Diagnostic and Therapeutic Implications of Relationships Between Fasting, 2-Hour Postchallenge Plasma Glucose and Hemoglobin A1c Values

Background  Increased fasting plasma glucose (FPG) and 2-hour postchallenge plasma glucose (PCPG) levels with normal hemoglobin A_1c (HbA_1c) levels are recognized as risk factors for cardiovascular disease. We undertook this study to determine the relationships between FPG and 2-hour PCPG levels over the normal HbA_1c range and to assess the need to control FPG and 2-hour PCPG levels to achieve HbA_1c targets recommended by the American Diabetes Association (ADA), International Diabetes Federation (IDF), and American College of Endocrinology (ACE).

Methods  The data of all healthy individuals with HbA_1c values less than 7.0% (N = 457) who underwent oral glucose tolerance tests between 1986 and 2002 for either screening as potential research volunteers (93%) or diagnostic purposes (7%) were analyzed.

Results  Of 404 individuals with normal HbA_1c levels (<6.0%), 60% had normal glucose tolerance, 33% had impaired glucose tolerance, 1% had isolated impaired FPG, and 6% had type 2 diabetes mellitus. Of 161 individuals without normal glucose tolerance, 80% had normal FPG levels. Both FPG and 2-hour PCPG levels increased as HbA_1c increased and were significantly correlated (r = 0.63, P<.001), but the 2-hour PCPG level increased at a rate 4 times greater than FPG and accounted for a greater proportion of HbA_1c. People who met the IDF and ACE HbA_1c targets (<6.5%) had significantly lower 2-hour PCPG levels than those who met the ADA target (<7.0%) (P = .03), whereas FPG levels were similar.

Conclusions  Most individuals with HbA_1c values between 6.0% and 7.0% have normal FPG levels but abnormal 2-hour PCPG levels, suggesting that an upper limit of normal for FPG at 110 mg/dL (6.11 mmol/L) is too high and that attempts to lower HbA_1c in these individuals will require treatment preferentially directed at lowering postprandial glucose levels.

N-terminal valine residues of erythrocyte hemoglobin become irreversibly glycosylated in proportion to circulating glucose concentrations, and the resultant product is commonly referred to as hemoglobin A_1c (HbA_1c).¹ Because of the half-life of the erythrocyte, the percentage of hemoglobin represented by HbA_1c provides an index of the average plasma glucose concentration during the previous 2 to 3 months.¹ Consequently, HbA_1c measurements have become the preferred method to monitor long-term glycemic control in patients with diabetes mellitus² and have been used in clinical trials to assess the efficacy of antidiabetic medications and the impact of therapeutic interventions on diabetic complications.^3-5

Data from both epidemiological studies^6,7 and controlled clinical trials^3-5 indicate that lower HbA_1c levels are associated with reduced risks for both microvascular and macrovascular diabetic complications. The American Diabetes Association (ADA) currently recommends an HbA_1c treatment target of 7.0% or less primarily to reduce microvascular complications.⁸ The International Diabetes Federation (IDF)⁹ and the American College of Endocrinology (ACE),¹⁰ taking into consideration cardiovascular disease (CVD), both recommend a treatment target of 6.5% or less. The Council for the Advancement of Diabetes Research and Education recommends the lowest HbA_1c achievable without unacceptable side effects.¹¹

Although HbA_1c levels accurately reflect long-term glycemia, several issues remain. One is the relative contributions of fasting plasma glucose (FPG) and postchallenge plasma glucose (PCPG) concentrations.¹² Such knowledge would be important when choosing among therapeutic options to achieve the different recommended glycemic targets, especially when HbA_1c values are near the upper limit of normal.^12-16

Another important issue concerns the use of HbA_1c,FPG, and 2-hour PCPG values for diagnostic purposes. In 1997, the ADA¹⁷ and the World Health Organization (WHO)¹⁸ revised their criteria for normal glucose tolerance (NGT): the upper limit of normal for FPG was reduced from 126 to 109 mg/dL (6.99-6.05 mmol/L). The FPG values between 110 and 126 mg/dL (6.11 and 6.99 mmol/L) were designated as impaired fasting glucose (IFG) by the ADA. Two-hour values during the standard oral glucose tolerance test (2-hour PCPG) diagnostic of impaired glucose tolerance (IGT) (values between 140 and 200 mg/dL [7.77 and 11.10 mmol/L]) and diabetes (values >200 mg/dL [11.10 mmol/L]) remained the same.

The ADA (but not the IDF and WHO) currently recommends that the FPG be used diagnostically in preference to the oral glucose tolerance test.¹⁷ An implicit assumption of this recommendation was that IFG would have the same significance as IGT regarding the risk for development of type 2 diabetes mellitus and its complications. However, several studies^19,20 have challenged this assumption. Furthermore, the ADA recommendation has been questioned, because numerous studies^12-14,21-24 have demonstrated that an appreciable number of individuals with a normal FPG level will have an abnormal 2-hour PCPG level and thus their condition will go undiagnosed and untreated.

The clinical significance of this is underscored by the fact that hyperglycemia has been identified as an independent and continuous risk factor for CVD,⁶ the major cause of mortality in people with type 2 diabetes mellitus.^25,26 Thus, in addition to patients with type 2 diabetes mellitus²⁵ or IGT,^13,19,27 it now appears that individuals with FPG and HbA_1c values in the upper normal range are also at increased risk for CVD.^20,28,29

This increased risk of CVD for individuals with HbA_1c levels in the normal range has been recently reinforced by the Norfolk cohort of the European Prospective Investigation of Cancer and Nutrition (EPIC-Norfolk) study.³⁰ This was a 2- to 4-year follow-up of 4662 men aged 45 to 79 years that found that individuals with HbA_1clevels between 5.0% and 5.4% had a 2.5-fold increased risk of dying of CVD compared with individuals with HbA_1c levels below 5.0%. Moreover, individuals with HbA_1cvalues between 5.5% and 6.9% had a 2-fold increase in overall mortality. Unfortunately, that study provided no data regarding FPG and 2-hour PCPG levels, and hence it was not possible to assess their relative impact.

We undertook the present study to determine the relative contribution of FPG and 2-hour PCPG levels to HbA_1c over the normal HbA_1c range, as well as those values analyzed in the EPIC-Norfolk study, and to assess the need to control FPG and 2-hour PCPG levels to achieve HbA_1c targets recommended by the ADA, IDF, and ACE.

From 1986 through 2002, data were systematically collected from all individuals undergoing standard oral glucose tolerance tests performed as recommended by the ADA³¹ in whom simultaneous HbA_1c levels were determined (N = 607).¹ Of these individuals, 457 had HbA_1c levels below 7.0%, and their data were selected for study. Most individuals (93%) were responders to advertisements for healthy volunteers for research studies, and the remainder (7%) were referred for evaluation of glucose tolerance status. Three individuals had been previously diagnosed as having diabetes but their condition was being managed by diet alone. None of the individuals were taking medications known to affect glucose tolerance. All were in apparent good health based on medical history, physical examination results, and routine laboratory tests. The patients were largely non-Hispanic white (59%); 13% were of Hispanic origin, 22% were of mixed Hispanic European, African American, and Native American background, and 6% were African American. Plasma glucose values were determined by a glucose analyzer (YSI Inc, Yellow Springs, Ohio, or Beckman Coulter Inc, Fullerton, Calif); the coefficient of variation for glucose values of 50 to 250 mg/dL (2.78-13.88 mmol/L) was approximately 1%. Hemoglobin A_1c values were determined by high-performance liquid chromatography (Bio-Rad Laboratories, Hercules, Calif; reference range, 4.0%-6.0%; coefficient of variation, 3.6%).

Data are given as mean ± SD unless otherwise specified and were analyzed using Statistica statistical software (1998 edition, Statsoft Inc, Tulsa, Okla). Normality of the distribution of HbA_1c values was assessed using the Kolmogorov-Smirnov test. Comparisons between groups were performed using analysis of variance followed by the Scheffé test. The contributions of changes in FPG and 2-hour PCPG levels to changes in HbA_1c levels were assessed using multiple linear regression. P≤.05 was considered statistically significant.

Using ADA, IDF, and WHO criteria,^20,32,33 243 individuals had NGT (Table 1). Their HbA_1c level was 5.05% ± 0.47%, resulting in an upper limit of normal (mean ± 2 SD) of 5.99%. The HbA_1cvalues were normally distributed and comparable to those found in large population-based studies such as the Third National Health and Nutrition Examination Survey (NHANES III) (mean ± SD, 5.3% ± 0.4%; N = 2284)³⁴ and the Telecom Study (5.05% ± 0.5%, N = 3240).³⁵ Thus, although not a population-based survey, our results appear to be representative of the general population. The mean ± SD FPG and 2-hour PCPG values were 85.7 ± 8.3 mg/dL (4.76 ± 0.46 mmol/L) and 103.1 ± 18.6 mg/dL (5.72 ± 1.03 mmol/L), respectively, yielding upper limits of normal of 102 and 140 mg/dL (5.66 and 7.77 mmol/L), respectively.

An additional 161 individuals had an HbA_1c level below the upper limit of normal (Table 1), but abnormal glucose tolerance. Of those with a normal HbA_1c level (n = 404), 86% with IGT had a normal FPG level, as did 63% with type 2 diabetes mellitus. In contrast, only 5 individuals (1%) had an abnormal FPG level with a normal 2-hour PCPG level (isolated IFG).

To assess the contribution of FPG and 2-hour PCPG levels to HbA_1c levels in the EPIC-Norfolk study and to evaluate the need for control of FPG and 2-hour PCPG levels for HbA_1c targets of the ADA, WHO, and IDF, we also included for analysis all subjects with HbA_1c levels between 6.0% and 7.0% (n = 53). None had NGT, 23 (43%) had IGT, 26 (49%) had type 2 diabetes mellitus, and 4 (8%) had isolated IFG. Of all those with abnormal glucose tolerance having an HbA_1c level less than 7.0% (n = 205), 152 (74%) had normal FPG levels (82% of those with IGT and 48% of those with type 2 diabetes mellitus), confirming the insensitivity of FPG for detecting abnormal glucose tolerance.^12-14,21-24

The data of all subjects (N = 457) were divided into deciles according to their HbA_1c level (Table 2). Figure 1 shows the distribution of these individuals according to categories of glucose tolerance. The frequency of abnormal glucose tolerance (IGT, isolated IFG, and type 2 diabetes mellitus) increased with HbA_1c from less than 10% in the first decile (4.33% ± 0.27%) to 100% in the tenth decile (6.46% ± 0.23%).

The FPG and 2-hour PCPG levels were significantly correlated (r = 0.63, P<.001) (Figure 2), and although both increased as HbA_1c level increased, 2-hour PCPG concentrations increased by a much greater extent (Figure 3). For every 1% increase in HbA_1c level, 2-hour PCPG level increased nearly 4 times as much as that for FPG (47 vs 12 mg/dL [2.61 vs 0.67 mmol/L]).

To determine the relative contribution of FPG and 2-hour PCPG to HbA_1c, multiple linear regression analysis was performed in which HbA_1c was the dependent variable and FPG, 2-hour PCPG, body mass index, sex, and age (Table 3) were independent variables. The overall correlation coefficient was 0.586 (P<.001). Only FPG (P<.001), 2-hour PCPG (P<.001), and age (P = .02) contributed significantly. The partial β regression coefficient for the 2-hour PCPG (.346) was nearly 1.5 times greater than that for the FPG (.238), indicating that increases in the 2-hour PCPG level explained approximately 50% more of the increase in HbA_1c than did the FPG.

Table 4 gives the characteristics of subjects subdivided by HbA_1c as in the EPIC-Norfolk study. Of 107 individuals with HbA_1c levels below 5.0%, 16% had abnormal glucose tolerance (15% had either IGT or isolated IFG, and 1% had type 2 diabetes mellitus). Of the 181 individuals with HbA_1c levels of 5.0% to 5.4%, 37% had abnormal glucose tolerance (32% had either IGT or isolated IFG, and 5% had type 2 diabetes mellitus). Of 169 individuals with HbA_1c values of 5.5% to 6.9%, 77% had abnormal glucose tolerance (53% had either IGT or isolated IFG, and 24% had type 2 diabetes mellitus).

Currently, both the IDF⁹ and ACE¹⁰ recommend a target HbA_1c of 6.5% or lower, whereas the ADA recommends a target of 7.0% or lower.⁸Table 5 compares FPG and 2-hour PCPG values of individuals with HbA_1c levels of 6.0% to 6.5% and 6.6% to 7.0%; FPG values of these groups were not significantly different from one another and were only slightly above the current upper limit of normal (110.9 mg/dL [6.16 mmol/L]). However, people with HbA_1c levels of 6.6% to 7.0% had 2-hour PCPG levels significantly greater than those with HbA_1c levels between 6.0% and 6.5% (225.8 ± 50.5 mg/dL vs 198.3 ± 55.8 mg/dL [12.53 ± 2.80 mmol/L vs 11.01 ± 3.10 mmol/L], P = .03). Moreover, their 2-hour PCPG levels were on average 2-fold greater than those of individuals with NGT (103.1 mg/dL [5.72 mmol/L], Table 1).

The key findings of this study are that with the reference range for HbA_1c, (1) FPG and 2-hour PCPG levels are significantly correlated and both increase as HbA_1c increases, but (2) 2-hour PCPG levels increase at a much greater rate than FPG levels and contribute more to the increase in HbA_1c levels; (3) 2-hour PCPG levels are a more sensitive indicator of abnormal glucose tolerance than either FPG or HbA_1c; and (4) the main difference in individuals satisfying HbA_1c goals recommended by the IDF and ACE (≤6.5%) compared with that recommended by the ADA (≤7.0%) relates to postprandial hyperglycemia.

The correlation we found between FPG and 2-hour PCPG, although highly significant (P<.001), was modest (r = 0.63), explaining approximately 40% of the variation. This could reflect the importance of unmeasured variables (eg, physical fitness), variability in the FPG and 2-hour PCPG levels,³⁶ and imprecision in the measurements of plasma glucose. Similarly, multiple linear regression indicated that FPG, 2-hour PCPG, and age explained only approximately 35% of the variation in HbA_1c. In addition to the previously mentioned factors, one must consider that only a single value was used to reflect postprandial hyperglycemia and more of the variation might have been explained had more sampling times been used, such as with continuous plasma glucose monitoring.³⁶ Nevertheless, it would have to be assumed that the measurements on a single day accurately reflected the average of the preceding 2 to 4 months, which is what the HbA_1c is thought to reflect.

To our knowledge, ours is the only study to examine the relationships between fasting and postprandial plasma glucose levels and HbA_1c values in people with HbA_1c values within the reference range. However, Monnier et al³⁶ recently examined the relative contribution of fasting and postprandial plasma glucose levels to day long hyperglycemia in people with type 2 diabetes whose HbA_1c values ranged from below 6% to above 9%. These investigators sampled plasma glucose levels at 8 AM, 11 AM, 2 PM, and 5 PM and divided their 290 subjects into quintiles based on HbA_1c levels; they found that as HbA_1c increased, the contribution of postprandial plasma glucose levels to day long hyperglycemia decreased from approximately 70% (lowest quintile) to approximately 30% in the highest quintile.

The patients studied by Monnier et al³⁶ in their lowest quintile had an HbA_1c level virtually identical to that of our subjects in our highest decile (ie, 6.45% vs 6.46%). Thus, our conclusion that postprandial hyperglycemia contributes approximately 1.5 times more to HbA_1c than do FPG levels in individuals with normal or near-normal HbA_1c values is consistent with the findings of Monnier et al³⁶ in their lowest-HbA_1c-quintile subjects (eg, a 70% contribution of postprandial hyperglycemia would represent 2.3-fold the contribution of fasting hyperglycemia). Indeed, in our subjects with HbA_1c values between 6% and 7% (Table 5), FPG levels were increased above normal by approximately 3 mg/dL (0.17 mmol/L), whereas their 2-hour PCPG levels were increased by approximately 70 mg/dL (3.89 mmol/L).

Our findings that most people with IGT have a normal FPG level and that in these individuals and those with NGT the 2-hour PCPG level contributes more to HbA_1c than the FPG^13-15 have several important clinical implications. These observations suggest that the increased risk for CVD found in people with IGT and normal HbA_1c levels could be attributable to postprandial hyperglycemia. This conclusion is further supported by our evaluation of FPG and 2-hour PCPG levels of individuals with HbA_1c values similar to those of the EPIC-Norfolk study. That study found that HbA_1c values between 5.0% and 5.4% were associated with a 2.5-fold increased risk of death from CVD and ischemic heart disease.³⁰ We found that 81% of individuals with HbA_1c values between 5.0% and 5.5% had a normal FPG level, whereas only 58% had a normal 2-hour PCPG level.

Impaired glucose tolerance is widely recognized as a precursor of type 2 diabetes mellitus^37-39 and a risk factor for CVD.^5,17,18 Pharmacologic and lifestyle interventions^37-40 have been shown to improve IGT and reduce the risk of developing type 2 diabetes mellitus. The IDF recommends that if IGT cannot be reversed by lifestyle changes, pharmacologic intervention should be considered.²⁰

Thus, identification of individuals with IGT is important. However, numerous studies^12-14,20-24 indicate that use of only FPG determinations, as currently recommended by the ADA, is suboptimal. Our data confirm these findings in that of all individuals in our database with IGT (n = 155), only 17% had an abnormal FPG level,^14,20,21,24 which supports the WHO and IDF^18,20 recommendations that the oral glucose tolerance be the main diagnostic procedure.

A recent IDF consensus panel suggested that one reason for the lack of sensitivity of the FPG in detecting IGT is that the upper current limit of normal for FPG may be too high.²⁰ Our data support this view. Individuals in our database with NGT had a mean FPG level of 85.7 mg/dL (4.76 mmol/L) (Table 1), a value identical to that found in a meta-analysis of 34 published studies, including NHANES III data (ie, 85.7 mg/dL [4.76 mmol/L]).⁴¹ Since the standard deviation of FPG in our study was 8.3 mg/dL (0.46 mmol/L), this implies an upper limit of normal (mean ± 2 SD) of 102 mg/dL (5.66 mmol/L), which is less than the currently accepted value of 109 mg/dL (6.05 mmol/L).

The ADA currently recommends an HbA_1c value of less than 7.0% as a target for acceptable glycemic control primarily based on the risk for microvascular complications.⁸ However, macrovascular complications account for most of the morbidity and mortality in people with type 2 diabetes mellitus.^25,26 On the basis of epidemiological data that suggest that a lower HbA_1c level may be needed to prevent these complications,^6,20 the IDF and ACE recommend an HbA_1ctarget of 6.5% or less.^9,10 Our data indicate that the difference in achieving IDF and ACE targets vs that recommended by the ADA would largely depend on reducing postprandial hyperglycemia, since individuals who met the IDF, ACE, and ADA targets had similar FPG levels and differed only in 2-hour PCPG levels. This conclusion is supported by the recent study of Monnier et al,³⁶ which showed that in type 2 diabetic patients with HbA_1c values that averaged less than 7.3%, postprandial hyperglycemia accounted for approximately 70% of day long hyperglycemia.

Correspondence: John E. Gerich, MD, Department of Medicine, University of Rochester School of Medicine, 601 Elmwood Ave, Campus Box MED/CRC, Rochester, NY 14642 (johngerich@compuserve.com).

Since this manuscript was accepted for publication, the ADA has reduced its upper limit of normal for FPG to 100 mg/dL.

Accepted for publication September 16, 2003.

This study was supported in part by grant 5MO1 RR-00044 from the Division of Research Resources, General Clinical Research Center, Bethesda, Md, and grant DK-20411 from the National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda.

We thank Mary Little for her excellent editorial assistance and the nursing and laboratory staff of the General Clinical Research Center for their superb help.