HIV Articles  
Back 
 
 
In utero nucleoside reverse transcriptase inhibitor exposure and signs of possible mitochondrial dysfunction in HIV-uninfected children
 
 
  AIDS:Volume 21(8)11 May 2007p 929-938
 
Brogly, Susan Ba,b; Ylitalo, Nathaliea,c; Mofenson, Lynne Md; Oleske, Jamese; Van Dyke, Russellf; Crain, Marilyn Jg; Abzug, Mark Jh; Brady, Michaeli; Jean-Philippe, Patrickj; Hughes, Michael Da,b; Seage, George R IIIa,c From the aCenter for Biostatistics in AIDS Research, Harvard School of Public Health, Boston, Massachusetts, USA
bDepartment of Biostatistics, Harvard School of Public Health, Boston, Massachusetts, USA
cDepartment of Epidemiology, Harvard School of Public Health, Boston, Massachusetts, USA
dPediatric, Adolescent, and Maternal AIDS Branch, National Institute of Child Health and Human Development, Rockville, Maryland, USA
eDepartment of Pediatrics, University of Medicine & Dentistry of New Jersey, Newark, New Jersey, USA
fDepartment of Pediatrics, Tulane University Health Science Center, New Orleans, Louisiana, USA
gDepartment of Pediatric Infectious Diseases, University of Alabama at Birmingham, Birmingham, Alabama, USA
hDepartment of Pediatrics (Infectious Diseases), University of Colorado School of Medicine and the Children's Hospital, Denver, Colorado, USA iChildren's Hospital, Columbus, Infectious Disease Department, Columbus, Ohio, USA
jHenry M. Jackson Foundation at Division of AIDS, National Institute of Allergy and Infectious Diseases, Bethesda, Maryland, USA.
 
"....there is equivocal evidence of an etiologic association between in utero NRTI exposure and MD (mitochondrial dysfunction) in HIV-uninfected children born of HIV-infected women. We conducted a study to estimate the independent association of in utero NRTI exposure and unexplained signs consistent with possible MD in HIV-uninfected children enrolled in the Pediatric AIDS Clinical Trials Group (PACTG) protocols 219 and 219C.....Twenty of 1037 children included in the primary had possible MD for a prevalence of 1.9% [95% confidence interval (CI), 1.2-3.0]; rates were 1.8% (95% CI, 1.1-2.9) and 2.9% (95% CI, 0.6-8.4) among 933 in utero NRTI exposed children and 104 ARV unexposed children, respectively....In summary, we did not observe an association between overall in utero ART exposure and possible MD. We did observe higher odds of first in utero exposure to 3TC, and to ZDV/3TC, in the third trimester among HIV-uninfected children with unexplained signs consistent with possible MD. This may demonstrate an etiologic association between third trimester initiation of certain NRTI and MD. Alternatively, our findings may be confounded by other causes of fetal injury such as higher maternal HIV RNA levels, psychoactive drug exposure, or late prenatal care. ARV prophylaxis is a critical component of the prevention of MTCT of HIV infection as recommended by the US Public Health Service [32]. The benefits of ARV prophylaxis outweigh any known risks to date. Many ARV drugs and therapeutic classes are available, and it is important to identify particular drugs or combinations that confer greater or lesser risk to the developing fetus. Additional studies that rigorously assess MD and allow for better control of confounding are needed....."
 
Abstract

Background: There is equivocal evidence of in utero nucleoside reverse transcriptase inhibitor (NRTI) exposure and the occurrence of mitochondrial dysfunction (MD) in HIV-uninfected children born of HIV-infected women.
 
Methods: The primary analysis included 1037 HIV-uninfected children born in 1991-2002 and enrolled in Pediatric AIDS Clinical Trials Group protocols 219/219C. Possible cases with unexplained signs of MD according to the Enquete Perinatale Francaise criteria were identified through retrospective review. Associations between overall in utero NRTI exposure, and trimester of first in utero NRTI exposure and possible MD were estimated with exact logistic regression.
 
Results: Cases (n = 20) were significantly more likely to be male and to be born in earlier years than non-cases (n = 1017). There was no association between overall in utero NRTI exposure and MD. In unadjusted models there were higher odds of first in utero exposure in the third trimester to lamivudine (3TC) [odds ratio (OR), 3.76 versus 3TC unexposed; 95% confidence interval (CI), 1.09-11.78] and to zidovudine (ZDV) and 3TC in combination (ZDV/3TC) (OR, 3.29 vs. ZDV/3TC unexposed; 95% CI, 0.96-10.25) among cases than noncases. When adjusted for year of birth the odds of first exposure in the third trimester to 3TC (OR, 10.57; 95% CI, 1.93-75.61) and ZDV/3TC (OR, 9.84; 95% CI, 1.77-71.68) were significantly higher among cases than non-cases. Incomplete data precluded control of possible confounding by maternal viral load and psychoactive drug use.
 
Conclusions: Our study suggests that first exposure to 3TC or ZDV/3TC in the third trimester may be associated with the occurrence of possible MD. Further studies that rigorously assess MD and better control confounding are needed.
 
Introduction
The use of combination antiretroviral therapies (ARV) during pregnancy and the neonatal period can reduce the rate of mother-to-child transmission (MTCT) of HIV from 25% [1,2] to less than 2% [2]. However, prevention of MTCT of HIV entails in utero and neonatal exposure to drugs with unknown toxicity, and there is some evidence that in utero nucleoside reverse transcriptase inhibitor (NRTI) exposure may cause mitochondrial dysfunction (MD) in a small number of HIV-uninfected infants [3].
 
The toxic effect of NRTI on mitochondrial DNA replication in various tissues in HIV-infected adults is established [4]. Inhibition of DNA polymerase-, the enzyme responsible for mitochondrial DNA replication in eukaryotes, by the pharmacologically active NRTI triphosphate is believed to be a mechanism of NRTI-induced mitochondrial toxicity, and evidence of multiple complex mitochondrial pathways is emerging [5,6].
 
In 1999, the Enquete Perinatale Francaise (French Pediatric Cohort, EPF) investigators described signs of MD in eight of 1754 HIV-uninfected children born of HIV-infected women, four of whom had been exposed to zidovudine (ZDV) in utero, and four to ZDV in combination with lamivudine (3TC) [7]. In a subsequent EPF study, rates of persistent MD of 0.3% (7/2644) in HIV-uninfected children exposed to NRTI in utero, and of 0% (0/1748) in HIV-uninfected children unexposed to NRTI were observed [3]. Further, the relative risk of possible or established MD was 2.5 times higher in children with in utero exposure to combination NRTI versus ZDV alone. However, these estimates were unadjusted for potential confounders, and were not differentiated by duration, timing, or specific NRTI exposures.
 
In contrast to the EPF findings, in a review of more than 20 000 children born of HIV-infected women followed in five American cohorts - about half of whom were exposed to ARV in utero, primarily ZDV alone - none of the 30 deaths among HIV-uninfected children were attributed to MD [8]. Likewise, no clinical manifestations of MD were found in 1008 HIV-uninfected children exposed to ARV in utero in the European Collaborative Study [9], and no difference in the risk of adverse neurological events was identified in an African study of 1501 infants exposed to ZDV in combination with 3TC (ZDV/3TC) versus placebo beginning at 36 weeks' gestation [10]. A recent African study observed no difference in the rate of increased lactate levels - a possible consequence of NRTI-induced mitochondrial toxicity - in children first exposed to ZDV or to ZDV/3TC at 32-36 weeks' gestation and children unexposed to NRTI [11]. Another study reported a significant increase in the risk of hyperlactatemia per increase in week of exposure to the NRTI, didanosine (ddI), but no association with ZDV or 3TC or ZDV/3TC [12].
 
Cases of possible MD in HIV-uninfected infants exposed to ARV in utero have been reported outside the EPF. In the USA severe MD was confirmed by biopsy and enzyme studies in an infant exposed to ZDV/3TC and to the non-NRTI (NNRTI), nevirapine, from 33 weeks gestation until delivery at 37 weeks, and to ZDV/3TC during the first 2 weeks of life [13]. In Italy an HIV-uninfected child exposed to ZDV in utero starting at 4 months gestation had biopsy-confirmed MD without identification of an etiology other than in utero ZDV exposure [14]. Finally, in the Netherlands MD was suspected but not confirmed by laboratory investigations in an HIV-uninfected infant exposed throughout gestation to two NRTI, stavudine (d4T) and ddI, and the protease inhibitor (PI) nelfinavir [15].
 
In summary, there is equivocal evidence of an etiologic association between in utero NRTI exposure and MD in HIV-uninfected children born of HIV-infected women. We conducted a study to estimate the independent association of in utero NRTI exposure and unexplained signs consistent with possible MD in HIV-uninfected children enrolled in the Pediatric AIDS Clinical Trials Group (PACTG) protocols 219 and 219C.
 
Discussion TOP
 
Clinical manifestations associated with MD appear to occur more frequently in infants born of HIV-infected women than in the general pediatric population, and there is some evidence that this may be due to in utero NRTI exposure [3]; non-human studies also support a possible etiologic association [21-23]. We did not observe a significant association between overall in utero NRTI exposure and clinical manifestations suggestive of possible MD. However, increased odds of first in utero 3TC and of first in utero ZDV/3TC exposure in the third trimester among HIV-uninfected children with signs consistent with possible MD were identified. We were unable to distinguish whether the increased odds was due to 3TC or to ZDV/3TC.
 
Others also have reported the occurrence of possible or definite MD associated with initiation of maternal ARV late in pregnancy. An American infant with confirmed MD first exposed to ZDV/3TC and nevirapine at 33 weeks' gestation had no identified genetic or virologic predisposition to MD other than in utero NRTI exposure [13]. In addition, three of eight cases of MD in the EPF initial case series also were first exposed to ZDV/3TC in the third trimester, and another case was first exposed to ZDV in the third trimester [7]. First exposure to 3TC and to ZDV/3TC in the third trimester may lead to MD in some children; all but one of our possible cases had neurological signs of MD. Studies of antiepileptic drug use during pregnancy have suggested that in utero exposure during the third trimester may partially account for the intellectual impairment and decreased head circumference observed in exposed children [24,25]. In our study only first exposure to 3TC or ZDV/3TC in the third trimester was significantly associated with possible MD; no significant association was detected for first exposure to 3TC or ZDV/3TC in the first or second trimester. Due to the limited use of other ARV in combination, we cannot rule out an etiologic association of other combinations and timings of in utero NRTI exposure and possible MD. Further, the CI of our estimates are wide due to the small number of cases.
 
It is possible that late initiation of ARV in the pregnant mother is confounded by other determinants of fetal injury. In utero cocaine exposure has been associated with cognitive impairments [26] and cardiovascular abnormalities [27], conditions included in our case definition. A study of the neurodevelopment of HIV-uninfected children exposed to ARV in utero and of HIV and ARV-unexposed children demonstrated a strong association between in utero exposure to cocaine, heroin, and tobacco exposure and low developmental scores [28]. Poor access to ARV, which may relate to late prenatal care, has also been observed in drug users [29]. We were unable to control adequately for psychoactive drug exposure, and we did not have data on timing of initiation of prenatal care. None of the children with MD in the EPF were exposed to illicit psychoactive drugs in utero [3].
 
The association between first exposure to 3TC in the third trimester and possible MD may be confounded by high maternal viral loads, which were observed among some of our cases first exposed to 3TC, and to ZDV/3TC, in the third trimester. A proposed mechanism may be a high level of cytokines in women with more advanced HIV disease and late ARV initiation; these cytokines may adversely affect placental function and injure the fetus. In addition, there is evidence that HIV-uninfected infants born of HIV-infected women have depletions in mitochondrial DNA in the absence of in utero ARV exposure [30].
 
Seventeen (1.8%) of 933 in utero NRTI exposed children included in our primary analysis met the EPF definition of possible MD, which is six times higher than the rate of MD of 0.3% observed among in utero NRTI exposed children in the EPF [3]. Three children without in utero ARV exposure had clinical manifestations of possible MD; our rate of possible MD in the NRTI-unexposed children was 2.9% versus 0% in the EPF cohort [3]. Our cases were identified through retrospective review of clinical signs, which likely has poor positive predictive value [3]. Unlike the EPF study, we did not have mitochondrial histological or enzymological studies or cerebral magnetic resonance imaging necessary for more definitive case identification. In an effort to reduce misclassification we excluded children whose only sign of MD was cognitive developmental delay - children whom the reviewing clinicians were less confident classifying as possible cases - from our primary analysis, and indeed, the estimated ORs were increased from those estimated when all cases were included. Our cases are perhaps best described as having a constellation of signs with unexplained etiology that could be consistent with MD according to the EPF criteria. In addition, our cases had abnormalities evident before 3 years of age and cases with later presentation of abnormalities were not included.
 
The higher odds of first exposure to 3TC, and to ZDV/3TC, were driven by six cases born in 1996, 1997, and 1999. Although we cannot explain the temporality of our findings, consistent results were obtained with various parameterizations of year of birth. A small proportion of HIV-uninfected children born of HIV-infected women in the USA enrolled in protocols 219 and 219C, and we cannot assess whether there was selective enrollment with respective to in utero NRTI exposure and/or possible signs of MD. Selection bias also could have arisen from differential discontinuation of follow-up.
 
The male predominance among our cases is concordant with previous reports. Only three of 12 cases reported in the EPF study were female (S. Blanche, personal communication 2005). A study of genetic mitochondrial disorders in the general pediatric population also demonstrated a male predominance (58% versus 42%) [31].
 
In summary, we did not observe an association between overall in utero ART exposure and possible MD. We did observe higher odds of first in utero exposure to 3TC, and to ZDV/3TC, in the third trimester among HIV-uninfected children with unexplained signs consistent with possible MD. This may demonstrate an etiologic association between third trimester initiation of certain NRTI and MD. Alternatively, our findings may be confounded by other causes of fetal injury such as higher maternal HIV RNA levels, psychoactive drug exposure, or late prenatal care. ARV prophylaxis is a critical component of the prevention of MTCT of HIV infection as recommended by the US Public Health Service [32]. The benefits of ARV prophylaxis outweigh any known risks to date. Many ARV drugs and therapeutic classes are available, and it is important to identify particular drugs or combinations that confer greater or lesser risk to the developing fetus. Additional studies that rigorously assess MD and allow for better control of confounding are needed.
 
Results
Twenty of 1037 children included in the primary had possible MD for a prevalence of 1.9% [95% confidence interval (CI), 1.2-3.0]; rates were 1.8% (95% CI, 1.1-2.9) and 2.9% (95% CI, 0.6-8.4) among 933 in utero NRTI exposed children and 104 ARV unexposed children, respectively.
Cases were significantly more likely to be male and to be born in an earlier year than noncases (Table 2). There were no significant differences in race, birth weight (< 2500 grams), prematurity (< 37 weeks gestation at birth), neonatal ARV prophylaxis [including neonatal ZDV monotherapy (75.0% versus 82.4%) or ZDV/3TC use (0.0% versus 4.3%)], mean log10 peak third trimester or delivery maternal HIV RNA copy number, or in utero tobacco, alcohol or other psychoactive drug exposure; however, information on maternal HIV RNA levels and in utero drug exposure was unavailable for 37% and 24% of mothers, respectively.
 
The maternal ARV regimens included combination ARV of NRTI, PI or NNRTI for eight (40.0%) cases and 449 (44.1%) noncases; three NRTI for one (5.0%) case and 15 (1.5%) noncases; two NRTI for two (10.0%) cases and 117 (11.5%) noncases; and a single NRTI for six (30.0%) cases and 336 (33.0%) noncases. Three of 20 cases and 101 of 1017 noncases were unexposed to any ARV in utero. There was no significant association between overall in utero NRTI exposure - for all NRTI considered together or specific NRTI - and possible MD in univariate or multivariate models (Table 3). Among cases and noncases exposed to ARV in utero, there was no significant difference in the median duration of in utero NRTI (22.9 versus 21.7 weeks), ZDV (22.6 versus 20.1 weeks), or 3TC (9.3 versus 21.8 weeks) exposure. In addition to the NRTI considered (Table 3 and Table 4), two cases also were exposed to d4T and abacavir in utero.
 
There were higher odds of first in utero 3TC and ZDV/3TC exposure in the third trimester among cases than non-cases in unadjusted analyses (Table 4). When adjusted for year of birth, the odds of first exposure to 3TC and to ZDV/3TC in the third trimester were higher in cases than non-cases (3TC: OR, 10.57; 95% CI, 1.93-75.61; ZDV/3TC: OR, 9.84; 95% CI, 1.77-71.68). However, no significant association was detected for first exposure to ZDV alone in the third trimester (OR, 1.72; 95% CI, 0.30-12.17). In secondary analyses where all 42 possible cases and all non-cases were included (Fig. 1), the adjusted ORs were 2.95 (95% CI, 0.91-10.90) and 3.46 (95% CI, 1.09-12.61) for overall in utero exposure to 3TC and to ZDV/3TC, respectively and 5.87 (95% CI, 1.61-22.31) and 5.59 (95%CI, 1.52-21.42) for first in utero exposure to 3TC and to ZDV/3TC in the third trimester, respectively.
 
The six cases first exposed to 3TC in the third trimester (Table 4) also were first exposed to ZDV/3TC in the third trimester; three had prior exposure to ZDV alone in the first or second trimester with initiation of 3TC in the third trimester. Although not statistically significant, peak maternal viral loads in the third trimester were higher among five cases first exposed to ZDV/3TC in the third trimester (maternal viral load was unavailable for one of the six cases) than among non-cases and cases with other distributions of in utero NRTI exposure (median log10 HIV RNA copies/mL 4.0 versus 3.4 and 2.9, respectively); and a higher proportion of the six cases was exposed to psychoactive drugs in utero (33.3% versus 15.9% and 7.1%, respectively). Psychoactive drug exposure, excluding alcohol, was higher in the three ARV unexposed cases than in non-cases and exposed cases.
 
Methods
Study population

PACTG protocol 219 enrolled and followed HIV-infected and HIV-uninfected perinatally exposed infants and children at clinics across the USA including Puerto Rico from May 1993 through August 2000 to study the long-term effects of in utero ARV exposure and the complications of pediatric HIV infection. Children currently or previously enrolled in a PACTG clinical trial and children whose mothers were enrolled in a PACTG clinical trial during pregnancy were eligible for participation. In September 2000 a revised protocol was initiated, PACTG 219C, and the eligibility criterion mandating enrollment in a PACTG clinical trial was removed. Participating institutions obtained approval from their respective institutional review boards for human research, and the child's parent or guardian provided written informed consent.
 
A total of 1732 HIV-uninfected children enrolled in protocol 219 or 219C through 31 August 2003; 1049 children enrolled in protocol 219, of whom 490 (46.7%) subsequently enrolled in protocol 219C, and 683 children enrolled in 219C only. The study population was restricted to 1220 HIV-uninfected children who enrolled in 219 or 219C prior to 2 years of age and had completed 1 year of follow up as of 31 August 2003. This age restriction was used to identify children who developed signs of possible MD by 3 years of age to be similar to follow up of children in the EPF [3], and because many children had not yet been followed beyond 3 years of age. A minimum of 1 year of follow-up was required to assess the persistence or resolution of signs of possible MD. As expected children in the study population were younger at enrollment than the 512 children not included (median age, 6.8 months; range, birth to 2 years; versus median age, 10.5 months; range, 1.3 weeks-14.9 years; P < 0.0001), enrolled earlier (median year of enrollment 1998 versus 2001; P < 0.0001), were followed longer before 3 years of age (median follow up, 2.0 years; range, 1-3 years; versus 5.8 months, range; 0 to < 1 year; P < 0.0001), and were more likely to have participated in protocol 219. There was no significant difference in sex, race/ethnicity, or the proportion of children known to have been exposed to ARV in utero in the two groups. However, children in the study population were significantly less likely to have been exposed to PI (27.3% versus 33.2%, P < 0.001).
 
Data collection
In protocol 219 visits occurred every 6 months for HIV-uninfected children less than 2 years of age, and annually thereafter until protocol completion, withdrawal, loss to follow up, or death. In protocol 219C visits occurred every 3 months for HIV-uninfected children less than 1 year of age, and annually thereafter until protocol completion, withdrawal, loss to follow up, or death. At enrollment and follow-up in 219 and 219C clinical diagnoses and diagnostic test results were recorded, and physical and neurological examinations were performed.
 
Neuropsychological testing was performed every 6 months to 3 years of age (protocol 219) or 2 years of age (protocol 219C) and annually thereafter using the Bayley Scales of Development I and II (1969 and 1993), the McCarthy Scales of Children's Intelligence (1972), and the Wechsler Preschool and Primary Scale of Intelligence Revised (1989) as age appropriate. Information on in utero ARV exposure, birth weight, and gestational age at birth was obtained at enrolment when available, and was supplemented with data collected in other PACTG studies (perinatal protocols 076, 082, 185, 249, 250, 255, 316, 332, 353, 354, 358, and 386). Maternal HIV RNA levels in the third trimester and at delivery were obtained from the perinatal protocols since these data were not collected in protocol 219 or 219C.
 
Identification of children with signs consistent with possible MD
To be included as a possible case of early MD children must have fulfilled the definition of MD proposed by the EPF investigators before 3 years of age [3]; additional signs, if present, that occurred after 3 years of age were included as supporting evidence for cases with signs before 3 years of age. To exclude children whose presentation could be explained by an etiology other than MD (e.g., congenital abnormality, severe prematurity), or who did not meet the EPF case definition of possible MD on further review (e.g., lack of persistent abnormal findings), a retrospective review of the medical histories recorded on the protocol case report forms was performed by clinicians (MA, MB, MC, LM, JO, PJ-P, RVD, NY) blinded to in utero ARV exposure. Of 110 children with a constellation of symptoms of unknown etiology that could be consistent with MD, 63 were classified as inconsistent because another etiology likely explained the signs of possible MD, the condition(s) subsequently resolved, or the child was deemed not to fulfill the case definition by 3 years of age; two had insufficient data to evaluate; and 45 had a constellation of persistent signs of unexplained etiology before 3 years of age that could be consistent with possible MD and were considered cases.
 
In utero ARV exposure
Gestational age at birth was determined by the last menstrual period, ultrasound, or pediatric assessment at birth; 40 weeks was used for children with an unknown gestational age at birth (26.3%). The start of the gestational period was determined by subtracting the infant's gestational age at birth from the infant's date of birth. In utero exposure was categorized as exposure to any NRTI, to specific NRTI, and to ZDV/3TC at any time in gestation and by trimester of first exposure.
 
Statistical analysis
Exact logistic regression (LogXact, version 7.0) [16] was used to examine the independent association of (i) overall in utero NRTI exposure (i.e., at any time during gestation), and (ii) trimester of first in utero NRTI exposure (the independent variables) and MD (the dependent variable). We assessed potential confounding by sex, race/ethnicity, year of birth, premature birth, neonatal ARV prophylaxis, peak maternal HIV RNA copy number during the third trimester or at delivery, and in utero psychoactive drug exposure (alcohol, tobacco, cocaine, and other drugs including heroin). Of the potential confounders considered, only year of birth affected the estimated odds ratios (OR) of in utero NRTI exposure and possible MD; sex was significantly associated with possible MD but not with in utero NRTI exposure. Therefore, the final models included indicators for in utero NRTI exposure and year of birth categorized in 1-year intervals.
 
Three of 45 identified possible cases had incomplete in utero ARV exposure information and were excluded from statistical analyses. Twenty two of the remaining 42 cases met the EPF case definition only because of one or more abnormal psychometric test results indicating major retardation of cognitive development (Fig. 1). The reviewing clinicians were less confident classifying these cases as having potential mitochondrial etiology without additional manifestations of MD given the high background rate of cognitive delay in high-risk, low socio-economic populations [17-20]. The primary analysis included 20 cases (Table 1) who had persistent signs consistent with possible MD other than, or in addition to, major retardation of cognitive development, and 1017 noncases not identified in the initial screening and with complete in utero ARV information (Fig. 1). Secondary analyses including all 42 cases and all 1079 noncases with complete in utero ARV information also were performed.
 
 
 
 
  icon paper stack View Older Articles   Back to Top   www.natap.org