The Normal 93: Linking the β-Cell and Insulin Sensitivity: Revisiting a 1993 Diabetes Classic by Kahn et al. Free

The Classics in Diabetes article for this month, “Quantification of the Relationship Between Insulin Sensitivity and β-Cell Function in Human Subjects: Evidence for a Hyperbolic Function,” was published in Diabetes in 1993 (1). This article provided a model for systemic glucose regulation that continues, more than three decades later, to be very useful for conceptualizing normal physiology and diabetic pathophysiology, and provides a framework against which to test new ideas. The article is based on a secondary analysis of parameters of glucose tolerance in 93 healthy individuals who had been studied over several years by Steven Kahn and his colleagues at the University of Washington. It has had a lasting influence because it provides simple, straightforward support for a negative feedback relationship between insulin sensitivity and insulin secretion. The demonstration in this article that the insulin response to a glucose challenge has a reciprocal, hyperbolic relationship to systemic insulin sensitivity was consistent with, and unified, experimental observations from the preceding two to three decades. Moreover, the proposal of a homeostatic system of insulin action that regulates blood glucose provided a model that could be used to interpret data from physiological experiments in animals and humans, as well as to develop novel hypotheses and interpretive challenges that have advanced the field. In the intervening years, the major conclusions from this article have become a core precept to explain the pathogenesis of type 2 diabetes mellitus (T2DM), in addition to informing therapeutics and pharmacologic development aimed at regulating glucose tolerance.

This month’s classic was published at a time of heightened interest in the pathophysiology of T2DM. Through the 1970s differences between what had been called juvenile and adult diabetes had become better defined, enough so that the two types of diabetes were given new names, and understanding the distinct mechanisms underlying the conditions became an imperative. The 1980s saw intense focus on insulin resistance, its association with obesity, and a likely role in impaired glucose tolerance and T2DM. The widespread application of techniques like the hyperinsulinemic-euglycemic clamp established the nearly universal presence of insulin resistance in T2DM patients (2–4). Moreover, the reduced glucose disposal in response to infusions of insulin observed in humans with obesity was experimentally connected to hyperinsulinemia (5), and the concurrence of these features in individuals with impaired glucose tolerance suggested that insulin resistance was the first step in the slide toward diabetes, i.e., the proximal factor in the pathogenesis of T2DM (3). Fasting and postprandial hyperinsulinemia in individuals with obesity with normal glucose tolerance was taken as evidence for β-cell adaptation to insulin resistance (6). Still, the frequent observation that individuals with T2DM had circulating insulin levels not dissimilar to those in control participants without diabetes suggested that this compensation was impermanent and liable to collapse. A common synthesis of these findings was that obesity and progressive insulin resistance led initially to hypersecretion of insulin but that over time β-cell exhaustion led to an inability to mount an insulin response sufficient to maintain euglycemia (3,7).

Despite growing appreciation for the role of insulin resistance in diabetes and other cardiometabolic syndromes, and progress in understanding the biochemical and molecular mechanisms of insulin signaling (7,8), a decline of β-cell function was widely acknowledged to be the final step leading to hyperglycemia. The apparently normal plasma insulin levels present in people with T2DM occurred at levels of blood glucose that would be expected to generate much greater responses from healthy β-cells (9), and among individuals with diabetes, it was not possible to distinguish the effects of hyperglycemia and insulin resistance on β-cell function. While there was evidence of an inverse relationship between insulin sensitivity and insulin levels (10), most studies did not include sufficient numbers of individuals to resolve the nature of the relationship. A study by Bergman et al. (11) suggested that the relationship was a negative hyperbola, but this proposal was based on a relatively small sample size. At this point, the Seattle group, in collaboration with Bergman, embarked on their analysis.

The signature finding in this classic article is the well-defined inverse, curvilinear relationship between insulin sensitivity and insulin responses to intravenous glucose. That this relationship was best described by a hyperbolic curve was taken as confirmation of an earlier proposal that the product of insulin sensitivity and the insulin response is a constant (11). This tight linkage supports a regulated system in which stimulus (insulin) and response (insulin receptor signaling) were matched to give a stable output, e.g., normal glucose tolerance. This relationship as previously proposed (11), had introduced the term disposition index (DI), the product of insulin secretion and insulin sensitivity, to describe it. The data from the 93 individuals without diabetes added power to this observation, along with a rigorous statistical approach that left little doubt that secretion and sensitivity were connected. While the data demonstrating the inverse, hyperbolic relationship were cross-sectional and thus fixed in time, the authors posited that β-cell function could shift in response to temporal changes of insulin sensitivity and provided some preliminary data to support that this was indeed the case. Several studies have since borne out this inference, and in fact, a solid argument can be made that in people whose sensitivity to insulin improves, β-cell function is adjusted downward to maintain glucose tolerance (12,13). There is also evidence suggesting that insulin sensitivity is blunted in cases of pathologically increased insulin secretion, to mitigate hypoglycemia in cases of insulinoma (14,15). Thus, the initial conjecture of Kahn et al. that there is adaptability of both the β-cell response and insulin sensitivity that preserves glucose homeostasis is valid at least in specific settings (1).

In reflecting on the impetus for this study and some of the challenges required for its completion, Steven Kahn recalls:

The finding that people with normal glucose tolerance have insulin responses matched to their level of insulin sensitivity provided a new way to conceptualize impaired glucose regulation. The hyperbolic relationship defined whether peripherally measured insulin responses were appropriate for the degree of insulin resistance, and a DI within the 95% CI for a population of individuals without diabetes predicted normal glucose tolerance with a high level of surety. β-Cell compensation for insulin resistance was not always complete; e.g., significant reductions of insulin sensitivity were mostly compensated for but often associated with modest increases of fasting and prandial glucose. However, only individuals who fell outside the normal range of the hyperbolic curve, i.e., below the 5th percentile of the DI, typically had impaired or diabetic glucose tolerance. Moreover, the model predicted, and experimental evidence supported, that even very insulin-resistant individuals would maintain normal glucose regulation if they could mount an insulin response sufficient to overcome sluggish insulin signaling. This latter observation suggested that defective insulin secretion, or at least the failure of β-cells to be able to appropriately compensate for insulin resistance, was a central pathogenic feature of T2DM (18,19). The centrality of defective insulin secretion as the essential feature of T2DM has been borne out in genome-wide association studies where the majority of genes associated with the disease relate to β-cell function (20).

Although an inverse relationship between insulin sensitivity and insulin responses can be demonstrated in people with diabetes with measures from an oral glucose tolerance test, it is less robust than in people without diabetes (21) (Fig. 1). This is in great part because the incretin effect is reduced in diabetes (22), resulting in both parameters being abnormally low in people with T2DM, truncating the range for analysis. In addition, glycemia, a potential regulatory variable of DI, directly affects measures of insulin secretion and action. Nonetheless, combining the determinants of glucose tolerance has value in assessing the course of people with dysregulated blood glucose (23,24). For example, a lower DI is predictive of the progression from normal to prediabetes and prediabetes to diabetes (21,25), and a lower DI has been demonstrated in normal glucose tolerant first-degree relatives of people with diabetes, who subsequently progress over time (26). Interestingly, Pima Indians who are noted to progress to diabetes in longitudinal studies clearly have a lower DI than those who do not progress at a time when both groups are normal glucose tolerant (17). In addition, the benefit of treatments to improve glycemia, whether improving insulin sensitivity with a thiazolidinedione (27,28) or insulin secretion with a glucagon-like peptide 1 receptor agonist (29), is best reflected in an improved DI.

While a predictable inverse relationship between insulin secretion and insulin sensitivity is now widely accepted as a component of normal physiology, and has had utility in understanding of diabetes, this model has limits. The mechanism for establishing and connecting the two parameters in this relationship has never been fully explained. And although the determination of DI supports a key role for β-cell insufficiency in T2DM, a specific lesion has not been identified. Lastly, while the hyperbolic relationship between the insulin response and sensitivity is clear in healthy and glucose intolerant humans, the nature of the relationship is less certain in those with diabetes, as the range of their insulin responses is so much narrower (21). Nonetheless, the work of Kahn et al., now some 30 years removed, has had a major impact on how glucose regulation is conceptualized and studied. While aspects of the model can be debated (29), the principles described in this article have become fundamental to considerations of glucose tolerance and remain widely useful today, befitting a Diabetes classic.