Inflammatory markers predict episodes of wheezing during the first year of life in Bangladesh

Abstract

Background

Environmental factors that influence wheezing in early childhood in the developing world are not well understood and may be useful in predicting respiratory outcomes. Therefore, our objective was to determine the factors that can predict wheezing.

Methods

Children from Dhaka, Bangladesh were recruited at birth and episodes of wheezing were measured alongside nutritional, immunological and socioeconomic factors over a one-year period. Poisson Regression with variable selection was utilized to determine what factors were associated with wheezing.

Results

Elevated serum IL-10 (rate ratio (RR) = 1.51, 95% confidence interval (CI): 1.22–1.87), IL-1β (RR=1.55, 95% CI: 1.26–1.93) C-reactive protein (CRP) (RR=1.41, 95% CI: 1.03–1.93) in early life, and male gender (RR =1.52, 95% CI: 1.27–1.82) predicted increased wheezing episodes. Conversely, increased fecal alpha-1-antitrypsin (RR= 0.87, 95% CI: 0.76–1.00) and family income (RR=0.98, 95% CI: 0.97–0.99) were associated with a decreased number of episodes of wheezing.

Conclusions

Systemic inflammation early in life, poverty, and male sex placed infants at risk of more episodes of wheezing during their first year of life. These results support the hypothesis that there is a link between inflammation in infancy and the development of respiratory illness later in life and provide specific biomarkers that can predict wheezing in a low-income country.

INTRODUCTION

Wheezing in early life in low income countries often occurs with acute lower respiratory infections, which are a leading cause of morbidity and mortality in children under five years of age [1,2]. Early childhood wheezing also increases the risk of later development of asthma, which in addition to increasing morbidity, places significant socioeconomic burden on already distressed populations [3,4]. Early life markers that predict the development of wheezing in infants in low-income populations are not well understood and could be useful in allowing early interventions and in understanding the pathogenesis of respiratory diseases. Systemic inflammation might provide such a marker. In high income countries autoimmune and inflammatory diseases often predispose to or co-occur with wheezing and asthma, however most studies have focused on adult populations [5–7]. Recent studies and a more nuanced understanding of the hygiene hypothesis, the idea that early life infections might help prevent the development of allergies and asthma, and in turn wheezing, has suggested that infections with some pathogenic organisms are in fact detrimental to later respiratory health [8–11]. Clinical and subclinical infection with enteric and non-enteric organisms is very common in children in low income countries and likely represents an environmental factor that contributes to a persistent inflammatory response in those children [12,13].

Pathogens encountered by children in these communities can significantly elevate serum levels of pro-inflammatory cytokines such as IL-1β and macrophage inflammatory protein (MIP1β) along with regulatory cytokines such as IL-10 [12,14,15]. In high income countries these cytokines have also been shown to be associated with obstructive respiratory diseases and wheezing [16,17]. In addition to inflammation stimulated by respiratory pathogens and exposures, children in developing country settings may have additional exposures leading to inflammation, especially in settings of poor sanitation and repeated exposure to enteric pathogens. Indeed, environmental enteropathy, an inflammatory condition of the intestine that can occur in individuals without evidence of overt diarrhea and disease, is also very common in children in low income countries and is thought primarily to be the result of poor sanitation and repeated infection with enteropathogens [12,13,18–20]. Environmental enteropathy is defined pathologically by inflammatory cell infiltration and villous blunting in the mucosa of the small intestine and is associated with gut barrier dysfunction and intestinal and systemic inflammation, including elevated serum IL-1β, and elevated C-reactive protein at multiple time points [13,19,21]. Markers of gut barrier dysfunction include, but are not limited to, α 1 antitrypsin (ATT). AAT in stool is commonly utilized as a marker of protein-losing enteropathy in children [22].

Thus, considering the potential impact of systemic inflammation on respiratory outcomes, we postulated that markers associated with environmental enteropathy, including serum inflammatory cytokines in the first weeks of life, might predict wheezing outcomes over the first year of life in children in Mirpur, Dhaka Bangladesh. To test this idea an exploratory analysis of the ability of early inflammation to predict wheezing was conducted in an independent cohort from the same community. 37.5% of children in this exploratory cohort had at least one instance of wheezing. Elevated serum IL-10 (top 25^th percentile) at 6 months of life was significantly associated with increased episodes of wheezing by one year of age (rate ratio (RR)= 1.48, 95% confidence interval (CI) 1.19–1.77). This preliminary data suggested that systemically elevated serum cytokines early in life might predict wheezing.

METHODS

Hypothesis generation cohort

A preliminary investigation of the association of serum cytokine levels with number of nebulizer treatments was performed in a small subset of children (147) from an earlier birth cohort study from the Mirpur area of Bangladesh. These 147 infants were enrolled after birth, beginning in January 2008, and followed until one year of age. The study period ended September 2009. Full details of the design of this birth cohort study have been described [23]. The results from this subset of 147 infants showing significant association of cytokine measures led us to the detailed investigation of biomarker effects on wheezing outcomes in the “ Performance of Rotavirus and Oral Polio Vaccines in Developing Countries” (PROVIDE) study cohort of 700 infants.

Study design and subjects

The characteristics of the PROVIDE study population and study design have been previously described [20,24]. Briefly, a birth cohort of 700 children from the Mirpur urban slum in Dhaka, Bangladesh were recruited and followed for 2 years. The objectives of the study were to determine the effect of environmental enteropathy and inflammation on oral vaccine performance and infant health. Our objective was to evaluate respiratory outcomes in the first year of life in this cohort. In the first year of life, 626 children remained in the study by month 6 and 609 by month 12 of follow-up. There were 4 deaths during the entire PROVIDE study which occurred before one year of age and were due to respiratory distress, acute diarrhea, neonatal sepsis, and cardio-respiratory failure. All studies were approved by the Ethical Review Committee of the ICDDR, B and the Institutional Review Boards of the Universities of Virginia and Vermont. Informed consent was obtained for all participating mothers and infants, including permission for repeated blood draws.

Measurement of Wheezing

Episodes of wheezing requiring clinician-administered nebulizer treatment were counted over the first year of life for each child in the study. Wheezing was assessed by the clinic physician as a continuous whistling or course airway sound during inspiration and/or exhalation [25]. An episode of wheezing was formally defined for this study as at least one nebulizer treatment (salbutamol, 0.1 mg per kilogram body weight) within a 10 day period [26]. In this study, our clinic served as the primary care clinic for the enrolled children and access to other healthcare was extremely limited, including home use of nebulizers.

Baseline characteristics and biomarkers

Baseline characteristics in this study included monthly family income, expenditure and maternal education. Biomarkers utilized in this analysis and the number of children available for each marker in both univariate and multivariate analyses are indicated in Table S1, along with clinical reference values when available. Mean± Standard Error (S.E.) for continuous biomarkers and percentage for categorical measures is also provided in Table S1. Only biomarkers measured at 6, 12, and 18 or 24 weeks were included in the univariate and multivariate analyses, and those children with missing biomarker measurements were excluded from the individual analysis. Cytokine biomarkers of systemic inflammation were categorized into the bottom 50^th percentile, 50^th–75^th, and 75^th and above, with the reference the bottom 50^th.

Measure of Sustained inflammation

C- reactive protein (CRP) was measured at weeks 6, 18 and 40. A CRP value was considered to be elevated if it was above the median value of CRPs at each time. Children were classified into four groups: Group 0 consisted of those in which none of their CRPs were elevated, Group 1 if one CRP measure was elevated, Group 2 if two CRPs were elevated and Group 3 if all three CRPs were elevated. Group 0 was considered as the reference in this analysis. Sustained inflammation with respect to CRPs was defined as if at least two CRP measures were elevated.

Statistical Analysis

Univariate Poisson regression was performed to evaluate the association of biomarkers from the PROVIDE study with episodes of wheezing. Based on univariate p values, the false discovery rate (FDR) was calculated to correct for multiple comparisons and those biomarkers with a FDR < 20% were considered as informative or important. The objective of the univariate analysis with FDR correction was to identify significant biomarkers individually that predicted episodes of wheezing over a one year period. While univariate analysis indicates the marginal effect of individual biomarkers, multivariable analysis can show joint effects of biomarkers. However, since there are nearly 50 predictors (Table S1) in this analysis, a situation of high-dimensional data, traditional multivariable analysis was not an optimal approach. Thus, a modern variable selection method with smoothly clipped absolute deviation (SCAD) regularization was utilized to identify a subset of informative biomarkers jointly associated with the wheezing outcome. The final model was determined using the Bayesian information criterion (BIC) and based on the biological plausibility. Results were confirmed with stepwise Poisson regression, not shown. In addition, univariate Poisson regression was performed to evaluate an association between sustained inflammation, as measured by number of elevated CRP tests, and wheezing. All statistical analyses were performed using SAS 9.3 (SAS Institute; Cary, NC) and the “grpreg” package in R 3.1 (www.r-project.org).

RESULTS

In the PROVIDE cohort 43.3% of children had at least one episode of wheezing over the first year of life (Figure 1). Univariate analysis (Table 1, S1) showed that increased risk of wheezing was associated with male gender (RR:1.52, 95% CI: 1.27–1.82), CRP (RR: 1.01, 95% CI: 1.00, 1.02), IL-10 (RR: 1.51, 95% CI: 1.22–1.86), IL-1β (RR: 1.55, 95% CI: 1.25–1.92), Mip1β (RR: 1.26, 95% CI: 1.02–1.56) and IL-7 (RR: 1.27, 95% CI: 0.98–1.63). Increased family income (RR: 0.98, 95% CI: 0.96–0.99), retinol binding protein, at 6 and 18 weeks of age (RR: 0.98, 95% CI: 0.97–0.99, and RR: 0.98. 95% CI: 0.98–0.99) and fecal alpha-1-antitrypsin at 12 weeks of age (RR: 0.87, 95% CI: 0.76–1.00) were associated with decreased episodes of wheezing.

Figure 1.

Number of episodes of wheezing requiring nebulizer treatment in the first year of life in the PROVIDE cohort.

Table 1.

Predictors of wheezing by one year of age (univariate Poisson regression)

Predictor/Biomarkers	Rate Ratio	95% CI, L	95% CI, H	FDR
Male gender	1.52	1.27	1.82	<0.01
Retinol binding protein at week 6	0.98	0.97	0.99	<0.01
Retinol binding protein at week 18	0.99	0.98	1.00	0.01
C reactive protein at week 18	1.01	1.00	1.02	0.14
alpha 1 antitrypsin at week 12	0.87	0.76	1.00	0.14
expenditure	0.98	0.97	0.99	0.03
income	0.98	0.97	0.99	0.01
IL-1 β week 18_2*	1.43	1.15	1.78	0.01
IL-1 β week 18_3	1.55	1.26	1.93	<0.01
IL-10 week 18_2	1.27	1.02	1.59	0.12
IL-10 week 18_3	1.51	1.22	1.87	<0.01
Mip1β week 18_3	1.26	1.02	1.56	0.12
IL-7 week 18_2	1.37	1.07	1.75	0.05
IL-7 week 18_3	1.27	0.98	1.63	0.17

^*Cytokine biomarkers were categorized as bottom 50th percentile, 50^th –75^th (2) and above 75^th (3).

Multivariable analysis demonstrated that male gender, IL-10, IL-1β, family income, retinol binding protein, and fecal alpha-1-antitrypsin were significant independent predictors of wheezing (Table 2, S1). Sustained inflammation, as defined by elevated CRP levels at two or three time points (CRP group 2, 3) was also positively and significantly associated with the number of episodes of wheezing in the first year of life (Table 3).

Table 2.

Predictors of wheezing (SCAD multivariable selection)

Predictors/Biomarkers	Rate Ratio
Male gender	1.46
Retinol binding protein at week 6 (mg/ml)	0.997
Retinol binding protein at week 18 (mg/ml)	0.999
Alpha-1-antitrypsin at week 12 (mg/g)	0.93
Income (1000 Taka)	0.98
IL-1 β at week 18, 50–75th percentile	1.07
IL-1 β at week 18, above 75th percentile	1.11
IL-10 at week 18, 50–75th percentile	1.03
IL-10 at week 18, above 75th percentile	1.06

Table 3.

Association of sustained inflammation with future wheezing.

CRP group	No. of Children (%)	Rate Ratio	95% CI, L	95% CI, H	P value
CRP group 0*	102 (17.9)	1.0 (ref.)
CRP group 1	178 (31.2)	1.09	0.81	1.46	0.59
CRP group 2	186 (32.6)	1.49	1.12	1.97	0.01
CRP group 3	104 (18.3)	1.41	1.03	1.93	0.03

^*CRP was measured at weeks 6, 18 and 40, and defined as elevated if the measure was above the median for each week. Group 0: non CRP measures were elevated, Group 1: 1 CRP measure elevated, group 2: 2 CRPs were elevated, group 3: all three CRPs were elevated.

DISCUSSION

The most important discovery of this work was the association of early life inflammation with increased risk of wheezing in the first year of life. This was found in two independent birth cohorts of slum dwelling infants in urban Bangladesh. Wheezing was common in both of these cohorts as expected, with 37.5% and 43.3% of children having at least one episode over a 1-year period [27]. Earlier work is supportive of a link of inflammation with wheezing. Increased IL-10 production from monocytes was predictive of a recurrence of wheezing in children from the Netherlands [17]. Children from Rotterdam who developed wheezing in the first year of life had an increased IL-10/IL-12 ratio [28]. Finally, pro-inflammatory cytokines were observed in bronchoalveolar lavage fluid from infants in Finland with frequent wheeze [29]. The PROVIDE study however represents the most comprehensive survey of inflammatory markers and environmental factors associated with risk of wheezing in infants in a tropical urban slum. It is unique in showing that systemic inflammation, as measured by increased levels of serum IL-1β and IL-10 and sustained C reactive protein (CRP), is predictive of an increased risk of wheezing in this population.

IL-1β is a pro-inflammatory cytokine that is significantly elevated during both enteric and respiratory infections while IL-10 is an anti-inflammatory cytokine that is often induced in response to pro-inflammatory cytokines such as IL-1β [15,30]. It was therefore anticipated that elevation in both of these cytokines would be observed in inflammatory states and would predict episodes of wheezing, as shown by our analysis. AAT in stool is commonly utilized as a marker of protein-losing enteropathy in children [22]. AAT was not clinically elevated in this cohort of children. However, it was expected that AAT would serve as a marker of increased inflammation and intestinal damage in this analysis, and that higher levels might be expected to predict increased wheezing, rather than decreased. Future studies will examine the implications of this unexpected finding.

Additional serum inflammatory markers predicted wheezing in the univariate analysis, including C reactive protein (CRP), IL-7, associated with T cell and macrophage activation, and Mip1β, associated with macrophage and neutrophil activation. Further supporting an association of inflammation and wheezing was the observation that an increased number of time points with elevated CRP predicted increased episodes of wheezing. This result suggests that sustained inflammation, known to occur in environmental enteropathy, is also an important marker of the development of wheezing [19].

Our univariate analysis selected serum retinol binding protein as a predictor of decreased episodes of wheezing. Our multivariable analysis also indicated that serum retinol binding protein could predict episodes of wheezing, but does not permit a strong statement of directionality. Retinol binding protein is a negative acute phase protein, and levels of it significantly decrease during infection and sepsis [31]. However, retinol binding protein levels also closely correlate with levels of dietary vitamin A, and can be utilized as a marker of vitamin A deficiency even during infection and protein malnutrition [32]. Children with vitamin A deficiency have previously been shown to be more likely to wheeze during acute respiratory infections [33]. Future studies will examine the etiology of the association between increased serum cytokines and level of retinol binding protein and wheezing to determine if enteropathogen burden or dietary vitamin A intake associate with serum retinol levels and wheezing in these children.

Low family income has in other studies also been associated with increased episodes of wheezing in the 1^st year of life, as has male gender[34,35]. In this study, these were independent of inflammation in multivariable analysis, suggesting that their potential to promote wheezing is through some mechanism other than inflammation. Maternal factors, including education, height and weight, were evaluated as potential factors influencing wheezing as previous studies have suggested that maternal health and education might influence childhood respiratory health[36]. However, we found no association between these factors and wheezing. Future studies will examine maternal factors and their influence on wheezing in more depth.

The strengths of our study included a large number of children in an isolated cohort and the collection of a large amount of data on subjects with repeated serum and fecal samples. Thus health care data could be utilized and a comprehensive set of biomarkers were able to be developed and evaluated. Nevertheless, this study had some limitations. The cause of the wheezing episodes was not measured, however, the young age of the population and the lack of association of serum IL-4 and IL-5 with wheezing, make a viral and not atopic origin of wheezing more likely. Another limitation was the high baseline level of inflammation in the population as a whole which may have masked the full contribution of inflammatory markers and their role in exacerbating wheezing episodes [18]. Finally, the two cohorts reported here were observational and cannot address causality but merely note the association between biomarkers of inflammation and wheezing. In future work it will be important to address the cause of early inflammation, which we hypothesize to be due to a combination of enteric and respiratory infections, so that directed interventions can be tested for their impact on mitigation of wheezing in infant and later childhood populations.

In summary, we have shown that systemic inflammation was associated with an increased risk of wheezing in the 1^st year of life in slum dwelling children from Bangladesh. This study demonstrated that IL-1β, IL-10, alpha-1-antitrypsin and C-reactive protein were important biomarkers in early life, predicting the later development of wheezing. The greater significance of this work lies in the demonstration that systemic inflammation may predispose to more severe respiratory outcomes later in life.

Highlights.

• Factors that influence wheezing in early childhood in the developing world are not well understood and may be useful in predicting respiratory outcomes.

• Children from Bangladesh were recruited at birth and episodes of wheezing were measured alongside nutritional, immunological and socioeconomic factors over a one-year period.

• Systemic inflammation early in life, poverty, and male sex placed infants at risk of more episodes of wheezing during their first year of life.