Cardiology Branch National Heart, Lung, and Blood Institute National Institutes of Health Bethesda, Maryland 20892
Address all correspondence and requests for reprints to: Michael J. Quon, M.D., Ph.D., Cardiology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Building 10, Room 8C-218, 10 Center Drive MSC 1755, Bethesda, Maryland 20892-1755. E-mail: quonm{at}nih.gov
Biological actions of insulin are essential for regulation and maintenance of glucose homeostasis. Insulin resistance (typically defined as decreased sensitivity or responsiveness to the metabolic actions of insulin) plays an important role in the pathophysiology of diabetes (1, 2). Insulin resistance is also associated with obesity (3) as well as hypertension, coronary artery disease, and dyslipidemias (4). Moreover, insulin resistance is a feature of a number of syndromes related to abnormal reproductive endocrinology, such as polycystic ovarian syndrome (5) and premature adrenarche (6, 7). Therefore, it is of great interest to quantify insulin sensitivity and resistance in humans to investigate the pathophysiology and epidemiology of major public health problems and to follow the clinical course of patients on various therapeutic regimens.
A host of methods have been developed to assess insulin sensitivity and insulin resistance in vivo. These include the hyperinsulinemic euglycemic glucose clamp technique (8), minimal model analysis of a frequently sampled iv glucose tolerance test (FSIVGTT) (9), and various indices derived from an oral glucose tolerance test (10, 11) or fasting glucose and insulin values (12, 13, 14, 15, 16). A number of variations on each of these approaches are available. For example, the glucose clamp technique can be performed under euglycemic, isoglycemic, or hyperglycemic conditions with or without infusion of tracer-labeled glucose (17). Likewise, minimal model analysis has been extended to analyze FSIVGTT modified by exogenous tolbutamide or insulin infusion (18) with or without infusion of tracer-labeled glucose (19).
The glucose clamp is generally regarded as the reference method for assessing insulin sensitivity in humans because it directly measures metabolic actions of insulin under steady state conditions. However, the clamp is also the most complicated method to implement because it requires simultaneous infusions of insulin, glucose, and potassium, multiple blood draws, and an experienced operator to adjust the glucose infusion appropriately over a 3- to 6-h time period. Minimal model analysis of an FSIVGTT is simpler to implement than the glucose clamp but still requires iv administration of glucose and insulin and multiple blood draws over a 3-h time course. Although minimal model results generally correlate with clamp measurements, identification of the minimal model index of insulin sensitivity in subjects with impaired insulin secretion (e.g. patients with diabetes) is problematic (15, 20). Moreover, there are systematic errors in minimal model estimates of glucose effectiveness and insulin sensitivity that may be due to oversimplified model representations of physiology (21, 22, 23). Simple indices of insulin sensitivity based on fasting glucose and insulin levels such as homeostasis model assessment (HOMA) (12) and QUICKI (15) are easily obtained and may be useful tools for large epidemiological studies. As discussed in a scholarly review on measurement of insulin sensitivity by Radziuk (25) in JCEM, the choice of an appropriate method to measure insulin sensitivity depends, in part, on the relative merits of each method for a particular application.
Recently, a number of studies have suggested that the fasting glucose to insulin ratio (G/I) may represent another useful method for assessing insulin resistance (13, 26, 27). However, unlike HOMA or QUICKI, which are based on the product of fasting insulin and glucose (12, 15, 28), G/I does not appropriately reflect the physiology underlying the determinants of insulin sensitivity. This issue has been nicely discussed in the context of the dynamics of an oral glucose tolerance test in a recent paper by Matsuda and DeFronzo (11). Similar arguments also apply to the G/I index obtained under fasting steady state conditions. In normal subjects who are fasting, glucose homeostatic mechanisms involving regulation of both hepatic glucose production and insulin secretion by pancreatic ß cells maintain glucose in the normal range. Under these steady state conditions elevations in fasting insulin levels (in the context of normal fasting glucose levels) correspond to increased insulin resistance. Indeed, in nondiabetic subjects, 1/(fasting insulin) is a well known proxy for insulin sensitivity that decreases as subjects become more insulin resistant (and fasting insulin levels rise) (29, 30, 31, 32, 33, 34). In the case of nondiabetic subjects, G/I is functionally equivalent to 1/insulin since the fasting glucose levels are similar for all subjects. This is the reason that the G/I ratio correlates with insulin sensitivity in nondiabetic patients with polycystic ovarian syndrome (13) or premature adrenarche (26, 27). Not surprisingly, in the paper by Vuguin et al. (26), a comparable correlation was observed when either the fasting insulin or G/I was compared with the minimal model index of insulin sensitivity.
The potential problems with using the fasting G/I ratio as a
physiologically appropriate index of insulin sensitivity become
apparent when fasting glucose levels are abnormal. This is easily
illustrated by comparing a normal subject with an insulin-resistant
nondiabetic subject and an insulin-resistant subject with type 2
diabetes (Table 1). As explained above,
when a normal subject is compared with a nondiabetic insulin-resistant
subject (whose fasting insulin level is elevated), simple indices of
insulin sensitivity based on fasting values such as 1/insulin, QUICKI,
and G/I are all decreased in the insulin-resistant subject when
compared with the normal subject, just as expected. Likewise, HOMA, an
index of insulin resistance, increases as expected. A diabetic subject
who has the same fasting insulin level as the nondiabetic
insulin-resistant subject is obviously even more insulin resistant
because the same level of insulinemia is not able to appropriately
compensate for fasting hyperglycemia. Importantly, in this diabetic
subject, the value for QUICKI is decreased even further and HOMA is
increased further, exactly as one might predict. However, 1/insulin
remains unchanged between the diabetic subject and the nondiabetic
insulin-resistant subject, and G/I paradoxically and erroneously
increases in the diabetic subject. Thus, QUICKI and HOMA both behave
qualitatively as expected across a broad spectrum of insulin
sensitivity and resistance. By contrast, the G/I ratio and 1/insulin
only behave appropriately in subjects with normal fasting glucose.
Indeed, G/I is functionally equivalent to 1/insulin under these
conditions, and there is no advantage to using G/I instead of
1/insulin. Because the fasting G/I ratio as an index of insulin
sensitivity is conceptually inappropriate, the use of 1/insulin in
nondiabetic subjects would be preferable. Moreover, it has previously
been shown that QUICKI is a superior index of insulin sensitivity
relative to the minimal model index or HOMA (15, 28).
|
Acknowledgments
Footnotes
Abbreviations: FSIVGTT, Frequently sampled iv glucose tolerance test; G/I, glucose to insulin ratio; HOMA, homeostasis model assessment; QUICKI, quantitative insulin-sensitivity check index.
Received April 25, 2001.
Accepted June 6, 2001.
References