Newborn Vitamin A Supplementation Does Not Affect Nasopharyngeal Carriage of Streptococcus pneumoniae in Bangladeshi Infants at Age 3 Months

Abstract

Nasopharyngeal (NP) carriage of S. pneumoniae (Spn) is a risk factor for pneumococcal disease and its transmission. We assessed the impact of vitamin A (VA) supplementation shortly after birth in reducing Spn colonization in early infancy in rural Bangladesh. We recruited 500 infants participating in a cluster-randomized trial that reported a 15% reduction in mortality following receipt of an oral dose of VA (52.25 μmol) compared to placebo. NP specimens were collected at the age of 3 mo to study the effect of VA on the prevalence of culture-confirmed Spn. Analyses were conducted by intention to treat. Spn carriage prevalence did not differ between VA and placebo recipients [OR = 0.83 (95% CI: 0.55–1.27); P = 0.390]. Spn carriage at the age of 3 mo was not lowered by VA given at birth. Results are similar to those from an Indian study in which impact on Spn carriage was assessed at the age of 4 mo [OR = 0.73 (95% CI: 0.48–1.10); P = 0.128]. The point estimate of the pooled effect size for the 2 studies is OR = 0.78 [(95% CI: 0.58–1.04); P = 0.095], which may imply a modest impact on carriage. If so, then the evidence thus far would suggest that Spn carriage reduction is unlikely to be a primary ancillary benefit of newborn VA supplementation.

Introduction

Spn⁷ is responsible for ~2 million childhood deaths annually in children <5 y of age, mostly in low-income countries. Approximately 90% of these deaths occur in infants with pneumonia (1). The pathogen also accounts for a high burden of early childhood sepsis, meningitis, and otitis media. In developing countries, undernutrition increases susceptibility to severe disease (2, 3). Efforts to control bacterial infections focus on prompt treatment with first-line antibiotics. However, the rapid spread of antibiotic-resistant strains has threatened to decrease the efficacy and increase the cost of treatment (4). The problem is of greatest significance in low-income countries where many bacterial infections are resistant to first-line antibiotics (5, 6). Furthermore, second-line therapies are often unavailable or unaffordable, making these infections difficult to treat, ultimately facilitating the spread of drug-resistant strains.

The nasopharynx is the primary reservoir for Spn in the human body and is the main source of infection (7, 8). Spn conjugate vaccines are efficacious in preventing invasive disease and in lowering the horizontal spread of infection by reducing NP carriage of vaccine strains (9, 10). However, concerns about the high cost of these vaccines, the limited-serotype protection, and the threat of replacement colonization with potentially pathogenic nonvaccine strains prevent their widespread use in developing countries (7, 11–13). In addition, the first dose of these vaccines cannot be administered before the second month of life, leaving young infants with partial, if any, immunity to Spn. Research on complementary strategies to prevent bacterial infection and block disease transmission in early infancy is needed.

VA plays a critical role as a modulator of the immune response; however, its role in the prevention of bacterial infections is not well understood. Data from developing countries suggest an apparent correlation between endemic VA deficiency and rapid colonization with Spn in early infancy, leading to high prevalence of carriage (5, 14, 15). Results from observational studies show that children with mild xerophthalmia, representing moderate to severe VA deficiency, may have a higher incidence and a greater mortality rate from acute lower respiratory infections (16, 17). Animal and in vitro studies have shown that low serum retinol concentrations are associated with keratinization of the respiratory mucosa, impaired antibody-mediated immunity, and diminished barrier defenses (18–20). These alterations may facilitate acquisition of bacterial colonization in early infancy and in turn may increase the risk of pneumococcal infection (7, 21, 22).

Evidence from community-based trials conducted in southern Asia suggests that newborn VA supplementation in areas of endemic deficiency can reduce early infant mortality by an average of 20% (23–26). The mechanisms underlying this observed effect are not well understood, but prior evidence suggests that prophylactic VA may be important in limiting the severity of infections rather than in decreasing their incidence (27). In addition, data from previous studies suggest that high-dose VA supplementation has no impact on the prevention of clinically diagnosed pneumonia in children over 6 mo of age (28, 29). However, results from a trial in south India showed that VA was associated with a decreased risk of Spn carriage and delayed the age of initial Spn acquisition in younger infants when supplemented at birth (30), which could in part help explain an associated lower risk of mortality in the early months of life. The evidence suggests a potential role for VA supplementation in reducing Spn carriage, which can reduce the risk of infections from specific bacterial pathogens. Further studies are needed to clarify the relationship between VA supplementation shortly after birth and risk of bacterial NP carriage in early infancy and yield clues to potential mechanisms of action. Nested within a large randomized, placebo-controlled trial, we evaluated the efficacy of newborn VA supplementation in preventing NP carriage with Spn among infants at age 3 mo in rural Bangladesh.

Materials and Methods

Study population.

The present study was nested into a concurrent substudy of a large, cluster-randomized, double-blind, placebo-controlled trial, called JiVitA-2, conducted in the districts of Gaibandha and Rangpur in northwestern Bangladesh from January 2004 to January 2007. The trial observed a 15% reduction in mortality in the first 6 mo of life among infants who received an oral dose of VA shortly after birth (26). The JiVitA-2 trial was itself nested, balanced across allocation arms, into the latter half of a placebo-controlled, double-blind, cluster-randomized trial, called JiVitA-1, that assessed the impact of weekly, low-dose VA and β-carotene supplementation during pregnancy and lactation on pregnancy-related mortality (31).

JiVitA-2 trial procedures.

Briefly, infants born to consenting mothers participating in the JiVitA-1 trial became eligible for inclusion in the newborn VA supplementation trial beginning in January 2004. Liveborn infants of consenting mothers were visited by staff members soon after birth and orally given oil drops from identical capsules containing 52.25 μmol VA or placebo. Anthropometric data were collected in the household using standard methods. Infant vital status was assessed weekly through 12 wk of age and again at ~24 wk and 12 mo of age. Information on infant feeding and morbidity patterns for the previous 3 mo was obtained at ages 12 and 24 wk. Additionally, in a contiguous 32-cluster study area (representing 3% of all trial clusters, balanced across supplement groups via randomization), infants were evaluated for clinical signs of illness and had heelstick dry blood samples taken at 12 wk of age primarily to validate differences in VA status between allocation arms. This subcohort of infants formed the sampling frame for the current nasopharyngeal carriage study.

In early December 2006, the Data Safety and Monitoring Board for the trial recommended that the study be closed to new enrollment. Newborn VA supplementation was stopped in January 2007 and the formal analytic sample for the JiVitA-2 trial was defined as comprising infants whose 24-wk vital status was known as of December 31, 2006 (26).

Nasopharyngeal carriage study enrollment.

The carriage study sought to detect a lower prevalence of pneumococcal carriage in infants at 3 mo of age who were dosed with VA compared to placebo shortly after birth. Using data from a previous study in south India (30), we estimated the prevalence of pneumococcal carriage in the placebo group at age 3 mo to be 60%. A sample size of 500 infants was sufficient to detect a minimum absolute difference of 17% (OR = 0.50) between treatment groups to test the null hypothesis that VA is not associated with pneumococcal carriage (α = 0.05, 1-β = 0.80, and design effect = 1.2).

The study was open to all infants participating in the JiVitA-2 trial who were born in the substudy area between January 2005 and December 2006 who were 12 wk of age. This period extended beyond the formal analytic period of the JiVitA-2 trial (26) to allow an opportunity to achieve the intended sample size of ~500 infants. Carriage assessments were conducted between April 2005 and March 2007 to coincide with the time when the last group of infants born in December 2006 were turning 12 wk of age. At the 12-wk postpartum visit, field staff informed parents of eligible children and sought their voluntary consent to participate. During this period, 503 infants aged 12 wk were eligible for enrollment in the carriage trial.

Ethical review.

The NP carriage study and JiVitA trial received ethical approval from the Bangladesh Medical Research Council and the Institutional Review Board of the Johns Hopkins Bloomberg School of Public Health. Infants were included in the carriage study and JiVitA-2 trial based on the provision of documented oral consent from parents or guardians.

Specimen collection.

Trained field workers collected one nasopharyngeal specimen from each infant enrolled in the carriage study at age 12 wk. Nasopharyngeal specimens were obtained following a set protocol. A small, flexible, rayon-tipped swab (Fisherbrand Calcium Alginate Swabs, Fisher Scientific) was inserted into one of the nares to the level of the posterior nasopharynx, a distance approximately midway between the tip of the nose and the ear lobe. The swab was left in place for 5 s before removal. Swabs were then inserted into media containing skim milk, tryptone, glucose, and glycerin media (32) and the specimens were transported to the field laboratory on ice. The specimens were stored at −20°C in a temperature-monitored freezer and transported on ice to the microbiology laboratory at Dhaka Shishu Hospital on a monthly basis for processing.

Laboratory procedures.

For the isolation of Spn, swabs were inoculated onto blood agar (Becton Dickinson) plates containing 5% sheep blood and 5 mg/L gentamicin. Plates were incubated at 37°C in 5% CO₂ for 18–24 h. Colonies exhibiting classic pneumococcal morphology were confirmed by optochin (Taxo, Becton Dickinson) inhibition or bile solubility testing. A pneumococcal reference strain (American Type Culture Collection 5603) was used for quality control. All culture-positive specimens were then subcultured and frozen in skim milk, tryptone, glucose, and glycerin media at −70°C until they were serotyped using Pneumotest kits (Staten Serum Institute). The kits use different combinations of type-specific pneumococcal rabbit antisera to identify 23 pneumococcal serogroups/serotypes of strains most frequently associated with invasive disease. They are serotypes 1, 2, 3, 4, 5, 8, 14, and 20 and serogroups 6, 7, 9, 11, 12, 15, 17, 18, 19, 22, 23, and 33. Serotyped isolates were further classified based on their inclusion in 3 current pneumococcal conjugate formulations: the licensed 7-valent vaccine (7vPnC, Wyeth Vaccines) as well as a 10-valent (10vPnC, GlaxoSmithKline) and a 13-valent (13vPnC, Wyeth Pharmaceuticals) undergoing evaluation for use in developing countries. The 7-valent vaccine includes serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F. The 10-valent vaccine includes all of the previous seven along with serotypes 1, 5, and 7F. The 13-valent vaccine includes all 10 serotypes with the addition of serotypes 3, 6A, and 19A.

The Kirby and Bauer disk diffusion method was used to screen pneumococcal isolates for sensitivity to first-line antibiotics, commonly prescribed in resource-poor settings for the treatment of pneumonia, including penicillin, cotrimoxazole, erythromycin, and chloramphenicol. All susceptibility testing was carried out using Mueller-Hinton media with 5% lysed sheep blood. Isolates were categorized as sensitive, intermediately resistant, or resistant according to the recommendations of the Clinical and Laboratory Standards Institute (33). Minimum inhibitory concentrations were determined using E-tests (AB Biodisk) to confirm penicillin resistance in Spn isolates that screened resistant to oxacillin.

Definition and measurement of outcomes.

The primary outcome of interest was the prevalence of culture-confirmed nasopharyngeal carriage of Spn at age 3 mo. Additional outcomes of interest were the prevalence of serotypes included in the 3 current Spn conjugate formulations and the prevalence of strains resistant to commonly used, first-line antibiotics.

Statistical analyses.

Analyses were conducted on the basis of intention to treat using Stata v.10 (Stata Corporation). Baseline characteristics of the treatment groups were compared using chi square tests for contingency data and t tests for continuous data. Bivariate logistic regressions models were constructed to evaluate the effect of treatment, baseline characteristics, anthropometric indices, infant morbidity pattern, and other variables of interest. For the bivariate and multivariate analyses, generalized estimating equations with the logit link function and a binomial random component were used to evaluate associations between potential risk factors, adjusted for clustering at the sector level.

OR and 95% CI were used to measure the association between treatment and NP colonization adjusted for the effect of the covariates in the model. Differences were considered significant at P ≤ 0.05.

Results

Enrollment and comparability.

Of 503 eligible infants, there were 3 refusals, leaving 500 infants enrolled into the carriage study. The VA (n = 275) and placebo (n = 225) groups were comparable with regard to most demographic, socioeconomic, and household factors (Table 1). The imbalance in sample size between groups reflects survival bias (survival until age 12 wk), the noted treatment effect of newborn VA supplementation on mortality during the 6 mo of life and chance. Among compared factors, infants in the VA group were more likely to come from households with electricity and to have had their mothers randomized to the placebo arm of the JiVitA-1 trial. Their mothers, however, were less likely than those of newborn VA recipients to have ever reported being night blind during pregnancy.

TABLE 1.

Characteristics of infants enrolled in the CHIPS study by treatment group, Gaibandha, Bangladesh

Characteristic	Placebo group (n = 225)	VA group2(n = 275)
	n (%)
Male	103 (45.8)	139 (50.6)
Muslim	208 (92.4)	257 (93.8)
Infant breastfed	220 (99.1)	274 (100.00)
Birth weight <2500 g	122 (55.0)	131 (48.7)
2 or more children <5 y in household	58 (25.8)	70 (25.6)
Mother with <1 y of formal education	70 (31.1)	82 (29.9)
Mother is illiterate	104 (46.2)	126 (46.0)
Mother has paid job*	96 (42.7)	144 (52.6)
Maternal night blindness during pregnancy*	16 (7.2)	40 (14.6)
Maternal supplementation arm*
β-Carotene	73 (32.4)	110 (40.0)
VA	67 (29.8)	94 (34.2)
Placebo	85 (37.8)	71 (25.8)
Electricity in household*	30 (13.3)	63 (23.0)
No latrine in household	78 (34.7)	83 (30.3)
Household with ≥3 living rooms	28 (12.4)	51 (18.6)
Family owns ≥1 bicycle	75 (33.3)	103 (37.6)
Family owns ≥1 cow	97 (43.1)	129 (47.1)

¹Characteristics were compared using Pearson's chi square test. *Different from placebo, < 0.05.

²VA, vitamin A.

Effect of VA on pneumococcal NP colonization.

The overall prevalence of pneumococcal carriage in the cohort at age 3 mo was 70.8%. The prevalence of Spn carriage in the VA and placebo groups was 69.1 and 72.9%, respectively. The crude OR of pneumococcal carriage after adjusting for the effect of clustering was 190, indicating a lack of association with newborn VA supplementation. Multivariate models were then constructed to control for the effects of baseline characteristics and other covariates; however, the point estimates from the multivariate analysis remained consistent with crude OR.

Effect of VA on colonization with invasive serotypes.

The prevalence of serogroups/serotypes included in the 3 current pneumococcal conjugate vaccine formulations, 7vPnc, 10vPnc, and 13vPnc, was 48.0, 49.4, and 50%, respectively. The carriage prevalence for 7vPnC (48.2 vs. 47.4%), 10vPnc (49.4 vs. 49.0%), and 13vPnc (50.0 vs. 49.5%) strains was comparable across treatment groups, assuming 100% cross-protection between serotypes within serogroups.

Effect of VA on carriage with antibiotic-resistant pneumococci.

Spn isolates were evaluated for susceptibility to cotrimoxazole, penicillin, chloramphenicol, and erythromycin. Among the 4 antibiotics evaluated, ~82.2% of isolates of the Spn strains were resistant to cotrimoxazole. In contrast, the prevalence of isolates that were resistant to chloramphenicol and erythromycin was 4.0 and 1.9%, respectively. All of the isolates were susceptible to penicillin. The carriage prevalence of resistant isolates in the VA and placebo groups was similar for cotrimoxazole (82.9 vs. 81.1%), erythromycin (1.2 vs. 2.1%), chloramphenicol (2.4 vs. 4.7%), and penicillin (0 vs. 0%).

Discussion

In this population-based, randomized trial cohort, we found a high prevalence of Spn carriage among infants at 3 mo of age. The prevalence of pneumococcal carriage in infants receiving VA at birth did not significantly differ from that of children in the placebo group. However, the point effect estimate is consistent with the results of an earlier study conducted in a population trial setting in south India, where, as in this Bangladesh population (26), newborn VA supplementation had significantly reduced infant mortality (30). In the Indian study, newborn VA supplementation significantly reduced the odds of colonization by 49% among infants aged 4 mo not colonized at age 2 mo. It was also associated with a 27% reduction in odds of colonization compared to the placebo group among children aged 4 mo. However, the finding was not significant. When the data from the current study are combined with that those from the Indian trial for children aged 4 mo, 476 the point estimate of the pooled effect size indicates that VA supplementation is associated with a 22% reduction in the odds of carriage [OR = 0.78 (95% CI: 0.58–1.04); P = 0.095]. These data may suggest a weak or modest impact on Spn carriage.

It should be noted that the preliminary results show no significant differences in mean serum retinol concentrations between treatment groups at age 3 mo (K. Schulze, Johns Hopkins School of Public Health, personal communication), which may explain why the association was stronger in the Indian trial. It further raises the possibility that if there is any benefit associated with VA with respect to Spn carriage, it is likely to be greatest in populations with more severe VA deficiency.

VA plays an important role in maintaining immunocompetence and the integrity of the epithelial tissue (34). Damage to epithelial tissue resulting from VA deficiency may reduce mucociliary clearance of bacteria (35, 36). In addition, low serum retinol concentrations have been correlated with impaired mucosal immunity in previous studies (18, 20). It is plausible that impairment of these defenses may result in increased bacterial adherence (37) and colonization of the nasopharynx (30). Supplementation may restore the integrity of the respiratory epithelium and help to reduce bacterial carriage. Decreasing infants’ susceptibility to carriage can reduce their risk of pneumococcal disease and reduce the circulation and transmission of pathogenic pneumococcal strains (7, 21, 22). Thus, it is plausible that neonatal VA supplementation may help to bolster the defenses of infants in settings of high infectivity and high risk of colonization, especially against a background of underlying VA deficiency.

The distribution of serotypes included in the 3 pneumococcal conjugate vaccine formulations being considered for use in developing countries was comparable across treatment groups. The prevalence of cotrimoxazole resistant Spn isolates was high and comparable to what has been reported in other areas in south Asia (5, 38). The pattern of antibiotic resistance for pneumococcal isolates was similar between the VA and placebo groups. The data suggest that the treatment effect is independent of the susceptibility of the isolates to the 4 antibiotics tested. This suggests that VA appears to neither decrease nor facilitate colonization with antibiotic-resistant pneumococcal isolates.

A potential limitation of the study was that the carriage data were only measured at age 3 mo. Bacterial carriage is a dynamic process and it is possible that, had we assessed carriage status at monthly intervals beginning at age 1 mo, we may have obtained different estimates of treatment effect. However, the results from the south Indian carriage study also suggest that the greatest reduction in carriage prevalence was in the interval between 2 and 4 mo postsupplementation.

In summary, results from this study show that newborn VA supplementation had no significant impact on the prevalence of pneumococcal carriage at age 3 mo. Pooling the results from this study with those from a previous trial conducted in India may suggest a weak effect of newborn VA supplementation on Spn carriage. At best, any effect is likely to be modest. The bigger question is what does a small impact on Spn carriage mean clinically and in terms of public health. The current evidence indicates that Spn carriage reduction is unlikely to be a major ancillary benefit of newborn VA supplementation.