INTRODUCTION
In areas of stable malaria transmission, residents acquire immunity by adulthood that limits infection and prevents severe disease due to Plasmodium falciparum. Immunity is compromised in first-time mothers, who frequently experience peripheral parasitaemia as well as dense accumulations of parasites and inflammatory cells in the placenta. Placental malaria is associated with poor outcomes, both for mothers who often develop severe anaemia and for newborns who are commonly underweight. Malaria-related low birth weight (LBW) is estimated to cause about 100 000–200 000 infant deaths each year (Guyatt and Snow, Reference Guyatt and Snow2001; Murphy and Breman, Reference Murphy and Breman2001).
Susceptibility to malaria infection and disease during pregnancy has been a longstanding mystery. Many observers presumed that ‘immunosuppression’ during pregnancy is required to prevent rejection of the foetal allograft, and that this also increased susceptibility to malaria. However, this hypothesis did not explain the peculiar epidemiology of pregnancy malaria due to P. falciparum, in which resistance increases over successive pregnancies when women live in an area of stable malaria transmission. Antimalarial prophylaxis confers few benefits in multigravid women (Greenwood et al. Reference Greenwood, Armstrong, Byass, Snow and Greenwood1992), indicating that immunosuppression is not increasing susceptibility during later pregnancies, and is therefore unlikely to be increasing susceptibility during first pregnancies.
To explain parity-related resistance, MacGregor speculated that local uteroplacental immune responses were required for protection, and that these might develop over successive pregnancies (McGregor, Reference McGregor1984). However, responses that localize to the uteroplacental unit have never been properly defined. Ultimately, the discovery that a distinct binding form of P. falciparum caused pregnancy malaria provided an explanation for many of the epidemiological and clinical features of this syndrome. The finding also offered new strategies for intervention.
A DISTINCT PARASITE PHENOTYPE CAUSES PLACENTAL MALARIA GLOBALLY
Among the four human malaria parasite species, only P. falciparum has the property of binding to endothelium and sequestering in deep vascular beds. Early observers deduced the nature and importance of parasite adhesion. Alphonse Laveran, who discovered the malaria parasite (Laveran, Reference Laveran1882), observed that the organ-specific syndromes of falciparum malaria (e.g. cerebral malaria) corresponded to the burden of malaria parasites sequestered in the affected tissue (Laveran, Reference Laveran1884). Detailed quantitative histological studies have supported this relationship (MacPherson et al. Reference Macpherson, Warrell, White, Looareesuwan and Warrell1985). Marchiafava and Bignami – Italian contemporaries and rivals of Laveran – deduced that parasite-infected erythrocytes (IE) bind to endothelial surfaces, leading to parasite sequestration (Marchiafava and Bignami, Reference Marchiafava, Bignami and Charles1894).
IE are known to bind numerous host molecules, such as CD36, ICAM-1, and the glycosaminoglycan (GAG) chondroitin sulfate A (CSA) among several others. IE collected from infected individuals commonly bind CD36, sometimes in combination with other receptors. IE collected from the human placenta have a distinct binding phenotype – they bind to CSA but do not bind to CD36 (Fried and Duffy, Reference Fried and Duffy1996). The discovery of CSA-binding parasites in the placenta was doubly surprising, since adhesion had previously been discounted as the basis for placental sequestration (Bray and Sinden, Reference Bray and Sinden1979). IE in the placenta accumulate throughout the intervillous spaces of the placenta, most without obvious attachment to the placental lining (Fried and Duffy, Reference Fried and Duffy1996). After the discovery of placental CSA-binding parasites, CSA was shown to be distributed throughout the intervillous spaces of infected placentas, and to co-localize with placental IE (Muthusamy et al. Reference Muthusamy, Achur, Bhavanandan, Fouda, Taylor and Gowda2004). These data support the notion that adhesion to CSA accounts for placental sequestration of parasites.
All studies of placental IE have identified adhesion to CSA as a common property of placental parasites. In addition to CSA, other placental receptors for parasite adhesion have been proposed, including hyaluronic acid (HA) and Fc receptors. Like CSA, HA is a GAG. Many preparations of HA contain CSA as a contaminant, and this CSA contaminant can support IE binding (Fried, Lauder and Duffy, Reference Fried, Lauder and Duffy2000; Valiyaveettil et al. Reference Valiyaveettil, Achur, Alkhalil, Ockenhouse and Gowda2001). In those studies that used pure HA without CSA contamination, binding of placental IE to placental cryosections was not inhibited by soluble HA (Fried et al. Reference Fried, Lauder and Duffy2000, Reference Fried, Domingo, Gowda, Mutabingwa and Duffy2006). A new study has found that placental IE do not bind to HA, and in detailed biochemical studies determined that HA was not detectable in placentas after removing umbilical cord tissue (Muthusamy et al. Reference Muthusamy, Achur, Valiyaveettil, Botti, Taylor, Leke and Gowda2007). This indicates that HA is not available in the intervillous spaces of the placenta to support IE binding. Curiously, enzymatic digestion of placental cryosections with specific hyaluronidases inhibits binding of placental IE (Rasti et al. Reference Rasti, Namusoke, Chene, Chen, Staalsoe, Klinkert, Mirembe, Kironde and Wahlgren2006). Further studies are required to substantiate a role for HA in placental sequestration.
Similarly, the role of Fc receptors in placental IE adhesion remains unclear. IE can adsorb immunoglobulin (Ig) G to its surface, and this prompted the hypothesis that IgG may act as a bridge for IE to bind to neonatal Fc receptor (nFcR) on the syncytiotrophoblast (Flick et al. Reference Flick, Scholander, Chen, Fernandez, Pouvelle, Gysin and Wahlgren2001). However, nFcR is not expressed on the syncytiotrophoblast surface, does not bind IgG at physiological pH (7·4), and therefore is unlikely to mediate IE adhesion. Similarly, the distribution and binding properties of other IgG-binding molecules expressed by the placenta, such as annexin, placental alkaline phosphatase and other Fc receptors, suggest that they do not support binding of IgG at the syncytiotrophoblast surface (Simister and Story, Reference Simister and Story1997). Studies of freshly collected placental IE have yielded conflicting data as to whether placental IE may use IgG to bind to the syncytiotrophoblast (Fried et al. Reference Fried, Domingo, Gowda, Mutabingwa and Duffy2006; Rasti et al. Reference Rasti, Namusoke, Chene, Chen, Staalsoe, Klinkert, Mirembe, Kironde and Wahlgren2006). More data are needed to resolve this controversy.
IMMUNITY TO PLACENTAL IE EXPLAINS PARITY-RELATED RESISTANCE
As noted above, resistance to pregnancy malaria is acquired over successive pregnancies. Among pregnant women living in areas of stable malaria transmission, first-time mothers have the highest frequency of parasitaemia and the most severe sequelae. The observation that placental IE have a distinct binding phenotype suggested a possible explanation for the parity-related pattern of susceptibility. Women may be naïve to CSA-binding parasites at the time of their first pregnancy, and then acquire antibodies over successive pregnancies that control this parasite form.
Immunoepidemiology studies have consistently supported this idea. Over successive pregnancies, women acquire antibodies that block binding of placental parasites (Fried et al. Reference Fried, Nosten, Brockman, Brabin and Duffy1998b), and antibodies that react with the surface of CSA-binding parasites (Ricke et al. Reference Ricke, Staalsoe, Koram, Akanmori, Riley, Theander and Hviid2000). Antibodies that agglutinate placental IE may increase over successive pregnancies, although placental IE agglutinate poorly compared to other parasites (Beeson et al. Reference Beeson, Brown, Molyneux, Mhango, Dzinjalamala and Rogerson1999; Maubert et al. Reference Maubert, Fievet, Tami, Cot, Boudin and Deloron1999). Cellular responses to CSA-binding parasites also increase as parity increases (Fievet et al. Reference Fievet, Tami, Maubert, Moussa, Shaw, Cot, Holder, Chaouat and Deloron2002).
Immunity to placental parasites or CSA-binding parasites is specifically related to improved outcomes. Anti-adhesion antibodies are related to protection from infection for the mother (Fried et al. Reference Fried, Nosten, Brockman, Brabin and Duffy1998b), and are related to increased birth weight and gestational age for the newborn (Duffy and Fried, Reference Duffy and Fried2003). Antibody that reacts with the surface of CSA-binding parasites is related to increased haemoglobin levels and increased birth weight among women with chronic forms of placental malaria (Staalsoe et al. Reference Staalsoe, Shulman, Bulmer, Kawuondo, Marsh and Hviid2004). These findings have strongly supported the idea of a pregnancy malaria vaccine that would elicit immunity to placental IE surface antigens before first pregnancy in order to improve maternal and foetal outcomes.
MALARIA HAS DIFFERENT OUTCOMES IN DIFFERENT WOMEN
First-time mothers are more susceptible to malaria than other mothers. However, the course of malaria also differs in first-time mothers. Placental parasite densities are significantly higher during first pregnancy. Inflammatory infiltrates in the placenta are frequently intense and sometimes massive during first pregnancy (Ordi et al. Reference Ordi, Ismail, Ventura, Kahigwa, Hirt, Cardesa, Alonso and Menendez1998), whereas placental malaria in multigravid women typically elicits little if any inflammatory infiltrate. Finally, the duration of placental malaria is longer in first-time mothers than other mothers, and the inflammatory cytokine responses in their placentas may extend beyond the resolution of parasitaemia (Fried et al. Reference Fried, Muga, Misore and Duffy1998a).
These different clinicopathological presentations profoundly modify the sequelae of pregnancy malaria for both the mother and her offspring. Well-known sequelae of pregnancy malaria, such as maternal anaemia and LBW offspring, are more frequent and severe during first pregnancy (Brabin and Rogerson, Reference Brabin, Rogerson, Duffy and Fried2001). However, the sequelae of placental malaria may be substantially greater than previously recognized, in part because there may be insidious effects that have been masked by parity.
Preeclampsia is a deadly hypertensive disorder of pregnancy that, like falciparum malaria, is most frequent during first pregnancy. A possible relationship between placental malaria and preeclampsia has been unclear. In Senegal, placental malaria was more common in women with preeclampsia (Sartelet et al. Reference Sartelet, Rogier, Milko-Sartelet, Angel and Michel1996), while studies in Kenya failed to find a relationship between pregnancy malaria and preeclampsia (Shulman et al. Reference Shulman, Marshall, Dorman, Bulmer, Cutts, Peshu and Marsh2001; Dorman et al. Reference Dorman, Shulman, Kingdom, Bulmer, Mwendwa, Peshu and Marsh2002). Data from Tanzania (Muehlenbachs et al. Reference Muehlenbachs, Mutabingwa, Edmonds, Fried and Duffy2006) now suggest that both earlier studies may have been correct, because placental malaria causes hypertension in some mothers but not others. The risk of hypertension in infected women is strongly influenced by parity and age.
In Tanzania, young first-time mothers with placental malaria had significantly increased risk of hypertension (Muehlenbachs et al. Reference Muehlenbachs, Mutabingwa, Edmonds, Fried and Duffy2006). Interestingly, hypertension in these women may result from a conflict between the mother and the foetus during the inflammatory response to placental infection. In older multigravid women, placental malaria lowered blood pressure, just as malaria lowers blood pressure in non-pregnant individuals. When data from all mothers were examined in aggregate, blood pressure measurements were similar between infected and uninfected women. The aggregate analysis masked the significant but opposing effects of placental malaria on blood pressure among different groups of women.
Similarly, important effects of pregnancy malaria on the offspring may be masked by an interaction with parity. In Cameroon, children born to mothers with placental malaria experienced parasitaemia more frequently than other children, although this relationship was not significant (Le Hesran et al. Reference Le Hesran, Cot, Personne, Fievet, Dubois, Beyeme, Boudin and Deloron1997). Studies in Tanzania confirm that the risk of parasitaemia is different among children born to infected women. However, the risk of parasitaemia may either increase or decrease, again depending on the parity of the mother. The frequency of parasitaemia increases among infants born to infected multigravid women, but the frequency decreases among infants born to infected first-time mothers (Mutabingwa et al. Reference Mutabingwa, Bolla, Li, Domingo, Li, Fried and Duffy2005).
Future studies will need to assess how pregnancy malaria modifies the risk of severe malaria and malaria-related death in offspring. Pregnancy malaria is related to perinatal mortality: chemoprophylaxis during first pregnancies reduces perinatal mortality by a quarter or more (Garner and Gulmezoglu, Reference Garner and Gulmezoglu2006), similar to the effect of insecticide-treated bednets (Gamble et al. Reference Gamble, Ekwaru, Garner and Ter Kuile2007). However, the total impact of placental malaria on post-neonatal mortality remains unclear. Because LBW is a strong risk factor for infant mortality, LBW due to placental malaria has been estimated to cause 100 000 or more infants deaths each year. If placental malaria is also modifying the risk of severe malaria during early childhood, then the impact of placental malaria on children's health in tropical countries may be much greater than we currently appreciate.
SHORT-TERM AND LONG-TERM SOLUTIONS
The impact of pregnancy malaria on the health of mothers and children is great. In areas of high malaria transmission, nearly all pregnant women are likely to be infected sometime during pregnancy (Brabin, Reference Brabin1983). Rates of placental malaria at the time of delivery can be as high as 70%. Some effects of placental malaria are well-known, such as maternal anaemia and LBW infants. Other effects of placental malaria may be insidious, such as maternal hypertension or modified risk of malaria in the offspring.
Unfortunately, the tools available to prevent pregnancy malaria are narrowing. Insecticide-treated bed nets are a useful tool, but only reduce malaria infection and disease by about a quarter to a third, although the size of this benefit may vary in different populations (Gamble et al. Reference Gamble, Ekwaru and Ter Kuile2006). Prophylaxis with chloroquine was a cheap and effective strategy until chloroquine-resistant parasites spread around the globe. Intermittent presumptive treatment during pregnancy (IPTp) with sulfadoxine-pyrimethamine (SP) replaced chloroquine as a preventative strategy in Africa. However, this strategy may soon falter with the rapid spread of SP-resistant parasites throughout the continent.
A vaccine is the best long-term strategy to prevent pregnancy malaria. Women develop resistance to pregnancy malaria over successive pregnancies, and this correlates with the acquisition of antibody against placental IE. A vaccine should reasonably target the surface proteins of placental IE to elicit protective immunity. The target(s) of these antibodies is (are) thought to be relatively conserved, because sera from multigravid women cross-react to placental and CSA-binding IE collected from around the world (Fried et al. Reference Fried, Nosten, Brockman, Brabin and Duffy1998b).
The gene var2csa is uniformly up-regulated in placental IE (Salanti et al. Reference Salanti, Staalsoe, Lavstsen, Jensen, Sowa, Arnot, Hviid and Theander2003). The VAR2CSA protein is a member of the PfEMP1 family of variant surface antigens, and it has become the primary focus of an international effort to develop a pregnancy malaria vaccine (see articles by Hviid and Salanti and by Scherf in this special issue). In addition to var2csa, a limited repertoire of transcripts (Francis et al. Reference Francis, Malkov, Oleinikov, Rossnagle, Wendler, Mutabingwa, Fried and Duffy2007) and IE membrane-associated proteins (Fried et al. Reference Fried, Hixson, Anderson, Ogata, Mutabingwa and Duffy2007) are also preferentially expressed by placental parasites, and may contribute to the protective efficacy of a vaccine.
The next steps to develop a vaccine will be to identify the protein domains and variants that are targeted by functional antibodies, and to express these as recombinant immunogens that sensitize women to the surface antigens of placental IE. A pregnancy malaria vaccine could be given to women at the outset of their reproductive years, in order to elicit immune responses that would be boosted during encounters with CSA-binding parasites during pregnancy. Such a vaccine may provide immunological protection from the worst effects of pregnancy malaria, and the benefits would be greatest among first- and second-time mothers.
ACKNOWLEDGMENTS
PED is supported by grants from NIH (R01 AI52059, U19 AI065664) and the Bill & Melinda Gates Foundation (Grant 29202) to study pregnancy malaria pathogenesis and interventions.
