Review of Cardiovascular Disease in HIV-Infected Women

Copyright: © 2016 Adekunle R, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.


Introduction
Individuals infected with Human Immunodeficiency Virus (HIV) are now living longer due to improved and expanded access to antiretroviral therapy (ART). However, all-cause mortality rates continue to be higher in HIV-infected patients than in the general population and there has been an increased percentage of non-AIDSdefining illnesses [1]. Coronary heart disease (CHD) represents the second cause of death among HIV-infected patients currently, and is expected to increase as the population ages [2].
Numerous studies report increased risk of CHD in HIV-infected people, but most of these studies were conducted in men or in predominantly male cohorts [3][4][5]. In the general population, men have been reported to have higher rates of CHD and at younger age than women [6,7]. The death rate from cardiovascular disease (CVD) has decreased among men, but continues to increase in women [8]. Recent statistics show that 42 percent of women who have heart attacks die within one year compared with 24 percent of men. As such, there has been increased attention in the scientific community as well as the general media surrounding missed opportunities to accurately diagnose CHD early and intervene in timely manner among women. It is unclear if the same increased rate of deaths due to CHD exist in HIV-infected women, and whether rates or risk of CHD differ between HIV-infected men and women, and between seropositive and seronegative women.
The purpose of our review was several-fold. We reviewed the literature on the rates of cardiovascular events and surrogate measures of atherosclerosis or CHD risk in HIV-infected women. We also reviewed rates of metabolic disease and other markers of inflammation and immune activation that could contribute to increased cardiovascular (CV) risk.

Methods
We searched PubMed and Embase for English-language articles published from January 1, 2005 -October 27th, 2015 using search terms "Cardiovascular disease, HIV, Women, " "Coronary heart disease, HIV, Women, " "Framingham risk and HIV and women, " "Acute MI, women, and HIV, " and "Acute myocardial infarction, HIV, and women. We also searched using terms "Carotid intimal thickness, HIV, women" and "Coronary atherosclerosis, HIV, women. " A total of 323 references were found, but 213 references were not selected because they were abstracts, poster presentations, correspondences, letters, or other report types. 76 more studies were excluded after reading the abstracts because they did not have findings specific to HIV-infected women. Four studies were removed from the review because they did not contribute any additional information than those included. Four were reviewed in the manuscript text but not included in the tables because they reported sample sizes of less than 100 participants. 12 relevant studies were included from references of selected manuscripts.
A total of 37 full-length articles were included in the final review ( Figure 1). Of note, there were two studies included in our review that reported on more than one end point, and therefore they appear in more than one table. Selected studies were presented in the tables and the remainder were described in the section corresponding to the relevant endpoint or risk factor for CHD discussed.
Study designs included one randomized clinical trial, 13 prospective observational cohorts, one retrospective analyses, 18 cross-sectional studies, and four case-control studies. Endpoints assessed were: CHD incidence and mortality, risk scores to assess risk for developing CVD, anatomical and physiological surrogate markers of CVD including carotid intimal media thickness (CIMT), pulse wave velocity (PWV), computed tomography coronary angiography (CTA) and ankle-brachial index (ABI), biomarkers of CVD, and alterations of the metabolic profile. Adjusted odds ratio (OR), hazard ratio (HR), relative risk (RR), along with associated confidence intervals (CI) and p-values were extracted and reported from all studies whenever available.

Clinical outcomes
The cardiovascular events reported in studies include acute myocardial infarction (AMI), unstable angina, ischemia heart failure, and ischemic stroke. Table 1 summarizes the six studies that reported clinical cardiovascular outcomes in HIV-infected women. In a study from the Partners HealthCare System of 3851 HIV-infected patients including 1172 women, there was RR of 2.98 (95% CI 2.33, 3.75; p<0.0001) of developing AMI for HIV-infected women compared to seronegative women whereas HIV-infected men had RR of 1.40 (95% CI 1.16, 1.67; p=0.0003) compared to seronegative men [9]. The Veterans Aging Cohort Study that included 2187 HIV-infected women demonstrated that incident CHD per 1000 person-years was significantly higher among HIV-infected women with incident rate (IR) 13.5 (95% CI 10.1, 18.1) compared to uninfected women with IR 5.3 (95% CI 3.9, 7.3; p<0.001). In addition, the median age to first CHD event was 49.3 years vs. 51.2 years for HIV-infected female veterans compared to HIV-uninfected female veterans (p=0.05) [10]. Similarly, a French study showed that the risk of MI was higher in both HIVinfected men and women, but the standardized morbidity ratio for HIV-infected women was 2.7 compared to 1.4 for HIV-infected men [11]. Chow et al. demonstrated that HIV-infected women have higher rates of ischemic stroke compared with HIV-infected men with HR 2.16 (95% CI 1.53, 3.04; p<0.001) for women vs. HR 1.18 (95% CI 0.95, 1.47; p=0.14) for men [12]. These studies confirm that HIV-infected women have increased rates of CHD compared to HIV-uninfected women, and in some instances may carry a greater risk of CHD than HIV-infected males.

Surrogate measures of cardiovascular disease
There are several risk assessment tools and non-invasive anatomical and physiological surrogate measures of CVD developed to assess the risk of developing CHD in the general population [13]. There were 28 studies reporting on risk scores to assess CVD or at least one surrogate

Design N Study Population Outcome Follow up (months) Results
Triant et al. [9].  measure of CVD in HIV-infected women. Table 2 lists selected studies that used these risk assessment tools and surrogate measures of subclinical CVD in the HIV population.
Risk scores to assess CVD risk: There were 11 studies included that reported on the Framingham risk scores (FRS) in HIV-infected women. The Framingham risk score has been the most common risk assessment tool used to evaluate the risk of developing CHD in HIV-infected men and women. Studies demonstrate that HIV-infected women have low Framingham risk scores [10,14,15], but there are no studies validating the FRS in this population. The Data Collection on Adverse Events of Anti-HIV Drugs (DAD) equation was derived to assess the potential increased risk of CHD in HIV-infected people using combination ART. Law et al. estimated the 3-year risk of myocardial infarction (MI) using the DAD equation in HIV-infected men and HIV-infected women and found that it was higher in men than in women, 0.92% (0.47 ± 1.42%) vs. 0.07% (0.05 ± 0.19%) [16]. Given the increased rates of CHD reported among HIV-infected women compared to HIV-infected men and to seronegative patients as described above, there is significant concern that the FRS and DAD equation may not be an accurate risk assessment tool for this patient population.

Carotid intimal media thickness (CIMT):
There were 11 studies included evaluating CIMT in HIV-infected women. Carotid Intimal Medial Thickness can be a useful marker of atherosclerosis and has been shown to correlate with CV risk factors and predict stroke and myocardial infarction in the general population [17]. Results of CIMT in HIV-infected women have been mixed with a few studies finding no significant difference between HIV-infected females and HIVuninfected females [15,18]. Increased CIMT has been associated with several CV risk factors, however. One study of 97 HIV-infected women noted that participants receiving protease inhibitors (PI) had increased CIMT compared to those not receiving PIs [18]. Another study found that in untreated HIV-infected women, there were no significant associations between lipid levels and CIMT, but in treated women higher levels of low-density lipoprotein cholesterol (LDL-c) and nonhigh density lipoprotein cholesterol (non-HDL-c) were associated with higher CIMT values [19]. Mangili et al. found that significantly more HIV-infected women than men had metabolic syndrome, and the mean common CIMT measurement was significantly higher among participants with metabolic syndrome than among those without metabolic syndrome (0.66 mm vs. 0.59 mm; p=0.005) [20]. The association between HIV infection, CV risk factors and CVD remains unclear based on the available studies in HIV-infected women.

Pulse wave velocity (PWV):
Pulse wave velocity is a surrogate marker of arterial stiffness that has been associated with CVD and  [21]. There is limited data on PWV in HIV-infected women, and only two studies reported on gender in their results. Eira et al. noted that there was a significant trend towards higher values of aortic PWV in HIV-infected women with diabetes or on ART among 18 HIV-infected women [22]. A Rwandan study of 276 HIV-infected women found no significant differences in measures of radial arterial wave reflection between HIV-infected and HIVuninfected women after adjustment for potential confounders and age [23]. More studies with larger numbers need to be conducted in order to obtain more conclusive results on the association between HIV infection and PWV in women.

Computed tomography coronary angiography (CTA):
There were three studies that evaluated CTA in HIV-infected women. Computed tomography coronary angiography has been used to evaluate coronary atherosclerosis in the general population, and its use has been promoted by recent studies [24,25]. There is a growing body of literature using the computed tomography coronary angiography (CTA) to assess subclinical CVD in HIV-infected individuals. Fitch et al. evaluated atherosclerotic plaque morphology among HIV-infected and HIV-uninfected women. The presence of coronary plaques was similar in both groups, but HIV-infected females had a significantly higher prevalence of non-calcified coronary artery plaques (35% vs 12% in female control subjects; p=0.04) which remained significant even after adjustment for CV risk factors such as age, race, FRS, smoking status, levels of triglycerides (TGs), HDL-c and LDL-c, and BMI. The percentage of coronary segments with non-calcified plaques was significantly higher in HIV-infected women (median 75% IQR 63%-100%) compared to HIV-infected men (median 50% IQR 3%-100%; p<0.05,) [26] which is important because non-calcified plaques are considered vulnerable and more susceptible to rupture leading to acute coronary syndromes [27]. Coronary artery calcium scores (CAC) have been found to be higher in HIV-infected patients with metabolic syndrome compared to those without metabolic syndrome [20]. The same study showed that HIV-infected women had higher rates of metabolic syndrome compared with HIV-infected men. Conversely, another study found that HIV-uninfected women had statistically significant higher percentage of CAC >100 (p=0.02) compared to HIV-infected women [26]. Other studies did not demonstrate gender differences in CAC [20,28]. From the limited studies of CTA in HIVinfected women, there is a suggestion that certain plaque morphologies may be greater in HIV-infected women compared to controls, but larger studies need to be performed to establish firmer conclusions.

Ankle-brachial index (ABI):
There is limited data on ABI in HIVinfected women, and only one study reported on gender in their results. Sharma et al evaluated ABI in 238 HIV-infected women and 97 HIVuninfected women. The presence of low ABI was too few for analysis (n=3, 0.9%), but it was noted that the presence of high ABI was more frequent than expected (n=23, 6.9) with an unclear clinical significance [29]. More studies with larger numbers need to be conducted to further evaluate the association between ABI and HIV infection in women and to develop meaningful clinically significant cut-off values.

Risk factors and metabolic indices
Several studies have evaluated traditional CHD risk factors, metabolic indices, and markers of inflammation in HIV infected women. Table 3 lists selected studies that reported on traditional CHD risk factors, markers of inflammation, and HAART. Table 4 lists selected studies that reported on metabolic indices that contribute to CHD risk.

Traditional risk factors
Studies report differing impact of traditional CV risk factors on CHD risk. There were six studies that evaluated traditional risk factors in HIV-infected women. Triant     but a lower prevalence of hypertension and obesity in HIV-infected women compared to HIV-uninfected women (p<0.05) [10]. Mulligan et al. found that 40% of both HIV-infected and HIV-uninfected young women were overweight or obese (BMI>25), but there was no difference between groups. On the other hand, fasting TGs and total cholesterol (TC) levels were significantly worse among HIV-infected participants than HIV-negative participants. The authors concluded that obesity and dyslipidemia were prominent among HIV-infected adolescent women, and when coupled with other CV risk factors may accelerate the lifetime risk of CHD and other adverse events [30].

Metabolic indices
There were 15 studies included that evaluated metabolic indices in HIV-infected women. Metabolic syndrome was noted to be particularly prevalent in HIV-infected women. There are several definitions for metabolic syndrome but they all include insulin resistance, obesity, atherogenic dyslipidemia (elevations in TGs or low HDL) and hypertension. Sobieszczyk et al. found that HIV infection was independently associated with metabolic syndrome with a rate of 33% vs 22% (p<0.0001) in HIV-positive women compared with HIV-negative women [31]. Janiszewski et al. investigated the association between waist circumference (WC) and TG levels and CV risk and observed that HIV-infected women with high TG and WC had higher rates of metabolic syndrome than any other group (p<0.05) [14]. Another study in intravenous drug users observed that the prevalence of metabolic syndrome was significantly higher among HIV seropositive females than among seropositive males (29% vs 10.3%, p=0.013) [32]. Mangili et al. showed that HIV-infected women had higher rates of metabolic syndrome compared to HIV-infected men, and mean common CIMT measurements were higher among those with metabolic syndrome than those without metabolic syndrome (0.66 mm vs. 0.59 mm; p=0.005) [20]. However, other studies did not consistently demonstrate increased rates of metabolic syndrome in seropositive women compared to men, nor that increased rates of metabolic syndrome was consistently associated with increased CHD risk ( Table 4).

Markers of inflammation and immune activation
There were eight studies that investigated markers of inflammation and immune activation in HIV-infected women. Conceivably, HIVinfected patients may have higher rates and increased risk of CHD because of increased levels of inflammation and immune activation than seronegative patients. This increased immune activation persists even in patients on ART who have achieved virological suppression and immune reconstitution with CD4 cell count >200 mm 3 [33]. Soluble CD163 (sCD163) is a monocyte-macrophage specific scavenger receptor cleaved from activated monocytes and macrophages during inflammation, and elevated levels have been associated with coronary artery disease in the general population and HIV-infected men suggesting this inflammatory and immune pathway may be involved in the pathogenesis of atherosclerosis [34,35]. Fitch et al. reported that sCD163 levels were significantly higher in HIV-infected females than HIV-infected and non-infected males [26]. 98% of the women were on  ART with a median duration of eight years and 84% had undetectable viral loads. Shaked et al. demonstrated elevated levels of sCD163 in HIV-infected and HIV/hepatitis C co-infected women, though 46% of total study cohort (29% of HIV+/HCV-cohort and 17% of HIV+/ HCV+ cohort) were not on ART and the median viral load was 885 copies/mL [36].
Kaplan et al. compared changes in several inflammatory markers in 127 HIV-infected women and matched control women before and after Highly Active Antiretroviral Therapy (HAART) initiation in the Women's Interagency HIV Study (WIHS). They found that HIVinfected women had statistically significant higher levels of tumor necrosis factor (TNF)-alpha, soluble interleukin (IL)-2 receptor, IL-10, monocyte chemoattractant protein (MCP)-1 and D-dimer prior to ART initiation, and though the levels of TNF-alpha and soluble IL-2 receptor decreased among HIV-infected women after ART these levels remained higher than in HIV-uninfected women. These studies demonstrating higher levels of plasma markers of inflammation and immune activation in HIV-infected women provide a potential mechanism by which these women could have increased rates of subclinical atherosclerosis and CHD [37].

Highly active antiretroviral therapy (HAART)
There were 12 studies that evaluated the association of HAART and CHD in HIV-infected women, and selected ones were listed in Table 3. Studies in predominantly male cohorts suggest that ART, particularly protease inhibitors, is an independent risk factor for increased CHD [38,39]. Abacavir (ABC) has also been implicated in increased risk of and rates of cardiovascular events though results have been mixed among studies, [5,[40][41][42][43]. These studies, however, again included predominantly male populations. In a study of 2187 women of whom 710 were HIV infected, Womack et al. found no association between ART and CVD outcomes [10]. Conversely, Chow et al. observed that longer duration of any ART regimen was associated with significantly decreased risk of strokes (HR 0.80; 95% CI 0.73, 0.88, p=<0.001) [12]. Therefore, it not clear whether ART exposure or specific antiretroviral medications (ARVs) truly increase rates of CVD in HIV-infected women.
Studies also evaluated the contribution of ART and specific ARVs to changes in traditional CHD risk factors and markers of inflammation. Estrada et al. demonstrated that HIV-infected women receiving nonnucleoside reverse transcriptase inhibitor (NNRTI)-based regimens had better lipid profiles than HIV-infected women on PI-based regimens [44]. Shaffer et al. evaluated CHD risk factors following initiation of NNRTI vs. PI-based ART in women from 7 countries in sub-Saharan Africa over 144 weeks. ART with either nevirapine or lopinavir/ ritonavir had similar lipid values with only TC increasing at 144 weeks (p=0.090) [45]. However, Tien et al. found no significant associations between ART and lipid values [46]. As such, the contribution of ARVs to lipid profiles remain uncertain.
Shikuma et al. looked at high sensitivity C-reactive protein (hsCRP) and its association with efavirenz (EFV) versus ABC over a 96 week period in HIV-infected women and men. Levels of hsCRP did not differ by gender at baseline, but by week 96 women had higher levels of hsCRP than men (median 6 mg/liter vs. median 1.6 mg/liter, p< 0.001), but no difference was noted between the EFV and ABC groups (p=0.38) [47]. Another study investigated the effects of ART on plasma markers of inflammation in HIV-infected and uninfected women and demonstrated that markers of inflammation decreased among HIV-infected women after ARV initiation, but levels of TNF-alpha and IL-2 remained elevated among HIV-infected women compared to uninfected women over time despite ART (p<0.0001 and p<0.01 respectively) [37]. It remains indeterminate to what degree plasma markers of inflammation and immune activation change with ART use and specific ARVs and whether these changes differ by gender.

Discussion
HIV-infected patients have increased rates of CHD compared to the general population, and these increased rates have been more frequently reported in HIV-infected men than in women. Freiberg et al looked at the veteran population and found an increase rate of cardiovascular events in HIV-infected male veterans compared with HIV-uninfected male veterans HR 1.48 (95% CI, 1.27-1.72) [3]. Similarly, Silverberg et al demonstrated a 44% increased risk of MIs among a predominately male HIV-infected population compared with a predominately male HIV-uninfected population [48]. Studies that included both male and female participants have mostly demonstrated that either gender was not a risk factor for development of CHD or that male rather than female gender was a risk factor for CHD in HIVinfected patients [9,[11][12]32,49,50]. Differences in gender as a risk factor for CHD among studies in HIV-infected patients may be due to differences in study population as well as the vast majority of studies in which women were underrepresented. What this review highlights though is that HIV-infected women do have increased rates of AMI and ischemic stroke compared to HIV-uninfected women, and may also be increased compared to HIV-infected men as observed in some studies despite women being projected to have lower CHD risk based on FRS.
Studies assessing CHD risk by measuring anatomical or physiological measures of subclinical atherosclerosis have reported mixed results. Of these, studies using CTA have consistently demonstrated increased coronary atherosclerosis particularly noncalcified plaques in asymptomatic HIV-infected women without known CV risk compared to uninfected women highlighting the presence of subclinical atherosclerosis in these women. Similarly, CTA has demonstrated high risk morphology plaque types in asymptomatic HIV-infected males [51].
Potential explanations for the increased rates of CVD and subclinical atherosclerosis may be partly explained by increased levels of inflammation and immune activation in HIV-infected women despite virological suppression on ART. While studies have suggested that ART, particularly PIs, increase the risk of cardiovascular disease in HIVinfected men, the contribution of ART to clinical or subclinical CHD in seropositive women remains uncertain due to conflicting results among the number of limited studies available. It appears unlikely, however, that disproportionate representation of traditional CHD risk factors and metabolic indices among HIV-infected women can explain well the observed increased rates of CHD given the inconsistent findings of the prevalence of CV risk factors reported among studies.
Potential reasons for conflicting results among studies of surrogate measures of subclinical atherosclerosis could be heterogeneous study designs and study populations with varying periods of followup. Additionally, different outcomes were measured making it hard to compare studies directly. Many studies were of short duration, which makes it difficult to detect associations with chronic disease processes with cumulative risk such as CVD. Finally, the median age in most studies of HIV-infected women was young posing a challenge in detecting subclinical CVD, which is generally considered a co-morbidity of aging. Nonetheless, it is important to identify and validate surrogate measures of subclinical CVD as these test results may be useful in further risk stratifying HIV-infected patients to maximize preventive measures to prevent clinical CHD events. The most promising surrogate measure of subclinical CHD currently may be the CTA, and larger studies to investigate this imaging modality in HIV-infected women seems warranted based on the suboptimal performance of other available surrogate measures. How to best assess the risk of CHD in HIV-infected women remains uncertain.
There are significant concerns that FRS does not accurately estimate CHD risk in HIV women. The D:A:D score has been suggested to be more accurate in HIV-infected patients because it accounts for history of exposure to ART [52]. However, it too has limitations because the original data used to develop the D:A:D score was obtained between 1999 and 2002 in a predominately male population and there have been dramatic changes in ART regimens since then. Thus it is not clear whether the D:A:D score will remain accurate, and no studies have evaluated it in the era of newer ARVs and specifically in women. When compared with the FRS, the Reynolds score was shown to be more accurate when assessing 10-year CV risk for women [53]. There are no studies using Reynolds score to assess CV risk in HIV-infected women, however. These limitations of our current risk assessment tools reinforce the need to identify and validate risk assessment tools and surrogate measures of clinical and subclinical atherosclerosis in HIVinfected women.
Elucidating the mechanisms for increased rates of CHD and increased risk of CHD in HIV-infected women remains a challenge due to the heterogeneity of study designs, study populations, sample sizes, and outcomes measured. There is a suggestion that metabolic syndrome characterized by insulin resistance may be more prevalent in HIV-infected women, and dyslipidemia appears to be more prominent though inconsistently so compared to HIV-uninfected women and HIV-infected men. Also, some markers of inflammation and immune activation appear to be elevated in HIV-infected women compared to HIV-uninfected women even after virological suppression has been achieved. sCD163 is of particular interest as it has been associated with coronary artery disease in the general population [54]. Another important potential factor contributing to premature, increased CHD risk that has not been extensively evaluated in the literature is the effects of menopause since HIV-infected women have been reported to undergo premature menopause secondary to loss of ovarian function at an earlier age [55,56].
There were several strengths of our review. It presented a comprehensive review of the current literature on the associations of HIV infection and CHD outcomes-namely, both clinical and subclinical surrogate measures of atherosclerosis-in women. We reviewed the prevalence of traditional CHD risk factors among HIVinfected women and compared them to HIV-uninfected women and HIV-infected men, and focused on other important potential mechanisms that may explain the reported associations with CHD, surrogate measures of CVD, and metabolic diseases that increase risk of CHD. Finally, the studies included ranged from urban hospital-based to rural community-based populations in the United States, Europe, Australia, sub-Saharan Africa, and Brazil thereby making the findings generalizable. One limitation of our review was that the review period was restricted to the past 10 years, though most data relevant to the current ART era were well-represented in that time period. Also, we did not perform a systematic meta-analysis but it would have been difficult to perform such an analysis given the scope of our review reporting on a wide range of endpoints such as clinical outcomes, surrogate measures of atherosclerosis, and risk factors.
In summary, HIV-infected women have increased rates of CHD compared to HIV-uninfected women and likely even HIV-infected men. However, there are no well-validated risk assessment tools or surrogate measures of subclinical CHD among HIV-infected to help identify high-risk women for targeting more intensive preventive measures. Additionally, our understanding of the biological and other reasons for the observed disparate CHD rates remains limited. Future studies that include large numbers of HIV-infected women with extended follow-up periods using surrogate measure of CVD and investigating pathogenic mechanisms underlying these observations are urgently needed.