Environmental Enteropathy

EE is thought (by most authors) to be the result of chronic exposure to enteropathogens, although the potentially synergistic and disruptive role of poor nutrition is increasingly being recognized. In addition to the histopathologic findings of villous blunting, the main components of EE are increased intestinal permeability (from impaired barrier function), mucosal inflammation, malabsorption, and systemic inflammation. The condition is considered distinct from tropical sprue and overt symptoms of diarrhea. Because it depends on exposure to unsanitary environments, it is common in both long-term residents and in travelers, and is reversible once the environment improves, EE is thought to be environmentally derived and likely widespread in low-resource settings.

However, the determination of a clear consensus definition for EE remains an elusive challenge. Because EE is without overt acute symptoms (although it may manifest in subacute weight loss and impaired growth or development over longer periods of time), the traditional gold standard for diagnosis has been intestinal biopsy to identify abnormalities in intestinal histology. Such an invasive diagnostic is infeasible in most research and many clinical settings and is also limited by potentially inadequate sampling, because the biopsied sample may not be representative of the whole intestine. However, the diagnosis of EE in the absence of these invasive procedures has proved challenging.

In response to the inability to regularly perform biopsies in healthy, “asymptomatic” individuals, recent research on EE has aimed to identify biomarkers to characterize EE that can be measured in stool, blood, or urine. At least 40 different biomarkers or metabolites have been investigated as potential indicators for EE that is clinically significant enough to result in growth faltering ( Table 1 ). These include markers of disrupted intestinal barrier or absorptive function (eg, lactulose and mannitol, rhamnose, or d -xylose absorption and excretion in the urine, alpha-1-antitrypsin in stool, tight junction components in plasma or intestinal tissue staining); translocation of microbes or their products (eg, lipopolysaccharide or anti-lipopolysaccharide antibody); intestinal inflammation (eg, myeloperoxidase, lactoferrin, calprotectin, or lipocalin in the stool); and systemic inflammation (eg, high sensitivity C-reactive protein or acid glycoprotein; serum amyloid A and other acute phase proteins). Other indicators include metabolites such as citrulline or tryptophan that may signal a healthy intestinal mucosa.

The biomarkers for these processes have varying specificity for EE-induced growth faltering. Lactulose absorption and excretion, like fecal alpha-1-antitripsin and plasma lipopolysaccharide markers, all reflect disrupted intestinal barrier dysfunction, which is a proximal component of EE. Conversely, markers of systemic inflammation have diverse causes and cannot specifically identify EE. Of course, measures of growth impairment alone, such as height-for-age z -score (HAZ) or stunting, have been included as markers in some studies, but they have poor specificity for EE because they may be due to nonintestinal causes. Like later impairments in cognitive development (especially in higher executive function or semantic fluency ), these measures are indicators of outcomes of EE and may occur too late in the disease process to be of timely diagnostic use to enable interventions. Tracking of growth trajectories may be useful to identify early decrements that could predict continuing growth deficits if EE conditions persist.

The identification of appropriate biomarkers for EE is challenged by their frequent comparison with nonspecific outcomes like linear growth rather than a true gold standard diagnostic. As noted, stunting is hypothesized to be an effect of EE and is highly multifactorial. The strongest predictor of stunting is birth size, which is necessarily unrelated to the development of EE in early life in the child (although likely relevant to EE in the mother). Therefore, single anthropometric measurements have poor sensitivity and specificity as a standard against which to validate biomarkers. However, growth trajectories, especially over the first 2 years of life (assessed as incremental changes in HAZ scores) may better reflect EE in early childhood. Validation against intestinal biopsy might be more specific, but is infeasible in many settings, as discussed.

In addition, the potential ubiquity of EE in low-resource settings makes it difficult to identify appropriate comparison groups. Because EE is acutely asymptomatic, it is unclear how to define “healthy” controls for a study in an environment conducive to EE and where the condition is expected to be common. Unlike the World Health Organization growth standards, for example, there are no international reference standards for EE biomarkers based on a representative group of children from diverse areas. Distributions of EE biomarkers in healthy children in high-resource settings are also largely not available to be used as reference standards. Even if this information were available, external reference populations could be inappropriate because they would differ on many other characteristics that would confound comparisons. In multiple papers assessing EE among children in low-resource settings, the authors referenced comparison biomarker values based on small studies in healthy adults that were not assessing EE directly. Clearly these reference values were suboptimal.

Beyond the uncertainties with biomarkers, the scientific community has not come to a consensus on what type of definitions are relevant for understanding EE. If EE is to be considered a disease, it will require a specific pathologic definition. Conversely, as a syndrome, EE could encompass multiple disease states. Arbitrary assemblages of morphologic or functional pathologies can have important acute or lasting consequences despite having multiple potential etiologies. Examples include pneumonia and diarrhea that are well-recognized as “disease” entities, even though they are composed of many component, more specific diagnoses, such as pneumococcal pneumonia or shigellosis. EE could be considered similarly as a set of functional pathophysiologic alterations, with varying degrees of morphologic pathology caused by diverse environmental determinants. However, distinguishing EE from similar pathologies as human immunodeficiency virus-associated enteropathy, for example, may be difficult. It is unclear whether and how similar presentations of enteropathy should be considered distinct when the conditions can be indistinguishable except for the underlying cause.

What needs to drive these definitions is the goal to identify EE as an entity for which its recognition and the development of effective interventions can improve long-term health outcomes that may range from growth to cognitive and even later life metabolic impairment. Both surveillance and clinical case definitions will be needed for future study of EE. A surveillance definition would be used to identify sentinel cases to inform population-level interventions, which would be relevant because EE seems to be widespread in low-resource settings. Because populations suffering from EE show a population shift in biomarker distributions, it may be difficult to make individual diagnoses, and population-level interventions may be most appropriate. In contrast, a clinical case definition will be needed for targeted treatment and other individual-level interventions.

Geographic differences in EE manifestations may also pose challenges. Variations in environmental exposures across low-resource settings likely result in different pathologies. Perhaps it is more important to describe an EE spectrum, where milder cases may not include systemic inflammation, for example, and more severe cases may be associated with villous blunting and poor growth outcomes. It may be necessary to distinguish between consequential EE, with observable poor outcomes, and inconsequential EE, in which a child may show abnormal histology or biomarker levels, but no related outcomes such as poor growth. In this conception, EE constitutes a risk factor for health outcomes, such as a significant decrease in HAZ in the formative early years of childhood or an impairment of normal cognitive development, especially in higher executive function. EE describes a state that is probabilistically associated with poor development (eg, growth, vaccine response, cognitive impairment), but is not a necessary or sufficient cause and may lead to no observable clinical outcomes in a significant proportion of cases.

As the field is quickly evolving, new biomarkers are being identified and refined to help enhance our understanding, if not the definition, of EE. In spite of the complex challenges in defining EE, improved clarity of this entity will help to develop appropriate surveillance and clinical case definitions against which to study risk factors, which will be vital to advancing our understanding and to testing, and ultimately investing in, potentially effective interventions.

Specific Enteric Infections

Overt diarrhea (defined as 3 or more unformed stools per day ), especially when prolonged or persistent, has predicted growth and even cognitive failure in many previous studies. However, reduced diarrhea rates observed in more recent studies have made overt diarrhea less useful in predicting growth and developmental outcomes. Despite these reductions, “subclinical” pathogen detection in stools continues to be common in many impoverished settings and has been associated with poor outcomes.

These infections are hypothesized to be a key contributor to EE, and asymptomatic carriage of known enteric pathogens may cause or aggravate linear growth faltering, even in the absence of recognized episodes of diarrhea. For example, asymptomatic excretion of enteroaggregative E. coli (EAEC) has been associated with linear growth faltering. The mechanism for the growth effect may be through subclinical gut inflammation, which has been associated with EAEC detection in MAL-ED (Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development Project), a birth cohort study performed at 8 sites in South America, sub-Saharan Africa, and South Asia (Rogawski and colleagues, manuscript in preparation). Carriage of Campylobacter spp. has also been associated with both linear and ponderal growth faltering. This association may be mediated through EE; Campylobacter detection was associated with increased markers of permeability (alpha-1-antitrypsin), intestinal inflammation (myeloperoxidase), and systemic inflammation (acid glycoprotein) in MAL-ED. Similarly, persistent Giardia detection in the first 6 months of life has been associated with malabsorption and reduced linear growth in two studies. Cryptosporidium parvum excretion has also been associated with growth faltering, although it is unclear if this association is driven by prolonged excretion after a symptomatic infection. These data suggest that a primary mechanism for the pathogenesis of EE may involve exposure to pathogens through poor hygiene practices and contaminated food and water, resulting in enteric infections.

Because these infections are commonly diagnosed in stool by culture or molecular methods, it is challenging to distinguish between colonization and infection. Many microorganisms can play both the roles of commensal and pathogenic organism and mere detection does not illuminate the multifactorial impact of the organism. Some organisms, like Clostridium difficile , Salmonella , and likely many other enteric bacterial pathogens, may change from commensal to pathogenic within an individual over time, and pathogenicity is highly dependent on the presence of other microorganisms and a wide range of environmental and genetically determined host factors. Further, the distinction between commensal and pathogenic is often made among organisms within the same species, as is the case for E coli . Pathogenic E coli are generally identified by an array of virulence factors, the significance of some of which have yet to be fully elucidated.

The identification, quantification, and attribution of pathogenic versus nonpathogenic organisms is complicated in endemic settings by differences in host resistance, normal microbiota, and acquired immunity. Organisms that are known causes of diarrhea outbreaks, such as waterborne Giardia outbreaks, are often not associated with diarrhea among children in endemic settings. Giardia , for example, was more commonly identified in nondiarrheal stools than diarrheal stools in both major studies of diarrhea etiology, MAL-ED and GEMS (Global Enteric Multicenter Study). Even organisms that are statistically associated with diarrhea in these studies are commonly found in nondiarrheal stools and “healthy” controls. New methods to use quantitative polymerase chain reaction (PCR) to incorporate quantity of pathogen have helped to distinguish cases from controls, but they remain imperfect and quantities significantly associated with case status necessarily only apply at the population level. Detections above these quantities cannot conclusively identify etiology in individual cases. In contrast, detections of these pathogens are rare in high-resource settings, which makes it easier to confidently assign etiology.

Pathogenic organisms must be considered in the context of the gut microbiota, a highly complex “organ” of the body which impacts myriad processes from metabolism to cognition. However, microbiota-related studies rarely consider the presence of specific pathogenic organisms, and it is unknown what bilateral impact they may have. Although the microbiota has been shown to be altered in many diseases states, it is unclear which compositions constitute beneficial microbiota and which represent dysbiosis. Diversity parameters for the microbiota (either alpha diversity within a sample or beta diversity between samples) can simplify the highly dimensional composition data, but high diversity is not uniformly positive. For example, while exclusive breastfeeding is recommended in early infancy, the predominant bacteria in the microbiota of breastfed infants are Bifidobacterium and Lactobacillus , whereas infants fed with formula milk have a more complex microbiota. Recently, measures of microbiota “maturity” have been developed to describe how malnourished children often have microbiota compositions that mirror the compositions of younger, healthy children. However, the organisms identified as discriminating for immaturity differ across populations, limiting the generalizability of the index. These discriminating organisms are not those classically considered as enteric pathogens, and it is unclear if they are harmful themselves or simply indicators for generalized dysbiosis.

Animal models can help to separate and elucidate the roles of specific pathogen infections versus the microbiota by experimentally controlling infections and the microbiota in specific diet and host contexts. For example, protozoal, bacterial, and viral infections in murine models have strikingly different effects in the context of specific nutrient deficiencies with normal microbiota. Protein deficiency greatly enhanced both the intensity of cryptosporidial infection as well as its impact on growth, documenting a bidirectional impact of worsened infection with protein malnutrition and conversely worsened growth with infection. Cryptosporidial infection in the setting of protein deficiency also caused impaired turnover of infected epithelial cells. Similar synergies with protein or zinc deficiency were seen with EAEC infections. Conversely, murine rotavirus infections were less severe in undernourished conditions, which may be because the intestinal villi were blunted and less efficiently provided the lactase needed to “uncoat” the virus as part of pathogenesis.

The microbiota also affect susceptibility to infection in mouse models. Early studies from the 1960s and 1970s showed the intestinal flora was antagonistic to Salmonella, Shigella, and Vibrio cholerae infection. Several other recent studies have found that a normal microbiota in mice successfully prevents colonization by Salmonella enterica serovar Typhimurium. Conversely, mice with altered microbiotas owing to antibiotic administration are more susceptible to intestinal infection and disease due to Salmonella and other enterobacteria such as E coli . One study showed a dose–response such that greater alterations to the microbiota led to higher colonization by S enterica serovar Typhimurium, with more severe inflammation and intestinal pathology. Further, modification of the microbiota through the antibiotic treatment of mice increased susceptibility to infection by vancomycin-resistant Enterococcus and C difficile.

In terms of the microbiota and susceptibility to viral infections, there are examples both where intestinal bacteria promote and are antagonistic to viral infection. For example, Bacteroides thetaiotaomicron and Lactobacillus casei have been shown to prevent infection of the intestinal epithelial cells by rotavirus in vitro. Similarly, mice with depleted microbiotas through antibiotic treatment or development in germ-free conditions are more susceptible to influenza compared with normal mice. On the other hand, the gastrointestinal microbiota has been shown to enhance replication and infection of other viruses. Antibiotic-treated mice were less susceptible to poliovirus compared with mice with normal microbiota, resulting in a mortality rate among normal mice twice that among antibiotic-treated mice. A similar study demonstrated that mouse mammary tumor virus, a retrovirus, was more efficiently transmitted in the presence of a rich microbiota, and correspondingly virus transmission to offspring was reduced in antibiotic-treated mice and germ-free mice.

Clinical studies have supported laboratory based evidence of the importance of the microbiota. Prior antibiotic treatment has been associated with increased susceptibility to E coli , Salmonella , Shigella , and Campylobacter infections and with longer duration of infection compared with patients who did not receive antibiotics. Antibiotic treatment also reduces the inoculum required to cause infection with Salmonella. The clear association between the microbiota and susceptibility to infection has led some researchers to suggest that people with an altered microbiota are functionally immunocompromised and less resilient against new and opportunistic pathogens and recurrent infections. The complex interactions in the gut between enteropathogens and the microbiota likely play a key role in development of EE.