In Reply: Insufficient Evidence to Diagnose In Utero Infection of HIV

To the Editors,

We thank Lou and Zhou for their interest in our recent case report and for the opportunity to clarify certain aspects of it. We do not agree with Lou and Zhou’s interpretation and believe that indeed this child was unequivocally infected in utero. Addressing the five points raised by Lou and Zhou in order:

1. It is not the case that in utero HIV is characterized by persistent HIV positivity from birth. The reason for this is that antiretroviral therapy (ART) taken by the mother in pregnancy that crosses the placenta, and prophylactic ART given from birth to children born to HIV-infected mothers, reduce plasma viremia, sometimes to beneath the level of detection, in in utero infected infants. Furthermore, in the era of effective combination ART (cART), it is quite possible for an HIV-infected infant with low-level infection to have numerous negative HIV tests, as illustrated in the well-characterized case of the Mississippi baby (1). A significant proportion of HIV-infected infants who achieve early and sustained virologic suppression after being treated with cART can have undetectable HIV DNA, RNA and HIV specific antibodies (2-5) in peripheral blood, therefore by standard clinical testing will appear HIV-uninfected.

2. We agree that maternal cells do cross into fetal circulation. However, to achieve the levels of 41 and 49 HIV DNA copies per million peripheral blood mononuclear cells (PBMC) that were observed in this child on days 1 and 10 of life, respectively, would require larger transfusions of maternal blood into the baby than arise in nature. A recent study, using DNA methylation to detect maternal contamination of fetal blood, showed that the levels of contamination were too low to be detected in 80-90% of cases, and, where detected, amounted to between 10-5 to 10-4 of cells (6). Assuming a maternal HIV DNA level of 1000 copies per million PBMC, 1 million PBMC per mL of maternal blood and an infant total blood volume of 240mL with 5 million PBMC per mL, it would require 20% of total infant blood volume to be maternal, compared to the level of <0.001% that is typically observed in reality (6). Consistent with this calculation is that we have used the GeneXpert total nucleic HIV PCR to test over 3000 infants in the first 48 hours of life who were born to mothers at high risk of in utero HIV transmission i.e. seroconversion during pregnancy or high plasma viremia, more than half of whom were born vaginally. In 97.6% of cases no HIV was detectable, and in only 0.1% of cases was HIV detected without HIV infection being subsequently confirmed by further tests.

3. We agree that breast-feeding can cause mother-to-child transmission of HIV, but not at 19 hours of life. The time between HIV infection and detectable HIV RNA and DNA in peripheral blood is 7-21 days (7). Therefore, the positive tests from day 1 of life could not have been due to HIV infection via breast milk.

4. We agree that phylogenetic clustering of mother and child Gag sequences shows mother-to-child transmission (MTCT) has occurred. When the sample is from the first day of life of the child, as described above, this confirms that in utero MTCT has occurred.

5. We agree that in utero HIV transmission is the least common route of HIV infection and is rare (7). Indeed, in our study in the province of KwaZulu-Natal, South Africa, the in utero MTCT rate is <1%. However, in KwaZulu-Natal maternal HIV seroprevalence is 40%, and the infant we describe was one of the 446 in utero HIV-infected infants born in this province in 2016 (8).

We therefore believe we can unequivocally conclude that the infant described was indeed infected with HIV in utero. The confusion with regard to the multiple negative tests in the first months of life highlight the new diagnostic uncertainty brought about by ART used for prevention and early treatment of HIV in infants.