Our principal finding is that placental malaria predicted a relatively weak IFNγ response to PPD a year after birth, which was also a year after the administration of the BCG vaccine. No difference was detectable at 6 months, probably because of the low response in both groups.
By contrast, we did not find an association between PM and responses to CMV. There are distinct differences in the presentation of BCG and CMV, as the BCG vaccine is administered within 24 h of birth and is present for a limited period of time. Gambian infants are usually infected with CMV several weeks after birth [28
], after which it establishes a persistent infection so that the host is exposed to CMV antigens for the rest of their life [29
]. The contrast between the associations with BCG and CMV suggests two mutually compatible explanations. The first is that the influence of PM on the response to BCG is stronger because the exposure to BCG occurred chronologically closer to the exposure to the exposure to placental malaria. The other is that the persistently high antigen exposure in CMV-infected infants drove the immune response to overcome any deleterious influence of PM.
The lack of effect on responses to CMV contrasts with earlier findings that malaria episodes are associated with low responses to another persistent herpesvirus, Epstein-Barr virus (EBV), and also with higher EBV viral loads [30
], implying poor immune control of viral replication. However, our study differs substantially from the studies that generated these findings as apart from the fact that the natural histories of CMV and EBV infection are very different, the children in our study were younger than most of those in the previous studies. Further, the previous studies were on children who either had a recent malaria episode or were frequently infected, while the infants in our cohort were exposed to PM and very few developed malaria in the subsequent year, implying little or no exposure to P. falciparum
We have previously reported that in BCG-vaccinated infants below 12 months, few CD4 T-cells produce IFNγ in response to PPD [33
], so it is likely that differences associated with PM were simply undetectable until the CD4 T-cells of the infants were producing enough IFNγ for a difference to be apparent.
The numbers were too small to analyse for the effects of acute, chronic and past PM. However, the relative proportions of the three types were similar in the infants sampled at six and 12 months, making it unlikely that the appearance of a difference at 12 months was due to a difference in the type of PM among the PM+ infants.
The sizes and weights of infants at birth in the PM+ and PM- groups were comparable, so we could exclude the confounding effects of premature birth and low birth weight which are well described for PM [1
]. The exclusion criterion of low birthweight also ruled out the effect of prenatal undernutrition, which is associated with poor immune development [34
] and while we have previously found an association between PM and congenital cytomegalovirus infection [36
], we found no evidence that congenital cytomegalovirus modulated the responses of CD4 T-cells to PPD [24
] in a much larger and more powerful study.
Testing for HIV was not carried out, but a survey carried out in the nearby town of Serrekunda from 2000 to 2001 found a prevalence of 1.0% among pregnant women [37
], and a recent screen of women in Sukuta confirmed the rates to be < 1% (unpublished), thus HIV was unlikely to be a serious confounder.
Helminth infection has been shown to modulate immune responses to mycobacterial antigens in BCG-vaccinated children [38
], but we found that intestinal helminth infections are not endemic in infants in the Sukuta area after a screen of fecal samples collected at 12 months of age (unpublished). However, the possibility has been raised that helminth infection during pregnancy may reduce the IFNγ response to PPD [41
], which we cannot eliminate as we did not screen the mothers for intestinal infections.
By 12 months of age, the PM+ infants were slightly smaller than the PM- infants. As there were no differences at birth, the difference at 12 months implies that PM was associated with reduced growth, which concurs with our findings from the entire cohort from which infants were selected for the analysis presented here [25
]. Similarly, the differences in IFNγ response emerged between 6 and 12 months, suggesting that PM was predictive of deleterious effects that remained present 12 months after birth.
Although no correlate of protection to tuberculosis has been identified, the ability to generate IFNγ-producing T-cells has been shown to be important, as defects in IFNγ signalling lead to a high susceptibility to tuberculosis disease [19
]. While it is possible that the reduced number of IFNγ-producing cells indicated a shift in response profile, rather than an overall loss of T-cell response, it is still likely that a reduced number of IFNγ-producing CD4 T-cells represents a reduced level of protection.
While we detected an association between PM and reduced immune response to BCG vaccination at birth, the scope of the study did not extend to establishing whether there was a direct causal link. Although PM makes infants more susceptible to subclinical P. falciparum
infection and episodes of malaria [5
], it is unlikely that the lower responses of PM+ infants were due to the immunosuppressive effect of repeated malaria episodes [43
] as we did not find a higher rate of malarial disease in PM + infants. While we could not exclude all possible confounders, no effect that we were able to measure showed any association with reduced responses other than exposure to PM.