This review underlines that the simultaneous assessment of both insulin secretion and sensitivity is necessary to reach appropriate conclusions. However, a great deal of attention must be placed on the choice of the experimental tests to be performed (27). The most important issue is that the two measurements must be as independent as possible. This may yield some problem when the disposition index has to be determined in large epidemiological studies. In this case, for instance, a widely used method to assess insulin sensitivity is HOMA (24), because it uses only the product of fasting insulin and glucose concentration. The corresponding index of beta cell function also uses basal insulin; thus, the two indices are strictly related being both proportional to the same basal insulin measurement. Their product therefore does not make much sense and should not be used. Other simple estimations of insulin sensitivity for epidemiological studies include the QUICKI calculation (25) still based on fasting glucose and insulin concentrations. These methods exhibited a good correlation with insulin sensitivity as determined by the euglycemic hyperinsulinemic clamp test, but their limitations should be recognized (27, 51).
To have a trustworthy measurement of insulin secretion, it is necessary also in large studies to use dynamic tests, such as the OGTT, which provide independent measurements of beta cell function, for example from the insulinogenic index (52) or from the ratio of the insulin area over the glucose area. For calculating the disposition index, these measurements can be used in conjunction with insulin sensitivity indices such as HOMA or other simple and probably more reliable methods. In particular, the method called OGIS (37) only needs three samples – fasting, 90 min and 2 h – and it has been used successfully for calculating the disposition index (53). The content of information from this simple test should be enough for large-scale studies, with a low cost/benefit ratio. However, the results must be interpreted having in mind all the assumptions and simplifications included in these methods.
In metabolic studies, when the number of subjects is not very high and the investigator needs a clear quantification of the parameters, possibly affected by the least error possible, dynamic tests, such as the euglycemic hyperinsulinemic clamp, the FSIGT, the arginine test or other tests, are mandatory. The burden for the clinical investigator and the discomfort of the subject under study increase, but the information gathered from these studies is quite large and reliable for the purpose of characterizing in detail the metabolic status of the single individual. Also in these cases attention should be focused on the simultaneous assessment of insulin secretion and sensitivity. For instance, the hyperglycemic glucose clamp yields a measurement of both processes, but insulin secretion, obtained from peripheral insulin concentration, may be intrinsically related to insulin sensitivity, as the same insulin concentration values are used also to assess insulin sensitivity. For medium size metabolic studies, measurement of beta cell function independently from the corresponding ones of insulin sensitivity, such as the first-phase insulin secretion indices from either the hyperglycemic clamp or the arginine test or the FSIGT are therefore advisable. In particular, in the insulin-modified FSIGT (29, 30), insulin secretion is calculated during the first 8–10 min (AIRG), while for the estimation of the parameter SI the delayed insulin action, i.e. after exogenous insulin administration from 20 min on, plays the major role. Plotting SI vs AIRG to obtain the disposition index is therefore correct. A methodological problem in metabolic studies in diabetic subjects is the confounding factor of hyperglycemia, which requires standardization and therefore in many cases more than one dynamic test (54).
For more precise measurements of insulin sensitivity in specific tissues, particular experimental protocols involving the use of sophisticated instrumentation must be carried out (among several: forearm arterovenous differences, nuclear magnetic resonance spectroscopy, positron emission tomography, multiple tracer dilution). For the assessment of pre-hepatic insulin release in the portal vein, measuring also C-peptide is necessary, while molecular biology studies allow the investigation of the fine mechanisms regulating insulin release directly from the beta cell. However, presentation and comments on these last techniques are beyond the aim of this review.
The same approach as in metabolic studies may be instituted for the clinician in an attempt to quantify disposition index in clinical practice. Here, however, an important issue is the reference values of the estimates, because both insulin sensitivity and insulin secretion are heavily influenced by age, ethnicity, body weight, fat distribution and other clinical conditions or treatment, and for accurate determination reference values would be required controlling for these confounders. Until such information is available, the use of a simple dynamic test, such as OGTT, might be sufficient.