The widespread use of antiretroviral combination regimens including HIV protease inhibitors (PI) has led to reduced morbidity and mortality in HIV-infected patients with advanced disease [1,2]. However, the early enthusiasm of clinicians about the rapid advance and virological response shown in clinical trials with highly active antiretroviral therapy (HAART) has recently been tempered. On one hand, the clinical manifestation of previously silent infections and autoimmune diseases caused by immune reconstitution has occurred, and on the other hand a syndrome of peripheral lipodystrophy, central adiposity, dyslipidaemia, and insulin resistance have been observed in patients treated with HIV PI [3-8]. Some patients developed hyperglycaemia and type 2 diabetes , and recent reports provided evidence for vascular complications after the initiation of HAART [10-12]. Carr et al.  hypothesized that PI treatment may lead to unspecific interactions with two proteins (cytoplasmatic-acid binding protein type 1 and low-density lipoprotein-related protein) that regulate lipid metabolism. The resulting inhibition is proposed to cause hyperlipidaemia, to contribute to central fat deposition and insulin resistance. In a recent report by Walli et al.  the association of peripheral insulin resistance and impaired glucose tolerance resulting from HAART has been demonstrated, but a detailed characterization of hyperlipidaemias as well as an assessment of the beta-cell function have not been provided. The occurrence of hyperlipidaemia, maturity onset diabetes and lipodystrophy with the consequences especially of atherosclerosis sheds new light on the use of PI. Interestingly, the HAART-associated lipodystrophy has striking similarities with the metabolic syndrome, called syndrome X. The understanding of the metabolic syndrome may thus have impact on the pathogenesis of lipodystrophy.
At present, the general consensus appears to be that patients with severe HIV disease should continue PI therapy where possible. Lipid-lowering therapy with statins and fibrates has been demonstrated to be successful , but the safety and efficacy of these drugs must be assessed in further studies. Statins are anti-inflammatory, primarily metabolized through cytochrome 3A4, and may lead to unfavourable drug interactions with PI. The aim of this study was to determine in detail the metabolic abnormalities and alterations of glucose metabolism and to evaluate possible vascular risk factors in HIV patients treated with HAART.
Thirty-eight (32 male, six female) HIV-1-infected patients receiving at least one HIV-1 PI were included and compared with 17 PI-naive HIV patients (15 men, two women) in an oral glucose tolerance test (OGTT). Co-administered antiretroviral therapy consisting of reverse transcriptase inhibitors (RTI) and/or non-nucleoside reverse transcriptase inhibitors (NNRTI) is listed in Table 1. Venous blood was taken after an overnight fasting for measurements of serum glucose, insulin, proinsulin, C-peptide, lipids, fibrinogen, homocysteine, and anticardiolipin antibodies (aCL). Plasma glucose, insulin, proinsulin, and C-peptide concentrations were determined before and 30, 60, 120, and 180min after oral administration of 75g glucose (Dextro-OGT; Boehringer, Mannheim, Germany). Blood glucose concentrations were measured by the hexokinase-glucose-6-phosphate-dehydrogenase method. Plasma concentrations of insulin and C-peptide were determined using a commercially available radioimmunoassay. For the measurement of proinsulin by microplate immunoradiometric assay, wells were coated with monoclonal mouse anti-human insulin/proinsulin antibody 3B1 (Biotrend, Köln, Germany) and incubated with human serum. Tubes were washed and incubated with monoclonal mouse anti-human C-peptide/proinsulin antibody PEP 01 (Novo Nordisk, Mainz, Germany) labelled with 125I. Recombinant human proinsulin was used as standard. IgG/IgM aCL were determined by a commercially available test (Pharmacia and Upjohn, Freiburg, Germany), homocysteine and fibrinogen levels were determined using commercial standard assays. CD4 cell counts were calculated by flow cytometry and HIV-RNA copies obtained by quantitative polymerase chain reaction (Amplicor HIV-1 Monitor Test; Roche Diagnostic Systems, Branchburg, USA).
Apolipoproteins [apoA-I, apo-AII, apoB, apoE and lipoprotein (a)] were analysed by nephelometric quantification using rabbit polyclonal antisera. The concentrations of lipid electrophoresis of very-low-density lipoprotein (VLDL), low-density lipoprotein (LDL), and high-density lipoprotein (HDL) cholesterol were determined using an electrophoretic and densitometric method (REP Lipoprotein-Kin; Helena Diagnostics, Germany). For the triglyceride assay the Peridochrom Triglyceride GPO-PAP Kit was used (Boehringer, Mannheim, Germany). For the total cholesterol assay the CHOD-PAP Kit (Boehringer) was used.
The results are presented as means ±SD (range). The statistical significance for lipid profiles and OGTT was determined by analysis using the non-parametric Mann-Whitney-U test. The c2 (Pearson) test was used for evaluating lipid differences in proportions between the groups. The areas under the curve (AUC) for the OGTT parameter were calculated according to the mathematic model of Thai . P values of <0.05 were considered to indicate statistical significance.
Characterization of glucose metabolism
Patients without PI had significantly higher CD4 cell counts, and a lower but not significant decrease in viral load (mean 0.41±1.69 log10 HIV-RNA copies/ml) since the initiation of antiretroviral therapy (n=10) compared with the PI-treated patients frequently on second and third-line therapy (mean decrease 1.36±1.32 log10 HIV-RNA copies/ml). A mean increase of CD4 cell count/μl since therapy was 84±197 in the PI-naive group compared with 122±29 in the PI-treated group. The other clinical characteristics are listed in Table 1. Eighteen (46%) patients on PI had impaired oral glucose tolerance with 2h post-load serum glucose levels between 7.8 and 11.1mmol/l, and five (13%) had diabetes according to 1985 WHO guidelines with 2h post-load serum glucose levels above 11.1mmol/l. Although four (24%) of the PI-naive patients were glucose intolerant, none had diabetes. Using the 1997 American Diabetes Association (ADA) fasting criteria , three patients on PI were glucose intolerant and two had diabetes, whereas all PI-naive patients had normal fasting glucose concentrations. Thirteen (45%) out of all hyperlipidaemic patients had impaired glucose tolerance according to the WHO criteria, although eight of the 26 (31%) patients with normal serum lipid levels revealed impaired glucose tolerance and two normolipidaemic patients demonstrated diabetes with over 11.1mmol/l 2h post-load.
Insulin response was delayed and increased in patients on PI with a maximum mean plasma insulin concentration of 784.5±693.6 (range 36.2-3294.7)pmol/l 120min post-glucose load in comparison with PI-naive patients with a maximum plasma insulin of 540.9±428.6 (range 65.2-1484.4)pmol/l already 60min post-glucose load (Fig. 1B). After the ingestion of glucose, C-peptide levels increased in PI-naive patients to mean peak concentrations of 6.6±3.0 (range 1.7-13.9)ng/ml already at 60min and returned to 3.9±2.0 (range 1.7-9.2)ng/ml at 180min. However, the PI-treated subjects presented with significantly higher mean fasting mean C-peptides of 3.4±2.2ng/ml (range 1.1-11.1, P<0.003), increasing up to higher and delayed peak concentrations of 10.7±5.0 (range 4.7-23.6)ng/ml at 120min (P<0.001), and continued on a markedly elevated level with 8.4 ±4.8 (range 2.4-24.9)ng/ml at 180min post-glucose challenge (P<0.0002) (Fig. 1C). In addition, the calculation of the AUC of the C-peptide curve revealed significant (P=0.001) higher total C-peptide release in the PI-treated group (Table 2).
The analysis of proinsulin concentrations revealed higher mean fasting values of 26.4±37.5 (range 2-167)pmol/l (P=0.006), and at each time point a significantly higher concentration with a peak of 170.0±144.0 (range 15-608)pmol/l in the PI-treated group compared with the mean maximum concentration of 78.8±66.8 (range 2-260)pmol/l in PI-naive patients at 120min (P=0.005) (Fig. 1A). Total pro-insulin secretion during OGTT, calculated as AUC, was also significantly higher (P=0.01) in the PI group (Table 2). Analysis of the serum glucose concentration revealed increased mean fasting glucose (5.5±1.7mmo/l, range 3.9-13.4) in PI-treated patients compared with 4.8±0.6 (range 3.3-5.5)mmol/l in PI-naive patients, but differences were not statistically significant. After the ingestion of the glucose, PI-treated patients reached higher maximum serum glucose levels and remained on high levels in the second phase at 120 and 180min (P=0.006 and P=0.002, respectively) (Fig. 1D). The AUC of glucose revealed higher total glucose levels during the OGTT in patients receiving PI (P=0.008) (Table 2).
More importantly, the hyperlipidaemic patients in the PI-treated group had more than three times the fasting proinsulin concentrations compared with normolipidaemic PI patients (34.9±43.4, range 5-167)pmol/l (P=0.008). Mean fasting insulin and particular 180min insulin concentrations were also markedly increased in hyperlipidaemic patients (136.0±136.3, range 21.7-687.9mmol/l and 611.0±514.5, range 29.0-1737.8mmol/l) in comparison with normolipidaemic patients with PI (55.3±15.6, range 36.2-86.9mmol/l and 182.3±141.6, range 43.5-521.4mmol/l; P<0.05 and P<0.005).
To evaluate the relative secretion response of proinsulin and insulin to glucose challenge, the proinsulin-to-insulin ratio during OGTT was determined. PI-naive patients showed a clear decrease of the proinsulin-to-insulin ratio (P=0.01) in the initial phase, indicating the dominating insulin release in comparison with proinsulin secretion. In contrast, the proinsulin-to-insulin ratio was only slightly decreased during the first phase of the OGTT in patients on PI, revealing early proinsulin and insulin secretion (Fig. 2). Re-analysis of the OGTT parameters and proinsulin-to-insulin ratio by removing the five diabetic subjects of the PI group still revealed statistically significant differences except for the 30min proinsulin level (data not shown). As the course of the C-peptide-to-insulin ratio during OGTT was not different in the PI-treated compared with the PI-naive patients, hepatic insulin clearance seems not to be affected in the PI group.
Characterization of lipid profiles
The analysis of the lipid pattern revealed that 27 (71%) of the PI-treated group had detectable hyperlipidaemia defined as fasting cholesterol >200mg/dl and/or triglycerides >200mg/dl. The fasting mean lipid parameters of both patient groups are demonstrated in Table 2. We also determined the type of hyperlipoproteinaemia using the Frederickson classification. Twelve patients (44%) presented with isolated hypertriglyceridaemia, two (7%) of them had type V, and 10 (37%) subjects had type IV hyperlipidaemia. Type IIb hyperlipidaemia, defined as an increase of both VLDL and LDL cholesterol, was found in 10 (36%) subjects and in addition, five (18%) patients presented with isolated hypercholesterolaemia thus classified as type IIa.
HAART was associated with significantly higher fasting cholesterol, triglycerides, LDL and VLDL levels (Table 2). Only four (24%) patients without PI presented with hyperlipidaemia, one with type IIb, one with type IIa and two with isolated hypertriglyceridaemia. These results were confirmed by characterization of the dyslipidaemias and assessment of apolipoprotein concentration (Table 2). An increased proportion of patients were found with elevated mean fasting serum values of apoB and apoE in the PI treatment group compared with PI-naive patients (P<0.005) (Table 3). Interestingly, an elevated concentration of lipoprotein (a) (>30mg/dl) as an important atherogenic risk factor was detected in six (16%) of the hyperlipidaemic patients, five of whom were treated with indinavir.
Assessment of other atherogenic risk factors
Fibrinogen was increased (>3.5g/l) in nine out of 26 patients; homocysteine was normal in all 27 and antinuclear antibodies (ANA) and anticardiolipin antibodies (aCL) was negative in all 23 tested patients receiving PI.
The aim of the study was to characterize the lipid profile and glucose metabolism with respect to pancreatic secretion in PI-treated patients and to elucidate possible risk factors for cardiovascular disease in this group. Because rare cases of abnormal fat deposition have also been described in patients receiving only RTI , the clinical body alteration may not be a direct side effect of PI at all. Elevated serum cortisol levels were suggested as a possible underlying mechanism in fat accumulation and metabolic abnormalities, but normal 24h urine cortisol levels in 25 of the PI-treated patients refute this explanation (data not shown). Moreover, excellent viral suppression under HAART regimens (below the limit of detection) has been proposed as a likely mechanism for lipodystrophy, but no correlation to absolute viral load nor to the decline of mean HIV-RNA levels in the PI-treated group were found. Recent data support a major role for PI in the pathogenesis of metabolic abnormalities since the replacement of a PI by an NNRTI (nevirapine) resulted in a partial improvement of serum lipids and fat distribution without a relapse of viral suppression or a decline in CD4 lymphocyte counts [17,18]. New-onset diabetes mellitus has been described with several PI commonly used to manage HIV disease . Walli et al.  proposed a stepwise model of a HAART-induced decrease in insulin sensitivity leading first to impaired glucose tolerance and finally in some cases to the manifestation of diabetes mellitus. The data demonstrate that patients on PI had a higher and prolonged output of insulin during the OGTT with delayed peak concentrations in the second phase of the test. In contrast, PI-naive patients respond with rapid insulin release in the first phase of OGTT after glucose ingestion.
More interestingly, proinsulin levels in the PI-treated group were significantly increased in comparison with untreated patients. The proinsulin-to-insulin ratio in this group was found to be relatively consistent during the first phase of the OGTT, suggesting the early secretion of immature proinsulin-rich granula. Increased secretion of proinsulin has been suggested as an indicator of beta-cell dysfunction, either in absolute terms in relation to insulin [19,20]. Hyperproinsulinaemia may have several explanations: as a secondary reaction to external factors, beta-cell distress, the increase of beta-cell secretory demand due to insulin resistance, the impaired maturation of proinsulin to insulin or compromised insulin secretion. Hyperglycaemia in individuals with impaired glucose tolerance and NIDDM might be associated with an increase of the proportion of proinsulin relative to insulin. It has also been suggested that beta-cell dysfunction rather than insulin resistance plays an important role in the future development of NIDDM in Caucasians with impaired glucose tolerance . These data partly support the two-step model for the development of diabetes proposed by Saad et al. . In the first step, the transition from normal to impaired glucose tolerance would depend mainly on the presence of insulin resistance. The second step, the worsening from impaired glucose resistance to diabetes, although accompanied by some further worsening of insulin resistance, is assumed to be primarily dependent on the development of beta-cell dysfunction. To support this hypothesis further, more direct measurements of beta-cell function need to be performed. It has recently been demonstrated that in elderly prediabetic individuals increased proinsulin concentration as an indicator of defective insulin secretion is associated with conversion to diabetes over a short time period [23,24]. The processing of proinsulin to insulin in beta cells is catalysed by proprotein convertases PC2 and PC3. It may be possible that PI for HIV-1 interact with enzymes required for the processing of proinsulin to insulin, resulting in an accumulation of proinsulin, and/or directly induce the secretion of immature proinsulin-rich granula of beta cells. Interestingly, in one HIV-1-infected patient, who developed diabetes while taking PI, there was no increase in insulin, C-peptide, and proinsulin secretion during OGTT (data not shown). After the discontinuation of PI, the patient resolved and became normoglycaemic within a few weeks, indicating a direct role of PI in the pathogenesis of HAART-induced diabetes.
These data confirm conclusively that a complex alteration in glucose and insulin metabolism exists in HIV patients on HAART. PI treatment was associated with a higher rate of diabetes mellitus, impaired glucose tolerance, hyperinsulinaemia, and early hypersecretion of proinsulin. This is important, because HIV infection itself was shown to cause an increase in insulin sensitivity of peripheral tissue and insulin clearance compared with control subjects . Significantly higher rates of the disproportional secretion of proinsulin and delayed insulin secretion were dominant in hyperlipidaemic HIV-patients receiving PI in comparison with normolipidaemic PI-treated patients. However, impaired glucose tolerance may exist without dyslipidaemia, although hyperproinsulinaemia is one of the main characteristics in patients with lipid abnormalities induced by HAART. The relationship between fasting hyperinsulinaemia and certain cardiovascular risk factors, such as blood pressure, obesity, raised triglycerides, and low concentrations of HDL cholesterol, has been demonstrated in both diabetic and non-diabetic individuals . It has been demonstrated that young, non- diabetic, male survivors of myocardial infarction are truly hyperinsulinaemic during an OGTT, and there is evidence for a close association between impaired glucose tolerance and proinsulin and coronary atherosclerosis [27,28].
In this context, the assessment of lipid metabolism in HIV-1-infected patients under HAART is of particular importance. Hypertriglyceridaemia and low plasma HDL cholesterol concentrations are the most common lipid abnormalities associated with both NIDDM and insulin resistance syndrome. Similarly, the treatment of HIV-1 infection with PI was associated with striking abnormalities in the lipid profiles compared with PI-naive HIV-1-infected patients. Significant higher fasting total cholesterol and triglyceride levels, and in a more comprehensive analysis higher LDL cholesterol and VLDL cholesterol levels, were characteristic for the patients receiving PI. Interestingly, hypertriglyceridaemia, especially in the late phase of the disease and to a lesser extent, has been seen in HIV-infected individuals since the early 1980s. The major causes of hypertriglyceridaemia were shown to be elevated rates of de-novo lipogenesis and the delayed clearance of triglycerides in the postprandial period . HAART had no significant influence on HDL cholesterol concentrations, and low plasma contents of HDL are already known to occur early in the disease . According to the described characteristic lipid profile associated with PI, evaluation of the apolipoproteins confirmed these results. HAART is thus accompanied by lipid abnormalities, which have been proved to be a significant risk factor for coronary heart disease, such as high total cholesterol and LDL cholesterol and low HDL cholesterol levels in a significant proportion of patients receiving PI. This is supported by a high percentage (60-70%) of HIV patients with elevated serum lipids and increased levels of lipoprotein (a) (>30mg/dl). In addition to other cardiovascular abnormalities common in HIV infection , this may contribute to severe cardiovascular events. Despite the fact that PI-naive patients partly received antiretroviral treatment (RTI, NNRTI), they demonstrate significant differences in glucose and lipid metabolism in comparison to patients on PI. Therefore we suggest that these effects are caused by PI, as supported by other studies [5,6,17,18]. Consequently, we recommend a comprehensive analysis of lipid profile and an OGTT before starting patients on PI treatment along with close monitoring during therapy. Furthermore, evaluation of current trials comparing PI-sparing drug regimens with treatment combinations including PI (e.g. the ATLANTIC Study ) will provide further information on the clinical relevance of the described metabolic abnormalities.
1. Mocroft A, Vella S, Benfield TL, et al
. Changing patterns of mortality across Europe in patients infected with HIV-1.Lancet
2. Palella FJ, Delaney KM, Moorman AC, et al
. Declining morbiditiy and mortality among patients with advanced human immunodeficiency virus infection.N EnglJ Med
3. Race EM, Adelson-Mitty J, Kriegel GR, et al
. Focal mycobacterial lymphadenitis following initiation of protease-inhibitor therapy in patients with advanced HIV-1 disease.Lancet
4. Behrens G, Knuth C, Schedel I, Mendila M, Schmidt RE. Highly active antiretroviral therapy.Lancet
5. Carr A, Samaras K, Burton S, Law M, Freund J, Chisholm DJ, Cooper DA. A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors.AIDS
6. Walli R, Herford O, Michl GM, et al
. Treatment with protease inhibitors associated with peripheral insulin resistance and impaired glucose tolerance in HIV-1 infected patients.AIDS
7. Lo JC, Mulligan K, Tai VW, Algren H, Schambelan M. ‚Buffalo hump‚ in men with HIV infection.Lancet
8. Miller KD, Jones E, Yanovski JA, Shankar R, Feuerstein I, Fallon J. Visceral abdominal-fat accumulation associated with use of indinavir.Lancet
9. Dube MP, Johnson DL, Currier JS, Leedom JM. Protease inhibitor-associated hyperglycemia.Lancet
10. Henry K, Melroe H, Huebsch J, et al
. Severe premature coronary artery disease with protease inhibitors.Lancet
11. Behrens G, Schmidt H, Meyer D, Stoll M, Schmidt RE. Vascular complications associated with the use of HIV protease inhibitors.Lancet
12. SoRelle R. Vascular and lipid syndromes in selected HIV-infected patients.Circulation
13. Carr A, Samaras K, Chisholm DJ, Cooper DA. Pathogenesis of HIV-1 protease inhibitor-associated peripheral lipodystrophy, hyperlipidaemia, and insulin resistance.Lancet
14. Henry K, Melroe H, Huebsch J, Hermundson J, Simpson J. Atorvastatin and gemfibrazol for protease-inhibitor-related lipid abnormalities.Lancet
15. Thai MM. A mathematic model for the determination of total area under glucose tolerance and other metabolic curves.Diabetes Care
16. American Diabetes
Association. Report of the Expert Committee on the diagnosis and classification of diabetes mellitus.Diabetes Care
17. Martinez E, Lozano L, Conget I, Casamitjana R, Gatell JM. Reversion of lipodystrophy after switching HIV-1 protease inhibitors to nevirapine.Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, February 1999 [abstract 670].
18. Ruiz L, Bonjoch A, Paredes R, et al
. A multi-center, randomized, open-labeled, comparative trail of the clinical benefit of switching the protease inhibitor (PI) by Nevirapine (NVP) in HAART-experienced patients suffering lipodystrophy (LD).Sixth Conference on Retroviruses and Opportunistic Infections
. Chicago, February 1999 [abstract LB14].
19. Williams DR, Byrne C, Clark PM, et al
. Raised proinsulin concentration as early indicator of b cell dysfunction. BMJ
20. Roder ME, Porte D, Schwartz RS, Kahn SE. Disproportionately elevated proinsulin levels reflect the degree of impaired b cell secretory capacity in patients with noninsulin-dependent diabetes mellitus.J Endocrinol Metab
21. Nijpels G, Popp-Snyders C, Kostense PJ, Bouter LM, Heine RJ. Fasting proinsulin and 2-h post-load glucose levels predict the conversion to NIDDM in subjects with impaired glucose tolerance: the Hoorn Study.Diabetologica
22. Saad MF, Knowler WC, Pettitt DJ, Nelson RG, Charles MA, Bennet PH. A two-step model for the development of non-insulin-dependent diabetes. Am J Med
23. Heine RJ, Nijpels G, Mooy JM. New data on the rate of progression of impaired glucose tolerance to NIDDM and predicting factors.Diabetes Med
24. Mykkänen L, Haffner SM, Kuusisto J, Pyörälä K, Laakso M. Serum proinsulin levels are disproportionately increased in elderly prediabetic subjects.Diabetologica
25. Hommes MJT, Romijn JA, Endert E, Schattenkerk JKME, Sauerwein HP. Insulin sensitivity and insulin clearance in human immunodeficiency virus-infected men.Metabolism
26. Haffner SM, Fong D, Hazuda HP, Pugh JA, Patterson JK. Hyperinsulinemia, upper body adiposity, and cardiovascular risk factors in non-diabetes.Metabolism
27. Jarrett RJ. The cardiovascular risk associated with impaired glucose tolerance.Diabetic Med
28. Båvenholm P, Karpe F, Proudler A, Tornvall P, Crook D, Hamsten A. Association of insulin and insulin propeptides with atherogenic lipoprotein phenotypes.Metabolism
29. Grunfeld C, Pang M, Doerrler W, Shigenaga JK, Jensen P, Feingold KR. Lipids, lipoproteins, triglyceride clearance, and cytokines in human immunodeficiency virus infection and the acquired immunodeficiency syndrome. J Clin EndocrinolMetab
30. Lipshultz SE. Dilated cardiomyopathy in HIV-infected patients.N Engl J Med
31. Katlama C, Murphy R, Johnson V, et al.The Atlantic Study. A randomised open-label study comparing two protease inhibitors (PI)-sparing antiretroviral strategies versus a standard PI-containing regimen.
Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, February 1999 [abstract 647].