Beta-cell dysfunction and glucose intolerance: results from the San Antonio metabolism (SAM) study

Abstract

Aims/hypothesis. Both insulin resistance and beta-cell dysfunction play a role in the transition from normal glucose tolerance (NGT) to Type 2 diabetes (T2DM) through impaired glucose tolerance (IGT). The aim of the study was to define the level of glycaemia at which beta-cell dysfunction becomes evident in the context of existing insulin resistance. Methods. Insulin response (OGTT) and insulin sensitivity (euglycaemic insulin clamp) were evaluated in 388 subjects in the San Antonio Metabolism (SAM) study (138 NGT, 49 IGT and 201 T2DM). In all subjects the insulin secretion/insulin resistance index (DeltaI/DeltaGdivided byIR) was calculated as the ratio of the increment in plasma insulin to the increment in plasma glucose during the OGTT divided by insulin resistance, as measured during the clamp. Results. In lean NGTs with a 2-h plasma glucose concentration (2-h PG) between 5.6 and 6.6 and between 6.7 and 7.7 mmol/l, there was a progressive decline in DeltaI/AGdivided byIR compared with NGTs with a 2-h PG less than 5.6 mmol/l. There was a further decline in DeltaI/DeltaGdivided byIR in IGTs with a 2-h PG between 7.8 and 9.3 and between 9.4 and 11.0 mmol/l, and in Type 2 diabetic patients with a 2-h PG greater than It. I mmol/l. Lean and obese subjects showed coincident patterns of relation of 2-h PG to DeltaI/DeltaGdivided byIR. Conclusion/interpreation. When the plasma insulin response to oral glucose is related to the glycaemic stimulus and severity of insulin resistance, there is a progressive decline in beta-cell function that begins in "normal" glucose tolerant individuals

To: Godsland IF, Jeffs JA, Johnston DG (2004) Loss of beta cell function as fasting glucose increases in the non-diabetic range. Diabetologia 47:1157–1166

“…A similar observation has been reported in children [2], although most previous studies have demonstrated a decrease in insulin sensitivity in adults with an FPG >6.1 mmol/l (110 mg/dl) [3,4]. The defect in first-phase insulin secretion is reminiscent of the decline in insulin secretion observed in normal glucose-tolerant subjects with 2-h plasma glucose concentrations well within the normal range, reported by Gasteldelli and colleagues [5]. Both observations [1,5] suggest that the current cut-off values that define normal glucose tolerance do not correspond to plasma glucose levels that identify the presence of physiological alterations in insulin secretion in vivo.…”

Insulin secretion and impaired glucose tolerance

“…We have shown elsewhere [3], using published data from some of the authors of Mari et al and DeFronzo et al [4], that glycaemic responses to an OGTT conform closely to a simple HOMA-like model [5] relating insulin secretion, insulin sensitivity and glycaemia. The model can be expressed as:…”

“…That this model provides a strong fit to data (r 2 =0.83 for both fasting and 2 h OGTT responses [4]) is not proof that it is fundamentally true or complete. However, it demonstrates that, in this data set at least, the model can explain all the variation that is not due to likely measurement and within-subject errors.…”

“…However, it demonstrates that, in this data set at least, the model can explain all the variation that is not due to likely measurement and within-subject errors. The maximum r 2 obtainable in a regression analysis is determined by the product of the reliabilities of the x and y variables [6]; a simulation using a synthetic raw data set derived from Table 2 of Gastaldelli et al [4], and incorporating commonly reported measurement errors for glucose (3% CV), insulin (8%) and clampderived insulin sensitivity (10%) produces estimated reliability coefficients of 0.87 and 0.99 for y and x variables respectively, which predict a maximum r 2 =0.86 (compared with r 2 =0.83 [4]). This simulation ignored other potential sources of error, including within-individual errors.…”

“…In an OGTT, R in represents the sum of glucose entry from the gut and endogenous glucose output averaged over the duration of the test, which would not be predicted to vary markedly, and does not appear to be an important omission from the model when applied to the data of Gastaldelli et al [4]. Therefore, within the model a DI is a function of 1/G plus measurement error (Eq.…”