Safety and risk assessment for the human superorganism

ABSTRACT

This editorial on safety evaluation and risk assessment for the human superorganism introduces a series of papers arguing for a fundamental shift in how we approach human health risk assessment. In this series emphasis is placed on the risk of infectious disease. Our 21^stst century understanding of human biology is that, as holobionts, we possess a majority of microbial cells and genes. In fact, our microbes fundamentally affect our interactions with the external environment, metabolism, physiology, and risk of both pathology and disease. As holobionts, we require our microbial partners for us to be both complete and healthy. Using the 20^th century understanding of human biology, we failed to capture the effects of the microbiome in evaluating health risks. But 21^st century risk assessments such as those associated with exposure to specific microbial pathogens need to include the human microbiome. Microbiome status will be central in determining the health risks for both individuals and populations.

KEYWORDS:

• microbiome revolution
• colonization resistance
• safety evaluation
• health risk
• paradigm shift

Introduction

Human safety evaluation combined with health risk assessments is utilized to better protect humans from external hazards of all varieties. Whether the threats come from microbial pathogens, environmental chemicals, physical agents (e.g., radiation), drugs, or food additives, the tools for identifying hazards and assessing risk from exposures have been used to identify human exposures that are likely to be problematic to both individuals and populations. Problematic exposures are likely to result in compromised human cells, tissues, organs, and/or physiological systems with an increased risk of pathogenesis and overt clinical disease. Of course, the goal of the safety evaluation is to link that with risk assessment to connect the dots that facilitate the best approximation of true health risks and to inform subsequent regulatory decisions (e.g., in the U.S. EPA, FDA, and USDA). The relationships among connections of these three components are all dependent on the presumptions that both the evaluation and risk assessment models used and the data from those models are highly relevant to human health.

In this editorial article, I discuss why a fundamental misunderstanding of human biology during the 20th century caused significant problems with safety evaluations and risk assessments, and why it is time to bring both in line with the new ideas of humans as a superorganism. I also introduce three papers which, taken together, present a paradigm shift in addressing the risk of human salmonellosis.

The human superorganism

The 20th century was a period where we unintentionally got the representation of humans very wrong. Because we viewed the healthiest humans as generally free of microbes, we designed safety evaluation and risk assessment models that unintentionally excluded or at least ignored a major part of us, the microbiome (Sleator 2010; Ericsson et al. 2015; Dietert 2016a). The most comprehensive data from validated protocols, whether used for chemical, drug, physical agent, or microbial safety, were based on what we now know as a non-existent, biologically incorrect, incomplete form of humans: the human as a single species, purified form of mammal. In failing to account for the effects of the human microbiome in modeling, we failed to analyze and include microbiome information from our test models, or failed to adequately control for what is now recognized as an important variation among microbiomes (Ericsson et al. 2015).

The human form is actually a composite organism, a holobiont, which includes a composite of cells and genomes (eukaryotic and microbial) (Catania et al. 2016; Miller 2016). The human superorganism has trillions of microbial cells and a significant majority of microbial genes guarding the boundary between the human eukaryotic cells and the external environment (Dietert 2016a). A less obvious issue with the superorganism is that of immunological self-tolerance. How does our immune system reconcile the compositional mix of species which is to some extent fluid? Among the thousands of species that reside on and in us, which will be tolerated and which will be rejected? Because of immune pruning of the microbiome, the postnatal immune system preconditioned by the maternal microbiome, diet, and local environment decides which species and strains stay in places like the gut and which will be rejected (Jain and Walker 2015; Kubinak and Round 2016). The immune–microbiome interface can be viewed as a final arbiter of the child's genetic composition and metabolic capacities. It is also a battleground site involving interactions among mutualistic microbes (both microbes and host benefit) and commensal microbes (no harm to the host) on one side versus pathogenic microbes on the other. In fact, because of the location of the human microbiome (at portals of exposure to food, water, pathogenic microbes, environmental toxicants, and drugs) and its critical role in barrier protection, immune regulation, metabolism, and physiological maturation and programming, it is no longer adequate to pursue risk assessments without the consideration of the microbiome.

Our thousands of different microbial species each have their own ecological niche, role, and range of contributions to the whole organism (Gupta et al. 2015; Kopac and Klassen 2016; Perez Perez et al. 2016). These need to be captured in both safety evaluation and risk assessment. The “microbiome revolution” of the 21st century, ushered in by The Human Microbiome Project (Turnbaugh et al. 2007; Blaser 2014), challenges the prior century's thinking about fundamental aspects of human biology, health, genetics, metabolism, environmental interactions, and safety.

A major reason that the microbiome cannot be ignored in safety evaluation and risk assessment is because it operationally connects human internal physiology with our external environment (Krautkramer et al. 2016) and may be the single, most critical factor determining health risks across the lifetime (Dietert and Dietert 2012; Dietert 2014; Dietert and Dietert 2015). Because of the location, the microbiome plays the role of gatekeeper or filter for the human mammal (Dietert and Silbergeld 2016). A part of this role is due to the primary physical locations of the human microbiome (e.g., gut, skin, airways, and urogenital tract) at the portals of exposure to the external environment. Besides the fact that human microbes encounter environmental agents via inhaled, oral, or dermal routes before our own mammalian cells, the microbes also respond to external agents. This impacts the status of the microbiome, which in turn affects the host brain and neurological system (Dinan and Cryan 2016), gut (Imhann et al. 2016), liver (Selwyn et al. 2016), bone (Hernandez et al. 2016), kidney (Pluznick 2016), endocrine organs and tissues (Kunc et al. 2016), reproductive system (Nelson et al. 2016), and immune system and host defense (Levy et al. 2016; Brown and Clark 2017).

In fact, a prior review suggested that the interlinkage of the microbiome, host barrier, and specialized host cells (e.g., gut epithelium) including underlying immune cells is so strong that it is more useful to think of these in both protection and therapy as a systems' biology unit (termed as the microimmunosome) (Dietert 2016b). It is very difficult to completely separate the status of the microbiome from that of the mucin layer, barrier, and epithelial cells, and the underlying immune cell populations. The cross regulation among these makes them function more as a unit at a level where health risks are to be considered. Figure 1 diagrams a systems' biology concept of this unit as it relates specifically to microbial pathogens and risk of pathogen-driven illness. It is important to recognize that a positive fecal sample following a bolus dose of pathogen exposure could reflect an infection likely to produce illness. But it could also result from a pathogen that never breaches the microimmunosome defense barrier (including the microbiome) and is simply in transit through the gut. The paper by Coleman et al. (2017) in this HERA Special Collection provides more details on the relationships among exposure, detection in fecal samples, transit, infection, and pathogen-induced illness.

Figure 1. The diagram illustrates the systems biology concept of the microimmunosome [microbiome, barrier functions (including mucin and eukaryotic cells in the lining), underlying immune system] as it is related to the risk of infection and illness following exposure to microbial pathogens. The microbiome is the front-line factor affecting the opportunity for microbial pathogens to produce disease. It is also the critical, missing piece when the health risk for the human superorganism is considered. Because individual microbes within the microbiome cross regulate each other and the individual components of the microimmunosome also cross regulate, the status of the systems' complex as a whole can be operationally useful from the perspective of health risks. In context of this series of papers, the issue becomes less a question of whether a pathogen is likely to reach the gut and instead focuses on the likely health outcomes once the pathogen encounters the microbiome and the microbe–host complex (the microimmunosome).

Re-envisioning human environmental health risks

Although the significance of mutualistic and commensal microbes was known to some bacteriologists decades ago (Smith 1982), the broader scientific community is only now becoming aware of their importance. Because of the vast array of functions within the microbiome, the safety assessment models for humans need to give priority to the microbiome (Spanogiannopoulos et al. 2016; Li et al. 2016; Wilson and Nicholson 2017). Prior data generated from mammalian-centric models have led us to potentially erroneous or uncertain conclusions about safety standards. In fact, to date, we appear to be in danger of not only persisting in this error but also extending it via the application of new in vitro and nano-technologies. It means that we need to redo safety assessment and connect those efforts to risk assessment that fully models the human superorganism. Support for the need to incorporate the microbiome into any safety evaluation and risk assessment is evident among recent investigations of health risk factors ranging from microbial pathogens (Kampmann et al. 2016), to drugs (Viswanathan 2013), environmental chemicals (Hu et al. 2016), radiation (Cui et al. 2016), and food additives (Chassaing et al. 2015). In these specific cases, it is the microbiome status and/or environmentally induced microbiome alterations that determine the health risk resulting from exposure.

Appropriateness of various in vivo and in vitro models has generally been judged against human development and physiology by contrasting human mammalian system and cells versus those of lab animal mammals, or vertebrates, and/or their cells. One example of this can be seen in the development of immunotoxicology testing protocols between 1980 and present (Vos 1980; Dietert 2010; Giannakou et al. 2016). However, as described by Dietert and Silbergeld (2015), what has been missing from most safety testing models is the consideration of the complete human-mammal plus the human microbiome (i.e., the human superorganism).

If the presence of the microbiome at the body's openings to the external environment can affect the risk of chemical and drug exposures, microbiota of the skin, airways, gastrointestinal and urogenital tracts are an even more direct factor in the health risks for exposure to microbial pathogens. In this special category, it is the microbe versus microbe interaction that occurs before the pathogen even encounters target human eukaryotic cells. This is what needs to be embedded in future microbial risk assessment models.

Beyond that it is important that superorganism-driven risk assessments are fully utilized in the processes leading to regulatory decisions. In a recent editorial, titled “Will there Ever Be a Role for Risk Assessment,” Hope (2009) discusses the fact that risk assessments are inherently based on the complexities of uncertainty and the notion of quantifiable non-zero risk for chemical hazards. Although the non-zero risk may be fully acceptable to both scientists and risk assessors, these complexities are often difficult for non-scientists to explain to the general public. Therefore, regulators will often choose not to fully apply risk assessments in the setting of standards but rather establish “bright lines” deeming that exposures above the line are generally hazardous, whereas exposures before the line are safe (Hope 2009). Obviously, this was an issue at the time of the 2009 editorial and may remain so today. In addition, for microbial hazards, regulators often make decisions based merely on the presence of pathogens in foods. However, as is true for chemical hazards, “the dose makes the poison.” A superorganism-based risk assessment would acknowledge that the growth of ingested Salmonella in the intestinal lumen to doses causing illness is prevented by the gut microbiota in healthy animals (Palmer and Slauch 2017, companion paper in this collection). Superorganism-based risk assessments will only translate into useful policies if the science is fully applied in reaching the final decisions.

Embracing colonization resistance

It has been more than 45 years since Van der Waaij et al. (1971) first applied the term “colonization resistance” in describing the inherent capacity of the host mutualistic and commensal bacteria to afford a natural line of defense against invading pathogens. Yet, we have started to realize the extent to which this microbiological principle impacts health protection, health risk assessment, and therapeutics. Colonization resistance, the capacity of core gut microbiota to resist invasion by harmful microbes, is not only a significant protection against pathogens (Brugiroux et al. 2016) but also a factor that must be considered in health risk assessment (Coleman et al. 2017, this issue). Additionally, the status of colonization resistance is malleable via core microbiota management (Brugiroux et al. 2016). For example, populations of core gut microbiota can be destroyed by antibiotic administration or potentially expanded such as with probiotic supplementation. Successful early strategies to enhance the level of colonization resistance used supplementation with core probiotic bacteria, a process sometimes called competitive exclusion or Nurmi Concept, to first overcome a 1970s outbreak of salmonellosis in poultry and later for other pathogens (Nurmi et al. 1992; Neal-McKinney et al. 2012). Similar strategies have been applied for other sectors of agriculture where useful outcomes can be measured as reduced infections or pathogen burden, improved barrier function, reduced inflammation, and even enhanced nutrient utilization and/or growth (Pollmann et al. 2005; Hai 2015; Angelakis et al. 2016; Wu et al. 2016).

Not surprisingly, the same microbiological principle of colonization resistance extends beyond agriculture to impact human health. In fact, in a recent publication in the journal Nature, researchers called for a broad microbial perspective on healthy human development and emphasized the fact that we need microbes for normal microbial communities as part of a healthy existence (Charbonneau et al. 2016). Microbial exposures during development are part of our necessary physiological/immunological maturation. For example, human breast milk itself is loaded with mutualistic, commensal, and potentially pathogenic microbes such as Streptococcus and Staphylococcus bacteria in addition to prebiotic oligosaccharides (Barile and Rastall 2013) and immune components (Fitzstevens et al. 2016).

Recently, human medicine has adopted several microbiome management strategies aimed at reducing the risk of infectious and non-communicable diseases (Avershina et al. 2016; Hashemi et al. 2016) and, in some cases, treating specific conditions (Staley et al. 2016). Among the microbe-based strategies is the use of probiotics in preterm infants to reduce the risk of necrotizing enterocolitis (Sawh et al. 2016). Metabolomics and meta-proteomics are likely to provide useful guidance on microbe–microbe and microbe–host interactions (Mao and Franke 2015; Vernocchi et al. 2016). The inclusion of the microbiome in human health protection and medical treatment is moving toward a fully envisioned idea of human-engineered biomes (Brüssow 2016).

There is considerable irony that, while the basic microbiological principles of colonization resistance have taken operational root and grown as part of agriculture, food production, and more recently human therapeutics, they have yet to be fully embraced in human environmental safety. On the contrary, we seem to be still enmeshed in the 20th century mantra that emphasizes a fear of microbes coupled with over-sanitation of our built environments and over-pasteurization of our food. We need to realize that microbes in our food are not always a problem. For example, the microbes in fermented foods are useful for improved microbiological and metabolic health (Marco et al. 2016).

A different view of risk–benefit including the microbiome

One can envision what elements of risk reduction for the superorganism might include. For example, the antibiotic destruction of the microbiome has long-term health consequences and would be less likely to occur in the absence of complementary therapies and modified benefit–risk calculation. Besides the long-term concern of antibiotic overuse in the pediatric setting relative to the epidemic of non-communicable diseases and conditions (Bailey et al. 2014; Johnson and Ownby 2017), the antibiotic destruction of the microbiome leads to differential recovery among patients, and this can affect the susceptibility for future infection by opportunistic pathogens. Isaac et al. (2016) found that the oral vancomycin treatment of humans results in severe damage to indigenous microbes, with inconsistent recovery among patients and increased vulnerability to future infections as outcomes. The researchers recommended that the cost–benefit calculations need to include the adverse outcomes involving the microbiome (Isaac et al. 2016). Beccatini et al. (2016) argue that preserving and utilizing beneficial functions of mutualistic microbiota should be a part of human safety, prevention, and treatment strategies when it comes to the risk of infection. A part of this is what has been termed the “live medicine” approach (Pamer 2016).

When the microbiome and colonization resistances are included in risk considerations of potential enteric infections with microbial agents like Salmonella, it is no longer simply an issue of whether a bolus of microbial cells can survive the stomach acid and be present to infect human gastrointestinal tract cells. Multidirectional interactions among Salmonella, the human microbiota, and host cells, in particular the immune cells, are critical in determining whether the pathogen succeeds in gut infection (Gart et al. 2016). Among the indigenous microbes blocking Salmonella success are butyrate-producing Clostridia bacteria, which are often destroyed via antibiotic treatments. While it could be argued that it takes a highly complex and rich gut microbiome to confer colonization resistance against Salmonella, decades of animal-based evidence suggest otherwise. Prebiotics can significantly change the dynamics of risk of infection with Salmonella as reflected in young chickens. Feeding of mannanoligosaccharides and xylooligosaccharides in conjunction with Salmonella challenge changed gut microbiota composition, host cytokine profiles, and colonization resistance against Salmonella infection (Pourabedin et al. 2017). Hence, it is not simply Salmonella exposure in a vacuum with low-level dosing that determines the risk of infection but rather, the total food consideration and existing microbiome that is likely to be more significant.

The technical papers

Three technical papers follow in this four-paper HERA special collection. In the second paper, Mechanistic Modeling of Salmonellosis: Update and Future Directions (Coleman et al. 2017), Margaret E. Coleman, Harry M. Marks, Richard C. Hertzberg, and Michele M. Stephenson provide an update on a mechanistic study of salmonellosis published nearly two decades ago considering expanding knowledge of the human microbiome in health and disease. The authors introduce evidence that is inconsistent with the common assumptions of low-dose linearity and non-threshold mechanisms (independent action or single-hit hypothesis). The validity of these simplifying assumptions for microbial dose–response assessment based on prior science policy is challenged by the recent expansion of “-omics” evidence as the “microbiome revolution” progresses and the innate immune defenses of the human superorganism are better characterized. The updated systematic investigation of the seven events of non-typhoid salmonellosis in humans incorporated recent scientific advances in understanding of the impact of the healthy superorganism and addressed fundamental misunderstandings of human biology and ecological systems. Four of the seven events include previously overlooked, but now well-documented, behaviors of the superorganism:colonization resistance (Event 1); tempering of initial pathogen growth by the extrusion of invaded host cells into the gut lumen (Event 5); competition with microbiota in gut lumen prior to inflammatory damage (Events 5 and 6); and the clearance of the pathogen from the inflamed gut (Event 7). Recent evidence of the synergistic, cooperative behaviors of the superorganism strengthens and extends the biological motivations for sublinear or convex dose–response curves and thresholds for illness in the microbial risk assessment. However, until the quantity and quality of mechanistic data available for tuberculosis and Clostridium difficile enteritis become available for salmonellosis, the consideration of possible empirical model forms (e.g., exhibiting threshold, sublinear, and linear low-dose behavior) is necessary to depict the uncertainty of the “true” dose–response relationship.

In the third paper, Mechanisms of Salmonella Pathogenesis in Animal Models (Palmer and Slauch 2017), Alexander D. Palmer and James M. Slauch present insights into recent advances in the knowledge of animal models for salmonellosis, focusing on studies with the classical mouse model (resembling systemic infection and septic shock consistent with human typhoid fever, not diarrhea) and with the streptomycin-pretreated mouse model (inflammatory diarrhea consistent with human non-typhoid salmonellosis). The streptomycin-pretreated mouse model provides a nuanced understanding of Salmonella intestinal infection with implications for future experimental studies to bridge the gaps in the knowledge for predicting human risks. These authors point out multiple lines of reasoning that challenge a common assumption by risk assessment teams: Salmonella replicate in the intestinal lumen prior to infection and induction of inflammation. After high oral doses (exceeding 10⁷ Salmonella), the cecum of streptomycin-pretreated mice is colonized within 2–6 h, and the inflammation is evident within 8 h post-exposure. The primary site of replication in the streptomycin-pretreated mouse model is intestinal epithelial cells in the lower intestine, inducing symptoms observed in humans (inflammation of distal ileum, cecum, and colon with edema of the submucosa and lamina propria, and with infiltration of neutrophils into the intestinal lumen). Significant gaps in the knowledge include the relevance and scaling of data from the streptomycin-pretreated mouse model to predict likelihood and severity of human illness; uncertainty about replication prior to invasion of a human host cell; and rates of growth within infected human gut cells and in the gut lumen both before and after inflammation.

In the fourth and final paper, Scientific Data and Theories for Salmonellosis Dose-Response Assessment (Marks and Coleman 2017), Harry M. Marks and Margaret E. Coleman describe an “expansive” risk assessment approach for characterizing dose–response relationships for salmonellosis, a self-limiting enteric illness with rare observations of severe or chronic complications. The expansive approach provides a range of possibilities (alternative data and theories) within the framework of present scientific knowledge of the human superorganism for communicating uncertainties transparently among risk assessors, decision makers, and stakeholders. Currently, many salmonellosis assessments rely on a narrow approach, use of conservative, non-threshold, low-dose linear models for dose–response assessment that represent fundamental misunderstandings of human biology before the “microbiome revolution” of the 21st century. The narrow approach commonly overestimates illness and ignores alternative human clinical data and theories consistent with the healthy human superorganism. The authors describe alternative data and theories (core and auxiliary) that incorporate variability and heterogeneity for the human and murine superorganisms, particularly innate colonization resistance lost with antibiotic-induced dysbiosis. Sublinear and threshold dose–response relationships appear consistent with multiple lines of evidence for salmonellosis as the knowledge of the human microbiome and its role in innate immunity expands.

As Next Generation microbiological methods advance, microbial risk assessment practices must also advance to more realistically model the superorganism in health and disease. Incorrect assumptions and policy choices of the past, intended to be public health protective, must be reexamined in light of scientific advances including the Human Microbiome Program. A concrete step to bring safety evaluations and risk assessment in line with new data and theories about the human as a superorganism involves an “expansive” approach, systematically considering alternative data and theories. Otherwise, unrealistic models will continue to overestimate illness, grossly underestimate uncertainty, and support a false confidence in the old paradigm of narrow risk approaches. Prior regulatory decisions, with costs and benefits of interventions linked to the old paradigm, also merit the re-examination for the relevance in light of the “expansive” approach. Specific model-directed studies consistent with alternative data and theories could support future extrapolations of existing dose–response relationships for human salmonellosis among healthy and dysbiotic populations.

Acknowledgments

The author thanks Margaret Coleman, Coleman Scientific Consulting, and Janice Dietert, Performance Plus Consulting, for their editorial suggestions.