Melding microbiome and nutritional science with early child development

Abstract

Programs for treating malnutrition in children should consider how food formulations affect postnatal gut microbiome development.

The pressing need to improve nutritional status globally is reflected in the United Nation’s top three Sustainable Development Goals: no poverty; zero hunger; and good health and well-being¹. Currently, one in three children under 5 years of age is either undernourished (wasted and/or stunted) or manifests their malnutrition as overweight². Part of the problem and part of the solution may come from the vast collections of microbes that establish themselves in the gastrointestinal tract beginning at birth.

30 million children worldwide suffer from moderate acute malnutrition, defined as a weight-for-length measurement 2–3 standard deviations below the median (weight-for-length z-score, −2 to −3) of a multinational reference cohort of 8,500 children³. DNA-sequencing technologies now allow comprehensive inventorying of the microbes present in the gut communities of infants and children. Application of these culture-independent methods to serially collected fecal samples from healthy members of birth cohorts uncovered a shared program of community assembly that is largely completed by the end of the second postnatal year. This program is perturbed in children with moderate acute malnutrition and is perturbed even more in severely malnourished children⁴. Preclinical studies suggested that disrepair of the microbial community is a contributing cause, not simply an effect, of malnutrition. This observation prompted a search for foods containing ingredients that could not only increase the abundance of underrepresented growth-promoting bacterial species but also reduce the proportion of bacteria that are inappropriately prominent⁵. In a recently reported randomized controlled trial, children with moderate acute malnutrition whose diets were supplemented with one such microbiota-directed complementary food (MDCF) formulation (called ‘MDCF-2’ in that study) exhibited faster rates of weight gain, an increased abundance of plasma-protein mediators of various facets of growth, and more-complete repair of their gut communities than that of children who received a standard ready-to-use supplementary food⁶.

These findings add to a growing body of evidence indicating that the nutritional value of food extends beyond those macro-nutrients and micro-nutrients that are directly absorbed by the host⁷ and highlight a need to better understand how interactions between the gut microbiome (both microorganisms and their gene products and metabolites) and molecules present in foods are linked to healthy growth. These results in children also give credence to the idea that disruptions in normal co-development of the gut microbiome and host can be repaired and that this facet of systems biology is pliable. The next step is clinical studies designed to determine the durability of microbiome repair and growth promotion attainable with a given MDCF formulation in a given population. Researchers should also consider the optimum timing for initiating treatment—i.e., how the time of onset and severity of the perturbation of normal microbiome–host co-development and the period of time during which that co-development is perturbed relate to the capacity to restore healthy growth—as well as whether the efficacy of a given MDCF formulation is generalizable across children from different geographic locales and/or with different co-morbidities.

One goal is to identify bioequivalent, geographically adapted MDCF formulations composed of inexpensive, locally available ingredients. This approach would provide incentives for local production to be cost-competitive with external manufacturers and would help ensure that the MDCF formulations are optimized for a given regional agricultural and consumer ecosystem. At present, creating biosimilar, ‘geo-adapted’ MDCFs requires a multistep journey analogous to the one used to create MDCF-2 (Fig. 1). The first step is to characterize features of the developing gut microbiomes of children with healthy growth phenotypes and those with malnutrition in the population targeted for treatment. The next step is to introduce microbial communities from malnourished children representative of the population of interest into gnotobiotic animals and thereby enable tests of MDCF prototypes that target microbial community members whose misrepresentation and/or aberrations in expressed metabolic functions need to be rectified in order to optimize co-development of the community and host. The final step is to advance candidate therapeutic food formulations that have satisfactory organoleptic properties to clinical studies in the very populations of malnourished children whose microbial communities were used to fashion the preclinical model.

Fig. 1 |. An approach for designing and testing MDCFs for treating childhood malnutrition.

A birth cohort study is first performed in a population in which the burden of disease is great (1). The network plots represent gut microbial communities sampled longitudinally; each node represents a bacterial strain, node size represents the abundance of that bacterial strain, and lines between nodes indicate covariation between the abundance of two strains. In the example here, assembly of a healthy gut community (top plot) is characterized by strong coupling between the green, purple and orange nodes, with a predominance of the strain represented by the green node. In contrast, community configuration in malnourished children (bottom plot) is characterized by a high abundance of the strain represented by the purple node and sparse coupling between the strain represented by the green node and other community members. representative gut communities from healthy or malnourished children are then introduced to young germ-free mice fed a diet that resembles one consumed by the donor population (2). The goal is to ascertain whether features of malnutrition are transmitted by malnourished donor communities to the recipient mice. Microbial targets (colored circles in panel 3) that contribute to growth, neurodevelopment, regulation of the immune system, bone biology and other features of host physiology are identified through the use of statistical methods that link microbial community components and facets of host physiology and metabolism. In the next step (3), the effects of various food staples, screened singly or in combination, on the fitness and expressed functions of these microbial targets are determined. Candidate MDCF prototypes are selected (4) on the basis of (i) their capacity to effect community repair (in this case, toward a state characterized by a high abundance of the strain represented by the green node and strong coupling between it and other members of the community) and (ii) the local availability, cultural acceptability, affordability and organoleptic properties of their ingredients. Selected lead formulations are advanced to pre-proof-of-concept and subsequently proof-of-concept studies, at least initially, in the same populations of malnourished children in which the therapeutic targets were identified (5).

Correlating changes in host physiology with the metabolism of MDCF ingredients by the microbiome should facilitate the identification of bioactive components of MDCFs, as well as enable critical evaluation of the efficacy and bioequivalence of different MDCF formulations. This research will also guide the development of new formulations whose ingredients are selected from databases of food staples with known geographic availability and whose chemical compositions have been delineated by new mass spectrometry–based methods⁸. Knowledge of these microbiome-active ingredients would also provide a foundation for the establishment of more-precise dosing requirements.

A primary outcome in randomized clinical trials testing the effects of nutritional interventions in children with wasting is the rate of change in weight gain or weight-for-length z-score. However, much as the composite disease-activity score captures disease activity or treatment response in rheumatoid arthritis, or liquid biopsy tests that utilize next-generation sequencing allow more-precise stratification of patients with advanced cancer, a more useful outcome in clinical trials would be a comprehensive, multi-featured readout of the effects of microbiome-directed therapeutics for childhood undernutrition that includes assessments of microbial community repair and host physiological state. For example, aptamer-based arrays currently available can simultaneously measure over 5,000 plasma-protein biomarkers and effectors of various aspects of host biology with great sensitivity over a large dynamic range and require only small amounts of blood⁶. Incorporating multi-featured readouts should help in determining how the rate and extent of microbiome repair achieved with MDCFs (or other approaches) correlate with changes in host systems and subsystems associated with musculoskeletal, immunological, metabolic and neurodevelopmental functions, rather than simple height and weight measurements alone. In contrast to the relative simplicity of anthropometric measurements, one obstacle to overcome will be the affordability and/or availability of such molecular diagnostic tests in the low-income settings where malnutrition is endemic.

Selecting primary and secondary outcome measures for trials involving comprehensive characterization of host and microbial community features creates challenges for downstream statistical analyses of the resulting datasets, including how to appropriately adjust for multiple-hypothesis testing⁹. Nonetheless, the lessons learned about what features are most informative to include as outcome measures will not only help in refining power calculations for future MDCF efficacy trials but could also benefit the development and testing of synbiotic formulations, composed of food plus microbes, that may be needed to restore gut communities that are so disturbed that repair is not achievable with MDCFs alone. Moreover, investment is needed in efforts to identify a sparse set of features that define microbial community organization and function and to develop the tools needed to measure them in a rapid and affordable fashion¹⁰. The resulting point-of-care diagnostics targeting both the microbiome and related host features would enable more-precise stratification of children who present with malnutrition and more-timely assessment of their responses to treatment. These diagnostics could also pave the way for adaptive clinical trials designed to optimize treatment and could serve as part of a preventative medicine initiative in which definitions of wellness are based in part on measurements of the state of development of the gut microbial community.

Although that envisioned approach for evaluating MDCFs brings a microbial and ‘multi-omics’ perspective to human-nutrition research, it presents a variety of challenges for organizations such as the WHO and UNICEF that formulate recommendations for treating malnourished children. These challenges include balancing a desire to aggregate children over broad age ranges (e.g., from 6 months to 24 or even 36 months) on the basis of their anthropometric scores with the inevitably more stratified view of human biology that is revealed when co-development of the microbiome and host is considered; how this balancing act is negotiated could have substantial therapeutic and policy implications¹¹. A list of other considerations for MDCF development that are related to science and technology, social and cultural considerations, and policy and governance is provided (Table 1).

Table 1 |.

Issues related to the development and deployment of MDCFs

Science and technology	Development of affordable diagnostics of nutritional status that include measurements of gut microbiome-host co-development
	Establishment of a reference database of the chemical compositions of world food staples (cultivars and varieties) to guide future development of microbiome-active nutritional products and/or supplements
	Investment in preclinical and clinical development pipelines for the design of geographically and/or culturally adapted MCDF formulations that promote growth and durable microbiome repair on the basis of knowledge of the structure-activity relationships between food components and microbes
Society and culture	Educational strategies that enhance compliance with complementary feeding practices designed to promote gut microbiome repair in malnourished children, and that ultimately are incorporated into preventative strategies for the promotion of healthy microbiome development
	Appropriate labeling, branding and advertising approaches for microbial community-directed food formulations that are consistent with best practices for nutritional products intended for infants and children
Policy and governance	Definition of data packages needed to support claims, mechanism of action, efficacy and safety of MDCFs for children
	Regulatory classification of MDCFs within and across national boundaries
	Ensuring access to affordable and sustainably produced MDCFs for populations who would benefit

Research and development in this area needs to be incentivized in ways that ensure equitable and widespread distribution of the fruits of such efforts. A promising system developed by Nutriset, the largest manufacturer of ready-to-use foods for malnourished children globally, relies on a franchise model that provides local stakeholders the intellectual property, tools and expertise to produce its therapeutic foods⁵. Such a model leverages patent protection to create financial incentives for local manufacturers by effectively making intellectual property inaccessible to large manufacturing companies in high-income countries that could potentially monopolize production. Although focusing on local production may sacrifice short-term economies of scale for sustainability, a campaign to develop and deploy MDCFs must include long-term strategies for developing nations to achieve nutritional autonomy¹².

Looking ahead, it is important to determine whether microbiome-directed therapeutic foods can be developed for women of childbearing age to maximize their nutritional health, minimize low birth weight in their offspring and help break the vicious cycle of intergenerational malnutrition. Additionally, can knowledge gained from identifying foods that repair the microbiome of children with malnutrition also be used to identify a temporal sequence of complementary foods that prevent failed microbial community development and thus promote healthy growth? These questions lie at the base of a new kind of food pyramid that considers the gut microbiome as part of nutrition itself.