Head circumferences of children born to HIV-infected and HIV-uninfected mothers in Zimbabwe during the preantiretroviral therapy era

Objectives: To describe the head growth of children according to maternal and child HIV infection status. Design: Longitudinal analysis of head circumference data from 13 647 children followed from birth in the ZVITAMBO trial, undertaken in Harare, Zimbabwe, between 1997 and 2001, prior to availability of antiretroviral therapy (ART) or cotrimoxazole prophylaxis. Methods: Head circumference was measured at birth, then at regular intervals through 24 months of age. Mean head circumference-for-age Z-scores (HCZ) and prevalence of microcephaly (HCZ < −2) were compared between HIV-unexposed children, HIV-exposed uninfected (HEU) children and children infected with HIV in utero (IU), intrapartum (IP) and postnatally (PN). Results: Children infected with HIV in utero had head growth restriction at birth. Head circumference Z-scores remained low throughout follow-up in IP children, whereas they progressively declined in IU children. During the second year of life, HCZ in the PN group declined, reaching a similar mean as IP-infected children by 21 months of age. Microcephaly was more common among IU and IP children than HIV-uninfected children through 24 months. HEU children had significantly lower head circumferences than HIV-unexposed children through 12 months. Conclusion: HIV-infected children had lower head circumferences and more microcephaly than HIV-uninfected children. Timing of HIV acquisition; influenced HCZ, with those infected before birth having particularly poor head growth. HEU children had poorer head growth until 12 months of age. Correlations between head growth and neurodevelopment in the context of maternal/infant HIV infection, and further studies from the current ART era, will help determine the predictive value of routine head circumference measurement.


Introduction
Children infected with HIV before or around the time of birth have faster disease progression than those infected through breastfeeding [1]. Despite avoiding HIV infection, HIV-exposed uninfected (HEU) children have higher mortality, morbidity and growth failure than HIVunexposed children [1][2][3]. HIV-infected children have poorer developmental outcomes than HIV-uninfected children, particularly when antiretroviral therapy (ART) is not initiated early [4].
In infancy, head growth is correlated with brain size [5] and is easily measured. In high HIV prevalence populations, microcephaly may predict poor neurodevelopmental outcomes [6] and is a sign of HIV encephalopathy [7]. Here, we report head growth among children followed from birth in Zimbabwe.

Methods
Between 1997 and 2000, 14 110 mother-infant pairs were recruited to a randomized trial of maternal/neonatal vitamin A supplementation in Zimbabwe, as previously described [8,9]. Exclusion criteria included multiple pregnancy, birthweight less than 1500 g, maternal plans to leave Harare and life-threatening conditions in the mother/child. ARTand cotrimoxazole were not available in public-sector health care at the time.

Ascertainment of maternal HIV status
All mothers were tested for HIVat baseline using enzymelinked immunosorbent assay (ELISA) and western blotting [8]. Mothers with negative HIV tests were retested at subsequent visits. Mothers with positive HIV tests were retested at the next visit to confirm status.

Ascertainment of infant HIV status
Cell pellets and plasma from HIV-exposed children were prepared from whole blood and stored at À70 8C. The last available sample was tested for HIV by ELISA on plasma if aged at least 18 months (GeneScreen; Sanofi Diagnostics Pasteur, Johannesburg, South Africa) or DNA PCR on cell pellets if aged less than 18 months (Roche Amplicor version 1.5; Roche Diagnostic Systems, Alameda, California, USA). If the last sample was positive, earlier samples were tested to determine timing of infection [8].
HIV infection/exposure status Children were classified into five categories, as previously described [1,3]: (1) HIV-unexposed: mother HIV-negative throughout study period. (2) HEU: mother HIV-positive at baseline; last available child sample HIV-negative. Children of mothers who seroconverted during follow-up were excluded. A time-varying analysis was used to categorize children, with those acquiring HIV intrapartum/postnatally categorized as HEU until their final HIV-negative test. To ensure that the HEU group did not become contaminated with IP/PN infants who died or became lost to follow-up before a positive HIV test could be obtained, infants were censored at their last negative HIV test.

Anthropometric data
Head circumference was measured at each study visit using methods described by Gibson [10]. Measurements were converted into head circumference-for-age Z-scores (HCZ) using WHO international standards [11].

Statistical analyses
Patterns of head circumference from birth to 24 months were described by plotting mean HCZ against age. Unpaired t tests were used to compare mean HCZ between groups. Odds ratios for microcephaly (HCZ < À2 SD) were calculated using HIV-unexposed infants as the reference category. Analyses were undertaken using Prism, version 6.0 (GraphPad, La Jolla, California, USA).  (12) 13.5 (14) 13.8 (9) 13.2 (7) 17.1 (7) 15. data). Numbers of children per time point in each group are shown in Table 1.

Ethical approvals
Mean head circumference-for-age Z-scores At birth, mean HCZ was below 0 in IU and HEU children, but not significantly different from 0 in HIVunexposed children (Table 1). HCZ in HIV-uninfected (HIV-unexposed and HEU) and PN children increased to 6 months of age, followed by a decline to 24 months, a trend not seen in IP and IU children (Fig. 1). As the increase in PN HCZ between 3 and 6 months may have been influenced by the addition of PN infants first testing positive at 6 months, we excluded these infants in a sensitivity analysis; a similar trajectory was seen (data not shown). IP and IU children had significantly lower HCZ than HIV-uninfected children throughout follow-up. PN-infected children had poorer head growth than HIVuninfected children except at 6 months. HEU children had statistically lower HCZ than HIV-unexposed children in the first year of life.
Among HIV-infected children, IU had the lowest HCZ, which progressively declined throughout follow-up, followed by IP children. During the second year of life, HCZ among PN children declined to reach a similar mean as IP-infected children (Fig. 1). As the PN group may have inadvertently included IP-infected infants testing PCR-negative at 6 weeks of age, two sensitivity analyses were undertaken. Infants first testing HIVpositive at 3 months of age were either grouped with IP infants or excluded from the analysis entirely, with no significant effect on results (data not shown).

Microcephaly
Microcephaly was more common among IU and IP children than HIV-uninfected children through 24 months. Overall, 11.1% of perinatally infected children had microcephaly at birth, and incidence remained high among IP and IU children throughout follow-up. HEU children had more microcephaly than HIV-unexposed children until 12 months. PN children had a higher incidence of microcephaly than HIV-uninfected children through most of the follow-up period; this reached statistical significance at 9 months of age (Table 1).

Discussion
The current study has three key findings. First, head growth was poor among HIV-infected Zimbabwean children between birth and 2 years of age; second, head circumference in these children was associated with timing of HIV acquisition; and third, head growth in HEU children was significantly poorer than in HIVunexposed children during the first year of life.
Overall, 11.1% of HIV-infected infants, compared with 5.4% of HIV-unexposed infants, had microcephaly at birth, highlighting the burden of HIV-associated fetal head growth restriction before ART availability. IU and IP children had lower mean HCZ and more microcephaly than HIV-uninfected children throughout 2 years of follow-up. Compared with HIV-uninfected children, PN children had significantly more microcephaly only at 9 months; the nonsignificant differences at other time points may result from the small number of PN children included in the analysis. Poor head growth in HIVinfected children has previously been reported from Rwanda [12] and the USA [13], and developmental delay among HIV-infected children is well recognized [4]. However, the association between head growth and neurodevelopment is less certain; although there are associations between postnatal head growth and developmental delay in preterm infants [14], there are limited data in the context of HIV [6]. An ART-era Zimbabwean study found no differences in head circumference between HIV-infected and HIV-uninfected infants at birth, but did show an association between microcephaly at birth and neurodevelopmental impairment [6]. In the USA, HIV-infected children had lower developmental scores and head circumferences than HIV-unexposed children at 24 months [15], although associations between development and head circumference were unclear. The impact of microcephaly in the context of HIV in sub-Saharan Africa warrants further investigation. In lowincome settings, in which developmental examinations are difficult to access, head circumference could be a helpful screening tool to highlight children who may benefit from formal neurodevelopmental assessment and intervention. Among HIV-infected children, head growth was associated with timing of HIV acquisition. IU children had particularly poor head growth, with mean HCZ scores that declined throughout the study period. Overall, those infected before birth had lower HCZ than children infected intrapartum or postnatally. This is consistent with previous findings from this cohort [1,3] and other studies [16] showing that IU children have particularly rapid disease progression. However, in contrast to previous findings, toward the end of the study period, IP outcomes were closer to PN than IU outcomes, highlighting the critical influence of intrauterine HIV transmission on head growth. Timing of HIV infection has previously been associated with neurodevelopmental outcomes [17], but children in the trial did not undergo developmental assessments.
Head growth in HEU children was significantly poorer than in HIV-unexposed children through the first year of life; it is plausible that we were underpowered to show significant differences in the second year of life. Similar findings were reported from Colombia, where head circumferences of 1-month-old HEU infants were significantly lower than HIV-unexposed infants, despite being in the normal range [18]. This finding may be specific to HEU infants in low-income and middleincome countries; in the USA, head circumference differences between HEU and HIV-unexposed infants at a mean age of 10 months did not reach statistical significance (HEU mean HCZ 0.18 vs. HIV-unexposed mean HCZ 0.23, P ¼ 0.75), although the study may have been underpowered (n ¼ 164) [19]. Studies comparing neurodevelopmental outcomes in HEU children have reported heterogeneous results [4]. After adjusting for socioeconomic status, HEU and HIV-unexposed children from Thailand had similar neuroanatomical measurements at 10 years of age [20]. However, HEU Thai children aged 2-12 years had lower IQs than HIV-unexposed controls, even after adjusting for confounders [21].
An important limitation of this study was that very low birthweight infants (<1500 g) were excluded from the original trial, meaning that the most severely growthrestricted infants could not be included in this analysis. Neurodevelopmental assessments were not carried out, meaning that the functional significance of restricted head growth remains uncertain. However, this study has several strengths. First, the HIV status of mothers and children was well characterized throughout follow-up, allowing a detailed insight into the impact of infection timing on head growth. Second, this is the largest HEU infant cohort to date, with a demographically similar HIVunexposed population. Whether differences in head growth are seen in the current era of preventing motherto-child transmission and early infant ART is uncertain, but clearly warrants investigation. Although it is plausible that maternal virological suppression and immune reconstitution on ART may reduce the effects of HIV infection on head growth, as shown in one South African study [22], ART may itself adversely affect head growth; one US study showed lower 12-month HCZ in children born to mothers using tenofovir-based compared with non-tenofovir-based regimens [23].
In summary, head circumference in early life was associated with HIV infection/exposure status. The reasons for poor head growth may relate to increased incidence of common infections [3], which may disturb the growth hormone-insulin-like growth factor-1 axis [24]; congenital coinfections, including cytomegalovirus [25]; or HIV itself. Since the first 1000 days (i.e. the period from conception through 2 years of age) is a developmentally sensitive period during which brain growth, neural pathways and synaptic pruning occur, the relationship between head growth and neurodevelopment requires careful evaluation. Future studies should seek to correlate head growth, neurodevelopmental status and neurobiological measures in HIV-infected and HEU infants in developing countries to determine the predictive value of microcephaly for long-term outcomes in these groups of vulnerable children.