Transitory impact of subclinical Shigella infections on biomarkers of environmental enteropathy in children under 2 years

TL;DR

This study found that subclinical Shigella infections in young children cause temporary increases in gut inflammation markers, but these effects do not last long enough to explain long-term growth issues.

Contribution

The study is the first to show that the impact of subclinical Shigella infections on environmental enteropathy biomarkers is transient, not sustained.

Findings

MPO concentrations increased during Shigella infection but returned to baseline within a month.

NEO and AAT concentrations showed delayed decreases but not sustained elevation.

The transient nature of biomarker changes suggests they are not reliable indicators of long-term growth effects.

Abstract

Clinical and subclinical Shigella infections among children living in low- and middle-income countries (LMICs) have been associated with long-term adverse effects such as impaired linear growth. The mechanism for the impact of subclinical infections has been theorized to occur through contributions to environmental enteropathy (EE). While Shigella has previously been associated with biomarkers of EE at the time of infection, we evaluated whether this impact was sustained after infections, which would support EE being the mechanism for the effects of Shigella on growth. A prospective birth cohort study of 1,715 children living in 8 different LMICs was conducted. Over the course of 24 months, monthly non-diarrheal stool samples were analyzed for subclinical Shigella infections through quantitative PCR methods. EE was reflected by elevated concentrations of 3 fecal biomarkers: myeloperoxidase (MPO), neopterin (NEO), and alpha-1-antitrypsin (AAT). MPO concentrations were found to be significantly higher by 0.30 ln(nm/mL) (95% CI: 0.23, 0.37) in the initial month of Shigella detection among stools with subclinical Shigella infections. After the Shigella infection, MPO concentrations declined throughout the following 6 months, and concentrations were lower by 6 months post-infection [MPO 6-month difference: -0.16 ln(nm/mL) (95% CI: -0.26, -0.04)]. Subclinical Shigella infections had no effect on NEO concentration levels within the initial month of Shigella detection but did decrease post-infection. Subclinical Shigella infections had no effect on AAT concentration levels until 6 months post-infection [AAT difference: -0.13 ln(mg/g) (95% CI: -0.24, -0.03)]. These findings did not differ by antibiotic use around time of index infection. The impact of Shigella on biomarkers of EE was not sustained, suggesting the negative association between Shigella and growth could be explained by the accumulation of time-limited rather than persistent effects on inflammation.

We evaluated whether the impacts of Shigella on linear growth among children in low-resource settings were reflected in sustained effects of Shigella infections on biomarkers of environmental enteropathy. The effect of Shigella on environmental enteropathy biomarkers was in fact transient, with myeloperoxidase concentrations returning to baseline within one month following the index Shigella infection. Therefore, myeloperoxidase and other EE biomarkers unlikely to be good markers of the sustained impact of Shigella on EE and growth.

Figures

Association of Shigella infections with EE biomarker concentration over time.Each plot shows EE biomarker natural log concentration differences and 95% confidence intervals comparing non-diarrheal stool samples with and without Shigella detection at month 0. Concentrations were measured monthly from the index Shigella-positive detection through 6-months post-detection.

Associations between Shigella infections and EE biomarker concentrations from time of index subclinical Shigella detection through 6-months post-detection, stratified by antibiotic use within the 15 days before or after Shigella infection.Each panel shows the respective EE biomarker natural log concentration differences comparing non-diarrheal stool samples with and without Shigella detection at month 0. Concentrations were measured monthly from the index Shigella-positive detection through 6-months post-detection. The black line represents no antibiotic use, and the red line represents antibiotic use within the 15-days before or after month 0.

Full text

Introduction

Shigella is an infectious bacterial pathogen which causes shigellosis, a high burden diarrheal disease that disproportionately affects children under 5 years living in low- and middle-income countries (LMICs) [1–3]. Shigella infection is generally thought to be defined by diarrhea and dysentery, but an analysis of the Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health (MAL-ED) study found that, among children under 2 years, only 15% of Shigella-attributed diarrheal stools were accompanied by dysentery and 64% of children that tested positive for Shigella never had a Shigella-attributed diarrheal episode [4,5]. Subclinical infections, despite having less acute clinical manifestations, have been associated with poor long-term outcomes comparable to clinical infections such as sustained impaired child growth throughout the first 2–5 years of life [6].

While the mechanism for how Shigella and other enteric infections could lead to these longer-term adverse effects is unknown, one proposed theory is through environmental enteropathy (EE) [7–9]. EE is a condition of the intestines characterized by inflammation, greater intestinal permeability, and shortening of villi from prolonged pathogen infections which results in malabsorption of nutrients [9,10]. While EE in the short-term is generally asymptomatic, the longer-term impacts have been theorized to include growth stunting, poor cognitive development, and poor vaccine effectiveness in children [9]. Shigella infections cause inflammation of the intestines which is theorized to lead to both the clinical symptoms of diarrhea and dysentery and the long-term impact on the intestines through EE [11].

Diagnosing EE is difficult and expensive as it requires invasive intestinal biopsies since physical symptoms can be general and sometimes not apparent [10,12]. Alternatively, there are several fecal biomarkers that capture multiple aspects of gut health that are commonly used in EE analyses [12]. Myeloperoxidase (MPO) reflects inflammation from neutrophil activity. Unlike lactoferrin and calprotectin, MPO secretion in the stool is not affected by whether the child is breastfed such that variability in MPO may be more reflective of intestinal disease activity [8,12–14]. Neopterin (NEO) reflects inflammation from TH1 immune activity, and NEO is resistant to being broken down in stool [8,12,13,15]. Alpha-1-antitrypsin (AAT) reflects intestinal permeability, is a clear indicator of protein loss, and is also resistant to being broken down in stool [8,12,13,16,17]. These biomarkers, among others, provide opportunities to non-invasively examine intestinal function and health.

Studies in LMICs have shown consistent findings of above average fecal biomarker levels that have been associated with poor linear growth [7,8,18]. Furthermore, both clinical and subclinical Shigella infections have been associated with elevated fecal biomarker concentrations [6–8,11,16,19]. Specifically, Shigella has been associated with significant fecal MPO elevations and a dose response relation of MPO increasing by 0.21 logs for every log of Shigella quantity detected [4,8]. While the evidence is consistent, one common limitation in these studies stems from the cross-sectional nature of the association between Shigella and biomarkers. For Shigella to affect longer-term outcomes like growth, we hypothesize the effects of Shigella on inflammation and EE would have to persist beyond the initial Shigella infection. This study evaluates the longitudinal effects of subclinical Shigella infections on fecal biomarkers of enteropathy to interrogate the potential mechanism for Shigella to impact growth through EE. Furthermore, even though Shigella is often treated with antibiotics, little is known on how antibiotics aid in the recovery of intestines from EE [20]. The second aim of this study is to provide insight about whether the effects of Shigella on EE differ depending on recent antibiotic therapy.

Methods

Ethics statement

The MAL-ED study collected data on and from human participants under the age of 2 along with parental consent. The study design was approved by the University of Virginia’s institutional review board. Approval was given within each of the 8 sites. The Bangladesh site had approval from the Ethical Review Committee, International Centre for Diarrhoeal Disease Research, Bangladesh. The Brazil site had approval from the Committee for Ethics in Research, Universidade Federal do Ceara, and the National Ethical Research Committee, Health Ministry, Council of National Health. The India site had approval from the Institutional Review Board, Christian Medical College, Vellore, and the Health Ministry Screening Committee, Indian Council of Medical Research. The Nepal site had approval from the Institutional Review Board, Institute of Medicine, Tribhuvan University, the Ethical Review Board, Nepal Health Research Council, and the Institutional Review Board, Walter Reed Army Institute of Research. The Peru site had approval from the Institutional Review Board, Johns Hopkins University, and the PRISMA Ethics Committee, Health Ministry, Loreto. The Pakistan site had approval from the Ethical Review Committee, Aga Khan University. The South African site had approval from the Health, Safety and Research Ethics Committee, University of Venda, and the Department of Health and Social Development, Limpopo Provincial Government. The Tanzania site had approval from the Medical Research Coordinating Committee, National Institute for Medical Research, and the Chief Medical Officer, Ministry of Health and Social Welfare. This study’s secondary analysis did not require approval from Emory University’s institutional review board, as it did not fall under human subjects research since the data from the primary MAL-ED study is public and deidentified.

Study population

We performed a secondary analysis of the MAL-ED prospective birth cohort study. From November 2009 through February 2012, infants were enrolled within 17 days of birth and followed for 24 months. There was a focus on resource-limited settings within LMICs, and at least 200 children were enrolled in each of the following 8 sites: Dhaka, Bangladesh; Fortaleza, Brazil; Vellore, India; Bhaktapur, Nepal; Loreto, Peru; Naushahro Feroze, Pakistan; Venda, South Africa; and Haydom, Tanzania. Healthy infants were eligible for enrollment if their caregiver reported they planned to live within the study area for the next 6 months and approved of twice weekly home visits for 24 months. Infants were excluded from study enrollment if they were hospitalized for anything besides a healthy delivery, were diagnosed by a medical doctor for a severe or chronic condition, were diagnosed with enteropathies, or weighed less than 1,500 grams at birth. Infants were also excluded if their family anticipated living outside the study area for more than 30 consecutive days within the first 6 months, if the infant was not a singleton, or if the mother was younger than 16 years of age or had another child enrolled in the study [21].

Data collection

Field MAL-ED researchers visited households twice weekly to collect basic health and dietary information as well as to complete disease surveillance measures. Sociodemographic information was collected at study enrollment, and updated every 6 months [21]. After 9 months, information on diet and breastfeeding was collected monthly rather than during the twice weekly visits [22]. Enteric pathogen infections and EE biomarker concentrations were measured in non-diarrheal stool samples collected monthly. Non-diarrheal samples were taken at least 3 days before or 3 days after a maternal reported diarrhea episode which represents a subclinical infection or post-diarrheal shedding [23]. These samples went under quantitative polymerase chain reaction (qPCR) as previously described for identification of 29 pathogens, including the following which were identified as the most prevalent pathogens in MAL-ED: adenovirus 40/41, astrovirus, Campylobacter, Cryptosporidium, Entercytozoon bieneusi, enteroaggregative Escherichia coli (EAEC), enterotoxigenic Escherichia coli (ETEC), typical enteropathogenic Escherichia coli (tEPEC), atypical enteropathogenic Escherichia coli (aEPEC), Giardia, norovirus, sapovirus, and Shigella [6]. To measure the EE biomarkers, quantitative ELISAs were run for MPO, NEO, and AAT on the non-diarrheal stool samples collected on months 1–12, 15, 18, 21, and 24 [21]. Data on recent antibiotic use of any drug class, cephalosporins, macrolides, and fluoroquinolones were collected during the twice weekly household questionnaires [23].

Statistical analysis

Subclinical Shigella infections were defined in non-diarrheal stool samples (i.e., in the absence of symptoms) by qPCR detection of the ipaH gene at a quantitative cycle threshold of less than 35. This could include subclinical infections caused solely by Shigella or by Shigella and another pathogen. We estimated the effects of Shigella infections on each EE biomarker at the time of the index Shigella infection through 6-months post-infection using multivariable linear regression. Specifically, for each non-diarrheal stool, we estimated the association of Shigella infection compared to no Shigella infection with MPO, NEO, and AAT concentrations respectively during the same (index) month and for each following month up to 6-months post-infection in separate models. Children without Shigella were frequently infected with other bacterial, viral, and/or parasitic pathogens as shown in Table 1. Each model was adjusted for country of residence, age at the time of the infection, sex, whether the child had been exclusively breastfed up until time of the infection, stool consistency, previous Shigella infection, and infections with other enteric pathogens. Recent previous Shigella infection was defined as Shigella detected in another stool sample from the prior 3 months. Infections with other enteric pathogens (coinfections) at the time of the index Shigella infection were included in the models if the pathogen was independently associated with the EE biomarker outcome. To determine which pathogens were associated with each EE biomarker, multivariable linear regression models were run to estimate the association between detection of each pathogen (defined by detection at a cycle threshold <35) with concentrations in the same stool of MPO, NEO, and AAT respectively while adjusting for country site and age. The pathogens statistically significantly (p < 0.05) associated with MPO were Campylobacter, Giardia, norovirus, sapovirus, EAEC, ETEC, tEPEC, and aEPEC. The pathogens statistically significantly (p < 0.05) associated with NEO were Giardia, norovirus, EAEC, and tEPEC. The pathogens statistically significantly (p < 0.05) associated with AAT were adenovirus, Giardia, norovirus, EAEC, ETEC, and tEPEC. The pathogens that were statistically significantly (p < 0.05) associated with the biomarker were included as covariates in the models for that biomarker.

Table 1: Characteristics of non-diarrheal stool samples from 1715 children in MAL-ED study.

In sensitivity analyses, we detrended biomarker concentrations by age to account for naturally higher concentrations in younger children, additionally adjusted the models for coinfections at the time of the biomarker outcome measurement, and further adjusted the models for Shigella detection at the time of the biomarker outcome measurement. Another sensitivity analysis subset to Shigella infections within the first 18 months of life only to allow for all included infections to have a complete 6 months of follow-up samples. A final sensitivity analysis limited the comparison of Shigella infections to viral infections only (i.e., bacterial and parasitic co-infections were excluded).

To determine if recent antibiotic treatment impacted the effect of Shigella on the concentrations of MPO, NEO, and AAT, the same linear regression models were run with an interaction term between Shigella infection and antibiotic use 15 days before or after the index Shigella infection. The different classes of antibiotics examined were cephalosporins, macrolides, and fluoroquinolones, and the last set of models looked at any antibiotic class used.

Results

EE biomarker concentrations through 6 months after Shigella infection

Over the course of 2 years, 1,715 children in the MAL-ED study contributed 34,654 non-diarrheal stool samples. Of these, 10% (N = 3,505 samples) tested positive for Shigella. Details of demographic characteristics associated with stool samples with and without Shigella detected are in Table 1. Biomarker concentrations were similar for infections that were solely caused by Shigella and infections that were caused by Shigella and at least one other pathogen (S2 Table), and there were no major differences between study sites (S3 Table).

In stools with subclinical Shigella infections, the inflammatory EE biomarker, MPO, was elevated by 0.30 ln(ng/mL) (95% CI: 0.23, 0.37) compared to stools without Shigella during the initial month of Shigella detection (Figs 1 and S1 and S1 Table). In the stools 1–4 months following the identification of the subclinical infection, the concentration of MPO was no different compared to those following stools without Shigella. After 5–6 months, MPO concentration levels were 0.18 ln(ng/mL) (95% CI: 0.00, 0.35) and 0.16 ln(ng/mL) (95% CI: 0.04, 0.26) lower following a Shigella infection compared to samples following no Shigella infection. There was no clear difference in NEO concentrations, the EE biomarker characterizing gut immunity, in the initial detection month between those with and without subclinical Shigella infections [NEO difference: 0.01 ln(nmol/L) (95% CI: -0.05, 0.07)]. The subsequent 6 months showed that stools following subclinical Shigella infections had lower concentrations of NEO compared to those not following an infection (Figs 1 and S1 and S1 Table). Similarly, in the initial month of detecting subclinical Shigella infections, there was no clear difference in concentrations of the EE biomarker AAT, reflecting intestinal permeability [AAT difference: -0.02 ln(mg/g) (95% CI: -0.08, 0.04)]. AAT concentrations were lower among stools following a subclinical Shigella infection, particularly after 6 months [AAT difference: -0.13 ln(mg/g) (95% CI: -0.24, -0.03)], compared to stools 6 months after no Shigella infection (Figs 1 and S1 and S1 Table).

Association of Shigella infections with EE biomarker concentration over time.Each plot shows EE biomarker natural log concentration differences and 95% confidence intervals comparing non-diarrheal stool samples with and without Shigella detection at month 0. Concentrations were measured monthly from the index Shigella-positive detection through 6-months post-detection.

In sensitivity analyses, the associations between Shigella infection and MPO biomarker concentrations showed little difference when adjusting for age, alternatively when detrending the biomarker concentrations for age, or by doing both. The results were also very similar when additionally adjusting for other infections at the time of the biomarker measurement, including subsequent Shigella infections (S2 Fig). Results were similar when restricting the analysis to stools collected at 18 months of age or younger (S3 Fig). Finally, results were similar when limiting comparison stools (i.e., those without Shigella) to stools with only viruses detected (S4 Fig).

Modification by antibiotic treatment

There were no consistent differences in the impact of Shigella on MPO, NEO, or AAT concentrations between children who were and were not treated with a cephalosporin, macrolide, fluoroquinolone, or any antibiotic in general with a 15-day range of the index Shigella infection (Fig 2). Children who took macrolides within the 15 days before or after the Shigella infection had lower MPO concentrations starting around 4 months after the infection compared to children without a Shigella infection who took macrolides. Specifically, MPO was significantly lower with Shigella among children with macrolide antibiotic use 15 days before or after infection at 5-months post-infection only. In contrast, while estimates were often imprecise, NEO concentrations were more elevated after Shigella infections when antibiotics were used. The impact of antibiotic use on Shigella’s impact on AAT concentrations was inconsistent over time.

Associations between Shigella infections and EE biomarker concentrations from time of index subclinical Shigella detection through 6-months post-detection, stratified by antibiotic use within the 15 days before or after Shigella infection.Each panel shows the respective EE biomarker natural log concentration differences comparing non-diarrheal stool samples with and without Shigella detection at month 0. Concentrations were measured monthly from the index Shigella-positive detection through 6-months post-detection. The black line represents no antibiotic use, and the red line represents antibiotic use within the 15-days before or after month 0.

Discussion

In the MAL-ED study, subclinical Shigella infections were associated with elevated MPO concentrations at the time of index infection, but concentrations returned to baseline by one month after the infection, showing no evidence of sustained impact on inflammation. Subclinical Shigella infections were not associated with elevated NEO or AAT concentrations within the initial month of infection nor the following 6 months. There was limited and inconsistent evidence that antibiotic use around time of infection modified the impact of Shigella. By 6-months post-infection, all 3 biomarker concentrations tended to be lower following stools with an index Shigella infection compared to those following no index infection. These associations were not explained by subsequent infections and may be consistent with Shigella’s capacity for immune modulation. Although acute Shigella infections can initially cause inflammation, in the convalescent phase Shigella could lead to chronic immune suppression that induces longer-term changes in mucosal immunity [24]. A better understanding of the biological mechanisms for the observed biomarker dynamics and whether such longer-term changes in immunity could contribute to the mechanism for the effect of Shigella on growth faltering is needed.

Despite the large sample size, the distributions of biomarker concentrations were variable within sites. Environmental enteropathy is multifactorial and diverse risk factors related to the environment and host susceptibility likely are important determinants of the observed biomarkers. We adjusted all models for site to account for the complex context-related factors that may have differed at each site and further adjusted for confounding variables of the Shigella-biomarker associations.

Elevated concentrations of MPO, NEO, and AAT have been associated with risk for growth stunting among young children [12]. Children naturally have higher levels of these biomarkers compared to adults, with concentrations decreasing with age. Even so, compared to high-income countries, children living in LMICs have further elevated levels [8,18]. The observed effects of Shigella infection on EE fecal biomarkers during the initial month of infection were consistent with prior research: MPO concentrations were elevated at time of infection and there was not a clear difference in NEO and AAT concentrations compared to those not infected [4,8,19]. However, the fact that MPO was only transiently elevated at the time of the index infection indicates that MPO is unlikely to be a good marker of the sustained impact of Shigella on EE and growth. In contrast, the kinetics of MPO following Shigella infections could make it a valuable marker of invasive bacterial enteric infection when children present to care [28].

Previous studies have found that antibiotic treatment can mitigate the impact of diarrhea, and specifically Shigella diarrhea, on growth [25,26]. Our findings provide evidence that the biomarker concentrations differ only slightly, if at all, between Shigella infections treated with fluoroquinolones, cephalosporins, and macrolides and those that were untreated. Neither specific information on what type or generation of antibiotics nor prescribed course of antibiotic were available in the primary study. Without this information, we are unable to verify if the correct dosage and treatment for Shigella infections were provided to participants. If participants were prescribed antibiotics that are outside the spectrum for treating Shigella, were given the wrong dosage, or did not follow the prescribed course, the antibiotics would not effectively treat Shigella, which could explain the lack of impact of antibiotic use. Also, because MPO concentrations returned to baseline one month following the index infection, we were unable to observe shorter term effects, for example whether antibiotics could hasten recovery from inflammation within the first weeks following infection.

The MAL-ED study design allowed for the unique longitudinal analysis of the effect of subclinical Shigella infections on EE fecal biomarker concentrations among children across 8 global sites, while accounting for age, breastmilk consumption, and other enteric infections which have been associated with inflammatory biomarkers [18]. This study was limited by not all non-diarrheal stool samples being tested for biomarker concentrations. Monthly samples in the second year of life were tested quarterly for biomarker levels which gave the infections in the first year of life a heavier weight in the analysis. Furthermore, because the biomarkers were not measured in diarrheal stools, we were unable to examine the potential sustained impact of Shigella diarrhea.There is lack of literature on healthy biomarker levels among children in both high-income countries and LMICs. While the levels between those with and without Shigella infections can be compared, it is not clear which concentration levels contribute to EE and/or growth impairment. Finally, the high frequency of coinfections among children in these populations may have obscured clear differences in biomarker concentrations between children with and without Shigella.

Many studies have investigated which enteric pathogens contribute to EE, and Shigella has consistently been found to be a high burden pathogen in LMICs with long-term effects on EE related to both symptomatic and asymptomatic infections [1,3]. This study concludes that if Shigella’s long-term effects are brought upon through a sustained impact on EE, these effects are not captured by sustained elevation of the biomarkers studied here. In fact, mechanisms for the more complicated long-term dynamics of these biomarkers need to be studied, perhaps in conjunction with biopsy studies that can better characterize the features of EE [27]. Stronger associations between enteric pathogens and EE biomarkers may arise when multiple pathogens are considered given many children in these regions experience co-infections, which may have a bigger impact on the long-term health of those infected. A better understanding of how Shigella affects EE and of the mechanisms for the long-term effects of Shigella on child growth and development is needed.

Supporting information

S1 TableMonthly EE biomarker concentration associations from time of index subclinical Shigella detection through 6-months post-detection.(PDF)

S2 TableBiomarker concentrations among Shigella infections and Shigella infections with coinfections.(PDF)

S3 TableBiomarker concentrations among the 8 MAL-ED study sites.(PDF)

S1 FigAssociation of Shigella infections with raw EE biomarker concentrations values over time.Each plot shows EE biomarker natural log concentration differences and 95% confidence intervals comparing non-diarrheal stool samples with and without Shigella detection at month 0. Concentrations were measured monthly from the index Shigella-positive detection through 6-months post-detection.(PDF)

S2 FigLongitudinal impact of Shigella infections on MPO concentrations by varying specifications of age and adjustment for subsequent infections.Each plot shows EE biomarker natural log concentration differences and 95% confidence intervals comparing non-diarrheal stool samples with and without Shigella detection at month 0. The dashed black line represents the modeling while adjusting for age which completely overlaps with the dashed orange line that represents the model adjusting for age and detrending for age. The green line is the model only detrending for age. The blue line is the model adjusting for age and other infections at the time of the biomarker measurement (i.e., subsequent infections). The purple line is the model adjusting for age, other infections at the time of the biomarker measurement (i.e., subsequent infections), and Shigella infections at the time of the biomarker measurement (i.e., subsequent Shigella infections).(PDF)

S3 FigLongitudinal impact of Shigella infections on biomarker concentrations among children 1–24 months of age and among children 1–18 months of age.Each plot shows EE biomarker natural log concentration differences and 95% confidence intervals comparing non-diarrheal stool samples with and without Shigella detection at month 0. The association of subclinical Shigella infection on fecal biomarkers among children younger the age of 24 months is represented by the black line. The association of subclinical Shigella infection on fecal biomarkers among children younger than the age of 18 months is represented by the dashed blue line.(PDF)

S4 FigLongitudinal impact of Shigella infections on biomarker concentrations compared with a control group consisting of stools with no Shigella infections and co-infections and compared with a control group consisting of no Shigella infections and only viral co-infections.Each plot shows EE biomarker natural log concentration differences and 95% confidence intervals comparing non-diarrheal stool samples with and without Shigella detection at month 0. The association of subclinical Shigella infection on fecal biomarkers among children with any co-infection is represented by the black line. The association of subclinical Shigella infection on fecal biomarkers among children younger with viral only co-infections (no bacterial or parasitic co-infections) is represented by the dashed green line.(PDF)