This evaluation of perinatal HIV transmission in a malarious area of western Kenya demonstrated that approximately 20% of infants born to HIV-infected mothers acquired HIV by 4 months of age, similar to rates reported in other sub-Saharan African settings (30
). Consistent with other studies, we found that maternal viral load in peripheral blood at the time of delivery and having an episiotomy or perineal tear were risk factors for perinatal MTCT (33,34
). Contrary to our expectations, we observed that women with placental malaria had lower rates of perinatal MTCT than women without malaria. However, women with high-density malaria had significantly higher rates of perinatal MTCT than parasitemic women with low-density malaria. In additional models, only low placental parasite density (1–9,999 parasites/μL of blood) was associated with protection from perinatal MTCT; high-density placental infection was either a risk factor for perinatal MTCT (in the model evaluating only women with malaria) or it was a neutral factor (in the model evaluating all women). Maternal viral load was slightly higher in women with high-density malaria, and, among transmitting women, mean viral load was eightfold higher if the woman had concurrent malaria, consistent with the suggestion that high-density malaria may be an important stimulus of viral replication (10
Our study had some important limitations. First, only healthy women were screened for this study; no women with AIDS or any known underlying chronic illness were enrolled. Although this eliminated potential conditions that could have complicated the analysis (e.g., a higher likelihood of additional concurrent infections), it restricted our study population to the “healthiest” women with HIV and likely resulted in an underestimate of the overall rate of perinatal MTCT in our study area. Some infants categorized as having acquired HIV through perinatal transmission may have actually acquired it through early breastfeeding transmission. However, including these infants would likely bias the results toward underestimating the magnitude of the observed risks. Our measurement of malaria was limited to microscopy examination of placental smears and could have been inexact; those with no evidence of malaria may have had very low-density infection and may have been misclassified as having no malaria. Such misclassification would be expected to bias our findings toward the null hypothesis. Because placental malaria can cause an inflammatory response in the placenta, the use of leukocyte count to calculate parasite density may have resulted in an underestimation. However, our cutoff of 10,000 parasites/μL for high-density placental parasitemia is essentially a relative measure based on the upper quintile of densities; therefore, we do not think that our estimation technique would have introduced any bias. Finally, in a study such as ours, loss to follow-up always has the potential to introduce bias. Approximately one third of infants enrolled in our study were lost to follow-up. However, as noted, these mothers and infants were generally similar to the study population included in our analysis, and we were unable to detect biases that would have affected our analysis.
Our observation of an association between low-density placental malaria and reduced perinatal MTCT has several possible explanations. First, over 16% of women reported self-treatment (typically for fever) with chloroquine during pregnancy. High-grade chloroquine resistance in this area is widespread, and its use is unlikely to clear placental infections but may reduce parasite densities. Chloroquine is known to have anti-HIV properties and to reduce HIV-1 replication and viral loads in adults (35
) and as such could potentially reduce the risk for perinatal MTCT. Although we identified no differences in the proportion of women with or without malaria who used chloroquine, and no association was found between chloroquine usage and MTCT, women identified as having malaria may have used chloroquine more frequently and at higher doses (information that was not collected) than women without malaria. A second possibility is that some of the women classified as having placentas negative for malaria parasites may actually have had low-level chronic malaria associated with inflammation and increased HIV replication. In contrast, women classified as having low-density malaria may have had recent malaria with minimal inflammatory response and thus no increase in HIV replication. Unfortunately, blood films are not capable of detecting inflammatory responses.
A more likely explanation is that a balance exists in the uterine-placental-fetal environment among malaria-induced antigen stimulation, HIV viral replication, maternal host immune response to both malaria and HIV, and the likelihood of MTCT. Although high viral load has been shown to increase the risk for MTCT, no more than half of exposed infants, even at high maternal viral load, become infected with HIV-1. These data suggest that other systemic or placental factors must be important in preventing HIV-1 transmission. Recent studies suggest that selected cytokines and hormones potentially affect HIV-1 transplacental transmission and that both innate and acquired protection play a role in MTCT (37
). First, malaria is known to induce disequilibrium in Th1 and Th2 cells, favoring the Th1 pathway (38
). T-helper responses are known to control HIV replication; hence, inducing Th1 response in the placental compartment could lead to reduced HIV-1 replication (41
). Indeed, a moderate increase was found in the Th1 cytokine interferon-γ response in the intervillous blood mononuclear cell responses of HIV-positive mothers with malaria as compared to HIV-positive mothers without malaria (43
). Second, leukemia inhibitory factor induces a potent inhibition of HIV replication, and this cytokine is upregulated in placentas of women who do not transmit HIV (44
). Malarial antigens may induce production of leukemia inhibitory factor that results in reduced rates of perinatal MTCT. Third, malarial antigen may result in altered chemokine production, which in turn can block chemokine receptors necessary for cellular HIV entry (45
). Recent immunologic studies conducted in a group of women who participated in this cohort showed that macrophage inflammatory protein (MIP)-1β, a chemokine known to block the entry of HIV-1, was significantly elevated in the intervillous blood plasma of women with placental malaria irrespective of their HIV status. In women with concurrent HIV and placental malaria, the intervillous blood plasma levels of MIP-1β were significantly elevated compared to HIV-negative women with malaria and HIV-positive women without malaria (46
). Thus, a complex balance of antigen stimulus and immune response may occur in the placenta, as demonstrated in recent studies focusing on the placental intervillous blood responses (38
). In this scenario, active malaria might lead to one of the local immune responses described. This response may control the density of parasitemia and may also provide immune control that limits perinatal MTCT. For most women with malaria (approximately 80% in our study), this resultant balance leads to “protection” with controlled (low-density) malaria and reduced perinatal MTCT. For a minority of women with malaria (approximately 20% in our study), the balance is tilted to inadequate control of malaria (i.e., higher density of malaria parasites with concomitant increased antigen stimulation of viral replication and higher viral load) and higher rates of perinatal MTCT. This finding is consistent with our observation that women with high-density placental malaria do have higher rates of perinatal MTCT than women with low-density placental malaria, but that they require much higher viral loads to achieve this transmission. Our findings suggest complex relationships between malaria, viral load, and perinatal MTCT. These findings will need to be confirmed and expanded upon in further research in different settings.
The finding of reduced perinatal MTCT in the presence of low-density, but not high-density, placental malaria does not suggest altering existing recommendations for the use of intermittent preventive antimalarial treatment in pregnant women in malarious areas of Africa (47
). For most dually (malaria and HIV) infected women, the overall outcome would have modest benefit in reducing perinatal MTCT, but for the minority with insufficiently controlled infections, perinatal MTCT would be increased. At this point, we cannot differentiate in advance which women will end up in which group. The important benefit of antimalarial treatment may be to reduce the likelihood of women having high-density placental malaria and in also helping reduce the other known adverse effects of malaria during pregnancy, including anemia, low birth weight, and prematurity (1
As noted earlier, this study was carried out in Kenya in a changing environment of perinatal HIV and malaria prevention and has been transformed into a program delivering short-course antenatal antiretroviral therapy to HIV-infected women and a system to support providing intermittent preventive antimalarial treatment during pregnancy according to the newly adopted national policy (48
). As more countries adopt policies of antiretroviral and antimalarial interventions in pregnancy, the interrelationship and ultimate benefit of these two interventions need to continue to be evaluated.