Background
Malaria during pregnancy poses a significant threat to both the mother and unborn child. For the mother, it increases the risk of illness, severe anaemia and death; for the unborn child it increases the risk of intra-uterine growth retardation and low birth weight, spontaneous abortion and stillbirth (reviewed in [
1]). Numerous epidemiological studies have highlighted the various aspects of malaria during pregnancy both in highly endemic regions (mostly from sub-Saharan Africa) and in regions where malaria transmission is low and sporadic (mainly from Asia) (see e.g. [
1‐
4]), and it has become clear that the pathology of malaria during pregnancy and its evaluation are highly dependent on the particular epidemiological setting, due to differences in acquired immunity in women reaching child-bearing age.
Acquired immunity to
Plasmodium falciparum, the most virulent agent of human malaria, is a gradual process by which individuals build up a repertoire of protective immune responses over years of repeated exposure. Although sterilizing immunity might never be attained, people living in malaria endemic areas seem to acquire protection against clinical malaria after a certain period of exposure [
5,
6] and to severe disease after only a few infections [
7]. In regions of sustained transmission, most women of child-bearing age are, therefore, clinically immune against the disease. Malarial infections during pregnancy in these regions are usually associated with placental sequestration of parasites adhering to glycosaminoglycan chondroitin sulphate A (CSA) through the expression of
var2csa [
8‐
11], a member of the
var gene family which encodes the immunodominant surface antigens and virulence factors PfEMP1 (
P.falciparum membrane protein 1); this phenomenon is also known as pregnancy-associated malaria or PAM.
Despite the vast overall diversity of
var genes, serum samples from women of different geographical backgrounds have shown that the antigens expressed in infected placentas are highly conserved [
12,
13]. As a consequence, immunity against PAM is rapid: usually after one or two episodes, women will be protected during subsequent pregnancies [
14]. This is also supported by field studies from highly endemic regions showing that PAM predominantly affects primigravidae, i.e. women in their first pregnancy, rather than multigravidae. This epidemiological picture does not, however, extend to low transmission areas where primi- and multigravid women can be equally affected by PAM. Also, pregnant women from regions with low and unstable transmission invariably become symptomatic upon infection, and as a consequence it is harder to differentiate between PAM and normal, i.e. non-PAM, malaria unless placental samples are examined and reported; this, however, is rarely the case in the published literature.
The assessment of the extra burden caused by PAM, therefore, has to take into account the particular transmission setting and is further complicated by at least two important factors. Firstly, women from highly endemic regions with placental malaria rarely show symptoms of disease yet can develop life-threatening anaemia; this can potentially lead to substantial underreporting of maternal and foetal mortality due to placental malaria. Secondly, the rates of mortality which can be attributed to PAM in low transmission settings are hard to assess not only because of the additional threat of clinical disease due to inadequate levels of protection but also because studies rarely report the regions' equivalent (age- and exposure adjusted) mortality rates for men or non-pregnant women. Crucially also, there is still a great need to evaluate the magnitude of the extra burden of PAM, and malaria during pregnancy in general, during epidemic outbreaks.
In this paper, historical data from the Punjab region in former British India is re-examined and used to demonstrate how to qualitatively assess the magnitude of PAM in an epidemic transmission setting by means of excess female mortality. The influence of the mode and intensity of transmission is then investigated by means of a simple mathematical model. Both the data and the model clearly demonstrate that epidemic malaria, especially when intermitted by periods of very low transmission, can have devastating effects on the female population. This calls for active intervention measures not only in highly endemic but also, or even in particular, in regions of unstable malaria transmission and those prone to epidemic outbreaks.
Discussion
The adverse effects of malaria during pregnancy have long been recognized, but not until relatively recently have they gained much attention in terms of public health and intervention measures. It is clear from earlier work that malaria during pregnancy takes a different dynamic as the intensity of transmission changes through location and time [
1,
2,
19]. There are also significant differences in its pathology and it is, therefore, useful to distinguish between placental malaria, or PAM, which specifically affects pregnant women and is caused by placental sequestration of
P. falciparum parasitized red blood cells, and 'normal' malaria infections during pregnancy which can also, yet to a lesser degree, be caused by other Plasmodium species with adverse consequences for the mother and unborn child [
20].
Placental malaria is the most common cause for low birth weight (LBW) in high transmission areas. It has been identified as the single greatest risk factor for neonatal mortality and contributor to infant mortality with an estimated 7-fold increase in the risk of still birth. Furthermore, it has recently been reported that children born to mothers with placental malaria also have an increased risk of malaria during their first few years of life [
21]. In unstable and low transmission settings, on the other hand, malaria infections are more likely to cause clinical disease. Consequently, preterm delivery caused by febrile episodes is the biggest risk factor for low birth weight (LBW) in children in these regions. It appears, therefore, that in low endemic and unstable transmission settings mothers are the ones mostly affected by malarial infections during pregnancy, whereas in regions where transmission is intense, the burden is carried mainly by the foetus. However, women with symptomatic infections are more likely to seek out medical attention (if at all available), thus potentially reducing the risk of both severe disease and the establishment of placental malaria. This could lead to an underestimation of the risk of PAM in low and unstable transmission regions, and may account for the low measures reported in those settings [
3,
22,
23].
Here, a method of using (historical) longitudinal data on excess female mortality has been used to qualitatively assess the impact of PAM on the female population in a malaria endemic area with considerable variations in transmission intensities. Although at this stage this can only be used as a proxy for the absolute level of mortality attributable to PAM, it clearly indicates a significant contribution of pregnancy-associated malaria to female mortality, especially during epidemic outbreaks. This approach is based on the observation that the ratio of female to male mortality is not constant over time but shows a strong correlation with transmission intensity. That is, if extra female deaths were simply due to complications during labour and/or infections during pregnancy, this number would expectedly be relatively constant and independent of the intensity of malaria transmission. In addition, excess female mortality is negatively and (possibly) non-linearly correlated with prior exposure to malaria as would be expected if it were attributable to PAM. Finally, the significant increase in excess mortality during an epidemic outbreak, compared to non-epidemic years, clearly demonstrates the existence and damaging effect of pregnancy specific malaria in a region where transmission is moderate and unstable.
A simple mathematical model demonstrates how overall female morbidity may be expected to increase with transmission intensity while shifting the contribution from women of all gravidity towards those in their first pregnancy. In low transmission settings there is a chance that women reach a non-reproductive stage without ever contracting PAM, whereas in hyper-endemic regions most women will contract the disease during their first pregnancy. Interestingly though, it was found that the contribution from multigravidae to excess female morbidity is not linear but peaks at low to medium levels of transmission and, as expected, diminishes towards to upper end of the endemicity spectrum. It is also clear from this analysis that the total average number of pregnancies impacts the magnitude of morbidity among females, at least in regions of low and unstable malaria (see Additional file
1). Thus, a reduction in the average number of pregnancies (for example through education on birth control) could have an impact on mortality in areas where transmission intensity is moderate but would have little effect in hyperendemic situations. This may, however, be altered by the spread of HIV as recent reports suggest that HIV infection can impair humoral immunity to placental type VSA [
24,
25]. If so, the age structure (and overall incidence) of PAM could shift in a direction, even in hyperendemic areas, that would warrant intervention through birth control.
The greatest effect of PAM on the female population observed within this framework was by incorporating seasonal variations in transmission intensity. Given sufficiently low background transmission rates, many women will go through successive pregnancies without getting infected, leaving a significant proportion of the reproductively active female population still susceptible to PAM. As a result, and in accordance with the data presented here, the observable excess in female morbidity during epidemic outbreaks is significantly higher than within a stable transmission setting and equivalent transmission intensity, and even more so as the level of transmission in the period preceding an epidemic is especially low.
These data also support the current conception of PAM as being particularly severe among females with low exposure to non-PAM malaria as compared to those living in hyperendemic areas. There are a number of reasons why this may be the case. In hyperendemic areas, women of childbearing age will typically have protective antibodies to the other PfEMP1 proteins encoded by the
var genes of parasites expressing placental binding
var2csa, and this may protect them from disease while allowing infection of the placenta (which still has severe consequences for the foetus); in areas of low transmission, women of childbearing age may not have immune responses against the other PfEMP1 proteins encoded by the infecting parasite and thus may experience disease as well as infection. Individuals living in areas of high transmission may also have antibodies to VAR2CSA itself without having sustained a placental infection [
26], although it remains to be shown if the antibody levels reported are protective against disease. Finally, there are of course a very large range of other immune responses that could prevent an outcome of severe complicated malaria upon infection with parasites expressing placental binding
var2csa. However, as men and women alike will lack these forms of protective immunity in areas of low transmission, the excess mortality attributed to PAM strongly implies that placental infection itself can greatly increase the chances of death among individuals with incomplete immunity. Indeed, if immunity to severe disease is acquired upon one or two infections [
7], the analyses presented in this paper indicates that this form of immunity clearly does not extend to PAM. The unusually conserved nature of VAR2CSA makes it an attractive vaccine candidate, but until this vaccine becomes available, ongoing and future prevention measures, such as intermittent preventive treatment (IPTp) against malaria, in particular during pregnancy (see e.g. [
27,
28]), have to be urgently employed not only in regions of extensive levels of transmission but equally in regions where malaria is more sporadic.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MR, SG, PB and APD devised and analyzed the model; MJB provided and analyzed the data and contributed to the writing of the manuscript. MR, APD and SG wrote the manuscript.