Effects of atazanavir/ritonavir and lopinavir/ritonavir on glucose uptake and insulin sensitivity

Abstract

We read with interest the article by Noor et al. [1] examining the differential effects on glucose metabolism of the combinations of atazanavir/ritonavir versus lopinavir/ritonavir. Their paper deals with an important issue, having both theoretical interest and clinical impact. We think that some comments concerning aspects of the experimental procedure, data analysis and interpretation of the results could be helpful to other readers. The euglycemic hyperinsulinemic clamp is the golden standard for the assessment of insulin sensitivity. This experimental approach requires that the glucose concentration is maintained at a constant level throughout the study [2]. However, in the study by Noor et al.[1], the glucose concentration increased remarkably (up to 100 mg/dl) during the initial portion of the clamp (their fig. 3a), a defect which may influence the clamp-derived estimate of insulin sensitivity. During the clamp, it is assumed that endogenously secreted insulin is inhibited and only peripherally administered insulin contributes to insulin elevation. If the glycemic increase triggered an endogenous insulin response, this likely influenced the degree of inhibition of endogenous glucose production because the acute elevation of the portal insulin level has been shown to have a long-lasting effect on the liver [3]. As a result, the clamp-derived estimate of insulin sensitivity would also be affected. Additional data on the C-peptide would be welcome to verify the time course of endogenous insulin secretion, as a bias cannot otherwise be ruled out. In analyzing the oral glucose tolerance test (OGTT), Noor et al.[1] used plasma glucose and insulin levels to calculate insulin resistance for each time point of the OGTT by means of the homeostasis model assessment (HOMA) (their fig. 3c). That, however, is a misuse of the HOMA method, which has been developed to provide an index of insulin resistance from the concentrations of glucose and insulin measured at the basal steady state [4]. This method cannot be applied to nonsteady-state conditions when insulin and glucose concentrations continuously change. Thus, the authors had no methodological support for calculating the HOMA index at each time point during the OGTT and performing statistical comparisons among groups. In discussing their data obtained in vivo, Noor et al.[1] faced the problem of reconciling a discrepancy between the euglycemic, hyperinsulinemic clamp and OGTT results. Although the clamp revealed a significantly different reduction in insulin sensitivity between atazanavir/ritonavir and lopinavir/ritonavir, the OGTT did not (their table 2). The authors claimed that this discrepancy was due to the fact that the OGTT reflects both liver and peripheral insulin sensitivity, whereas the glucose clamp only reflects the latter component. This argument is questionable for the following reasons. First, a patent contradiction immediately emerges. If the OGTT reflects both liver and peripheral insulin sensitivity, whereas the clamp only reflects the latter, the OGTT would be expected to be more capable than the clamp of detecting differences in insulin sensitivity among the study groups. This contradicts the experimental evidence, which shows the opposite. Second, the claim that the clamp only reflects peripheral insulin sensitivity has no grounds. The index of clamp-based insulin sensitivity used by Noor et al.[1] is the ratio M/I, where M is the exogenous glucose infusion rate and I is the insulin concentration, both measured at the end of the clamp. It is worth emphasizing that the glucose infused during the clamp allows a constant plasma glucose concentration to be maintained because it balances exactly the increase in glucose disappearance (Rd) and the reduction in endogenous glucose production (EGP). Thus, the M-value measures the ability of insulin to stimulate peripheral glucose uptake, as well as to inhibit glucose production (M = ΔRd − ΔEGP, where Δ is the change from pretest level) [5]. In the study by Noor et al.[1], the inhibition of endogenous glucose production could not be assumed to be identical in the study groups because protease-inhibitors are known to affect liver insulin sensitivity [6]. The contribution of glucose uptake can only be isolated from that of endogenous glucose production if a glucose tracer is concurrently infused along with unlabeled glucose [5], but Noor et al.[1] did not use a tracer. It can be concluded that the true reason for the discrepancy between the clamp and OGTT results remains to be elucidated. In conclusion, we hope that the above considerations may prove useful to clinicians in the interpretation of studies assessing the effects of antiretroviral drugs on carbohydrate metabolism using methods such as the euglycemic hyperinsulinemic clamp, the HOMA, or the OGTT.