Glycemic Variability: A Hemoglobin A_1c–Independent Risk Factor for Diabetic Complications

Diabetes affects an estimated 20.8 million individuals in the United States, 7% of the current population, and the lifetime risk of developing diabetes for those born in the year 2000 is 35%.^{1 - 2} Many of these individuals will develop diabetes-specific microvascular pathology in the retina, renal glomerulus, and peripheral nerve and accelerated atherosclerotic macrovascular disease affecting arteries that supply the heart, brain, and lower extremities. In both type 1 and type 2 diabetes, large prospective clinical studies have shown a strong relationship between time-averaged mean levels of glycemia, measured as hemoglobin A_1c (HbA_1c), and diabetic complications.^{3 - 4} These studies are the basis for the American Diabetes Association's current recommended treatment goal that HbA_1c should be less than 7%.⁵ However, only about a third of patients diagnosed as having diabetes achieve that goal.⁶ Even fewer reach the target level for HbA_1c of 6.5% advocated by the American College of Endocrinology.⁷

The advent of self-monitored blood glucose (SMBG) made it possible for patients with type 1 diabetes to achieve significantly lower levels of HbA_1c.³ The American Diabetes Association recommends SMBG 3 or more times daily for patients receiving multiple insulin injections.⁵ However, there is no specific recommendation for the frequency of SMBG in patients with type 2 diabetes who are receiving oral agents or medical nutrition therapy.

In this issue of JAMA, Saudek and colleagues⁸ assess the evidence underlying both HbA_1c testing and SMBG and arrive at evidence-based conclusions about their optimal use. The authors found disagreement among published studies about whether there is a relationship between frequency of SMBG and HbA_1c, such that some studies have found no benefit of SMBG whereas others found benefit, with a reduction in HbA_1c of 0.4% among patients who performed SMBG.⁹ Given this reduction, the estimated corresponding reduction in risk of complications in patients with type 2 diabetes would be only 12.5% for those whose initial HbA_1c was 8%.⁴

Considering these facts, an important issue is whether greater risk reduction of diabetic complications can be achieved without further lowering of HbA_1c. The study by Monnier and colleagues¹⁰ in this issue of JAMA provides new data to help address this issue. Over the past 35 years, several major molecular mechanisms have been implicated in glucose-mediated vascular damage. Each of these mechanisms has been studied independently of the others, with no apparent common element linking them. Recent discoveries have made clear that all of these seemingly unrelated mechanisms may arise from a single, hyperglycemia-induced process: the overproduction of the reactive free radical molecule superoxide. It now appears that mitochondria, the principal energy-generating organelles in the cell, are required for initiation of hyperglycemia-induced superoxide production, which can, in turn, activate a number of other superoxide production pathways that may amplify the original damaging effect of hyperglycemia. Increased free fatty acid oxidation in mitochondria also produces superoxide.^{11 - 13}

In their study involving patients with type 2 diabetes not using insulin and with poor glycemic control (mean [SD] HbA_1c of 9.6% [1.3%]), Monnier et al measured 24-hour excretion of 8-iso prostaglandin F_2α (8-iso PGF_2α), an indicator of free radical production derived from esterified arachidonic acid in cell membranes.^{14 - 15} At the same time, patients used a continuous blood glucose monitoring system, and fasting glucose levels and HbA_1calso were determined. Among those with diabetes, there was a linear correlation between increased free radical production and the magnitude of glucose fluctuations, calculated as the mean amplitude of glycemic excursion (MAGE), a quantitative tool first described in 1970.¹⁶

There was no significant correlation between free radical production and the 24-hour mean glucose concentration, fasting plasma glucose levels, or even the HbA_1c level. As reported previously,¹⁷ levels of this indicator of free radical production were nearly 2-fold higher in patients with diabetes compared with levels in nondiabetic controls. However, among those with diabetes, 8-iso PGF_2α levels were 4 times higher in patients with greatest glycemic variability compared with patients having the lowest glycemic variability. Moreover, acute glucose variability as estimated by MAGE was a strong predictor of total free radical production, whereas postprandial blood glucose level area under the curve was not. This finding is important, given the intense interest in the possible role of postprandial hyperglycemia in the pathogenesis of diabetic atherosclerosis.¹⁸

Is this level of increased free radical production enough to cause clinically significant damage? Our research group has observed that exposure of lean nondiabetic participants to the same magnitude of glycemic excursion reported by Monnier et al, using hyperglycemic clamps for similar periods of time, causes a dramatic free radical–induced decrease in activity of a major antiatherogenic endothelial cell enzyme, prostacyclin synthase (M.B. et al, unpublished data). Prostacyclin synthase prevents both the initiation of atherosclerosis and the progression and nature of atherosclerotic plaques through actions affecting endothelial cells, monocytes/macrophages, and vascular smooth muscle cells.¹⁹

These new data reported by Monnier et al may explain how in the Diabetes Control and Complications Trial the diabetic retinopathy risk at identical sustained levels of HbA_1c was significantly reduced by intensive treatment.²⁰ For example, in the subgroups with a sustained HbA_1c of 9% for the entire study duration, the risk for retinopathy was reduced by more than 50% in the intensive control group, even though these 2 subgroups of patients had the same HbA_1c. Similar analyses have not yet been published from the UK Prospective Diabetes Study. Since continuous glucose monitoring is not yet feasible for routine, long-term patient care,⁸ an important clinical issue is whether physicians can estimate glycemic variability using SMBG monitoring data. Monnier and colleagues did not present data correlating the SMBG values obtained concomitantly with continuous glucose monitoring with the magnitude of free radical production. Such data would be invaluable for devising algorithms for calculating clinically meaningful estimates of glycemic variability from SMBG downloaded from patients' glucose meters. Most currently available glucose meters are able to calculate total and time-specific means and standard deviations.²¹

The new findings reported in this issue of JAMA by Monnier and colleagues,¹⁰ if confirmed in larger studies, have enormous clinical implications. First, these data suggest that in patients with type 2 diabetes, SMBG should be performed with increased frequency to monitor glycemic variability, regardless of the effect on HbA_1c. Second, the findings suggest that different therapeutic strategies now in use should be evaluated for their potential to minimize glycemic excursion, as well as for their ability to reduce HbA_1c.

The timing and frequency of SMBG necessary to optimally minimize glycemic variability need to be determined by future investigations. Additional work is needed to determine the best pharmacologic strategies for minimizing glycemic variability and the increased free radical production it causes. Wider use of real-time continuous glucose monitoring in clinical practice would provide the required monitoring tool to minimize glycemic variability and superoxide overproduction. In the midst of a global diabetes epidemic, efforts to assess and minimize glycemic variability as a risk factor independent of HbA_1c may help reduce the staggering burden of diabetic complications.