Cardiac Status of Children Infected With Human Immunodeficiency Virus Who Are Receiving Long-term Combination Antiretroviral Therapy: Results From the Adolescent Master Protocol of the Multicenter Pediatric HIV/AIDS Cohort Study

Importance Prior to contemporary antiretroviral therapies (ARTs), children infected with human immunodeficiency virus (HIV) were more likely to have heart failure. This study suggests that highly active ART (HAART) does not appear to impair heart function.

Objective To determine the cardiac effects of prolonged exposure to HAART on HIV-infected children.

Design In the National Institutes of Health–funded Pediatric HIV/AIDS Cohort Study's Adolescent Master Protocol (AMP), we used linear regression models to compare echocardiographic measures.

Setting A total of 14 US pediatric HIV clinics.

Participants Perinatally HIV-infected children receiving HAART (n = 325), HIV-exposed but uninfected children (n = 189), and HIV-infected (mostly HAART-unexposed) historical pediatric controls from the National Institutes of Health–funded Pulmonary and Cardiovascular Complications of Vertically Transmitted HIV Infection (P²C²-HIV) Study (n = 70).

Exposure Long-term HAART.

Main Outcomes and Measures Echocardiographic measures of left ventricular (LV) function and structure.

Results The 325 AMP HIV-infected children had lower viral loads, higher CD4 counts, and longer durations of ART than did the 70 HIV-infected children from the P²C²-HIV Study (all P < .001). The z scores for LV fractional shortening (a measure of cardiac function) were significantly lower among HIV-infected children from the P²C²-HIV Study than among the AMP HIV-infected group or the 189 AMP HIV-exposed but uninfected controls (P < .05). For HIV-infected children, a lower nadir CD4 percentage and a higher current viral load were associated with significantly lower cardiac function (LV contractility and LV fractional shortening z scores; all P = .001) and an increased LV end-systolic dimension z score (all P < .03). In an interaction analysis by HIV-infected cohort, the HIV-infected children from the P²C²-HIV Study with a longer ART exposure or a lower nadir CD4 percentage had lower mean LV fractional shortening z scores, whereas the mean z scores were relatively constant among AMP HIV-infected children (P < .05 for all interactions).

Conclusions and Relevance Long-term HAART appears to be cardioprotective for HIV-infected children and adolescents.

In the early 1990s, children infected with human immunodeficiency virus (HIV) usually did not receive antiretroviral therapy (ART), or they only received monotherapy with zidovudine. They often had abnormalities in left ventricular (LV) structure and function that were some of the strongest predictors of subsequent mortality.¹ These abnormalities occurred before the widespread use of highly active ART (HAART) and likely have multiple causes, such as HIV-infected myocardial cells, myocardial cells coinfected with cardiotropic viruses, and cardiotoxicity from pharmacologic agents and inflammatory cytokines.^2-4 Abnormalities included dilated cardiomyopathy with depressed LV contractility and LV dilation, heart failure, and aortic dilation.^1,3,5-11 Similar abnormalities associated with cardiomyopathic changes have been observed in animal models and in individuals who became infected with HIV as adults.^12-21 In utero exposure to 2-drug nucleoside reverse transcriptase inhibitor therapy in HIV-exposed but uninfected (HEU) children was associated with mitochondrial dysfunction, which could have negative cardiac effects, although a study of in utero exposure to zidovudine monotherapy in HEU children found no deleterious cardiac effects.^22,23

The effects of HIV and ART on the cardiovascular system of HIV-infected children are not completely understood. Children infected with HIV are routinely exposed to ART for many years, including in utero exposure to HAART while the cardiovascular system is still developing. The effects of ART and HIV on the cardiovascular system of children may be interactive; however, the direction and magnitude of such effects are unknown.

Children and adolescents offer a unique opportunity to study the physiologic mechanisms of HIV-associated cardiomyopathy because they are less likely than adults to be exposed long term to such confounding factors as hypertension, smoking, obesity, diabetes mellitus, and coronary atherosclerosis. The effect of HIV infection and ART on children suggests that subclinical abnormalities of cardiac structure and function may eventually result in symptomatic cardiomyopathy in adulthood as they become exposed to the other cardiovascular risk factors already mentioned.^1,3-11,24-26

To date, the largest echocardiographic study of the effects of HIV infection and ART on the cardiac status of children was the Pulmonary and Cardiovascular Complications of Vertically Transmitted HIV Infection (P²C²-HIV) Study, which was funded by the National Heart, Lung, and Blood Institute and conducted from 1990 to 1997.^3,27 Subclinical increased LV mass and decreased LV contractility were early predictors of mortality for HIV-infected children in the pre-HAART era.^1,7 The echocardiographic measurements were associated with postmortem diagnoses of cardiomegaly.^1,6,28 The P²C²-HIV Study also found that HIV-infected children older than 5 years of age were more likely to die of cardiac causes or wasting syndrome than of pulmonary disease.²⁸

To identify the cardiovascular effect of ART (in addition to the cardiovascular effect of HIV infection) on children, we compared 3 groups of children who had echocardiograms: (1) perinatally HIV-infected children who were exposed to HAART; (2) HEU children; and (3) perinatally HIV-infected children from the P²C²-HIV Study who were relatively unexposed to ART. Data from the perinatally HIV-infected children and HEU children were obtained from the Adolescent Master Protocol (AMP) of the PHACS network.

The AMP is a prospective cohort study, conducted at 14 US sites, designed to evaluate the effect of HIV infection and ART on the development of children and adolescents with perinatal HIV infection and to compare these children with a control group of HEU children. The study enrolled children 7 to 16 years of age between March 2007 and November 2009. The study protocol was approved by the institutional review board of each participating site and the Harvard School of Public Health (Boston, Massachusetts). Written informed consent was obtained from the parent or legal guardian, and written informed assent was obtained from older participants, per local institutional review board rules.

We also analyzed data from a cohort of HIV-infected participants, 7 years of age or older, from the P²C²-HIV Study, born before April 1, 1985, in a period when ART was much less intensive. The P²C²-HIV Study was approved by local institutional review boards. The detailed methods for the P²C²-HIV Study are published elsewhere.²⁷ The P²C²-HIV Study data set was obtained by the PHACS Data and Operations Center through a data use agreement approved by the National Institutes of Health. Data elements were similar to those in PHACS.

Our analysis included 3 study groups: (1) perinatally HIV-infected children from the AMP who were exposed to HAART (n = 325); (2) HEU children from the AMP who typically were exposed to prenatal combination ART and short-term neonatal ART prophylaxis (n = 189); and (3) perinatally HIV-infected children from the P²C²-HIV Study who were relatively unexposed to ART (n = 70). Children were excluded from analysis if they had congenital cardiac abnormalities, such as a large atrial septal defect or a ventricular septal defect with pulmonary stenosis.

At each semiannual AMP study visit, information about participants and their families was gathered through clinical interviews and chart reviews. Current health status was ascertained through physical and laboratory evaluations. The ART regimens with start and stop dates were abstracted. Highly active ART was defined as any regimen that included 3 or more ART drugs from 2 or more ART drug classes. The Centers for Disease Control and Prevention (CDC) clinical disease classification was determined at study entry and at each study visit.²⁹

The PHACS-trained site staff obtained a protocol-based single M-mode echocardiogram (recording the short axis of the LV from the parasternal short-axis view and the LV transverse view), with concurrent blood pressure and heart rate measurements. To improve reliability, all echocardiograms were centrally remeasured at the echocardiographic core laboratory at Boston Children's Hospital in Massachusetts.

Children in the P²C²-HIV Study had serial echocardiograms, clinical data collected from chart review, and laboratory evaluations collected in a manner similar to that in the AMP study. For this analysis, to make the age groups as similar as possible, we used the most recent echocardiogram (at least 6 months prior to death) and associated clinical data from the P²C²-HIV Study children who were 7 years of age or older at time of echocardiography. All echocardiograms for the P²C²-HIV Study children were obtained with criteria consistent with those of the AMP and were centrally remeasured at the same site by the same investigator as for the AMP study, at Boston Children's Hospital.

We compared demographic and anthropomorphic covariates among the 3 groups using χ² tests for categorical variables and Kruskal-Wallis tests for continuous measures. Measures of HIV disease severity and ART characteristics (latest available measures before or at the time of echocardiography) were compared between the HIV-infected children in the AMP and the HIV-infected children in the P²C²-HIV Study using Wilcoxon rank sum tests or χ² tests. We also compared the characteristics of children with echocardiograms with those of children without echocardiograms, to assess the representativeness of our study population.

To minimize the burden on participating children, their families, and site staff while maximizing efficiency, the study was designed to obtain echocardiograms from 500 children to provide 80% power for detecting differences in mean z scores of 0.26 or greater. Echocardiography was therefore discontinued in December 2010 when the target was achieved. Echocardiographic z scores were calculated using normative data from an established reference cohort of healthy children from Boston Children's Hospital not known to be infected with HIV or exposed to HAART, with adjustment for age and body-surface area, as appropriate.³⁰ The distributions of z scores were compared among the 3 groups (HIV-infected children from the AMP, HEU children from the AMP, and HIV-infected children from the P²C²-HIV Study) using analysis of variance (df = 2) from general linear regression models, both unadjusted and adjusted for age, sex, race, ethnicity, and other potential confounders, such as body mass index z score.³¹ Within each group, overall 1-sample t tests were performed to determine whether echocardiographic z scores differed from those of the normal pediatric reference population.

General linear regression models were also fit to data from both groups of HIV-infected children to evaluate the association between characteristics of ART regimens (duration of ART and type of current ART regimen) and measures of HIV disease severity (CD4⁺ T lymphocyte percentage [CD4 percentage]), HIV-1 viral load (VL), and CDC classification (N, not symptomatic; A, mildly symptomatic; B, moderately symptomatic; and C, severely symptomatic). Each measure of HIV disease severity and each characteristic of ART were evaluated for association with echocardiographic z scores, with adjustment for age, sex, race/ethnicity, and body mass index z score. Potential differences in the association between HIV disease severity measures and echocardiographic measures by cohort, representing pre-HAART era vs HAART era, were investigated via inclusion of interaction terms. Analyses were conducted using SAS version 9.2 (SAS Institute Inc) and were based on data submitted as of January 2012.

Of the 678 children in the PHACS AMP study, 451 were HIV-infected children, and 227 were HEU children. Of these 678 children, 514 (76%) (325 of 451 HIV-infected children [72%] and 189 of 227 HEU children [83%]) had echocardiograms available for analysis, slightly exceeding the target of 500 children. The AMP children with echocardiograms were demographically similar to those without echocardiograms. Seventy HIV-infected children with echocardiograms obtained at 7 years of age or older were included from the P²C²-HIV Study (Table 1).

The children in the P²C²-HIV Study were younger than the HIV-infected children and HEU children in the AMP at the time of echocardiography (mean age, 9.7 years vs 13.0 years vs 11.0 years, respectively; P < .001). In the P²C²-HIV Study group, there was a significantly higher proportion of Hispanics and a lower proportion of blacks than in either of the AMP groups (P < .001). The HIV-infected children in both the AMP and the P²C²-HIV Study were significantly shorter and lighter than the AMP HEU children.

The AMP HIV-infected children had significantly less advanced HIV disease, in terms of VL, CD4 counts, and CD4 percentage, at the time of echocardiography than did the HIV-infected children in the P²C²-HIV Study (Table 2). The AMP HIV-infected children were more likely to have had a prior AIDS-defining condition (CDC classification C) than were the HIV-infected children in the P²C²-HIV Study (P < .001). At the time of echocardiography, 89% of AMP HIV-infected children and 17% of HIV-infected children in the P²C²-HIV Study were receiving HAART (P < .001), and 80% of AMP HIV-infected children had received HAART for more than 5 years, whereas none of the HIV-infected children in the P²C²-HIV Study had received HAART for more than 5 years. The AMP HIV-infected children were also significantly more likely to have received a protease inhibitor than were the HIV-infected children in the P²C²-HIV Study.

Regarding LV function, the HIV-infected children in the P²C²-HIV Study had the lowest adjusted mean LV contractility z score (−1.56; Table 3). Both HIV-infected cohorts had lower mean LV fractional shortening z scores than that of the AMP HEU cohort, with the HIV-infected children in the P²C²-HIV Study having the lowest mean z score (−1.94). For LV mass, LV end-diastolic dimension, and LV end-systolic dimension mean z scores, the 3 cohorts had significantly different values, with the AMP-HEU cohort having the lowest and the P²C²-HIV cohort having the highest. The AMP-HEU and AMP-HIV cohorts had lower end-diastolic septal wall thickness and LV end-systolic wall stress z scores than did the P²C²-HIV cohort. Differences in mean z scores also had clinical significance: 31 of the 70 HIV-infected children (44%) in the P²C²-HIV Study met the definition for cardiomyopathy (either a z score of less than −2 for LV fractional shortening or greater than 2 for LV dimension), whereas 12 of the 325 HIV-infected children in the AMP and 3 of the 189 of the HEU children in the AMP met the definition for cardiomyopathy (P < .001).

Children with a nadir CD4 percentage of less than 15% and those with a VL of greater than 5000 copies had significantly lower LV contractility, LV fractional shortening, and higher LV end-systolic dimension and LV end-systolic wall stress z scores than did those with less severe disease (Table 4). In addition, children with current VLs greater than 5000 copies had significantly higher end-diastolic septal thickness. Children with a prior AIDS-defining condition (CDC classification C) had shifts in mean z scores similar to those already described (Table 4). In multivariable models including all 3 HIV disease measures from Table 4, a current VL of greater than 5000 copies/mL and a nadir CD4 percentage of less than 15% remained significant independent predictors for these echocardiographic measures, whereas CDC classification C was no longer associated with these echocardiographic parameters.

Relationships between certain HIV disease severity measures and some echocardiographic parameters differed between the 2 HIV-infected cohorts. In the P²C²-HIV cohort, those with a longer duration of ART had lower mean LV fractional shortening z scores. However, in the AMP-HIV cohort, the mean z scores for LV fractional shortening were constant, regardless of duration of ART (interaction, P = .04). In addition, the HIV-infected children in the P²C²-HIV Study with lower nadir CD4 percentages had worse cardiac function, on average, with lower mean z scores for LV fractional shortening and higher mean z scores for LV end-systolic dimension. However, for the HIV-infected children in the AMP, these z scores (Figure) appeared relatively constant over levels of nadir CD4 percentage (interaction, P = .001 and 0.001, respectively). Thus, the overall shifts in mean z scores associated with a nadir CD4 percentage of less than 15% shown in Table 4 are primarily attributable to associations within the P²C²-HIV cohort, for which a much higher percentage of children fell into this low immunological category. For other measures of HIV disease severity, such as CDC classification C and current VL, there was no evidence of interaction for any echocardiographic parameter. Thus, the overall measures of association with HIV disease severity shown in Table 4 apply to both HIV-infected cohorts.

We compared echocardiographic measures of LV structure and function between a cohort of HIV-infected children from an earlier treatment era who were relatively unexposed to ART (from the P²C²-HIV Study) and a contemporary cohort of HIV-infected children who were more exposed to ART (usually HAART; from the AMP). We also compared both of these cohorts with a contemporary cohort of HEU children and determined the associations of HIV disease severity with echocardiographic characteristics in the HIV-infected cohorts.

In the P²C²-HIV cohort, increased LV mass, dimension, wall thickness, and wall stress and decreased LV fractional shortening and contractility all predicted mortality.^1,7 After about 5 to 6 years of age, these children were more likely to die of chronic cardiac disease than pulmonary disease.²⁸ A subset of this cohort developed clinical heart disease.⁴ These same echocardiographic parameters in the AMP-HIV cohort were closer to normal, and none of the children had symptomatic heart disease. This finding suggests an overall cardioprotective effect of long-term HAART. Whether this effect is the result of better immunological health, more normal somatic growth, direct killing of intramyocardial HIV, or other mechanisms is unclear.

The HIV-infected children in the P²C² HIV Study who had a longer duration of ART tended to have lower LV fractional shortening. In this group, lower LV fractional shortening was associated with a lower nadir CD4 percentage. Increased LV end-systolic dimension z score, which is associated with myocardial contractility, was also associated with a lower nadir CD4 percentage. In contrast, LV fractional shortening z scores appeared relatively constant over all durations of ART and over all nadir CD4 percentages for the HIV-infected children in the AMP. We believe these differences suggest that some children in the P²C²-HIV cohort with evidence of substantial immunosuppression had already experienced substantial cardiac damage before they received effective ART regimens, if they received any ART at all. This conclusion is consistent with previous reports from the P²C²-HIV Study showing that many of the deleterious cardiac changes occurred in the first few years of life in children with marked immunosuppression without effective ART or multidrug therapy.^1,3,6

The longitudinal National Heart, Lung, and Blood Institute Cardiac Highly Active Antiretroviral Therapy II (CHAART-II) Study compared echocardiographic measurements between a HAART-exposed, HIV-infected cohort (3-16 years of age), which was contemporary to the AMP cohort, and the much less ART-exposed P²C²-HIV cohort.³² During the early years of follow-up, the CHAART-II HAART-exposed children had consistently higher LV fractional shortening and contractility, lower LV wall stress, and more normal LV mass and LV wall thickness than the children in the P²C²-HIV cohort.³² At longer follow-up, the better LV function in terms of both fractional shortening and contractility was even more marked when compared with the P²C²-HIV cohort.³² However, some of the structural differences between the cohorts, such as for LV mass or dimension, became attenuated with longer follow-up.³¹

Consistent with our findings, the results from the CHAART-II Study suggest that HAART-exposed, HIV-infected children maintain normal cardiac function for at least the first 10 years of follow-up, whereas HIV-infected children from the pre-HAART era had progressive deterioration in function.³² However, differences in cardiac structure between the 2 groups appeared to lessen over time. Another group of children who were exposed to cardiotoxins are survivors of childhood cancer who were treated with anthracyclines.³³ These survivors have shown an initial maintenance of or improvement in cardiac health that was followed by a deterioration in cardiac status more than 10 years after exposure.³⁴ Therefore, HIV-infected children and adolescents currently receiving HAART may benefit from a follow-up echocardiogram to determine the trajectory of their cardiac status, although the cost-effectiveness and optimal timing of a follow-up echocardiogram are unknown.

This study was a cross-sectional analysis, so no causal inferences can be made. Also, analyzing only 1 echocardiogram per participant does not allow us to know whether their cardiac status was static, improving, or deteriorating. We also did not examine the effects of individual ART or specific combinations of these drugs on cardiac structure or function. Finally, the Boston Children's Hospital reference cohort used to calculate the z scores was very different in terms of racial distribution and socioeconomic status from both of the HIV-infected cohorts.

Children with more severe HIV disease appear to experience more serious adverse effects on cardiac structure and function. Our results indicate that the current use of combination ART, usually HAART, appears to be cardioprotective in HIV-infected children and adolescents. This finding is even more relevant in the developing world where the prevalence of HIV disease in children is much higher. The long-term cardiac health of the HIV-infected children in the AMP in the HAART era is unknown.³⁵ A recent report³⁵ of HIV-infected adolescents and young adults receiving long-term HAART found that, while traditional echocardiographic measures of cardiac function were normal, measures of cardiac strain and strain rate were significantly impaired. Sims et al³⁵ suggest that these echocardiographic measures could be prognostic of long-term myocardial dysfunction in this population. Examining the associations between individual ART agents and combinations of ART might identify optimal ART regimens, both in terms of optimizing HIV outcomes and protecting long-term cardiac health and represents a future research opportunity using this data set. Finally, an analysis of renal function data from these 2 HIV-infected pediatric cohorts, relative to the cardiac findings from both the P²C²-HIV and AMP-HIV cohorts, could be a potentially interesting next research step in better characterizing the potential contributing factors to cardiac outcomes in pediatric HIV-infected populations.

Correspondence: Steven E. Lipshultz, MD, Department of Pediatrics, University of Miami Leonard M. Miller School of Medicine, PO Box 016820, Miami, FL 33101 (slipshultz@med.miami.edu).

Accepted for Publication: October 17, 2013.

Published Online: April 22, 2013. doi:10.1001/jamapediatrics.2013.1206

Author Contributions: Dr Lipshultz had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Lipshultz, Williams, Wilkinson, Van Dyke, Shearer, Rich, and Hazra. Acquisition of data: Van Dyke, Dooley, Scott, and Colan. Analysis and interpretation of data: Lipshultz, Williams, Wilkinson, Leister, Van Dyke, Shearer, Rich, Hazra, Kaltman, Jacobson, Rabideau, and Colan. Drafting of the manuscript: Lipshultz, Williams, Wilkinson, Rich, Kaltman, Dooley, and Rabideau. Critical revision of the manuscript for important intellectual content: Lipshultz, Williams, Wilkinson, Leister, Van Dyke, Shearer, Rich, Hazra, Kaltman, Jacobson, Scott, and Colan. Statistical analysis: Williams, Leister, Shearer, and Jacobson. Obtained funding: Williams, Van Dyke, and Rich. Administrative, technical, and material support: Lipshultz, Wilkinson, Shearer, Rich, Kaltman, Dooley, Rabideau, and Colan. Study supervision: Lipshultz, Williams, Wilkinson, Van Dyke, Hazra, and Kaltman.

PHACS Group Members: The following institutions, clinical site investigators, and staff participated in conducting PHACS AMP in 2012: William Shearer, Mary Paul, Norma Cooper, and Lynette Harris (Baylor College of Medicine, Houston, Texas); Murli Purswani, Mahboobullah Baig, and Anna Cintron (Bronx Lebanon Hospital Center, New York); Ana Puga, Sandra Navarro, Doyle Patton, and Deyana Leon (Children's Diagnostic and Treatment Center, Fort Lauderdale, Florida); Sandra Burchett, Nancy Karthas, and Betsy Kammerer (Boston Children's Hospital, Massachusetts); Ram Yogev, Margaret Ann Sanders, Kathleen Malee, and Scott Hunter (Ann and Robert H. Lurie Children's Hospital of Chicago, Illinois); Andrew Wiznia, Marlene Burey, and Molly Nozyce (Jacobi Medical Center, Bronx, New York); Janet Chen, Latreca Ivey, Maria Garcia Bulkley, and Mitzie Grant (St Christopher's Hospital for Children, Philadelphia, Pennsylvania); Katherine Knapp, Kim Allison, and Megan Wilkins (St Jude Children's Research Hospital, Chicago, Illinois); Midnela Acevedo-Flores, Heida Rios, and Vivian Olivera (San Juan Hospital/Department of Pediatrics, Puerto Rico); Margarita Silio, Medea Jones, and Patricia Sirois (Tulane University Health Sciences Center, New Orleans, Louisiana); Stephen Spector, Kim Norris, and Sharon Nichols (University of California, San Diego); Elizabeth McFarland, Emily Barr, and Robin McEvoy (University of Colorado Denver Health Sciences Center); Arry Dieudonne, Linda Bettica, and Susan Adubato (University of Medicine and Dentistry of New Jersey, Newark); and Gwendolyn Scott, Patricia Bryan, and Elizabeth Willen (University of Miami, Florida).

Funding/Support: The study and PHACS were supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development with cofunding from the National Institute on Drug Abuse, the National Institute of Allergy and Infectious Diseases, the Office of AIDS Research, the National Institute of Mental Health, the National Institute of Neurological Disorders and Stroke, the National Institute on Deafness and Other Communication Disorders, the National Heart, Lung, and Blood Institute, the National Institute of Dental and Craniofacial Research, and the National Institute on Alcohol Abuse and Alcoholism, through cooperative agreements with the Harvard School of Public Health (grants HD052102, 3 U01 HD052102-05S1, and 3 U01 HD052102-06S3) and the Tulane University School of Medicine (grants HD052104 and 3U01HD052104-06S1). Data management services were provided by the Frontier Science and Technology Research Foundation, and regulatory services and logistical support were provided by Westat, Inc.

Role of the Sponsors: Representatives of the National Institutes of Health were part of the study team, and therefore the sponsor was involved in the design and conduct of the study; in the collection, analysis, and interpretation of the data; and in the preparation, review, or approval of the manuscript.

Disclaimer: The conclusions and opinions expressed in this article are those of the authors and do not necessarily reflect those of the National Institutes of Health or the US Department of Health and Human Services.

Additional Contributions: We thank the children and families for their participation in PHACS and the individuals and institutions involved in the conduct of PHACS.