Author + information
- Received April 29, 2010
- Revision received July 9, 2010
- Accepted August 24, 2010
- Published online January 4, 2011.
- Steven E. Lipshultz, MD*,* (, )
- William T. Shearer, MD, PhD†,
- Bruce Thompson, PhD§,
- Kenneth C. Rich, MD∥,
- Irene Cheng, MS§,
- E. John Orav, PhD¶,
- Sulekha Kumar, MD∥,
- Ricardo H. Pignatelli, MD‡,
- Louis I. Bezold, MD‡,
- Philip LaRussa, MD#,
- Thomas J. Starc, MD, MPH**,
- Julie S. Glickstein, MD**,
- Sharon O'Brien, MD‡‡,
- Ellen R. Cooper, MD††,
- James D. Wilkinson, MD, MPH*,
- Tracie L. Miller, MD* and
- Steven D. Colan, MD§§
- ↵*Reprint requests and correspondence:
Dr. Steven E. Lipshultz, Department of Pediatrics (D820), Leonard M. Miller School of Medicine, University of Miami, P.O. Box 016820, Miami, Florida 33101
Objectives The aim of this study was to investigate the possible effects of antiretroviral therapy (ART) in utero on cardiac development and function in human immunodeficiency virus (HIV)–negative children.
Background ART reduces vertical HIV transmission. Long-term cardiotoxicity after in utero exposure to ART is unknown in children but has occurred in young animals.
Methods Using a prospective multisite cohort study design, echocardiograms taken between birth and 24 months were compared in 2 groups of HIV-negative infants of HIV-positive mothers: 136 infants exposed to ART (ART+) and 216 unexposed infants (ART−).
Results Mean left ventricular (LV) mass z-scores were consistently lower in ART+ girls than in ART− girls: differences in mean z-scores were −0.46 at birth (p = 0.005), −1.02 at 6 months (p < 0.001), −0.74 at 12 months (p < 0.001), and −0.79 at 24 months (p < 0.001). Corresponding differences in z-scores for boys were smaller: 0.13 at 1 month (p = 0.42), −0.44 at 6 months (p = 0.01), −0.15 at 12 months (p = 0.37), and −0.21 at 24 months (p = 0.21). Septal wall thickness and LV dimension were smaller than expected in ART+ infants, but LV contractility was consistently about 1 SD higher at all ages (p < 0.001). In ART+ infants, LV fractional shortening was higher than in ART− infants; girls showed a greater difference.
Conclusions Fetal exposure to ART is associated with reduced LV mass, LV dimension, and septal wall thickness z-scores and increased LV fractional shortening and contractility up to age 2 years. These effects are more pronounced in girls than in boys. Fetal ART exposure may impair myocardial growth while improving depressed LV function.
Abnormalities of left ventricular (LV) structure and function are associated with human immunodeficiency virus (HIV) infections (1–6), possibly with antiretroviral therapy (ART) (7–10), and even mild abnormalities independently predict mortality in HIV-infected children (3–5). HIV-exposed but negative newborns show depressed LV contractility that persists at 5 years of age, whereas HIV-positive newborns experience chronic and progressive abnormalities in LV structure and function (6).
Mitochondrial abnormalities have been reported in animals and children exposed to ART, with inconsistent results (7–12). Such abnormalities could cause cardiomyopathy in exposed children, although the clinical implications of mitochondrial abnormalities in HIV-negative infants exposed to ART are uncertain (7–12).
Nearly 10,000 ART-exposed and HIV-exposed but negative infants are born annually in the U.S. but are not routinely followed for suspicion of heart disease (8–10,13).
To study the cardiac effects of perinatal exposure to ART, we compared 2 cohorts of HIV-negative children born to HIV-positive mothers.
In the CHAART-1 (Cardiovascular Status of HAART Therapy in HIV-Exposed Infants and Children) cohort study (14), 166 HIV-negative infants exposed to ART in utero were prospectively identified between June 2003 and January 2006 from the WITS (Women and Infants Transmission Study) of 2,842 infants born to HIV-positive women (15). CHAART-1 participants came from 5 of the 7 WITS sites; 4 also participated in the P2C2 HIV (Pediatric Pulmonary and Cardiac Complications of Vertically Transmitted HIV Infection) study. CHAART-1 participants were enrolled in WITS and were age 2 years or less so that primary end points could be measured at age 2 years. CHAART-1–recruited children were followed until loss to follow-up, the child withdrew, December 2006, or whichever came first.
Maternal exclusion criteria for CHAART-1 children included diabetes, phenylketonuria, a Mendelian or chromosomal defect, a heart defect requiring medication or surgery, or pregnancy exposures to chemotherapy, radiation, or drugs associated with heart disease in offspring.
The P2C2 HIV study prospectively described cardiovascular disorders in children born to HIV-positive mothers enrolled between 1990 and 1994 at 6 centers (16). Of the 463 HIV-negative infants enrolled during gestation or within 28 days after birth, 262 had no in utero or post-natal ART exposure and were the non-ART-treated controls. The P2C2 HIV study showed that independently of ART, HIV-negative children born to HIV-positive women have abnormal LV structure and function, making them an essential control group when assessing the impact of ART therapy in the CHAART-1 study (6).
CHAART-1 sites received institutional review board approval to collect serial echocardiograms. Written informed consent was obtained from parents or guardians.
Prospective data were collected from the CHAART-1 and P2C2 HIV datasets. Cardiac function was evaluated through serial echocardiograms starting at birth or the visit closest to CHAART-1 enrollment and ending on December 31, 2006. Echocardiograms were scheduled at 6 intervals from 1 to 48 months of age. The corresponding P2C2 HIV collection was scheduled at 14 intervals from birth to 54 months. In this report, we analyze echocardiographic data for participants up to age 2 years.
The P2C2 HIV echocardiographic acquisition protocol was followed (2,14). Echocardiographic data were digitized by the same independent, blinded cardiologist who had measured the previous P2C2 HIV and Boston control echocardiographic data (17).
CHAART-1 compared 2 primary echocardiographic end points at age 2 years with respect to in utero ART exposure: LV mass and fractional shortening. Alpha was set at 0.025 to correct for multiple comparisons. All other analyses were carried out at the 0.05 alpha level and should be considered as suggestive findings in need of future confirmation.
General linear models were used to determine whether ART independently predicted changes in these 2 measures and to identify any ART-covariate interactions. Serial echocardiographic measurements from birth to 2 years were analyzed using generalized estimating equations (18). Twins and infants whose maternal ethnicity was other than African American, Caucasian, or Hispanic were excluded.
The echocardiographic measures were standardized to z-scored deviations using Boston infants and children as the normal reference population (2). Z-scores were used as an adjustment mechanism to correct echocardiographic measurements for heart growth and development. Z-scores of zero correspond to an average heart parameter measurement, negative to below average, and positive to above average. Numeric values correspond to changes in standard deviation units. The methodology is identical to the growth charts versus age for height and weight in children that are provided by the Centers for Disease Control and Prevention. As long as the comparison is made with populations that are not markedly different from the normal control reference population with regard to factors such as body mass index, level of habitual exercise (i.e., exclusion of elite athletes), and blood pressure, no differences in z-scores are anticipated.
Interaction terms of ART exposure with sex, ethnicity, and age that were significant at 0.05 were included in the final models. Alpha at 0.05 determined statistical significance in all analyses, except for the primary end points (see the preceding discussion), because these were hypothesis-generating analyses. The effect of ART exposure on primary end points was analyzed without adjustments and with adjustment for sex, ethnicity, significant maternal characteristics, and significant sex-ART interaction.
All analyses were performed using SAS version 8.2 (SAS Institute Inc., Cary, North Carolina).
Demographic and clinical characteristics
Data were collected from 136 ART-exposed, HIV-negative infants from the CHAART-1 centers and from 216 non-ART-exposed, HIV-negative infants from the P2C2 HIV study (Table 1) (19). Infant sex and maternal ethnicities were similar. However, the maternal CHAART-1 cohort used tobacco (22% vs. 37%) and illicit drugs (15% vs. 29%) less, and their infants had higher birth weights. CHAART-1 mothers experienced vaginal bleeding during pregnancy (25% vs. 3%) and cesarean-section delivery (40% vs. 18%) more frequently. Immunologic studies were similar, except for a higher mean CD8 T-lymphocyte count in the P2C2 HIV mothers (p = 0.04) (Table 1). In utero ART exposure for the 136 CHAART-1 infants was 6 (4%) for mono-ART, 11 (8%) for combination ART (2 or more ART drugs, without a non-nucleoside reverse transcriptase inhibitor or protease inhibitor), and 119 (88%) for highly active ART (3 or more ART drugs, including a non-nucleoside reverse transcriptase inhibitor or protease inhibitor).
Data included 234 serial echocardiograms from CHAART-1 infants (mean 1.7 per child) and 516 serial echocardiograms from P2C2 HIV infants (mean 2.4 per child). Mean systolic and diastolic blood pressure z-scores were higher in ART-exposed male infants compared with unexposed male infants (Table 1). Adjustment variables for each analysis in a table are presented in the footnote of that table.
Primary end point evaluations
The unadjusted LV mass z-score for all CHAART-1 infants at age 2 years was 0.53 SD less than in the P2C2 HIV cohort (p = 0.006). LV mass and ART remained associated after adjusting for sex, sex-ART interaction (p = 0.009), ethnicity, and maternal vaginal bleeding (p = 0.01). Both the unadjusted and adjusted LV mass z-scores were smaller in the CHAART-1 group when pooling across sex; most of the difference was among girls, who showed a greater reduction (Table 2).
The unadjusted LV fractional shortening z-score for CHAART-1 infants was 0.45 SD greater than that of the P2C2 HIV infants (p = 0.01) at age 2 years. For this outcome, there was no sex-ART interaction. LV fractional shortening and ART remained significantly associated after adjusting for sex and ethnicity. Both the unadjusted and adjusted LV fractional shortening z-scores were higher in the CHAART-1 group, with girls showing a greater but nonsignificant difference (Table 2).
Longitudinal analyses of cardiac outcomes
LV fractional shortening and LV mass end points were similar to the outcomes at age 2 years, except that the sex-ART interaction for LV fractional shortening was more prominent and statistically significant. The sex-ART differences for these 2 end points were present by age 6 months and remained at age 2 years.
Adjusted LV end-systolic dimension z-scores were lowest in the ART-exposed group, with the greatest differences in girls (0.52 SD lower in the ART-exposed than the non-ART-exposed girls). Adjusted LV end-diastolic dimension z-scores were lower in the ART-exposed girls at age 6 months and age 2 years (0.44 SD and 0.51 SD, respectively). LV contractility was 0.72 SD and 1.11 SD higher in ART-exposed boys and girls, respectively. End-diastolic septal wall thickness was 0.82 SD and 1.42 SD lower in ART-exposed boys and girls, respectively. For the cardiac end points without sex-by-ART interaction, results were combined across sexes (Table 3).
Other treatment interactions for cardiac end points
Cardiac outcomes had statistically significant sex-by-treatment or time-by-treatment interactions (Figs. 1 and 2). Among CHAART-1 boys, mean LV mass was higher at 1 month and decreased over 6 months; in P2C2 HIV boys, it increased over 6 months (Fig. 1A). The CHAART-1 girls were consistently about one-half an SD below the average of the P2C2 HIV girls (Fig. 1B).
LV fractional shortening measurements for CHAART-1 boys changed from being above (but still below normal) at 1 month to approximately the same (and normal) at age 2 years compared with P2C2 HIV boys (Fig. 1C). For girls, the difference between the adjusted means of the 2 cohorts was greatest at 1 month and remained statistically significant at age 2 years (0.87 SD, p < 0.001) (Table 2, Fig. 1D). LV end-diastolic dimension (Figs. 1E and 1F), end-systolic dimension (Figs. 1G and 1H), contractility (Figs. 1I and 1J), and end-diastolic septal wall thickness (Figs. 1K and 1L) differed by sex and ART treatment (Table 2).
LV end-diastolic posterior wall thickness (Fig. 2A), heart rate (Fig. 2B), and weight-for-height (Fig. 2C) z-scores differed by cohort (Table 3). CHAART-1 infants consistently had higher LV end-diastolic posterior wall thickness, but their z-scores remained approximately one-half an SD below zero (p < 0.01). Heart rate measures were 0.18 SD higher in the CHAART-1 infants at all ages (p = 0.002). The heart rate z-scores in the CHAART-1 infants increased from 0.36 SD at birth to 1.06 SD at 2 years (p < 0.01). Weight-for-height z-scores were 0.79 SD higher in the CHAART-1 infants than in the P2C2 HIV infants at 1 month only (p < 0.001), with the CHAART-1 measurements above normal and the P2C2 HIV measurements below normal (p < 0.01). Both cohorts improved to normal after 1 month.
LV end-systolic posterior wall thickness (Fig. 2D), end-systolic septal wall thickness (Fig. 2E), and LV afterload (Fig. 2F) showed a qualitative (crossing) interaction between age and exposure to ART (Table 3), resulting in significant associations with ART only at certain ages. LV end-systolic posterior wall thickness measurements of both cohorts were below normal at 1 month, with higher values for the CHAART-1 cohort. Both cohorts improved to normal and became approximately the same at age 2 years. End-systolic septal thickness z-scores from the CHAART-1 cohort were 0.40 SD lower than the P2C2 HIV measurements during the first year (p < 0.05). Measurements from both cohorts became approximately the same at age 2 years. LV afterload showed the CHAART-1 cohort starting at 1 month with an average z-score 0.47 SD lower than the P2C2 HIV measurements and ending approximately the same as the P2C2 HIV measurements at 12 months and age 2 years.
Infants of HIV-positive mothers exposed to ART showed improved LV contractility and fractional shortening during the first 2 years of life compared to a non-ART-exposed cohort of HIV-negative infants born to HIV-positive mothers. However, ART exposure was also associated with reduced LV mass, septal thickness, and dimension, which were also below normal. In other settings, these 3 changes can lead to progressive LV dysfunction (20–22). The long-term cardiac effects of fetal exposure to ART are unclear. Our findings are consistent with the hypothesis that control of myocardial growth differs from control of myocardial function and that ART exposure is associated with less myocardial growth but better myocardial function.
Higher LV contractility leads to a smaller LV end-systolic volume and a higher wall thickness and hence lowers LV wall stress at end-systole. A decrease in LV wall stress reduces the stimulus to LV hypertrophy, and hence, lower LV mass is the anticipated response to higher LV contractility.
Our ART-exposed cohort had better maternal health than did the P2C2 HIV cohort and better infant LV contractility and fractional shortening. LV fractional shortening normalized in both groups by age 2 years, except for CHAART-1 girls, and LV contractility was better in the CHAART-1 group and remained normal or slightly above normal across all visits. Adding the significant maternal variables to the final multivariate models did not change the ART exposure effects on echocardiographic parameters, suggesting that ART itself, not improved maternal health, affects LV function either directly or indirectly. We incorrectly speculated that improved maternal health resulted in improved LV contractility and fractional shortening (6).
As in rodents (23,24), ART exposure inhibited myocardial growth, being associated with decreased septal thickness and LV dimension and mass. These changes suggest an overall loss of cardiac tissue with ART exposure, an inability of the septum to grow in response to increasing body surface area, or both. Weight for height was not significantly different by ART exposure with follow-up, suggesting that body surface area–normalized z-scores for these variables are appropriate.
Although our results suggest that ART is associated with early improved LV function, longitudinal changes in other factors may lead to increasing mechanical stress, such as lower LV mass, that may result in increased LV afterload, which could compromise long-term LV function, a finding found in anthracycline-treated childhood cancer survivors (20–22). Mechanistically, these changes could be induced by apoptosis and by decreased hypertrophic signaling mechanisms related to ART growth-inhibiting effects, ART-associated deoxyribonucleic acid mitochondrial mutations, or gene polymorphism shifts (23–25), mechanisms similar to that for pediatric cancer survivors with doxorubicin-associated cardiomyopathy, a well-characterized example of late cardiotoxicity after early childhood drug exposure (26).
Perinatal myocardiocyte injury and death are significantly related to inflammation (27). Fetal inflammatory response syndrome is associated with increased neonatal morbidity (28). Immune reconstitution inflammatory syndrome in ART-exposed HIV-infected mothers may heighten inflammation that contributes to myocardiocyte injury and reduced neonatal LV mass (29).
Exposure to ART was also associated with higher heart rates, which have been associated with the rapid progression of childhood HIV-associated disease (30). Sustained tachycardia, as observed with ART exposure, may be cause or effect, but with secondary LV hypoplasia may explain our LV growth findings. Mild LV restrictive disease, with a reduced capacity for LV dilation and with a secondary tachycardia, is also possible. Reduced end-diastolic volume as the initial limiting factor to cardiac output with secondary compensatory tachycardia is certainly the anticipated response to detraining and to a restrictive cardiomyopathy, so this sequence is also plausible. Alternatively, the decreased LV growth may be secondary to a direct ART effect, such as mitochondrial deoxyribonucleic acid damage from ART-associated oxidative stress, resulting in LV hypoplasia that contributes to the observed increased LV contractility and heart rate, especially in girls (23–25).
The effects of ART exposure on LV mass, dimension, fractional shortening, and septal thickness were more pronounced in girls. These differences might predict early and advanced cardiovascular disease in girls as they age. Our findings are similar to the effects of zidovudine effects in mice, in which females were more sensitive to ART than males (31). Similarly, newborn girls have greater myocardial injury marker elevations than newborn boys (27). Doxorubicin cardiotoxicity suggests an increased vulnerability of the developing female myocardium (20–22).
The magnitude of the z-score differences in this study may appear to be small relative to those that guide the daily clinical decisions by cardiologists. However, their magnitude is consistent with findings in doxorubicin-treated pediatric cancer survivors (20–22), who 20 to 30 years later have a 15-fold increase in the rate of heart failure compared with siblings and an 8.2-fold increase in cardiac mortality compared with U.S. population mortality data (32,33).
Zidovudine and its metabolites cross the placenta and are found in the fetal heart (34). The hypothesis that zidovudine may be associated with cardiac mitochondrial dysfunction is supported by animal studies and by limited clinical data (35–39). Yet, zidovudine monotherapy showed no effect on the same LV structure and function measurements in the present study in either HIV-negative or -positive children (11,12). Follow-up studies of HIV-negative infants born to HIV-positive women have found neither detrimental effects of perinatal zidovudine exposure nor clinically important cardiovascular disease (8–10,12,40–42).
Combination ART (96% of CHAART-1 infants had combination therapy) has early and marked effects on LV structure and function. In rodents, combination ART produces more mitochondrial deoxyribonucleic acid mutations than monotherapy (23–25,31). In contrast to nucleoside reverse transcriptase inhibitors, protease inhibitors in general do not cross the placenta, suggesting that the effects seen likely result from nucleoside reverse transcriptase inhibitor combinations rather than protease inhibitors. Combination ART compared with a nucleoside reverse transcriptase inhibitor alone in HIV-negative children was associated with more lymphocyte abnormalities (43). Other abnormalities in ART-exposed, HIV-negative children include anemia, neutropenia, thrombocytopenia, lower CD4 and CD8 T-lymphocyte counts, and increased micronucleated erythrocytes (43–45).
The P2C2 HIV study (1991 to 1996) and the CHAART-1 study (2003 to 2006) were performed in different eras (before and after highly active ART, respectively). This nonconcurrent design was used for ethical reasons, because ART therapy could not be withheld from pregnant women. However, the cardiac protocol, clinical sites, study personnel, and the person measuring echocardiograms were largely common to both studies.
As previously noted, a potential limitation of the use of z-scores is if the study population differs significantly from the reference population in terms of anthropomorphic measures. We found no evidence of such differences between our study population and the reference group.
HIV virulence might have increased in the decade between the studies, causing CHAART-1 measurements to deviate more from normal than P2C2 HIV measurements. However, the CHAART-1 mothers had no evidence of more advanced disease.
Fetal exposure to ART is associated with reduced LV mass, dimension, and septal wall thickness and higher LV fractional shortening and contractility during the first 2 years of life. We speculate that in utero exposure to ART may impair myocardial growth while initially improving LV function, although LV function was less than normal. These effects are more pronounced in girls. In the U.S., with more than 100,000 HIV-negative infants exposed to ART, these findings clearly indicate a need for long-term monitoring of these infants to better define the mechanism of these effects and to evaluate their long-term clinical importance.
The CHAART-1 study is supported by grant R01 HL72705 from the National Heart, Lung, and Blood Institute to Dr. Lipshultz. The authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- antiretroviral therapy
- human immunodeficiency virus
- left ventricular
- Received April 29, 2010.
- Revision received July 9, 2010.
- Accepted August 24, 2010.
- American College of Cardiology Foundation
- Lipshultz S.E.,
- Easley K.A.,
- Orav E.J.,
- et al.
- Al-Attar I.,
- Orav E.J.,
- Exil V.,
- Vlach S.A.,
- Lipshultz S.E.
- Lipshultz S.E.,
- Easley K.A.,
- Orav E.J.,
- et al.
- Foster C.,
- Lyall H.
- Shearer W.T.,
- Quinn T.C.,
- LaRussa P.,
- et al.,
- Women and Infants Transmission Study Group
- Lipshultz S.E.,
- Easley K.A.,
- Orav E.J.,
- et al.
- Liang K.Y.,
- Zeger S.L.
- Lipshultz S.E.,
- Lipsitz S.R.,
- Sallan S.E.,
- et al.
- Lipshultz S.E.,
- Walker V.E.,
- Torres S.M.,
- et al.
- Shearer W.T.,
- Lipshultz S.E.,
- Easley K.,
- et al.
- Mertens A.C.,
- Yasui Y.,
- Neglia J.P.,
- et al.
- Chow H.H.,
- Li P.,
- Brookshier G.,
- Tang Y.
- The Perinatal Safety Review Working Group
- Pacheco S.E.,
- McIntosh K.,
- Lu M.,
- et al.,
- Women and Infants Transmission Study
- Feiterna-Sperling C.,
- Weizsaecker K.,
- Buhrer C.,
- et al.