| |
Endothelial Function [Impaired] in HIV-Infected Persons
|
| |
| |
Clinical Infectious Diseases May 1 2006;42:1325-1332
".....We conclude that HIV-infected persons have substantial impairment of endothelial vasomotor function and that this impairment is worse among persons with elevated levels of HIV replication, particularly IDUs. In the present study, impaired endothelial function was not associated with PI therapy; however, it was associated with a lower -HDL triglyceride level, which is a presumed marker of metabolic imbalance. Prospective studies are needed to determine whether effective antiretroviral treatment can lead to improved endothelial function in HIV-infected persons....
.... In multivariate analysis, independent predictors of impaired FMD were current injection drug use (FMD difference, 3.3%; 95% CI, 0.9%-5.7%; P = .007) and the a-HDL triglyceride level (FMD difference, 3.2%; 95% CI, 1.3%- 5.1%; P = .002). This model explained 26% of the variability in FMD among the HIV-infected patients. However, when current injection drug use was removed from the model, HIV load became a significant predictor of impaired FMD (FMD difference, 2.1%; 95% CI, 0.2%-4.0%; P = .04)....
....The present study showed a strong association between HIV-1 infection and endothelial dysfunction. This finding is consistent with several studies that have described the association of coronary artery disease with HIV infection. Joshi et al. have documented significant inflammation involving the coronary arteries of children infected with HIV. This inflammatory state is a well-known precursor of the cascade of events leading to arteriosclerosis.....The present study also showed a strong correlation between current injection drug use and impaired FMD among HIV-infected persons. Although multiple comparisons were performed, the strength of this association was such that it is not likely to be attributable to chance. Injection drug use (mainly cocaine use) is a well-known risk factor for coronary artery disease. Repetitive toxic effects of cocaine or other illicit substances could damage the endothelium....When injection drug use was removed from the multivariate model, HIV load became a significant predictor of FMD. This finding suggests that at least part of the effect of current drug use is mediated by the HIV load.....On the other hand, the present study did not substantiate the observation by Stein et al. [25] that PI-induced dyslipidemia and insulin resistance were associated with endothelial dysfunction........our analysis did not show any significant worsening of FMD values associated with lipid status or presence of insulin resistance, and patients receiving PI treatment displayed slightly better endothelial function than did patients who were not receiving PI therapy...."
Table 3. Multivariate analysis of risk factors for impaired endothelial function among 302 study subjects.
Authors: Anthony Solages,1 Joseph A. Vita,2 David J. Thornton,1 Jessica Murray,3 Timothy Heeren,3 Donald E. Craven,4,5 and C. Robert Horsburgh, Jr.1,4
Divisions of 1Infectious Diseases and 2Cardiology, Department of Medicine, Boston University School of Medicine, and Departments of 3Biostatistics and 4Epidemiology, Boston University School of Public Health, Boston, and 5Department of Infectious Diseases, Lahey Clinic, Burlington, Massachusetts
Potential conflicts of interest. All authors: no conflicts
ABSTRACT
Background. Several reports have suggested an increased risk of coronary disease in human immunodeficiency virus (HIV)-infected patients receiving protease inhibitors (PIs). Impaired endothelium-dependent vasodilation is a putative surrogate marker of coronary atherosclerotic disease.
Methods. The present study evaluated the effect of HIV infection and antiretroviral treatment on endothelial vasomotor function, by assessing brachial artery flow-mediated dilation (FMD). A total of 75 HIV-infected patients were compared with 223 control subjects who were presumed to be HIV uninfected.
Results.
HIV-infected patients had significantly impaired FMD, compared with control subjects (mean ± SD, 7.3% ± 4.4% vs. 11.1% ± 6.3%; P < .0001).
When adjustments were made for smoking status, sex, and body mass index, the difference between the 2 groups remained statistically significant (P < .01).
In a cross-sectional analysis of the HIV-infected patients, we found significant associations between FMD and current injection drug use, hazardous drinking, HIV load, and a-high-density lipoprotein triglyceride levels, but not PI therapy.
In a multivariate analysis, only current injection drug use and a lower a-high-density lipoprotein triglyceride level were significantly associated with FMD.
Conclusions. HIV-infected patients have significant impairment of endothelial function, and this impairment is worse among those with elevated levels of HIV replication, particularly injection drug users.
INTRODUCTION
Since the introduction of antiretroviral therapy, mortality among persons with AIDS has decreased substantially [1-5]. As people live longer with HIV disease, they develop chronic manifestations of HIV infection, such as lipodystrophy, dyslipidemia, and glucose intolerance [6-10]. More recently, concern has arisen that the onset of coronary artery disease is also accelerated in HIV-infected patients [11]. Several studies have tried to demonstrate an association between the endocrine abnormalities related to PI use and an increased risk of coronary artery disease among HIV-infected patients [12-18]. However, this association remains controversial [19].
The pathogenesis of atherosclerotic disease in patients with HIV infection is unknown. Endothelial dysfunction is an early event in atherogenesis [19-21], and brachial artery ultrasound is a well-established noninvasive method of assessing endothelium-dependent vasodilation [22-25]. Some studies have demonstrated an association of endothelial dysfunction with PI use in HIV-infected persons [25-26], whereas other studies have not [27]. Therefore, we compared the endothelial function of HIV-infected persons with that of persons without HIV infection, using brachial artery ultrasound. In addition, we examined factors that were associated with endothelial function in HIV-infected patients.
Patient population.
HIV-infected patients from the Boston Medical Center Infectious Disease Clinic were enrolled in the present study. The patients were recruited from a pool of patients participating in a longitudinal study of hepatitis C virus infection. A control group of persons who were presumed to be HIV uninfected and who did not have clinically defined diabetes mellitus (i.e., the patient had a fasting glucose level of >126 mg/dL or was taking medications for hypoglycemia), hypertension (the patient had a blood pressure measurement of >140/90 mm Hg or was taking blood pressure-lowering medications), or cardiovascular disease was drawn from a database of subjects previously studied at Boston Medical Center; this cohort has been described elsewhere [28]. The Boston Medical Center Institutional Review Board approved the study, and all participants provided written, informed consent. We excluded from the study pregnant women, patients receiving hemodialysis, and patients with uncontrolled hypertension at the time of the study visit. Study subjects were defined as receiving a PI if they had been receiving a PI regimen for at least 3 consecutive months at the time of the study visit. Subjects were defined as not receiving a PI if they did not meet this criterion. Metabolic syndrome was defined by the presence of at least 3 of the following factors, as defined by the National Cholesterol Education Panel [29]: central obesity, as specified by waist measurement; presence of dyslipidemia with low high-density lipoprotein (HDL) levels; high triglyceride levels; evidence of a high fasting blood glucose level; and the presence of hypertension.
DISCUSSION
The present study showed a strong association between HIV-1 infection and endothelial dysfunction. This finding is consistent with several studies that have described the association of coronary artery disease with HIV infection [12, 34]. Joshi et al. [35] have documented significant inflammation involving the coronary arteries of children infected with HIV. This inflammatory state is a well-known precursor of the cascade of events leading to arteriosclerosis. Furthermore, HIV infection by itself may be considered a procoagulant state. Various biological markers of endothelial cell dysfunction that may contribute to this procoagulant environment have been described elsewhere [36-41].
The present study also showed a strong correlation between current injection drug use and impaired FMD among HIV-infected persons. Although multiple comparisons were performed, the strength of this association was such that it is not likely to be attributable to chance. Injection drug use (mainly cocaine use) is a well-known risk factor for coronary artery disease [42]. Repetitive toxic effects of cocaine or other illicit substances could damage the endothelium, leading to repeating cycles of cell loss and a decreased ability of the endothelium to release nitric oxide in response to a physiological stimulus. It is also possible that cocaine or other drugs may be factors in the down-regulation of endothelium-dependent vasorelaxation. In support of this notion, Havranek et al. [43] used plethysmography to compare the forearm blood-flow response to intra-arterial acetylcholine in long-term cocaine users with that in control subjects. The mean forearm blood flow was lower in long-term cocaine users than in control subjects. In our cohort, we identified 12 current drug users (28%) in the non-PI group, compared with 3 current drug users (9%) in the PI group. This potentially could explain the higher FMD value in the PI group.
When injection drug use was removed from the multivariate model, HIV load became a significant predictor of FMD. This finding suggests that at least part of the effect of current drug use is mediated by the HIV load. This mediation is biologically plausible, because current drug use is known to adversely affect adherence to antiviral medications, and poor adherence would be expected to be associated with a higher HIV load. Of note, 12 (80%) of the current drug users studied had viral loads that were not completely suppressed. Our results therefore support the observation by Blum et al. [44] that HIV load may be an independent risk factor for endothelial dysfunction.
On the other hand, the present study did not substantiate the observation by Stein et al. [25] that PI-induced dyslipidemia and insulin resistance were associated with endothelial dysfunction. Although the results reveal a slightly worse lipid profile among subjects in the PI-treated group, our analysis did not show any significant worsening of FMD values associated with lipid status or presence of insulin resistance, and patients receiving PI treatment displayed slightly better endothelial function than did patients who were not receiving PI therapy. Although 57% of our study population was black, compared with 5% of the study population evaluated by Stein et al. [25], and although an effect modification by race was observed, this effect modification does not explain the lack of effect of PI on FMD.
The role of metabolic abnormalities in HIV-infected patients receiving PI therapy and the potential association of such abnormalities with accelerated cardiovascular and cerebrovascular disease in HIV-infected patients therefore remain controversial [45]. In a recent study, improvement in the lipoprotein profile occurring in association with the use of pravastatin in HIV-infected patients receiving a PI did not lead to significant improvement in FMD [46]. Some small case series [11, 12, 47] and large retrospective cohorts [13-14] have demonstrated an excess of cardiovascular events associated with PI therapy among HIV-infected persons. However, the largest observational study with the longest follow-up to date failed to confirm these observations [18].
Although HIV infection appears to be associated with substantial impairment of endothelial function, the degree to which this impairment translates into increased risk for cardiovascular disease in persons with HIV infection is still unknown. Modena et al. [23] prospectively evaluated brachial artery FMD in 407 postmenopausal women with a new diagnosis of hypertension but with no known atherosclerosis. Failure to note improvement in FMD after 6 months of antihypertensive therapy was an independent predictor of coronary events occurring during the next 5 years. If FMD has the same predictive value in subjects with HIV infection, then the subjects who we studied are at substantial risk for cardiovascular disease. On the other hand, if the decreased FMD values that we observed improve when the HIV load is reduced, then the long-term risk for cardiovascular disease may be small.
The present study had several potential limitations. Both diet and exercise can potentially affect FMD. We had patients fast before the study, so the short-term effects of the last meal eaten should not be an issue, but we cannot exclude the possibility that long-term diet might have varied among the patient groups. Similarly, exercise affects risk factors that influence endothelial function, such as HDL and glucose levels, and it also may have direct effects on the endothelium. In addition, it is possible that an association between endothelial function and PI therapy was missed because of confounding by indication (i.e., persons who were at risk for developing impaired endothelial function were preferentially prescribed a regimen that did not contain a PI). However, the patients who were not receiving PIs were not, for the most part, those who had experienced failure of a PI regimen, and clinical predictors of dyslipidemia during receipt of PI therapy have not been identified.
We conclude that HIV-infected persons have substantial impairment of endothelial vasomotor function and that this impairment is worse among persons with elevated levels of HIV replication, particularly IDUs. In the present study, impaired endothelial function was not associated with PI therapy; however, it was associated with a lower -HDL triglyceride level, which is a presumed marker of metabolic imbalance. Prospective studies are needed to determine whether effective antiretroviral treatment can lead to improved endothelial function in HIV-infected persons.
RESULTS
Clinical characteristics. Seventy-six patients who were seropositive for HIV-1 (confirmed by Western blot analysis) were enrolled in the study. Results for 1 study subject were not analyzed because of the poor quality of the imaging studies. Data for 227 subjects presumed to be HIV-1 uninfected were selected from the database of studies previously performed at Boston Medical Center. The demographic and clinical characteristics of the 2 groups are presented in table 1. HIV-infected patients were older, were more likely to be black, and had more abnormal lipid profiles.
Brachial artery ultrasound results. HIV-infected patients had significantly lower FMD, compared with HIV-uninfected control subjects (7.3% ± 4.4% vs. 11.1% ± 6.3%, respectively; P < .0001). Non-endothelium-dependent NMD did not differ between the 2 groups (17.4% ± 8.2% vs. 20.1% ± 9.1%, respectively; P = .11). Table 2 displays FMD and NMD values according to the subjects' risk factors. Older age (>40 years), smoking, male sex, higher BMI, higher LDL level, and lower HDL level were all associated with significantly lower values for both FMD and NMD. Patients with triglyceride levels of >200 mg/dL also had lower FMD values, but they demonstrated no significant difference in NMD, compared with patients with triglyceride levels of <200 mg/dL.
Predictive value of endothelial function. As shown in table 3, independent predictors of endothelial dysfunction in multivariate modeling were HIV status, smoking, sex, and BMI. This model accounts for 27% of the variability in FMD as measured by brachial artery ultrasound. HIV-infected patients had significant endothelial dysfunction, compared with HIV-uninfected control subjects, for an adjusted parameter estimate of 3.6% (95% CI, 1.8%-5.2%; P < .01).
Factors associated with FMD among HIV-infected patients. To further examine the effect of HIV infection on endothelial function, we performed an analysis of factors associated with FMD among the HIV-infected patients. Forty of the subjects were current or former injection drug users (IDUs). Of these subjects, 15 were current IDUs and 25 were former IDUs. The median duration of injection drug use was 24 years (range, 4-34 years). Of the 15 current IDUs, 12 (80%) were using heroin and 8 (53%) were using cocaine. Of the 25 former IDUs, 22 (88%) had used heroin and 22 (88%) had used cocaine. Of the 75 subjects, 32 were receiving a PI regimen, and 43 were receiving a non-PI regimen. Distributions of demographic characteristics, hepatitis C virus status, injection drug use, and smoking status were similar in PI and non-PI groups (table 4). Patients receiving a PI regimen had higher LDL levels than did patients who were receiving a non-PI regimen (120 ± 36 mg/dL vs. 103 ± 71 mg/dL, respectively; P = .03). Overall PI therapy was not associated with the metabolic syndrome or with worsening endocrine abnormalities. However, patients receiving PI regimens were more obese, as documented by waist measurements (P = .04).
Table 5 shows the results of univariate associations between demographic characteristics and laboratory variables, FMD, and NMD among the HIV-infected patients. In addition to the factors shown in the table, the following parameters were also examined: waist size; waist-to-hip ratio; fasting glucose and fasting insulin levels; and hemoglobin A1c, high-sensitivity C-reactive protein, apolipoprotein B, total cholesterol, pre-Beta very low density lipoprotein cholesterol, total triglyceride, LDL, HDL, lipoprotein (a) cholesterol, lipoprotein (a) triglyceride, and Beta low-density triglyceride levels. None of the factors were statistically significantly associated with FMD. HIV load was significantly associated with FMD, as were use of injection drugs during the past year and hazardous drinking (defined by a score of >8 on the Alcohol Use Disorders Identification Test [AUDIT] questionnaire). There was an association between PI use and FMD, but it did not reach statistical significance, and FMD was better in the PI group than in the non-PI group. In stratified analysis, the mean FMD was 2.4% higher in the PI group, compared with the non-PI group, among black subjects, but it was only 0.8% higher in the PI group, compared with the non-PI group, among nonblack subjects.
In multivariate analysis, independent predictors of impaired FMD were current injection drug use (FMD difference, 3.3%; 95% CI, 0.9%-5.7%; P = .007) and the a-HDL triglyceride level (FMD difference, 3.2%; 95% CI, 1.3%- 5.1%; P = .002). This model explained 26% of the variability in FMD among the HIV-infected patients. However, when current injection drug use was removed from the model, HIV load became a significant predictor of impaired FMD (FMD difference, 2.1%; 95% CI, 0.2%-4.0%; P = .04).
Study protocol. Using a standardized questionnaire, we recorded patient data on smoking history, family history of coronary artery disease, diabetes mellitus, and hypertension. We collected blood samples and measured fasting blood glucose and insulin levels, the high-sensitivity C-reactive protein level, and the fasting lipid profile and lipoprotein values. We also measured the body mass index (BMI; calculated as weight in kilograms divided by the square of height in meters), the waist-to-hip ratio, and the brachial-to-ankle ratio. All study subjects fasted overnight and took their most recent antihypertensive medication at least 24 h before the study. Subjects refrained from exercising on the day of the study visit, and they refrained from smoking for at least 4 h before the study.
We used flow-mediated dilation (FMD), as measured by brachial artery ultrasound, as an indicator of endothelial function. Peak hyperemic blood flow was estimated to be the product of the vessel cross-sectional area, flow velocity, and heart rate, by use of the cardiac cycle with the largest flow velocity integral after cuff deflation. The percent increase in blood flow during hyperemia was expressed as the percent increase in flow from baseline.
We performed FMD measurements according to an established protocol [30]. In brief, study subjects were positioned in a comfortable supine position. After a 15-min rest, the baseline end-diastolic brachial artery diameter and brachial artery flow velocity above the antecubital fossa were recorded. Ischemia of the forearm was induced by inflating a blood pressure cuff on the upper arm for 5 min at a pressure of at least 100 mm Hg above the systolic blood pressure. Peak hyperemic flow was recorded within 15 s after cuff release, and the brachial artery diameter was recorded 60 s after cuff release. After reactive hyperemia developed, we allowed 10 min of rest for restoration of baseline conditions. To assess vascular response to an exogenous source of nitric oxide, we recorded brachial artery diameter and velocity before and after sublingual administration of nitroglycerin (0.4 mg). Nitroglycerin was not given if a patient had a systolic blood pressure of <100 mm Hg, had a history of adverse reaction to nitrates, refused to take nitroglycerin, or had used sildenafil within 7 days of the study. The same technician performed ultrasound for all study patients and was blinded to their HIV treatment.
Statistical analysis. We analyzed data by use of SAS software, version 8.02 SE (SAS). We reported the descriptive statistics as mean values ± SD. FMD and nitroglycerin-mediated dilation (NMD) were expressed as the percentage increase from the baseline value. Differences between groups were evaluated using the nonparametric Wilcoxon rank test. We used a stepwise regression to select covariates that affected FMD among HIV-infected patients and control subjects. On the basis of current published literature on endothelial dysfunction [31-33], age, BMI, smoking, total cholesterol level, fasting blood glucose level, and sex were included in the model. HIV serologic status was also included in the model, as were any variables that were associated with FMD (P .15). We used a similar approach to identify covariates that affected FMD among the HIV-infected patients.
|
|
| |
| |
|
|
|
