Why Most Microbiome Therapies Never Reach the Clinic, and What Decides the Ones That Do

Microbiome science has produced an enormous catalog of promising therapeutic candidates, yet only a handful of microbial products have ever been approved for patients. A new review published in Frontiers in Cellular and Infection Microbiology on 27 October 2026 asks a deceptively simple question: what actually determines whether a microbial therapeutic reaches the clinic? The answer, according to authors Nives Hribernik, Harsh Maan and Tatyana L. Povolotsky, is that biological sophistication matters far less than the ability to achieve durable colonization, demonstrate meaningful clinical benefit and satisfy biocontainment and regulatory requirements.

The review, a narrative synthesis rather than a systematic review, draws on controlled trials, regulatory decisions and mechanistic studies published between 2020 and 2026, gathered from PubMed and Web of Science. The authors deliberately selected examples to illustrate distinct translational outcomes rather than to provide comprehensive coverage. They compare three therapeutic strategies: traditional probiotics, next-generation probiotics derived from commensal organisms, and engineered live biotherapeutics built with synthetic biology tools. Examples are drawn from metabolic disease, oncology and the vaginal microbiome.

The clearest success story is recurrent Clostridioides difficile infection. Two microbiome therapeutics, Rebyota, given rectally, and Vowst, an oral spore-based product, were approved by the US Food and Drug Administration in 2022 and 2023 respectively to prevent recurrence after antibacterial treatment. The authors attribute this success to a rare alignment of favorable factors: a defined pathogen, a clear post-antibiotic treatment window, a meaningful clinical endpoint and substantial unmet need. Even so, approval does not eliminate evidentiary uncertainty. The 2024 American Gastroenterological Association guideline recommends fecal microbiota-based therapies only conditionally and rates the certainty of evidence as low.

By contrast, engineered probiotics for inflammatory gastrointestinal disease are often more mechanistically sophisticated but far less clinically mature. In murine colitis, an Escherichia coli Nissle 1917 strain engineered to overexpress catalase and superoxide dismutase and protected in a chitosan-alginate formulation reduced oxidative stress and supported epithelial repair, but controlled human efficacy has not been shown for such products. The review’s central observation is stark: the most clinically advanced products are not the most extensively engineered, but those for which the therapeutic target, treatment window, delivery route and clinical endpoint align.

Metabolic disease illustrates why demonstrating microbial activity is not enough. Synlogic engineered the E. coli Nissle strains SYNB1618 and SYNB1934 to consume phenylalanine in the gut of patients with phenylketonuria. In the 14-day Synpheny-1 study, exposure to orally administered deuterium-labeled phenylalanine fell by 34 percent and 43 percent respectively, providing clear evidence of target engagement. Yet the pivotal Synpheny-3 trial was discontinued in February 2024 after an internal review indicated it was unlikely to meet its primary endpoint, a decision unrelated to safety. The program demonstrated biochemical activity in humans, but this did not translate into meaningful clinical benefit, and because detailed analyses were not disclosed, the precise reasons remain speculative.

A different limitation emerged in a 12-week randomized, double-blind Phase 2 trial of live Akkermansia muciniphila in 58 participants with overweight or obesity and type 2 diabetes. Body weight and HbA1c declined in both groups with no significant between-group difference overall. In prespecified subgroup analyses, however, participants with low baseline abundance of the organism colonized efficiently and lost body weight, fat mass and HbA1c, whereas those with high baseline abundance colonized poorly and showed no comparable improvement. Fecal-transfer experiments in germ-free mice reproduced this pattern. The finding shows that baseline microbiome composition can determine whether supplementation alters community structure or host phenotype at all.

In oncology, the evidence is clinically relevant but still preliminary. In a prospective cohort of 338 patients with advanced non-small-cell lung cancer treated with immune-checkpoint inhibitors, baseline intestinal A. muciniphila was associated with higher objective response rates and longer overall survival in multivariable analyses, independently of PD-L1 expression and other established clinical factors. Whether this association can be translated into therapeutic benefit is now being tested in early interventional studies. Separately, in randomized trials, Clostridium butyricum MIYAIRI 588 improved progression-free survival when combined with immune checkpoint therapy in patients with metastatic renal cell carcinoma, despite no corresponding shift in the measured microbiological endpoints. The authors caution that association between a microorganism’s abundance and treatment outcome does not establish that administering it will improve that outcome.

The review uses the vaginal niche as a case study because it offers a more tractable setting than most gut indications. The healthy vaginal ecosystem is typically low in microbial diversity and often dominated by Lactobacillus species, particularly Lactobacillus crispatus, whose lactic acid production maintains a low pH that restricts the growth of Gardnerella and other bacterial vaginosis-associated anaerobes. The distinction is ecological rather than taxonomic: L. crispatus is native to the vaginal niche, where it can form a stable protective community, whereas conventional oral probiotic strains introduced into the gut persist only depending on the recipient’s microbiome. The strongest efficacy evidence comes from LACTIN-V, a vaginal live biotherapeutic containing L. crispatus CTV-05. In a Phase 2b trial of 228 women who had completed vaginal metronidazole gel, bacterial vaginosis recurred by week 12 in 30 percent of LACTIN-V recipients and 45 percent of placebo recipients, with benefit still evident at week 24. Notably, benefit was achieved despite incomplete and variable long-term colonization, showing that clinical efficacy does not necessarily require durable engraftment in every recipient.

Recent studies aim to improve colonization consistency through rational strain selection, ecological support, optimized release and selective removal of competing organisms. The Phase 1 VIBRANT trial evaluated multi-strain L. crispatus products after metronidazole in 91 participants with bacterial vaginosis; product strains were detected in 66 percent of active-group participants during the first five weeks, and 49 percent of initial colonizers still carried them at week 12, usually as dominant members, with detected participants showing a lower risk of recurrence, though the trial cannot establish that engraftment itself caused the lower recurrence. In a randomized study of 70 healthy premenopausal participants, the vaginal synbiotic VS-01 combined three L. crispatus strains with growth-supporting substrates in a slow-release tablet; among participants lacking an L. crispatus-dominant microbiome at baseline, 90 percent converted to a dominant community by day 21 against 11 percent on placebo. Meanwhile, Gardnerella-targeted endolysins such as PM-477 and its successor BNT331-EL depleted viable Gardnerella by at least 94 percent in preclinical and ex vivo work, but BNT331-EL also reduced Lactobacillus iners by an average of 92 percent, and neither compound has yet shown clinical efficacy in humans.

The authors integrate these observations into a translational-readiness framework that evaluates microbial therapeutics across five dimensions: target specificity, colonization and exposure, safety and containment, demonstrated benefit and regulatory precedent. The framework is qualitative, derived from the cases reviewed, and has not been externally validated; it is intended to structure prioritization rather than to predict outcomes. Two often-overlooked factors receive particular attention. Biocontainment safeguards such as kill switches, auxotrophic dependencies and inducible expression systems remain essential, yet the long-term stability and evolutionary behavior of engineered circuits in the human microbiome remain insufficiently characterized. And while engineered strains have progressed to Phase 3 under existing regulatory frameworks, demonstrating that development pathways exist, no jurisdiction has yet approved an engineered live biotherapeutic. The limitation, the authors conclude, is one of regulatory precedent rather than the absence of a regulatory pathway. Across all applications, they argue, future trials should account for baseline features that predict microbial establishment, define the duration of in-host activity required for each strategy, and integrate safety and regulatory planning early in development.

Subject of Research: Translational barriers facing natural and engineered probiotic therapeutics

Article Title: A translational-readiness framework for natural and engineered probiotics

Article References: Hribernik, N., Maan, H., & Povolotsky, T. L. (2026). A translational-readiness framework for natural and engineered probiotics. Frontiers in Cellular and Infection Microbiology, 16, Article 1965075. https://doi.org/10.3389/fcimb.2026.1965075

Image Credits: AI Generated

DOI: 10.3389/fcimb.2026.1965075

Keywords: microbiome, probiotics, live biotherapeutics, Akkermansia muciniphila, vaginal microbiome, bacterial vaginosis, Lactobacillus crispatus, Clostridioides difficile, synthetic biology, clinical translation, biocontainment, regulatory approval