Modeling Pathogenesis of Type 2 Diabetes

National Institute Of Diabetes And Digestive And Kidney Diseases

Investigators

Linked publications & trials

Abstract

In previous reports we have described our model for T2D pathogenesis. It builds on the foundational model (Topp et al, J. Theor. Biol. 2000; 206(4):605-19), which posited that moderate but persistent increases in blood sugar mediate negative feedback to increase insulin secretion by increasing beta-cell mass, either by increased replication or reduced apoptosis. However, if that increase fails to occur or is inadequate to restore normal glucose homeostasis, further increases in glucose raise it to a level where it becomes toxic to beta cells. Instead of negative feedback, there is then positive feedback, which causes a catastrophic loss of beta-cell mass and T2D. In addition to quantitative refinements to more accurately reflect the measured dynamics of T2D progression in humans and rodents, we included regulation of beta-cell function, in two distinct forms, in addition to mass. Data show that such changes are more rapid and more extensive than changes in mass, especially for humans, for whom beta-cell replication is very slow after adolescence. The model captures many key features of T2D progression, including the sudden deterioration of glucose control after a long period of gradual worsening (threshold behavior) and the fact that prevention is generally much easier than reversal, but drastic interventions, such as bariatric surgery and extreme caloric restriction can reverse established disease. The model has been further extended to track fasting and post-prandial glucose, rather than just average daily glucose, which is important because individuals differ in which aspect of glucose deviates first from normal. In addition, the model can be paused at any point during progression over years to simulate glucose tolerance tests, both oral (OGTT) and intravenous (IVGTT). We also used the model to investigate the hypothesis that high insulin causes insulin resistance rather than the other way around, as we assume in our model. Although there is strong evidence that hyperinsulinemia does contribute to insulin resistance, model simulations suggest that this plays at best a minor role in the development of diabetes. We have used the model to address an area in the literature that has been conflicted with regard to the greater risk of T2D among African-Americans (AA) compared to White Americans (WA). Measures of insulin resistance using the Minimal Model (Bergman et al, Am J Physiol.1979 236:E667) have interpreted as showing that AA are more insulin resistant. In contrast, measures using the euglyemic hyperinsulinemic clamp, considered the gold standard, generally do not show differences by race. We applied our model of glucose-insulin homeostasis to generate synthetic individuals with defined degrees of insulin resistance and beta-cell responsiveness to acute intravenous glucose stimulus. We then analyzed the synthetic data using the minimal model and simulated clamps. The main result was that individuals with higher glucose responsiveness may be artifactually reported as more insulin resistant. We also simulated oral glucose tolerance tests (OGTTs) for the same synthetic individuals and compared their characteristics to OGTTs in the literature and identified signatures that indicate a genuine difference in insulin resistance in individuals with high glucose responsiveness or no difference. Examples of both signatures are found in the literature. We concluded that caution needs to be used in interpreting differences in insulin sensitivity from the minimal model between groups that differ strongly in beta-cell glucose responsiveness. The results are detailed in Ref. #1. We continued to provide support to the long-term project of Dr. Anne Sumner (NIDDK) to develop methods of screening for T2D and T2D risk in Africa, where there are many challenges to using standard methods. Although it is tempting to use fasting plasma glucose because it is easy to obtain, it generally does not work well for people in Africa and African immigrants living in the US. Dr. Sumner found that the glucose at the two-hour point of an OGTT, though more resource intensive to acquire, works well and is highly reproducible for detecting T2D, but only modestly reproducible for detecting pre-diabetes. This is in part because individuals near the clinically defined threshold for prediabetes may transition back and forth across the diagnostic line multiple times before settling down to clear, sustained prediabetes, as reported in Ref. #2. Motivated in part by the findings in Ref. #2 we have been working on developing a new screening marker based on our model for glucose-insulin homeostasis. We have found that the disposition index (DI), originally derived from the minimal model, which represents the degree of insulin secretion relative to insulin resistance, is a better marker than glucose. The fundamental reason is that glucose increases very slowly as individuals transition from normal glucose tolerance to prediabetes, whereas the DI drops markedly during that transition because its fall reflects the underlying physiological defect of impairment of insulin secretion relative to insulin resistance long before glucose itself has risen substantially. This has great potential relevance to other groups at risk for T2D in addition to people of African descent. A paper is in preparation and will be discussed in a future report.