Increased Carotid Intima-Media Thickness Associated with Antibody Responses to Varicella-Zoster Virus and Cytomegalovirus in HIV-Infected Patients

Objective We investigated the relationship of the Herpesviridiae with inflammation and subclinical atherosclerosis in HIV-infected patients. Methods Prospective study including virologically suppressed HIV-infected patients. IgG antibodies against herpesviruses, carotid intima-media thickness (cIMT), endothelial function through flow-mediated dilatation (FMD) of the brachial artery, and blood atherosclerosis biomarkers (hsCRP, TNF-α, IL-6, MCP-1, MDA, sCD14, sCD163, VCAM-1, ICAM-1, D-dimer, and PAI-1) were measured. Results 136 patients with HIV viral load <200 copies/ml were included. 93.4% patients were infected with herpes simplex virus type-1, 55.9% with herpes simplex virus type-2, 97.1% with varicella-zoster virus, 65.4% with human herpesvirus-6, 91.2% with cytomegalovirus, and 99.3% with Epstein-Barr virus. Previous AIDS diagnosis was associated with higher cytomegalovirus IgG titers (23,000 vs 17,000 AU, P = 0.011) and higher varicella-zoster virus IgG titers (3.19 vs 2.88 AU, P = 0.047), and there was a positive correlation of the Framingham risk score with IgG levels against cytomegalovirus (Spearman's Rho 0.216, P = 0.016) and Herpes simplex virus-2 (Spearman's Rho 0.293, P = 0.001). IgG antibodies against cytomegalovirus correlated in adjusted analysis with the cIMT (P = 0.030). High seropositivity for varicella-zoster virus (OR 2.91, 95% CI 1.05–8.01, P = 0.039), and for cytomegalovirus (OR 3.79, 95% CI 1.20–11.97, P = 0.023) were predictors for the highest quartile of the cIMT in adjusted analyses. PAI-1 levels were independently associated with cytomegalovirus IgG titers (P = 0.041), IL-6 and ICAM-1 levels with varicella-zoster virus IgG (P = 0.046 and P = 0.035 respectively), and hsCRP levels with Herpes simplex virus-2 IgG (P = 0.035). Conclusion In virologically suppressed HIV-infected patients, antibody responses against herpesviruses are associated with subclinical atherosclerosis, and with increased inflammation and coagulation biomarkers.


Introduction
Cardiovascular disease has emerged as an important cause of morbidity and mortality in HIV-infected patients. Among the involved causes, in recent years systemic inflammation and immune activation have gained attention as central factors in the pathogenesis of HIV-related atherosclerosis [1]. While HIV replication has been considered the major trigger of the immune system, persistent inflammation and immune activation have also been found in patients receiving effective antiretroviral therapy (ART) [2], and in elite controllers [3]. This suggests that additional causes may elicit immune activation and might therefore be involved in the accelerated course of atherosclerotic disease in virologically controlled HIV-infected patients [3].
There is accumulating evidence that certain infectious agents, like Herpesviridiae, play a role in the pathogenesis of atherosclerosis [4][5][6]. In HIV-infected patients, a relationship with subclinical atherosclerosis has been reported for cytomegalovirus and herpes simplex virus type-2 (HSV-2) [7,8,9]. To assess the role of further members of the herpesvirus family, we investigated their relationship with carotid intima-media thickness (cIMT), endothelial function through flow-mediated dilatation (FMD) of the brachial artery, and several cardiovascular biomarkers in virologically suppressed HIV-infected patients.

Methods
We conducted a prospective study including consecutive HIVinfected patients at increased risk for herpesvirus infection cared for in an HIV outpatient clinic. Accordingly, patients who acquired the HIV through sexual transmission were invited to participate in the study. To avoid the confounding effect of HIV replication, only patients with virological suppression, defined as a viral load ,200 copies/ml, were included in the study.

Ethics statement
Patients who agreed signed an informed consent before inclusion. The local ethics committee (Comité É tico del Hospital General Universitario de Elche) approved the protocol. The study protocol was in accordance with the Declaration of Helsinki of good clinical practice guidelines

Evaluation of the endothelial function and the carotid intima-media thickness
Endothelial function was evaluated by measuring FMD of the brachial artery as detailed elsewhere [11]. Measurements were performed by an experienced examiner blinded to the subject's information. Reproducibility was assessed by the same examiner in 10 healthy subjects (8 men, age 4468.5 years) who had FMD measured twice, at an interval of 2 hours. Median (interquartile range [IQR]) FMD was 8.15% (5.26-12.1%). Median (IQR) intraobserver intersession percentage of variation for brachial artery diameter was 2.38% (0-5.71), within the reported range. cIMT measurement was performed as previously described [12]. Measures were taken from both common carotids and bulb portions. Total cIMT was calculated as the mean of all measurements, and analyzed both as a continuous variable, and categorized in quartiles.

Statistical analyses
Bivariate Spearman's correlation and Kruskal-Wallis tests were used to assess the association of IgG antibodies against the herpesviruses with the cIMT, FMD and blood biomarkers. Multivariable linear regression and logistic regression analyses were performed to adjust the models for the variables found to be associated in univariate analyses: Framingham risk score, HCV, and previous AIDS diagnosis. To be included in regression analyses, all continuous variables were transformed on the natural log scale (log e ) due to their highly skewed distribution.
Quantitative and qualitative serology tests results are shown in Table 1. Previous AIDS diagnosis was associated with higher IgG anti-cytomegalovirus titers (23,000 vs 17,000 AU, P = 0.011) and higher IgG anti-varicella-zoster virus titers (3.19 vs 2.88 AU, P = 0.047). There was a positive correlation of the Framingham risk score with IgG levels against cytomegalovirus and HSV-2 ( Table 2). Hepatitis C was associated with lower IgG titers against cytomegalovirus (13,000 vs 19,000 AU, P = 0.047), HHV-6 (1.21 vs 1.76 AU, P = 0.019) and a trend to lower titers against HSV-2 (0.27 vs 8.61 AU, P = 0.058). No differences in IgG levels were found according to hepatitis B serostatus or antiretroviral therapy composition. No correlation was found between CD4 cells and antibody levels against any of the herpesviruses (data not shown).

Carotid intima-media thickness
There was a significant positive correlation between total cIMT and IgG antibodies against cytomegalovirus (P = 0.004), and a trend to a correlation with varicella-zoster virus (P = 0.070) and HSV-2 (P = 0.071) IgG levels ( Table 2). IgG cytomegalovirus antibodies also correlated with the cIMT at all locations measured: right common cIMT (Spearman's rho 0.241, P = 0.008), left common cIMT (Spearman's rho 0.197, P = 0.031), right bulb (Spearman's rho 0.199, P = 0.034), and left bulb (Spearman's rho 0.236, P = 0.012). VZV IgG antibodies correlated with the right bulb cIMT (Spearman's rho, 0.198, P = 0.029). Linear regression analyses adjusted for the Framingham score, HCV, and previous AIDS diagnosis, showed a persistent association of IgG cytomegalovirus antibodies with total cIMT (P = 0.030) and with the cIMT at different locations (P = 0.027 and P = 0.012 for left bulb and right common cIMT, and P = 0.087 for left common cIMT).
When total cIMT was categorized in quartiles, a significant increase of IgG antibodies against cytomegalovirus (P KW = 0.003), and against varicella-zoster virus (P KW 0.028) were found with increasing total cIMT quartiles ( Figure 1). For HSV-2, there was a significant difference in IgG antibody levels between the 1 st and 3 rd total cIMT quartiles (P = 0.016). No other significant associations were found for the other herpesviruses analyzed. The highest cIMT quartile ($0.98 mm) was selected to define subclinical atherosclerosis, and IgG antibody levels above the median to define high seropositivity. In multivariate logistic regression analyses adjusted for the Framingham score, HCV, and previous  Table 3 for the biomarkers associated with IgG antiviral responses against cytomegalovirus or varicella-zoster virus in univariate or multivariate analyses, showed a small effect size in the association between the cIMT and the antibody responses (Table 3).
IgG antibody titers against herpesviruses tended to be higher in patients with carotid plaques, except for HHV-6, but the difference only reached statistical significance for cytomegalovirus: median (IQR) IgG 17,000 (11,500) AU in patients with plaques vs 14,000 (3115-19,500) AU in patients without plaques, P = 0.007. Plasma biomarkers hsCRP levels correlated with HSV-2 (P = 0.002) and cytomegalovirus (P = 0.03) antibody titers ( Table 2). After adjusted linear regression, only the association between hsCRP levels and HSV-2 antibody titers continued to be significant (P = 0.035). The biomarkers IL-6 and ICAM-1 correlated with IgG antibodies against varicella-zoster virus (P,0.001 for both), and this correlation was confirmed after adjustment (P = 0.046 and P = 0.035 respectively). Cytomegalovirus IgG antibody levels also correlated with PAI-1 in unadjusted (Spearman's rho 0.182, P = 0.042) and adjusted analyses (P = 0.041).

Endothelial function
No statistically significant correlations were found between FMD of the brachial artery values and IgG antibody levels for any of the herpesviruses studied.

Discussion
We explored the relationship of several Herpesviridiae with subclinical atherosclerosis measured with the cIMT and FMD, and different biomarkers of inflammation, endothelial activation, coagulation, and oxidative stress in HIV-infected patients. To date, no such extensive Herpesviridiae evaluation had been carried out to investigate their association with so many different surrogate markers of atherosclerosis in virologically-suppressed patients. Our study shows a significant and robust association between IgG antibody response against cytomegalovirus and cIMT measurements. A high seropositivity for varicella-zoster virus and for cytomegalovirus were associated with subclinical atherosclerosis. Infection with those viruses was also accompanied by a concomitant elevation of inflammation and coagulation biomarkers. Finally, anti-HSV-2 titers were associated with inflammation measured with hsCRP, but the relationship with cIMT was not confirmed in adjusted analysis. Cellular specific anti-cytomegalovirus response had been found to be associated with the cIMT in HIV-infected patients [7]. Our results show that, in addition, a high humoral anti-cytomegalovirus response, which might reflect either an increased inflammatory individual response or the persistence of a higher load of the virus, is a predictor of increased cIMT. The higher cytomegalovirus antibody levels in patients with carotid plaques reinforce its relationship with carotid atherosclerosis. Findings from this study support those described in the general population [13,14]. In HIVinfected patients, cytomegalovirus antibody titers have been related to decreased artery distensibility and carotid lesions in women, but no association was found with the cIMT measurements [8]. Apart from sex, the majority of African American/ black and of viremic women in that study could have contributed to explain differences with ours. Interestingly, cytomegalovirus serologic response was accompanied by elevated PAI-1 levels. In vitro data support a loss of anticoagulant and the acquisition of procoagulant properties in endothelial cells infected with herpesviruses, including cytomegalovirus, which can also indirectly induce a prothrombotic state by increasing binding sites for inflammatory cells [5].
Varicella-zoster virus IgG antibodies were not as firmly correlated with the cIMT as cytomegalovirus antibodies were. However, a high serological response to varicella-zoster virus was associated with subclinical atherosclerosis. To the best of our knowledge, the relationship of this prevalent herpesvirus with atherosclerosis had not been previously described in HIV or in immunocompetent patients. This association remained after adjusting for serological response to cytomegalovirus, suggesting that it was independent from cytomegalovirus immune effects, though there may be numerous common mechanisms involved in atherosclerosis interacting each other. Additional data corroborating this finding are warranted, as well as the elevation of IL-6 and ICAM-1 accompanying varicella-zoster virus serological response, and their potential contributing role in atherogenesis. The small effect size of these two biomarkers in the model for the association of varicella-zoster virus with the cIMT, and also of the biomarkers associated with cytomegalovirus, reflects the multiple pathways contributing to atherogenesis, and probably that the antibody response to the viruses comprises several of such proatherogenic mechanisms, and therefore the association with atherosclerosis may be stronger than that of any biomarker alone.
No independent relationship between seropositivity to HSV-2 and the cIMT was demonstrated in our study, although higher coronary calcium levels have been described in HSV-2-infected patients [9]. Previous studies in HIV-negative patients could neither demonstrate this relationship [14]. Nevertheless, HSV-2 seropositivity was associated with hsCRP levels, and therefore its contribution to atherogenesis cannot be excluded.
Limitations of the study are the cross-sectional nature of the associations studied. Strengths are the large variety of herpesviruses evaluated and the widely validated surrogate markers of atherosclerosis studied.
In summary, in this extensive study of the Herpesviridiae infecting humans, our data suggest a relationship of high seropositivity to varicella-zoster virus and cytomegalovirus with subclinical atherosclerosis. The concomitant finding of elevated inflammation and hypercoagulability biomarkers accompanying herpesvirus infection in virologically suppressed patients suggests a possible involvement of these mechanisms as contributing pathogenic factors. Our results illustrate a plausible scenario where several potentially interrelated pathogens, and pathogens' immune and inflammatory responses coexist in HIV-infected patients with subclinical atherosclerosis.