College of American Pathologists
Printable Version

  A1c: added oomph to diagnosing diabetes


CAP Today




July 2010
Feature Story

Anne Paxton

“A1c” isn’t a company listed on the New York Stock Exchange. But if it were, the word on Wall Street would probably be “Buy.”

Contrary to some expectations, in its new clinical practice recommendations released in January, the American Diabetes Association opted not to crown hemoglobin A1c as the preferred laboratory test for diagnosing diabetes. But in advocating the use of the assay as an additional means of diagnosing the nation’s fastest growing chronic disease, the ADA made a move that will affect millions of people.

Clinical laboratories can expect that the ADA’s Standards of Medical Care in Diabetes—2010 will markedly increase their HbA1c test volume, says Robert Christenson, PhD, professor of pathology and of medical and research technology at the University of Maryland School of Medicine and director of the core laboratory, University of Maryland Medical Center. And he believes that’s all to the good. “Given the ADA’s stature, and the other parties involved, I think the recommendation is a very exciting development.”

While there is little controversy over the value of HbA1c in diagnosis, the recommendation could significantly alter the number of people deemed to have diabetes or pre-diabetes, and debates continue over reference ranges and other issues. In the meantime, diabetes-testing experts say, clinical laboratories need to be prepared for the preanalytic and analytic issues that A1c’s new status will bring.

Long a standard test for monitoring diabetes, A1c now officially takes its place alongside the traditional fasting plasma glucose and the oral glucose tolerance test (OGTT) as a valid and reliable means of diagnosing diabetes and identifying pre-diabetes. A1c, of course, has a huge advantage over these two other tests: Rather than require patients to fast for two or even eight hours, it can be performed without advance warning. Says Richard M. Bergenstal, MD, ADA president of medicine and science: “A1c offers an amazing opportunity to get people diagnosed where they are, when they are.”

In its new criteria for diagnosis of diabetes, the ADA now approves an A1c of 6.5 percent or higher, in addition to the previously existing criteria (a fasting plasma glucose of 126 mg/dL or higher, or a two-hour plasma glucose of 200 mg/dL or higher during an OGTT). A random plasma glucose of 200 mg/dL or higher in a patient with classic symptoms of hyperglycemia is also considered diagnostic. In addition, the ADA recommends that an A1c of 5.7 percent to 6.4 percent be considered an indicator that blood glucose levels are in the pre-diabetes range.

Dr. Bergenstal, executive director of the International Diabetes Center at Park Nicollet Health Services in Minneapolis, does not consider the ADA’s inclusion of A1c in the standard of care to be a radical change. “All we’re doing is saying we’ve had three ways to make a diagnosis of diabetes—fasting glucose, OGTT, and random glucose—and we want to add a fourth as another possible means. We look at it as a move that’s just logical and will allow more people to be screened for diabetes.”

The logic has to do with the A1c assay’s many advantages over glucose testing. “In some ways, A1c is a more reliable test,” says Rich Furlanetto, MD, PhD, medical director for endocrinology at Quest Diagnostics Nichols Institute, Chantilly, Va., and medical advisor for HemoCue. “From ¬≠laborator¬≠ian s’ point of view, it has less biological variability; there are fewer preanalytical factors to interfere with it. Patient preparation is easier and intercurrent illnesses don’t affect the result. You could be in a doctor’s office with a cold, and we couldn’t do a glucose reliably, but we can do an A1c that is reliable.”

Unfortunately, there is a lot more variability in glucose than there is with other laboratory tests, Dr. Furlanetto notes. “Glucose jumps all over the place in normal people during the day. The samples have a lot of biological variability, and then, unless the samples are run immediately, they start deteriorating even when they’ve been collected properly. There’s almost no way of preventing glucose deterioration, and that adds to the complexity.”

Using A1c mitigates a lot of the limitations of glucose testing, Dr. Christenson says, though it comes at a cost. “A1c is going to be more expensive than garden variety glucose.” Depending on the carrier, a fasting glucose may be $5, as opposed to $15 for an A1c, he notes. “So it can be a big difference.” Nevertheless, he emphasizes, better accuracy for diagnosing diabetes is important. “It’s a pretty devastating thing to be told you have diabetes. So I think the wider use of A1c for diagnosis is a leap forward.”

Determining the effect of the ADA’s recommendations on physicians’ choice of tests for diagnosis is complicated by the fact that many doctors already use A1c in this way. A 2005 survey of physicians at the annual meeting of the American College of Physicians found that 93 percent screen for diabetes routinely, 49 percent use A1c for screening, and 58 percent use A1c for diagnosis.

“A1c is not going to replace glucose testing,” Dr. Christenson says. “It will be an add-on because glucose is a component of the basic metabolic profile. But many clinicians already choose to use A1c largely because of the fasting element with glucose. If you have a 10 AM appointment and you’ve fasted—especially for folks with diabetes—it’s a very big problem.”

A1c has been well recognized as the standard for diabetes care, Dr. Furlanetto says, but many people who have diabetes remain undiagnosed. “There’s been a continual effort on the part of physicians to get these patients diagnosed, and when you heard about what A1c did, that it’s an average, the reasonable question was: Why can’t we use it for diagnosing? And a lot of physicians used it for that.” Dr. Furlanetto presented a study examining A1c results at the ADA meeting in 2007. “About half of the A1cs submitted to Quest did not have an ICD-9 code indicating the patient had diabetes. Those test orders may have been coming from physicians who did not code them as being from someone with diabetes, but there were substantial numbers obviously being done for screening reasons. And while there were lots of data about A1c levels in diabetes patients, the information to correlate A1c with a diagnosis of diabetes wasn’t there. The new guidelines provide this information.”

David B. Sacks, MB, ChB, medical director of clinical chemistry at Brigham and Women’s Hospital and associate professor of pathology at Harvard Medical School, made a similar finding in reviewing practices at his hospital about a year and a half ago. “Just looking at the distribution of A1c measurement, we found 50 percent of them were below 6.5 percent and about a third were below six percent. We assumed that those with A1cs below six percent were probably being screened, because there are very few people with diabetes who have A1cs below 6.1 percent.”

The problem has been that without recommended cut-points, clinicians haven’t known what values to call diabetes or not diabetes, and they were looking for guidance, says Dr. Bergenstal. “They were probably using that test more than any other test because fasting is really not very convenient—but every doctor was using a different interpretation of results.” Without official cutoffs, Brigham and Women’s Hospital physicians were operating in a diagnostic gray area, Dr. Sacks found. “If one asked them what cutoffs they were using, they sort of shrugged their shoulders.”

The Food and Drug Administration has approved A1c only for monitoring, not for diagnosis, but this does not seem to be a consideration for the ADA, Dr. Christenson says. “It does not appear that whether the FDA cleared it or not for the indication was really an issue in the ADA’s decision. Seems that they looked at the clinical data, and if it were approved by the FDA and they didn’t think it appropriate, they would go the other way as well.” Physicians can be reluctant to use tests that the FDA hasn’t approved, he concedes. “Having said that, a clinician can order a test for an off-label use when indicated, just as they can use a drug for an off-label use.”

The major change that has made A1c become acceptable for diagnosis is the degree of standardization, Dr. Bergenstal says. “Since 1997, there’s really been an extensive effort clearly throughout the U.S. to do the NGSP [National Glycohemoglobin Standardization Program] certification. All labs have the ability to standardize their methods and report their results out based on the Diabetes Control and Complications Trial [DCCT], and it’s estimated that more than 90 percent of all labs now are standardized to do that.” The assays have gotten better too.

Beyond that, he says, just in the past two years the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) has gone a step further with another specific, more precise way to measure A1c. “So we now have ways we’re going to continue this standardization globally for those labs that wish to participate.”

Nichole L. Korpi-Steiner, PhD, DABCC, associate director of core laboratories at UMass Memorial Medical Center, Worcester, also notes that proficiency testing data have demonstrated the effectiveness of the method standardization. “CAP Surveys have shown great improvement in A1c results,” she says.

The CAP has taken two steps in its Surveys that have helped improve A1c assays, Dr. Sacks says. “The first was that CAP switched to accuracy-based grading for its Surveys. In the past, if a laboratory’s method had bias, the peer-group mean would pull them through. With an A1c assay, for example, if you used method A, this meant your grading was based on the mean of everybody else who used method A. This would have major implications for treating people with diabetes because it could tell you a seven is an eight.”

The second step the CAP took was to tighten the criteria for acceptability of grading and for passing. “So initially the criterion was 15 percent. Now this year it is eight percent, and that’s forced the manufacturers to be a lot more accurate, because people are going to fail if they are outside CAP acceptance criteria.” In fact, Dr. Sacks has met with some manufacturers who have said some of their methods are never going to achieve accuracy to eight percent, so they have basically phased them out. These changes in standards by the College have had a sizable influence on improving methods standardization, accuracy, and precision, Dr. Sacks says.

When it added A1c to the diagnostic roster, the ADA decided to dial back the key recommendation of the International Expert Committee whose members—appointed by the ADA, the European Association for the Study of Diabetes, and the International Diabetes Federation—began meeting in 2008 to consider the current and future means of diagnosing diabetes in non-pregnant individuals.

The Expert Committee’s members had reached a stronger consensus in their 2009 report: that although no assay can be considered the gold standard in diagnosing hyperglycemia, the A1c assay “may be a better means of diagnosing diabetes than measures of glucose levels.” A1c, the committee said, has several technical advantages, both preanalytic and analytic, over laboratory measures of glucose, and it provides a reliable measure of chronic glycemia, which correlates with the risk of long-term diabetes complications.

The ADA recommendation was scaled down from that of the International Expert Committee for practical reasons, Dr. Bergenstal says. “They thought A1c was such a good test that the others should be done away with. If you do an A1c today and one tomorrow, there will be much less variation than in glucose or glucose tolerance between today and tomorrow. So it’s actually more precise and a better assay than all the others, they said.”

“But the ADA took the Expert Committee’s very nice document and said, ‘Our job is to put this into practice for the large community—to make it sensible and practical on implementation. So we think it would be better to add the test, not get rid of the others.’”

A concern of countries outside the U.S. over the International Expert Committee’s recommendation to replace fasting glucose with A1c, Dr. Bergenstal says, was that in many parts of the rest of the world, good operating A1c tests are not available. “They were scared they would be considered second-class citizens if they were doing fasting glucose, so the ADA took that to heart. It said nothing is wrong with fasting glucose if it’s done carefully in a good lab. A1c is a bit more accurate and a better test, but if you don’t have those facilities, it’s fine.”

Despite opting against A1c as the preferred test, the ADA has shifted ground considerably with its 2010 recommendations. As recently as 2003, the organization went back and re-evaluated its 1999 position that A1c should not be used for diagnosis, and its followup report concluded it was “still premature” to add A1c. Dr. Sacks says: “The World Health Organization and the IDF also issued guidelines in their most recent published guide in 2006 saying A1c was not suitable. But in 2009, the International Expert Committee’s report completely changed the ADA’s mind.”

David M. Nathan, MD, professor of medicine at Harvard Medical School and director of the Diabetes Center at Massachusetts General Hospital, said the International Expert Committee chose 6.5 percent as a reasonable A1c cut-point on the basis of the association with risk for retinopathy. In order not to over-diagnose diabetes, “The Expert Committee decided to err on the side of specificity rather than sensitivity,” Dr. Nathan said in a March 2010 webinar sponsored by the American Association for Clinical Chemistry. As reassuring supporting evidence, he noted that at A1cs of less than 6.5 percent, moderate retinopathy is virtually absent.

In the end, the ADA also decided to use the detection of microvascular complications as a cutoff to settle the issue—though the issue isn’t yet settled, Dr. Bergenstal says. It was decided that 6.5 percent was appropriate because it is the level “at which you have a high probability of detecting moderate retinopathy; therefore, they said that’s the cutoff for diabetes.”

Still, there is controversy among endocrinologists about the reference ranges for A1c and other issues. Shortly after the ADA released its recommendations, the Endocrine Society issued a statement expressing support for the addition of A1c as an option to diagnose diabetes because of A1c’s close correlation with microvascular complications and its ease of use. But, the society said, clinicians and the population at large will need to understand certain caveats.

The major one: Since A1c fosters a more restrictive definition of diabetes, “there will be a significant number of patients who would have been diagnosed with diabetes with plasma glucose criteria who now would not be diagnosed with diabetes using A1c.” That may have an impact on the number of patients who would be treated according to the society’s recommendations for cardiovascular risk management.

Endocrinologists overall supported the recommendations, Dr. Bergenstal says. “Everybody pretty much agreed with the 6.5 percent cutoff for diagnosing diabetes. As you’d expect, there was some controversy on the pre-diabetes or ‘at risk’ guidelines. Some say we should have the glucose test in addition, some wanted to qualify it other ways, and some wanted to study it further to set the right number.” He thinks more consensus will be built over time on the at-risk category.

“Several studies have demonstrated that A1c is a heterogeneous molecule,” Dr. Korpi-Steiner points out. “And there are demographic factors that can influence results such as age, gender, and race. So although the general range of A1c is four to six percent in healthy populations, it’s very difficult to determine what a healthy population is these days. In my mind, further clinical studies are needed to better define that.”

It has proved challenging to set reference ranges for A1c as compared with glucose, Dr. Christenson says. “With glucose measurements, there’s a ‘hockey stick’ distribution. With an outcome of retinopathy, the curve is pretty flat at the lower end, up to around 100, then it goes up all of a sudden between 100 and 230. That’s one reason why the ‘pre-diabetic’ designation emerged, because at 85 you’re at about the same relatively low risk as at 75, while if you’re at 105 you’re at greater risk of diabetes than at 95.”

By contrast, though diabetes is defined as “an abnormal state of glucose metabolism associated with heightened risk for relatively specific long-term complications,” the glycemic levels detected by A1c and glucose are distributed in a continuum, Dr. Nathan noted in his AACC Web presentation. Except in some very high prevalence populations, he said, the tests can’t readily separate “normal” from “abnormal.”

Within constraints like those, the ADA set its reference ranges based on as much hard science as possible, Dr. Bergenstal says. “When you look at the diagnosis of diabetes, there isn’t one number that all of a sudden dramatically indicates it, so we looked at the risk for retinopathy, which is really specific for diabetes. In all the studies of what A1c levels see a jump in incidence in eye disease, it was pretty clear that 6.5 percent was the number. So the diagnosis of diabetes really should be based on the risk of developing one of the hallmark complications of diabetes.”

The cutoff for “high risk” or pre-diabetes was a little different, Dr. Bergenstal says. “The 5.7 number says if you don’t do anything, you have a high likelihood of developing diabetes over the next five years. Not that you have to start medication, but just start counseling about lifestyle, habits, and exercise. That’s the number all the literature said at least doubled your risk and probably more than that.” There were differences of opinion, “but we felt we got it right, because half said it should be higher and half said it should be lower.”

Generally, people who object to the use of A1c for diagnosis do so for two main reasons, Dr. Sacks explains. “One is that there are all these other things that affect A1c. For example, published evidence suggests that age and race affect A1c concentration. However, there is no consensus that we should have age-related differences or race-related differences for monitoring diabetes, and the clinical significance of these variations remains to be established.”

The second reason is that A1c is more expensive. “I was recently in Tanzania, and there were people from 25 different countries attending the course, and they indicated that in many of their countries you couldn’t get A1c easily.” Running a single sample of glucose is less costly than running an A1c.

But the key advantage of A1c in diagnosis and screening—that the person need not be fasting—essentially brings the cost down, Dr. Sacks points out. “The ADA recommends diabetes screening for anyone 45 or older. That’s particularly relevant when a person that age shows up for a routine physical. You can say to him or her, ‘We need to do blood work,’ and we can measure A1c, whereas if you use fasting glucose, you have to tell the patient to return the next morning.”

“So one could say from a cost perspective to the patient, it’s more expensive to have to return the next day fasting, just in time lost from work and so on.” One could also argue, Dr. Sacks says, that paying for the complications of the 40 percent of people who have diabetes but haven’t been diagnosed would be much more costly than employing a convenient A1c test to reduce that percentage.

The A1c reference range issues remain a significant problem, Dr. Furlanetto says. The Endocrine Society statement, citing extensive data collected over the years by the federally sponsored National Health and Nutrition Examination Survey, or NHANES, noted that using an A1c over 6.5 percent predicts 19.7 million people in the U.S. with diabetes, but if you use fasting plasma glucose of greater than 126 mg/dL, there would be 21.5 million with diabetes. And with an OGTT of greater than 140 mg/dL, there would be 26.5 million with diabetes.

These are non-overlapping populations, Dr. Furlanetto points out—and that demonstrates that just changing the name of a condition doesn’t change the risk for complications of diabetes. “While the current guidelines were set at an A1c of 6.5 percent based on microvascular complications, in fact that doesn’t take into account macrovascular complications, especially cardiovascular disease and stroke.”

Similar considerations apply to the recommended cutoffs for pre-diabetes, Dr. Furlanetto says. “Using the NHANES data and the criteria for high risk for diabetes of an A1c of 5.7 to 6.4 recognizes 12 percent of the U.S. population as at high risk. But using fasting plasma glucose of 100 to 125, that’s 25 percent of the population, and they don’t overlap; if you use either one or the other it’s 33 percent of the population. So doctors are likely to feel they have an easier test to do, but just because A1c identifies fewer people doesn’t mean it’s more tailored or more targeted.” The other side of that, Dr. Furlanetto adds, is that patients with an impaired fasting glucose may take comfort in the fact that their A1c is “normal”—but their sense of security would be a false one.

Although the issue of reference ranges is a hot topic, Dr. Sacks says, he’s confident that diabetes experts will figure it out. “But it will take a while.”

Laboratories have already begun to tackle some of the analytic issues posed by the ADA recommendations. For example, the ADA guidelines stipulate that either a glucose or A1c assay, if used for diagnosing diabetes, must be done in the central lab, Dr. Furlanetto says. “Point-of-care instruments are not sufficient. Except for a couple of meters like Quest’s HemoCue, point-of-care glucose has too much error, and there is no A1c point-of-care meter approved for diagnosing diabetes.” Point-of-care devices are not appropriate for making a diagnosis of either diabetes or pre-diabetes, Dr. Bergenstal says. “That doesn’t mean they might not be someday, but not right now.”

Greater use of A1c testing will lead to a push for faster turnaround time for A1c, Dr. Christenson predicts. “Some folks are running A1c once a day, and now it will have to be run a lot more frequently than that. There will be increased demand, and clinicians are going to want to know the answer quickly. So if you don’t have it in a main analyzer yet, you will.”

One caveat about use of A1c, Dr. Furlanetto says, is that for patients with rapidly developing type 1 diabetes, A1c results have not caught up with what the actual blood sugar is. “But you don’t need A1c to diagnose that. Usually it’s so apparent a clinical condition that the test is not needed.”

Dr. Furlanetto says a standard for pediatrics will be forthcoming. “They will be looking at lots of kids to find out if A1c is really useful in diagnosing mainly type 2 diabetes.” Different issues apply with pregnant women, so he does not expect that recommendations for using A1c to diagnose gestational diabetes will be developed.

In the case of people who have an A1c of less than 6.5 percent but who do have an abnormal fasting plasma glucose, “they’re at risk and they need to be identified, they need their lipids lowered and their blood pressure cared for, just the same as before the criteria were changed,” Dr. Furlanetto says. He notes that the Endocrine Society recognizes that these are non-overlapping populations and has said that if a patient appears to be at risk for diabetes and the A1c is less than 6.5 percent, those people should still undergo glucose testing.

To confirm a fasting glucose, OGTT, or A1c indicating diabetes, the test has to be repeated on a different day, presumably a week or two later, Dr. Furlanetto says. “That’s because things like intercurrent illnesses can affect glucose levels. In the case of A1c, intercurrent illnesses don’t affect the results, but you still want to verify the result is true because you want to make sure the diagnosis is correct.”

Another important point is that the ADA strongly encourages labs to be NGSP-certified. “And they are doing that.” Dr. Bergenstal says. “It’s up to each lab. But we’re hoping that over time, the lab reports will reflect the new standards, so that when a value comes back 15 or above, it’s marked with a star to indicate that the ADA says this is consistent with a diagnosis of diabetes, and others are high risk under ADA standards. We’re hoping there will be common communications of these values.”

Dr. Korpi-Steiner points to another analytical issue laboratories should be aware of: the potential for assay interference. “Some variants such as hemoglobin E, S, and C do interfere with select A1c assay methods, while others are unaffected by variants. So it’s really important to consider the patient population being tested and prevalence of variants in order to choose the most appropriate A1c methodology.” Immunoassay, cation exchange chromatography, and boronate affinity method are the three main methodologies, and all perform differently on this score.

For example, she says, “If you have a patient population with a high prevalence of hemoglobin S, or sickle cell disease, then you may want to choose a method that is capable of identifying the variant as present, and therefore not releasing the result.” The NGSP website ( lists many A1c assay methods and which hemoglobin variants interfere with a given method.

Whether the ADA A1c recommendations will cause a surge in the number of diagnoses of diabetes is difficult to predict, Dr. Korpi-Steiner says. “Some patient conditions preclude A1c testing. Patients with conditions that shorten red blood cell lifespan such as hemolytic anemia and select hemoglobinopathies preclude A1c testing because this may lead to spurious results and misdiagnosis.” Demographic factors including age and ethnicity may also influence A1c values, she says. “Studies suggest that A1c values increase with age and that blacks have higher A1c ranges than whites. So A1c values should be interpreted cautiously to prevent misdiagnosis of diabetes.” Reporting of estimated average glucose (eAG) with the A1c value is also recommended by the ADA, and may be useful for clinicians to correlate with fasting plasma glucose and clinical evidence to ensure correct interpretation and diagnosis. “In my opinion,” Dr. Korpi-Steiner says, “what’s required is that clinicians be well educated to know when it’s appropriate to use A1c for testing and when not to.” The ADA, the specialty societies, and clinical laboratories should all be active in educating clinicians on this subject, she adds.

If there is a drawback to A1c testing, Dr. Korpi-Steiner says, it is the need for further standardization. “There have been great efforts in standardizing A1c assays, but there still exist discrepancies between the NGSP-certified methods aligned to DCCT methods versus methods aligned to the IFCC reference methods. So in my opinion a focus on global harmonization of methods and reference ranges is warranted to improve the quality of A1c testing.”

In the meantime, the ADA’s recommendations are winning wide acceptance in clinical practice. “The A1c recommendations came out in January, and a lot of people are saying, ‘Gosh, that is a convenient test and an appropriate test,’” Dr. Bergenstal says.

The gap that must be filled is enormous. “One in every three people born today will develop diabetes 1 in their lifetime in the current trajectory. There is a Medicare Diabetes Screening Act right now, so that anybody under Medicare is entitled to a blood sugar screening for diabetes every year. It’s in the law, but only 20 percent of those individuals are getting it,” Dr. Bergenstal says. The ADA is now working on getting the Medicare regulations to incorporate the A1c assay as well as fasting blood sugar.

Anne Paxton is a writer in Seattle.

Related Links Related Links