Prevention is better than cure.....

In sub-Saharan Africa, most HIV infections are heterosexually transmitted, and women are more likely to become infected than men [1]. Most HIV infections in children are acquired through mother-to-child transmission (MTCT) [2, 3]. Rates of perinatal transmission range from 13% to 40% among women not treated for their HIV infection [4]. Early child mortality among HIV-positive and HIV-negative children born to HIV-infected mothers have been reported to be 4 and 1.3–2 times higher, respectively, than that among children born to HIV-negative women [5] (A.M.v.E., personal observation).

There is geographic overlap between HIV and malaria in sub-Saharan Africa. In areas where malaria is endemic, pregnant women have a higher risk of malaria than nonpregnant women, and this risk is further increased among HIV-infected women [6]. Although these malaria infections are usually asymptomatic, parasites sequester in the placenta, and placental malaria is associated with low birth weight (LBW), intrauterine growth retardation, prematurity, and maternal and infant anemia [7–9].

Malaria is a major contributor to infant mortality in sub-Saharan Africa, directly through infant illness and indirectly through placental malaria and its adverse effects. The effect of the combination of maternal HIV infection and malaria during pregnancy on infant survival is not clear [6]. An early study in Malawi suggested that placental malaria was a risk factor for postneonatal infant mortality (PNIM) in children born to HIV-infected women [10]. Malaria, like other coexisting infections, can cause temporary increases in HIV load [11]. It was hypothesized that PNIM could have been increased either because placental malaria increased MTCT of HIV-1 or because placental malaria enhanced the progression of HIV disease in infected infants directly or indirectly (if placental malaria would be a marker for another disease leading to infant death, such as infant malaria). The HIV vertical transmission study in Kisumu was designed to examine these questions. We previously reported that placental malaria was not associated with increased MTCT of HIV but, by contrast, was found to be protective [12]. In the present article, we study the effect of placental malaria, infant malaria, and anemia on PNIM among the infants.

SUBJECTS AND METHODS

Study site and screening procedures. The study was conducted at the Nyanza Provincial General Hospital in Kisumu, western Kenya; enrollment lasted from June 1996 to August 2000, and infant follow-up ended in August 2001. Screening procedures have been described elsewhere [12]. Briefly, healthy pregnant women with an uncomplicated singleton pregnancy of 32 weeks' gestation were invited to participate. After informed consent was obtained and HIV counseling performed, a questionnaire on medical and obstetric history was completed, and blood was obtained for HIV testing. All screened women received routine antenatal care [13] and were encouraged to deliver at the hospital; 50% did, which reflects the low rate of deliveries at health care facilities in this area (37.4%) [14]. For women who gave birth in the hospital, placental smears and blood for maternal hemoglobin and viral load determinations were collected. Within 24 h of birth, infants were weighed, and their gestational age was assessed [15].

Enrollment in the follow-up study. All HIV-seropositive mothers and their normally delivered, live-born infants were eligible for enrollment in the follow-up study. One month postpartum, maternal blood was collected for the determination of CD4+ cell counts. Infants were scheduled to be seen every 4 weeks until the age of 1 year or death. During routine visits, we obtained information on the health of the infant during the preceding month. Capillary blood was collected by finger stick for a blood smear, hemoglobin determination, and polymerase chain reaction (PCR) testing for HIV. CD4+ cell counts were not determined in infants. Infants with malaria parasitemia were treated, regardless of clinical symptoms, with an appropriate antimalarial (sulfadoxine-pyrimethamine [SP] or amodiaquine). Clinical staff treated infants with anemia or complaints in accordance with their findings [9]. If a routine visit was missed, we visited the participant's home to determine whether the caretaker remained interested in participating and to assess the vital status of the infant. For infant deaths, we obtained additional information using verbal autopsy.

Laboratory procedures. Details on standard procedures for blood smears, hemoglobin determination, and maternal HIV testing have been reported elsewhere [12]. Maternal CD4+ cell counts were assessed using commercial, dual-label monoclonal antibodies (Becton-Dickinson Immunocytometry) and standard fluorescence-activated cell sorting analysis after whole-blood lysis [16]. The maternal HIV-1 load was determined in the delivery sample using the Roche Amplicor HIV-1 monitor test (version 1.0; Roche Diagnostics), which has a quantification limit of 400 viral copies/mL. HIV testing of the infants was done by PCR of proviral DNA extracted from peripheral blood mononuclear cells [17].

Definitions.

PNIM was a death occurring between 29 and 364 days of delivery. HIV-positive infants were infants of HIV-seropositive mothers who had 2 consecutive positive PCR tests with the first positive PCR test at age 4 months; HIV-negative infants were infants of HIV-seropositive mothers who had 2 consecutive negative PCR tests and for whom the PCR test at the last visit was negative. Infants for whom we had insufficient PCR data to determine their status were excluded. To obtain a homogenous group of HIV-infected infants, we excluded infants who had their first positive PCR test after the age of 4 months; the different timing of HIV transmission may have resulted in a different immunological status and susceptibility to infectious diseases.

Malaria in infants was defined as the presence of asexual-stage parasites in thick smears, independent of the presence or absence of clinical signs and symptoms and independent of species. Moderate to severe anemia among infants was defined as a hemoglobin level <8 p =" .2),">

Ethical review.

The study protocol was approved by the review boards of the Kenya Medical Research Institute, Nairobi; the Centers for Disease Control and Prevention, Atlanta, GA; and the Academic Medical Center of the University of Amsterdam, The Netherlands. During the study years, programs for the prevention of MTCT had not yet been introduced in the hospital; this started late 2000, after enrollment of infants had finished [12].

RESULTS :
Description of the study population. Infants from 835 HIV-seropositive women were enrolled. Of these, 265 infant/mother pairs were excluded: 165 infants never made a follow-up visit, 77 infants visited once only, 6 infants had an indeterminate HIV status despite 2 visits, and 17 infants seroconverted after the age of 4 months. The 265 excluded infants had a significantly lower mean birth weight (table 1). Although 27 of the excluded infants were known to have died, none of the 17 infants who seroconverted after the age of 4 months died. The prevalence of placental malaria among excluded infants who died or survived was 29.6% and 23.5%, respectively (P = .5).

Table 1. Characteristics of included and excluded HIV-seropositive mothers and their infants for the analysis of factors associated with postneonatal infant mortality, Kisumu, Kenya, 1996–2001.

The majority of the mothers of included infants had a maternal CD4+ cell count 500 cells/L (table 1). Of the 570 included infants, 67.7% completed the study. Primiparae and mothers with premature or HIV-negative infants were less likely to complete the study (P = .04, P = .01, and P = .04, respectively); women who did not complete the study had a higher median CD4+ cell count and a lower median log10 viral load (P = .05 and P = .01, respectively).

Placental malaria was detected in 141 women (24.7%). All infections were with Plasmodium falciparum, and there were 3 mixed infections with P. malariae. A blood smear and hemoglobin data were not available for 5 and 85 visits, respectively, of the 5083 visits that were made by the infants. Thirteen percent of blood smears were positive; 7 mixed infections with P. malariae and P. falciparum were detected, and all other infections were with P. falciparum. Moderate to severe anemia was diagnosed at 5.2% of the routine visits.

PNIM. Of the 570 infants, 75 (13.2%) died during the postneonatal period. One-third of the deaths occurred at the age of 4–5 months (table 2). As expected, PNIM was significantly more common among HIV-positive infants than among HIV-negative infants (figure 1 and table 2). Factors associated with PNIM in univariate and multivariate Cox regression models are shown in table 3. The adjusted HR (AHR) for placental malaria overall was 0.77 (95% confidence interval [CI], 0.43–1.39). However, effect modifications were observed between placental malaria and infant HIV status (P = .06 for the interaction term) and between infant anemia and infant HIV status (P = .03 for the interaction term).

Table 2. Postneonatal infant deaths among infants of HIV-seropositive women by infant HIV-status, Kisumu, Kenya, 1996–2001.
Figure 1. Kaplan-Meier survival curve of postneonatal infant mortality by infant HIV status and placental malaria, Kisumu, Kenya, 1996–2001. Log rank tests: A vs. B, P = .7; A vs. C, P = .2; A vs. D, P < .001; B vs. C, P = .4; B vs. D, P < .001; C vs. D, P = .07. Neg, negative; pos, positive. Table 3. Risk factors for postneonatal infant mortality by infant HIV-status among infants of HIV-seropositive women, Kisumu, Kenya, 1996–2001. Placental malaria. HIV-positive infants born to mothers with placental malaria tended to be more likely to survive than those born to mothers without placental malaria (AHR, 0.34 [95% CI, 0.10–1.10]; P = .07, log-rank test) (table 3 and figure 1). Placental malaria was not associated with PNIM in HIV-negative infants (AHR, 1.29 [95% CI, 0.62–2.66]) (table 3). Infant anemia and infant parasitemia. Moderate to severe infant anemia was a risk factor for PNIM among HIV-negative infants but not among HIV-positive infants (table 3). There was no interaction between the infant's HIV status and the effect of infant malaria: malaria parasitemia at a routine visit was a protective factor against PNIM among both HIV-infected and HIV-uninfected infants (table 3). Among HIV-positive infants who died, median survival among 10 infants with at least 1 episode of malaria was 274 days, compared with 134 days among 29 infants without any episodes of malaria (P < .01, Mann-Whitney U test). Among HIV-negative infants, median survival was 263 days (10 infants) and 160 days (26 infants), respectively (P = .08, Mann-Whitney U test). Parasite prevalence among infants generally increases during the first year of life, and it increased in this study population, from 7% at 4 weeks to 22.0% at 44 weeks of age. To assess whether the association between infant malaria parasitemia and PNIM was not merely a proxy for age, we restricted the analysis to infants who completed the first 6 visits; the overall AHR with respect to malaria parasitemia and PNIM remained indicative of protection (0.53 [95% CI, 0.18–1.56]). Infants born to mothers with placental malaria have been reported to have a higher risk of infant malaria [18] and anemia [9]. We therefore determined whether the observed association between infant malaria and PNIM could partially be explained by the effect of placental malaria, which, as shown above, was also associated with PNIM, although the direction of the association differed between HIV-negative and HIV-infected infants. Removal of placental malaria from the models did not substantially change the HR between PNIM and infant parasitemia or anemia (AHR, 0.36 [95% CI, 0.10–1.23] and AHR, 5.06 [95% CI, 1.99–12.90] for parasitemia and anemia, respectively, in HIV-negative infants; AHR, 0.34 [95% CI, 0.08–1.44] and AHR, 1.02 [95% CI, 0.30–3.42] for parasitemia and anemia, respectively, in HIV-positive infants). We examined whether the association between malaria parasitemia was affected by treatment with SP, the antimalarial that was predominantly used in the study (75%) to treat parasitemia. The inclusion of a variable indicating SP use during the month preceding the visit did not change the association between infant malaria and PNIM in any of the models (AHR, 0.72 [95% CI, 0.31–1.67] and AHR, 0.43 [95% CI, 0.13–1.41], for HIV-positive and HIV-negative infants, respectively, adjusted for the factors in the adjusted models in table 3). Because infant parasitemia is associated with anemia, we further examined the effect of these variables in the model among HIV-negative infants. We created a variable indicating whether both, either, or neither of these factors were present at the visit. Compared with the absence of malaria parasitemia or of moderate to severe anemia, the combination of moderate to severe anemia and parasitemia was not significantly associated with PNIM (AHR, 1.13 [95% CI, 0.15–8.53]; P = .9), whereas anemia alone (i.e., without parasitemia at the visit) remained a significant factor (AHR, 5.80 [95% CI, 2.19–15.38]; P < .01), and malaria parasitemia alone continued to be associated with a protective effect (AHR, 0.49 [95% CI, 0.12–2.08]; P = .3). Maternal geometric mean viral load and PNIM. Infant HIV infection and death were associated with a significantly higher maternal viral load (table 4). Placental malaria was not associated with a significantly higher viral load. However, among women who transmitted HIV, maternal viral load was significantly higher when placental malaria was present. Among HIV-positive infants, the geometric mean maternal viral load among the 14 survivors of mothers with placental malaria (16,846 copies/mL) was not significantly different from the viral load among the 31 infants who died and whose mothers did not have placental malaria (8561 copies/mL; P = .4), whereas it was significantly higher than the maternal viral load of the 54 survivors whose mothers did not have placental malaria (3849 copies/mL; P = .02), which indicates that placental malaria may change the association between a high maternal viral load and HIV progression in the infected infant. The mothers with placental malaria of the 2 HIV-infected infants who died had very high maternal viral loads, and this was significant, compared with the HIV-infected survivors of mothers without placental malaria. Table 4. Associations between the geometric mean maternal viral load at the time of delivery, placental malaria, and postneonatal infant mortality, Kisumu, Kenya, 1996–2001. Probable causes of death. For 63 deaths, a probable cause of death was assigned using verbal autopsy. There were no significant differences by infant HIV status, but numbers were small (figure 2). Figure 2. Probable causes of postneonatal infant deaths by infant HIV status among infants of HIV-seropositive women by use of verbal autopsy, Kisumu, Kenya, 1996–2001. Verbal autopsy information was reviewed independently by 3 medical workers (clinical officers or doctors), who each assigned a diagnosis. If 1 diagnosis was given by 2 reviewers, it was assigned as the cause of death. If all 3 causes were different, a fourth reviewer was used to adjudicate. If no verbal autopsy information was available, the cause of death was labeled as undetermined. Malaria and anemia were combined, and HIV-positive infants were compared with HIV-negative infants (P = .06, 2-tailed Fisher's exact test). DISCUSSION Our results underscore that the complex interaction between HIV and malaria is challenging to study. By contrast to the earlier study by Bloland et al. [10], we did not find that placental malaria was a risk factor for PNIM among infants of HIV-seropositive women. Overall, the risk on PNIM was actually lower, although not significantly so, in infants born to HIV-infected mothers with placental malaria (AHR, 0.77 [95% CI, 0.43–1.39]), and this remained so after adjustment for potential confounders, including maternal CD4+ cell counts and LBW. These results are consistent with those of another study [19] that reported a nonsignificant protective effect of placental malaria against PNIM among infants born to HIV-seropositive women (HR, 0.39 [95% CI, 0.11–1.38]). We previously reported that MTCT was significantly lower in infants born to mothers with placental malaria than in those without placental malaria [12]. Several explanations can be considered for the association between placental malaria in HIV-positive women and increased survival time of HIV-infected infants. First, we speculate that prenatal exposure to malaria may modulate the immune system to generate protective immune pathways (involving innate or classical T cell–mediated pathways), which may slow the progression of HIV-1 and contribute to prolonged survival. There is considerable evidence that maternal exposure to malaria modulates immune responses to malaria in utero [20–23]. Second, it has recently become known that HIV-mediated disease progression is correlated with the down-regulation of a regulatory T cell (Treg) subset (Foxp3+CD4+CD25hi T cells), which may be critical for maintaining CD4+ cell counts [24]. It has been shown recently that placental malaria increases the number of Treg cells in cord blood [25], and this may help protect against any rapid decrease in CD4+ cell counts in infants born to HIV-positive mothers with placental malaria. Third, HIV/malarial coinfection can modulate the cytokine environment in the placenta and/or fetus, which may reduce the initial HIV load or slow HIV replication in the fetus. We have shown previously that HIV/malarial coinfection up-regulates (compared with HIV-positive malaria-negative women) some chemokines, including macrophage inflammatory protein (MIP)–1, that can compete with the CCR5 receptor for HIV entry [25, 26]. We were also interested in establishing whether malaria in the HIV-positive infant was associated with enhanced HIV disease progression. Infant malaria parasitemia was not a risk factor for PNIM—indeed, median survival was higher among HIV-infected children with at least 1 documented malaria episode than in children with no malaria, consistent with the results of a previous study in Uganda [27]. Because all infections were treated regardless of the presence of symptoms, frequent treatment with SP may have provided time windows of prophylaxis and reduced subsequent malaria and morbidity in these infants; SP can both treat and protect against malaria because of the long half-life of its components [28]. SP may have an effect on other infections as well (Streptococcus pneumonia, Toxoplasma gondii, and Pneumocystis jiroveci). However, when we included the use of SP in the models, it did not change the direction of the association between infant malaria and PNIM, which indicates that the use of SP is not an explanation for this association. It is possible that low-grade infection with malarial parasites in infants triggers immune responses to malaria that also result in reduced infant HIV-1 loads. Malaria is known to shift the immune system toward Th1-type responses, and progression of HIV-infection has been associated with a shift from a Th1-type to a Th2-type response [29]. In addition, malarial infection can activate chemokines such as MIP-1 and MIP-1, among others [30], that can compete for the HIV entry receptor CCR5 and thereby slow the progression of HIV infection in infants. As explained earlier, malarial infection can also activate the Treg population, which can modulate HIV-associated disease progression [31]. Another possibility is that malaria infection activated innate immune responses, which can help to control HIV-associated opportunistic infections, leading to increased survival time. Among HIV-negative infants of HIV-seropositive women, moderate to severe infant anemia was a significant risk factor for PNIM. Infant anemia may be the culmination of a sequence of events whereby maternal HIV infection, placental malaria, infant malaria, and SES may all contribute. Whatever the contributing factor, moderate to severe anemia in infants can be detected and should be responded to, particularly among infants of HIV-positive women. Our study had several limitations. First, we only enrolled HIV-infected women with no AIDS-defining symptoms; PNIM among women with progressed HIV disease is likely to be higher than our present findings. Our measurement of parasitemia by microscopic examination of smears was inexact compared with other methods, may have misclassified particularly low-density parasitemias, and did not identify intervillous inflammation, but this would have resulted in bias toward the null. The number of HIV-infected infants born to mothers with placental malaria was small; this precludes more-definitive answers and does not rule out a chance finding. An infant needed to make at least 2 follow-up visits for us to be able to make an HIV diagnosis, and loss to follow up was considerable, which may have biased our results. However, the reported PNIM among HIV-infected infants was comparable to results obtained in eastern Africa from a pooled analysis of trials [32], and, consistent with previous reports, a low CD4+ cell count was a significant risk factor for PNIM [32, 33]. Although excluded infants had a lower mean birth weight, the percentage of LBW was similar among both groups, and other characteristics between excluded and included infants were similar as well. Deaths among the excluded infants were not associated with placental malaria. Furthermore, placental malaria was not linked with known factors that have been associated with infant HIV disease progression, such as maternal viral load and maternal CD4+ cell count [12, 33, 34]. Our findings about the effect of placental malaria and infant malaria on PNIM are consistent with reports from other studies [19, 27]. In summary, in this study population of infants of HIV-seropositive women, PNIM was associated with infant HIV infection, moderate to severe infant anemia, LBW, low SES, and low maternal CD4+ cell count. Our results do not confirm the findings of Bloland et al. [10] and do not support our hypothesis that placental malaria may enhance HIV disease progression in the HIV-infected infant. By contrast, we found that infants born to HIV-infected mothers with placental malaria do not have an increased risk of PNIM and that malaria in the HIV-infected infant may further contribute to a protective effect from PNIM, which suggests a complex relationship between maternal and infant immune responses to malaria and HIV. The role of placental malaria in the context of HIV and its effect on PNIM merits further study.