Authors Sharon Cox (corresponding author) and Betty Kirkwood, Nutrition and Public Health Intervention Research Unit, LSHTM, Keppel St, London, WC1E 7HT, UK. Tel.: +44 207 958 8132; E-mail: Sharon.Cox@lshtm.ac.uk, Betty.Kirkwood@lshtm.ac.uk Trine Staalsoe and Lars Hviid, Centre for Medical Parasitology, Copenhagen University Hospital (Rigshospitalet) and Institute for Medical Microbiology and Immunology, University of Copenhagen, Copenhagen, Denmark. Tel.: +45 3545 7375; E-mail: firstname.lastname@example.org, email@example.com Judith Bulmer, Department of Pathology, University of Newcastle, Royal Victoria Infirmary, Newcastle upon Tyne, NE3 4PH, UK. Tel.: +44 191 222 7144; E-mail: J.N.Bulmer@ncl.ac.uk Harry Tagbor and Eleanor Riley, Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, Keppel Street, London, WC1E 7HT, UK. Tel.: +44 207 299 4751; E-mail: Harry.Tagbor@lshtm.ac.uk, Eleanor.Riley@lshtm.ac.uk Chris Frost, Medical Statistics Unit, LSHTM, Keppel St. London, WC1E 7HT, UK. Tel.: +44 207 927 2242; E-mail: Chris.Frost@lshtm.ac.uk
Background Vitamin A supplementation is believed to enhance immune responses to infection but few studies have assessed its effects on anti-malarial immunity, especially during pregnancy when women are at increased risk from both vitamin A deficiency and pregnancy-associated malaria. The pathological effects of malaria in pregnancy are believed to be due to the sequestration of parasites in the placenta mediated via binding of variant surface antigens (VSA) expressed on the surface of P. falciparum infected red blood cells to placental chondroitin sulphate A (CSA).
Methods We conducted a randomized double-blind controlled trial of vitamin A supplementation in 98 primigravid Ghanaian women to investigate the effects of vitamin A supplementation on levels of IgG antibodies binding to VSA of a clinical, P. falciparum placental isolate and to two isolates selected (or not) for adherence to CSA in vitro (anti-VSACSA IgG or anti-VSA IgG). Placental malarial infection was determined by placental blood smear and histology.
Results Vitamin A supplementation was non-significantly associated with a decreased risk of active or chronic-active placental malarial infection compared to past, resolved infection at delivery, as determined by histology (OR = 0.42, P = 0.13 – adjusted for level of education). After adjustment for differences in baseline values, levels of anti-VSACSA IgG to a placental, CSA-adherent isolate (EJ-24) but not to two isolates selected for CSA-adhesion in vitro (FCR3CSA and BusuaCSA), were significantly lower in women receiving vitamin A supplementation than in women receiving placebo (P = 0.002). There was no apparent effect of vitamin A supplementation to levels of Ab to non-CSA-adherent parasite isolates.
Conclusions The data suggest that the reduction in the levels of anti-VSACSA antibodies to the known placental malaria isolate may reflect reduced intensity or duration of placental parasitaemia in women receiving vitamin A supplementation. These observations are of potential public health significance and deserve further investigation.
Subclinical vitamin A deficiency – i.e. low plasma retinol or liver stores in the absence of ocular signs – is associated with increased mortality from infection (McLaren & Frigg 2001). Community trials of vitamin A supplementation have led to large reductions in infection-related child mortality (Beaton et al. 1993), suggested to be due to an effect on progression to severe disease rather than on the prevalence or incidence of infection (Ghana-VAST-study-team 1993). Evidence suggests this effect may be mediated through an impact of vitamin A on immune function (McLaren & Frigg 2001; Stephensen 2001).
Vitamin A deficiency in animal models of malaria infection is associated with increased parasitaemia and increased mortality (Krishnan et al. 1976; Stoltzfus et al. 1989) and, in apparent agreement with this, a randomized controlled trial of vitamin A supplementation showed a decreased risk of clinical malaria in children in Papua New Guinea (Shankar et al. 1999) although no effect was reported in an earlier trial in Ghana (Binka et al. 1995).
Pregnant women are at particular risk of vitamin A deficiency (West 2002) and are also at increased risk of pregnancy-associated malaria and its associated complications, including severe maternal anaemia, still birth and low birth weight (McGregor et al. 1983; Steketee et al. 2001). Increased susceptibility to malaria is attributed to generalized suppression of cell-mediated immune responses (Roberts et al. 1996; Smith 1996) and/or preferential sequestration of parasitized red blood cells (pRBC) in the placenta (Fried & Duffy 1996; Andrews & Lanzer 2002). Placental pRBC bind to complex polysaccharides, for example chondroitin sulphate A (CSA) and hyaluronic acid (HA), in the placental intervillous space. Binding is mediated by variant erythrocyte surface antigens (VSA) encoded by highly polymorphic, multigene families within the parasite genome, members of which also mediate binding (via numerous cell surface receptors) to vascular endothelium throughout the body, reviewed in (Kyes et al. 2001). VSA mediating sequestration of pRBC in the placenta are antigenically distinct from those that mediate pRBC sequestration elsewhere in the body (Beeson et al. 1999; Maubert et al. 1999; Ricke et al. 2000; Staalsoe et al. 2001).
Parity-dependent development of immunity to pregnancy-associated malaria has been linked to acquisition – during first and second pregnancies – of anti-VSA antibodies that specifically block pRBC sequestration in the placenta (Fried et al. 1998; Staalsoe et al. 2001). However, exposure to novel VSAs during first and subsequent pregnancies requires the generation of primary immune responses and such responses may be inhibited by vitamin A deficiency. To determine the effect of vitamin A deficiency/supplementation on acquisition of antibodies to pregnancy-associated VSAs we have evaluated anti-VSA Ab responses in a cohort of primigravid women enrolled in a randomized double-blind placebo-controlled trial of vitamin A supplementation.
Study area and subjects
Primigravid pregnant women were recruited from antenatal clinics at Nkoranza District Hospital and three rural health clinics in Brong Ahafo region, Central Ghana from March to June 2001. In this area malaria transmission is perennial and intense (Browne et al. 2000). The prevalence of severe, subclinical vitamin A deficiency (26% breast milk retinal <20 μg/dl; arthur, unpublished) indicates that vitamin A deficiency is a significant public health problem in this community (WHO 1996). Primigravid women were eligible for inclusion in the study if they were resident within the study area, in good health and less than 24 weeks pregnant, as estimated by fundal height measurements. If women were too early in their pregnancies for accurate palpation, pregnancy was confirmed by a urine test. According to available health records, none of the women were suffering from HIV infection or tuberculosis; diagnostic tests were not performed, but the estimated prevalence of HIV infection amongst antenatal attendees in the Brong Ahafo region for 2000 was 1.6% (Ministry of Health 2001).
This study was a randomized double-blind placebo controlled trial. Using balanced block randomization 100 women were randomized (eight women per block) to one of eight sets of capsules, four of which contained vitamin A. The coding was unknown to the investigators, until after completion of the trial and the capsules were identical in appearance. The capsules were given weekly and contained 10 000 IU of vitamin A as retinyl palmitate in groundnut oil, plus tocopherol as a preservative or groundnut oil and tocopherol only in the placebo capsules. It was estimated that the study had >90% power at a 5% significance level to detect 30% differences in plasma anti-VSACSA concentrations between the vitamin A and placebo group, assuming an Ab level (expressed as a percentage of the positive control) of 78.0 and a standard deviation of 28.6; assumptions were based on Ab data obtained from a highly comparable study of Cameroonian primigravid women (Staalsoe et al. 2001).
Capsules were taken at home under supervision of a field worker each week from enrolment until 6 weeks post-partum. All study women were actively encouraged to comply with the hospital policy for antimalarial prophylaxis and nutritional supplementation. In practice, all women received a full treatment regime of CQ at recruitment and daily iron and folate supplements were provided free of charge through the hospital pharmacy. Women who were found to have peripheral parasitaemia or moderate or severe anaemia at any of the follow-up visits were provided with treatment, free of charge, at the local hospital or health centre.
Of the 100 women randomized, two were later excluded, due to an early miscarriage and a false pregnancy (Figure 1). Of the remaining 98 women, only one (in the placebo group) was lost to follow up prior to delivery but 10 women (seven in the placebo group, three in the vitamin A group) missed the late pregnancy visit (planned to be within the last month of gestation based on antenatal records) due to premature or earlier than expected deliveries. Placental samples were collected from 76 out of 78 study women who delivered in local health facilities. Before the final blood collection at 6 weeks post-partum two women were lost to follow up through migration (one in each group) and one woman died (vitamin A group) (Figure 1). Birth weights were extracted from routine data collection for women delivering at local health facilities.
Venous blood samples (4 ml) were collected into heparinized vacutainers (Greiner) at recruitment, in late pregnancy and at 6 weeks post-partum. At the same time, finger prick blood samples were obtained for assessment of plasma haemoglobin (Hb) concentration (HemoCue) and determination of malarial parasitaemia by microscopic examination of Giemsa-stained thick blood films. Blood films were independently examined by two observers and the numbers of parasite infected erythrocytes counted per 200 leukocytes. Parasite densities were generated from the mean of the two readings and using an assumed leukocyte count of 8000 leukocytes per μl of blood. Plasma retinol was measured in baseline samples by high performance liquid chromatography (Bankson et al. 1986) from plasma stored in UV protected vials for a maximum of 5 months at −30 °C.
Placenta samples were collected from women delivering in one of the study health facilities as previously described and classified according to published criteria (Bulmer et al. 1993). In brief, samples approximately 1–2 cm3 were cut from the maternal side of the placenta and placed in 60 ml of 10% formalin. Samples were embedded in wax and sections 1 μm thick were cut onto slides, processed and stained with Giemsa and haematoxylin and eosin Y. All slides were independently assessed by two examiners (JNB and SEC). In addition, thick blood smears were made from blood dripped from the excised placental samples.
Three genetically distinct isolates of P. falciparum were used. FCR3 is an established laboratory isolate (Jensen & Trager 1978), Busua was originally collected from a non-immune Danish man presenting with uncomplicated malaria upon return from the coastal region of Ghana (Staalsoe et al. 2004) and EJ-24 was isolated from the placenta of a woman in the Ashanti Region of Ghana (Salanti et al. 2003). The genotypic identities of all isolates were regularly confirmed by genotypic profiling at the polymorphic msp-1, msp-2 and glurp loci (Snounou et al. 1999). Maintenance of parasite cultures and selection of isolates for adhesion to CSA were as previously described (Cox et al. 2005). Confirmation of adhesion to CSA of the placental parasite isolate and of the CSA-selected parasite lines, and of the sex dependent recognition of VSACSA, is reported elsewhere (Cox et al. 2005).
Plasma levels of VSA-specific IgG were measured by flow cytometry as previously described (Staalsoe et al. 1999). In brief, purified late stage parasites were stained with ethidium bromide, sequentially incubated with human test plasma, goat anti-human IgG (Dako) and fluorescein isothiocyanate-conjugated rabbit anti-goat immunoglobulin (Dako) and analysed by flow cytometry (EPICS XL-MCL, Coulter Electronics). The binding of anti-VSA IgG was quantified as the mean fluorescence index (MFI). For each parasite line, all test and control plasma samples were tested on the same day using parasites from the same batch. Pooled plasma from Ghanaian, malaria-exposed men was used as a negative control for the CSA-adherent parasite lines. Pooled plasma from multigravid, pregnant, Ghanaian women, previously demonstrated to have high levels of anti-VSA IgG (Ricke et al. 2000), was used as a positive control.
Flow cytometric data were analysed by WinMDI software and statistical analyses performed in Stata 7 (Stata Corporation, Texas, USA). Levels of anti-VSA IgG for each sample were measured as MFI and expressed as a percentage of the mean of the MFI from replicate analysis of a pool of plasma samples from Ghanaian multigravid women (positive control). For each parasite line differences between the treatment groups in the levels of anti-VSA IgG at late-pregnancy and post-partum were estimated using analysis of covariance regression models to control for differences at recruitment (baseline). In view of the uneven distribution of levels of education and period of vitamin A supplementation – due to variation in stage of gestation at enrolment – between the treatment groups (Table 1), adjusted differences were estimated from analysis of covariance models that controlled for both of these factors (treated as four and three level categorical variables respectively) and additionally allowed for an interaction between baseline level and enrolment group. Level of education was included in the model as it was not only unevenly distributed between the treatment groups but was also independently associated with levels of anti-VSA IgG, possibly as a result of behavioural or socio-economic differences and consequent variation in risk of malaria infection. There was no evidence of an interaction between enrolment time and treatment group (which would indicate a dose–response effect of the vitamin A) and thus was not included in the reported model. A Mantel Haenszel test was used to test the association between active placental malarial infection at delivery and peripheral parasitaemia at the late pregnancy blood collection, whilst controlling for possible effects of vitamin A supplementation. Odds ratios for the association between vitamin A supplementation and placental malarial infection or low birth weight (<2.5 kg) were calculated using logistic regression, adjusted for baseline imbalances in level of education as described above.
Table 1. Baseline characteristics of women randomly allocated to receive vitamin A supplementation or placebo
Placebo (n = 50)
Vitamin A (n = 48)
BMI, body mass index; MUAC, mid upper arm circumference.
* Gestation was estimated by palpation and measurement of fundal height.
† Although malarial transmission was perennial, transmission peaked in the months at the end and after the two rainy seasons.
‡ Enrolment. Women who enrolled at different times during their pregnancies had different numbers of weeks between the data collections at enrolment and late pregnancy: early, more than 16 weeks; Mid, 12–16 weeks; Late, 7–11 weeks.
Mean plasma retinol (μg/dl) (SD)
30.6 (9.1) n = 47
28.5 (11.1) n = 42
Mean haemoglobin (g/dl) (SD)
9.76 (1.6) n = 50
9.26 (1.3) n = 47
Mean BMI (kg/m2) (SD)
21.3 (1.8) n = 49
21.2 (2.0) n = 48
Mean MUAC (cm) (SD)
26.4 (2.6) n = 50
25.8 (2.0) n = 47
Mean age (years) (SD)
21.0 (2.9) n = 50
21.0 (2.9) n = 48
Mean gestational age (weeks) (SD)*
15.0 (5.6) n = 50
17.0 (4.3) n = 48
Temperature >37.4 °C
Recruited in peak malaria transmission†
Resident in Nkoranza town
Level of education
Individual informed consent was obtained from all study participants. This study was approved by the Ethical Review Committee of the London School of Hygiene and Tropical Medicine and the Ministry of Health, Ghana.
Table 1 shows the baseline characteristics of the two groups. Although mean levels of most variables were similar, there were some reasonably large differences: the most marked being in educational level and gestational age at enrolment. Table 4 also shows that levels of anti-VSACSA IgG to the FCR3CSA parasite line differed between the treatment groups at baseline. Figure 1 shows the extent of the data available at baseline and at each follow up visit. There were considerably fewer data available for the placebo than the vitamin A group at the late pregnancy follow up, due to both missed visits resulting from earlier than expected or premature deliveries and from insufficient plasma volumes available for the laboratory analyses.
Table 4. Effect of vitamin A supplementation on anti-VSA IgG* levels at late pregnancy and post-partum
Vitamin A group
Vitamin A vs. placebo adjusted for baseline IgG levels
Vitamin A vs. placebo adjusted for baseline IgG levels, gestational age at enrolment and level of education
Difference (95% CI)
Difference (95% CI)
* Levels of anti-VSA IgG are expressed as a % (mean MFI of study women × 100/mean MFI for pooled plasma from multigravid women).
−3.6 (−9.0, 1.8)
−13.5 (−24.1, −2.9)
0.1 (−4.6, 4.6)
−7.8 (−18.2, 2.6)
0.61 (−7.0, 8.2)
−6.2 (−14.2, 1.7)
−1.3 (−9.1, 6.6)
−6.8 (−15.9, 2.2)
−3.2 (−8.6, 2.2)
−4.0 (−10.4, 2.3)
−2.2 (−6.8, 2.3)
−4.3 (−9.9, 1.3)
−3.4 (−9.7, 2.9)
−3.0 (−10.3, 4.3)
−0.3 (−4.8, 4.2)
−1.0 (−6.6, 4.7)
1.9 (−1.0, 4.8)
2.2 (−1.2, 5.6)
−0.5 (−4.5, 3.5)
−1.3 (−6.0, 3.3)
Nutritional status at baseline (as indicated by the variables reported in Table 1) was similar in the two treatment groups. At enrolment, 55% (23/42) and 47% (22/47) of women were marginally vitamin A deficient (plasma retinol <30 μg/dl), whilst 17% and 13% were severely deficient (<20 μg/dl) in the placebo and vitamin A groups respectively. Mild to moderate anaemia (plasma Hb <11 g/dl) was common [80% (40/50) and 91% (43/47)] but severe anaemia (<7 g/dl) was observed in only 6% and 8% of the placebo and vitamin A groups respectively. Gestational ages, estimated from fundal height measurements or back-calculated from recorded gestational age at delivery, ranged from 4 to 29 weeks (median = 16 weeks) at enrolment and 25–36 weeks (median = 31 weeks) at the late pregnancy blood collection and did not differ significantly between the treatment groups.
Compliance was uniformly high in both groups with 99.1% of all scheduled doses of vitamin A or placebo being received. Women who enrolled early in their pregnancies had a greater number of total weekly doses of vitamin A or placebo and these were similar between the treatment groups ranging from 16 to 37 (median = 27) for the vitamin A group and 15–38 (median = 24) for the placebo group.
Vitamin A supplementation and malarial infection
Peripheral parasitaemia was common at recruitment and late pregnancy but less common post-partum. There was no apparent effect of vitamin A supplementation on the prevalence or density of peripheral parasitaemia (Table 2). Seventy-six placental samples were collected – 38 from each treatment group – and thick blood smears of placental blood were prepared from 58 of these. On histological examination, placental samples were classified into one of four categories: not infected [no evidence of parasites or haemozoin (malarial pigment) deposition], active infection (parasites and/or pigment in the intervillous spaces, pigment in monocytes, but not in fibrin), active-chronic infection (as for active infection but including pigment in fibrin and/or the chorionic villous synctiotrophoblast) or past-chronic infection (parasites not present and pigment confined to fibrin) (Table 3). A large proportion of placental samples showed evidence of either active or past malarial infection (81% in the placebo group and 92% in the vitamin A group, Table 3). Microscopy of thick blood smears from placental blood revealed the presence of parasites in only 22% (13/58) of women. By comparison with placental histology, examination of placental blood smears picked up only two out of six placentas with active infection, 8 out of 27 with active-chronic infection and 2 out of 33 with past infection. In addition, one histology sample classified as negative was positive by placental blood smear. Although the numbers are small these results confirm previous observations that examination of placental blood smears may seriously underestimate the prevalence of placental malaria infection (Rogerson et al. 2003). There was no significant association between active placental malarial infection at delivery and peripheral parasitaemia measured at the late pregnancy blood collection (χ2 = 2.22 P = 0.14, controlling for an effect of treatment group).
Table 2. Prevalence of peripheral malaria and maternal anaemia by treatment group
Vitamin A group
Vitamin A vs. placebo
Vitamin A vs. placebo controlling for level of education
GMPD, geometric mean parasite density.
* Logistic regression.
† Linear regression.
‡ Linear regression controlling for baseline levels.
OR = 1.0 P = 0.99*
OR = 0.74 P = 0.53*
P = 0.34†
P = 0.16†
P = 0.62‡
P = 0.36‡
OR = 0.69 P = 0.51*
OR = 0.54 P = 0.37*
P = 0.89†
P = 0.50†
P = 0.81‡
P = 0.90‡
Table 3. Prevalence of placental malaria, mean birth weight and incidence of low birth weight by treatment group
Vitamin A group
Vitamin A vs. placebo
Vitamin A vs. placebo controlling for level of education
* Logistic regression.
† Linear regression.
Placental malarial infection
All grades of infection
Active and active-chronic vs. past infection and not infected
OR = 0.58 (0.23–1.46) P = 0.25*
OR = 0.67 (0.24–1.83) P = 0.43*
Active and active-chronic vs. past infection – excluding women with no evidence of infection
OR = 0.42 (0.16–1.13) P = 0.087*
OR = 0.42 (0.13–1.30) P = 0.133*
Mean birth weight (kg)
P = 0.69†
P = 0.72†
Low birth weight (<2.5 kg)
OR = 2.34 (0.75–7.8) P = 0.14*
OR = 2.8 (0.75–10.5) P = 0.13*
Although the prevalence of malarial infection during pregnancy was high it appeared to be largely asymptomatic. At recruitment 47% of women with peripheral parasitaemia also had fever (temperature >37.4 °C), but at late pregnancy, although the prevalence of peripheral parasitaemia increased, none of the women had fever. In addition, only 8% of women with peripheral parasitaemia at recruitment and 5% of those at late pregnancy reported having felt unwell in the previous 2 days.
In contrast to a lack of effect of vitamin A supplementation on peripheral parasitaemia, we observed that although women receiving vitamin A supplementation had more placental infections overall (92%vs. 81%) they appeared to be much less likely than those in the placebo group to have active or active-chronic placental infection at delivery compared to past, resolved infection (OR = 0.42) although this difference did not reach statistical significance (P = 0.13, Table 3). However, any effect of vitamin A supplementation on the prevalence of active placental infection at delivery did not appear to translate into a lower prevalence of anaemia or low birth weight (Tables 2 and 3).
Vitamin A supplementation and levels of anti-malarial antibody
Levels of plasma IgG Ab to VSA expressed on the surface of five different parasite lines were asessed. As reported previously, in vitro selection of two parasite lines for adhesion to CSA led to substantially higher levels of CSA-adhesion compared to the parental lines and these parasites were preferentially recognized by IgG from multigravid women and not IgG from men (Cox et al. 2005).
Plasma levels of anti-VSACSA IgG varied among primigravid women and also varied among the three CSA-adherent parasite strains (Figure 2). Overall, Ab levels in study women tended to increase from baseline to post-partum and this was most evident for antibodies to FCR3CSA; changes in levels of anti-VSA IgG during pregnancy and lactation, independent of vitamin A supplementation, have been reported in detail elsewhere (Cox et al. 2005). To determine the effect of vitamin A supplementation on VSA Ab levels, levels of anti-VSA IgG were expressed as a percentage of the mean MFI value for pooled plasma from multigravid Ghanaian women.
Anti-VSA IgG values were available for 79 (n = 78 for BusuaCSA) of the 88 women with plasma samples collected in late pregnancy (n = 34 in the placebo group and n = 45 in the vitamin A group) and for 90 (n = 89 for BusuaCSA) of the 95 samples collected post-partum (n = 45 in each group). As each of the three CSA-adherent parasite isolates express antigenically distinct VSA (Cox et al. 2005), the effect of vitamin A supplementation was tested for each parasite line individually.
Table 4 shows the mean levels of anti-VSA IgG for each parasite line, at each of the three data collections, by treatment group together with estimated differences between the two groups. Levels of anti-VSACSA IgG binding to the placenta-derived EJ-24 parasite line were lower in the vitamin A group than in the placebo group in late pregnancy, with this effect being statistically significant when differences in educational level and enrolment time were allowed for (P = 0.002). Similar but less marked reductions in the levels of anti-VSACSA IgG to the other CSA-adherent parasite lines were observed but these were not statistically significant. At post-partum, the lower levels of IgG to the FCR3CSA and BusuaCSA parasite lines among women receiving vitamin A compared to placebo was similar to that observed at late pregnancy, and were not statistically significant. For IgG to EJ-24, the adjusted difference between treatment groups was less than in late pregnancy and was not statistically significant. For the parental, non-CSA adherent lines, there was no consistent or significant change in IgG levels during the study period. In conclusion, there was no evidence of an effect of vitamin A supplementation on Ab recognition of parasite expressed VSA, with the important exception of EJ-24, which is the only parasite, which is known unequivocally to sequester in placental tissue.
Associations between anti-VSACSA Ab levels and peripheral and placental malaria infection
As reported previously, a positive association between levels of anti-VSACSA IgG and the presence of peripheral parasitaemia was observed in early pregnancy but not at late pregnancy or post-partum (Cox et al. 2005). Furthermore, as reported previously, there was a trend for IgG levels to the placental isolate EJ-24 to be higher in women with resolved placental malaria infections than in women with active placental infection, at delivery (Cox et al. 2005). To test for an effect of vitamin A supplementation on the interaction between anti-VSA Ab levels and risk of placental malaria infection, categorization of the women into vitamin A or placebo groups was included in the statistical model; the effect of this was to reduce the apparent protective effect of anti-EJ-24 antibodies [77.3 (95% CI 71.8–82.8) vs. 71.8 (95% CI 66.6–77.0), P = 0.069]. Using the same statistical model, there was no evidence of an interaction between vitamin A supplementation and placental infection on the levels of anti-VSACSA IgG (P = 0.28) and, furthermore, the presence or absence of placental infection had no significant effect on the relationship between vitamin A supplementation and Ab levels.
The purpose of this study was to determine whether vitamin A supplementation enhanced the development of immune responses to parasite antigens that are predicted to mediate parasite sequestration in the placenta; if this were the case, then vitamin A supplementation might be a useful intervention to help reduce the deleterious effects of pregnancy-associated malaria. Although this was a randomised trial, there were some differences between the treatment groups in baseline levels of anti-VSACSA IgG and in two parameters (level of education and time of enrolment) that could conceivably influence the outcome of the trial. We have re-examined the methods and procedures used for allocation of women to treatment or placebo groups and find no reason to believe that these differences were due to a loss of blinding or systematic failure of randomization. To reduce any bias, we have adjusted for these factors in the analyses of the effects of vitamin A supplementation. In addition, blood samples were obtained from fewer women in the placebo group than in the vitamin A supplementation group at late pregnancy. Again, we believe that this occurred simply by chance but, given the small size of the study, the lack of this data may have substantially reduced our chances of detecting significant differences between the groups.
Despite these limitations, somewhat surprisingly, our study suggests that vitamin A supplementation of subclinically vitamin A deficient women may inhibit rather than enhance the primary immune response to VSA. This finding was particularly noticeable for the placental isolate EJ-24, which, by definition, is a parasite known to sequester in the placenta. Similar effects of vitamin A supplementation were seen on levels of antibodies to the two in vitro CSA-selected lines; that these effects were not statistically significant may be due simply to chance in this rather small study. However, the VSAs expressed by the three parasite lines are antigenically distinct (Cox et al. 2005) and it is possible that not all women were equally exposed to parasites expressing the different antigens; this is especially likely for infection with parasites expressing the BusuaCSA antigen as very few women had Ab levels to this parasite that were above the level of the negative control.
Given the evidence that vitamin A supplementation is associated with decreased mortality from infections in populations with subclinical vitamin A deficiency (Beaton et al. 1993) and associated with increased cell-mediated and humoral immune function in animal models (Carman et al. 1992; Molrine et al. 1995; Stephenson et al. 1996; Wiedermann et al. 1996; Stephensen 2001), it seems unlikely that vitamin A supplementation specifically suppressed the immune response to VSACSA. [Although vitamin A supplementation has been reported to decrease seroconversion to live measles vaccination when given at 6 months (instead of the more normal 9 months) (Semba et al. 1995), it is thought that this effect is limited to younger children in whom high levels of maternal Ab are present (Benn et al. 1997).] Rather, we hypothesise that vitamin A supplementation might reduce levels of Ab to VSA of placental parasites by reducing the prevalence of active, unresolved placental malarial infection at delivery and thus reducing the stimulus for production of specific anti-VSACSA. In support of this hypothesis, we did indeed find that women receiving vitamin A supplementation were less likely to have active placental malaria at delivery; again, the lack of significance of this effect may be due to the small size of the study. Vitamin A may protect against placental malaria by enhancing cellular immune function, as suggested by our finding of significant effects of vitamin A supplementation on pro-inflammatory cytokine responses (Cox et al. manuscript submitted for publication). The evidence for this hypothesis might have been strengthened if anti-VSACSA IgG levels had been higher in women with active placental infection whereas, instead, we observed that anti-VSACSA IgG levels were modestly (and non-significantly) lower in women with active placental infection than in women with resolved infections. On the other hand, if anti-VSACSA IgG does in fact protect against placental malaria infection as previously reported (Duffy & Fried 2003; Staalsoe et al. 2004), then levels are likely to be higher in women who have resolved their infection than in women who have not. However, due to the design of the study, anti-VSACSAIgG was measured some weeks before delivery and measurements are thus not directly contemporary with the placental infection data. Further studies are required to separate out cause and effect in the relationship between placental infection and anti-VSACSAIgG levels.
Placental histology indicated that the vast majority of the study women were exposed to malarial infection during pregnancy; weekly vitamin A supplementation appeared to reduce the ratio of active to past infections by more than twofold. Although the evidence for this effect in this small study is quite weak, the size of the effect and its potential public health significance if confirmed indicate that the hypothesis – that vitamin A supplementation protects against active placental malaria late in pregnancy – deserves to be tested in a larger trial.
Given that the effect of vitamin A supplementation on risk of active placental malaria failed to reach significance, it is not surprising that we did not find any significant effect of vitamin A supplementation on birth weight or maternal hemoglobin levels; the study was not designed, and was clearly underpowered, to assess these as outcomes. However, although mean birth weight did not differ between the treatment groups, there was a suggestion that vitamin A supplementation was associated with an increased risk of low birth weight. Very few studies have assessed the effect of vitamin A supplementation on birth weight, but in HIV positive women, vitamin A supplementation was found to have no effect (Fawzi et al. 1998) or to significantly increase birth weight (Kumwenda et al. 2002). Clearly larger studies are required to explore these findings further.
In summary, in this rather small study of the effects of vitamin A supplementation on malaria in pregnancy, we have found that vitamin A supplementation appears to reduce both the risk of active pacental malaria infection at delivery and the levels of antibodies specific for antigens that mediate parasite sequestration in the placenta. The true magnitude of these effects, the link between them, and their clinical significance, are all currently unclear but we believe that our findings justify further, larger, trials of vitamin A supplementation in women at risk of placental malaria infection and the consequent effects of this on maternal and child health.
We thank the computer centre staff at KHRC, especially Boniface Batosona and Seeba Amenga-Etego, Kofi Tchwum the laboratory technician, and the fieldworkers Kwasi Osei Owusu, Betty Baazing, Doris Asare and Prosper Owusu. We also thank Eddy Addo (Nutrition Unit, Noguchi Memorial Institute of Medical Research, University of Ghana) for the assessments of plasma retinol, Maxwell Appawu (Parasitology Unit, NMIMR, University of Ghana) for assessment of the malaria blod films and Kirsten Pihl and Anne Corfitz (CMP, Denmark) for excellent technical assistance with the flow cytometry analysis. We would also like to express our immense appreciation for the co-operation of all of our study participants.
The study was funded by means through by the following institutions and scholarships: MRC UK (G78/6363), University of London Central Research Fund, Nestle Foundation, Commission of the European Communities (QLK2-CT-2001-01302, PAMVAC), Danish Medical Research Council (SSVF 22-02-0571), Denis Burkit Study Award, Jeremy Chadwick Travelling Fellowship, Parkes Foundation.