Expert consensus document: The International Scientific Association for Probiotics and Prebiotics (ISAPP) consensus statement on the definition and scope of prebiotics

In previous iterations of the term prebiotic, 'selectively' was interpreted as referring mostly to lactobacilli and bifidobacteria. Specific stimulation of bifidobacteria (bifidogenesis) was considered a prebiotic effect. Early research on gut microbial ecology was based on culture methods, which we now know were insufficient to reveal the complexity of prebiotic-induced microbial changes. Molecular-based methods, which have since identified a broader range of members of the gut microbial community, have enabled the appreciation that more bacterial genera might utilize some prebiotic substrates, by fermentation and other metabolic pathways. These microorganisms can vary depending upon the host and ecosystem under consideration. Hence, it is recognized today that prebiotic effects probably extend beyond bifidobacteria and lactobacilli, but to meet the selectivity criterion of a prebiotic, the range of microorganisms affected must be limited. To this end, in two human studies that used high-throughput sequencing, bifidobacteria responded to prebiotic use22,27. However, other groups such as Faecalibacterium prausnitzii also increased in abundance in one trial22 and in another study Anaerostipes spp. were additionally elevated, whereas Bilophila spp. decreased27. Both studies used high-throughput sequencing to confirm selectivity of the prebiotic fermentation. Selectivity does not necessarily mean effects on just one microbial group; a selective effect could extend to several microbial groups, just not all. A prebiotic, in addition to having a selective effect on microorganisms, must also evoke a net health benefit. The guiding principles are that microorganisms affected and metabolites produced are considered to be beneficial and linked to a defined health aspect.

Envisaging every scenario is challenging. But, for example, is a product a prebiotic if its intake increases microbial production of butyrate? Short-chain fatty acids (SCFAs), such as acetate, propionate and butyrate, and some other compounds, are recognized as having mechanistic links to health outcomes28,29. If the effect is a measurable benefit to host health, distinct from a control, it would constitute a 'prebiotic effect'. To verify that the product itself is prebiotic, experiments would have to demonstrate that the product is selectively utilized, in this case by showing that a defined range of butyrate-producing microorganisms grow because of the product. Alternatively, the product might stimulate growth of other members of the microbiota, releasing metabolites that in turn stimulate butyrate production by other microorganisms. This phenomenon could constitute a 'cross-feeding effect'. The net result is still selective in that propagation of particular microorganisms led to this overall health effect. However, if pathogenic microorganisms are involved in butyrate generation and a negative consequence occurs for the host, then it cannot be termed a prebiotic. This distinction makes it important to determine both function and composition of the gut microbiota involved.

Similarly, prebiotics for use in the gut microbiota of humans should not form gas distension issues after ingestion; as such, their fermentation must be selective and preferably include genera that are not gas formers (such as Clostridium). This consideration points unequivocally towards the need for selective metabolism. Notably, neither bifidobacteria nor lactobacilli manufacture gas in their metabolism16.

Moreover, it is implicit that such influences on host health be determined in mixed microbial ecosystems containing the full microbiota of interest (that is, in vivo). Making inferences on prebiotic effects from pure or co-culture experiments is inadequate. Similarly, any conclusion regarding prebiotic activity must be based on an assessment of the full microbial diversity, not simply increased abundance of gut bifidobacteria or lactobacilli, for example. The best techniques available need to applied, particularly as the microbiome field has benefited greatly from molecular-based technological advances. These techniques would include high-throughput sequencing, including metagenomics, which demonstrates quantifiable changes in the microbiota. Similarly, metabonomic assessments, such as NMR or mass spectrometry, in appropriate biological materials can identify metabolic responses to prebiotics and help determine concomitant functionality of the microbiota.

A number of fermentable carbohydrates have been reported to convey a prebiotic effect, but the dietary prebiotics most extensively documented to have health benefits in humans are the non-digestible oligosaccharides fructans and galactans30. These oligosaccharides are preferentially metabolized by bifidobacteria16. A phenomenon explained by structure to function relationships; the linkage bonds in FOS and GOS can be readily degraded by degraded by β-fructanosidase β-galactosidase enzymes, respectively, which are prevalent in bifidobacteria. This genus also seems to preferentially metabolize the chain length size typical of oligosaccharides; that is, a degree of polymerization (DP) between 4 and 30 (Refs 31,32). Importantly, having the appropriate transport machinery to capture and deliver these substrates into the microbial cytoplasm is a key requirement and contributes to the selectivity of prebiotics in the target sites33 and emphasizes their ability to do so in a competitive environment in mixed culture ecosystems such as the human gut.

Substrates that affect composition of the microbiota through mechanisms not involving selective utilization by host microorganisms are not prebiotics. These substrates would include antibiotics, minerals, vitamins and bacteriophages, which are not growth substrates, even though their intake might alter microbiota and metabolic composition.

Certain soluble fermentable fibres are candidate prebiotics34, and some other types of dietary fibre can be prebiotic, provided that they are selectively utilized by the host microbiota and promote health. Categorizing fibres as prebiotics is complicated by the fact that a dietary fibre can be a prebiotic in one host but not another. For example, cellulose can be considered a prebiotic in ruminants but not in humans, as the latter's intestinal microbiota only poorly utilize β(1→4) linked D-glucose polysaccharides35. Furthermore, a substrate qualifying as a prebiotic might also depend on the target site. For example, xylitol can be considered as a prebiotic in the oral cavity, but has not been shown to be prebiotic elsewhere16,18.

Among the first group of substances recognized for their ability to influence gastrointestinal health were the oligosaccharides present in human milk. Human milk oligosaccharides (HMOs) are particularly important for the development of the newborn baby's intestinal microbiota and metabolic and immunological systems, which have consequences for health later in life36,37. Consumption of mother's milk containing these HMOs clearly increases the proportion of HMO-consuming Bifidobacteriaceae and Bacteroidaceae28. Bifidobacterium longum subsp. infantis (B. infantis) is the only Bifidobacterium spp. that has specifically evolved machinery to degrade the complete repertoire of HMOs. Other Bifidobacterium spp. predominant in adults, mainly B. longum subsp. longum, B. adolescentis and B. lactis, lack many of the enzymes necessary to directly utilize HMOs effectively38,39.

HMOs might indirectly affect composition of the intestinal microbiota by modulating immune responses and also have metabolism-independent mechanisms of action in the infant gut40. In particular, fucosylated and sialylated HMOs can prevent adhesion of pathogens to the intestinal epithelium through a competitive mechanism that ultimately protects the neonate from infection41,42. The main issues for this discussion are the following. Is there evidence that HMOs confer a health benefit in humans through the host's microbiota selectively utilizing them, and therefore fulfilling the prebiotic definition? And if compounds equivalent to HMO (or bovine milk oligosaccharides, BMOs) were to be produced by enzymatic synthesis, fermentation or extraction, could they still be considered as prebiotic?

The ability of HMOs, BMOs or synthesized compounds to act as a substrate for the selective growth of beneficial bacteria, such as Bifidobacterium spp., would be supportive evidence of a prebiotic nature43. To confirm their status as a prebiotic, a controlled human study showing selective growth of bifidobacteria resulting in a health benefit is also needed. However, the use of such compounds for in vivo studies is limited to only a few reports. In one study, a chemically synthesized compound, 2′-fucosyllactose (2′FL), equivalent to the naturally occurring 2′FL in HMO, was added to formula milk along with GOS. Although safe for infants, the 2′FL treatment provided no net difference in weight, length, head circumference and other measures compared with human milk over a 4-month period44. In another study by the same group, infants fed formula with 2′FL plus GOS had immune responses similar to breast-fed infants in that both groups had lower levels of inflammatory cytokines than infants fed formula plus GOS45. However, effects on the microbiota were not reported in this study. In a third study, 2′FL and another synthesized HMO, lacto-N-neotetraose, were administered to adults46. The treatments were well tolerated and led to an increase in abundance of Bifidobacterium spp. Collectively, these studies provide an incomplete assessment of the prebiotic properties of these synthesized versions of HMOs. Although 2′FL is utilized by B. infantis as well as some strains of B. longum subsp. longum and B. breve46,47, the ecological context (that is, infants versus adults) might dictate whether these HMOs are indeed prebiotic. Moreover, having structural equivalence to specific HMOs does not infer functional equivalence to the constellation of HMOs in milk48. Thus, for now, it is acceptable to state that some HMOs are candidate prebiotics.

Plant polyphenols constitute a class of compounds that can also meet the criteria of prebiotics, although far more studies in the target host are required. An estimated 90–95% of dietary polyphenols are not absorbed in the small intestine and, therefore, reach the colon49 where they undergo extensive biotransformation by the colonic microbiota. Increasing evidence indicates that health benefits associated with polyphenol consumption depend on microbial utilization and the metabolites produced, rather than on parent compounds50.

This evidence expands the prebiotic concept beyond non-digestible oligosaccharides such as FOS and GOS. However, evidence for these emerging prebiotics is scarce relative to the fructans and galactans16 and more studies measuring health benefits are required to fulfil their prebiotic status.

As selective utilization of a prebiotic by host microorganisms is key to its physiological effects, metabolic results of this utilization must, by deduction, be the main drivers. Some organic acids, for example, are principal end products of non-digestible carbohydrate or dietary fibre fermentation by host microorganisms. The main SCFAs (≥95%) generated mostly in the colon (humans) and caecum (rodents) as a result of several bacterial metabolic pathways are acetate (two carbon, C2), propionate (C3)) and n-butyrate (C4). These SCFAs are crucial for intestinal health and their activity can subsequently influence sites distant to the gut, with different SCFAs having varying functions. SCFAs can modulate certain aspects of metabolic activity including colonocyte function, gut homeostasis, energy gain, the immune system, blood lipids, appetite and renal physiology, as reviewed elsewhere16,51,52.

In a study published in 2017, 13C-labelling was used to show that colonic-administered acetate, propionate and butyrate were systemically available at 36%, 9% and 2%, respectively, with conversion of acetate into butyrate (24%) by the colonic microbiota53. Bifidobacteria, often stimulated by specific prebiotics, do not produce butyrate, so a probable scenario is that cross-feeding by other bacteria must have resulted in production of this SCFA. Much has been reported about the benefits of butyrate in the gut and beyond54, leading to the potential of known butyr