Progressive Loss of β-Cell Function Leads to Worsening Glucose Tolerance in First-Degree Relatives of Subjects With Type 2 Diabetes

The subjects were 33 (9 male and 24 female) Caucasian, nondiabetic, first-degree relatives of individuals with type 2 diabetes. Twenty-six subjects had participated in the American Diabetes Association’s Genetics of Non-Insulin Dependent Diabetes Mellitus (GENNID) Study (14) and returned 7 years later for the follow-up study. These 26 subjects had both a frequently sampled intravenous glucose tolerance test (FSIGT) and an oral glucose tolerance test (OGTT). An additional 7 individuals had not participated in the GENNID Study, and their first assessment was done by FSIGT only, whereas their follow-up study 9 years later involved both an FSIGT and an OGTT. At both times they were studied, all subjects were apparently healthy and had not been counseled or studied between the two assessments. The study was reviewed and approved by the Human Subjects Review Committee at the University of Washington, and written informed consent was obtained from each participant.

Weight, height, waist, and hip circumferences and blood pressure were measured at the first assessment for the 26 GENNID Study subjects and at the second assessment for all 33 subjects. A tolbutamide-modified FSIGT was performed to quantify insulin sensitivity, insulin release, and intravenous glucose tolerance (4). Briefly, glucose (11.4 g/m² body surface area) was infused intravenously over 1 min, and 20 min later tolbutamide (125 mg/m² body surface area) was administered intravenously over 30 s. Three basal blood samples were drawn before glucose injection, and 32 samples were drawn over the 240 min after glucose administration for glucose and insulin measurements.

A 75-g OGTT was performed after a 10-h overnight fast. Samples were drawn at baseline and every 30 min after glucose ingestion up to 120 min. Subjects were classified as having normal glucose tolerance (NGT), IGT, or diabetes (15).

Plasma glucose was measured using the glucose oxidase method. Plasma insulin concentrations were measured by radioimmunoassay (4), in which the primary antibody detects insulin, proinsulin, and proinsulin conversion intermediates on an equimolar basis. Fasting plasma triglycerides and total cholesterol were determined by enzymatic analytical chemistry. HDL cholesterol was measured after heparin–manganese sulfate precipitation. LDL cholesterol was calculated using the Friedewald equation. Fasting adiponectin was measured as described previously (16).

Insulin sensitivity was quantified as the insulin sensitivity index (S_I) using the minimal model of glucose kinetics. The minimal model also provides an estimate of glucose effectiveness at basal insulin (S_g), which can be used to calculate glucose effectiveness at zero insulin (GEZI) as S_g − (basal insulin × S_I) (13). Insulin release was quantified as the acute insulin response to glucose (AIR_g), the mean of the incremental insulin response between 2 and 10 min after glucose injection.

The disposition index (DI), which provides a measure of β-cell function, was calculated as S_I × AIR_g, based on the known hyperbolic relationship between these two variables (4), a relationship that is not affected by sex or differences in glucose tolerance (4,17). Percentile rankings of DI were calculated on the basis of data from our laboratory (4). As the relationship between S_I and AIR_g is nonlinear and to allow plotting of the data from initial and follow-up assessments on the hyperbolae for the control cohort, geometric means were calculated for S_I, AIR_g, and DI.

The glucose disappearance constant (K_g) provides an estimate of intravenous glucose tolerance. It was calculated as the slope of the linear regression line relating the natural logarithm of the glucose concentration over time from 10 to 19 min after glucose administration. The trapezoidal rule was used to calculate the incremental area under the curve for glucose (AUC_G) during the OGTT as a measure of oral glucose tolerance.

Data are presented as arithmetic means ± SD unless otherwise indicated. Comparisons of data obtained at the two time points were performed by paired t test or by a Mann-Whitney U test. Relationships between continuous variables were assessed using Spearman’s correlation coefficient and by multiple regression analysis. Multiple linear regression analysis was also performed to determine whether progression to IGT was related to a given variable at the second assessment while adjusting for the baseline value. A two-sided P ≤ 0.05 was considered significant.

At the first assessment, subjects were 37.9 ± 12.1 years of age. The time between the first and second assessments was 7.3 ± 1.2 years and during this period the 33 individuals remained apparently healthy.

As shown in Table 1, the average weight gain was 7.6 kg. Thus, although the cohort’s average BMI placed them in the range of overweight at the first assessment, at follow-up their average BMI was in the obese range. Waist circumference increased, without a change in hip circumference. In parallel with this increase in central adiposity, LDL cholesterol and triglycerides increased, as did systolic blood pressure (Table 1). Fasting glucose or insulin levels did not change.

At their first assessment, subjects had an S_I of 4.6 ± 2.7 × 10⁻⁵ min⁻¹/pmol), which did not change during follow-up (P = 0.56) (Table 1), consistent with the unchanged fasting insulin levels. Adiponectin also did not change over time (P = 0.24) (Table 1). In contrast, insulin release, measured as AIR_g, was 16% lower at the second assessment (P = 0.08). β-Cell function, determined as DI, decreased by 22% over the same period (P = 0.02) (Table 1). Glucose effectiveness did not change over time whether measured at basal insulin (P = 0.22) (Table 1) or when corrected to zero insulin (P = 0.39) (Table 1).

We examined the percentile ranking of DI for these high-risk subjects using equations previously published for this relationship (4). The first-degree relatives had a percentile ranking of 15 at their first assessment, and this decreased to 3 at the second assessment (Fig. 1A). In line with the decrease in β-cell function, intravenous glucose tolerance (K_g) decreased significantly (P = 0.03) (Table 1).

We examined whether the evolution of insulin sensitivity, insulin secretion, or glucose effectiveness was correlated to the change in weight or in waist circumference over time. We observed no correlation between weight gain or increased abdominal contour and insulin sensitivity. In contrast, the change in weight was inversely correlated with the change in DI (r = −0.36, P = 0.02), whereas the change in waist circumference was inversely correlated with the changes in AIR_g (r = −0.50, P = 0.02) and DI (r = −0.46, P = 0.07). No significant correlations were found with glucose effectiveness.

Oral glucose tolerance, measured as AUC_G, decreased over the 7-year follow-up period (P = 0.0003) (Table 1). The 2-h glucose level also increased (P = 0.03) (Table 1), and this was associated with changes in glucose tolerance categories. Thus, among the 16 subjects who had NGT at the first assessment, 6 developed IGT 7 years later, whereas 10 retained NGT. Among the 10 subjects with IGT at the first assessment, two reverted to NGT, four still had IGT, and 4 progressed to diabetes.

To further assess the variables determining the progression from NGT to IGT, we examined the 16 subjects with NGT at the first assessment. The 10 subjects who maintained NGT during the follow-up period (nonprogressors) were compared with the 6 subjects who developed IGT (progressors). At the first assessment, when all subjects had NGT, progressors and nonprogressors values for S_I, AIR_g, DI, or S_g did not differ significantly, whereas GEZI was lower (Table 2). Over time, S_I declined by 12% in the progressors and by 3% in the nonprogressors, whereas AIR_g decreased by 25% in the progressors compared with 11% in the nonprogressors. Thus, β-cell function, determined as DI, decreased by 38% in the progressors and by 20% in the nonprogressors (Table 2). These differences in the responses translated to a decline in the percentile rankings for DI from 21 to 5 in the progressors and from 45 to 35 in the nonprogressors (Fig. 1B). At the end of follow-up, β-cell function was lower in the progressors than in the nonprogressors (P = 0.05). Further, progression to IGT was associated with a lower DI at follow-up after adjustment for the baseline DI (regression coefficient −0.598, P = 0.04). Thus, the decline in DI was greater for progressors than for nonprogressors (518 vs. 289). S_I and AIR_g, however, were not associated with progression to IGT in similar linear regression models. As regards glucose effectiveness, GEZI declined by 10% in the progressors and by 11% in the nonprogressors. Thus, although this parameter was significantly different at the first assessment, at the second assessment this measure only tended to be lower in the progressors (P = 0.10) (Table 2). Adiponectin levels were not different between progressors and nonprogressors at the first or second assessment (Table 2).

In keeping with the evolution of β-cell function, intravenous glucose tolerance and oral glucose tolerance did not differ between the progressors and nonprogressors at the first assessment. However, both were significantly lower in the progressors at follow-up (P = 0.03 and P = 0.001, respectively) (Table 2).

As in the subjects who initially had NGT, in the 10 subjects who initially had IGT, the percentile rankings for DI changed in line with the conversion to different glucose tolerance states. Thus, in the four subjects who progressed to diabetes, the percentile ranking decreased from 4 to 1, whereas in the six subjects who did not progress, it evolved from 6 to 9.

The relative importance of insulin resistance and β-cell dysfunction in the development of type 2 diabetes has been a long-standing debate (1–3,5,6,18,19). Contributing to this debate has been the lack of recognition that, as with all endocrine systems, insulin release is regulated by a feedback loop between the insulin-sensitive tissues and the β-cell (4). Thus, when insulin sensitivity decreases, the insulin response should increase reciprocally such that β-cell function is not changed, and glucose tolerance is maintained. In the current longitudinal study, this reciprocal response did not occur, reflecting a decline in β-cell function over time. This decline was associated with a decrease in glucose tolerance in the whole cohort and the development of diabetes in 4 of the 33 subjects. Among the 16 subjects who initially had NGT, 6 developed IGT. Progression from NGT to IGT was associated with a reduction in β-cell function, which was significantly different from the response in the 10 subjects who retained NGT. Thus, a progressive loss of β-cell function characterizes the deterioration in glucose tolerance that occurs in high-risk individuals, even at an early stage of the disease process.

A number of groups are at increased risk of developing type 2 diabetes, including individuals with a family history of the disease, those with IGT, older individuals, and women with a history of gestational diabetes or polycystic ovary syndrome. Cross-sectional studies in which β-cell function was assessed by considering the insulin response relative to insulin sensitivity have demonstrated that these groups have reduced β-cell function (7,8,18,20). However, few longitudinal studies in these high-risk groups have been performed. In one study in 155 offspring of diabetic parents (2), it was concluded that insulin sensitivity was the major determinant of progression. However, this analysis did not account for the modulating effect of insulin sensitivity on the insulin response. Evidence that a decline in β-cell function is a critical determinant of deteriorating glucose tolerance comes from longitudinal studies in Pima Indians (19) and in postmenopausal women (21) and from the U.K. Prospective Diabetes Study (22). Recently, a role for impaired β-cell compensation in the progression to IGT and diabetes was also demonstrated in the Insulin Resistance Atherosclerosis Study (23) and in Hispanic women (24).

The mechanisms underlying the progressive decline in β-cell function are not fully understood. It may be related to a genetic predisposition compounded by environmental exposure such as increased caloric intake and the development of obesity. Interestingly, we found that the development of central adiposity was associated with loss of β-cell function, suggesting that changes in central or visceral fat–derived factors may predispose to β-cell dysfunction in high-risk individuals. Similarly, others have found that central body fatness and the increase in fat over time were the major predictors of a decline in the homeostasis model assessment insulin secretion index in women with a family history of diabetes (25).

The subjects in the current study were relatively insulin resistant, in keeping with studies in offspring of diabetic parents (2). Although subjects gained weight over time, their insulin sensitivity did not decline, consistent with data from the Pima Indians, in whom little change in insulin sensitivity occurred in subjects who gained weight and progressed to diabetes (19). This result may indicate that insulin resistance has been established and is near maximal so that additional weight gain does not discernibly reduce insulin sensitivity. Consistent with this finding, adiponectin did not change over time. β-Cell function was also poor at the first assessment (percentile ranking 15), and decreased to percentile ranking 3 during follow-up. Thus, insulin resistance is likely to be involved in the pathogenesis of type 2 diabetes, but the progressive loss of β-cell function appears to be the critical determinant for disease progression from NGT to IGT and then to type 2 diabetes.

Our observations were made using insulin measurements. Because insulin clearance is affected by insulin sensitivity, with clearance being less in insulin-resistant individuals (26), our findings could be said to be even more striking. Using C-peptide measurements, we probably would have found an even greater change over time. The difference between measurements using insulin and C-peptide also raises the interesting concept that DI is a measure of both changes in β-cell responsiveness and modulation of hepatic insulin extraction. The aim of this integrated adaptation is to reduce pancreatic workload while simultaneously trying to maintain glucose tolerance. Our findings highlight the severity of the β-cell defect, as the reduced insulin clearance is unable to adequately compensate to maintain hyperinsulinemia and thus glucose tolerance.

We also evaluated the possible role of glucose effectiveness, an important determinant of glucose metabolism (10). Although we did not observe a significant change in this variable over time, it was lower at the first assessment in the subjects who subsequently progressed. This finding is in keeping with those of Martin et al. (12) and suggests that a reduction in insulin-independent glucose uptake may also contribute to progression to diabetes. Further work in this area using larger cohorts will help define whether this is an important contributor.

The Diabetes Prevention Program (DPP) (27) and the Finnish Diabetes Prevention Study (28) both demonstrated that a lifestyle intervention decreases the risk of development of diabetes by 58%. Further, the DPP found that metformin reduced the risk of development of diabetes by 31%, whereas the STOP-NIDDM, Troglitazone in Prevention of Diabetes (TRIPOD), and Diabetes Reduction Assessment with Ramipril and Rosiglitazone Medication (DREAM) studies reported 25, 50, and 60% risk reductions with acarbose, troglitazone, and rosiglitazone, respectively (29–31). The improvement in glucose tolerance with intervention may be the result of improved β-cell function along with enhanced insulin sensitivity, as observed after weight loss (32) and lifestyle and metformin interventions in the DPP (33). An improvement in insulin sensitivity and preservation of β-cell function was also demonstrated in the TRIPOD study (29). Thus, we believe that individuals with a family history of type 2 diabetes should be closely followed, and serious consideration should be given to interventions with lifestyle modification or medications to try to prevent the development of hyperglycemia that is the result of both insulin resistance and, importantly, a progressive loss of β-cell function.

This work was supported by the Medical Research Service of the Department of Veterans Affairs; National Institutes of Health Grants DK-02654, DK-17047, DK-026687, RR-37, and RR-16066; and the American Diabetes Association. M.C. was supported by a fellowship of the Belgian American Educational Foundation, J.V. by a fellowship from the Spanish Ministry of Science and Technology, and R.L.H. by a Juvenile Diabetes Research Foundation Fellowship. S.E.K. is the recipient of an American Diabetes Association Distinguished Clinical Scientist Award.

We thank the subjects, nursing staff of the Diabetes Research Group, Ruth Hollingworth and Katherine Cueno for their participation in the study, Francine Grandsard and Ivan Cnop for help with mathematical analyses and Françoise Féry for comments during manuscript preparation.