点击显示 收起
1 From the Bandim Health Project, INDEPTH Network, Bissau, Guinea-Bissau (ABF, AR, and PA); the Bandim Health Project, Statens Serum Institut, Copenhagen, Denmark (PA and CSB); the Department of Infectious Diseases, Aarhus University Hospital, Aarhus, Denmark (ABF); the Department of Pathology, Herlev University Hospital, Copenhagen, Denmark (IML); the Department of Biostatistics, University of Aarhus, Aarhus, Denmark (BMB); and SEAMEO, Regional Center for Community Nutrition, University of Indonesia, Jakarta, Indonesia (JGE)
2 Supported by the European Commission, INCO-program (contract no. ICA4-CT-2002-10053), the March of Dimes, and The Danish Medical Research Council (contract no. 22-04-0118). The Bandim Health Project received support from DANIDA and the Danish National Research Foundation. PA holds a research professorship grant from the Novo Nordisk Foundation. 3 Reprints not available. Address correspondence to AB Fisker, Bandim Health Project, INDEPTH Network, Bissau 2300, Guinea-Bissau. E-mail: a.fisker{at}bandim.org.
ABSTRACT
Background: The effect of vitamin A supplementation (VAS) at birth on subsequent vitamin A status has not been studied.
Objective: The objective was to study the effect of 50 000 IU vitamin A administered with BCG vaccine at birth on vitamin A status in both sexes.
Design: Within a randomized placebo-controlled trial of VAS, we obtained blood from 614 children at 6 wk of age and from 369 mother-infant pairs at 4 mo of age. We assessed vitamin A status on the basis of serum retinol-binding protein (RBP) and measured serum C-reactive protein to monitor for concurrent infections.
Results: RBP concentrations indicated vitamin A deficiency in 32% of the children at age 6 wk and in 16% at age 4 mo. VAS was not associated with higher RBP concentrations overall or in either sex. However, the effect of VAS varied with maternal education (P for interaction = 0.004): At age 6 wk, VAS was associated with higher (9%; 95% CI: 2, 17%) RBP concentrations in children of noneducated mothers but not in children of educated mothers. Overall, RBP concentrations increased between 6 wk and 4 mo of age. The increase correlated inversely with the number of diphtheria-tetanus-pertussis (DTP) vaccines received in the interval (P = 0.009), particularly in girls (P for interaction = 0.01) and in vitamin A recipients (P = 0.01).
Conclusions: Overall, VAS at birth had no effect on vitamin A status. However VAS may temporarily improve vitamin A status in the subgroup of children of noneducated mothers. In vitamin A recipients, subsequent DTP vaccines affected vitamin A status negatively. The main trial was registered at clinicaltrials.gov as NCT00168597.
Key Words: Vitamin A supplementation • BCG vaccine • retinol-binding-protein • acute phase reaction • sex differences • diphtheria-tetanus-pertussis vaccine
INTRODUCTION
Vitamin A supplementation (VAS) at birth may reduce child mortality, particularly in boys (1, 2; CS Benn, BR Diness, A Roth et al, unpublished observations). No previous trials have studied the effect of VAS at birth on subsequent vitamin A status.
Within a randomized placebo-controlled study of the effect on mortality of 50 000 IU vitamin A administered with BCG at birth (CS Benn, BR Diness, A Roth et al, unpublished observations), we aimed to examine the effect on vitamin A status at 6 wk and 4 mo of age in both sexes. Vitamin A status was assessed on the basis of serum retinol-binding protein (RBP) concentrations.
The present trial (CS Benn, BR Diness, A Roth et al, unpublished observations) as well as a similar trial of VAS at birth from Zimbabwe (3) did not find the expected beneficial effect on mortality that was observed in Indonesia (2) and India (4). We previously hypothesized that VAS and diphtheria-tetanus-pertussis (DTP) vaccines may interact negatively (5). In the present study, therefore, we examined whether VAS administered with BCG at birth interacted with subsequent DTP vaccines to affect the vitamin A status of the child.
SUBJECTS AND METHODS
Study population
The Bandim Health Project (BHP) has a demographic surveillance system in 6 suburban districts of the capital of Guinea-Bissau, Bissau. The present study was carried out within a randomized, double-blind, placebo-controlled trial of the effect on mortality of VAS plus BCG at birth (CS Benn, BR Diness, A Roth et al, unpublished observations).
In brief, women living in the study area were invited to participate if they had a child that was due to be BCG-vaccinated, either before discharge from the maternity ward of the national hospital or when coming for the first time to 2 of the 3 health centers in the study area. Inclusion criteria were weight 2500 g and no overt illness or malformation. After deciding to participate in the study, the mothers drew a lot determining whether their children should receive VAS (50 000 IU retinyl palmitate in vegetable oil with 20 IU vitamin E as antioxidant) or placebo (the same amount of vegetable oil and vitamin E) plus BCG. The BCG vaccine (0.05 mL; Statens Serum Institut, Copenhagen, Denmark) was given as an intradermal injection in the left deltoid area by a trained study nurse. Vitamin A and placebo were prepared by a Danish pharmacy, which held the code until all children had been followed for 1 y. At enrollment, information on baseline anthropometric measures and socioeconomic background factors was collected.
Guinea-Bissau was classified as an area with subclinical vitamin A deficiency (VAD) by UNICEF (6), but no surveys of vitamin A status have been conducted (7). The prevalence of HIV-1 was 3–5% among women of fertile age in the study area in the period of the study (unpublished sentinel surveillance data). A large proportion of pregnant women in this area are screened for HIV to prevent vertical transmission. Uncontrolled vertical transmission would be 25% (8). Because of the ongoing vertical-transmission control program, we expect <1% of the children to be infected with HIV-1. The protocol was presented to the Ministry of Health in Guinea-Bissau and the Danish Central Ethical Committee for approval.
Blood sampling
We collected blood samples in a subgroup of children enrolled in the study between 2004 April 13 and May 19 (dry season, defined as December-May) and between 2004 August 1 and November 28 (rainy season, defined as June-November). Children were visited and blood samples were collected at 6 wk of age if found at home. Of the children enrolled in the rainy season, those included within the first 10 d of life were visited again at 4 mo of age, and blood samples were collected from the infants and the mothers. The restriction to within 10 d of birth was chosen to reduce variability in time-since-supplementation because of different ages at enrollment. Blood was collected from the mother only if the child was breastfed. We aimed to include 400 children. At the visits, we interviewed the mothers about their infant's feeding, recent medications, and vaccinations. Blood samples were obtained by finger prick.
Handling and processing of blood samples
Blood was sampled into a dry tube and placed in a plastic box protected from sunlight at ambient temperature (28–32 °C). The blood was separated within 4 h, and the serum was frozen at –20 °C until analyzed for RBP and C-reactive protein (CRP) concentrations. RBP and CRP were measured with a sandwich enzyme-linked immunosorbent assay technique at the SEAMEO, Regional Center for Community Nutrition, University of Indonesia, Jakarta, Indonesia (9). In brief, wells were coated with RBP antibodies (DAKO) and CRP antibodies (DAKO), diluted serum was added to the wells, H2O2-coupled RBP and CRP antibodies were added after incubation and washing. After another round of incubation and washing, the color substrate was added. After development of sufficient color, the reaction was stopped by adding phosphoric acid. The concentration was assessed by spectrophotometry (9). All concentrations were measured twice, and the average determined. If the 2 measurements differed by >20%, a second set of double measurements was carried out.
The blood concentration of retinol is under homeostatic control over a large range of vitamin A storage concentrations (10). However, the circulating concentration of retinol decreases with marginal vitamin A reserves. Serum retinol concentrations of <0.70 and 1.05 µmol have been used to define subclinical VAD and marginal vitamin A stores, respectively. Retinol is transported in the bloodstream bound to RBP in a 1:1 complex, but the saturation of RBP tends to be higher with high retinol concentrations (11, 12). Hence, higher cutoffs for RBP than for retinol may be used when assessing the vitamin A status. To correct for <100% saturation of RBP with retinol, we measured retinol by HPLC in 26 of the samples. With retinol determined by HPLC as the standard, the corresponding values of RBP were plotted and the linear relation was as follows: RBP = 0.7901 x retinol by HLPC + 0.2808 µmol/L (R2 = 0.79). We therefore used the cutoffs 0.83 and 1.11 µmol RBP/L to define the groups with subclinical VAD and marginal vitamin A stores, respectively. CRP was measured to monitor for ongoing infection at the time of the blood sampling (13).
Statistical analysis
We used multiple linear regression models to evaluate the effect of VAS on RBP concentrations at 6 wk and 4 mo of age. Log transformation of RBP concentrations yielded a better normal distribution assessed by inspection of inverse normal plots; results are therefore given as geometric means and relative changes in RBP. Logistic regression was used to evaluate the effect of VAS on the proportion of subjects with low RBP (equivalent to 0.70 µmol retinol/L).
All analyses were stratified by sex, and interactions between sex and VAS were tested in the models. We attempted to control for ongoing infection using 2 different methods: 1) the widely used approach of excluding samples with a CRP concentration >5 mg/L and 2) including log-transformed CRP concentrations in the model. Hence, we were able to examine the effects of VAS on RBP concentrations and proportions with low RBP concentrations controlled for CRP without excluding any sample from the analysis. Samples with a CRP concentration below the detection level (0.1 mg/L) (10 samples) were assigned half this value.
We tested all background variables (Table 1) and reports of recent medication for association with vitamin A status and the effect of VAS on vitamin A status by including them in univariate regression models and in multivariable regression models including the interaction terms. We controlled for maternal vitamin A status for breastfed children sampled at 4 mo by including the logarithm of maternal RBP concentration in the model. Finally, we examined the association between change in RBP and number of DTP vaccines according to treatment group and sex. All linear models were evaluated visually to ensure that the residuals were normally distributed and that they showed no variation with the covariates.
View this table:
TABLE 1. Baseline characteristics of the participating children1
RESULTS
At 6 wk of age, 960 children were eligible for sampling. We visited the households of 753 (78%) of the children and sampled 614 (82%) of the children (vitamin A: 79%; placebo: 84%). Proportions of children visited, not visited, and sampled did not differ between the vitamin A and placebo groups (Figure 1). There were no significant differences between the vitamin A and placebo groups with regard to baseline characteristics (Table 1). Also, there were no differences between visited and nonvisited children with regard to baseline characteristics (data not shown).
FIGURE 1.. Flow chart of children in the study.
Of the 506 infants included in the study after 1 August 2004 and sampled at 6 wk of age, 441 (87%) were eligible for sampling at 4 mo of age. The remaining children had been included in the study after 10 d of age. We visited 440 of the children at least once within 10 d of 4 mo of age and obtained sufficient serum from 369 (84%).
RBP concentrations 6 wk after supplementation
We found no overall effect of VAS on vitamin A concentrations 6 wk after supplementation (Table 2). We found no sex difference in the RBP concentrations or the proportion of low RBP, nor did we find any overall difference in the effect of supplementation between boys and girls. As expected, an acute phase reaction was associated with a lower RBP: a 2-fold increase in CRP was associated with an odds ratio (OR) of 1.26 (95% CI: 1.17,1.37) for low RBP (P < 0.001). Adjustment for CRP did not change the effect of VAS (Table 2). Exclusion of samples with CRP concentrations >5 mg/L excluded more boys than girls from the analysis (135 samples; 22%): 76 from boys (26%) and 59 from girls (19%) (P = 0.04, chi-square test) but gave similar results (Table 2). Limiting the analysis to the samples from children included in the study within the first 10 d of life (87%) did not affect the results.
View this table:
TABLE 2. Effect of vitamin A status at birth on retinol-binding protein (RBP) concentrations at 6 wk of age1
We examined the effect of different background variables thought to be associated with vitamin A status. In a regression analysis, children enrolled in the dry season had 7% (11; CS Benn, BR Diness, A Roth et al, unpublished observations) higher RBP concentrations than did children enrolled in the rainy season (P = 0.005). Anthropometric measures or maternal midupper arm circumference was not associated with RBP at 6 wk of age, nor were any of the other background factors in Table 1. The proportion of children who had received medication (mainly chloroquine, penicillin, paracetamol, or a combination thereof) on the day before sampling (vitamin A: 14%; placebo: 16%) did not differ between groups (P = 0.20, chi-square test). These children had a higher CRP concentration (P < 0.001), but medication use did not affect RBP concentrations (P = 0.68) or the effect of VAS on vitamin A status (P = 0.91).
Maternal education interacted significantly with treatment group (P = 0.004); VAS was associated with an increased concentration of RBP in children of noneducated mothers, whereas there was no effect of supplementation in children of educated mothers. Similarly, the proportion of children with a low RBP concentration was reduced in supplemented children of noneducated mothers, but not in supplemented children of educated mothers (Table 3). There were no other interactions between background factors and VAS (data not shown).
View this table:
TABLE 3. Effect of vitamin A supplementation (VAS) at birth on retinol-binding protein (RBP) concentrations at 6 wk of age by maternal education1
Twenty-two percent of the children sampled at 6 wk had a DTP vaccine before sampling: vitamin A: 25% (n = 72); placebo: 19% (n = 60) (P = 0.08, chi-square test). DTP-vaccinated children were more likely (42%) to have elevated CRP concentrations than were children not vaccinated with DTP (16%) (P < 0.0001). VAS did not affect RBP concentrations among DTP-vaccinated (relative difference: –1%; 95% CI: –9%, 8%) or DTP-unvaccinated (relative difference: 1%; 95% CI: –3%, 6%) children.
RBP concentrations 4 mo after supplementation
At 4 mo of age, there was no overall difference between the vitamin A and placebo groups in RBP concentrations, nor did the effect of VAS differ between boys and girls. However, adjusted for CRP, boys had 4% (95% CI: 0%, 9%) higher RBP concentrations than girls. The proportion of children with low RBP did not differ between the VAS and placebo groups (Table 4). As observed at 6 wk of age, exclusion of samples with elevated CRP concentrations excluded more boys than girls from the analysis: 105 samples (28%); 58 from boys (34%) and 47 from girls (24%) (P = 0.03, chi-square test).
View this table:
TABLE 4. Effect of vitamin A supplementation at birth on retinol-binding protein (RBP) concentrations at 4 mo of age1
Maternal RBP concentrations were available for 357 of the 360 breastfed children (355 mothers). Three samples had insufficient serum for analysis. The mean maternal RBP concentration was 1.66 µmol/L (95% CI: 1.62, 1.70). Only one of the mothers had an RBP concentration indicating VAD; 18 (5%) had an RBP concentration indicating marginal vitamin A stores. Educated mothers had 6% (95% CI: 1, 11) higher RBP concentrations than did noneducated mothers. The RBP concentration in children varied with maternal RBP concentration: an increase in maternal RBP by 30% was associated with a 3% (95% CI: 1, 6) increase in RBP of the child (P = 0.007).
Change in RBP between 6 wk and 4 mo
RBP concentrations were available for 367 children at 6 wk and 4 mo of age; during this interval, 70% had experienced a rise in the RBP concentration (mean rise in RBP = 0.13 µmol/L; 95% CI: 0.10, 0.16). The proportion of children who had an increase in RBP and the magnitude of the rise did not differ between the vitamin A and the placebo groups, nor did it differ between boys and girls. However, the magnitude of the rise in RBP depended in a linear manner on the number of DTP vaccines that the child had received between the 2 blood samples (Table 5).
View this table:
TABLE 5. Effect of vitamin A supplementation (VAS) at birth, sex, and number of diptheria-tetanus-pertussis (DTP) vaccines on change in retinol-binding protein (RBP) concentrations between 6 wk and 4 mo of age1
Between 6 wk and 4 mo of age, 51% of the sampled children had received 3 doses of DTP vaccines, 42% received 2 doses, 7% received 1 dose, and 1% received no doses. The proportions did not differ between the VAS and placebo groups (P = 0.73) but did differ with maternal education; children of educated mothers had received more doses (P < 0.001). Overall, the rise in RBP was reduced by 0.06 µmol/L (95% CI: 0.01, 0.10) for each dose of DTP (P = 0.009). This effect was only seen in girls (0.12 µmol/L; 95% CI: 0.05, 0.19) per dose of DTP and not in boys (0.00 µmol/L; 95% CI: –0.06, 0.06) (P for interaction = 0.01). Also, controlled for the effect of sex, VAS recipients had a 0.12 µmol/L (95% CI: 0.03, 0.21) larger decrease per dose of DTP than did the placebo recipients (P for interaction = 0.01) (Table 5). Exclusion of the 47 children who had received a DTP vaccine before the 6-wk sample or exclusion of the 169 children who had received a DTP vaccine within 2 wk before the 4-mo sample did not change the estimates. Control for time since the last DTP vaccination, time between the samples, or maternal RBP concentration did not change the estimates either. Control for CRP concentrations or RBP concentration at 6 wk of age reduced the magnitude of the fall, but the conclusions were unaffected. Grouping children with 0–2 DTP vaccines in a single group did not change the interactions between sex and DTP and between VAS and DTP (P = 0.001 and P = 0.05, respectively).
DISCUSSION
Six weeks after VAS, the effect on vitamin A status was limited to the children of noneducated mothers. There was no measurable effect of VAS on RBP concentrations at 4 mo of age. When planning the present study of vitamin A status, we hypothesized that the effect might differ between boys and girls and with subsequent vaccines. There was no sex-differential effect of VAS on vitamin A status. However, the effect did seem to depend on the number of DTP vaccines received after VAS in a manner that differed between boys and girls.
The reported duration of the effect of single high-dose VAS varies. In observational studies, supplementation with 200 000 IU in children >12 mo of age has been reported to affect vitamin A status at 1–2 mo but the effect is not sustained to 4–6 mo after supplementation (14, 15). Others have reported an effect of 200 000 IU but not of 100 000 IU (given in a randomized trial) for up to 6 mo (16).
It has been customary to adjust measurements of RBP and retinol concentrations for ongoing infections by excluding children with CRP concentrations for >5 mg/L. We attempted to correct for a lowered RBP due to a subclinical infection by including CRP concentrations in the regression models, thereby not excluding any observations. We believe that this method has advantages over the method of excluding children with CRP concentrations >5 mg/L. First, as pointed out by others (17), inflammation may affect circulating vitamin A concentrations even when CRP concentrations are <5 mg/L, at least in young children. Second, in populations with a high prevalence of infections, exclusion of samples with CRP concentrations >5 mg/L may lead to the exclusion of a large group of infected children who also have low vitamin A status; thus, the actual prevalence of VAD would be underestimated (18, 19). Third, comparisons of vitamin A status between groups may be underestimated or confounded if differences in vitamin A status affect the prevalence of infections.
We noted that the exclusion of samples with CRP concentrations >5 mg/L resulted in the exclusion of more boys than girls at 6 wk and 4 mo of age. Control for CRP was also associated with a larger rise in RBP concentrations in boys than in girls. Hence, the association between sex and RBP concentration may be confounded by CRP. To our knowledge, a differential relation between CRP and RBP in boys and girls has not been described previously. It could be speculated that differential associations between CRP and RBP in boys and girls may be due to differences in the severity or duration of the phases of infections. If the effect of elevated CRP on RBP concentration is indeed different for boys and girls, correction for elevated CRP concentrations may either give rise to or mask sex differences in RBP concentrations and thus further complicate the assessment of vitamin A status by circulating RBP or retinol concentrations.
Applying the often used cutoff of 0.70 µmol/L retinol for VAD, we found 32% and 16% of the children to be vitamin A deficient at 6 wk and 4 mo of age, respectively. It should be noted, however, that this cutoff is based on measurements of children older than 6 mo of age (20). Overall, the mothers in this population seemed less deficient than expected: the vitamin A status was equivalent to that of American women (21) despite the fact that WHO-recommended postpartum supplementation (22) has not been implemented in Guinea-Bissau. Although the observed prevalences below this cutoff seem high, they are still less than the 60% seen for 6-wk-old Ghanaian children (23) and the 83% in Nepali infants at 3 mo of age (24). Hence, Guinea-Bissau does probably not have a VAD problem of the same magnitude and we cannot exclude that VAS would have affected vitamin A status more in more deficient populations. It should be noted, though, that we only included children with a weight >2500 g. Low-birth-weight children are more likely to be vitamin A deficient (25). Thus, our values will tend to underestimate the prevalence of VAD in our population.
It is not unexpected that the effect of supplementation was most marked among noneducated mothers because they have a lower RBP concentration and therefore have a lower vitamin A concentration in their milk (26). Furthermore, these children may have been born with a lower vitamin A status (27). This finding corroborates previous studies in which vitamin A–deficient children were more likely to have an improved vitamin A status after VAS (14-16).
In the main study, we found no overall effect of VAS on mortality (CS Benn, BR Diness, A Roth et al, unpublished observations). This is in contrast with the findings of 2 previous studies from Asia (2, 4) but is similar to the findings of a recent study conducted in HIV-negative women from Zimbabwe (3). It could be speculated that this lack of effect on mortality in our study and in the Zimbabwean study was due to a low prevalence of VAD. However, there are several indications that this may not be the case. In the Indonesian study, vitamin A status among mothers was comparable with that of American women (2). In our main study, boys and children enrolled and supplemented in the dry season benefited from VAS (CS Benn, BR Diness, A Roth et al, unpublished observations). Because we found that boys had a better vitamin A status than did girls at 4 mo and vitamin A status was better among children enrolled in the dry season, the data does not support that a beneficial mortality effect of VAS depends on a high level of VAD. This notion is also supported by the observation of Beaton et al (28) that there was no relation between the baseline prevalence of xerophthalmia and the relative effect of vitamin A. Hence, the effect of VAS on mortality does not seem to be due exclusively to correction of VAD, but may also depend on immunomodulatory effects.
A child following the recommended vaccination schedule would have received all 3 doses of DTP before the 4-mo sample, whereas children not adhering to the schedule might have received fewer doses. We found that children of educated mothers had more doses. Because educated mothers had higher RBP concentrations, it is surprising that the rise in RBP concentration depended inversely on the number of DTP vaccinations. The association cannot be explained by an acute phase reaction leading to a lowered RBP because of a recent vaccine, because the exclusion of samples with recent DTP vaccines or correction for CRP made no difference. It may be important that DTP vaccines affected RBP concentrations negatively in girls and in VAS recipients, which resulted in a significant decrease in vitamin A status in girls who received VAS. We found no previous reports of the effect of DTP vaccination on vitamin A concentrations, but a small study of vitamin A concentrations before and after measles vaccine also found a larger decrease in girls than in boys (29).
If VAS and DTP vaccines interact to reduce RBP concentrations, particularly in girls, it may explain why we found higher mortality rates in girls than in boys after VAS at birth when the children reached the age of DTP vaccination (1; CS Benn, BR Diness, A Roth et al, unpublished observations). In the study of VAS at birth from India (4), the authors examined mortality after subsequent vaccinations (30). It was not possible to separate children who received BCG first and then DTP as in our study. However, for girls who received no BCG but received the DTP vaccine after enrollment, mortality was reduced to a larger extent in the placebo group than in the VAS group, whereas the opposite was the case for boys.
In conclusion, at 6 wk of age the effect of VAS on circulating RBP concentrations was limited to children of noneducated mothers. At 4 mo of age there was no overall difference between the vitamin A and placebo groups. DTP vaccinations appeared to affect the rise in RBP concentrations negatively, particularly in girls who received VAS.
ACKNOWLEDGMENTS
The authors' responsibilities were as follows—CSB, PA, IML, and AR: designed the study and obtained funding; CSB and PA: initiated enrollment in the randomized trial; IML and ABF: collected blood samples and supervised the fieldwork and data entry; BMB: provided statistical support; and ABF: wrote the first draft of the manuscript. All authors contributed to the final version. None of the authors declared a conflict of interest.
REFERENCES