosis and risk assessment without additional, time consuming and costly testing [33].

The individual glycation variation could be an added on benefit in assessing complication risk. The hemoglobin glycation index; the difference between observed and predicted HbA1_C levels, can be linked to an increase in risk of developing complications [34].

The epidemiologic landscape of diabetes would alter if HbA1_C is picked as the diagnostic tool of choice. However a report exists indicating that using HbA1_C instead of FPG would not significantly change this scenario and 97.7 % of subjects would not be categorized differently. Additionally, approximately half of the subjects with a FPG ≥7 mmol/L (≥126 mg/dL) had HbA1_C values of 6.00–6.49 %. This meant that strict monitoring and intervention was required.

The gold standard, at present, is the combination of FPG, 2 h-PG and HbA1_C but is not feasible or recommended for a variety of reasons especially on a large scale. HbA1c appears more feasible.

Issues with HbA1c to diagnose diabetes

There has been much debate about using HbA1_C to diagnose diabetes. One concern is that HbA1_C is a measure of gylcation and not a direct measurement of glycaemia. There are many physiological and pathological states influencing HbA1_C. HbA1_C is altered when three basic mechanisms are affected; amount of glucose entering red blood cells, red blood cell age in circulation and the glycation rate.

There is heterogeneity in the glucose concentration gradient across the red blood cell membrane and the average lifespan of red blood cells, even in healthy individuals. In pathological states affecting red blood cell turnover, such as an increase or decrease in erythropoiesis with or without haemloysis, HbA1_C levels are affected. Care has to be taken when analyzing HbA1_C results from patients who have liver or renal failure or haemoglobinopathies.

HbA1_C has also faced concerns that it will identify a population set of diabetics different from those that have been identified by measurements of plasma glucose. The objective of any diagnostic criteria is to identify people at risk of complications and would benefit from interventions, not to detect people with particular plasma glucose values. However, multiple criteria for diagnosing can create confusion.

HbA1_C may vary with age and ethnicity [35]. People of African ethnicity may have higher HbA1_C levels than Caucasians [35, 36]. Evidence indicates HbA1_C levels correlate differently with plasma glucose amongst South Asians or Inuits compared to Caucasians [23]. The reasons are unknown but differences in genetic determinants of haemoglobin glycation, the glucose balance between extracellular and intracellular environment and red blood cell turn over are considered [37]. This might necessitate ethnicity specific thresholds to use HbA1_C for diagnosis.

Abnormal haemoglobin levels could affect HbA1_C values and depending on the method used, not all assays can differentiate between pathological and glycated haemoglobin. Different races exhibit different frequencies of haemoglobin disorders; 10 % of African-Americans have a haemoglobin C trait that could interfere with HbA1c assay [38]. The National Glycohemoglobin Standardization Program website (NGSP, http://www.ngsp.org/interf.asp) provides current information on how the most common haemoglobin variants such as HbS, HbC and HbE interfere with twenty methods that measure HbA1_C, although some methods remain unaffected. However, if the patient is a homozygous or double heterozygous for a haemoglobin variant then no method can measure HbA1_C because it is not present.

Standardization of measuring HbA1_C has been improved by adopting the NGSP protocols but not all countries have achieved this. Plasma glucose concentration is also difficult to assay with consistent accuracy. It is estimated that up to 12 % of patients can be misclassified in the diagnosis of diabetes due to laboratory instrument error in measuring glucose [31, 39].

Depending on the study, the probability of a person meeting the plasma glucose criteria for diabetes and having an HbA1_C ≥6.5 % can range between 17–78 % [40]. The discordance is wide and varies between different populations [23]. The different methodologies used can partly explain the difference but even so if HbA1_C becomes used globally, the epidemiology of type 2 diabetes mellitus would be altered.

The diagnosis of diabetes based on HbA1_C misses a large proportion of asymptomatic early cases of diabetes that are only detected by the OGTT. FPG has shown to be more sensitive than HbA1_C [41]. People with impaired glucose tolerance cannot be detected by HbA1_C and it is this group that can benefit most by diabetes prevention programs [42, 43]. This is of particular importance when considering evidence that this group has an ~40 % chance of increased mortality [44]. Early diagnosis of diabetes is of the utmost importance.

Studies show using HbA1c and FPG and/or OGTT identify different diabetic groups [45]. HbA1_C ≥6.5 % identifies 30–40 % of previously undiagnosed diabetics but FPG estimates 50 % and the OGTT detects 90 % of diabetics [46, 47].

A large proportion of newly diagnosed based on current glucose criteria have HbA1c ≤6.5 %. In the Finnish Diabetes Prevention study, the sensitivity of HbA1c ≥6.5 % to diagnose diabetes was 39 % so 61 % of newly diagnosed diabetics had an HbA1c ≤6.5 % [48]. This means that 61 % of patients with diabetes could be diagnosed later and risk complications.

The decision on a single, optimal threshold for HbA1_C values has been more difficult than expected. The expert committee recommended a cut-point of 6.5 % and was based on evidence that in diverse populations, retinopathy seemed to increase above this particular level [49]. However, this recommendation was based on an unpublished analysis of data derived from the DETECT-2 study and cited as a personal communication by one of the committee members, Stephen Colagiuri [9, 50]. Another article that discussed the DETECT-2 study clearly pointed out that “currently available statistical and mathematical models were not able to identify a clear cut-point. It also states that the cut-points for diabetes are a combination of the available evidence and expert consensus [51].

Statements such as these summarize the dilemma posed when a cut-point for diabetes is selected which is based on retinopathy. The expert committee [49] pointed towards older studies that analyzed the relationship of retinopathy to FPG, 2-h PG and HbA1_C [52, 53]. These studies observed that retinopathy appeared to increase when HbA1_C levels of 6 % in the National Health and Nutrition Examination Survey (NHANES) III, 6.2 % in the Pima Indian Study and 6.3 % in a study done in the Egyptian population [54]. The 1066 people studies as part of the 2005–2006 NHANES cohort showed that increased retinopathy at A1C ≥5.5 % [55]. A Japanese study has also suggested that the optimal cut-point should be 5.3–5.5 % [56]. A South Asian study pointed out the concept of a threshold effect. The authors of this study report a continuous linear relationship between HbA1_C and the most common microvascular complications. It was found, in the population studies, that mild to moderate retinopathy was rare below an HbA1_C of 6.6–7 %. This study concluded that an HbA1_C of 6.6 % may represent an optimal cut-point for diagnosing diabetes based on the risk of mild to moderate retinopathy but not for other microvascular complications [12]. A review of three population-based cross-sectional studies of the relationship between retinopathy and FPG concluded that there is “inconsistent evidence” for a uniform glycemia threshold for prevalent and incident retinopathy. The analyses also suggested the existence of a continuous relationship between glucose levels and retinopathy, including moderate retinopathy [57]. The Multi-Ethnic Study of Atherosclerosis (MESA), also examined the relationship between HbA1_C and retinopathy and found a continuous, non-inflected relationship between these two parameters with no threshold effect [57].

The discrepancies amongst studies about the threshold effect could be partly due to older studies relying on direct ophthalmoscopy or single field retinal photographs which are not as reliable as new methods for detecting early stage retinopathy [58]. There are also ethnic groups that show a higher prevalence of retinopathy at relatively lower glycemia levels. An Australian study showed that individuals who had not been diagnosed with diabetes and had retinopathy, had a higher risk of being diagnosed with diabetes 5 years later [29]. This suggests that retinopathy can start at glucose concentrations below the threshold for the diagnosis of diabetes. The association of retinopathy with “pre-diabetes” was also observed in the Diabetes Prevention Program [59] and in the 2005–2006 NHANES study; both studies found that 8 % of individuals with FPG levels below the diagnostic threshold for diabetes had retinopathy [60].

HbA1_C also has issues when it is correlated with glycaemia in ‘non-diabetic’ populations. The Dutch New Hoorn study found that HbA1_C correlated well with FPG and 2-h PG in people with diabetes diagnosed by criteria before 2010. However, the correlation of HbA1_C with FPG and 2-h PG in the general population was significantly lower. The authors concluded that HbA1_C is affected by aging, genetic traits, differences in red blood cell lifespan especially for non-diabetic individuals [61].

Glyco-oxidation products and glycated products are present on α and β globins but are more abundant in diabetics. The methods measuring glycated haemoglobin cannot distinguish between glycated and glyco-oxidated species on HbA1_C [62]. In type 2 diabetics with chronic complications there is a higher abundance of glyco-oxidation products which might affect the diagnosis of diabetes [63].

2-h PG and HbA1_C are better predictors than FPG for future cardiovascular complications but derived by multivariate analysis, only 2-h PG remains statistically important [23].

In terms of expense, HbA1_C is an expensive test. Many patients also require other tests such as lipid or hepatic profile during the fasting state, at the same time, and adding glucose is not a major issue. Most laboratories can test before average working hours.

HbA1_C is unaffordable or even unavailable in many low and middle income countries [64, 65]. Setting HbA1_C as the gold standard for diagnosing diabetes creates a divide between societies that have or have no access to HbA1_C testing.

Conclusion

Diabetes places a heavy global burden of disease and the definition of it is of the utmost importance. HbA1_C has many attractive attributes and can be used as an appropriate diagnostic test when an internationally standardized assay is used and clinical conditions that affect it are accounted for. When a diagnostic test is used for diabetes, the limitations of the choices currently available must be clearly understand. The use of clinical judgment and patient preference are needed to select the appropriate test.

Until the challenges against using HbA1_C have not been better met, using HbA1_C as the sole diagnostic test could potentially lead to misclassification and errors. It must be used thoughtfully and preferably in combination with traditional glucose criteria when screening for and diagnosing diabetes.

Ethics policy

This article does not contain any studies with human or animal subjects performed by any of the authors.

Conflict of interest

The author declares that they have no conflict of interest.