A Step Backward—or Is it Forward?

The time may have come—or at least will soon arrive—when the diagnostic categories related to diabetes and its close relatives, impaired glucose tolerance (IGT) and impaired fasting glucose (IFG), should be reconsidered. IGT, IFG, and the combination of the two are now considered pre-diabetes because, compared with normal glucose tolerance, they signal an increased risk of developing diabetes (1). Before 1979, when the National Diabetes Data Group defined IGT (2) and declared it to be an official diagnostic category, one either did or did not have diabetes and nondiagnostic glucose levels were de facto “normal.” A middle ground was recognized by some physicians, but it was not enshrined in all epidemiological studies or in clinical practice.

The addition of IFG in 1997 for the purpose of creating a state equivalent to IGT both added to the middle ground and raised awareness and appreciation of intermediate levels of glucose intolerance (3). There was now a distinctive group of individuals whose fasting plasma glucose (FPG) lay between normal (<110 mg/dl) and the new 1997 lower diabetes cut point of 126 mg/dl and who may not meet the IGT criterion of 140–199 mg/dl 2 h after a standard oral glucose load of 75 g (3). A further refinement of IFG in 2003 altered the lower IFG cut point from 110 to 100 mg/dl, a change recommended to make both intermediate states equivalent and to define a more realistic upper limit of normal (4).

Recently, however, more and more studies have demonstrated that at glycemic levels above normal but within the range of either IFG or IGT, there is an increasing risk of crossing the diabetes line (FPG 126 mg/dl and/or a 2-h post–glucose load level of 200 mg/dl) within a few years (1). These latter numbers are based on the association of IFG and IGT with the later development of diabetic retinopathy (3); however, even this pathologically based dividing line between diabetes and pre-diabetes may be challenged by the Diabetes Prevention Program report finding an increased prevalence of retinal microaneurisms in individuals with IGT (5). Also, a recent study in young Israeli men reports that at fasting glucose levels now considered normal, there is still a significantly increased risk of developing diabetes within a few years (6). In the range of FPG between 81 and 99 mg/dl, there is a threefold progressive increase in the 6-year incidence of diabetes (6). Likewise, in the range of 2-h plasma glucose levels between <85 and 125–139 mg/dl after an oral glucose load, the risk of incident diabetes over 7–8 years increases from 3.0 to 10.7% (7). Furthermore, within these currently defined normal ranges of both FPG and 2-h plasma glucose, a reduction in insulin secretion relative to insulin resistance, i.e., a reduction in physiological β-cell function, is observed as plasma glucose increases (8), offering a pathophysiological continuum with the more marked β-cell dysfunction characterizing IGT, IFG, and, finally, diabetes. In addition, FPG and especially 2-h plasma glucose levels below the diabetes cut points (9–14), as well as their accompanying A1C levels (15), are reported to predict cardiovascular disease and mortality. In short, the diagnostic situation is much more fluid than published categories and their cut points would suggest.

Moreover, the use of IGT in particular as an absolute cut point for enhanced risk of diabetes, i.e., pre-diabetes, is undercut by an undesirably high degree of irreproducibility (3). Patients can move back and forth between IGT and normal glucose tolerance in both the short-term and the long-term. For example, in one recent population-based study, 39% of individuals with IGT on an initial oral glucose tolerance test were classified as normal (and 13% as diabetic) on a second oral glucose tolerance test performed 2–6 weeks later (16). Obvious clinical reasons for this phenomenon such as changes in weight, exercise patterns, preceding carbohydrate intake, drug use, and comorbidity are not always identifiable, but alteration in β-cell function (as noted above) seems a likely pathophysiological candidate.

Reversion from diabetes to nondiabetes was also reported in 12% of one surveyed population after 7–8 years; 90% of the reverters were newly diagnosed at baseline examination (17). In another study (18), 30 months after an initial diagnosis of diabetes, 42% of subjects diagnosed by FPG ≥126 mg/dl and 72% diagnosed by 2-h plasma glucose ≥200 mg/dl were nondiabetic.

The term “dysglycemia” has been advocated to describe the state of having a level of plasma glucose above that at which adverse outcomes are likely to occur (19). Glucose, a molecule once thought to be a critical blessing, essential to life, now is thought to also cast a dark shadow on life, even within the supposedly normal range. We may all be on our way to diabetes and/or to cardiovascular disease (CVD) at different paces dependent on our genome, our environment, our culture, and our lifestyle. The only thing preventing us from developing diabetes may be the interposition of some other fatal disease or adoption of a stringent lifestyle that defeats obesity and the ill effects of sedentary behavior.

Given this new appreciation that the level of glucose, like that of LDL cholesterol, may have no true cut point to separate the range of no risk from the range of risk, it may be completely arbitrary and biologically unsound to define a state of pre-diabetes if we are all pre-diabetic to a greater or lesser degree. Even our confidence at defining normal glucose levels seems to be challenged regularly (20).

Thus, it would seem that the time has arrived for considering a reorganization of our diagnostic categories. We could discard IFG and IGT altogether and go back to one disease category labeled “diabetes” with the current cut points of 126 mg/dl and 200 mg/dl. The long-term results of follow-up of the Diabetes Prevention Program cohort (21) may even someday incline us to lower these numbers if people who remain in the current IFG or IGT categories show more and more evidence of complications during careful follow-up.

Within the category of nondiabetes, physicians should, however, recognize a gradient of risk for developing diabetes and/or CVD related to rising plasma glucose levels and say, as is done with LDL cholesterol, “the lower the better.” Physicians could indicate to their patients that there is a gradient of risk beginning at an FPG of ∼80 mg/dl, show them where they are on this gradient, and begin encouraging lifestyle modification at a time much earlier than is practiced now.

To aid physicians, we could create nomograms with odds ratios for the risk of developing diabetes for every 10 mg/dl increment in FPG above 80 mg/dl as the referent of 1.0. A similar nomogram with an appropriate baseline and increments can now be constructed for A1C (15), as standardization of the assay has proceeded well (22) and we are deriving accurate and precise estimates of equivalence between A1C and average glucose levels (23). Moreover, it now seems likely that an A1C level can be established for the diagnosis of diabetes (24–27). Even more appropriate and relevant are global risk assessment algorithms that address both the risk of diabetes and the risk of developing CVD by integrating all the major risk factors for both conditions (see Diabetes PHD at http://www.diabetes.org/diabetesphd/default.jsp?WTLPromo=Home_flash_030408PHD).

Of course, physicians, nurses, and other professionals should do their utmost to employ lifestyle modification therapy in all whose health care is their responsibility because so many adults (and increasingly children) are at risk. The presence of other risk factors for diabetes and associated adverse outcomes, such as family history, age, overweight/obesity, a history of gestational diabetes mellitus, dyslipidemia, and elevated blood pressure should intensify efforts at lifestyle modification. Periodic measurement of FPG (or A1C) should be employed to determine how successfully each patient is faring with his/her dysglycemia, much as we now measure lipid and blood pressure levels for analogous reasons.

An important issue is whether drugs should be used to treat dysglycemia to achieve the lowest glucose level possible. Available clinical trial evidence shows some effectiveness in preventing or delaying progression from IGT and/or FPG to diabetes of metformin (28), acarbose (29), rosiglitazone (30), orlistat (31), and glipizide (32), at least for 2–4 years. The risk-to-benefit and cost-to-benefit ratios for the indefinite use of drug therapy below the current cut points of diabetes are unclear, although recent guidelines indicate a place for pharmacologic treatment in very select individuals at risk (33). Just as more aggressive use of blood pressure–lowering drugs in populations with blood pressure levels previously considered without risk is being studied (34), we should determine the long-term benefit of glucose lowering when levels have not reached those of diabetes onset.

The above changes require many issues to be addressed. For example, because all of the prevention trials were short term and glucose tolerance tends to worsen progressively, even though there was a significant in-trial reduction in the risk of developing diabetes, in most cases the majority remained at risk. Will we be able to identify with certainty from pathophysiological studies which patient will benefit from which drug, or would multiple drug therapy be necessary, increasing risk and cost?

What could be the negative consequences of returning to a simple diagnostic categorization of diabetes? Now that physicians, nurses, nutritionists, residents, and medical students have been encouraged to think IGT and IFG, to drop these defined states would require a reeducation program, and change in itself encounters challenges and opposition (35). To a lesser extent, some public reeducation might also be necessary, though the terms IFG and IGT do not yet seem to be commonly used by laypeople. Does the evidence that IFG and IGT stem from somewhat different pathophysiologic processes (36,37) preclude doing away with the terms to achieve a clinically simplified paradigm? Also, because IGT and IFG have frequently been used as risk factors in epidemiologic research and IGT has been a baseline state for therapeutic research, they could retain their research significance in future studies for some time.

Another problem is that all the important diabetes prevention trials to date, including tests of lifestyle modification, have started with cohorts who had IGT (28,38); many of the subjects undoubtedly had IFG as well (28). Would diabetes be as preventable (or as easily delayed) if these methods were used in the general population with lower glucose levels, even if used in higher risk groups such as those with a family history of diabetes? Would they work as well in the younger as in the older segments of the populations or in those who meet current BMI definitions of overweight and obese as in those who do not? Would microvascular and CVD complications never develop? Should we delay changing the classification scheme until we get the answers to these questions, when it is possible we will not know the answers for many years, if at all?

Another important consideration would be whether health care insurers could be expected to pay for professional time and effort to teach and support preventive lifestyle modifications if patients could not be given a formal diagnosis of pre-diabetes based on plasma glucose levels. Current public and political concern about the inexorable rise in health care costs might oppose widespread, potentially very costly, preventive measures in the entire population until they are of solidly proven benefit. Are we ready to tackle this complex issue? Do we still need to absorb the lessons of IGT and IFG and continue to determine their relative relationship with a growing list of biologic and demographic variables? Perhaps a numerical global risk assessment score for diabetes and CVD employing all risk factors measurable in the physician's office can be used to trigger insurance coverage for intensive lifestyle modification therapy (28).

In our judgment, continuing to debate the correct cut points now seems of little value. Despite an endless list of studies that could be done to better define IFG and IGT, it is clear that the days of IFG and IGT as clinical entities are limited, and one can predict their lifespans will be shortened, not lengthened, by the accumulation of knowledge. We predict that it will actually be a step forward and not backward—scientifically if not historically—when IGT and IFG become outmoded. Perhaps it is now time to reexamine our current categorical classifications of plasma glucose levels. Our future may turn out to be our past.