Home >> ALL ISSUES >> 2017 Issues >> HbA1c shows its mettle in predicting diabetes risk

HbA1c shows its mettle in predicting diabetes risk

image_pdfCreate PDF

Anne Paxton

December 2017—The longitudinal Framingham Heart Study, which first identified the concept of risk factors and made serum LDL cholesterol a household name, could help increase the celebrity status of HbA1c, with the Oct. 26 publication of a new study in Diabetes Care.

International and national organizations since 2010 have recognized HbA1c as a valid way to diagnose abnormalities in glycemia and diabetes mellitus. But there has been less consensus on its use as a screen for elevated diabetes risk.

It has been shown that elevated HbA1c and elevated fasting glucose are better at diabetes prediction than fasting glucose alone. But is HbA1c associated with incident diabetes independently, such that HbA1c results can identify individuals with high diabetes risk? That was the question addressed in the Diabetes Care retrospective study “Prediction of type 2 diabetes by hemoglobin A1c in two community-based cohorts,” in which the authors reviewed extensive data collected on subjects of the Framingham Heart Study and the Atherosclerosis Risk in Communities (ARIC) study (Leong A, et al. doi.org/10.2337/dc17-0607).

Based on that data, the authors found that HbA1c predicts diabetes in different common scenarios and is useful for identifying individuals with higher diabetes risk in the short and long term. HbA1c is an “accurate and convenient test [that] has a central place in Type 2 diabetes prevention efforts nationally and worldwide,” the authors conclude.

The Diabetes Care study “is one of the first to demonstrate the additive value of HbA1c in improving the ability to diagnose future diabetes mellitus in combination with fasting glucose,” says study coauthor Michael J. McPhaul, MD, medical director for endocrinology and metabolism at Quest Diagnostics’ Nichols Institute.

For this study, the researchers focused on middle-aged participants in the Framingham and ARIC studies who did not have diabetes—11,244 whites and 2,294 blacks—and determined whether those subjects developed diabetes in the short term (within eight years) or the long term (after 20 years). A total of 3,315 subjects developed diabetes after 20 years, and their initial HbA1c results, it turned out, were highly predictive of whether they did develop diabetes. For each percentage-unit increase in HbA1c, the odds of developing diabetes increased fourfold.

Those results held in four real-world scenarios where HbA1c is commonly used: the “HbA1c only” scenario in which age and sex are the only extra factors taken into account; the “HbA1c plus fasting laboratory tests” scenario; the “HbA1c plus clinic visit” scenario; and the “HbA1c plus fasting laboratory tests plus clinic visit” scenario. In all of these models, higher HbA1c was associated with increased type 2 diabetes risk in participants, with and without high fasting glucose.

The study authors’ intent was not primarily to demonstrate that HbA1c predicts diabetes, says study coauthor James B. Meigs, MD, MPH, professor of medicine at Harvard Medical School and one of the investigators of the Framingham Heart Study. “We knew when we started that HbA1c predicts diabetes, so that was our ‘straw-man’ hypothesis. In this paper, we wanted to frame the study around how well HbA1c predicts in common clinical settings. We showed it did work in all the settings we examined.”

“We concluded HbA1c is a useful test; there are different scenarios where people get tested with HbA1c, and it works in all. So let’s use it.”

HbA1c and fasting glucose measurements, while typically concordant, are sometimes discordant, and the importance and implications of that discordance have not been systematically studied, Dr. McPhaul points out. If the two measures were represented by circles in a Venn diagram, their sets of results would largely overlap, but there would be slivers on either side representing cases in which one result but not the other indicates diabetes or “normal.”

At Quest, Dr. McPhaul’s charge is to update and improve on existing offerings in the areas of diabetes and metabolic disorders, and he has collaborated with other institutions to explore the relationships of different measurement phenomena in diabetes. The new study is the result of a collaboration with Dr. Meigs.

“We decided to model the utility of adding HbA1c to the existing diabetes risk score, which is one of a number of calculated algorithms created as a spinoff of the Framingham cohort study,” Dr. McPhaul says. “It’s a relatively simple algorithm that integrates a number of demographic features, including diastolic hypertension and family history of diabetes, to give a risk score for developing diabetes within eight years.”

Dr. McPhaul

Dr. McPhaul

When first developed in 2007, that risk score did not include HbA1c. “I always found that kind of puzzling,” Dr. McPhaul says. So he and other researchers began to look at how much the change in the capacity of HbA1c would change the overall predictiveness of the algorithm. “We also decided to look at HbA1c as an independent variable. We simply analyzed the data in a way that allowed us to include all the elements that were part of the original Framingham score, but also kept a measure of how well and in what way HbA1c, as an independent variable, would affect the risk over time.”

The researchers decided to use four typical screening scenarios for this analysis. “We didn’t want some arcane, highly constrained single view of the way to see the impact of these different parameters,” Dr. McPhaul says. Several of the different models borrowed from the original Framingham risk score to include demographics, laboratory data only, or both. Conducting a retrospective analysis of data and information that was collected prospectively, “we added into this the HbA1c measure to see what impact HbA1c would have on the model.” The goal was to find the best combination for providing a cost-effective means of assessing diabetes risk, with a minimum of material information needed to add into the model.

Fasting glucose and HbA1c are different angles on exactly the same problem, Dr. McPhaul says. “Fasting glucose is a snapshot, looking at people under the best of all possible circumstances, of how well their body can regulate and maintain glucose at a normal level, while HbA1c is a very specific correlate to an average glucose over time.” The Diabetes Care study brings into focus that these two pieces of information are not simply interchangeable; each contains information that complements the other, he says.

An early surprise of this research project was the predictive capacity of HbA1c in subsets of individuals who had no element suggesting they were going to have diabetes in the future, based on other risk factors. “You could see that there were people who were predicted to be at risk who did develop diabetes, but have normal fasting blood sugar.” Dr. McPhaul has heard physicians say to patients in the past that slight elevations of HbA1c don’t have consequence. But “I don’t think slight elevations of HbA1c have necessarily been viewed as they should be viewed, which is as an early warning sign.”

The Diabetes Care study showed that individuals with normal HbA1c, whatever their other test results, are at the lowest risk of developing diabetes and people with the highest HbA1c have the highest risk, Dr. McPhaul says. “Whether you’re black, whether you’re white, whether you were in ARIC, whether you were in Framingham, these observations hold true in both of these very large cohorts.” Given these findings for two racially diverse cohorts, he adds, “I have no reason to believe that these observations would not be true if modeled on any other similarly sized population of whatever ethnic or racial composition.”

Addressing the controversy among international groups over the best role for HbA1c in identifying risk was beyond the scope of the Diabetes Care study, Dr. McPhaul notes. “What the study does point to clearly is that abnormal HbA1c is something that people should pay attention to. Even if a screening result includes only a moderately elevated HbA1c, people should take that seriously and recognize that they should be even more urgent and direct in addressing their risk factors through improving their level of exercise and decreasing their weight.”

The study authors compare the use they recommend for HbA1c to the uses of prostate-specific antigen and LDL. Since the risk estimates and prediction equations were obtained from large population-based cohorts from two major U.S. ethnicities with two decades of follow-up, HbA1c results “can be used in clinical laboratory reports, similar to the reporting of high values of PSA and LDL that are supplemented by their associated estimated risk for prostate cancer or cardiovascular disease,” the study notes.

This comparison is simply a matter of recognizing that the risks and benefits of HbA1c are ones that have to be judged carefully within the context of the decisions to be made, Dr. McPhaul says. “The measurements of lipids or PSA can be judged as valuable enough to warrant their use in specific circumstances: the evaluation of patients. In the case of HbA1c, it should be recognized as not simply an equivalent of fasting glucose but something that can add to the information that fasting glucose provides.”

Dr. McPhaul also believes that the Diabetes Care study solidly makes the case for the contribution of HbA1c to risk assessment. “The breadth of population that has been studied here pretty well establishes that this is not a one-time observation and one study. It demonstrates that individuals with elevations of HbA1c have the highest risk of future diabetes no matter what their glycemic status is.”

But the greater utility of research like the Diabetes Care study will be getting people to pay attention to the test results, Dr. McPhaul says. “The challenge is not really research. It’s more implementation: getting physicians and patients to act on the data they have in front of them through diet and exercise. That’s the need we have in front of us as a society, so we don’t end up spending the equivalent of the GDP of Greece on diabetes every year.”

Since the American Diabetes Association and the World Health Organization recommended HbA1c for screening and diagnosis of diabetes in 2010 and 2011, respectively, “we’ve seen major increases in HbA1c testing and its use for diagnosis,” says Elizabeth Selvin, MPH, PhD, a coauthor of the study and a professor of epidemiology at Johns Hopkins Bloomberg School of Public Health, who has published extensively on HbA1c and the epidemiology of prediabetes and diabetes. “But for prediabetes, the subject of the Diabetes Care study, there is much more controversy regarding an optimal single definition.”

To start with, there is no universally agreed-upon definition of prediabetes. The ADA and WHO recommend different ranges of fasting glucose results to define who has prediabetes. Although the two organizations do agree on ranges for two-hour glucose, Dr. Selvin notes, that test is not used as widely as the other tests. “So in the end, that means there are five different definitions of prediabetes that are currently employed. And that’s definitely sowing confusion in this field.”

CAP TODAY
X