Markers of environmental enteric dysfunction are associated with neurodevelopmental outcomes in Tanzanian children

Abstract

Background

Chronic exposure to enteropathogens may result in environmental enteric dysfunction (EED), a subclinical condition associated with poor child growth. Growth faltering is strongly associated with poor neurodevelopment, and occurs during sensitive periods of postnatal brain development. We investigated the role of novel EED biomarkers, systemic inflammation and micronutrient status on neurodevelopment in Tanzanian children.

Methods

Non-stunted subjects with 6 week and 6 month blood samples and neurodevelopmental measures (n=107) were included in this study. Samples were tested for biomarkers of gastrointestinal function (citrulline, antibodies to lipopolysaccharide (LPS) and flagellin), micronutrient status (iron, retinol binding protein (RBP), and vitamin D), systemic inflammation (C-reactive protein (CRP) and alpha-1-acid glycoprotein (AGP)), and growth (insulin-like growth factor (IGF-1) and insulin-like growth factor binding protein 3 (IGFBP-3)).

Results

Cognitive scores at 15 months were associated with higher concentrations of 6 month anti-LPS IgG (β=1.95, p-value=.02), anti-flagellin IgA (β=2.41, p-value=.04), and IgG (β=2.99, p-value=.009). Higher receptive language scores were positively associated with anti-flagellin IgG (β=0.95, p-value=0.05), and receptive language and gross motor scores were positively associated with citrulline at 6 months (β=0.09, p-value=.02; β=0.10, p-value=.03, respectively). Gross motor scores were positively associated with RBP at 6 months (β=1.70, p-value=.03). Markers of systemic inflammation were not significantly associated with neurodevelopment.

Conclusions

Plasma citrulline, a marker of gastrointestinal mucosal surface area, and vitamin A status were associated with higher gross motor development scores. Novel markers for EED, but not inflammation, were positively associated with cognitive scores, suggesting a possible mechanistic pathway involving immune response and neuroprotection.

Introduction

More than 200 million children worldwide do not fulfill their developmental potential, with consequences of deficits in school achievement, adulthood economic productivity and the intergenerational transmission of poverty (1–3). Socio-cultural factors and poverty give rise to multiple interacting risk factors that impact the sensorimotor, socioemotional and cognitive language dimensions of neurodevelopment (4). Modifiable biological factors in mental development include infection (malaria, diarrhea, gastrointestinal infections) and poor nutrition, and are risk factors also shared by poor growth. There exists a strong association between stunting and neurodevelopment (5), and the postnatal window during which neurodevelopment occurs coincides with the period in which the most detrimental effects of stunting have been observed (6), suggesting a common etiology. Moreover, the high prevalence of stunting can persist despite nutritional, sanitation, and healthcare interventions, and the effect size for nutritional interventions on cognitive, language and motor development are small (7, 8).

Emerging associations between growth faltering, gut dysfunction, and systemic and intestinal inflammation have provided new insight on the potentially shared etiologies of poor growth and neurodevelopment. The subclinical condition environmental enteric dysfunction (EED) impacts the structure and function of the small bowel, and is hypothesized to be the result of chronic exposure to poor sanitation and hygiene, conditions frequently met in resource-challenged countries. EED is characterized by changes to the villous structure of the small intestine, compromising nutrient absorption and the pathogenic barrier (increased permeability and inflammatory cell infiltrate), and impaired gut immune function, where diarrhea (and a specific entereopathogen profile) is not a necessary component (9, 10). Early studies on Gambian infants observed that linear growth was strongly associated with markers of intestinal permeability, but not with dietary inadequacy or diarrhea (11). Estimation of the quantity and burden of asymptomatic enteric infections in resource poor countries has only recently begun, suggesting that there is a constant, chronic exposure to multiple enteropathogens beginning shortly after birth (12). These and other observations suggest a mechanism for stunting and neurodevelopment that diverges from dietary inadequacy and diarrhea-mediated causes (13).

EED represents a potentially important mechanistic target for investigating the role of nutrition and poor sanitation in growth and development. To determine the role of micronutrient status, inflammation, and EED on neurodevelopment, we evaluated nutritional, growth, inflammatory, and novel EED biomarkers as factors for child neurodevelopment in a cohort of young Tanzanian children.

Study design and participants

Participants in this study were a subset of a parent 2×2 factorial randomized, double-blind, placebo-controlled trial (clinicaltrials.gov NCT 00421668) designed to examine whether daily administration of multivitamins and/or zinc to infants born to HIV-negative women from 6 weeks of age for 18 months reduced the risk of infectious disease morbidity and growth faltering, compared with placebo (N=2400) (14). A secondary goal of the trial was to evaluate neurodevelopment in a subset of the participants.

The trial took place in peri-urban Dar es Salaam, Tanzania. Trial methodology has been previously described (14). Briefly, mothers of potentially eligible infants were recruited into the study in one of two ways: 1) pregnant women ≤ 34 weeks gestation presenting at one of three prenatal clinics in Dar es Salaam were informed about the study and consented prenatally; or 2) women were recruited from the labor ward of Muhimbili National Hospital within twelve hours of delivering a healthy, singleton baby. In both cases, written informed consent was obtained and mothers were asked to present at a study clinic within 1–2 weeks of delivery for HIV testing. Maternal HIV status was determined using 2 sequential enzyme-linked immunosorbent assays (ELISA) that used the Murex HIV antigen/antibody (Abbott Murex) followed by the Enzygnost anti-HIV-1/2 Plus (Dade Behring) or the Enzygnost HIV Integral II Antibody/Antigen (Siemens). Any discrepancy between the first and second ELISA was resolved by a Western blot assay. Consenting mothers who were confirmed to be HIV-negative were enrolled into the study and their infants were randomized between 5–7 weeks of age. Infants of multiple births and infants with congenital anomalies or other conditions that would interfere with the study procedures were excluded. Birth characteristics were obtained immediately following delivery. We used reference data from Oken et al. (2003) to calculate the percentile of birth weight for each completed week of gestation and defined small-for-gestational age as ≤ 10th percentile (15). At the time of randomization, clinical examination was performed by a study physician; history of morbidity and infant feeding practices was conducted by a study nurse, and anthropometric measurements (i.e., length, weight, head circumference) were recorded. Mothers and children were followed from the time of randomization for 18 months, until the child’s death, or until loss to follow-up. Blood samples were drawn again at 6 and 12 months of age.

Participants in this study were subjects who had previously participated in a randomized trial of zinc or multiple micronutrients. A sub-sample of 247 children from the parent trial was selected for neurodevelopment assessment from a single research site (Magomeni Hospital) due to training and space restrictions. Of these, 107 children who also were non-stunted at 6 weeks of age (length-for-age z-score (LAZ) ≥ −2), had a blood sample at 6 weeks and 6 months of age available, and had at least one subsequent anthropometric assessment were selected for inclusion in this substudy. The Bayley Scales of Infant and Toddler Development- third edition (BSID-III) was administered at 15 months to evaluate neurodevelopment. A specialist in child neurology (DCB) travelled to Dar es Salaam to train the BSID-III test administrators and conducted observational evaluations to address quality control. Tests were administered in Kiswahili by two trained Tanzanian nurses during the child’s clinic visit. The BSID-III includes separate raw scores that reflect the infant’s cognitive functioning, expressive and receptive language skills, as well as fine and gross motor capabilities. We elected to use raw, unstandardized scores because the BSID-III has not been validated in the Tanzanian context, and it may not be appropriate to compare our study population’s scores with the US reference population used for standardization.

Biomarker Analyses

Biomarkers associated with enterocyte mass, mucosal permeability, and systemic inflammation were selected to capitulate the hallmark changes in small bowel morphology associated with EED. Peripheral blood samples taken at 6 weeks, 6 months and 12 months of age were tested for biomarkers of gut dysfunction (citrulline, anti-LPS, and anti-flagellin), nutrient absorption (iron, retinol binding protein (RBP), and vitamin D), systemic inflammation (C-reactive protein (CRP) and alpha-1-acid glycoprotein (AGP)), and growth (insulin-like growth factor (IGF-1) and insulin-like growth factor binding protein 3 (IGFBP-3)).

Anti-flagellin and anti-LPS-specific IgA and IgG levels were quantitated by ELISA, as has been previously reported (16). Microtiter plates were coated overnight with purified E. coli flagellin (100 ng/well), or purified E. coli LPS (2 μg/well). Serum samples from study participants diluted 1:200 were applied to wells coated with flagellin or LPS. After incubation and washing, the wells were incubated either with anti-human IgA (KPL) or IgG (GE Healthcare) coupled to horseradish peroxidase. Quantitation of total immunoglobulin was performed using the colorimetric peroxidase substrate tetramethylbenzidine, and optical density (OD) was read at 450 nm with an ELISA plate reader. Data are reported as OD corrected by subtracting background (determined by readings in samples lacking serum). Underivatized citrulline levels in serum specimens were evaluated following ultra-centrifugation based sample pre-treatment in the Clinical and Epidemiological Research Laboratory in the Department of Laboratory Medicine, Boston Children’s Hospital. The separation selectivity of citrulline was obtained using a C18 BEH column and the ion pairing agent pentadecafluorooctanoic acid on a Waters Acquity UPLC System. Positive electrospray ionization and a Waters Quattro premier triple-quad mass spectrometer are used to detect citrulline by its characteristic 176-70 and 176-113 m/z transitions. Analysis of ferritin, RBP, CRP, and AGP was performed using a combined ELISA method in the laboratory of Dr. Juergen Erhardt in Germany (17). Vitamin D (25-hydroxyvitamin D) was quantitated by high performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) using an API-5000 (AB Sciex, Foster City, CA) at Boston Children’s Hospital. High sensitivity CRP (hsCRP) concentration was determined using an immunoturbidimetric assay on the Roche P Modular system (Roche Diagnostics - Indianapolis, IN), using reagents and calibrators from Roche. IGF-1, IGFBP-3, and AGP (in an expanded sample set), was measured by an ELISA method from R&D Systems (Minneapolis, MN) at Boston Children’s Hospital. CD4 and CD8 T cell counts were measured at 12 months in a subset of children using FACSCalibur system (Becton Dickinson) to evaluate immunologic response.

Testing for AGP and CRP were conducted in 2 separate laboratories in semi-overlapping sample populations. Results from sample IDs tested by both laboratories were highly correlated (N=42; CRP R²=0.97, AGP R²=0.87), therefore all acute phase marker results were combined, where the hsCRP (Roche) results were taken as the preferred measure given improved accuracy at lower concentrations.

Statistical Analyses

For this analysis, 107 children were included who had a BSID-III score, biomarker data on at least one time point, and all covariate data. Since iron deficiency and anemia were important to adjust for in our models and because of incomplete overlap of sample testing, the effective number of observations used in a given model reflects a subset of the maximum (n=107) where the biomarker, iron deficiency and anemia variables were all nonmissing. Raw BSID-III scores were taken separately as 5 indices of developmental outcomes (cognitive, expressive and receptive language, fine and gross motor skills).

Data were double entered using Microsoft Access software at the central study site, then converted to SAS datasets and uploaded to a secured UNIX-based server for analysis. Descriptive statistics were used to summarize the baseline characteristics of the study population, biomarker results, and BSID-III raw scores. Frequencies were reported for the categorical variables and the mean ± standard deviation (SD) for continuous variables. Multivariate linear regression models to estimate the effect of each biomarker separately on BSID-III score. Included as covariates in adjusted models were BSID-III examiner, study arm, maternal mid-upper arm circumference (MUAC), maternal age, infant sex, infant age, baseline LAZ and weight-for-age z-score (WAZ), exclusive breastfeeding (at 6 weeks or at 6 months), any respiratory infection or diarrhea diagnosis, anemia (Hb < 11g/dL), iron deficiency (serum ferritin < 12ug/L), and inflammation (CRP > 5mg/L or AGP > 1g/L). Covariates were selected for inclusion in the adjusted models based on previous associations with the outcome or potential influence on predictors. Models that evaluated ferritin (CRP, AGP) as continuous predictors of BSID-III score did not include the iron deficiency (inflammation) binary variable. Stratified analyses were performed to determine the effect of receiving zinc in the parent trial. All analyses were performed using SAS software (version 9.3; SAS Institute). Institutional approval was granted by the Harvard T.H. Chan School of Public Health Human Subjects Committee, the Muhimbili University of Health and Allied Science Committee of Research and Publications, the Tanzanian National Institute of Medical Research, and the Tanzanian Food and Drugs Authority. Over the course of the study, a Data Safety and Monitoring Board met twice annually.

Results

The subset of children included for this analysis (n=107) did not statistically differ on most maternal baseline characteristics from the parent trial participants (N=2400), except for spending more on food (daily food expenditure per individual < 1000 Tanzanian Shillings (TShs) 14.7% versus 29.5%, p-value=0.001), and fewer people in the household (3.2 versus 3.6, p-value=0.04), (Table 1). Additionally, this subset did not differ on most infant characteristics, except for fewer having preterm births (4.7% in the subset versus 13.4%, p-value=0.02) and lower weight-for-length z-scores −0.17 versus 0.13, p-value=.02 and more diarrhea and/or respiratory infections (no diarrhea or respiratory infections: 36.5% versus 49.6%, 1 to 3: 54.2% versus 46.1%, or 3 or more: 9.4% versus 4.3%, p-value=.005) (Table 2).

Table 1.

Maternal baseline characteristics

Maternal Characteristics	Mean ± SD or N (%) N=107
Age, years	26.0 ± 4.6
Formal education (years)
0	1 (0.9)
1–7	82 (77.4)
≥ 8	23 (21.7)
Employment
Housewife without income	63 (60.0)
Housewife with income	36 (34.3)
Other	6 (5.7)
Has personal income	97 (39.3)
Married or cohabitating with partner	96 (91.4)
Prior pregnancies
None	34 (32.4)
1–4	68 (64.8)
≥ 5	3 (2.9)
Mid-upper arm circumference (cm)	26.5 ± 3.6
Socioeconomic Characteristics
Daily food expenditure/person/house < 1000 TShs	15 (14.7) 1
Household possessions
0	29 (27.4)
1–3	60 (56.6)
≥ 3	17 (16.0)
Number of people in household	3.2 (1.3) 2

¹Compared to those with no BSID-III score and/or biomarker data, P=0.001

²Compared to those with no BSID-III score and/or biomarker data, P=0.04

Table 2.

Child baseline characteristics and BSID-III raw scores at 15 months

Child characteristics	Mean ± SD or n (%) n=107
Male	53 (49.5)
Low birth weight, <2500g	2 (1.9)
Born preterm, <37 weeks	4 (4.7) 1
Apgar score ≤ 7 at 5 minutes after birth	0 (0)
Anthropometrics at 6 weeks
Length-for-age Z score	−0.15 ± 0.97
Stunting	02
Weight-for-length Z score	−0.17 ± 1.203
Wasting	10 (9.4)
Weight-for-age Z score	−0.28 ± 0.90
Underweight	3 (2.8)
Exclusive breastfeeding at 6 weeks	66 (61.7)
Exclusive breastfeeding at 6 months	0 (0)
Respiratory infections or diarrhea diagnosis ≤ 15m
0	39 (36.5)4
1–3	58 (54.2)
≥ 3	10 (9.4)
Raw BSID-III scores at 15 months
Cognitive Scale	50.15 (3.3)
Expressive Language	19.74 (2.6)
Receptive Language	19.03 (2.0)
Fine Motor	34.83 (2.9)
Gross Motor	47.94 (2.2)
Age at testing (mo)	14.66 (0.4)

¹Compared to those with no BSID-III score and/or biomarker data, P=0.02

²Length-for-age Z score < −2, selection criteria included non-stunted at baseline

³Compared to those with no BSID-III score and/or biomarker data, P=0.02

⁴Compared to those with no BSID-III score and/or biomarker data, P=0.005

Most mothers received education only up to the age of about 13 (76.3% had 1–7 years of education), and were married housewives without income (Table 1). Children in the study were mostly delivered at term (≥ 37 weeks), birth weight was above 2500g, and at baseline (6 weeks of age), were anthropometrically average (Table 2). Univariate statistics for raw BSID-III scores were not significantly different between children that did or did not have a plasma sample.

Parameter estimates and p-values in unadjusted and adjusted linear regression models evaluating the role of biomarkers at 6 months on BSID-III score at 15 months are presented in Table 3. Six month anti-LPS IgG, anti-flagellin IgA and IgG levels were all positively associated with cognitive scores (β=1.95, p-value=0.02; β=2.41, p-value=0.04; β=2.99, p-value=0.009, respectively) and flagellin IgG antibodies were positively associated with receptive language scores (β=0.95, p-value=0.05). Citrulline levels at 6 months was positively associated with receptive language and gross motor scores (β=0.09, p-value=0.02; β=0.10, p-value=0.03, respectively). RBP at 6 months was positively associated with gross motor scores (β=1.70, p-value=0.03). A small negative association was found between ferritin and CRP at six months and gross motor scores (β=−0.008, p-value=0.02; β=−0.07, p-value<0.001, respectively).

Table 3.

Relationship between Biomarkers at 6 Months on BSID-III Scores at 15 Months in Tanzanian Infants

	N	Cognitive		Receptive Language		Gross Motor
		Unadjusted Beta (p-value)	Adjusted1,2 Beta (p-value)	Unadjusted Beta (p-value)	Adjusted1 Beta (p-value)	Unadjusted Beta (p-value)	Adjusted1 Beta (p-value)
anti-LPS IgA	65	1.03 (0.16)	1.39 (0.07)	0.39 (0.20)	0.56 (0.08)	−0.41 (0.19)	−0.55 (0.11)
anti-LPS IgG	65	1.30 (0.10)	1.95 (0.02)	0.42 (0.21)	0.58 (0.10)	−0.24 (0.78)	−0.36 (0.35)
anti-Flagellin IgA	65	2.36 (0.04)	2.41 (0.04)	0.47 (0.32)	0.34 (0.48)	−0.75 (0.12)	−0.85 (0.10)
anti-Flagellin IgG	65	2.05 (0.06)	2.99 (0.009)	0.71 (0.12)	0.95 (0.05)	0.65 (0.93)	−0.03 (0.95)
Citrulline	57	0.04 (0.66)	0.13 (0.18)	0.09 (0.01)	0.09 (0.02)	0.07 (0.06)	0.10 (0.03)
RBP	76	0.62 (0.71)	−0.78 (0.68)	0.84 (0.20)	0.73 (0.33)	1.74 (0.009)	1.70 (0.03)
Ferritin3	76	0.01 (0.05)	0.01 (0.07)	−0.001 (0.70)	−0.001 (0.64)	−0.004 (0.14)	−0.008 (0.02)
CRP4	76	0.008 (0.85)	0.02 (0.72)	0.003 (0.85)	0.007 (0.73)	−0.07 (<0.001)	−0.07 (<0.001)
AGP4	76	0.83 (0.32)	0.85 (0.34)	−0.02 (0.95)	0.01 (0.97)	−0.64 (0.06)	−0.60 (0.11)

¹Covariates: study arm (zinc only, zinc+multivitamin, multivitamin only, placebo), BSID-III examiner; maternal age, mid-upper arm circumference; infant sex, age, baseline LAZ and WAZ, exclusive breastfeeding at 6w or at 6m, any respiratory infection or diarrhea diagnosis, anemia (Hb < 11 g/dL), inflammation (CRP > 5mg/l or AGP > 1.0 g/l), iron deficiency (serum ferritin < 12 ug/l)

²Number of people in household significant in adjusted model

³Not adjusted for iron deficiency

⁴Not adjusted for inflammation

Biomarkers at 6 weeks or 12 months (timepoints 1 and 3) were not associated with neurodevelopment at 15 months of age. Anemia at 6 months was associated with fine motor scores in unadjusted (β=−1.18, p-value=0.02) and adjusted models (β=−1.05, p-value=0.03). BSID-III scores were not associated with IGF-1, IGFBP-3, and vitamin D levels at any time point, nor the 12 month CD4 or CD8 T cell count measures. Treatment arm was not associated with neurodevelopment independently of any biomarker.

Discussion

In this study designed to evaluate the relationship between novel biomarkers of EED, nutritional and inflammatory biomarkers, with neurodevelopmental outcomes in Tanzanian children, we found that serum immunoglobulins towards flagellin and LPS were significantly associated with cognitive, receptive language, and gross motor scores. Higher blood concentrations of anti-LPS IgG, anti-flagellin IgA and IgG at 6 months were consistently associated with higher 15 month cognitive scores. Additionally, we found that citrulline was positively associated with receptive language and gross motor scores, and RBP was also positively associated with gross motor scores. Finally, we identified a very small negative association between ferritin and CRP and gross motor scores, and no association between AGP, vitamin D, IGF-1, or IGFBP-3 and BSID-III score.

Normally, the bacterial products LPS and flagellin are present in high density in the gut lumen but they are generally excluded from absorption by the gut epithelium. These molecules are potent ligands for Toll-like receptors 4 and 5 (TLR4 and TLR5) and the NLRC4 inflammasome that operate critically in innate immunity and strongly affect gut function in both health and disease (18–20). Given the high density of these molecules in the gut lumen and their microbial origin, they are credible biomarkers for gut barrier function. Hence children with higher blood concentrations of these immunoglobulins are thought to have had greater systemic exposure to gut bacteria as a consequence of increased gut permeability (16, 21).

We therefore had initially anticipated that subjects with higher peripheral markers of EED would be at risk for lower scores of neurodevelopment (22). Our findings of a positive association between these antibodies and cognitive scores may instead indicate that children with higher cognitive scores have a more robust immune response. Microglial cells function as key defenders against invading pathogens within the central nervous system, whose activation, whether by neurodevelopmental processes or by inflammation, leads to secretion of a panoply of cytokines and chemokines with resultant effects on synapse development and plasticity (thus learning and memory) (23). TLRs are expressed on microglia, astrocytes, oligodendrocytes, and neurons, and have been implicated in neuroinflammation in response to infection (24). More recently TLRs have been implicated in neurogenesis and neuroplasticity, suggesting potential involvement in cognitive processes; studies using TLR-deficient mice and TLR agonists to evaluate cognitive performance have identified deficits in learning and memory, some perhaps independently of neuroinflammatory-mediated pathways (for a review, see (25)). Effects of TLR activation are cell-specific and are modified by co-activation of TLRs (26); it is of note that in our population, multiple TLR ligands may be present at once, and given the various cell-types involved in the complex process of cognition, it remains unclear the precise effects EED has on neurodevelopment.

Similar to our findings, Jiang et al. identified a significant positive association between elevated IL-4, a cytokine associated with Th2-like immune response, and cognitive BSID-III score in Bangladeshi infants (27). Flagellin has been shown to induce a Th2-biased response and it is known that the ligation of TLR5 with flagellin has downstream effects resulting in the transcriptional activation of genes involved in antibacterial function, immune cell chemoattractants, and heat-shock proteins (among other stress-induced genes), indicating cytoprotective potential (28–31). Additionally, LPS-induced TLR4 signaling in microglia had been shown to be responsible for expression of genes indicative of neuroprotection (32). It is hypothesized that short-term immunologic response may be neuroprotective, whereas chronic or repeated exposure to resulting cytokines and chemokines potentially induce functional alterations or prolonged microglial activation with compromising effects (33, 34). Repeated measurements at subsequent time points will be critical to understanding the true nature of the exposure as will forthcoming developments in studies of neuroimmunomodulation.

Our lack of findings for significant associations between systemic inflammatory markers and neurodevelopmental outcomes bears comment. A proposed inflammatory mechanism by which EED influences growth and development is through the release of pro-inflammatory cytokines by innate immune cells and subsequent hepatic production of acute phase proteins (e.g., CRP, AGP) by products of bacterial translocation (10, 35, 36). Additionally, recent research supports a role for peripheral immune response as a mediator of central nervous system functioning, with demonstrated effects in both learning and plasticity (37, 38), sickness behavior and depression (39) and cognitive decline (40). Maternal prenatal and infant nutritional status may further modify neurodevelopment through peripheral activation of the immune system (41, 42). Given such observations, we had hypothesized that peripheral inflammation would be associated with lower neurodevelopmental scores.

The utility of including inflammatory markers in nutrition and development-related studies is well described, however, given the vast number of possible markers, their pleiotropic effects, and their mutability by co-exposures and genetic variation, the interpretation of results has been a topic of considerable interest (43). CRP and AGP are often included in studies as indicators of inflammation because they are easy to measure, have an established response following infection or tissue injury, and may be useful for the detection of subclinical inflammation. Although we did not find evidence of an effect of systemic inflammation on neurodevelopment in this study, it is possible that inflammatory events occurring interim to our measurement (6 weeks and 6 months) contributed to these findings. Alternatively, different cutoffs for CRP and AGP may be required, or use of other markers may improve the evaluation of systemic inflammation on neurodevelopment. Timing, duration of exposure, and magnitude of response are all critical factors when interpreting profiles of inflammatory proteins and disease (44), and the search for biomarkers of inflammation that capture the nuances of subclinical conditions in global public health applications is well under way (45).

Taken together, our results indicate that 6 month markers of vitamin A status (RBP) and GI mucosal surface area (citrulline) are associated with better gross motor scores. Further, our analyses identified a positive association between citrulline levels at 6 months and receptive language. Citrulline, a non-essential amino acid synthesized from glutamine and proline by the enterocytes of the small intestine, has recently emerged as a candidate biomarker of enterocyte mass and absorptive capacity (46–48). We and others (49) have linked serum citrulline levels with GI absorptive capacity in children with intestinal failure, identifying significant correlations between citrulline concentrations and bowel length, dependence on parenteral nutrition (50), and the occurrence of catheter-related bloodstream infections, reflecting impaired intestinal barrier function (51). The results from our investigation provide new insight about GI health and language and motor development.

The results of our study provide new insight for the potential roles of gut dysfunction and immunomodulation, independent of inflammation, in early child development. To our knowledge, this is the first such study of these EED markers and neurodevelopmental outcomes conducted in a low-resource setting. Nonetheless, there are many limitations of our study that make these findings preliminary and in need of confirmation. Our p-values were not corrected for multiple testing, given the nature of the exploratory investigation. Further, it was not possible to distinguish whether the source of the flagellin and LPS antibodies were maternal- or infant-derived. Maternally derived immune factors (transplacentally or via breast milk) may be responsible for the observed associations. The selection criteria for this study required children to be non-stunted at baseline. Given the known association between growth faltering and poor neurodevelopment, the children selected for inclusion likely represent healthier children with better neurodevelopmental scores than children in the population from which the subjects were drawn. This may represent an upward bias in our BSID-III scores and nutritional assessments, and a downward bias in measures of systemic inflammation. Additionally, despite using raw (unstandardized) BSID-III scores in our analyses, it should be acknowledged the inherent limitations of adapting the BSID-III developmental assay to a different context than for that which it was developed. Further, since the assays used to detect antibodies to LPS and flagellin in our study preferentially recognize gram-negative bacteria like E. coli, the total burden of exposure to pathogens may have been greater than the subset measured in our study, leading us to potentially underestimate the effect of EED on neurodevelopment. Finally, our sample size was limited, and we did not find an association between markers at 6 weeks and 12 months of age with neurodevelopment status at 15 months.

Conclusions

In this preliminary study, we found that markers of environmental enteric dysfunction were associated with neurodevelopment in children. Our evaluation of novel biomarkers for EED and systemic inflammation on cognition potentially implicate an alternative pathway involving immune response and neuroprotection via TLRs 4 and 5. Further, we show that markers of gastrointestinal health and nutritional status may be associated with gross motor development. Future studies of early childhood development should incorporate measures of gastrointestinal and nutritional status to confirm these findings.

Abbreviations

AGP

Alpha-1-acid glycoprotein

BSID-III

Bayley Scales of Infant and Toddler Development- third edition

CRP

C-reactive protein

EED

Environmental enteric dysfunction

ELISA

Enzyme-linked immunosorbent assays

HIV

Human immunodeficiency virus

IGF-1

Insulin-like growth factor

IGFBP-3

Insulin-like growth factor binding protein 3

LAZ

Length-for-age z-score

LPS

Lipopolysaccharide

MUAC

Middle-upper arm circumference

RBP

Retinol binding protein

TLR4

Toll-like receptor 4

TLR5

Toll-like receptor 5

TShs

Tanzanian shillings

WAZ

Weight-for-age z-score