INTRODUCTION
Infection caused by human immunodeficiency virus (HIV) is the fifth-leading cause of death among adults worldwide1. The policy of universalizing access to antiretroviral therapy (ART) has been determined to impact on patient survival, changing the profile of the epidemic characterized by the reduction of new cases and mortality1. Despite this, mortality is still 3 to 15 times higher in the population living with HIV than the general population2–4. Although the mortality rate can be attributed to infectious complications, more than half of these deaths are due to non-infectious causes. With the improvement of ART-associated survival, a transition between infectious and non-infectious complications was noted, increasing the importance of chronic degenerative events5–10. The reduction in life expectancy among people living with HIV (PLWHIV) is associated with an increased risk of a series of non-AIDS complications, including heart disease, cancer, liver, kidney diseases, bone diseases, and neurocognitive disorders10,11. HIV carriers are at 1.5 to 2 times higher risk of cardiac events than the general population3,8,11,12–21. This increase may be related to chronic immune-inflammatory activity and immunosenescence in HIV3,6,9, in addition to the classic risk factors. Inflammatory biomarkers are generally elevated in PLHIV12,22–30. These markers have been associated with an increased risk of cardiovascular disease (CVD), opportunistic conditions, and other causes of mortality7,12,13,22,31–35. Persistent immune activation and inflammation associated with classical factors play an important role in the etiology of CVDs17. The identification of subclinical coronary artery disease, assessed by measuring the carotid intima-media thickness (CIMT), considered as a substitutive marker, may represent an early diagnostic strategy in this population18,23–26. The CIMT of PLHIV is thicker than that in the uninfected population25,27–29. In this context, the present study evaluated the profiles of the inflammatory biomarkers related to CVDs and the intima-media thickness as predictors of CVD risk in human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS) patients before antiretroviral therapy.
METHODS
This observational, cross-sectional study was performed among PLWHIV before starting ART, between 2014 and 2015 in a specialized service.
Ethical considerations
The study protocol was approved by the Research Ethics Committee of the Federal University of Minas Gerais (no. 12649713.4.0000.5149).
The inclusion criteria were: age between 18 and 60 years, and both genders with a diagnosis of HIV infection without the prior use of ART. The exclusion criteria were: lack of completion of the protocol for the collection of clinical or laboratory data, pregnancy, or history of chronic non-controlled diseases.
The following information was collected: sociodemographic (age, gender, race, education, and marital status), lifestyle (smoking, alcoholism, and alcohol abuse through the CAGE questionnaire)36, systolic and diastolic blood pressure, anthropometric data (weight and height), and biochemical parameters as fibrinogen, 25-hydroxyvitamin D, lipidogram, blood glucose, polymerase chain reaction (PCR) results, in addition to viral load (HIV-PCR), and LTCD4+ lymphocyte count.
The levels of biomarkers related to CVDs were measured using the Human Cardiovascular Disease (Merck Millipore) kit according to the manufacturer’s recommendations37. Ultrasonography of the carotid arteries was also performed by a single trained examiner, within 10 days after initiation of the treatment. The measurement of CIMT followed the current technical principles26,37–39. The values of intima-media thickness were expressed in terms of vascular age, based on the vascular age calculator (VAC)38.
In the descriptive analysis, qualitative variables were summarized as absolute and relative frequencies and quantitative variables were summarized as means and standard deviations or medians and interquartile ranges. The assumption of normality for the probability distribution of the variables was verified using the Shapiro-Wilk test. Comparisons between two quantitative variables were performed using the Kruskal-Wallis test for non-normally distributed variables and analysis of variance for normally distributed variables. The tests were used to compare serum levels of the biomarkers (ADAMTS13, D-dimer, GDF-15, myoglobin, sICAM-1, MPO, p-selectin, lipocalin-2/NGAL, sVCAM-1, and SAA) among the categories of viral load and Framingham risk score. Biomarkers and viral load correlations were analyzed using Spearman’s rank correlation test. Receiver operating characteristic (ROC) curve analysis was used to investigate cardiovascular risk prediction and verify the association of biomarkers with cardiovascular risk and viral load. The log-binomial model was used to evaluate whether biomarkers were predictors of cardiovascular risk. The univariate model of each marker was adjusted with the risk variable (yes or no); markers that had p-values ≤ 0.20 were considered as candidates for the multivariate model. A multivariate model was fitted with all candidates and step-by-step the markers with higher p-values were withdrawn until all significant markers remained at the 0.05 level. The quality of fit of the model was assessed by the deviance and quality measures of the model, i.e., Akaike information criterion, corrected version of the Akaike information criterion, Bayesian information criterion, and consistent Akaike information criterion. The multivariate linear regression model was used to investigate the correlation between the study variables and the thickness of the intima media. It was initiated by univariate adjustment of the simple linear regression model of each group of variables. Within each group, the candidate variables (with p ≤0.20) were selected, adjusting the multivariate models for each group individually with their respective candidate variables and step-by-step, with the variables with the highest p-values being removed until they were all significant at the 0.05 level. The quality of fit was verified by residual analysis (normality, homoscedasticity, and independence). In all applied tests, significance levels of 5% were used.
RESULTS
In this study, out of 105 patients who were screened, 5 did not meet the inclusion criteria, and 9 did not undergo laboratory data collection. Among the 91 patients included in the study, 67 (73.6%) underwent carotid ultrasonography and measurement of biomarkers.
The median age of the participants was 32 (26–41) years, the majority being men who have sex with other men (MSM). Regarding comorbidities, the prevalences of hypertension, diabetes, and dyslipidemia were 16.4%, 3% and 70.1%, respectively. The median viral load was 42,033 (16,189-101,069) copies/mL and the median LTCD4+lymphocyte count was 426 (262–561) cell/mm³. The median time between diagnosis of HIV infection and initiation of antiretroviral therapy was 4.6 months, ranging from 1 to 301.7 months, as shown in Table 1.
TABLE 1: Clinical and laboratory characteristics of treatment-naïve people living with HIV, assessed in a reference center, Belo Horizonte, State of Minas Gerais, March 2014 – September 2015.
| Variables | Number=67 |
|---|---|
| Age | |
| median (interquartile range) | 32 (26; 41) |
| Gender, n (%) | |
| female | 12 (17.9) |
| male | 55 (82.1) |
| Race, n (%) | |
| white | 26 (39) |
| nonwhite | 41 (61) |
| Education, n (%) | |
| illiterate | 0 (0.0) |
| elementary/ middle school | 8 (12.0) |
| high school | 17 (25.4) |
| college | 42 (62.7) |
| Marital status, n (%) | |
| single | 54 (80.6) |
| married | 13 (19.4) |
| Alcoholism | |
| yes | 45 (67.2) |
| no | 22 (32.8) |
| Alcohol abuse (CAGE), n=45 | |
| yes | 4 (8.9) |
| no | 41 (91.1) |
| Tabagism | |
| yes | 12 (17.9) |
| no | 55 (82.1) |
| SBP (mmHg) | |
| median* | 120 (110;125) |
| DBP (mmHg) | |
| median* | 80 (70;80) |
| BP, n (%) | |
| normal, (SBP<120) and (DBP<80) | 18 (16.9) |
| prehypertension (SBP 120-139) or (DBP 80-89) | 38 (56.7) |
| hypertension (SBP >140) or (DBP >90) | 11 (16.4) |
| Blood glucose (mg/dL), n (%) | |
| ≤ 99 normal | 65 (97.0) |
| > 99 abnormal | 2 (3.0) |
| Triglycerides (mg/dL), n (%) | |
| < 150 normal | 47 (70.1) |
| ≥ 150 abnormal | 20 (29.9 |
| HDLc (mg/dL), n (%) | |
| male, n=55 | |
| ≥ 40 normal | 18 (32.7) |
| < 40 abnormal | 37 (67.3) |
| female, n=12 | |
| ≥ 50 normal | 3 (25) |
| < 50 abnormal | 9 (75) |
| Total cholesterol (mg/dL), n (%) | |
| < 200 normal | 53 (79.1) |
| ≥ 200 abnormal | 14 (20.9) |
| LDLc (mg/dL) n (%) | |
| < 160 normal | 61 (91.0) |
| ≥ 160 abnornal | 6 (9.0) |
| Viral load (copies/mL), n (%) | |
| < 50 | 0 (0.0) |
| 50 -1,000 | 5 (7.5) |
| 1,000-100,000 | 45 (67.2) |
| > 100,000 | 17 (25.4) |
| Viral load | |
| median | 42,033 (16,189-101,069) |
| CD4 count (cells/mm³) n (%) | |
| ≤ 200 | 10 (14.9) |
| 201-350 | 14 (20.9) |
| 351-500 | 19 (28.4) |
| > 500 | 24 (35.8) |
| CD4 | |
| median | 426 (262-561) |
| Time between diagnosis and initiation of treatment (months) | |
| median* | 4.60 (2.10;10.80) |
| BMI (kg/m²) | |
| median* | 24.00 (21.67;26.85) |
| BMI classification, n (%) | |
| < 18.5 underweight | 1 (1.5) |
| 18.5-24.9 normal | 46 (68.7) |
| ≥ 25 overweight/obesity | 20 (29.8) |
CAGE: the CAGE questionnaire; SBP: systolic blood pressure; DBP: diastolic blood pressure; BP: blood pressure; HDLc: high density cholesterol; LDLc: low density cholesterols; CD4: lymphocytes CD4+; BMI: body mass index. *Interquartile range.
The mean CIMT was 494.08 (± 96.84mm). The mean vascular age was 33.2 ± 18.9 years, one year longer than the chronological age (32 years), but with no statistical significance. The majority of participants had a low cardiovascular risk (94%) as measured by the Framingham score (Table 2). According to the consensus, individuals with CIMT values greater than or equal to the 75th percentile are considered as high risk and indicate an increased CVD risk23. Thus, after the reclassification, considering the CIMT percentiles, 13 (19.4%) patients had medium/high cardiovascular risk, and 54 (80.6%) patients had low cardiovascular risk. Considering the CIMT measure for the estimation of cardiovascular risk was statistically significant (p = 0.02).
TABLE 2: Carotid ultrasonography findings and Framingham score of people living with HIV before antiretroviral therapy treatment in a reference center, Belo Horizonte, State of Minas Gerais, March 2014 – September 2015.
| Variables | Number=67 |
|---|---|
| Vascular age calculator | |
| mean±SD | 33.21±18.87 |
| Difference between chronological age and VAC | |
| mean±SD | -0.01±15.05 |
| Mean CIMT | |
| mean±SD | 494.08±96.84 |
| CIMT percentile according to standard table n (%) (Stein et al., 2008) | |
| <2.5 | 0 (0.0) |
| 2.5-10 | 5 (7.5) |
| 10.1-25 | 8 (11.9) |
| 25.1-50 | 22 (32.8) |
| 50.1-75 | 21 (31.3) |
| 75.1-90 | 4 (6.0) |
| 90.1-97.5 | 6 (9.0) |
| >97.5 | 1 (1.5) |
| Framingham risk rating | |
| low | 63 (94.0) |
| moderate | 1 (1.5) |
| high | 3 (4.5) |
| Framingham risk reclassification of individuals with CIMT >75 percentile according to the consensual recommendation (Stein et al., 2008)* | |
| low | 54 (80.6) |
| moderate | 9 (13.4) |
| high | 4 (6.0) |
HIV: human immunodeficiency virus; SD: standard deviation; VAC: vascular age calculator; CIMT: carotid intima-media thickness. * p=0.0198.
ROC curve analysis showed that body mass index was the only predictor of higher cardiovascular risk (p = 0.03) (Table 3). D-dimer and p-selectin presented predictive tendencies (p = 0.06 and 0.06, respectively). D-dimer showed a significant correlation with viral load level, (r = -0.389, p <0.01). P-selectin was uncorrelated to the intima-media thickness, and without statistical significance, i.e. β = -5.57 (95% confidence interval: -12.31-1.18), p = 0.10. There was an association between measured thickness and cardiovascular risk and age.
| Variable | AUC | 95% CI of AUC | p-value |
|---|---|---|---|
| BMI | 0.69 | 0.52; 0.86 | 0.03 |
| VL | 0.64 | 0.49; 0.80 | 0.11 |
| sVCAM-1 | 0.51 | 0.35; 0.67 | 0.89 |
| ADAMTS13 | 0.55 | 0.37; 0.73 | 0.60 |
| D-Dimer | 0.33 | 0.18; 0.48 | 0.06 |
| GDF15 | 0.44 | 0.28; 0.61 | 0.53 |
| sICAM | 0.37 | 0.22; 0.52 | 0.16 |
| MPO | 0.64 | 0.48; 0.80 | 0.11 |
| P-Selectin | 0.33 | 0.17; 0.49 | 0.06 |
| Lipocalin-2NGAL | 0.50 | 0.31; 0.70 | 0.99 |
| SAA | 0.53 | 0.35; 0.71 | 0.74 |
| Myoglobin | 0.53 | 0.35; 0.72 | 0.71 |
BMI: body mass index; VL: viral load; 95% CI: confidence interval 95%; AUC: area under the curve; CI: confidence interval; sVCAM-1: vascular adhesion molecule1; ADAMTS13: Disintegrin and Metalloproteinase with Thrombospondin motifs; GDF15: growth differentiation factor 15; sICAM: soluble molecule of intercellular adhesion; MPO: myeloperoxidase; 2NGAL: vascular growth factors angiopoietin; SAA: serum amyloid A.
DISCUSSION
The majority of participants were male and MSM with a mean age of 32 years. Being a young population, there were low prevalences of arterial hypertension and diabetes mellitus. However, the prevalence of smoking was higher than that reported in Brazilian national surveys40–43. In addition, we found a high (70%) prevalence of dyslipidemia, mainly with high density lipoprotein cholesterol reduction, despite the young population and prior ART44, which can be explained, among other factors by the alteration in the maturation of this lipoprotein caused by the direct action of HIV45,46. All efforts should be made to promote health and disease prevention in this population, aiming to reduce the classical factors of chronic-degenerative diseases, mainly smoking cessation. Regarding the immunological statuses of the patients, the majority had LTCD4+ counts above 400 cells/mm³, considered adequate and higher than those reported in another study in Brazil46,47. There was no association between biomarkers and cardiovascular risk.
There was an association between the measured thickness, the cardiovascular risk, and the age. Age showed a positive correlation with the intima-media thickness; older patients had thicker intima-media. This result corroborates findings from the literature on risk factors of mean CIMT25.
The sample size may have been insufficient to show the relationship between biomarker changes and other classic cardiovascular risk factors, such as age, tabagism and immunological status, as evidenced in the literature7,12,32.
In this cross-sectional study, associated factors and effects were evaluated simultaneously. Therefore, this study only allows the establishment of associations between events, and it is not possible to define causality. No sedentary lifestyle was evaluated in this population, and no association could be made between physical activity and cardiovascular risk in the described population. Another limitation of the study is the absence of a control group of healthy participants in order to compare serum levels of studied biomarkers, due to the absence of reference values in the literature.
In this study, we noted a high (70.4%) prevalence of dyslipidemia in relation (60.3%) to the general Brazilian population. This level of change was found prior to the individual’s exposure to ART and tended to increase significantly with the initiation of therapy, since most ART negatively modify the user’s long-term lipid profile, gradually increasing cardiovascular risk over time.
There was no association between cardiovascular biomarkers and carotid atherosclerosis, similar to other studies that did not find such an association48. We identified early changes in the intima-media thickness and the increase in the vascular age calculated in the majority of participants, similar to other studies49. This result points to the role of HIV itself in the pathogenesis of atherosclerotic disease since the population studied was young, most of them were without classical risk factors, and were all without previous ART exposure. Prospective studies are needed to confirm this tendency. The intima-media thickness was directly associated with the age of the participants, a finding that corroborates those of the current literature, and shows that age is a risk factor for the increase in intima media thickness. The intima-media thickness seems to be a good marker for subclinical arteriosclerosis in the HIV population, as it is already in the general population. We suggest that prospective studies evaluate the effectiveness of inclusion of this variable in the care protocols for PLWHIV.
