Ecological management of the microbiota in patients with cancer

Abstract

The composition of the intestinal microbiota influences cancer treatment outcomes, although the best way to use this knowledge to improve cancer care remains unclear. In this Review, I synthesize the current understanding of host–microbiota dynamics in cancer patients, and propose the integration of microbiota management guided by ecological principles in cancer care. Ecological management of the microbiota emphasizes the preservation of microbial populations — and the benefits they provide to the host — from the disruption caused by treatments such as chemotherapy and prophylactic antibiotics. The microbiota can be routinely and longitudinally monitored in patients using proven non-invasive methods, such as 16S ribosomal RNA amplicon sequencing. Longitudinal microbiome data can be processed with innovative computational tools based on principles of mathematical ecology to predict the risk of microbiota-related complications, guide treatment choices that minimize microbiota disturbance, and restore microbial populations damaged by cancer treatment. Routine microbiome monitoring could also generate extensive datasets for human-based research, which may lead to new microbiota-targeted interventions that improve responses to cancer treatments, including immune-checkpoint inhibitors. Applying ecological approaches to manage microbiota could enhance cancer care and improve patient outcomes.

Introduction

The human microbiota is composed of trillions of microorganisms that include bacteria, viruses, fungi and protozoa, and reside in various organs such as in the gastrointestinal system, skin and oral cavity¹. In the absence of disease, these microbial ecosystems can have beneficial effects on the host, such as improving their metabolism, protecting them against pathogens and enhancing their immunity. Similarly to other ecosystems, however, microbiota can be affected by disturbances. Strong disturbances — such as disease, treatment and even changes in diet — can shift the species composition, impair the benefits provided to the host and contribute to pathological conditions in various organs, such as atopic dermatitis², periodontitis³, bacterial vaginosis⁴ and inflammatory bowel disease⁵, some of which are associated with cancer risk.

Cancers and their treatments disturb the microbiota in patients^6–8. The altered microbiota can, in turn, affect host features (such as immune status⁹) and influence cancer treatment and, consequently, patient outcomes¹⁰. Although the precise cellular and molecular mechanisms underlying these interactions remain to be fully understood, growing evidence indicates that integrating microbiota management approaches into cancer treatment could improve patient outcomes (FIG. 1).

FIG. 1 |. Microbiota, immune system and cancer: an interconnected ecosystem affected by treatment.

a. The microbiota of patients with cancer constitutes a complex ecosystem that interacts with the immune system and is influenced by cancer and its treatments. Key immune cells, including neutrophils, lymphocytes and monocytes, engage in bidirectional interactions with the microbiota and cancer cells, shaping response to treatment and disease progression. b. Studies in patients undergoing allogeneic haematopoietic stem cell transplantation (alloHSCT) have provided insights on how cancer treatment disrupts the microbiota and how remediation strategies can restore immune balance. Treatments used for the management of patients with cancer, including chemotherapy, immunosuppressive drugs and also prophylactic antibiotics, can disturb the microbiota composition and deplete its biodiversity^15,82. Autologous faecal microbiota transplantation (autoFMT) can help to restore the microbiota¹⁰⁴, which in turn might influence immune recovery by modulating neutrophil, lymphocyte and monocyte populations⁹.

Approaches rooted in ecology could fill the gap caused by the lack of mechanistic understanding. Ecology is the branch of biology that studies how complex ecosystems function through the interactions among constituent species in changing environments. The principles of ecology complement those of cellular and molecular biology, and translating them to the clinical oncology setting is essential for managing the microbiota in patients (TABLE 1). Ecology-based management strategies, originally developed for macroscopic ecosystems, can provide the foundation for maintaining the microbiota amid disturbances. In this regard, the seven pillars of ecosystem management¹¹ offer guiding principles that can be adapted from the context of macroecology to the ‘microecology’ of patients with cancer (TABLE 2). These pillars emphasize the need for dynamic, patient-specific microbiota monitoring, maintaining key microbial functions rather than focusing solely on biodiversity, and leveraging the inherent resilience of the microbiota to treatment-related disturbances. Molecular tools first developed by microbial ecologists to study microbial dynamics^12,13 have subsequently been applied to study human microbiomes¹⁴ and are now being tested in clinical settings¹⁵, where they could be deployed to track the dynamics of microbial ecosystems in patients with cancer and monitor the disturbances caused by cancer treatment. Computational models that describe the dynamics of macroscopic ecosystems¹⁶ can also be applied in the context of microbial dynamics^17–21 and expanded to predict the risk of microbiota-related complications, inform treatment decisions, minimize damage to the microbiota and guide precision interventions to restore the microbial functions damaged by cancer treatment. In this Review, I explore advances in understanding the roles of microbiota in oncology from the past decade and propose a framework for the integration of precision management of the microbiota into oncology care.

TABLE 1 |.

Ecology terms translated to the management of the microbiota in oncology

Ecology term	Definition adapted to the microbiota of patients with cancer	Application to the management of microbiota in patients with cancer
Biodiversity	The variety of microbial species present in a complex microbiota population. A higher biodiversity is often associated with superior outcomes, although this view is not fully accurate.	Biodiversity indices, such as those proposed by Shannon116 and Simpson117, can be easily calculated from microbiome profiling data and provide a summary metric of the state of microbiota in a patient. Tracking how biodiversity changes (for example, through daily faecal sample analysis) can inform oncologists about microbial resilience and potential vulnerabilities during treatment.
Disturbance	Events that disrupt microbiota composition and function, such as antibiotic use, exposure to chemotherapy or dietary changes.	Treatments used for the management of patients with cancer, especially prophylactic antibiotics and chemotherapy, often shift microbiota composition and function. Monitoring disturbances can guide timely interventions to restore the microbiota, such as faecal microbiota transplantation104.
Resilience	The ability of the intestinal microbiota to recover from disturbances, maintaining or quickly regaining its original state.	Resilience can be the objective of clinical interventions using designed mixtures of commensals that not only recover microbial functions damaged by cancer treatment but also form a population that can persist upon future disturbances.
Adaptive management	A strategic approach that involves continuously monitoring and adjusting interventions on the basis of the current state and dynamics of the microbiota. This strategy might be used to optimize treatment outcomes in patients with cancer.	Real-time microbiota monitoring enables precision interventions, such as modifying antibiotic regimens or administering defined microbial consortia to improve response to cancer treatment.
Functional redundancy	The presence of multiple microbial species that perform similar functions within a host. Redundancy can ensure that essential functions are preserved even if some species are lost owing to disturbances.	Oncologists can leverage functional redundancy to maintain beneficial ecosystem services (for example, butyrate production for gut barrier integrity) despite fluctuations in microbial composition. This principle supports targeted interventions that enhance key microbial functions.
Ecosystem services	The key functions and benefits that microbiota provide to the host, including nutrient absorption, protection against pathogens and modulation of the immune system.	Specific microbiota-derived metabolites (such as short-chain fatty acids or tryptophan derivatives) influence immune responses and treatment efficacy. Cancer treatments that affect the microbiota can compromise their ability to deliver services to the host.
Trophic interactions	The feeding relationships between different microbial species, which influence the flow of nutrients and energy through the microbiota, and are crucial for maintaining its balance and function.	Cancer treatments can disrupt trophic interactions, with unintended consequences such as the expansion of opportunistic pathogens. Understanding these interactions can guide microbiota-based interventions.
Niche	The specific role or function of a microbial species within the microbial ecosystem. Each species occupies a niche that includes not only what it does but also how it responds to other microbes and the physical environment.	Certain microbes, such as Akkermansia muciniphila, are linked to enhanced responses to immune-checkpoint inhibitors25. Identifying and supporting their niches could help to optimize treatment strategies.

TABLE 2 |.

The seven pillars of ecosystem management translated to oncology

Pillar	Application to the management of microbiota in patients with cancer
Ecosystem management reflects a stage in the evolution of social values and priorities	Managing microbiota in patients with cancer reflects a broader shift in medical priorities, which emphasize precision medicine, patient-centred care and public health. These approaches seek to balance individual treatment outcomes with public health goals, such as antibiotic stewardship and control of infections in oncology wards.
Ecosystem management is place-based, and the boundaries of the place must be clearly and formally defined	Effective microbiota management in patients with cancer is patient-specific, focusing on the individual’s unique microbiota composition and their health status. In gut-associated cancers and in patients with cancer receiving immune checkpoint inhibitors, the primary ‘place’ is the intestinal microbiota, which directly affects cancer progression and response to systemic treatment. The oral, lung, skin, or vaginal microbiota might also have important roles in other cancers. Defining the appropriate ‘place’ for microbiota management should consider local and systemic effects30.
Ecosystem management should achieve the desired benefits	The goal of microbial management in oncology is to optimize microbial functions to support the effects of treatments for cancer and overall health. These approaches can include interventions such as supplementation with defined mixtures of probiotic microbes, dietary modifications to enhance microbiota resilience and function, and rational administration of prophylactic antibiotics in patients with cancer.
Ecosystem management can exploit the ability of ecosystems to respond to natural and artificial stressors	The resilience of microbiota to cancer treatments and other interventions must be leveraged while recognizing its limits. Strategies should minimize collateral damage from chemotherapy and antibiotics, and support microbial recovery and stability.
Ecosystem management might not necessarily result in an emphasis on biological diversity	High microbial diversity in the intestine is often associated with superior outcomes and loss of diversity is linked to inferior outcomes15, although in some contexts (such as vaginal cancer) higher diversity can be detrimental35. In many cases, the emphasis should be on preserving specific taxa (such as Faecalibacterium and Akkermansia in patients undergoing allogeneic haematopoietic stem cell transplantation)9 rather than on maintaining overall diversity.
The sustainability of interventions should be clearly defined	The benefits from cancer treatments, such as high efficacy and low toxicity, must be weighed against potential disadvantages, including the risk of disrupting the microbiota and long-term effects in overall health.
Scientific information is crucial, but is only one element in a decision-making process of public or private choice	Collaborative decision-making involving health-care providers, patients and their carers is essential to tailor microbiota management strategies to each patient’s needs and values.

Ecosystem management has seven pillars¹¹. The integration of these principles in the routine management of the microbiota in patients with cancer can result in precision approaches that improve patient outcomes.

Host–microbiota: interconnected ecosystems

The human microbiota is composed of microbial populations — each adapted to a specific organ or tissue niche, modulated by factors such as pH, oxygen levels and nutrient availability²². The species compositions of these populations can fluctuate in response to physiological changes, dietary shifts and external challenges, including opportunistic infections by pathogens and treatment with antibiotics, but tend to stay within certain composition limits²³. The stability of the microbial compositions are thought to be partially imposed by human host tissues via mechanisms that sense and control compositional shifts in the microbial populations, and that ensure that the microbiota continues providing benefits to the health of the host (BOX 1). In patients with cancer, the disease and its treatment can disturb both the microbial populations and the mechanisms of host control over their dynamics. Cancer drugs, radiation and antibiotics can affect specific microorganisms and shift microbial populations beyond normal limits. Physiological perturbations in patients with cancer can affect their body’s ability to control microbial ecosystems, exacerbating complications and influencing responses to treatment²⁴.

BOX 1 |. Host control of microbial ecology.

Evolutionary theory challenges the conventional view of the intestinal microbiota as harmonious and stable mutualisms among microbes and also with their host. Natural selection shapes the host and its microbiota, but does not necessarily result in increased mutual benefit. Ecosystem services reflect microbial adaptations to the host environment (for example, in the gut) that can also influence host health. According to this theory, natural selection favours host mechanisms that keep the ecosystem under tight regulation^118,119 through detection of changes in microbial services and subsequent responses mediated by host immunity or physiological changes that affect the composition of microbiota. Some well-characterized mechanisms include:

• Short-chain fatty acids that the host can detect via G-protein-coupled receptors, which can trigger anti-inflammatory responses and strengthen the intestinal barrier^120,121.

• Tryptophan metabolites produced by intestinal microbes that activate the aryl hydrocarbon receptor and modulate immune responses as well as the integrity of the intestinal lining^122,123.

• Bile acids that interact with receptors such as farnesoid X nuclear receptor, thereby modifying bile acid synthesis and, in turn, affecting metabolism and immune responses, which can exacerbate treatment-related complications such as graft-versus-host disease^124–126.

• Diarrhoea, a drastic host response to intestinal damage caused by enteric pathogens resulting in a massive discharge of intestinal microbes¹²⁷.

Natural selection also shapes the microbiota; the strategies they evolve to survive and reproduce can sometimes come at the expense of the host’s health. For example, some strains of Vibrio cholerae can produce diarrhoea-causing toxins that help it spread to other hosts¹²⁸. Vibrio cholerae can also use its type VI secretion system to induce host intestinal movement and expel competing bacteria¹²⁹.

Understanding all the processes that affect the microbiota in humans and that are relevant to oncology presents challenges, although efforts to characterize these processes are underway. For example, the depletion of beneficial microbes such as Akkermansia muciniphila, a key member of the intestinal microbiota, has been associated with inferior clinical responses to immune-checkpoint inhibitors²⁵. Conversely, specific microbial activities can hinder the effects of other cancer treatments. For instance, bacterial β-glucuronidase enzymes can worsen the gastrointestinal toxicity of the DNA topoisomerase I inhibitor irinotecan, complicating treatment for up to 40% of patients taking this anti-cancer medication²⁶.

Researchers might be tempted to test interventions that specifically promote microbes considered beneficial while suppressing those considered detrimental, hoping to improve the efficacy and reduce the toxicity of oncology treatments. Nevertheless, interventions that ignore microbial ecology and the mechanisms of host control are unlikely to succeed in the long term. Strategies to improve the production of short-chain fatty acids (SCFAs) by intestinal bacteria are an example of such an intervention. SCFAs can be seen as a beneficial ecosystem service that generally reduces host inflammation and can modulate mechanisms involved in cancer (for example, they can inhibit histone deacetylases)²⁷. Therefore, increasing the abundance of SCFA producers in the patients’ intestines seems a logical approach. However, SCFA production results from intricate trophic interactions among intestinal bacteria and is context dependent. Primary degraders, such as certain Bacteroides spp., initially break down complex fibres into simpler molecules, and secondary degraders use those products to generate SCFAs including butyrate and propionate²⁸. Moreover, many bacteria, such as members of the Eubacterium, Faecalibacterium and Roseburia genera, produce the same compounds, including butyrate, which indicates that redundancies occur in the ecosystem²⁹. Improving a beneficial ecosystem service such as SCFA production might, therefore, require more actions than simply introducing specific SCFA producers: ecological factors such as the simultaneous interactions among various members and the redundancy of the microbial ecosystem must be carefully considered.

Role of microbiota beyond the intestine in cancer

The link between the microbiota and cancer was first explored in the gut, where it can influence cancer development both locally and systemically³⁰. The gut microbiota is part of the tumour microenvironment in colorectal cancer, directly influencing disease progression and immune responses³¹. In non-gastrointestinal cancers, gut microbes can still modulate outcomes indirectly through their effects on host metabolism and immunity. Metabolic products from the gut microbiota, including SCFAs and tryptophan derivatives, modulate systemic immunity and can influence cancer progression in other organs³². Alterations in gut microbial metabolism can influence host oestrogen metabolism by increasing the reactivation and reabsorption of oestrogens; this may elevate systemic oestrogenic activity, potentially contributing to the development of hormone-sensitive cancers such as breast cancer³³. Integrating microbiome analysis into oncology practice could provide novel insights into such cross-organ microbial interactions³⁰. The gut microbiota has a role both locally and systemically³⁰. Additionally, there is growing evidence that the microbial populations of other organs also influence cancer and its treatment.

Vaginal microbiota.

The vaginal microbiota has a distinct composition, typically dominated by Lactobacillus spp. which produce lactic acid and maintain a low pH that limits pathogenic colonization. Disruptions in this microbiota leading to increased biodiversity and reduced dominance of Lactobacillus have been associated with persistent infection with human papillomavirus and an increased risk of cervical cancer^34,35. Given the potential implications for screening and treatment, microbiota-targeted interventions might be worth integrating in the management of gynaecological malignancies.

Lung microbiota.

The composition of the lung microbiota contains more Proteobacteria than other microbiota populations and its biodiversity can increase with disturbances like smoking and decrease in patients with a history of chronic bronchitis³⁶. Several studies have reported differences in the microbiome of lung cancer patients, including correlations with metastasis; however, the exact differences observed may vary among studies^37–40. For example, a study that sequenced the microbiome of lung tissues, including tumor sites, observed that although the overall lung microbiota composition appeared broadly similar between patients with non-small cell lung cancer and non-cancer controls, there were differences in the relative abundance of rare taxa, such as Lactobacillus rossiae, Bacteroides pyogenes, Paenibacillus odorifer, Pseudomonas entomophila, Magnetospirillum gryphiswaldense, and the fungus Chaetomium globosum⁴¹. Compositional differences like these, once validated by future studies, could reveal possible modulators of lung cancer progression.

Upper aerodigestive microbiota.

The microbiota of the upper aerodigestive tract, including the oral and pharyngeal regions, may play a role in carcinogenesis³¹. Oral microbiota have been associated with an increased risk of head and neck squamous cell carcinoma, with species such as Fusobacterium nucleatum linked to tumour progression and a poor prognosis by promoting immune evasion through interference with T cell-mediated responses⁴². This mechanism is similar to that of F. nucleatum in colorectal cancer⁴³, which is better studied despite F. nucleatum being an oral, not a gut, bacterium⁴⁴. The presence of F. nucleatum in saliva has potential as a non-invasive diagnostic tool for confirming malignancy in patients with suspicious findings^45,46.

Skin microbiota.

The skin microbiota interacts with local immune cells to regulate inflammation and cutaneous homeostasis⁴⁷. A preclinical study suggests that commensal bacteria in the skin, such as Staphylococcus epidermidis, can influence tumorigenesis and immune responses in melanoma, and that engineered bacteria can enhance antitumour immunity by activating local dendritic cells and promoting infiltration of CD8⁺ T cells into tumours⁴⁸. Future studies should explore the use of engineered skin bacteria as adjuvants to immune-checkpoint inhibitors in patients with melanoma. Additionally, systemic microbial effects (such as gut-derived metabolites influencing immune responses) might have an indirect role in cutaneous malignancies. Modulating the gut microbiota with dietary interventions, such as administering inulin, a complex polysaccharide metabolized by specific intestinal microbes that facilitates SCFA production, can improve T cell responses and the efficacy of immune-checkpoint inhibitors in mice⁴⁹. In the future, synthetic biologists might use genetically engineered bacteria to deliver cancer therapeutics like vaccine delivery mechanisms, stimulating immune responses targeted against a specific tumour located in a distant organ^50,51.

The gut microbiome and alloHSCT: a well-studied role

Allogeneic haematopoietic stem cell transplantation (alloHSCT) is one of the earliest treatments designed to leverage the immune system for cancer therapy and is almost exclusively used in patients with haematological malignancies such as leukaemias, lymphomas and multiple myeloma⁵². AlloHSCT seeks to restore the blood cell production system in patients with cancer by replacing diseased or dysfunctional bone marrow with healthy, nonmalignant stem cells from a compatible donor, thereby reconstituting physiologically normal haematopoietic function. The newly established immune system can recognize and attack residual leukaemia cells, a beneficial effect known as the graft-versus-leukaemia response, but it can also target the patient’s own tissues, leading to graft-versus-host disease (GvHD), which primarily affects the skin, gastrointestinal tract and liver⁵². Successfully guiding patients through alloHSCT requires intensive pretransplantation conditioning with approaches including chemotherapy, radiation, immunosuppressive drugs to mitigate GvHD and antibiotic prophylaxis to prevent opportunistic infections. These interventions can cause collateral damage to the intestinal microbiota, which in turn can affect treatment outcomes⁵³. In addition, conditioning regimens cause a period of neutropenia, a substantial loss of systemic immunity that can last for several weeks⁵⁴.

Disturbances in the intestinal microbiota of patients undergoing alloHSCT are among the most dramatic alterations of microbial populations associated with cancer treatment. AlloHSCT recipients are closely monitored, with clinical records providing a detailed timeline of vital parameters such as white blood cell counts, body temperature, stool consistency and microorganisms detected in blood samples. The addition of regular faecal microbiome analysis to these rich longitudinal metadata has provided unique insights into the extreme microbial dynamics and their relevance in treatment outcomes.

Several studies have linked shifts in the microbiota of patients with cancer undergoing alloHSCT to an increased risk of cancer relapse⁵⁵, GvHD⁵⁶ and bloodstream infections⁵⁷, as well as unfavourable overall survival (OS)¹⁵. Microbiome data collected over 15 years in a single cancer center⁵⁸ can amount to an extensive microbiome dataset of >10,000 faecal samples from >1,000 patients, together with detailed metadata about treatments and other clinically relevant events; such data provide a unique opportunity to study aspects including antibiotic-driven shifts, immune system–microbiota interactions and microbiota-related complications directly in patients undergoing alloHSCT. The extreme microbial disturbances observed in these patients reveal the key role of the intestinal microbiota in clinical outcomes (including immune recovery), and prevention of GvHD and opportunistic infections. These insights underscore the potential benefits of microbiota management strategies leveraging ecological principles.

Extreme ecosystem disturbances

Patients undergoing alloHSCT experience substantial losses of intestinal biodiversity⁵⁹. The biodiversity starts dropping before transplantation and can remain low for several weeks after engraftment. Antibiotic prophylaxis is the main cause of biodiversity loss, by decreasing intestinal commensal anaerobes, and enabling the rise of mono-populations of antibiotic-resistant facultative anaerobes⁶⁰. The species that benefit from these massive disturbances, include bacteria such as Enterococcus faecium, Escherichia coli and Klebsiella pneumoniae, and fungal pathogens from the Candida genus. These opportunistic microbes tend to carry genes specialized for their benefit that can render them resistant to antibiotics but also exploit specific nutritional niches without contributing ecosystem services that can result in the prevention of GvHD, such as secondary bile acid production⁶¹. After dominating the intestinal niche, opportunistic microbes can breach the intestinal barrier and translocate to the blood, causing bloodstream infections — a particular problem during the neutropenic period^59,62–64. The loss of intestinal commensal microbes can also leave patients susceptible to colonization by anaerobic pathogens such as Clostridioides difficile, which do not necessarily expand to dominate the ecosystem but produce potent toxins that can cause severe intestinal damage⁶⁵.

Effects on the host

The conditioning regimens that precede alloHSCT result in neutropenia. Transplant engraftment subsequently restores white blood cell counts and immune function. These extreme fluctuations in circulating immune cells and the extreme shifts in microbiota populations that occur simultaneously provide an opportunity to study the interactions between both⁹.

The computational analysis of the daily changes in circulating neutrophil, lymphocyte, and monocyte counts, and of longitudinal microbiome tracking from a large cohort of patients undergoing allo-HSCT, revealed that the abundance of certain gut bacteria, such as Faecalibacterium, Akkermansia, and members of the Lachnospiraceae family (phylum Firmicutes, class Clostridia), including taxa often labeled as ‘Ruminococcus 2’ in 16S ribosomal RNA (rRNA) gene profiling using the SILVA taxonomy⁶⁶, can lead to faster recovery of immune cell counts⁹. This network of interactions was inferred by a computational approach based on principles of mathematical ecology. Although the analysis can detect the direction and the strength of bacterium-immune interactions, it cannot explain the underlying mechanism. Still, the bacteria detected to drive faster neutrophil increases could be leveraged in interventions that improve immune recovery after alloHSCT, even if the exact mechanisms underlying those ecological interactions remain unclear. Conversely, loss of Faecalibacterium and Ruminococcus can increase the risk of GvHD⁶⁷, suggesting a complex link between intestinal bacteria and host immunity. Future work that uncovers the mechanisms by which specific microbes drive immune system dynamics could lead to applications in other treatments, such as leveraging the links between the intestinal microbiota and melanoma ICI responses⁶⁸.

The mechanisms of interaction may lie on the ability of specific intestinal microbes to improve the integrity of the host intestinal barrier, preventing increased intestinal permeability (commonly referred to as ‘leaky gut’ syndrome) and modulating systemic inflammatory responses. Bacteria from the Blautia genus and the Lachnospiraceae family can produce butyrate, which helps to maintain the physical integrity of the intestinal barrier and promotes infiltration of regulatory T cells essential for immune tolerance^69,70. The specific contribution of different taxa to these ecosystem services is a reminder that, despite the importance of maintaining microbiota biodiversity, adapting microbiota management approaches to the alterations relevant to each patient might be even more important.

Shifts in microbial metagenomes

The metagenome comprises all genes within a microbiota, shaping its function and the beneficial ecosystem services it provides to the host. Some of these functions, such as SCFA production and bile acid metabolism, are performed by enzymes encoded by genes conserved among closely related species. These functions can be inferred from phylogenetic profiling, such as the integration of 16S rRNA amplicon sequencing, with predictive bioinformatics tools such as Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2)⁷¹. Nevertheless, relying on reference genomes to predict functional pathways might fail to capture strain-level variations, novel functions present in under-represented taxa, or functions encoded in mobile genetic elements, such as plasmids and bacteriophages, which are crucial for microbial adaptations to dynamic environments.

Some key features of microbes, including antibiotic resistance and virulence factors, can be investigated using metagenomic sequencing and databases such as the Comprehensive Antibiotic Resistance Database⁷² and the Virulence Factor DataBase⁷³. Antibiotic resistance affects the effectiveness of prophylaxis against infections, a leading cause of mortality in patients with cancer⁷⁴. Interestingly, some antibiotic resistance genes are more important than others, which means that the presence of specific genes may matter more than the number of all genes associated with antibiotic resistance (the resistome.) For example, in patients undergoing alloHSCT, higher intestinal biodiversity is correlated with a more diverse resistome but a lower risk of breakthrough bloodstream infections⁶⁴. This paradox arises because many commensal microbes naturally harbour resistance-associated and virulence-associated genes, yet their presence does not necessarily translate into pathogenicity or prophylaxis failure. Specific genes can have direct implications. For example, vanA, which enables vancomycin resistance in Enterococcus spp.⁷⁵, can facilitate intestinal dominance of these species under vancomycin prophylaxis and lead to gut-borne bloodstream infection⁷⁶. Sometimes, even single-nucleotide variations can affect patient outcomes. For example, in a patient who died from multidrug-resistant sepsis after alloHSCT, resistance to aztreonam was linked to the presence of a nalD mutation in the pathogen Pseudomonas aeruginosa detected by whole-genome sequencing of blood culture isolates⁷⁷.

Disturbances in transkingdom ecology

The rise of fungal pathogens, particularly Candida spp., in patients undergoing alloHSCT highlights the importance of understanding transkingdom ecology under conditions of immunosuppression. Longitudinal analysis of patients undergoing alloHSCT revealed that Candida infections in the bloodstream are usually preceded by intestinal expansion of this fungus, which can colonize the niche of the commensal bacterial populations impacted by antibiotic prophilaxis⁷⁸. Patients who have expansion of Candida parapsilosis complex species have higher mortality rates even from causes independent of candidaemia⁶². Furthermore, C. parapsilosis can exhibit heteroresistance to micafungin, a trait characterized by a low-frequency subpopulation of resistant cells that often goes undetected with standard antimicrobial susceptibility tests, which drives intestinal dominance and subsequent candidemia. The detection of heteroresistance on the basis of genomic features of fungal isolates could inform clinical decisions and address prophylaxis failure⁷⁹.

Microbiota management in oncology

The microbial dynamics observed in patients undergoing alloHSCT are particularly dramatic, although similar patterns of microbiota disruption are increasingly reported in patients with cancer receiving chemotherapy, ICIs and other treatments. These observations suggest that the insights on microbiota management strategies obtained from studies of patients undergoing alloHSCT might have broader applicability in oncology.

Monitoring of faecal microbiota

The combination of 16S rRNA amplification and next-generation DNA sequencing was originally developed my microbial ecologists to study complex bacterial populations without requiring culture-based methods⁸⁰. This 16S rRNA amplicon sequencing approach is now a widely used method to profile the composition of the gut microbiota⁸¹. Faecal sample collection is a non-invasive procedure that enables repeated monitoring of the microbiota in patients with cancer receiving treatment. This approach offers valuable insights into the dynamics of microbial populations^7,15,64.

Still, 16S rRNA amplicon sequencing has important limitations; it lacks the resolution needed to distinguish closely related species or strains, and is restricted to bacterial taxa, overlooking archaea, fungi and viruses. Functional inferences drawn from 16S rRNA amplicon sequencing based on predictive bioinformatics tools such as PICRUSt2 (Ref.⁷¹) are helpful, but they can introduce uncertainties compared to direct functional assessments from shotgun metagenomics (as described later)⁸¹. Despite these limitations, studies conducted in the past few years support 16S rRNA amplicon sequencing as a clinically informative approach. For example, in patients undergoing alloHSCT, the biodiversity of gut microbiota assessed by 16S rRNA amplicon sequencing correlates with patient outcomes and those with the lowest biodiversity at engraftment have inferior OS. This correlation was statistically significant in a small cohort of patients (n = 80) from a single centre⁸² and later confirmed in a larger (n = 1,362), multicentre study¹⁵.

Beyond biodiversity, specific bacterial shifts have prognostic significance. Patients with dominance of Enterococcus (>30% relative abundance) have a ninefold increased risk of developing vancomycin-resistant Enterococcus bacteraemia⁵⁹, whereas those with proteobacterial dominance (for example, E. coli, K. pneumoniae and P. aeruginosa) have a fivefold increased risk of Gram-negative bloodstream infections⁵⁷. Conversely, reduced abundances of Clostridia commensals, lower levels of butyrate-producing bacteria, and shifts in the ratio of strict to facultative anaerobes are associated with an increased risk of GvHD⁶¹.

Beyond 16S rRNA amplicon sequencing

Despite the insights that 16S rRNA amplicon sequencing provides, other approaches can be used to monitor the microbiota for oncology-related applications. Fungal, viral and archaeal communities contribute to host–microbiota interactions and thus to disease outcomes but are not captured by 16S rRNA amplicon sequencing. For example, a transkingdom analysis of microbiota dysbiosis in colorectal cancer (CRC) detected bacterial–fungal co-abundances, such as those between Clostridium saccharobutylicum and Talaromyces islandicus, in the microbiomes of CRC patients⁸³. In patients undergoing alloHSCT, fungal amplicon sequencing of the internal transcribed spacer 1 (ITS1) region of ribosomal DNA has revealed that Candida expansions reached a relative abundance of nearly 100% in faecal samples, sometimes 10 days before bloodstream infections⁷⁸.

Viral populations, particularly bacteriophages, have crucial roles in shaping microbial ecosystems and mediating cross-species transfer of antibiotic resistance genes⁸⁴. However, the contribution of viruses in the microbiota dynamics of cancer patients, for example, alloHSCT patients treated with antibiotic prophylaxis, remains largely unexplored. A virome analysis based on total RNA extraction followed by viral sequence enrichment using a probe capture-based system⁸⁵ was used to compare the fecal viromes of 12 colorectal cancer (CRC) patients, before and 6 months after surgery, with those of 26 non-cancer controls⁸⁶. CRC patients exhibited altered viromes, characterized by increased co-abundance network connectivity and disrupted virus–bacteria co-abundances. Notably, some of these alterations persisted even six months after surgery, suggesting lingering dysbiosis that could potentially influence patient outcomes.

Shotgun metagenomics provides a more-comprehensive view of microbial communities relative to 16S rRNA amplicon sequencing by sequencing all the genetic material in a sample, enabling species-level and strain-level resolution and functional annotation. This approach enables the identification of genes involved in virulence, antibiotic resistance and metabolic pathways relevant to cancer therapies⁸⁷, although its higher cost and data-processing requirements limit implementation in routine clinical practice. Long-read sequencing technologies are emerging as alternatives to shotgun metagenomics⁸⁸ and standard short-read 16S rRNA amplicon sequencing⁸⁹, offering improved species- and even strain-level resolution⁹⁰. Long reads also reduce assembly biases, which are systematic errors that occur during the reconstruction of genomes from short sequences.

Spatial microbiomics is another rapidly growing area in microbiology research, enabling the mapping of microbial communities within their tissue microenvironments³¹. Single-cell sequencing techniques promise to revolutionize our understanding of the heterogeneity of microbiota by revealing information on each individual cell in a population of thousands of cells⁹¹. Though still in the early stages, these technologies could evolve into powerful tools for understanding microbiota–host interactions in clinical oncology with unprecedented detail.

Finally, isolating organisms from faecal samples or positive blood cultures for whole-genome sequencing is a time-tested approach that can complement more sophisticated sequencing-based approaches to study processes such as the evolution of mechanisms of resistance to antibiotics in a given patient or to track down infection sources^63,77,92,93.

Metabolomics is a complementary tool to sequencing, enabling the characterization of the small molecules produced by microbial ecosystems. Fecal metabolites can influence host physiology, immune responses and treatment efficacy. For example, indole-3-propionic acid, produced by Lactobacillus johnsonii and Clostridium sporogenes, enhances the antitumour activity of ICIs in mouse models by modulating the stemness of CD8⁺ T cells⁹⁴. The integration of metabolomic and metagenomic data will be crucial for advancing microbiota-based precision oncology.

Despite growing evidence that the microbiota influences cancer progression and treatment responses, microbiome-derived biomarkers have not yet been widely adopted in routine oncology practice. Emerging methods that improve resolution and functional characterization of the microbiome — such as long-read sequencing and integrated multi-omics — offer a promising path to bridge this translational gap and enable the development of clinically actionable, microbiota-informed strategies for patient stratification and therapy optimization.

Visualizing patient data

Visualization aids in interpreting both large-scale clinical microbiota datasets and patient-specific microbial dynamics. A stacked bar plot of microbial compositions can provide a patient-level view (FIG. 2). Analyzing microbiome trends across large patient cohorts to distill clinically actionable insights requires different approaches, such as dimensionality reduction techniques. Traditional methods, such as Principal Coordinate Analyses (PCoA)⁹⁵, have been widely used. The TaxUMAP, an algorithm to project large microbiome datasets onto a 2-D scatter plot that takes into account the phylogenetic tree of gut bacteria, enabled the construction of a detailed atlas of the faecal bacteria composition in >1,000 patients undergoing alloHSCT. This visualization revealed disruptions of intestinal populations, such as dominance of Enterobacteriaceae or Enterococcus, which increase the risk of bloodstream infections⁶⁴.

FIG. 2 |. Longitudinal microbiota and clinical data from a patient undergoing alloHSCT.

This timeline displays the clinical course of a patient undergoing allogeneic haematopoietic cell transplantation (alloHSCT) relative to the day of cell infusion (day 0). The timeline includes the timing of cancer treatments (for example, melphalan or fludarabine phosphate), immunosuppressive drugs (methotrexate) and antibiotics (such as vancomycin, ciprofloxacin, aztreonam and amikacin); and tracks white blood cell (WBC) counts, maximum body temperature (Tmax) and the occurrence of a blood culture testing positive for Escherichia coli, indicating bacteraemia. The stacked bar plots show the relative abundances of bacterial taxa in faecal samples, highlighting relevant shifts in the composition of the microbiota composition, such as the increase in E. coli preceding bacteraemia and subsequent microbial recovery. This personalized timeline illustrates the utility of 16S ribosomal RNA amplicon sequencing and visualization with clinical metadata to monitor patients and anticipate infections, informing decision-making in the management of patients with cancer. The timeline was constructed from published microbiome and clinical metadata⁵⁸.

Additionally, analyzing the changes in the microbiota composition reveals how specific treatments, such as prophylactic antibiotics or chemotherapy, disturb the microbial ecosystem⁷. Understanding how clinical interventions drive changes in microbiota composition may help predict disruptions before they lead to adverse health outcomes.

Ecological network modelling

A computer model able to accurately predict how a patient’s microbiota will shift in response to cancer treatment would bring immediate clinical benefits. Such a model would help to identify the risk of developing microbiota-related complications (such as gut-derived bloodstream infections), select treatments that are less likely to disrupt the microbiota and guide the computer-based design of microbial management interventions to improve patient outcomes. Nevertheless, despite the importance of models to predict microbial dynamics in oncology and other areas, creating accurate such models continues to pose a substantial challenge for microbial ecologists⁹⁶.

Ecological network models posit that the interactions between bacterial species can be used to predict the microbiota compositional dynamics. The generalized Lotka–Volterra model has been widely used for this purpose^17–21. This model can parameterize serial data to quantify pairwise interactions between microbial species as well as the effect of environmental perturbations, such as treatments and changes in diet, on these interactions. However, the generalized Lotka–Volterra model has well-known limitations. It assumes pairwise interactions that remain static across varying conditions and cannot describe nonlinear dependencies⁹⁷. Many factors that influence microbial dynamics in clinical settings, including antibiotic use, dietary changes and immune status, might require more-flexible approaches.

In the past few years, neural ordinary differential equation models⁹⁸ have been developed that might provide a powerful alternative to the generalized Lotka–Volterra model. These models can infer microbial dynamics directly from serial data without predefining interaction structures, making them well-suited for capturing complex ecology. Neural ordinary differential equation models are a flexible approach that, in principle, may be expanded to include metagenomic functional profiling and host immune parameters, thereby improving the accuracy in predictions of shifts in microbiota during cancer treatment.

The acquisition of vast longitudinal patient datasets is crucial in clinical applications and to improve ecological network models, which have a substantial role in advancing our basic understanding of microbial biology. For example, network models parameterized with data from patients undergoing alloHSCT and from experiments in mouse models have identified a species of commensal bacteria that prevents C.difficile colonization⁶⁵. Furthermore, the countervailing hypotheses that are emerging within the field should be integrated in a unified network model (BOX 2)

BOX 2 |. Towards a unified theory of how microbiota affect host physiology.

Several hypotheses have emerged that describe distinct yet potentially overlapping roles of the intestinal microbiota in cancer progression and response to therapy. Ecological network models could provide a structured approach to test these hypotheses and refine a unified theory.

• The two competing guilds theory suggests that intestinal microbial ecosystems can stabilize into two distinct community structures, one favouring health and the other associated with disease¹³⁰. This model captures the idea of resilience and bistability in microbial ecosystems but lacks direct mechanistic explanations for how the microbiota contributes to the initiation of the disease. Ecology network models can be used to study key interspecies interactions that drive the formation and persistence of these guilds and identify ecological tipping points — shifts in microbial composition that might precede oncogenic or treatment-resistant states.

• The TOPOSCORE theory posits that specific topological arrangements of bacterial species within the gut microbiota correlate with response to immune-checkpoint inhibitors (ICIs)¹³¹. Initially developed using samples from patients with non-small cell lung cancer, this theory proposes that metagenomic networks, rather than individual taxa, are more predictive of therapeutic responses. Computer simulations with models such as neural ordinary differential equation models⁹⁸ can infer population dynamics and predict how interventions, such as dietary changes or microbial-directed therapies, might reconfigure these topologies to improve patient outcomes.

• A third major perspective, focused on pro-inflammatory versus anti-inflammatory bacteria, describes the role of the microbiota in regulating systemic myeloid signalling and ICI efficacy¹³². Certain bacteria, such as Prevotella, Alistipes, and Bacteroides species, promote systemic inflammation and immune suppression, in part through LPS-driven inflammatory signaling and the enrichment of genes for mucus degradation and endotoxin biosynthesis. These microbial features correlate with elevated neutrophil-to-lymphocyte ratios and poor response to PD-1 blockade. Other bacteria, including Bifidobacterium spp., Lachnospiraceae, and Ruminococcaceae, are associated with reduced systemic inflammation and improved ICI efficacy. Network-based computer models incorporating immune–microbiota interactions can potentially help quantify these effects.

These hypotheses are not mutually exclusive. A unified ecological model might incorporate elements of all three theories, capturing stability, topological features, and host immune modulation.

Other machine learning approaches might also prove instrumental for managing the microbiota of patients with cancer. For example, discrete-time survival models⁹⁹ have been used to predict bloodstream infection risks in patients undergoing alloHSCT⁶⁴. The discrete-time survival model is a flexible approach that could be expanded with sophisticated models, such as neural networks¹⁰⁰, to leverage longitudinal sequencing data (as it becomes available) together with clinical metadata, and more-accurately compute the risk of patients developing treatment-related complications on the basis of their evolving microbial dynamics.

With the growing availability of microbiome datasets, computational models will continue to have a crucial role in clinical oncology. Future directions include integrating multiomics data — combining microbiome sequencing, metabolomics, immune profiling and patient health records — to develop precision strategies for the management of microbiota. Model interpretation will be essential for extracting novel biological insights; techniques such as SHapley Additive exPlanations¹⁰¹ and Sparse Identification of Nonlinear Dynamics¹⁰² will help researchers interpret complex predictive models of microbial dynamics and clinicians translate them into actionable treatment strategies. By incorporating advanced computational models and larger datasets, microbiota management in oncology can evolve from a descriptive field into a predictive and proactive discipline.

Designing precision interventions

Microbiota-targeted interventions that could be used in oncology range from broad, unspecific approaches such as faecal microbiota transplantation (FMT)^{103,104 105,106} to more-precise, rationally designed consortia of live bacteria. The most common intervention to restore the composition and function of the host microbiota is heterologous FMT, in which faecal matter from a healthy donor is introduced into a patient’s gastrointestinal tract. Heterologous FMT has demonstrated efficacy in overcoming resistance to anti-PD-1 antibodies in patients with melanoma by increasing the abundance of Firmicutes, including Lachnospiraceae and Oscillospiraceae, which are associated with improved responses to anti-PD-1 antibodies¹⁰⁵. This intervention enhances activation of CD8⁺ T cells while reducing the frequency of IL-8-expressing myeloid cells, both of which are associated with better outcomes in patients receiving ICIs. Heterologous FMT, however, carries risks including pathogen transmission¹⁰⁷ and variability in donor microbiota composition.

To mitigate the limitations of heterologous FMT, researchers have explored the use of autologous (auto) FMT, particularly in patients undergoing alloHSCT¹⁰⁴. AutoFMT involves collection of a patient’s own faecal microbiota before undergoing intensive treatments; the microbiota are then reintroduced to aid in microbial recovery. This strategy eliminates risks associated with donor material by reintroducing microbiota previously adapted to the patient.

Rationally designed microbial consortia consist of selected bacterial strains that can colonize and functionally restore microbial composition. Unlike traditional probiotics, which introduce generic species, engineered consortia can be designed with both ecological stability and host interactions in mind¹⁰⁸. Such approaches are being researched for oncology applications, particularly immune modulation in animal models¹⁰⁹. Of note, the term probiotic might be inadequate to describe these therapeutic formulations, considering that it is often associated with over-the-counter supplements of uncertain efficacy. Instead, live biotherapeutic products is a more-precise term for microbiota-based therapies¹¹⁰.

Bacteriophage therapy is also emerging as a promising strategy for modulating the microbiota. Phages can be engineered to target specific bacteria while preserving the ecosystem, providing a precise method to decrease pathogen populations in the microbiota of immunocompromised patients with cancer. Preclinical studies have demonstrated the potential of phage therapy to deplete bacteria linked to poor responses to ICIs¹¹¹.

Synthetic biology and genetic engineering are also opening new avenues in the field of microbial management. Bacteria engineered to produce immune-modulating metabolites or capable of colonizing tumours are being explored as therapeutic tools¹¹². These targeted strategies could help to avoid risks associated with whole-community transplants, such as FMT.

Finally, diet has a crucial role in shaping the composition of the microbiota¹¹³. The administration of prebiotics, such as inulin and resistant starches, selectively enhances the growth of specific bacteria that affect host phenotypes¹¹⁴. These dietary interventions are akin to ecological management strategies, providing resources for specific microbes to thrive while suppressing the expansion of undesirable species, such as opportunistic pathogens. For example, fibre-rich diets promote the development of SCFA-producing bacteria¹¹⁵, essential for maintaining gut barrier integrity and reducing inflammation, both essential factors in oncology.

Conclusions

The microbiota in the gut and other tissues are influenced by cancer treatment and in turn affect cancer outcomes, presenting a potential area of intervention in oncology. The exact molecular and cellular mechanisms underlying these interactions remain elusive, although ecological principles can help to address this knowledge gap. Mathematical ecology, which precisely defines concepts such as biodiversity and interactions, provides mathematical formalisms and computational models of ecosystem dynamics, opening the path to applying ecosystem-management principles in clinical oncology settings (FIG. 3). Routine monitoring of microbiota has potential to bring immediate benefits through the early identification of risk of complications from treatment as well as guiding targeted interventions. Embracing microbial ecology within oncology could revolutionize how we approach cancer care, leveraging the microbiota as a powerful tool in patient management. Given the rapid advances in microbiology research and technology, one can barely imagine a future in which ecosystem-based management of the microbiota is not a key component of integrative cancer treatment.

FIG. 3 |. Integrating microbiota monitoring with computational models to manage the microbiota of patients with cancer.

Computational models can assess the risk of microbial-related complications in patients with cancer during treatment, enabling early interventions to mitigate adverse outcomes. Mathematical ecology-based dynamic modelling of microbiota responses to cancer therapy and antibiotics can optimize treatment choices to minimize disruptions to the gut microbial ecosystem. Finally, rationally designed microbial consortia and targeted interventions can restore microbiota functions impaired by cancer treatments, improving patient outcomes.

Key points.

• Cancers and their treatments affect the patient’s microbiota, which in turn can influence immune responses and clinical outcomes.

• Site-specific microbiota not only in the intestines, but also in tissues including the lungs and skin, affect local immune responses and the progression of cancers in those tissues.

• Allogeneic haematopoietic stem cell transplantation illustrates the substantial disturbances that treatment for cancer can cause on the host microbiota, affecting outcomes.

• Integrating ecology-based approaches for microbiota management in oncology care can reduce complications and enhance the efficacy of cancer treatment.

• Integrating 16S ribosomal RNA amplicon sequencing with metagenomics, metabolomics and other emerging technologies can provide a comprehensive understanding of the role of microbiota in cancer.

• Precision interventions for the management of microbiota, including faecal microbiota transplantation and computer-designed microbial mixtures, might improve cancer treatment responses.