Oral glucose tolerance: relationship with hemoglobin A1c

Diabetes Research and Clhffcal Practice, 3 (1987) 343-349 Elsevier

343

DRC 00146

Oral glucose tolerance: relationship with hemoglobin Ale H a r o l d S. Starkman, J. Stuart Soeldner and R a y E. Gleason Elliot P. Joslin Research Laboratory, Department of Medichle, Brigham and Women's Hospital, Harvard Medical School, New England Deaconess Hospital, Joslhl Clinic Division, and William E. Beetham Eve Unit, Joslin Diabetes Center. hlc., Boston, MA, U.S.A. (Received 26 November 1987, revised received 7 April 1987, accepted 30 April 1987)

Key words: Oral glucose tolerance; Insulin physiology; Hemoglobin Ale; Non-insulin-dependent diabetes mellitus; (Diabetes mellitus - - diagnosis); (Diabetes mellitus - - pathophysiology)

Summary Fifty-two normal non-obese males and 77 male offspring of two diabetic parents, aged 15-72 years, were studied to identify possible reasons for discordance between the oral glucose tolerance test (OGTT) and hemoglobin A1¢ (Hb Aa¢). Subjects were classified into four study groups: group 1 (n = 83), normal OGTT and normal Hb A~¢; group 2 (n = 19), normal OGTT and abnormal Hb Axe; group 3 (n = 9), abnormal OGTT and normal Hb Aa¢; and group 4 (n = 18), abnormal OGTT and abnormal Hb Ale. Glucose and insulin response were analyzed in each study group. Group 2 showed slightly higher mean glucose areas during the first 60 min of OGTT testing when compared with group 1 (P < 0.01). Insulin levels, insulin areas and insulin/glucose regression coefficients on group 2 did not differ significantly from group 1 during OGTT. Group 3 showed significantly higher mean blood glucose levels than subjects in group I (P < 0.0001) or group 2 (P < 0.001), but significantly lower mean blood glucose levels than subjects in group 4 (P < 0.04) throughout the OGTT. During the OGTT, group 3 showed marked absolute hyperinsulinism when compared with all other groups (P ~< 0.002). Also, relative hyperinsulinism in group 3 was suggested by the elevated insulin/glucose regression coefficient (1.86 + 0.50) when compared with group 1 (1.17 -4- 0.09), group 2 (0.92 + 0.011) or group 4 (0.55 + 0.17) (P ,N< 0.05). In this group mild oral glucose intolerance associated with normoglycemia on a day-to-day basis as indicated by a normal Hb Ale appears to be due to a state of insulin resistance and hyperinsulinemia.

Address for correspondence: Harold S. Starkman, M.D., 100 Madison Avenue, Morristown, NJ 07960, U.S.A. Supported by NIH Grants AM-09748, National Research Services Award from NIADDK, 5T32 AM-07260, the Massachusetts Lions Eye Research Fund, and the Joslin Diabetes Center, Inc., Boston, MA, U.S.A. Presented at the American Diabetes Association meeting, June 14, 1983, San Antonio, TX, U.S.A.

Introduction The use of both oral glucose tolerance testing [1] and hemoglobin Ale (Hb Ale) concentration [2-4] has been advocated for the detection of impaired glucose tolerance and mild type II diabetes mellitus in asymptomatic subjects. Although the results of

0168-8227/87/$03.50 © 1987 Elsevier Science Publishers B.V. (Biomedical Division)

344 these two testing methods are correlated with coefficients as high as 0.86 [4], agreement between oral glucose tolerance tests (OGTT) and Hb Ale levels has been variable and inconsistent. Elevations of Hb AI~ have been documented in up to 4% of subjects with normal OGTT [3]. Conversely, in those subjects with abnormal O G T T , elevations in Hb Ale have been noted in only 43-65% [2-4]. This discrepancy between O G T T and Hb A~¢ results has been attributed to such factors as irreproducibility of the O G T T on multiple testing [3], the criteria used for test interpretation [2], and the non-physiologic nature of the blood sugar rise during O G T T when compared to longer-term elevations necessary to produce changes in Hb A ~ [4]. Few data exist concerning glucose/insulin response in subjects showing agreement or disagreement of O G T T and Hb At~ test results. We hypothesized that individuals with an abnormal oral glucose tolerance test, but a normal fasting Hb A~¢, had insulin insensitivity with an adequate compensatory insulin secretion to provide for normoglycemia on a meal-to-meal basis, but had a relatively insufficient insulin secretory reserve to provide for a normal blood glucose response during the stress of the 100-g OGTT. Therefore, blood glucose, serum insulin and glucose/insulin relationships during O G T T were studied.

Subjects and methods

Stlbjccls A total of 129 male subjects were studied, including 77 offspring of two type II diabetic parents and 52 non-obese controls. Their ages ranged from 15 to 72 years. Subjects on medications or with medical conditions known to affect glucose tolerance at the time of testing were excluded.

Methods All subjects underwent a 4-h oral glucose tolerance test (OGTT). Hb Ale determination was performed on the fasting blood sample, The O G T T was performed after 3 days on a high carbohydrate (200300 g/day) diet and an overnight fast with a glucose dose of 100 g. Blood samples were obtained fasting and at 15, 30, 45, 60, 90, 100, 150 and 240 rain post glucose. Blood glucose levels were measured by a Technicon AutoAnalyzer using the ferricyanide method. Insulin levels were measured using a double-antibody radioimmunoassay technique [5]. Hb Ale was measured using a high-pressure liquid chromatography (HPLC) technique [6,7]. The O G T T s were classified as normal or abnormal by previously established criteria based upon 212 normal control subjects (Table 1). For each specific decile age group of non-obese (less than 120%

TABLE I UPPER LIMITS OF BLOOD GLUCOSE (mg/dl) IN NORMAL MALES FOR 100-g ORAL GLUCOSE TOLERANCE TEST (MEAN + 2 SD) The sum of increments of blood glucose in excess of mean + 2 SD to demarcate abnormality (97.5 + percentile)is >/12. Age (years)

10-19 20-29 30-39 40-49 50-59 60 +

n

20 124 47 13 4 4

Fasting

94 90 90 93 89 114

Time interval 30

60

90

120

180

146 168 162 153 189 198

156 161 165 166 186 206

127 129 128 156 159 219

113 113 117 146 148 178

117 96 97 95 113 144

345 TABLE 2 COMPOSITION AND CHARACTERISTICS OF SUBJECTS (n = 129) IN THE STUDY GROUPS AS DEFINED BY OGTT AND Hb AI~ CRITERIA Results are expressed a means + SEM.

Age (years) Percent of ideal weight Hb A~¢ (%)

Group 1 (n = 83)

Group 2 01 = 19)

Group 3 (n = 9)

Group 4 (n = 18)

35,0 + 1.3 105,8 + 1.8 5.0 4- 0.04

45.1 4- 3.1 113.2 + 6.2 6.1 + 0.08

47.5 + 3.5 116.2 4- 5.8 5.1 + 0.09

47.6 :t: 2.1 115.2 + 5.1 6.5 + 0.25

of ideal weight [8]) normal controls, the mean + 2 SD for blood glucose was calculated at fasting as well as at 30, 60, 90, 120 and 180 rain post glucose. Individual O G T T s from the 212 control subjects were then evaluated by subtracting the appropriate mean + 2 SD from the glucose levels obtained at the corresponding times. The sum of positive increments (where the O G T T glucose levels exceeded the criterium levels) was calculated for each and a frequency distribution of these positive increments was obtained. Any subject with a sum of positive increments greater than or equal to that at the 97.5 percentile was classified as having an ' a b n o r m a l ' O G T T [9]. The classification of H b Ale was based upon the frequency distribution of H b A~c results obtained from 39 normal male controls studied in this laboratory. The 97.5 percentile was 5.8%, and a H b Ale greater than or equal to this level was classified as 'elevated'. Thus, statistically equivalent criteria were devised for both the O G T T and the H b A~c level which targeted the upper 2.5% as ' a b n o r m a l ' or 'elevated'. All subjects were classified into one of four subgroups based on their O G T T and H b A ~ resuits, as follows: group 1, normal O G T T and H b Ate; group 2, normal O G T T , elevated H b A~¢, group 3, abnormal O G T T , normal H b Ate; and group 4, abnormal O G T T and H b A~¢. The demographic information for each group is summarized in Table 2. Glucose and insulin levels and glucose/insulin relationships during O G T T were compared amongst the groups.

Early glucose and insulin responses are represented by areas under the respective curves above fasting between 0 and 60 min for the O G T T . Total responses are represented by corresponding areas between 0 and 240 min post glucose for O G T T . In addition, the regression coefficient (b) of insulin upon glucose was calculated for each test. This statistic provides a measure of the insulin response per mg/dl of glucose stimulus during the glucose tolerance test. D a t a files were stored on a VAX/780 computer and statistical analyses were performed on an IBM 4341 computer using the SAS programming package [11]. In addition to standard statistical analysis, an analysis of covariance correcting for age and percent ideal weight was performed using the SAS general linear models procedure. Post hoc comparisons a m o n g the four subgroup means were made using a least squares means method. The alpha level for statistical significance was 0.05, and all results noted as being significant had a probability of 0.05 or less.

Results During the O G T T , the mean fasting blood glucose level in group 4 was significantly higher than in the other groups (P < 0.001) (Fig. 1). Mean blood glucose levels after oral glucose administration were significantly higher at each sampling time in group 4 subjects when compared to subjects in all other groups (P ~< 0.008). Mean glucose responses in

346

:°oI

,,

group 3 subjects were less than those in group 4, yet greater than those in group 1 from 15 to 210 min (P ~< 0.03), and in group 2 from 30 to 210 min (P ~< 0.02). Mean blood glucose levels in group 2 subjects differed significantly from those in group 1 only at 30 min when the group 2 value was significantly higher (P < 0.03). During the first 60 min of the OGTT, the largest mean early glucose response, as represented by the area under the glucose curve above fasting, was noted in group 3 followed closely by group 4. When compared to the other groups (Fig. 2) these early responses were significantly higher than those seen in groups 1 (P < 0.0001) and 2 (P < 0.003). The glucose area was significantly higher in group 2 when compared with group I (P < 0.01). When the mean areas under the total glucose curve above fasting from 0 to 240 rain were compared, the results were similar (Fig. 2) to those for early glucose response.

',,,,

'601 ,ooI

4O 20

0

30

60

90 120 150 TIME (minutes)

180 210

240

Fig. 1. Glucose levels (mean 4- SEM) during O G T T by study group (n = 129). Numerous statistically significant differences are enumerated in the text.

6 , 2 4 9

24,000 22,000

20,000

18,000 --NS--

16,000

--NS--

----~04

-.ooa -

---'.81---

I

~i

14,00(3

01-

)1-

12,00C 10,00(3

-.04

BOOC 600C

±

-NS-"-~008 - -

--01---.0001---,0001--

-=002 --NS-~0002-

I

:003--.O005---NS--

400( 2000 1234 GLUCOSE AREA 0-60mln [mg/dllmin)

1234 INSULINAREA 0-60 rain (M.U/ml/min)

J_

234 GLUCOSEAREA O-240mm {rag/all/rain) 1

,1

234 1234 INSULIN AREA SLOPEOF REGRESSION 0-240rain Insulin upon Glucose [M.U/mi/min) lM,U/ml per mg/dl)

Fig. 2. Glucose, insulin and insulin/glucose relationships during OGTT by study group (n = 129). The units of glucose areas are in mg/dl × min, the units of the insulin areas are ,uU/ml x min, and the unit for the slope of the regression of serum insulin upon blood glucose is #U/ml per mg/dl.

347

180 160

& _z co z

8C

6C "...~ GROUPI GROUP 2

o

"

was significantly less than that in group 1 (P < 0.04). There was no significant difference in early insulin response when groups 1 and 2 were compared. There were no statistically significant differences in mean total insulin responses when groups 1, 2 and 4 were compared from 0 to 240 min. Significant differences in the mean insulin/glucose regression coefficients (b, mU/ml of insulin per mg/dl of glucose) were also noted during the OGTT among the study subgroups (Fig. 2). The highest mean b value was seen in group 3 followed by groups 1, 2 and 4. The mean b value for group 3 was significantly higher when compared to all other groups (P <~ 0.04). In addition, the mean regression coefficient for group 1 was significantly higher than that for group 4 (P < 0.002). There was no significant difference in mean regression coefficients when groups 1 and 2 were compared.

.~o 6o " 9o " ~:~o ,~,o' ,~o' 2io 'z,;,o TIME (mlnules)

Fig. 3. Insulin levels (mean + SEM) during OGTT by study group (n = 129). Statistically significant differences are described in the text.

Although there were no significant differences in fasting insulin levels when groups 1 and 2 were compared, fasting insulin in group 3 was significantly higher than in all other groups (P ~< 0.0004) (Fig. 3). In addition, group 4 fasting insulin levels were significantly higher than those in groups 1 (P < 0.008) and 2 (P < 0.01). The fasting hyperinsulinism noted in group 3 persisted throughout the OGTT and the mean insulin levels were consistently elevated when compared with the other three groups (P ~< 0.02). Subjects in group 4 showed significantly higher mean insulin levels than subjects in group 1 at 240 min (P < 0.002) and subjects in group 2 from 180 to 240 min (P < 0.05). Mean insulin levels in groups 1 and 2 were not significantly different. During the first 60 min of the OGTT, the mean early insulin area above fasting was higher in group 3 than in any other group (P ~< 0.008) (Fig. 2). The next highest early insulin response was noted in group 1. The group 4 mean early insulin response

Discussion

In our study, 28 of 129 subjects (21.7%) showed discordance between the results of their OGTT and Hb Ale. Of the 92 subjects with a normal Hb Ale, 9 (9.8%) had blood glucose abnormalities during OGTT. In contrast, of the 37 subjects with an elevated Hb AI¢, 18 (48.6%) had abnormal OGTTs. Alternatively, of the 102 subjects with normal OGTT, 19 (18.6%) had elevations of Hb A~c as opposed to 18 of the 27 (66.7%) with abnormal OGTT. These results are similar to those reported by other investigators [3,4], and reflect the difficulty of using Hb AI~ as the sole criterium for diagnosis of impaired glucose tolerance or diabetes mellitus. Our data showed significant differences in insulin/glucose relationships during the OGTT among the study subgroups. As expected, subjects in group I (normal OGTT and Hb A~c) showed normal glucose and insulin dynamics and those in group 4 (abnormal OGTT and Hb Aac) showed significantly elevated blood glucose levels associated with a delayed and prolonged insulin response [12,13]. Notable, however, were the dynamics in the OGTT/Hb A~c discordant groups. Subjects in group 2, characterized by a normal OGTT and ele-

348 vated Hb Ale showed slight elevations of blood glucose levels during OGTT when compared to those with normal OGTT and Hb Ale, but significantly lower blood sugar levels than subjects in groups 3 and 4. The mean insulin levels in group 2, however, were not significantly different from those in normal subjects. In contradistinction, subjects in group 3 (with abnormal OGTT, but normal Hb At~ determinations) showed glucose levels significantly higher than those in groups 1 and 2, but lower than those in group 4. These blood glucose levels were accompanied by insulin levels that were significantly higher than in any other study group. Interestingly, this hyperinsulinism is both absolute and relative to the associated blood glucose levels. These data support the hypothesis that the discordance between OGTT and Hb A~ results may in fact be related to alterations in insulin/glucose relationships. Abnormalities in OGTT reflect a defect in glucose utilization after an acute non-physiologic glucose load as opposed to elevation of Hb At~ which appears to reflect long-term ambient blood glucose levels. Thus, subjects with normal Hb At~ values, but abnormalities in OGTT, may have a peripheral defect in insulin sensitivity well compensated for in daily life, but revealed during an acute oral glucose challenge. Tiros, the hyperinsulinism and increased blood sugars in group 3 suggest the possibility of a target organ defect with an associated 'insulin resistance'. Such previously described defects include reduced number of insulin receptors or binding ability as well as post-receptor abnormalities [14,15]. Subjects in group 2 with normal OGTT and elevated Hb A1¢ appear to have slightly higher blood sugars during OGTT (increased early and total glucose responses) with relative hypo-insulinism (decreased early and total insulin responses) though absolute insulin levels appeared to be normal. Perhaps these subjects tend to have slightly higher day-to-day blood glucose levels due to their relative hypo-insulinism and this in turn causes an elevation in Hb AI~. Although subjects with concordant normal OGTT/Hb Ate and concordant abnormal OGTT/Hb A~¢ represent distinct ends of the spec-

trum of glucose homeostasis, the clinical relevance of the two intermediate groups remains unclear. Perhaps these discordant groups are stages in the development of type II diabetes, or alternatively subtypes of type II diabetes mellitus. Further longitudinal studies are needed to more clearly delineate the nature of this observed discordance.

Acknowledgements We wish to thank Ms M. Wacks, Ms T.M. Smith, R.N., Ms C. Labrie, R.N., Mrs. M. Grinbergs and Mrs. D. Shen for their aid with the technical aspects of this research project. In addition, we would like to thank Ms R. Russo and Ms L. Ryan for their assistance in preparing the manuscript.

References I Seltzer, H.S., Diagnosis of diabetes. In: Diabetes Mellitus: Theo O, and Practice, Medical Examination Publishing Co., New York, 1983, pp. 415-450. 2 Dunn, P.J., Cole, R.A., Soeldner, J.S. and Gleason, R.E., Reproducibility of hemoglobin A1, and sensitivity to various degrees of glucose intolerance, Ann. hltern. Med., 91: 390396, 1979. 3 Koenig, R.J., Peterson, C.M., Kilo, C., Cerami, A. and Williamson, J.R., Hemoglobin A i, as an indicator of the degree of glucose intolerance in diabetes, Diabetes, 25: 230-232, 1976. 4 Santiago, J.V., Davis, J.E. and Fisher, F., Hemoglobin Ale levels in a diabetes detection program, J. Clin. Endocrhwl. Metab., 47: 578-580, 1978. 5 Soeldner, J.S. and Slone, D., Critical variables in the radioimmunoassay of serum insulin using the double antibody technique, Diabetes. 114: 771-779, 1965. 6 Cole, R.A., Soeldner, J.S., Dunn, P.J. and Bunn, H.F., A rapid method for the determination of glycosylated hemoglobins using high pressure liquid chromatography, Metabolism. 27: 289-301, 1978. 7 Cole, R.A., Soeldner, J.S., Dunn, P.J. and Bunn, H.F., An automated method for the determination of hemoglobin Ale and total fasting hemoglobin using high pressure liquid chromatography. In: G.L. Hawk (ed.), Biological Biomedical Applications of Liquid Chromatography, Chapter 35, Marcel Dekker, New York, 1979, 569-577. 8 New Weight Standards for Men and Women, Statist. Bull. Metropolitan Life hTsurance Co., 40: I-6, 1959. 9 Kahn, C.B., Soeldner, J.S., Gleason, R.E., Rojas, L., Ca-

349 merini-Davalos, R.A. and Marble, A., Clinical and chemical diabetes in offspring of diabetic couples, New Engl. J. Med.. 281 : 343-347, 1969. 10 Ganda, O.P., Day, J.L., Soeldner, J.S., Connon, J.J. and Gleason, R.E., Reproducibility and comparative analysis of repeated intravenous and oral glucose tolerance tests, Diabetes. 27: 715-725, 1978. 11 Statistical Analysis System, Userk Guide: Statistics, A.E. Ray (ed.), SAS Institute, Cary, NC, 1982. 12 Seltzer, H.S., Allen, E.W., Herron, Jr., A.L. and Brennan, M.T., Insulin secretion in response to glycemic stimulus: re-

lation of delayed initial release to carbohydrate intolerance in mild diabetes mellitus, J. Clin. hlvest., 46: 323-335, 1967. 13 Reaven, G.M., Insulin resistance and insulin secretion in patients with chemical diabetes: implications concerning the pathogenesis of idiopathic diabetes mellitus. In: R.A. Camerini-Davalos and B. Hanover (eds.), Treatment of Earl)' Diabetes, Plenum Press, New York, 1978: pp. 187-200. 14 Olefsky, J.M., The insulin receptor: its role in insulin resistance of obesity and diabetes, Diabetes, 25:1154-1165, 1976. 15 Kahn, C.R., Insulin receptors and syndromes of insulin resistance, Diabetes Care. 5: 98-101, 1982.