Getting to the HAART of insulin resistance

There is now abundant evidence that HIV protease inhibitors (PI) as a class make a powerful independent contribution to the ‘metabolic syndrome’ that accompanies the long-term use of highly active antiretroviral therapy (HAART) regimens. Clinical studies have demonstrated that PI therapy is strongly associated with the development of dyslipidemia and insulin resistance [1–6], and that short-term use of PI is sufficient to cause these abnormalities even in the absence of body composition changes associated with the ‘lipodystrophy syndrome’[7]. Two published studies involving HIV-seronegative individuals have also shown that HIV infection is not a prerequisite condition for the effects of PI on lipid and glucose metabolism [8,9]. However, there are a number of critical questions that remain unresolved at this time. Firstly, what is the relationship between PI therapy and the three components of the ‘metabolic syndrome’ – dyslipidemia, insulin resistance, and visceral adiposity – and which of these interrelated outcomes is the primary pathogenic event? Secondly, how do PI, or the metabolic syndrome they cause, interact with nucleoside analogue reverse transcriptase inhibitors (NRTI) in the pathogenesis of subcutaneous fat wasting in individuals with lipodystrophy? The answers to these questions are fundamental to an understanding of the pathogenesis of the HIV lipodystrophy syndrome. In this setting, the study presented by Kosmiski et al. in this issue of AIDS[10] provides some welcome insights. `Insulin resistance’ refers to an impaired ability of insulin-sensitive tissues (such as liver, muscle, and subcutaneous adipose tissue) to respond appropriately to increased levels of insulin in the post-prandial phase. Insulin coordinates post-prandial metabolism, stimulating the uptake and utilization of carbohydrates as metabolic fuel, while directing fats towards uptake and storage, for later use in the ‘fasted’ state. Release of fatty acids from adipose stores (lipolysis) is inhibited by insulin, whereas fatty acid and triglyceride synthesis is stimulated. Hence, insulin resistance is associated with defects in post-prandial metabolism in which there is impaired uptake of fatty acids by adipocytes and other sites of fat storage, and inappropriate release of fatty acids in the fed state. As shown in Fig. 1, insulin resistance is associated with a characteristic dyslipidemia in which triglyceride-rich lipoproteins are abundant [11,12], a phenotype that is also seen in the familial partial lipodystrophy syndrome [13] as well as in HIV lipodystrophy [1–6,14,15].Fig. 1.: (a) The lipid metabolism response to insulin. Metabolism is directed towards fat storage in adipose tissue in the post-prandial phase, and inhibition of fatty acid release. Upregulation of lipoprotein lipase, and inhibition of lipolysis (via hormone-sensitive lipase) is accompanied by decreased hepatic very low density lipoprotein (VLDL)-cholesterol production. (b) Insulin resistance. In this situation, inappropriate fatty acid release and decreased lipoprotein lipase activity, as well as increased hepatic VLDL production, increase levels of triglyceride-enriched lipoproteins (chylomicrons and VLDL-cholesterol), and alter post-hepatic metabolism so that ‘small-dense’ Low density lipoprotein (LDL)-cholesterol is formed, and high density lipoprotein (HDL)-cholesterol is decreased. FFA, free (non-esterified) fatty acid; TG, triglyceride/triacylglycerol; LPL, lipoprotein lipase; IDL, intermediate density lipoprotein; CETP, cholesteryl ester transfer protein; LCAT, lecithin:cholesterol acyltransferase.Establishing the temporal and causal relationships between insulin resistance and dyslipidemia has been hampered, however, by the fact that while dyslipidemia is readily assessed in the clinical setting, fasting insulin and glucose levels – and calculations of insulin resistance using these values in the HOMA model – provide only an imprecise estimate of insulin resistance [16]. More accurate techniques such as the hyperinsulinemic euglycemic clamp [9] and intravenous glucose tolerance test with frequent sampling of plasma glucose and insulin [17] have been used in small studies, in which insulin resistance has been demonstrated following short-term indinavir therapy (< 8 weeks) in both HIV-infected [17] and seronegative [9] individuals. Now, Kosmiski and colleagues have assessed insulin resistance using an insulin-modified frequently sampled intravenous glucose tolerance test in 32 individuals on various HAART regimens, and have demonstrated marked insulin resistance in PI-treated individuals with increased visceral adiposity as a component of the lipodystrophy syndrome [10]. The cross-sectional nature of the study does not allow for an exploration of the temporal relationships between insulin resistance and visceral fat accumulation, but as the authors state, evidence from other studies in which insulin resistance was rapidly induced by PI therapy [9,17] supports other research [18,19] proposing that insulin resistance precedes, and contributes to, visceral adiposity. It is clear that insulin resistance and post-prandial lipid/fatty acid metabolism are intimately linked [11,12] (see Fig. 1), so that a primary defect in insulin signaling will inevitably lead to dyslipidemia. It must be borne in mind, however, that an alternative hypothesis also exists in which this temporal association is reversed, and a defect in post-prandial lipid metabolism leads to insulin resistance (as would be predicted if lipoprotein-receptor-related protein [20], or the acylation-stimulating protein pathway [21], are involved in pathogenesis). Hence, while the findings of Kosmiski et al.[10] are consistent with a body of in vitro research indicating that PI can directly impair insulin signaling in insulin-responsive tissues at pharmacologic doses [22–26], providing a plausible basis for the hypothesis that insulin resistance is the primary event in the PI-associated metabolic syndrome, further elucidation of these mechanisms requires a careful dissection of these metabolic pathways in vivo. It is also possible that the primary defect induced by select PI may be different. For example, there is evidence for early and pronounced insulin resistance in the presence of indinavir, both in vivo[9,17] and in vitro[23,25,26], while the induction of significant increases in triglyceride-rich lipoproteins after 2 weeks’ exposure to ritonavir therapy in HIV-seronegative individuals [27] is more suggestive of defective post-prandial lipid metabolism. Insulin resistance was not assessed in this study. Irrespective of whether insulin resistance or defective post-prandial lipid metabolism is the initiating event, a central physiological feature of this phenotype is increased turnover of fatty acids, and inappropriately high levels of systemic fatty acids in the post-prandial phase [28], as demonstrated in studies of dynamic in vivo lipid metabolism in the presence of PI-containing HAART [14,15]. The finding that resting energy expenditure (REE) is strongly correlated with insulin resistance in the study by Kosmiski et al.[10] is of great interest in this context. A similar association between HAART and increased REE has been described previously by Shevitz et al.[29], who observed that increased REE did not correlate with markers of virological or immunological status in the presence of HAART. The probable basis for this elevated resting metabolic rate, which as the authors point out is a feature of other insulin resistant states including congenital lipodystrophy syndromes [10], is partial ‘uncoupling’ of mitochondrial respiration induced by elevated tissue levels of long chain fatty acids. This effect of fatty acids is well characterized, and involves multiple members of a family of inner mitochondrial membrane anion transporters [30] (Fig. 2). (It is interesting to observe in passing that the latest addition to this family is a deoxynucleotide transporter that is likely to be a significant determinant of NRTI transport into the mitochondrial matrix [31]). While the uncoupling of substrate utilization from ATP production in mitochondria is usually conceptualized as a mechanism for generating heat in thermogenic brown adipose tissue, it is likely that in adult humans these processes are used to decrease mitochondrial efficiency in a controlled way in response to the presence of increased fatty acid delivery. In this way, the generation of harmful reactive oxidative species is minimized [32], mitochondrial metabolic processes that may otherwise become inhibited by excessive substrate availability and/or ATP production are maintained [33], and fatty acids are removed from the mitochondrial matrix where they may be deleterious [33,34]. Interestingly, the UCP2-UCP3 gene is a candidate locus in a region that is strongly linked to resting metabolic rate in humans [35].Fig. 2.: Fatty acid-induced partial uncoupling of mitochondrial respiration. The formation of ATP (`metabolic energy’) from metabolic fuels such as fatty acids and glucose (not shown), is dependent on the creation of a proton gradient across the inner mitochondrial membrane by Complexes I, III and IV. This is then used by Complex V (ATP synthase) to drive the conversion of ADP to ATP in the mitochondrial matrix. Protonated fatty acids (see inset, denoted ‘COOH’) are able to diffuse through the inner membrane, and donate a proton before being transported back to the cytosol by one of a family of closely related mitochondrial membrane anion transporters. This can be performed by ‘uncoupling proteins’ (UCP-1, UCP-2, UCP-3), as well as the adenine nucleotide translocase (ANT), dicarboxylate carrier, and the aspartate/glutatmate antiporter. The presence of excess H+ in the matrix dissipates the inner membrane proton gradient, decreasing the efficiency of mitochondrial respiration (i.e., more substrate is required to create each ATP). Note also that fatty acids do not normally enter the mitochondrion, and must be converted to acyl-CoA derivatives before being transported into the matrix by a carnitine transporter (CPT-1).The findings of Kosmiski et al.[10] may provide a new way of looking at the interaction between NRTI and PI in the pathogenesis of subcutaneous fat wasting. There is increasing evidence from longitudinal clinical studies that the development of fat wasting is greatly enhanced when NRTI and PI are used in combination, compared to when either class is used alone [36,37], suggesting that the interaction between these drug classes is the most powerful determinant of this outcome. The role of NRTI in the pathogenesis of fat wasting is beyond the scope of this article, and has been reviewed elsewhere [38]. However, given that mitochondrial dysfunction appears to play a prominent role in long-term toxicities associated with NRTI use [39], the in vivo demonstration that PI-associated insulin resistance and increased fatty acid flux may substantially affect mitochondrial function by decreasing the efficiency of oxidative metabolism provides a possible mechanism to explain the overlapping toxicities of NRTI and PI. These findings move us one step closer to elucidating the proximal pathogenic mechanisms of the metabolic syndrome induced by the PI class, which in turn will clarify the relationship between NRTI and PI in the pathogenesis of the lipodystrophy syndrome as a whole. As suggested by the authors, meaningful comparisons of different PI on lipid and glucose/insulin metabolism, and their subsequent effects on body composition changes, will be provided by longitudinal in vivo studies that examine these outcomes and their temporal relationships to the introduction (or cessation) of PI therapy, or to interventions (e.g., PPAR-γ agonists). This study also highlights the importance of utilizing appropriate diagnostic tools in the assessment of insulin resistance, adipose tissue distribution (including objective measures of the visceral fat depot), and the particular dyslipidemic phenotype that accompanies insulin resistance and defective post-prandial lipid metabolism.