The wonders of lactate
Very recently, our laboratory has made a major improvement in the use of the minimal model to assess insulin sensitivity and evaluate its importance in metabolic control. During the FSIGT, as it is practiced, there is initially a large increase in glucose concentration, due to the glucose injection, there is an endogenous finite insulin response in normal individuals, and the insulin is then increased again at 20 min after the exogenous insulin injection. We considered the question of the fate of the injected glucose before the secondary insulin injection. In the absence of a large increase in insulin, much of the glucose is disposed by an insulin-independent mechanism, a process we termed “glucose effectiveness.” Tissues which can dispose of glucose independent of a large dynamic insulin response include the brain, the liver, and red blood cells. In fact, hepatic glucose uptake is independent of the dynamic insulin response . We have therefore hypothesized that the major site of glucose uptake during the FSIGT is by the liver. What is the fate of the glucose in the liver during the FSIGT? Under resting conditions, little of the glucose will be anaerobically metabolized, as there is little need for additional energy. Liver glycogen synthesis under these conditions will proceed slowly . Therefore it is likely that much of the glucose will proceed down the glycolytic pathway, and exit the liver as lactate. It is possible, therefore, that the plasma lactate levels during the FSIGT reflect the ability of the liver to take up glucose in the hyperglycemic situation. The rate of liver glucose uptake is limited by the phosphorylation of glucose in the liver — which is limited by the activity of glucokinase in the liver. We have modeled the relationship between plasma glucose and lactate during the FSIGT (Figure 15.7; ), and this model yields a parameter which represents the activity of glucokinase in the liver. We therefore can suggest that the activity of glucokinase can be estimated from the FSIGT. This activity is important because it is likely the major component of the glucose effectiveness.
There is now evidence that glucose effectiveness is under control by the CNS ; the latter concept has emerged from the use of the minimal model to assess liver glucose uptake. The latter studies, conducted in collaboration with M. Schwartz and his colleagues, utilize the minimal model in the rodent. In fact, we have collaborated with Alonso and O’Donnell and colleagues in their development of a minimal model procedure in the mouse . Additionally, a genetic variant has been identified for the expression of glucokinase regulating protein, and it is possible in the future that the use of the minimal model accessing lactate will be used for identifying new risk variants for T2DM and other insulin-resistance syndromes.
Hyperbolic relationship between insulin sensitivity and insulin secretion: the disposition index
Diabetes, characterized by elevated blood glucose concentration, is reflective of an inability of the patient to regulate the plasma glucose concentration within normal limits. How is this regula- tion maintained in the normal individual?
Insulin resistance is common in Westernized societies. Yet, most insulin-resistant individuals do not have diabetes. Diabetes is prevented by the compensatory characteristics of the pancreatic β cells. Insulin sensitivity is reduced in many normal as well as pathophysiologic states. Among the normal states are pregnancy and puberty; pathologies include obesity and infection. Pathologies include not only T2DM, but hypertension, colon cancer, and polycystic ovarian syndrome.
Diabetes is prevented in many states of severe insulin resistance by the robust compensatory function of the cells of the pancreas. In fact, there is a predictable β-cell response to insulin resistance (Figure 15.8) :
SI × AIRglucose = DI
where the three variables are defined as in Table 15.1. This equation is a rectangular hyperbola, which predicts that for a given decrease in insulin sensitivity, there will be a proportionately equal and opposite increase in β-cell sensitivity to glucose stimulation. The disposition index defines the position of the hyperbolic curve on the sensitivity/secretion plot. If the DI is high, then the β cells mount a robust response to a decrease in SI; if the DI is low, such a value is reflective of unresponsive β cells. Subjects at risk for T2DM have a low disposition index [89,106–109]. There is evidence that the DI value is determined by genetics as well as environment [82,110,111]. The minimal model, therefore, provides a method for assessing insulin sensitivity in the intact organism by combining a relatively straightforward experimental protocol with the power of the digital computer to model real data. Software to calculate parameters has been available for some years making the methodology available to all clinical investigators. The method continues to be used in laboratories throughout the world.
Surrogate measures of insulin sensitivity
Laboratory methods for measuring insulin action have been shown to be accurate. However, under certain conditions application of these methods is difficult if not impractical. Some studies call for assessment of insulin action in even hundreds of thousands of subjects. Of course, the glucose clamp is not amenable to larger studies. The minimal model has been used to measure insulin sensitivity and DI in studies of populations in excess of 1000 [94,110,112], but not in multiples of that number of subjects. Are there simpler but accurate methods for use in large populations? We shall focus on two classes of “surrogate” measures of insulin sensitivity: indices based upon fasting measures, and indices based upon the OGTT.
Kahn and his colleagues have reported a curvilinear relationship between fasting insulin and insulin sensitivity , one that is similar to stimulated insulin release. In an individual with healthy β cells, because there is a defined relationship between secretion and insulin sensitivity, it could be argued that measuring one is as good as measuring the other. Thus, if insulin sensitivity is low, fasting insulin will be elevated, and vice versa. Therefore in a group of subjects with similar β-cell function we may expect elevated insulin to reflect insulin resistance. Similarly, with healthy β cells, insulin resistance will be reflected in elevated fasting insulin, as well as elevated β-cell secretory response to stimuli such as glucose, arginine, GLP-1, or other secretagogues. It is arguably appropriate to utilize fasting insulin or a provocative index of β-cell function as a surrogate with the following caveat: fasting insulin per se, or other indices based upon the fasted state, should be considered accurate surrogates for insulin sensitivity only when comparing between or among individuals with equivalent functional activity of the pancreatic β cells .
The downside associated with using fasting insulin (or other related indices of insulin secretion) becomes clear if we consider comparing among groups or individuals with differing islet secretory function. Let us consider as an example comparison of fasting insulin between normal individuals and individuals with IGT. The latter group is not only insulin resistant, but is charac- terized by a β-cell defect of at least 50% [115–117].
Comparing the normal subjects with insulin-resistant subjects with IGT, plasma insulin concentration, or insulin response to glucose could be identical. Thus, the fasting insulin concentration will not represent an accurate reflection of insulin sensitivity comparing individuals or groups for whom cell function is not identical. In practice, individuals with similar β-cell function are not often compared; it is more usual to compare normal subjects with those at risk for disease (e.g. IGT [115–117], gestational diabetes after term [87,107], first-degree relatives of type 2 diabetic patients [118,119]).
The previous discussion demonstrates that fasting insulin per se cannot be considered an accurate surrogate for insulin sensitivity (or resistance) under most conditions. Nevertheless, there is a substantial literature using the fasting insulin value for just this purpose [120–122]. At the very least, the careful observer must consider with care reports of fasting insulin as a surrogate for insulin resistance. If there is even a latent subtle β-cell defect in a group of subjects, any reported value of fasting insulin will likely underestimate the degree of insulin resistance.
Other surrogate measures
Besides fasting insulin itself, there are various other surrogate measures based upon fasting values alone, which have been exploited in population studies. The first to appear in the literature was the index proposed by Dr. Peter Bennett. Others include the homeostasis model assessment index insulin resistance measure (HOMA), and the “QUICKI” [123–136]. Note that these indices are indeed similar; they represent insulin resistance as proportional to either the product or the inverse of the product of fasting insulin and fasting glucose. In fact, in nondiabetic subjects plasma glucose differs little compared to large variance in fasting plasma insulin. Therefore, in non- diabetic subjects, both HOMA index and Bennett’s index are approximately proportional to the fasting insulin concentration. As discussed, similar to fasting insulin, these indices will not reflect insulin sensitivity accurately in nondiabetic subjects with differing β-cell function.
Quon and colleagues introduced the “QUICKI” index . Quon et al. equate insulin resistance to the inverse of the sum of the logarithms of fasting plasma glucose and fasting plasma insulin. The value of insulin resistance emerging from this index will depend upon the units of glucose and insulin, and the relative importance of these components will depend upon whether they are expressed in European (SI) units (mM glucose and pM insulin) or in units used in the United States (mg dL−1 glucose and μUmL−1). Also, despite the logarithmic transformation, QUICKI values exhibit nonlinear proportionality to fasting plasma insulin and will, like fasting insulin, underestimate insulin resistance in a population with a latent decrease in β-cell function. In fact, QUICKI and HOMA are formally identical to each other, as evidenced in the measures made in a large number of individuals (Figure 15.9).
The HOMA index of insulin resistance is extremely simple to calculate ((fasting glucose × fasting insulin) / 22.5). Therefore it has been used in a very large number of publications. However, it must be used with great caution. Because the fasting insulin is dependent not only upon insulin sensitivity, but also on insulin secretory function as well as metabolic clearance of insulin, the HOMA is a mixed index dependent upon all these physiologic parameters, and is not a direct measure of insulin sensitivity. The limitations of this measure have recently been documented in a series of studies [137,138]. Xiang and colleagues have shown that HOMA does not detect longitudinal changes in insulin resistance in a population at risk for diabetes . It fails to detect insulin resistance associated with surgical trauma . In a very recent study from our group done by Ader et al. , we examined whether HOMA could detect insulin resistance associated with a fat diet in the canine model. We measured insulin sensitivity with the euglycemic clamp, with the minimal model, and with HOMA. While the clamp and FSIGT values were strongly correlated, there was no significant correlation between the HOMA index and insulin sensitivity measured by either clamp or FSIGT. More significant, there was a correlation between HOMA and insulin response (Figure 15.10). Therefore based upon this study, HOMA results must be treated with a degree of skepticism, regarding insulin sensitivity per se.
Thus, indices based upon fasting measurements of insulin alone or fasting insulin plus fasting glucose should not be considered accurate indices of insulin resistance comparing groups with possible inequalities in β-cell health. Because we are often comparing between individuals or groups with differing islet function, it must be concluded that fasting insulin, as well as other surrogate measures based upon fasting values alone, may not be accurate. Certainly they are inaccurate compared to measures available from laboratory procedures such as the euglycemic clamp or the minimal model.