- Top of page
- Materials and methods
Non-AIDS-related morbidities such as hypertension, cardiovascular disease (CVD), malignancy, and renal, liver and bone diseases have emerged as increasing clinical problems in HIV-infected patients . In fact, non-AIDS-related mortality today exceeds AIDS-related mortality in populations with access to antiretroviral therapy (ART) . A premature ageing process has been suggested to occur in HIV-infected individuals for which several contributing factors have been proposed, including viral replication, drug toxicity, lifestyle factors, and persistent immune activation with increased cell proliferation and apoptosis as well as elevated levels of pro-inflammatory markers .
Primary HIV infection is characterized by massive T-cell depletion in the gastrointestinal mucosa with subsequent enhanced translocation of bacterial products such as lipopolysaccharide (LPS) and flagellin from the intestinal lumen into the systemic circulation [3, 4]. LPS is a potent inducer of immune response and inflammation through the innate immune system. Soluble CD14 (sCD14) is a marker of monocyte activation and is shed from monocytes upon LPS stimulation . Microbial translocation has been suggested to be a major driver of HIV-associated immune activation through stimulation of Toll-like receptors (TLRs) . Measures of T-cell-related immune activation independently predict disease progression and mortality in patients receiving and not receiving ART [6, 7]. Moreover, gut epithelial barrier dysfunction  as well as high levels of sCD14  predicts mortality in HIV infection. Immune activation and microbial translocation are reduced, but often not normalized despite prolonged effective ART with achievement of viral suppression [4, 10-12]. Importantly, increased cardiovascular risk has been linked to lack of CD4 cell count restoration despite effective ART [13, 14], which in turn is associated with low nadir CD4 cell counts , persistent microbial translocation  and immune activation .
Hypertension occurs frequently in HIV-infected populations, and is a major cause of myocardial infarction , non-AIDS-related mortality , and cardiovascular and all-cause mortality . In the large D:A:D (Data Collection on Adverse Effects of Anti-HIV Drugs) study, predictors of new-onset hypertension such as older age, male gender and higher body mass index (BMI) were similar to those in the general population, but markers related to immune activation were not included in this study . Other studies have shown an association between low nadir CD4 cell counts and the development of hypertension after initiation of ART [19, 20], although the mechanisms have not been addressed. Whether immune activation and microbial translocation play a role in the relationship between immunodeficiency and hypertension is not known. Our group has previously identified low nadir CD4 cell count (i.e. <50 cells/μL) as a predictor of sustained hypertension in the present HIV-infected cohort , and in this substudy we wanted to explore microbial translocation as a possible pathogenetic link. Thus, we chose to include patients with previous severe immunodeficiency as well as patients with preserved immunocompetence throughout their ART-naïve course of HIV infection. The aim of the present study was to test the hypothesis that markers of microbial translocation and subsequent monocyte activation before initiation of ART could predict hypertension in HIV-infected individuals.
- Top of page
- Materials and methods
Characteristics of the controls and the HIV-infected patients at T0, including stratification of the 16 HT and 26 NT patients at T0, are given in Table 1. The HIV-infected study population included 19% non-Caucasians, and they had higher triglyceride levels, lower cholesterol and lower BMI compared with the Caucasian controls. All patients were ART-naïve at T0, whereas 50% and 64% were receiving ART at the time of inclusion at T1 and T2, respectively.
In the control group, LPS and sCD14 plasma levels did not correlate with age, BMI, BP or triglycerides (data not shown). Plasma levels of LPS correlated with sCD14 in the HIV-infected population (Pearson's r=0.48; P = 0.001), but not in the control group. In the HIV-infected study population, LPS and sCD14 were both negatively correlated with nadir CD4 cell count (r = −0.57; P < 0.001 and r = −0.66; P < 0.001, respectively). A diagnosis of AIDS within 1 year of nadir occurred in two-thirds of the lowN patients.
Elevated plasma levels of LPS and sCD14 in HIV-infected subjects with hypertension
Plasma levels of LPS and sCD14 were strongly correlated in patients with HT (Spearman's rho=0.62; P = 0.011), but not in patients with NT or controls (Fig. 2a). Plasma levels of LPS (P < 0.001) and sCD14 (P = 0.024) at T0 were elevated in patients with HT compared with those with NT, and in both HT and NT patients compared with controls (P < 0.001 for all) (Fig. 2b and c). Furthermore, there was a stepwise increase in the number of HT subjects through tertiles of LPS (P = 0.001) and sCD14 (P = 0.007) (Fig. 3).
Figure 2. Within-group correlations and between-group comparisons of plasma levels of lipopolysaccharide (LPS) and soluble CD14 (sCD14). (a) Correlations between plasma levels of LPS and sCD14 at the time of nadir in HIV-infected patients with hypertension (HT), HIV-infected patients with normotension (NT) and controls. (b, c) Plasma levels of LPS and sCD14 are elevated in patients with HT compared with NT patients, and in HT and NT patients compared with controls. Horizontal lines in dot plots indicate the median level.
Download figure to PowerPoint
Figure 3. Increase in the number of hypertensive HIV-infected patients through tertiles of lipopolysaccharide (LPS) and soluble CD14 (sCD14). P-values refer to χ2 linear-by-linear association.
Download figure to PowerPoint
LPS and sCD14 as independent predictors of elevated blood pressure
In univariate analyses, plasma LPS correlated with MAP at T1 and T2 (r = 0.57; P < 0.001 and r = 0.48; P = 0.001, respectively) (Fig. 4), with SBP at T1 and T2 (r = 0.52; P < 0.001 and r = 0.43; P = 0.004, respectively), and with DBP at T1 and T2 (r = 0.57; P < 0.001 and r = 0.46; P = 0.002, respectively). Furthermore, sCD14 correlated with MAP at T1 and T2 (r = 0.39; P = 0.012 and r = 0.38; P = 0.013, respectively) (Fig. 4) as well as with DBP at T1 and T2 (r = 0.45; P = 0.003 and r = 0.41; P = 0.007, respectively), but not significantly with SBP either at T1 (r = 0.27; P = 0.083) or T2 (r = 0.29; P = 0.060). Univariate linear regression analyses for MAP at T1 are presented in Table 2.
Figure 4. Correlations between plasma levels of log lipopolysaccharide (LPS) and log soluble CD14 (sCD14), respectively, at the time of nadir and mean arterial pressure (MAP) at time period 1 (T1).
Download figure to PowerPoint
Table 2. Univariate and multivariate linear regression analyses with mean blood pressure at time period 1 as outcome; also presented are adjusted β coefficients for lipopolysaccharide (LPS) and soluble CD14 (sCD14), respectively, when a fourth variable was added to the multivariate regression model
|Characteristics at T0||Unadjusted unstandardized coefficient β [95% CI]||t||p||Model with LPS||t||p||Model with sCD14||t||p|
|Adjusted unstandardized coefficient β [95% CI]||Adjusted unstandardized coefficient β [95% CI]|
|LPS (+10 pg/mL)||1.05 [0.52–1.57]||4.03||<0.001||0.86 [0.31–1.41]||3.17||0.003|| || || |
|sCD14 (+100 ng/ml)||0.40 [0.00–0.80]||2.17||0.036|| || || ||0.40 [0.10–0.80]||2.58||0.014|
|Age (+1 year)||0.45 [0.11–0.78]||2.65||0.012||0.29 [−0.03–0.61]||1.84||0.073||0.36 [0.037–0.69]||2.26||0.030|
|Male gender||8.43 [0.33–16.54]||2.10||0.042||2.56 [−5.05–10.17]||0.68||0.500||7.03 [−0.54–14.59]||1.88||0.068|
| || || || ||Basic model with additional adjustment for each of the following variables|
| || || || ||Adjusted unstandardized coefficient β [95% CI] for LPS (+10 pg/mL)||t||p||Adjusted unstandardized coefficient β [95% CI] for sCD14 (+100 ng/ml)||t||p|
|Caucasian ethnicity||8.08 [−0.45–16.64]||1.91||0.063||0.96 [0.40–1.53]||3.46||0.001||0.40 [0.10–0.80]||2.56||0.015|
|Current smokinga||2.46 [−4.51–9.44]||0.71||0.479||0.88 [0.32–1.44]||3.20||0.003||0.40 [0.10–0.80]||2.53||0.016|
|BMI (+1 kg/m2)||1.01 [−0.19–2.21]||1.70||0.097||1.08 [0.43–1.73]||3.36||0.002||0.50 [0.10–0.80]||2.80||0.008|
|Triglycerides (+1 mmol/L)||3.39 [−1.14–9.91]||1.51||0.138||0.86 [0.24–1.48]||2.82||0.008||0.40 [0.10–0.70]||2.41||0.021|
|eGFR (+1 ml/min)||−0.19 [−0.39–0.01]||−1.95||0.058||0.78 [0.21–1.34]||2.78||0.009||0.50 [0.10–0.80]||2.84||0.007|
|Glucose (+1 mmol/L)||2.93 [−2.91–8.77]||1.01||0.317||0.85 [0.28–1.41]||3.04||0.004||0.40 [0.10–0.70]||2.54||0.015|
|Current ARTa||2.63 [−4.33–9.59]||0.76||0.449||0.91 [0.24–1.58]||2.76||0.009||0.40 [0.00–0.80]||2.10||0.042|
|Log HIV RNA (+1 copies/mL)||−0.76 [−3.53–2.01]||0.56||0.583||1.06 [0.47–1.65]||3.65||0.001||0.50 [0.10–0.90]||2.79||0.009|
|Nadir CD4 count (+10 cells/μL)||−0.05 [−0.16–0.07]||−0.79||0.433||0.92 [0.30–1.55]||2.98||0.005||0.50 [0.10–0.90]||2.41||0.021|
|Nadir CD4 count <50 cells/μL||−4.0 [−11.19–3.23]||−1.12||0.271||1.05 [0.32–1.79]||2.91||0.006||0.40 [0.10–0.80]||2.58||0.014|
Both LPS and sCD14 remained independent predictors of MAP at T1 after adjustment for age and gender (Table 2). Moreover, for each 10 pg/ml increase in LPS (range 66–272 pg/ml), the increment in BP at T1 was 1.03 [95% confidence interval (CI) 0.23–1.83] mmHg for SBP (P = 0.013) and 0.78 (95% CI 0.29–1.26) mmHg for DBP (P = 0.002) after adjustment for age and gender. For each 100 ng/ml increase in sCD14 (range 135–4218 ng/ml), the increment was 0.40 (95% CI 0.10–0.70) mmHg for DBP (P = 0.004) after adjustment, but with no independent association with SBP (P = 0.101). Similarly, for each 10 pg/ml increase in LPS, the increment in BP at T2 was 0.75 (95% CI 0.20–1.29) mmHg for MAP (P = 0.009), 0.97 (95% CI 0.15–1.80) mmHg for SBP (P = 0.022) and 0.63 (95% CI 0.15–1.11) mmHg for DBP (P = 0.011) after adjustment for age and gender. For each 100 ng/ml increase in sCD14, the increment in BP at T2 was 0.40 (95% CI 0.10–0.70) mmHg for both MAP and DBP (P = 0.008 and P = 0.007, respectively) and 0.50 (95% CI 0.00–1.00) mmHg for SBP (P = 0.036) after adjustment for age and gender. The multivariate regression models including LPS and sCD14 yielded R2 of 0.17–0.29 and 0.19–0.28, respectively.
When a fourth variable was added to the multivariate regression model, i.e. ethnicity, viral load, nadir CD4 cell count, triglyceride concentration, BMI, eGFR, glucose concentration, smoking or current ART, plasma levels of LPS and sCD14 remained independent predictors of MAP at both T1 (Table 2) and T2. In addition, LPS independently predicted SBP and DBP, whereas sCD14 independently predicted DBP but not SBP at T1 and T2 (data not shown).
- Top of page
- Materials and methods
In the present study, LPS and sCD14 levels were measured at the time of nadir CD4 cell count, selecting both patients with advanced immunodeficiency and preserved immunocompetence at the time of nadir. LPS and sCD14 levels were both independent predictors of later BP levels and were associated with subsequent sustained hypertension. These findings could not be explained by cardiometabolic risk factors, viral load, nadir CD4 cell count or later use of ART. Our data suggest that microbial translocation and immune activation might contribute to increased BP and the development of hypertension.
The significant association between LPS and hypertension found in this study can be explained by several mechanisms. First, LPS is well known to induce endothelial dysfunction , among other mechanisms by activating the endothelium through a receptor complex consisting of TLR4, CD14 and myeloid differentiation (MD)-2 , and by inducing apoptosis of endothelial cells . Endothelial dysfunction and impaired arterial elasticity are present throughout the atherosclerotic process , and an association of LPS with carotid atherosclerosis has been described in the general population  as well as in HIV-infected individuals . In non-HIV-infected cohorts, we and others have shown that pro-inflammatory cytokines  and sCD14 are associated with arterial stiffness  which is mechanistically linked to hypertension.
In HIV-infected individuals, viral replication has been associated with endothelial dysfunction . Furthermore, increased LPS levels are associated with activated T cells , consistent with a recent study reporting that a predominance of such cells was associated with arterial stiffness in HIV-infected women . Taken together, these findings suggest that microbial translocation and subsequent activation of macrophages and T cells may contribute to HIV-associated endothelial dysfunction and arterial stiffness, which over time may cause hypertension.
Other mechanisms may also be involved. LPS has been shown to activate the sympathetic nervous system in experimental settings , and high autonomous nervous activity and elevated levels of noradrenaline have been linked to impaired ART response and enhanced replication of HIV . Furthermore, the renin-angiotensin system is involved in LPS-induced endothelial dysfunction, and can be reversed by the angiotensin-converting enzyme-inhibitor enalapril .
Nadir CD4 cell count was associated with LPS and sCD14 in our study. In fact, microbial translocation could explain the reported independent association between low nadir CD4 cell count and the development of hypertension [19-21]. However, other mechanisms could also be at play. Hypertension has previously been related to inflammation [37, 38]. Immune activation is linked to HIV progression, and could be driven by factors other than microbial translocation, for example viral coinfections, other opportunistic infections and amplified HIV replication during immunodeficiency . Dyslipidaemia during advanced HIV infection  could have an adverse effect on endothelial cells, as could viral toxicity . Furthermore, the combination of severe immunodeficiency and subsequent ART-mediated immune reconstitution could possibly have an adverse influence on BP [19, 20].
There are limitations in this study that need to be acknowledged. First, the small sample size increases the risk of statistical type II errors. However, the risk of type I errors is lower, and the significant association between microbial translocation and elevated BP levels is likely to be reliable. Secondly, the study was not originally designed for the purpose of exploring a link between microbial translocation and hypertension. Consequently, the study lacks systematic BP measurements at T0. Because of limited availability of plasma samples in all groups except the NT highN group, the HT and NT patients could not be matched across the two nadir strata, either in number or in other characteristics. However, adjusting for relevant covariates in linear regression analyses did not alter the main result. Thirdly, measurement of LPS has many pitfalls, including technical difficulties and possible interference from plasma turbidity . A fasting state was advised but not formally recorded, and thus post-prandial hypertriglyceridaemia could have influenced the LPS levels, in particular in samples with elevated triglycerides. Furthermore, storage of plasma samples over a long period of time could possibly have an adverse influence. Fourthly, the controls were recruited with the purpose of adding context to the data. Because of the lack of matching between the HIV-infected cohort and controls, the comparisons should be interpreted with caution. Finally, although all samples were run in duplicate, inter- and intra-assay variability cannot be excluded.
Our study also has several strengths. The close association between LPS and BP was supported by the finding of a similar predictive power for sCD14, which is pathogenetically linked to LPS . The longitudinal design in the assessment of hypertension status, the thorough validation of BP measurements, and the use of standardized BP readings and sustained hypertension as predefined endpoints are additional strengths.
The majority of patients with a low nadir CD4 cell count reached nadir at the time-point at which ART was introduced (1996). Thus, the study population does not reflect the situation today, and this study should therefore be considered a proof of concept study. Prospective studies should be performed in populations on stable ART in which non-AIDS-related complications are an increasing clinical problem .
Hypertension is a non-AIDS-related complication which deserves particular clinical attention. As our data show that hypertension was closely associated with markers of microbial translocation in HIV-infected individuals, a key aspect of HIV pathogenesis may be linked to the development of hypertension. We propose that the gastrointestinal mucosal barrier may present as a potential therapeutic target for the prevention of future hypertension and long-term cardiovascular complications in HIV infection, and this needs to be explored in future studies.