What is unknown in using microbiota as a therapeutic?

Introduction

Since the explosive interest in the study of gut microbiota and the report of using fecal microbiota transplantation (FMT) in recurrent Clostridioides difficile infections (CDI) about a decade ago, FMT has been widely tested for a number of gastrointestinal conditions including irritable bowel syndrome, inflammatory bowel diseases, hepatic encephalopathy, as well as nonintestinal conditions including obesity, metabolic syndrome, and even autism.¹ It is one of the most effective strategies to change the intestinal ecology and restore a healthy microbiota. Several guidelines and consensus reports on the use of FMT for refractory or recurrent CDI have been released by international professional bodies.^2-4 Recently, a national registry of FMT in the United States has been set up and reported on the efficacy and safety of this treatment in CDI.⁵ Yet there are still missing gaps in our knowledge and experiences with FMT, as well as other related microbiota-based therapies. Before we jump into a widely, and somehow indiscriminatory, use of FMT in various conditions, much experimental works and properly conducted trials are needed.

Undetected unknowns

Using the current sequencing technology, a much wider and deeper identification of the microbiota populations in the gut can be achieved. Numbers of known and unknown members of the human microbiota identified from population-wide and large-scale metagenomic assembly studies in multiple body sites indicated that human microbiota is composed of at least 22 known phyla with 4930 species-level genomes and 2.85 M functionally annotated genes.⁶ As for the gut microbiota, the Unified Human Gastrointestinal Genome collection has identified 204 938 nonredundant genomes from 4644 gut prokaryotic species,⁷ expanding our existing reference databases of microbial genomes. Despite several recent culture-based studies,^{8, 9} most of the genomes were represented by uncultured species.⁷ Moreover, at the deepest level of hidden diversity, there are still microbes not readily captured by current technology, so-called undetected unknowns. These include low abundance but potentially crucial taxa, such as keystone species that affect the microbial community disproportionately large compared with their relative abundance.¹⁰ There are complexities in profiling these uncommon bacterial or other nonbacterial microbes, due to their lower abundance, lack of complete reference genomes, difficulty in microbiological culture, and limited toolkits for their investigations.¹¹

Despite these difficulties, there have been attempts to evaluate these less well-known species from our gut. Culturomic methods have been employed to create diverse culture conditions for new bacteria, many have previously been thought to be unculturable.^{12, 13} Alternations in both the enteric virome and mycobiome have been described in diseases including colorectal cancer^{14, 15} inflammatory bowel diseases,^16-20 and diabetes mellitus.^{21, 22} With regards to FMT, bacteriophages and fungi have been found to be associated with FMT efficacy in CDI patients, with successful transplants being associated with a stable virome core,²³ reductions in Caudovirales,²⁴ and Candida albicans in the FMT recipients.²⁵ These data suggest the importance of nonbacterial components and the trans-kingdom relationships in CDI and its FMT treatment.^26-28 Nevertheless, our understandings of these nonbacterial microbiota remained at metagenomic level at best, and their potential functional roles have not yet been fully explored. Without a thorough understanding of the composition of microbiota, it is difficult to further improve the efficacy of FMT through therapeutic modulation, in addition to the safety concern of the nonbacterial species being transferred.

Colonization unknowns

Besides incomplete knowledge of normal gut microbiota composition, the other major challenges in using FMT is the engraftment of microbes in the gastrointestinal mucosa.^{29, 30} Stool samples for microbiota studies may not accurately reflect on the sessile population of microbiota in the mucus,³¹ normally located at the outer layer³² but may penetrate to interact with the epithelial and immune cells of the host in diseases like inflammatory bowel diseases.³³ If anything, the fecal microbiota represents a microbial population more easily shed off from the gastrointestinal epithelium. In a study that used single nucleotide variant to monitor microbial strains in a FMT clinical trial, 37.6% of the donor-specific strains were retained at 3 months, and the colonization success was greater for conspecific species than new species.²⁹ Long-term colonization of bacterial and viral species has been reported.^34-37 Individual patterns of microbiota resistance and donor-recipient compatibilities were observed,²⁹ although some donor microbiota harboring high microbial diversity and Prevotella to Bacteroides ratio were dominantly effective in engraftment and may constitute “super-donors.”³⁸

Antimicrobial pretreatment and repeated dosing have been advocated to improve clinical response of the FMT treatment. Repeated FMT dosing has also been employed with variable success in irritable bowel syndrome, metabolic diseases such as obesity, and in recurrent CDI.^39-41 On the other hand, antibiotics prior to FMT may open up niche for microbial engraftment in an ecologically clean state and hence enhance the success of the therapy. This concept has been demonstrated in a mouse model study by Ji et al,⁴² although more data are necessary to establish the need of antibiotic pretreatment in human.^{29, 43, 44} Preprocedural proton-pump inhibitors has uncertain benefits and may therefore be more dispensable.^{45, 46} Further work needs to be done in this space to understand the mechanism of microbial engraftment on gastrointestinal epithelium and the interactions between microbes in colonization of the surfaces.

Functional unknowns

Many studies have looked at the compositions of microbiota in healthy and diseased individuals. This is relatively simple given the current advanced sequencing technology. Nevertheless, defining a normal microbiota may not be as simple as finding what is there and what is not there. In general, high microbial diversity, balanced composition of Bacteroidetes versus Firmicutes, high concentration of fecal butyrate and low abundance of C. albicans are considered features of a “healthy” microbiota, although there might not exist a single healthy state.^{47, 48} What is more challenging is elucidating the functional potential encoded in the overall microbiota. To understand how gut microbial communities are involved in FMT and whether key microbes can be manipulated to improve therapeutic outcome, one must understand the functions of microbes involved in the therapeutic mechanisms beyond the species level.²⁹ It is known that strain-level variations are pervasive among microbial species, exhibiting diverse functional capacity enabling intraspecies strains to behave differently and sometimes antagonistically.⁴⁹ For example, a study characterizing hundreds of Lachnospiraceae isolates revealed significant interspecies and intraspecies genomic diversity.⁵⁰ Strains of the same species, such as Ruminococcus gnavus and Bacteroides fragilis, are known to differ in their capacity in carbohydrate utilization, promotion of inflammation, and production of a carcinogenic metalloprotease toxin.^{6, 50, 51} Without knowing the functional mechanisms of microbes in the therapy, one cannot be sure which are the critical microbial constituents. This will impact on finding the right donor for the FMT.

While single microbial constituents are key to the understanding of microbial community, studies have shown that microbes coexist or co-exclude in such a way that their synergic functions or exclusive interactions might play an important role in the host–microbiota interaction. Consideration for this interactive network using an integrative multibiome approach is needed, to study the disease microbiota which may not be appreciated analyzing a single microbial group.⁵² The highly individualized pattern of colonization should also be appreciated, when microbial modulation is introduced either through FMT²³ or probiotic consumption.⁵³ The mechanisms of synergism or antagonism are far from clear at this stage.

Indication unknowns

Multiple randomized controlled trials have proved the efficacy of FMT for treating recurrent CDI in about 90% of the cases.^{54, 55} While this success has generated optimism for using FMT to treat other microbiota-associated diseases, it seems to have picked on recurrent CDI as a low-hanging fruit. Positive therapeutic effects of FMT have been observed in ulcerative colitis,⁵⁶ irritable bowel syndrome,^{57, 58} alcoholic use disorder,⁵⁹ and hepatic encephalopathy,⁶⁰ with moderate success compared with recurrent CDI. Nevertheless, the role of FMT is less established for other conditions, including Crohn's disease,⁶¹ metabolic disorders,^62-64 and autism spectrum disorder,⁶⁵ especially when we focus at the clinical outcomes of these studies. There is also heterogeneity in donor recruitment, sample testing, and treatment protocols between various clinical trials,⁶⁶ which hinder the adoption of FMT in routine clinical use.⁶⁷ Together with its safety concern, FMT has not yet been approved by the US Food and Drug Administration for use outside the Investigational New Drug settings, except for CDI cases not responding to standard treatments.

Efficacy and adversity unknowns

Without a thorough understanding of the microbial role in disease pathogenesis, it is hard to predict the treatment efficacy and more importantly, the potential adversity. To date, the use of shotgun metagenomics screening to predict FMT success in CDI and other disease are mostly confined to academic research due to its cost and timescale.⁶⁸ Among various factors, key determinants of FMT success include donor selection and sample administration, as well as recipient parameters including their clinical status, medications, and baseline microbiota.⁶⁹ There is also evidence that matching between donors and recipients may be important to optimize FMT efficacy.³⁰ Host factors such as dietary habit, lifestyle factors, difference in innate immunity, and xenobiotic exposure may affect the donor and recipient matching.⁷⁰ A more comprehensive profiling of gut microbiota samples allows safer selection of donor samples and an improved understanding of which taxa contribute the most to the success of therapy. Changes in a-diversity and b-diversity of microbiota population after FMT could be a predictor of response.⁷¹ However, as most studies only look at short-term effects of FMT on disease improvement over weeks and months, little is known about the long-term efficacy of FMT especially for chronic diseases and conditions.

Besides efficacy, the medium to long-term (6 months and beyond) adversity data of FMT remain relatively scanty. Recent evidence suggested that long-term engraftment of donor bacterial and viral species into the recipients is possible.^{30, 37} There are reported cases of diarrhea, abdominal pain, bloating, and constipation in more than 10% of recipients after FMT.^{5, 72} But more serious complications such as bacteraemia and local infections (urinary tract infection, pneumonia, and gastrointestinal infections) have also been reported leading to hospitalization and even death.^{73, 74} There were also concerns of potential transmission of pro-carcinogenic bacteria during FMT, with sustained detection of Escherichia coli strain harboring the carcinogenic colibactin toxin.^{75, 76} It is unclear how much of these adverse events are directly or indirectly related to the transfer of donor microbiota or the procedure of FMT itself. A registry of FMT usage and its outcome is mandatory to keep surveillance of the potential hazards of this procedure.⁵

Statistical unknowns

Microbiota studies require statistical analysis based on hypothesis testing. There are three components in the model of dynamic interactions between host, microbiota, and environment (including intervention). The two main themes of statistical hypothesis include (i) to characterize the relationships between microbiota features and the clinical parameters and treatment outcomes and (ii) to identify potential genetic, dietary, and other environmental factors that are associated with the microbiota. In most studies, standard statistical t-tests and non-parametric Wilcoxon rank-sum tests are used to compare the diversity between groups or the microbial abundance. When comparing more than two groups, the one-way anova or its non-parametric equivalent of Kruskal–Wallis test is often employed. However, some of these tests are used without resolving the issue of data dispersion, that may require transforming and modeling due to the high dimensionality, the association sparsity, and interdependence between microbial taxa.^77-79 Most assume that the microbiota data are independent, but this is probably not true as coexistence and co-exclusion of microbiota are well known. The other limit of statistical analysis is the question of inferring correlation as causality. To model causative effects of microbiota data, both appropriate design and suitable statistical models are needed.

Unknowns in other microbiota therapeutics

Fecal microbiota transplantation is one of the most effective ways to change the intestinal microbiota; nevertheless, the many unknowns have posed challenges to its widespread use in routine clinical care. As such, other therapeutic approaches are being studied, to modify the gut microbiota and treat diseases like FMT. These include FMT-based products derived from donor stool, such as frozen stool capsules, lyophilized microbes, bacterial consortia, or purified spores isolated from stool.⁸⁰ They have been tested for treating recurrent CDI^{81, 82} and other conditions^{83, 84} with variable successes. Other non-FMT-based approaches have been developed, such as direct strategies to enrich certain beneficial bacteria by probiotics, to deplete harmful bacteria by antibiotics, or indirect strategies to change the gut microbiota through prebiotics or personalized nutrition. As these methods may have more pervasive effects on the microbiota, there are also efforts to develop more targeted interventions, such as bioengineered commensals, bacteriophage therapy, or gene editing therapy to modify specific microbes.^{80, 85}

While it is not the scope of this review to go through these individual approaches, it is worth noting that many share the same unknowns as FMT. FMT-based therapeutics, such as lyophilized microbes or purified spores, have similar unknowns with their colonization capability, intraspecies functionality, efficacy, and adversity. Probiotics are limited by their modest effects on the microbiota, limited efficacy, and variable colonization patterns.⁵³ Antibiotics, on the other hand, have pervasive and untargeted effects on the microbiota in addition to their own side effects.⁸⁶ Novel engineering approaches, such as bioengineered commensals⁸⁷ or phage therapy,⁸⁸ offer attractive alternatives to probiotic and antibiotic treatments by having more targeted effects. There has been recent success in using phage therapy to eliminate cytolytic Enterococcus faecalis in models of alcoholic hepatitis⁸⁹ and protumoral Fusobacterium nucleatum in enhancing chemotherapy for colorectal cancer.⁹⁰ Nevertheless, continued efforts are needed to overcome various difficulties, such as the need for phage and host selection, viral isolation and preparation, efficacy and adversity studies, and regulatory hurdles.⁹¹

Putting unknowns into research equation

We have discussed several major areas of unknowns pertaining FMT as a biotherapeutic remedy. While these may limit the wider adoption of FMT, they represent some of the most fundamental questions in the microbiota science. These include key questions on the role of nonbacterial microbes, interkingdom networks, colonization patterns, predictors of FMT success, and long-term efficacy and safety concerns, spanning across the abovementioned areas of unknowns (Table 1). Finding answers to these questions will help understand the mechanisms of FMT, inform its long-term efficacy and safety, facilitate its regulatory steps for clinical uses, and enhance development of related therapies.

Conclusion

Even though there are still a lot of missing gaps in our knowledge of gut microbiota and its behavior, FMT has already been extensively used in various gastrointestinal and extraintestinal conditions, either as a last resort or experimental therapy. Multiple levels of unknowns in microbiota research and application of FMT needs much work to elucidate before this therapy can be used effectively and safely.⁹² This article is not meant to discourage the use of FMT but to caution against inadvertent applications, thinking that this is a “natural” and “organic” therapy that is usually harmless. With the blooming of FMT therapy as a business in therapeutics, we must not forget that the first principle in Medicine is “Do No Harm” to our patients.