Evaluation of Risk for Late Language Emergence after In Utero Antiretroviral Drug Exposure in HIV-Exposed Uninfected Infants
Download the PDF here |
Full Text below - The Pediatric Infectious Disease Journal Publish Ahead of Print
NIH - For Immediate Release: Thursday, July 18, 2013
Anti-HIV drugs in pregnancy not linked to children's language delays - full text/pdf below this press release
The combinations of anti-HIV drugs recommended for pregnant women do not appear in general to increase their children's risk for language delay, according to a study from a National Institutes of Health research network.
Children exposed to HIV in the womb and whose mothers received combinations of anti-HIV drugs during pregnancy were no more likely to have language delays than were children exposed to HIV in the womb and whose mothers did not receive these recommended treatments, the study found. In both groups, about 25 percent of the children had language delays by 2 years of age, suggesting that the delays were not associated with the anti-HIV drugs taken during pregnancy.
The findings allay concerns in the medical community that the drug combinations could affect the developing fetal brain in ways that cause language delays. Typically, these combination treatments include three or more drugs from at least two drug classes. For a woman who is HIV-positive and pregnant, recommended combination therapies treat the infection and greatly reduce the chance that the virus will spread to the fetus. Previous studies suggested that the drugs used to treat pregnant women might contribute to language delays in infants and toddlers, even those who remained HIV-negative.
However, the researchers concluded that one drug sometimes used in the combination treatments should be monitored. Children whose mothers received combination therapy containing the drug atazanavir were more likely to have language delays at 1 year of age than were the other children in the study. These children appeared to catch up to their peers. The researchers noted that these effects were not seen in children in the atazanavir group at age 2.
"Anti-HIV combination therapies do not appear to be linked to language delays, but it's prudent to monitor children exposed to HIV in the womb for signs of language delay," said study co-author George Siberry, M.D., of the Maternal and Pediatric Infectious Disease Branch of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), one of the NIH institutes that conducted the study. "Until there is a better understanding of what contributes to the delays, it's important to monitor the language development of children in this group carefully, and refer them for language therapy at the first sign of a delay."
Dr. Siberry collaborated with first author Mabel L. Rice, Ph.D., of the University of Kansas, Lawrence; Paige L. Williams, Ph.D., of the Harvard School of Public Health, Boston; Howard J. Hoffman, of the National Institute on Deafness and Other Communication Disorders (NIDCD), part of NIH; and colleagues at the National Institute of Mental Health (NIMH), also part of NIH; Albert Einstein College of Medicine, New York City; Northwestern University Feinberg School of Medicine, Chicago; Keck School of Medicine of the University of Southern California, Los Angeles; and Tulane University School of Medicine, New Orleans.
In addition to the NICHD, NIDCD and NIMH, seven NIH institutes and funding agencies supported the study: National Institute on Drug Abuse, National Institute of Allergy and Infectious Diseases, Office of AIDS Research, National Institute of Neurological Disorders and Stroke, National Heart, Lung and Blood Institute, National Institute of Dental and Craniofacial Research, and National Institute on Alcohol Abuse and Alcoholism.
The findings appear online in the Pediatric Infectious Disease Journal. (Full TEXT Below)
As part of an ongoing protocol known as the Pediatric HIV/AIDS Cohort Study, the researchers examined the effect of typical combination treatments.
Around the time of the children's first and second birthdays, the researchers administered questionnaires to caregivers, asking them to assess their child's language development. At age 1, the caregivers indicated whether their children could respond to a list of simple words, and whether they could point with a finger or make other gestures to communicate. The assessment for 2-year-olds asked caregivers to choose words the child had spoken or responded to, as well as other milestones of language use common to children at this age.
The researchers evaluated the language skills of nearly 800 children. They were able to collect assessments at study visits at either ages 1 or 2 for 70 percent of the children, and 40 percent were assessed at BOTH ages 1 and 2.
The researchers found that 26 percent of the 1-year-olds were significantly behind typical children their age with regard to emerging language skills. Communicating at age 2 requires a different set of skills, as words increasingly replace gestures. Yet, communication among 23 percent of 2-year-olds, too, was significantly delayed, compared with peers, the researchers found.
The researchers also conducted an analysis of individual drugs in the combination treatments the mothers received. Atazanavir is among the preferred medications for the treatment of pregnant women, and its use is increasing, the authors noted. Mothers of about 20 percent of the children in the study took this drug as part of their treatment. The researchers found that 1-year-olds who had been exposed to atazanavir were nearly twice as likely to experience language delays as those who were exposed to combination regimens without atazanavirs. However, they did not find evidence of increased language delays among the 2-year-olds whose mothers took atazanavir.
Evaluation of Risk for Late Language Emergence after In Utero Antiretroviral Drug Exposure in HIV-Exposed Uninfected Infants
The Pediatric Infectious Disease Journal Publish Ahead of Print
Rice, Mabel L. PhD; Zeldow, Bret MS; Siberry, George K. MD, MPH; Purswani, Murli MD; Malee, Kathleen PhD; Hoffman, Howard J. MA; Frederick, Toni PhD, MSPH; Buchanan, Ashley MS; Sirois, Patricia A. PhD; Allison, Susannah M. PhD; Williams, Paige L PhD; for the Pediatric HIVAIDS Cohort Study (PHACS)
Background: Combination antiretroviral (cARV) regimens are recommended for pregnant women with HIV to prevent perinatal HIV transmission. Safety is a concern for infants who were HIV-exposed but uninfected (HEU), particularly for neurodevelopmental problems, such as language delays.
Methods: We studied late language emergence (LLE) in HEU children enrolled in a US-based prospective cohort study. LLE was defined as a caregiver-reported score <= 10th percentile in any of 4 domains of the MacArthur-Bates Communicative Development Inventory for one-year-olds and as >=1 standard deviation below age-specific norms for the Ages and Stages Questionnaire for two-year-olds. Logistic regression models were used to evaluate associations of in utero cARV exposure with LLE, adjusting for infant, maternal, and environmental characteristics.
Results: 1,129 language assessments were conducted among 792 one- and two-year-old children (50% male, 62% black, and 37% Hispanic). Overall, 86% had in utero exposure to cARV and 83% to protease inhibitors. LLE was identified in 26% of one-year-olds and 23% of two-year-olds, with higher rates among boys. In adjusted models, LLE was not associated with maternal cARV or ARV drug classes in either age group. Among cARV-exposed one-year-olds, increased odds of LLE was observed for those exposed to atazanavir (aOR=1.83, 95% CI=1.10-3.04), particularly after the first trimester (aOR=3.56, p=0.001), compared with atazanavir-unexposed infants. No associations of individual ARV drugs with LLE were observed among two-year-olds.
Conclusions: In utero cARV exposure showed little association with LLE, except for a higher risk of language delay observed in one-year-old infants with atazanavir exposure.
Early language acquisition appears during the first year as gestures (showing intent to communicate), rudimentary understanding of spoken language, and first spoken words at 12-18 months, with rapid vocabulary growth from two years onward. Delayed onset of language can be described as late talking and/or late understanding of spoken language, relative to age expectations. The term "Late Language Emergence" (LLE) defines a delayed onset of language as low performance, relative to age norms, for first word use or understanding, or both, in the general population of young children, controlling for multilingual exposures1. In the general population, predictors of early gestural communication include maternal, environmental, and child factors1-3, and are not necessarily the same as predictors of early verbal productions2, perhaps due to different neurologic requirements for gestures versus spoken words. Recent studies highlight the vital importance of the prenatal period on infants' subsequent development4,5. For example, maternal alcohol use during pregnancy is known to influence children's neurodevelopment, including risk for language impairments6,7.
In a previous study, we found that school age children with perinatal HIV infection (PHIV) and those HIV-exposed but uninfected (HEU) had a higher risk for language impairments compared to children without HIV exposure8. The risk for language impairments was higher for children with concurrent cognitive or hearing impairments, but the PHIV group did not differ from the HEU group. Among children with PHIV, those with poorly controlled HIV or with cognitive or hearing impairments had three times the risk for language impairments compared to those with controlled HIV and those with no cognitive or hearing impairments.
For children born more recently to mothers with HIV, in utero exposure to combination antiretroviral drugs (cARV) is common9, resulting in very low risk for HIV transmission to the infant. Increasingly, newer generations of drugs within existing classes of antiretrovirals (ARV), and newer classes of ARVs are being used as part of these cARV regimens. Among studies that have been conducted to evaluate neurodevelopmental functioning in children born to mothers with HIV infection, none have specifically addressed language impairment, despite the fact that language and hearing impairments may reflect some of the most sensitive indicators of ARV-associated toxicities10. Hence, as these regimens evolve over time, it is important to understand whether in utero exposure to cARV, or to particular drugs in the regimen, increases a child's risk for language impairment. If so, it is important to know if the risk is evident early in development given the known benefits of early intervention10.
A recent study11 of HEU infants ages 9-15 months reported no consistent association of individual ARVs with any domain of the Bayley Scales of Infant and Toddler Development,
Third Edition (Bailey-III)12, except for the Language domain. One drug, atazanavir, was associated with lower scores on the Language domain in both primary analyses and in sensitivity analyses omitting the intermediaries of prematurity and small for gestational age (SGA). This study was conducted with the same cohort under investigation here. Further investigation is needed to determine if the association of atazanavir and early language acquisition is evident with more detailed measurements of early language acquisition and if exposure to atazanavir predicts risk of LLE when controlled for other factors, such as cognitive delay and hearing impairment. Selective impairment of language acquisition in infancy would help clarify whether language impairment is apparent from onset or if it reflects a cumulative effect apparent only later in development. Further evidence of an association with an individual drug in cARV could clarify possible mechanisms of drug effects early in development. The overall objective of the current study was to evaluate further the safety of maternal ARV use on language acquisition in the first two years of life, by considering the association of LLE with cARV, ARV drug classes, and individual ARV drugs, controlling for other relevant caregiver, environmental, and child factors.
Study Population and Language Functioning:
The CDI and ASQ were obtained from groups with similar backgrounds (Table 1). For the CDI, 51% were female, 70% black, and 33% Hispanic; 71% had annual household income below $20,000; 98% of caregivers were the infants' biological mothers and 32% of caregivers had not graduated from high school. The percent of children from the Dynamic cohort was higher for those assessed by the CDI (71%) than for the ASQ (51%). Within both age groups, 43% were exposed to cARV in the first trimester, increasing to 86% by the third trimester. For the CDI, 12% were exposed in utero to an NNRTI, and 83% to a PI. ARV exposures were similar for two-year-olds.
Valid data were obtained for 577 children on the CDI at the 12-month visit (age range: 9.0-18.9 months) and for 552 children on the ASQ at the 24-month visit (age range: 20.1-34.2 months) (see Figure). Some children (n=215) in the Static cohort were enrolled into the study at the 24 month visit and did not have the 12-month CDI assessment; others (n=240) had not yet reached 24 months of age and had only the 12 month visit. A total of 337 children had both the CDI and ASQ assessments. Mean (standard deviation [SD] and 95% confidence intervals [CI]) percentiles on CDI subtests were: Phrases Understood, 54.1 (27.7, 51.8-56.4); Vocabulary Comprehension, 44.5 (30.0, 42.0-47.0); Word Production, 47.6 (21.3, 45.9-49.3); and Total Gestures, 46.9 (28.3, 44.6-49.2). The mean percentiles were significantly below the population norms of 50 for the latter three CDI subtests, but above norms for the Phrases Understood subtest. The mean (SD) summary score was 193.0 (83.9, 186.1-199.9). A total of 149 (26%, 22.3-29.6%) one-year-olds met the criteria for LLE; 6% (4.1-8.1%) (n=34) were classified with severe LLE. For the 552 ASQ assessments, the mean (SD) Total Score was 45.3 (15.6, 44.0-46.6). Among these 2-year olds, 128 (23%, 19.7-26.9%) were classified with LLE; 78 (14%, 11.3-17.3%) with severe LLE. Among the 337 children with both CDI and ASQ assessments, the correlations between CDI subtests at 12 months and the ASQ Total Score were as follows (Spearman's ): Phrases Understood (0.29), Vocabulary Comprehension (0.29), Total Gestures (0.32), Word Production (0.27), and Summary Score (0.37), all of which were statistically significant.
Association of LLE with In Utero ARV Exposure and Other Factors:
Among 1-year olds, boys had higher odds of LLE (adjusted odds ratio [aOR]=1.54, p=0.03) than girls (Table 2). Infants whose caregiver had a verbal IQ <85 had over twice the odds of LLE (aOR=2.10, p=0.003). Other variables associated with higher odds of LLE and retained in the adjusted model at the 12-month visit were multilingual exposure, combination neonatal ARV prophylaxis (vs zidovudine alone), and low maternal CD4% prior to labor and delivery; also retained was maternal VL>400 copies/mL early during pregnancy, associated with lower odds of LLE. Boys also had higher odds of LLE at the 24-month visit (aOR=2.00, p=0.002). Low maternal CD4% during pregnancy and caregiver health problems were associated with increased odds of LLE at 24 months, but infants whose caregivers reported that poor health limited their daily activities had lower odds of LLE. At both the 12-month and 24-month visits, there was no association of LLE with cARV or ARV drug class in either unadjusted or adjusted models (Table 3).
Association of LLE with Exposure to Individual ARVs:
Among 1-year old children with in utero cARV exposure, the odds of LLE were increased for those exposed to atazanavir (aOR=1.83, p=0.02) but the association was not significant at 24 months (Table 3). There was also an increased odds of LLE associated with saquinavir exposure at 12 months (aOR=2.74, p=0.03) but at 24 months the effect changed directions and was not statistically significant (aOR= 0.15, p=0.06). Drug exposure differed for atazanavir and saquinavir; 21% of 1 year-olds and 19% of 2 year-olds had in utero exposure to atazanavir compared to only 5% of both age groups for saquinavir. When examining the timing of atazanavir exposure, there was a threefold increase in odds of LLE associated with later atazanavir use (first use during the second/third trimester) compared to absence of atazanavir exposure (aOR=3.56, p=0.001). Among two-year-olds with in utero cARV exposure, there were no significant associations between LLE and any individual ARV drug based on either overall exposure (Table 3) or first trimester exposure (data not shown).
The association of LLE with atazanavir based on the CDI remained in sensitivity analyses adjusting for preterm birth and SGA (aOR=1.83, p=0.02) but was attenuated when adjusted for cognitive delay and hearing problems (aOR=1.67, p=0.06). There was no association of LLE with first trimester exposure to any individual ARV (data not shown). When ASQ analyses were restricted to those between 21 and 27 months of age (corresponding to the visit window for the ASQ; n=485) results were essentially unchanged. In sensitivity analyses accounting for study site effects on the CDI, we observed weak within-site correlations in LLE at 1 year (r=0.03) and the associations between LLE and ARV exposures remained similar in magnitude and significance. The association of atazanavir with LLE remained significant (aOR=1.86, p=0.04) while the association with saquinavir was no longer significant (aOR=2.95, p=0.08).
Among 2-year olds with in utero cARV exposure, there were no significant associations between LLE and any individual ARV based on either any exposure (Table 3) or first trimester exposure (data not shown).
Association of Language Scores with In Utero ARV Exposure and Other Factors:
A summary of GEE linear regression models for the CDI summary score for 1-year olds and the ASQ Total Score for 2-year olds is shown in Table 4. In both unadjusted models (not shown) and after adjustment for potential confounders, there was no association of in utero cARV exposure or ARV class exposure, either at any time during pregnancy or during the first trimester, on mean scores within either age group. In addition, there was no significant difference in mean scores by regimen (cARV with PI vs. cARV without PI vs. non-cARV regimens, data not shown). Exposure to a cARV regimen without PI was associated with a marginally higher mean Total Score on the ASQ (6.7 points higher, p=0.07) as compared to a non-cARV regimen.
In evaluation of separate subscales of the CDI (data not shown), there was a significant decrease of 7.7 points (p=0.02) for the Total Gestures subscale among 1-year olds exposed to PIs; this decrease in mean score was observed in those with exposure to PIs later in pregnancy (second or third trimester) compared to those not exposed to PIs. For the Word Production subtest, atazanavir exposure was associated with a significantly lower score in both unadjusted and adjusted models (adjusted mean decrease = 4.9 points, p = 0.04). In contrast, no association was observed with the Total Gestures subscale for atazanavir or any other ARV drug. Among cARV-exposed 2-year olds, no significant associations between individual ARV agents and mean ASQ Total Score were observed after adjustment for other covariates.
Most of the covariates identified as predictive of LLE in the logistic regression models (Table 2) were associated with the continuous summary scores (Table 4). Boys had lower mean language scores than girls and significantly higher odds of LLE. Maternal/caregiver performance IQ <85 was associated with significantly lower CDI summary scores. Maternal immune status during pregnancy (low CD4% or detectable VL) was associated with the CDI summary score and the ASQ Total Score consistent with the logistic regression models, although the relationship with viral load was counterintuitive (early maternal VL>400 copies/mL was associated with higher mean scores). Limitations in caregiver health, including fatigue and difficulty caring for the child, were associated with lower mean scores on the CDI and ASQ and with increased odds of LLE. Maternal alcohol use during pregnancy was associated with a 21.3 point decrease in mean CDI summary score after adjustment for other factors but was not a risk factor for LLE within either age group.
An earlier study of school age children enrolled PHACS found that children perinatally exposed to HIV were at high risk for language impairments (36%), and the risk was similar for children with and without HIV infection. The results of the current study extend the concern for language delay to HEU infants and toddlers. LLE was identified in 26% of one-year-olds and in 23% of two-year-olds, prevalence estimates somewhat higher than reported in previous studies in the general population1,2. Reilly et al 2 reported 20% with LLE based on the CDI at age 2, for a sample of children which included those with multilingual exposure (6.6%, similar to the 6.8% in this study). Zubrick et al1 reported 13% with LLE on the ASQ at age 2, excluding children with multilingual exposure.
The associations of LLE with some demographic and maternal characteristics were generally consistent with previous studies of infants in the general population and of those enrolled in SMARTT11. Boys had lower language scores and were more likely to be identified with LLE at both 12 and 24 months of age 1,2. Multilingual exposure has been associated with LLE in 2-year olds 2. We observed that lower maternal/caregiver verbal ability was associated with higher risk of LLE for 1-year old infants, while Feldman et al3 found lower maternal education to be associated with higher scores on the CDI in 1-year olds and noted similar outcomes in other studies. The apparent discrepancy may be attributable to different patterns of prediction for the low end of performance versus the full range of performance in one-year-old infants, or for greater sensitivity of maternal verbal intellectual functioning as a predictor variable. Although maternal alcohol use was associated with a decrease in mean CDI score at 12 months, it was not a risk factor for LLE within either age group. This may be because adverse effects of maternal alcohol use on children's early language acquisition are more likely to occur for children with cognitive impairment or hearing loss, factors controlled for in the LLE analyses. Maternal ARV use during pregnancy showed little association with LLE, except that atazanavir was associated with significantly increased odds of LLE at age 12 months: the effects
were more evident for word productions than for gestural communications. This finding is of particular importance because use of this medication during pregnancy is increasing (from 2% in 2004 to 20% in 2009)9 and atazanavir is recommended as a preferred ARV drug for use in pregnancy20. Increased usage of atazanavir is a reflection of its once daily dosing regimen, low pill burden and lower rate of metabolic, particularly lipid, adverse effects. Further, an association of in utero atazanavir exposure with lower scores on the Bayley-III12 language domain was observed in a separate investigation of neurodevelopmental outcomes at age 12 months among 374 infants from the same cohort studied here11. This association should be examined in other cohorts of HEU infants.
Atazanavir crosses the placenta at an estimated ratio of 20%21. In pregnant women, as in non-pregnant adults and children, atazanavir (usually in combination with boosting ritonavir) is commonly associated with elevated serum unconjugated bilirubin22 . Normally, fetal unconjugated bilirubin crosses the placenta to maternal circulation and then is conjugated and eliminated by the mother's liver. However, atazanavir-related impairment of maternal bilirubin conjugation and elevated maternal unconjugated bilirubin levels may result in higher levels of fetal unconjugated bilirubin levels through much of the latter half of pregnancy. Neonatal hyperbilirubinemia at varying levels has been associated with risk of developmental delay in multiple domains, including language23-25. Neurodevelopmental problems associated with neonatal hyperbilirubinemia in some of these studies include milder and variable deficits (unlike classic kernicterus) and may be detected in the first year of life but often are not in evidence at older ages. Fetal and neonatal elevations in bilirubin associated with maternal atazanavir use20 are lower than the extreme neonatal hyperbilirubinemia linked to neurologic injury. Neither maternal nor infant bilirubin measurements were collected in our study. It is possible that chronic fetal exposure to milder elevations of bilirubin during the second and third trimesters could result in subtle neurotoxicity detected later as LLE in these language evaluations. This is further supported by data from animal studies of developing neurons suggesting potential neurotoxicity at much lower bilirubin levels,26 mediated, in part, through mitochondrial toxicity27. These mechanistic hypotheses should be tested.
The absence of association of atazanavir exposure with LLE at 24 months of age is reassuring, suggesting that the effect is transient, mild or compensated with ongoing development. In our sample, among 83 one-year-olds with LLE and available for follow-up assessment at two years, 34 (41%) still had LLE (data not shown). Some of those with resolution of LLE may have improved language functioning due to receipt of early intervention services, which increased from 6% to 10% between ages one and two years in our cohort. An alternative explanation is that the ASQ is not as sensitive as the CDI for detecting associations with individual ARV exposures, a limitation of our study protocol. The CDI assesses four domains, and multiple items on each domain, whereas the ASQ provided a total score from six items; the greater amount of information from the CDI could result in greater variation in the individual scores thereby increasing sensitivity as compared to the ASQ. Longer-term language outcomes in this cohort and investigation of associations of atazanavir exposure with language, hearing, and other developmental outcomes in other cohorts will be essential.
Our data suggest careful clinical screening for LLE in infants who are HEU in the first two years of life. As new ARV agents are approved and become more widely used, in combination with NRTIs or other regimens, continued evaluation of language functioning and other neurodevelopmental outcomes are warranted for ensuring the safety of treatments for pregnant women with HIV.