HIV type 1 infection, and not short-term HAART, induces endothelial dysfunction
AIDS:Volume 23(5)13 March 2009p 589-596
Francisci, Danielaa; Giannini, Silviab; Baldelli, Francoa; Leone, Mariob; Belfiori, Barbaraa; Guglielmini, Giuseppeb; Malincarne, Lisaa; Gresele, Paolob
aDivision of Infectious Diseases, Department of Experimental Medicine and Biochemical Sciences, Italy
bDivision of Internal and Cardiovascular Medicine, Department of Internal
Medicine, University of Perugia, Perugia, Italy.
Objective: Ischemic cardiovascular events increasingly occur during long-lasting HIV infection and are attributed either to the infection itself or to the use of HAART. Endothelial dysfunction and platelet activation are markers of atherosclerosis. Our aim was to assess whether patients with chronic HIV infection present endothelial dysfunction and whether this is the consequence of infection or of HAART.
Design: Fifty-six HIV-infected patients were studied in a retrospective cohort study before and 3, 6, 12 and 24 months after starting HAART with protease inhibitors (n = 28) or nonnucleoside reverse transcriptase inhibitors (n = 28), and compared with 28 age-matched and sex-matched healthy controls, and with 10 naive HIV-infected patients studied at diagnosis and after 12 months of untreated infection.
Methods: Soluble endothelial and platelet activation markers were measured in plasma by flow cytometry.
Results: Soluble P-selectin, soluble vascular cell adhesion molecule-1, monocyte chemoattractant protein-1 and von Willebrand factor were significantly higher in HIV-infected patients than in healthy controls, whereas soluble CD40 ligand and tissue type plasminogen activator were within normal range. During follow-up, soluble vascular cell adhesion molecule-1, monocyte chemoattractant protein-1 and von Willebrand factor but not soluble P-selectin decreased progressively, without significant differences between protease inhibitors and nonnucleoside reverse transcriptase inhibitors treatment. In naive, untreated patients, increased plasma markers of endothelial dysfunction were confirmed at diagnosis, with no changes upon follow-up.
Conclusion: Chronic HIV infection, and not its pharmacological treatment, induces alterations of markers of endothelial function. Short-term treatment with HAART reduces some markers of endothelial dysfunction, with no differences between protease inhibitors and nonnucleoside reverse transcriptase inhibitors.
Since the introduction of HAART, in the mid-1990s, HIV infection has turned from a rapidly fatal disease into a chronic condition. Before the era of HAART, the estimated median survival for a young person infected with HIV was 7 years, whereas currently it reaches about 35 years . Consequently, complications involving the cardiovascular system, previously not manifested, have emerged.
Recent studies suggest that the incidence of coronary heart disease is at least three-fold increased in HIV-infected patients as compared with the general population  and that the relative risk of acute myocardial infarction is 1.75 in HIV-infected patients after adjustment for age, sex, race, hypertension, diabetes and dyslipidemia .
However, it is still debated whether cardiovascular complications are a consequence of the HIV infection itself or are due to the long-term use of HAART. The latter, in fact, is known to modify the lipid profile towards a proatherogenic pattern, increasing the levels of total serum cholesterol, triglycerides, low-density lipoprotein and very-low-density lipoprotein .
The use of protease inhibitors (PIs) was associated with an increased risk of myocardial infarction, with an adjusted relative rate of 1.16 per year of exposure . However, the effect of PIs on the lipid profile probably does not fully explain the increased risk of myocardial infarction .
On the contrary, a pathologic role of HIV infection itself and not of HAART is suggested by the observation that in the Strategies for Management of Anti-Retroviral Therapy (SMART) study, persons undergoing episodic antiretroviral therapy had an increased risk of cardiovascular events identical to that of patients undergoing continuous therapy [6,7].
Inflammation appears to be an important pathogenic event in the progression of atherosclerosis and in the rupture of the plaque's fibrous cap, which is usually the key event leading to acute ischemic accidents. Infection-induced chronic inflammation may thus contribute to the increased incidence of cardiovascular events in HIV-infected individuals. Endothelial dysfunction, an early marker of atherosclerosis with predictive value for cardiovascular events [8-10], could be the link between HIV infection and atherosclerosis. Indeed, an increased expression of adhesion molecules and inflammatory cytokines and, namely, of tumour necrosis factor-α (TNF-α), interleukin-6 (IL-6), vascular cell adhesion molecule-1 (VCAM-1), intercellular cell adhesion molecule-1 (ICAM-1), P-selectin (P-sel), monocyte chemoattractant protein-1 (MCP-1) and von Willebrand factor (VWF), all molecules involved in the recruitment and adhesion of leukocytes at sites of atheroma initiation, has been reported during chronic HIV infection [11-15].
Activated platelets may also play a role in HIV-induced atherosclerosis through the expression and release of mediators that, in turn, can activate the endothelium and support the adhesion of leukocytes to the inflamed vessel wall . Previous studies reported increased plasma levels of markers of platelet activation [s and soluble CD40 ligand (sCD40L)] and of procoagulant platelet-derived microparticles, of circulating activated platelets expressing membrane bound P-sel, CD63 and aminophospholipids, in HIV-infected patients and the effect of antiretroviral treatment on platelet activation was controversial [12,15].
Aim of our study was to evaluate markers of endothelial and platelet activation in HIV-infected patients as compared with healthy individuals, to assess whether these markers are affected by HAART therapy and to see if there is a difference between treatment with PIs or with nonnucleoside reverse transcriptase inhibitors (NNRTIs) for their effects on these markers.
HAART has reduced the risk of potentially fatal opportunistic infections and has significantly prolonged the average lifetime of HIV-infected people. As a consequence, long-term complications, and in particular cardiovascular events, have emerged as a clinically significant issue and have became the object of several studies. One unsolved question is whether HIV-infected patients have an increased cardiovascular risk as compared with uninfected people because the infection-associated chronic inflammation produces vascular changes [11,20] or because the antiretroviral agents produce a shift of the lipid profile to a proatherogenic pattern .
Previous findings from the Data Collection on Adverse Events of Anti-HIV Drugs (DAD) study indicate that the time of exposure to PIs is correlated with an increased risk of myocardial infarction, in part, due to treatment-induced dyslipidemia . On the contrary, several studies have demonstrated that the HIV infection itself is associated with an increased cardiovascular risk, probably as a consequence of the activation of inflammatory pathways. Indeed, other infectious agents too, bacterial or viral, have also been claimed to induce an inflammatory condition that contributes to atherogenesis (Helicobacter pylori, Chlamydia pneumoniae, cytomegalovirus and hepatitic C virus). Human atherosclerotic plaques, for instance, may show signs of infection by microbial agents such as C. pneumoniae . Moreover, HIV can directly activate endothelial cells, changing their phenotype from anti to prothrombotic ; in addition, the HIV-associated membrane glycoprotein gp120 can activate human arterial smooth muscle cells in vitro causing the release of procoagulant tissue factor .
In recent years a prominent role for inflammation in the pathogenesis of atherosclerosis has emerged and, based on this evidence, circulating inflammatory molecules, of endothelial or platelet origin, have been identified as markers of atherosclerotic risk .
The aim of our study was to investigate a series of markers of inflammation related to the development of atherosclerosis in HIV-infected patients and to evaluate the relative role of infection itself or of HAART in these changes.
Our results confirm that in HIV-infected patients signs of endothelial dysfunction are present. In fact, we have found that HIV-infected patients have increased circulating levels of sVCAM-1, MCP-1, sP-sel and VWF. sVCAM-1 and MCP-1 are two markers of endothelial activation [24,25], whereas sP-sel and VWF are markers of both endothelial and platelet activation . However, it was reported that, most of the circulating sP-sel in atherosclerotic animal models comes from the endothelium . On the contrary, we observed no signs of in-vivo platelet activation, in fact, no alterations in the plasmatic levels of sCD40L, a marker of in-vivo platelet activation , were evident. Moreover, also t-PA, a marker of activation of the fibrinolytic system , was normal. Higher t-PA plasma levels in HIV-infected patients with fat redistribution or in AIDS stage were reported [29,30], and a positive association between cholesterol and t-PA has been described . In our study, t-PA antigen levels were not raised either in naive or in treated patients, despite the increase of total cholesterol and triglycerides observed during HAART.
Our data also show that soluble markers of endothelial dysfunction were not enhanced, but on the contrary were reduced, by the short-term use of HAART, as shown by a significant decrease of sVCAM-1, MCP-1 and VWF levels starting from the sixth month of pharmacologic treatment. The role of HAART in the progressive decrease of endothelial dysfunction markers is demonstrated by the observation that HIV-infected patients with elevated levels of endothelial activation markers, who were not antiretroviral treated, did not show any decrease of these markers after 12 months of untreated infection.
Our study has some limitations. First, it is a retrospective study, thus raising the risk of selection bias. However, the selection of patients was randomly carried out from our database based on predefined selection criteria, thus excluding selection bias. Moreover, patients were followed up at several time intervals for up to 24 months, strengthening our observations and providing a longitudinal view on their endothelial function.. Second, samples were collected over an extended time period, thus raising the risk of possible variability due to stability problems with some of the markers analysed. However, we found that sVCAM-1, MCP-1 and VWF decreased during the 24 months observation period in treated patients and that lower values were found in the most recently collected samples, thus making unlikely any bias due to the instability of the stored samples. Moreover, the same parameters measured in untreated patients and collected over the same period showed stability of the values between baseline and 12 months of follow-up, thus supporting the hypothesis that the observed variation was due to HAART. Third, we have measured soluble markers of endothelial dysfunction, and not a direct functional parameter of endothelial function, like flow-mediated vasodilatation, and the abnormalities we observed might reflect HIV-associated inflammation rather than endothelial dysfunction. However, the analytes we evaluated are known to contribute to vascular inflammation and to the progression of the atherogenetic process and are almost invariably associated with impaired endothelium-mediated vasodilatation [9,10,23,32]. Moreover, a recently published study reports an improvement of endothelial function after antiretroviral therapy when measured by brachial artery flow-mediated dilatation (FMD). FMD increased significantly after 4 and 24 weeks of HAART with similar changes induced by three different HAART regimens (PI-sparing, NNRTI-sparing and nucleoside reverse transcriptase inhibitors-sparing), findings in perfect agreement with our results . Fourth, the sample size of the patient groups was relatively small. However, the statistical power of the test we applied ranged from 57.0 to 98.3%, making us quite confident with the results.
Previous studies have already reported increased levels of markers of endothelial activation in HIV-infected patients and also the decrease of some of these markers with HAART ; however, our study adds new information because of the longer follow-up upon HAART treatment (24 months), which can be a crucial aspect, given that prolonged and uncorrected lipid abnormalities induced by HAART have been suggested to be the cause of higher cardiovascular risk in the longer term [3-5,34]. Moreover, we simultaneously studied a group of untreated HIV-infected patients followed up for 12 months, which is an important control in the evaluation of the time-course of these atherosclerosis-related parameters. Another important observation from our study is that the effect of HAART on soluble markers of endothelial dysfunction was independent from the lipid profile changes; in fact, even if total cholesterol and triglycerides increased during PIs treatment, a decrease in the levels of sVCAM-1, MCP-1 and VWF occurred, with no correlation between the two changes. A lack of correlation between improvement of FMD and lipid profile changes was observed also in the study by Torriani et al. . On the contrary, a correlation was evident between HIV RNA and MCP-1, sVCAM-1 and VWF changes, further suggesting that the major role in the observed alterations is played by viral infection.
In conclusion, our study shows that chronic HIV infection itself, and not its pharmacologic treatment, induces an increase of soluble markers of endothelial dysfunction. A short-term treatment with HAART reduces some parameters of endothelial dysfunction, with no differences between PIs and NNRTIs.
Based on the present data, a prospective study to evaluate the long-term effects of different HAART regimens on endothelial activation appears to be warranted, given that a late, drug-related, increase of cardiovascular risk in the DAD study has been observed .
Demographic, epidemiological and immunovirological data
Demographic and epidemiological data for 56 HIV-infected patients at baseline are summarized in Table 1. No relevant differences were found between patients receiving PIs (93% lopinavir/ritonavir and 7% nelfinavir) and NNRTIs (86% efavirenz and 14% nevirapine). Immunovirological data and lipid profile at baseline and after 3, 6 and 12 months of treatment are summarized in Table 2. Patients treated with PIs, but not those treated with NNRTIs, showed a significant increase of total cholesterol and triglycerides after starting HAART. Baseline CD4+ cell count in HIV-infected patients was 206.6 ± 196.6 cells/μl without significant differences between NNRTIs-treated and PIs-treated patients (P = 0.16); no other relevant differences were found. The second group of 10 naive HIV-infected patients for which no HAART treatment was started, all in CDC A1 , had a CD4+ cell count of 630.2 ± 261.6 cells/μl and a significantly lower viral load than the 56 treated patients at baseline (3.87 ± 0.75 vs. 4.96 ± 0.52 log10 copies/ml plasma, P < 0.001). No other relevant differences were found between the two patient groups.
Markers of endothelial and platelet activation at baseline
Of the eight markers measured in plasma, IL-6 and IL-8 were unmeasurable in all the samples assessed because below the detection limit of the assay.. Levels of t-PA and sCD40L in HIV-infected patients (2099.6 ± 160.8 and 2201.8 ± 214.2 pg/ml, respectively) did not differ from the levels observed in the control group (1960.8 ± 212.7 and 2257.4 ± 292.7 pg/ml, respectively). On the contrary, plasmatic levels of sP-sel, sVCAM-1, MCP-1 and VWF were significantly higher in HIV-infected patients at baseline, that is before starting HAART, as compared with healthy controls (Fig. 1). Patients and controls were matched for smoking history, but not for hypertension. However, if hypertensive HIV-infected patients were excluded from the analyses, sP-sel, sVCAM, MCP-1 and VWF were still significantly increased as compared with controls, suggesting that hypertension was not the cause of the observed increase in the HIV-infected patients.
Effect of HAART on endothelial activation
During treatment with HAART, plasma sVCAM-1, MCP-1 and VWF levels decreased progressively. This reduction was statistically significant starting from the sixth month of treatment, without any significant difference between PIs and NNRTIs. On the contrary, sP-sel levels remained elevated during HAART (Fig. 2). Both sVCAM-1 and MCP-1 normalized, that is no longer differed between treated HIV-infected patients and healthy controls, after 24 months of HAART. On the contrary, VWF levels decreased during HAART but not to the levels of normal controls even after 24 months. In the second group of patients with HIV infection who did not necessitate antiretroviral therapy, plasma levels of sP-sel, sVCAM-1 and MCP-1 at diagnosis were also higher than in healthy controls, but they remained elevated after 12 months of infection, differently from that observed in HAART-treated patients (Fig. 3). VWF levels in the group of naive HIV-infected patients not requiring HAART were also higher as compared with healthy controls (183.1 ± 41.4 vs. 106.2 ± 8.5; one-way ANOVA: P = 0.2; Student's t-test: P < 0.05). However, differently from HAART-treated patients, in these patients VWF levels remained elevated after 12 months of untreated infection (Fig. 3). HIV-RNA significantly correlated with sVCAM-1 (P < 0.05, r = 0.16, n = 197), MCP-1 (P < 0.001, r = 0.26, n = 197) and VWF levels (P < 0.0001, r = 0.31, n = 195). No significant correlations were instead observed between either serum cholesterol or triglycerides and sVCAM-1, MCP-1 and VWF levels.
Patients and design of the study
We carried out a retrospective cohort study on HIV-infected outpatients followed at our centre between January 2002 and December 2005 and who were about to start a therapeutic regimen with at least one PI or an NNRTI.
During this period 131 patients started a HAART treatment, 58 with NNRTIs and 73 with PIs. We initially selected all those patients treated with NNRTIs of which cryopreserved plasma samples had been collected at regular intervals (3 ± 1, 6 ± 1, 12 ± 1 months) for at least 1 year, and who satisfied the following criteria: age 18 years at least, absence of opportunistic infections and other infectious diseases, no change of treatment over the first 12 months.
Of the 58 patients treated with NNRTIs, four had been lost to follow-up, 16 were excluded because they changed treatment due to virologic failure, toxicity or no adherence, six because they had opportunistic infections or other infectious diseases, four because plasma samples were not available, thus leaving 28 NNRTI-treated patients to be analysed for the study. Then, an identical number of age-matched and sex-matched patients treated with PIs were selected from the 73 followed in the same period at our centre, and who satisfied the same enrolment criteria.
Out of the 56 studied patients, in a subgroup of 20 patients (10 exposed to PIs and 10 to NNRTIs) samples collected 24 months after starting therapy were also analysed, in order to have a longer follow-up. Demographic and clinical information were collected by reviewing medical files.
A second control group of 10 HIV-1-infected, naive patients, all in the A1 disease stage according to the Centre for Disease Control and Prevention (CDC) classification who did not require antiretroviral therapy because of a CD4+ cell count more than 500 cells/μl, was selected and samples collected at diagnosis and after 12 months of follow-up were analysed.
In addition, a cross-sectional comparative study with healthy donors was carried out.
Plasma samples from 28 healthy, HIV-1-negative blood donors, matched for age, sex and smoking history, chosen among staff members of the clinic, collected in the same period, were used as controls. Informed consent was obtained from all patients and controls.
Peripheral whole blood from patients and healthy individuals was withdrawn in vacutainer tubes containing EDTA and centrifuged at 1000 g for 10 min.. Platelet-free plasma was separated and cryopreserved at -80°C for subsequent quantification of endothelial and platelet activation markers and of viral load.
Quantification of HIV-1-RNA copy numbers in plasma
Plasma viral load was determined by quantitative reverse PCR using Cobas Amplicor HIV-1 Monitor Test, version 1.5, Ultrasensitive (Roche Diagnostic, Indianapolis, Indiana, USA). The limit of detection was 50 copies/ml plasma.
Quantification of the CD4+ lymphocytes subset
CD4+ cell count was carried out in peripheral whole blood collected in EDTA by flow cytometry.
Quantification of von Willebrand factor
Soluble VWF antigen was quantified by ELISA (Hyphen BioMed, Neuville-sur-Oise, France).
Simultaneous quantification of endothelial and platelet activation markers by flow cytometry sCD40L, sP-sel, tissue type plasminogen activator (t-PA), sVCAM-1, IL-6, IL-8 and MCP-1 in plasma samples were measured by flow cytometry using the FlowCytomix human cardiovascular 7plex kit (Bender MedSystems, Vienna, Austria), that allows the simultaneous measurement of seven different parameters in the same plasma sample, according to the manufacturer's instructions . Briefly, we used two sets of beads of different size (4.4 and 5.5 μm) consisting of five and two bead populations internally labelled with different intensities of a fluorescent dye that emits at 690 nm when excited by an argon laser.
Seven different bead populations were detectable based on their size and on their fluorescence intensity. A mixture of the seven different bead populations, each coated with an antibody directed against one of the analytes to be detected, was incubated with the plasma samples or with several dilutions of a mixture of standards and with a biotin-conjugated secondary antibody. Subsequently, streptavidin-phycoerythrin was added and samples were finally analysed in a EPICS XL-MCL flow cytometer (Coulter Corporation, Miami, Florida, USA), as described . The instrument was equipped with an argon laser operating at 488 nm. The seven bead populations were differentiated by their size and their internal dye intensity, measured using a 675 nm band pass filter. The streptavidin-phycoerythrin fluorescence intensity of each bead population, detected using a 575 nm band pass filter, was directly related to the respective antigen concentration in the sample, thus allowing the simultaneous quantification of seven different analytes. Standard curves for each analyte were generated and data were analysed by the FlowCytomixPro 1.0 software. The detection limit of the assay was 62.0 ng/ml for sP-sel, 82.0 pg/ml for sCD40L, 12.0 ng/ml for sVCAM-1, 41.0 pg/ml for MCP-1, t-PA and IL-6 and 123.0 pg/ml for IL-8.
Results are expressed as mean ± SEM. Parameters showed a normal distribution. Differences between controls and patients were analysed by one-way ANOVA or in selected cases by the unpaired Student's t-test; differences between replicate measures over time in treated patients were assessed by using GLM-ANOVA followed by the Bonferroni's post hoc test. All statistical analyses were performed using the NCSS and PASS packages (Statistical System, Kaysville, Utah, USA). The statistical power of the tests applied ranged from 57.0 to 98.3%.
A P-value of less than 0.05 was considered as statistically significant. Correlation analyses were carried out with the Spearman's test.