Cocaine use modifies the association between antiretroviral therapy and endothelial dysfunction among adults with HIV infection

Abstract

Cocaine is commonly used among HIV-infected people, and may worsen HIV disease progression. In addition, existing evidence suggests a link between antiretroviral regimens and endothelial dysfunction. This study aimed to examine whether the associations of antiretroviral therapy (ART) regimens with endothelial dysfunction may be modified by cocaine use in adults with HIV infection. Between 2003 and 2014, 466 HIV-positive participants residing in Baltimore, Maryland, were enrolled in a study investigating comorbidities associated with HIV/ART. The associations between various risk factors and endothelial dysfunction indicators were examined by robust regression models fitted for the overall subjects and cocaine subgroups, separately. Duration of nonnucleoside reverse-transcriptase inhibitor (NNRTI) based therapy was negatively associated with plasma vWF:Ag levels in cocaine non-users (β=−0.715, SE=0.220, p<0.05). However, cocaine users on longer-term NNRTI-based regimens had greater plasma ET-1 concentrations than their counterparts (β=0.003, SE=0.001, p<0.05). Additionally, current cigarette smoking was significantly positively associated with ET-1 concentrations in both cocaine non-users (β=0.609, SE=0.164, p<0.05) and cocaine users (β=0.331, SE=0.086, p<0.05). In conclusion, cocaine use modified potential effects of NNRTI-based therapy on biomarkers of endothelial dysfunction. These findings suggested that reduction in cocaine use may improve endothelial function in HIV-infected cocaine users.

1. INTRODUCTION

Although new HIV infections have continuously decreased in the past years, it was estimated that more than 1.1 million people were living with HIV in the US at the end of 2015, half of whom were adults age 50 years and over¹. The development and widespread practices of combination antiretroviral therapy (ART) have significantly improved life expectancy of people infected with HIV. Nowadays, immunodeficiency-related events account for approximately 20% of deaths in this population² and non-AIDS chronic diseases have become dominant health conditions³.

Mounting evidence has suggested that endothelial dysfunction may play an important role in the onset and development of coronary artery disease (CAD)^4–6. Endothelial dysfunction refers to impaired vasodilation responses as well as a proinflammatory and prothrombotic state⁷. Endothelin-1 (ET-1) is the most potent constrictor of vascular smooth muscle known in humans and may lead to endothelial dysfunction by decreasing nitric oxide (NO) bioavailability⁷. Another sensitive and reliable marker of endothelial dysfunction is von Willebrand factor (vWF)⁸. vWF is synthesized in vascular endothelial cells and megakaryocytes⁸. vWF antigen (vWF:Ag) measures circulating mature vWF in plasma and may increase when endothelial damage occurs⁹.

Laboratory and clinical studies have indicated associations between antiretroviral drugs and significant endothelial dysfunction^10,11. Our prior study in Baltimore, Maryland found that median ET-1 level of cocaine users (0.96 pg/mL, IQR 0.71–1.36 pg/mL) was significantly higher than that of cocaine non-users (0.72 pg/mL, IQR 0.58–1.06)¹². In addition, our cash-based incentive intervention study observed significant decreases in plasma ET-1 concentrations among HIV-infected participants who quitted or reduced cocaine use at the 6-month follow-up¹³.

Cocaine use is prevalent among HIV-infected people¹⁴, and may aggravate progression and survival in HIV infection^13,15. Our previous publications have demonstrated that cocaine use affected the associations of HIV/ART with myocardial steatosis and subclinical CAD^16,17. However, few studies have ever examined possible modification effects of cocaine use on ART-associated endothelial dysfunction in people with HIV infection. This study sought to investigate whether chronic use of cocaine may modify the associations between ART treatment and biomarkers of endothelial dysfunction by analyzing baseline data from a group of HIV-infected adults who resided in Baltimore, Maryland, US.

2. MATERIALS AND METHODS

2.1. Study participants

A total of 592 HIV-positive adults ≥21 years were recruited between August 2003 and September 2014 to participate in an observational study investigating the effects of HIV infection, antiretroviral drugs, substance use, and other factors on subclinical atherosclerosis. None of the participants had any clinical symptoms related to CAD or any diagnosis of CAD. Nor did the participants report a previous history of statin use. HIV infection status was evaluated by enzyme-linked immunosorbent assay (ELISA) and determined by Western blot test. The final analyses included 466 participants who had complete data on plasma vWF:Ag and ET-1 levels. The study protocol was approved by the Institutional Review Board at the Johns Hopkins School of Medicine, and all procedures involved in this study followed institutional guidelines. Written informed consent was obtained from all the participants.

2.2. Substance use and ART

The interviews at baseline were performed by trained research staff. Information on participants’ demographics, specific lifestyle behaviors, medical history, and HIV medications was collected. Participants were asked to recall the durations and frequencies of cocaine and other illicit drug use. Cocaine non-users were defined if a participant had never used cocaine. Chronic use of cocaine was considered with continuous administration of cocaine ≥4 times per month by any route for ≥ 6 consecutive months. Cocaine users with use of other substance, e.g., alcohol, opiates, methamphetamine, or benzodiazepines, were included. Those that used cocaine infrequently, i.e., less than 4 times per month, or less than 6 consecutive months, were excluded from the final analyses.

Participants’ medical history and medications were self-reported and further confirmed by medical chart reviews. Information on both names and durations of ART regimens was collected. Durations of ART regimens were presented in months. ART regimens were summarized into three groups: nonnucleoside reverse-transcriptase inhibitors (NNRTIs), protease inhibitors (PIs), and nucleoside reverse-transcriptase inhibitors (NRTIs).

2.3. Plasma von Willebrand factor antigen (vWF:Ag) and endothelin-1 (ET-1)

Plasma for vWF:Ag and ET-1 assessment was prepared by centrifugation of citrated blood samples at 2500 rpm for 15 minutes at 4℃. Plasma samples were stored at −80 °C until analyzed. vWF:Ag was measured by the Asserachrom VWF:Ag ELISA kit with an intra assay coefficient of variation of 3.5% (Diagnostica Stago, Inc.). ET-1 was evaluated using the Quantikine ET-1 ELISA kit (R&D Systems, Inc.), of which the intra assay coefficient of variation was 4.0%.

2.4. Potential confounders

Participants’ blood pressure was measured three times in the seated position and an average of the three readings was used (mmHg). Body mass index (BMI, kg/m²) was calculated based on body weight and height measured by nurses at the clinical research unit. Blood samples were collected after an overnight fast. Blood glucose was measured using spectrophotometry (SP) by Quest Diagnostics (Baltimore, MD). Details on serum preparation and lipid assessment can be found elsewhere¹⁸. Baseline HIV viral loads and CD4 cell counts were abstracted from medical charts.

2.5. Statistical analysis

All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC). Continuous variables, including age, BMI, lipid profiles, blood pressure, and durations of ART regimens, were described with medians and interquartile ranges (IQRs). Proportions were calculated for categorical demographic/lifestyle variables. Comparisons between cocaine users and non-users in those non-normally distributed continuous variables were made by the nonparametric Wilcoxon rank-sum test. The categorical variables were compared using the chi-square test.

Linear regression models were fitted to examine the associations of clinical and lifestyle risk factors with plasma vWF:Ag and ET-1 levels in the total sample, cocaine use subgroup, and cocaine non-use subgroup, separately. Given the presence of non-normal distributions and outliers of our data, robust regression models rather than multiple linear regression models were used to provide robust estimates of parameters because robust regression models were based on the least trimmed squares (LTS) estimation method, and less sensitive to outliers and high leverage points¹⁹.

The unadjusted associations between endothelial dysfunction indicators and individual risk factor were evaluated by univariate robust regression models. The risk factors that were examined included age, sex, BMI, current cigarette smoking, current alcohol use, years for cigarette smoking, years for alcohol consumption, fasting glucose, triglycerides, total cholesterol, high density lipoprotein cholesterol, systolic and diastolic blood pressure, baseline HIV viral load level, baseline CD4 cell count, months of PI use, months of NRTI use, and months of NNRTI. Predictive variables with a significance level of p<0.15 in those univariate models remained in the multivariate robust regression models, which were used to investigate the combined effects of these risk factors on plasma vWF:Ag and ET-1 levels. After repeating the process of deleting, refitting, and verifying, the final models only contained the variables that made significant contributions. As for the factors that were no longer statistically significant based on the Wald statistic, they were removed in a stage wise manner from the multivariate robust regression models. All the statistical tests were two-sided and a p value less than 0.05 was regarded as statistically significant.

3. RESULTS

3.1. Selected characteristics by the use of cocaine

Of the 466 subjects with HIV infection, 302 were cocaine users, 61.2% were males, 62.9% were current cigarette smokers, and 35.2% had current alcohol consumption (Table 1). The median age of the participatory subjects was 47 years (IQR 41–52 years). The median plasma level was 204.90% (IQR 149.00–276.80%) for vWF:Ag and 1.51 pg/mL (IQR 1.07–1.93 pg/mL) for ET-1. As compared to cocaine non-users, cocaine users were significantly more likely to currently smoke, smoke cigarette and consume alcohol for longer years, and have greater HIV viral loads at baseline (all p<0.05). Cocaine users also had significantly greater plasma vWF:Ag and ET-1 levels than cocaine non-users (both p<0.05). Sex, race/ethnicity, BMI, blood pressure, lipid profiles, and months of ART treatments did not significantly differ between two cocaine subgroups (all p>0.05). The percentage of patients on PI-based, NRTI-based, or NNRTI-based therapy was 58.4%, 72.8% and 34.3%, respectively.

TABLE 1.

Characteristics of study participants by cocaine use status (n=466)^a

Characteristics	All			Cocaine use			Cocaine non-use			P valueb
N (%)	466			302	(64.8)		164	(35.2)
Age (years)	47	(41–52)		48	(43–52)		43	(33–51)		<0.001
Male (%)	285	(61.2)		187	(61.9)		98	(59.8)		0.647
Race (%)
Non-Hispanic black	438	(94.0)		285	(94.4)		153	(93.3)		0.157
Non-Hispanic white and others	28	(6.0)		17	(5.6)		11	(6.7)
Current cigarette smoking (%)	293	(62.9)		227	(75.2)		66	(40.2)		<0.001
Years of cigarette smoking	22	(5–32)		27	(20–35)		5	(0–20)		<0.001
Current alcohol use (%)	164	(35.2)		112	(37.1)		52	(31.7)		0.246
Years of alcohol use	16	(2–30)		20	(7–30)		5	(0–16)		<0.001
Body mass index (BMI, kg/m2)	25.3	(22.3–29.5)		25.3	(22.5–29.1)		25.3	(22.0–30.7)		0.930
Systolic BP (mm Hg)	118	(108–128)		118	(108–129)		118	(109–128)		0.794
Diastolic BP (mm Hg)	72	(65–81)		73	(67–81)		71	(63–81.5)		0.060
Fasting glucose (mg/dL)	85	(78–92)		84	(78–91)		85	(78–93)		0.569
Total cholesterol (mg/dL)	161.0	(140.0–186.0)		159.0	(137.0–184.0)		166.0	(142.0–188.0)		0.059
Triglycerides (mg/dL)	105.5	(74.0–155.0)		109.0	(78.0–160.0)		103.0	(66.0–144.0)		0.065
HDL–C (mg/dL)	50.0	(39.0–62.0)		50.0	(38.0–63.0)		50.0	(42.0–60.0)		0.582
CD4 count at baseline (cells/mm3)	346	(214–548)		333	(206–555)		384	(274–528)		0.207
HIV RNA load at baseline (copies/mL)	1027	(202–32730)		2750	(202–52620)		202	(202–18764)		0.049
Duration of ART therapy (months)	36.0	(5.9–92.2)		44.0	(5.9–91.1)		35.0	(8.9–94.1)		0.815
Duration of PI therapy (months)	12.0	(0–60.0)		16.4	(0–60.0)		12.0	(0–59.6)		0.134
Duration of NRTI therapy (months)	26.0	(0.9–76.0)		29.0	(1.9–72.0)		24.0	(0.9–82.5)		0.875
Duration of NNRTI therapy (months)	0	(0–14.0)		0	(0–12.0)		0	(0–19.5)		0.070
vWF:Ag (%)	204.90	(149.00–276.80)		220.15	(163.90–286.80)		183.90	(125.75–261.20)		<0.001
Endothelin-1 (pg/mL)	1.51	(1.07–1.93)		1.55	(1.16–2.05)		1.34	(1.02–1.70)		<0.001

^aMedian (interquartile range) was presented for continuous variables, and proportion (%) was presented for categorical variables.

^bNonparametric Wilcoxon rank-sum test was used to compare continuous variables. Chi-square test was used to compare categorical variables.

3.2. The associations between durations of ART regimens and plasma vWF:Ag levels, stratification by the use of cocaine

The associations of lifestyle and clinical risk factors with vWF:Ag measurements were presented in Table 2. As 94% of the subjects were African Americans, race/ethnicity was not adjusted for in the univariate and multivariate regression models. Only those variables with significance or p<0.15 in the univariate analyses were shown in the table. The multivariate robust regression analysis of the overall sample showed that age, sex, years of cigarette smoking, total cholesterol, fasting glucose, and duration of NNRTI use were significantly associated with plasma vWF:Ag levels (all p<0.05). In those who had never used cocaine, only diastolic blood pressure (β=1.711, SE=0.520, p<0.05), total cholesterol (β=−0.462, SE=0.189, p<0.05), and duration of NNRTI use (β=−0.715, SE=0.220, p<0.05) were independently associated with vWF:Ag levels. However, current cigarette smoking (β=1.511, SE=0.378, p<0.05) and fasting glucose (β=0.305, SE=0.143, p<0.05) but not duration of NNRTI use (p>0.05) were significantly associated with vWF:Ag levels in cocaine users.

TABLE 2.

The associations of demographic and clinical characteristics with plasma von Willebrand factor antigen (vWF:Ag) levels by cocaine use status, robust regression

	Univariate model				Final multivariate model
Characteristics	Estimate	SE	P value		Estimate	SE	P value
	All
Age	1.684	0.436	<0.001		1.225	0.503	0.015
Female	21.664	8.034	0.007		29.405	8.114	<0.001
Current cigarette smoking	28.994	7.906	<0.001
Years of cigarette smoking	1.457	0.265	<0.001		1.145	0.297	<0.001
Years of alcohol use	0.761	0.281	0.007
Long-term cocaine use	28.563	8.129	<0.001
Systolic BP (mm Hg)	0.475	0.229	0.038
Diastolic BP (mm Hg)	0.887	0.326	0.007
Fasting glucose (mg/dL)	0.336	0.127	0.008		0.322	0.124	0.010
Total cholesterol (mg/dL)	−0.292	0.108	0.007		−0.260	0.103	0.012
HDL–C (mg/dL)	−0.381	0.210	0.070
Duration of PI therapy (months)	0.117	0.078	0.136
Duration of NNRTI therapy (months)	−0.405	0.134	0.002		−0.391	0.132	0.003
	Cocaine use
Age	1.807	0.694	0.009
Female	19.722	9.901	0.046
Current cigarette smoking	20.501	10.817	0.058		1.511	0.378	<0.001
Years of cigarette smoking	1.510	0.374	<0.001
Years of alcohol use	0.747	0.346	0.031
Fasting glucose (mg/dL)	0.327	0.146	0.026		0.305	0.143	0.033
Total cholesterol (mg/dL)	−0.200	0.124	0.107
Triglycerides (mg/dL)	0.086	0.059	0.145
	Cocaine non-use
Age	1.067	0.608	0.080
Female	26.705	13.287	0.044
Current cigarette smoking	25.363	13.303	0.057
Years of cigarette smoking	1.013	0.512	0.048
Systolic BP (mm Hg)	0.643	0.375	0.087
Diastolic BP (mm Hg)	1.269	0.525	0.016		1.711	0.520	0.001
Total cholesterol (mg/dL)	−0.333	0.196	0.090		−0.462	0.189	0.015
Triglycerides (mg/dL)	0.187	0.092	0.042
HDL–C (mg/dL)	−0.792	0.367	0.031
Duration of NNRTI therapy (months)	−0.805	0.232	<0.001		−0.715	0.220	0.001

3.3. The associations between durations of ART regimens and plasma ET-1 concentrations, stratification by the use of cocaine

Table 3 presented the results of multivariate robust regression analysis on plasma ET-1 concentrations. Only those variables with significance or p<0.15 in the univariate analyses were included in the multivariate analyses. Age (β=0.012, SE=0.003, p<0.05) and current cigarette smoking (β=0.324, SE=0.059, p<0.05) were significantly associated with ET-1 concentrations in the pooled sample. In cocaine users, multivariate robust regression models showed that age (β=0.024, SE=0.006, p<0.05), total cholesterol (β=−0.002, SE=0.001, p<0.05) and duration of NNRTI use (β=0.003, SE=0.001, p<0.05) were independently associated with plasma ET-1 concentrations. However, baseline CD4 cell count (β=−0.001, SE=0.000, p<0.05) but not duration of NNRTI use (p>0.05) was significantly associated with ET-1 concentrations in those who had never used cocaine. Current cigarette smoking was significantly associated with ET-1 concentrations in both cocaine non-users (β=0.609, SE=0.164, p<0.05) and users (β=0.331, SE=0.086, p<0.05).

TABLE 3.

The associations of demographic and clinical characteristics with plasma endothelin-1 (ET-1) concentrations by cocaine use status, robust regression

	Univariate model				Final multivariate model
Characteristics	Estimate	SE	P value		Estimate	SE	P value
	All
Age	0.012	0.003	<0.001		0.012	0.003	<0.001
Current cigarette smoking	0.317	0.060	<0.001		0.324	0.059	<0.001
Years of cigarette smoking	0.012	0.002	<0.001
Years of alcohol use	0.005	0.002	0.040
Long-term cocaine use	0.239	0.062	<0.001
Body mass index (BMI, kg/m2)	−0.007	0.005	0.149
Duration of PI therapy (months)	0.001	0.001	0.026
Duration of NRTI therapy (months)	0.002	0.001	0.009
Duration of NNRTI therapy (months)	0.002	0.001	0.068
	Cocaine use
Age	0.017	0.006	0.003		0.024	0.006	<0.001
Current cigarette smoking	0.275	0.089	0.002		0.331	0.086	<0.001
Years of cigarette smoking	0.011	0.003	0.000
Total cholesterol (mg/dL)	−0.002	0.001	0.062		−0.002	0.001	0.049
Duration of PI therapy (months)	0.001	0.001	0.081
Duration of NRTI therapy (months)	0.002	0.001	0.027
Duration of NNRTI therapy (months)	0.004	0.001	0.003		0.003	0.001	0.004
	Cocaine non-use
Current cigarette smoking	0.310	0.080	<0.001		0.609	0.164	<0.001
Years of cigarette smoking	0.006	0.003	0.058
Systolic BP (mm Hg)	−0.004	0.002	0.095
HDL–C (mg/dL)	−0.005	0.002	0.054
CD4 count at baseline (cells/mm3)	−0.0006	0.0003	0.065		−0.001	0.000	0.008

4. DISCUSSION

In this study we observed that (1) duration of NNRTI-based therapy was significantly negatively associated with plasma vWF:Ag levels in cocaine non-users; (2) duration of NNRTI-based therapy was significantly positively associated with plasma ET-1 concentrations in those with cocaine use; and (3) current cigarette smoking was significantly positively associated with ET-1 concentrations in both cocaine non-users and users. To the best of our knowledge, this is the first study to investigate the potential modification effects of cocaine use on ART-associated endothelial dysfunction in adults with HIV infection.

NNRTIs bind to HIV-1 reverse transcriptase (RT) in a hydrophobic pocket near the enzyme’s catalytic site, and highly specifically inhibit HIV-1 RT²⁰. This binding may induce conformational changes and cause reduced catalytic activities of HIV-1 RT²⁰. Unlike NRTIs which achieve antiviral activities based on phosphorylation²¹, NNRTIs do not require phosphorylation, and are recommended as alternatives to NRTIs and PIs in HIV-infected patients prone to atherosclerosis²². However, recent clinical studies reported increased oxidative stress assessed by serum F2-isoprostaine concentrations in those patients receiving NNRTI regimen efavirenz^23,24. Moreover, efavirenz was found to decrease tight junction proteins, promote superoxide anion, and increase endothelial permeability in endothelial cells of human coronary artery²⁵.

Existing studies have reported that high plasma vWF:Ag levels are associated with greater risks of CAD, ischemic stroke, and even all-cause and cardiovascular mortality^26–28. These studies were conducted in general, non HIV-infected populations. Despite increasing evidence on negative effects of PIs and NRTIs on vascular function, the findings on endothelial toxicities of NNRTI-based regimens are inconclusive^22,23. A prospective study among 82 treatment-naïve HIV-infected patients observed a significant improvement in endothelial function measured as increased flow-mediated dilation (FMD) of the brachial artery after initiating ART treatment including NNRTIs²⁹. A hospital-based cohort study indicated that PIs but not NNRTIs were associated with development of premature atherosclerosis³⁰. Our overall sample suggested that plasma vWF:Ag levels were significantly lower among HIV-positive people who used NNRTIs for longer time. No such association was found for plasma ET-1 concentrations in our pooled sample. However, a 12-month prospective study indicated that NNRTI regimen efavirenz was more likely to decrease FMD than PIs in HIV-infected patients²³. The AIDS Clinical Trials Group (ACTG) Study A5202 also reported significantly greater likelihood of ischemic events in tenofovir (NRTI)/emtricitabine (NRTI)/efavirenz (NNRTI) group as compared to other treatment groups³¹. Such discrepancies across studies may be due to potential effects of modifying or confounding factors that are not accounted for in the analyses, e.g., substance use.

A main result of this study was that the associations between NNRTIs and endothelial dysfunction may be modified by cocaine use status of HIV-infected individuals. Plasma vWF:Ag levels were significantly negatively associated with months of NNRTI therapy among cocaine non-users, which disappeared in cocaine users. However, plasma ET-1 concentrations were predominantly increased with duration of NNRTI use in the individuals with cocaine use but not in those that never used cocaine. These results were consistent with our findings presented in recent publications. Our previous research has observed that cocaine use may acerbate ART-associated myocardial steatosis and coronary stenosis^16,17.

Although existing studies have indicated adverse impact of cocaine use on endothelial function, the molecular mechanisms that underlie the intervening effects of cocaine use on NNRTIs and their associated endothelial dysfunction need a further investigation. Our prior research has detected the association of cocaine use with endothelial damage measured as elevated ET-1 concentrations, and observed noncalcified plaque regression at 12-month follow-up in cocaine users who reduced their cocaine use^12,13,32. In human aortic endothelial cells, long-term exposure to cocaine suppressed NO production and endothelial NO-synthase expression³³. Cocaine use may also increase the concentrations of inflammation biomarkers such as C-reactive protein (CRP) and serum amyloid A^32,34. Systematic inflammation levels in HIV-infected people may be reduced by ART regiments including NNRTIs but remain higher than those in HIV-negative counterparts³⁵. Randomized trials showed that despite decreases in inflammation biomarkers such as tumor necrosis factor-α and IL-6, CRP levels increased at 24 and 96 weeks than at baseline in HIV-infected patients administered with efavirenz (NNRTI)^36,37. Therefore, we postulated that inflammation process, particularly those pathways activated via CRP, may be involved in the possible interactions between cocaine use and NNRTI-based regimens with regard to development of endothelial dysfunction. Furthermore, different effects of cocaine in ET-1 and vWF:Ag were observed in this study. ET-1 and vWF:Ag are synthesized via distinct pathways and loosely correlated under some circumstances^38,39. Different aspects of endothelial dysfunction that ET-1 and vWF:Ag reflect may partly explain the observed discrepancies^40–42.

Our data also showed that current cigarette smoking was significantly associated with increased vWF:Ag levels in cocaine users as well as increased ET-1 concentrations in both cocaine users and non-users. It is estimated that approximately 50–70% people with HIV infection are current smokers⁴³. In our study, the prevalence of current cigarette smoking was 75.2% in cocaine users and 40.2% in cocaine non-users (p<0.05). Existing literature has linked cigarette smoking with various non-AIDS conditions⁴⁴. A randomized clinical trial of 1504 current smokers reported significantly improved endothelial function indicated by increased FMD of the brachial artery after smoking cessation⁴⁵. All these findings have emphasized the importance and priority of smoking cessation for management and intervention of cardiovascular diseases in HIV-infected people.

This study has several limitations. First, some participants used several illicit drugs. We examined other illicit drugs (e.g., marijuana and heroin) as potential confounders and found that they were not statistically significant in the regression models. Second, the data we analyzed were cross-sectional. Therefore, we were unable to examine the relationships between ART therapy, cocaine use, and endothelial dysfunction indicators over time, and our findings cannot lead to a causal conclusion. Third, evidence has shown that physical activity is related to vWF:Ag and ET-1 levels^46,47. However, data on regular physical activity levels were not collected for our participants. Given high correlations between body weight status and physical activity⁴⁸, BMI was used as a proxy and adjusted for in our regression models. Fourth, we did not have a direct test to measure endothelial dysfunction such as flow-mediated dilatation (FMD). Additionally, vWF propeptide (vWFpp) was not measured in this study. Therefore, we cannot differentiate chronic and acute endothelial activation by calculating the ratios between vWF:Ag and vWFpp⁴⁹. Lastly, substance use was measured based on self-report data in this study. Therefore, we cannot verify the information on nonuse of cocaine.

5. CONCLUSIONS

The main findings of this study indicated that potential influence of NNRTI regimens on endothelial cells may be modified by chronic cocaine use. While NNRTI has a beneficial effect on vWF:Ag in cocaine non-users, which disappears in cocaine users, an independent negative effect of NNRTI on ET-1 has been found in cocaine users. Future studies such as longitudinal research and clinical trials are needed to verify these findings. Medication adherence is usually compromised in HIV-infected populations who receive polypharmacy⁵⁰. Considering that a large percentage of HIV-infected people report using cocaine, reduction in or abstinence from cocaine may be regarded as a supplementary approach to vascular protective medications to prevent endothelial dysfunction in HIV-infected individuals who use cocaine.