Microbiota-centered interventions to boost immune checkpoint blockade therapies

Immune checkpoint blockade (ICB) therapies are monoclonal antibodies that target and block the actions of key immune regulators, such as programmed cell death protein 1 (PD-1), its ligand PD-L1, or cytotoxic T lymphocyte–associated protein 4 (CTLA-4). They have revolutionized cancer treatment by enhancing the immune system’s ability to recognize and attack tumor cells and have extended patient survival in multiple malignancies, including melanoma (Larkin et al., 2015), lung (Carbone et al., 2024; Hellmann et al., 2019), kidney (Tannir et al., 2024), and bladder (Sharma et al., 2019) cancers. However, primary resistance to ICB remains a challenge, with only about one third of treated patients achieving durable, long-term clinical benefits. Additionally, ICB therapies are frequently associated with immune-related adverse events (irAEs), which may require additional medications for management or even lead to therapy discontinuation if they become severe (June et al., 2017; Marin-Acevedo et al., 2019; Postow et al., 2018). Biomarkers that predict clinical response and methods that effectively ameliorate high-grade irAEs without compromising treatment efficacy are critical unmet medical needs.

Ongoing research continues to demonstrate that the gut microbiome can modulate ICB efficacy and the incidence and severity of irAEs. Particularly for the gastrointestinal irAEs, which represent the most common toxicity associated with ICB (Abu-Sbeih et al., 2020), microbiome modulation may be an effective strategy to alleviate symptoms. This complex community of microorganisms—including bacteria, archaea, eukaryotes, fungi, and viruses—lives in the gastrointestinal tract (Donaldson et al., 2016) and influences systemic and tumor immunity (Derosa et al., 2022; Kim et al., 2023). Therefore, gut microbial dysbiosis - which can be caused by comedications such as antibiotics (Elkrief et al., 2024b) and proton pump inhibitors (PPIs) (Lopes et al., 2023), the malignancy itself (Yonekura et al., 2022), lifestyle factors (e.g., diet and exercise), and age (Thomas et al., 2023; Zhernakova et al., 2016) - can significantly impair the effectiveness of immunotherapies. Therefore, identifying and addressing dysbiosis through novel interventions focused on the gut microbiota could potentially be an effective means of enhancing the therapeutic impact of ICB treatments.

This review focuses on the most widely used and extensively studied ICB therapies, specifically PD-1 blockade and anti-CTLA-4 therapies. We begin by exploring the gut microbiome’s role in modulating ICB efficacy. Next, we briefly discuss methods to detect gut dysbiosis. Building on this, we will examine microbiota-centered interventions (MCIs) that could be implemented to augment the antitumor effects of ICB and improve the likelihood of patients responding to these therapies. We conclude by discussing unresolved challenges and the possible applicability of new technologies in the field.

The gut microbiome’s significance and impact on cancer therapies were first reported by Viaud et al. (2013) in the context of immunogenic chemotherapy (e.g., metronomic cyclophosphamide, oxaliplatin) and Iida et al. (2013) in the context of immunotherapy and chemotherapy (CpG-oligodeoxynucleotide with IL-10 receptor blockade or oxaliplatin). Thomas Gajewski in Chicago, Jennifer Wargo at the MD Anderson Cancer Centre, and our laboratory at the Gustave Roussy Cancer Centre in Paris later published seminal studies in the same 2018 issue of Science, demonstrating direct links between gut microbiota composition and the success of ICB in different cancer populations, including melanoma, non–small-cell lung cancer (NSCLC), and urothelial and kidney cancers (Gopalakrishnan et al., 2018; Matson et al., 2018; Routy et al., 2018b). Since then, three lines of evidence have solidified the gut microbiota’s role in shaping ICB treatment outcomes.

First, antibiotic use near the start of ICB therapy is linked to worse outcomes. This observation has been demonstrated in multiple studies involving thousands of patients (Daillère et al., 2020a; Derosa et al., 2022; Elkrief et al., 2019; Hakozaki et al., 2020; Routy et al., 2018b, 2024). For instance, one study found that antibiotic use before ICB treatment was associated with decreased progression-free survival (PFS) in patients with advanced melanoma (Derosa et al., 2018). Additionally, following the first meta-analysis of nearly 12,000 patients reported by Derosa et al. (2021), a second meta-analysis of ∼46,000 patients confirmed that antibiotics taken within a critical window (60 days before to 42 days after the start of ICB therapy) had the most pronounced detrimental effect on treatment outcomes (Elkrief et al., 2024b). Another meta-analysis showed that PPIs—which also affect gut bacterial diversity—can similarly have a negative influence on ICB therapeutic outcomes (Lopes et al., 2023). Antibiotics and PPIs are well recognized for disrupting gut microbial balance (Silva et al., 2025), with antibiotics in particular often leading to dysbiosis. The observed negative impact on ICB outcomes strongly suggests a causal link between gut microbiome composition and the success of these therapies.

The second line of evidence is that researchers have identified specific bacterial species and strains associated with responses to ICB. Certain bacteria are linked to favorable outcomes, including the mucin-consuming anaerobic bacterium Akkermansia muciniphila (Derosa et al., 2022; Routy et al., 2018b), Faecalibacterium prausnitzii (Gopalakrishnan et al., 2018), and Bifidobacterium spp. (Sivan et al., 2015), particularly Bifidobacterium longum (Matson et al., 2018). These species are enriched in the gut microbiome of ICB responders and contribute to enhanced antitumor effects by promoting cytotoxic T cell infiltration and activating immune signaling pathways involving pattern recognition receptors and the type I IFN pathway. Other bacteria like Enterococcus hirae (Viaud et al., 2013), Bacteroides fragilis (Vétizou et al., 2015), and Eubacterium sp. AM28-29 (Thomas et al., 2023) have been linked to improved responses in preclinical and clinical settings, often through mechanisms such as dendritic cell maturation and cytokine production (including IL-12). Additionally, families like Lachnospiraceae, Turicibacteraceae, and Prevotellaceae are more abundant in breast cancer patients achieving pathologic complete responses to trastuzumab-containing regimens (Di Modica et al., 2021).

In contrast, some bacterial species are associated with resistance to ICB. These include Clostridium hathewayi (also known as Hungatella hathewayi) and members of the group XIV Clostridiaceae family, which are enriched in patients with primary resistance and are often linked to prior antibiotic use (Daillère et al., 2020b; Derosa et al., 2020). Similarly, Bacteroides ovatus, Prevotella copri (renamed Segatella copri), Ruminococcus gnavus (renamed Mediterraneibacter gnavus), and Clostridiumsymbiosum are negatively correlated with ICB efficacy in certain contexts, despite associations with positive outcomes with chemotherapy or other monoclonal antibody therapies in other studies (Di Luccia et al., 2024; Peters et al., 2019; Routy et al., 2018b). Finally, the outgrowth of species in the Enterocloster genus following antibiotic cessation can potentiate the trafficking of immunosuppressive regulatory Th17 cells from the gut to distant tumor sites, thereby limiting antitumor immune responses and impacting ICB treatment performance (Fidelle et al., 2023). These findings further underscore the complex interplay between gut microbiota and cancer therapies and highlight the potential for targeted microbiome modulation to optimize ICB efficacy.

The last line of evidence comes from preclinical fecal microbiota transplantation (FMT) studies, which have provided a more direct link between the gut microbiome and ICB response. Tumor-challenged mice that received FMT from patients responsive to anti-PD-1 had significantly reduced tumor growth compared with mice receiving FMT from nonresponders (Derosa et al., 2020; Gopalakrishnan et al., 2018; Routy et al., 2018b). Additionally, researchers have shown that anti-PD-1 efficacy can be restored in mice that received FMT from nonresponders by supplementing them with specific bacterial species associated with favorable responses, such as A. muciniphila (Derosa et al., 2018; Routy et al., 2018a). These findings have sparked interest in investigating whether FMT could be used as an adjunct therapy to boost the ICB response rate, which will be discussed in more detail below.

Together, these three lines of evidence—antibiotics, direct species effects, and FMT—highlight the gut microbiome’s significant influence in shaping the outcomes of ICB therapy. Clearly, there is a need to find methods of modulating the gut bacteria and identifying which patients would benefit from such therapies.

The broader literature uses the term “dysbiosis” to primarily mean a change or imbalance in the microbiota, including alterations in composition and diversity, loss of beneficial or keystone species, blooms of harmful pathogens, and sometimes shifts in metabolic functions, all of which can negatively impact host health (Hooks and O’Malley, 2017). This broad definition belies important implications, not least of which is that dysbiosis could be either a result of the disease (diagnostic) or a causative agent of the disease. Moreover, the term “imbalance” in this context lacks a precise definition. Does it simply refer to changes in taxonomic composition, or is it better understood as a weakening of host functions that regulate the availability of resources for microbial growth (Winter and Bäumler, 2023)? Hence, a more refined definition is urgently required to better tailor therapeutic options for cancer patients.

Regardless, we can still correlate dysbiotic markers with disease status and treatment outcomes. We have previously published comprehensive reviews on methods for detecting dysbiosis (Silva et al., 2025; Thomas et al., 2023). Here, we will focus only on the most critical aspects.

Techniques like 16S ribosomal RNA sequencing and shotgun metagenomics can profile the composition of the gut microbiota. These approaches also enable the computation of alpha diversity (within-sample diversity) and beta diversity (between-sample differences), thus providing a broad overview of microbial community structure. However, although indices like Shannon or Chao1 can show that immunotherapy responders possess greater microbial diversity (Gopalakrishnan et al., 2018), it has become clear that these metrics alone do not fully capture how individual taxa drive or mitigate disease outcomes, as evident in lung or kidney cancer patients (Gupta et al., 2020). Consequently, researchers employ machine learning (ML) tools (e.g., Random Forest classifiers) and multivariate analyses such as partial least squares discriminant analysis and linear discriminant analysis effect size to identify key taxa associated with treatment responses (Knights et al., 2011; Segata et al., 2011; Wang and Lê Cao, 2023).

Although ML algorithms have improved the predictive accuracy of treatment outcomes, challenges remain regarding the reproducibility and generalizability of microbiome-derived signatures across patient cohorts and immunotherapy regimens (Gunjur et al., 2024; Lee et al., 2022). Indeed, multiple studies have highlighted that no single microbial species can reliably predict treatment response across all contexts, suggesting a complex interplay among microbiome–immunotherapy interactions (Gunjur et al., 2024; Lee et al., 2022). Novel deep learning architectures, such as DeepGeni, have been introduced to enhance both the generalizability and interpretability of microbiome-based classifiers. These models can also help identify microbiota with potential mechanistic relevance to therapy response (Oh and Zhang, 2023). Thus, integrative strategies are essential for understanding the specific bacterial species, genera, or families, and their interactions, that shape ICB responses.

One of the most common strategies has been to examine the prevalence or relative abundance of specific microbial taxa in fecal samples. As noted above, several studies have shown that responding patients typically harbor a rich, diverse microbiome, with higher abundance of beneficial (or healthy) bacteria from the Lachnospiraceae and Oscillospiraceae families (e.g., Ruminococcus, Dorea, Coprococcus, Eubacterium, Roseburia, or Faecalibacterium genera, notably F. prausnitzii), as well as archaea and A. muciniphila. Collectively, these microbes support gut barrier fitness and promote insulinosensitivity. In contrast, nonresponders frequently exhibit a low alpha diversity (poor taxa richness) and relative overdominance of tolerogenic or proinflammatory bacteria such as oral taxa (e.g., bacilli or Streptococcaceae, Veillonellaceae, or Eggerthellaceae family members), Proteobacteria, or Enterocloster spp., which are commonly found across various chronic inflammatory disorders. Indeed, ∼50% of cancer patients inherently exhibit a highly altered gut microbial composition due to aging, systemic changes caused by cancer, or comorbidities and comedications (Thomas et al., 2023; Yonekura et al., 2022).

The Akkermansia species are notable for their correlation with ICB treatment outcomes. Derosa et al. (2022) demonstrated that A. muciniphila (SGB9226) exhibits a nonlinear effect on overall survival in lung and urinary tract malignancies; both its complete absence and excessively high abundance (>the 77th percentile, i.e., >4.779%) correlated with dismal overall survival in these patients. Notably, this “abnormal” relative abundance is accompanied by an increased “exfoliome” (exfoliated host cells and DNA in the stool) and relative dominance of the Enterocloster genus (Birebent et al., 2025). This trichotomization of A. muciniphila and its impact on enteric health suggest the existence of at least two distinct types of dysbiosis and indicate that the condition may be more complex than previously assumed.

Standard ecological or taxonomic metrics do not fully capture these complex nonlinear patterns. This shortcoming recently prompted the development of more nuanced tools, including the Gut Microbiome Health Index (GMHI) and topological score (TOPOSCORE) (Derosa et al., 2024; Gupta et al., 2020). The GMHI classifies a patient’s likelihood of having an unhealthy gut by weighing the relative abundance of “health-prevalent” and “health-scarce” microbes. Compared with standard ecological indices, GMHI has outperformed the Shannon index in distinguishing healthy individuals from those affected by a wide range of conditions (Gupta et al., 2020), though it has yet to be tested extensively in immuno-oncology settings (Silva et al., 2025).

Meanwhile, our laboratory has developed the TOPOSCORE, a patient-specific metric that predicts the likelihood of immunotherapy response across multiple indications. It uses metagenomics data from stool samples to assess gut microbial composition and analyzes species-interacting groups (SIGs) linked to either resistance (dysbiosis, SIG1) or response (eubiosis, SIG2) to immunotherapy (Derosa et al., 2024). SIG1 consists of 37 bacterial species that have been linked with treatment resistance, including members of the Enterocloster genus, and the Streptococcaceae, Veillonellaceae, and Lactobacillaceae families. In contrast, SIG2 encompasses 45 species associated with favorable treatment outcomes, including members of the Lachnospiraceae and Oscillospiraceae families. The TOPOSCORE is the personalized normalized scored difference of SIG1 and SIG2 bacteria that also integrates the relative abundance of A. muciniphila. This network-based approach provides a comprehensive assessment of gut dysbiosis and, by extension, helps predict patient outcomes following ICB by distinguishing microbiome profiles indicative of treatment response or resistance. It not only simplifies the complex interplay of microbial ecosystems but may enable precision stratification, and, providing prospective real-world validation, may allow clinicians to personalize immunotherapy strategies by using MCIs.

Examining the levels of metabolic by-products produced by the gut microbiota or the host–microbe interactions may offer an additional means of detecting dysbiosis. For example, short-chain fatty acids (SCFAs, e.g., acetate, propionate, and butyrate) are potent immunomodulators derived from the fermentation of dietary fiber by gut bacteria (Bachem et al., 2019). Butyrate is known to promote T cell differentiation and effector function (Bachem et al., 2019; Kim et al., 2014). It can also promote the differentiation of colonic regulatory T cells (Tregs) by enhancing histone H3 acetylation at the Foxp3 promoter (Mann et al., 2024), thus having implications for limiting the incidence of gastrointestinal irAEs. When dysbiosis occurs, the abundance of SCFA-producing strains may diminish in stools (e.g., Megasphaera massiliensis, and members of the Lachnospiraceae and Ruminococcaceae families) leading to reduced fecal SCFA levels and impaired Treg cell function, along with compromised effector T cell responses (Silva et al., 2025; Thomas et al., 2023).

Likewise, an accumulation of immunosuppressive secondary bile acids (e.g., lithocholic or deoxycholic acid) or shifts in trimethylamine N-oxide can signal a microbiome tipping into an unbalanced state—often accompanied by compromised inflammatory control and tumor responsiveness (Silva et al., 2025; Thomas et al., 2023; Varanasi et al., 2025). Altered bile acid metabolism can also influence the abundance of soluble mucosal addressin cell adhesion molecule 1 (sMAdCAM-1), which has been linked to antibiotic-induced gut dysbiosis (Fidelle et al., 2023). MAdCAM-1 is expressed in high endothelial venules found in the lamina propria and gut-associated secondary lymphoid tissue (Briskin et al., 1997) and enables retention of α4β7⁺ T cells in the gut (Gorfu et al., 2009; Ogawa et al., 2005). Fidelle et al. (2023) demonstrated that antibiotic treatment downregulates MAdCAM-1 ileal expression in mouse models and the intestinal biopsies of patients collected during routine endoscopic examinations. This downregulation has been associated with alterations in bile acid metabolism due to the “blooming” of Enterocloster spp. following antibiotic cessation. In mouse models, downregulation of MAdCAM-1 ileal expression allowed the exodus of enterotropic α4β7⁺ Treg and regulatory Th17 cell subsets from the gut, which then homed to extraintestinal tumors and promoted resistance to anti-PD-1 treatment. In the clinic, the level of sMAdCAM-1 was demonstrated to be a strong prognostic factor for the ICB response and clinical outcome in NSCLC, kidney, and bladder cancer patients (Fidelle et al., 2023). Thus, monitoring sMAdCAM-1 levels in blood samples could serve as an additional tool for assessing gut dysbiosis.

Evaluating gut dysbiosis through an integrated approach that combines microbial and metabolic analyses will provide oncologists with a valuable means of identifying patients at higher risk of failing to respond to ICB therapies. Once these patients are identified, clinicians can next determine whether they might benefit from MCIs, administered either before or concomitant with ICB therapy.

MCIs are any intervention that aims to modulate the gut microbiome from a dysbiotic state to a more balanced, eubiotic state, consequently enhancing antitumor immunity or mitigating irAEs. They include dietary modifications, FMT, and biological supplementation (e.g., probiotics), and represent a promising approach to improving the outcomes of ICB therapies. By restoring microbial diversity, promoting the growth of beneficial bacterial species, or correcting dysbiosis, MCIs can reshape the gut environment to favor immune activation and potentially amplify the efficacy of ICBs. This emerging therapeutic strategy also offers a personalized approach to cancer treatment. A theoretical workflow from patient diagnosis to MCI decision is demonstrated in Fig. 1. Research is continuing to unravel the intricate relationship between the microbiome and ICB function. Thus, MCIs are gaining recognition as critical adjunct therapies for optimizing treatment outcomes and are being investigated in multiple clinical trials (Table 1).

Diet is intimately linked to the gut microbiome’s composition and can rapidly shift its taxonomic structure in healthy individuals (David et al., 2014). A recent study of ∼24,000 participants found individuals who consumed both meat and plants had higher levels of Alistipes putredinis, Bilophila wadsworthia, and Roseburia torques, all of which have been linked to inflammatory bowel disease and an overall decrease in SCFAs. In contrast, participants who adhered to a vegan diet had higher abundance of species in the Lachnospiraceae family, Butyricicoccus spp., and Roseburia hominis, all of which have roles in fiber degradation and SCFA production (Fackelmann et al., 2025). Additionally, the American Gut Project—a large-scale, crowdfunded citizen science initiative focused on analyzing the human microbiome—collected samples from 10,000 individuals, which were then correlated with lifestyle, health, and disease questionnaires. Self-reported high-plant consumers exhibited a significantly more diverse gut microbiome (Gamrath et al., 2025). Lastly, preclinical mouse studies found that ketogenic diets significantly altered the microbiome’s taxonomic composition by promoting the outgrowth of intestinal A. muciniphila and Roseburia intestinalis at the relative expense of the Lactobacilli genus (Ferrere et al., 2021).

Therefore, dietary changes may be the most straightforward (and least invasive) MCI, as increased fiber intake, reduced consumption of processed sugars and fats, and limited alcohol use can encourage the outgrowth of SCFA-producing bacteria (Silva et al., 2025). Indeed, diet has been shown to have a profound impact on immunotherapy responses; consumption of plant-based foods has been positively correlated with better response to anti-PD-1 therapy in a cohort of Polish melanoma patients, while consumption of dairy products had the opposite effect (Pietrzak et al., 2022). Additionally, melanoma patients in the USA who consumed >20 g of dietary fiber per day before treatment had significantly better PFS than those patients who consumed less fiber. This finding has been recapitulated in murine studies, as tumor-challenged mice fed a fiber-rich whole-grain diet had delayed tumor growth during anti-PD-1 therapy compared with mice fed a fiber-low diet (Spencer et al., 2021).

Given these findings, the next logical step is to evaluate the effects of dietary changes on the outcomes of patients receiving ICB therapies. Several clinical trials are underway to evaluate diet on ICB efficacy (Table 1). NCT04552418 is a pilot study that is supplementing starch to patients diagnosed with solid cancer to improve dual ICB therapy outcomes. NCT05805319 aims to improve dietary fiber intake among NSCLC patients to increase SCFAs, while the fasting mimicking diet for reducing immune related adverse events for cancer patients on immune checkpoint inhibitors (“FMD-ICI Trial,” NCT06438588) is investigating the effect of a proprietary blend (Xentigen) that mimics periodic fasting (Brandhorst et al., 2024) on the occurrence of irAEs. Both trials are actively recruiting participants.

Another option worth investigating is the Mediterranean diet, which consists of ingredients native to the countries around the Mediterranean Sea. It includes a variety of plant foods like vegetables, fruits, nuts, seeds, berries, legumes, herbs, and spices. It also features animal proteins like fish, meat, cheese, and other fermented dairy products from locally raised animals, along with “healthy” fat from olive oil. This traditional diet also incorporates ancient grains, honey, and wine (Gamrath et al., 2025; Sofi et al., 2008). Sofi et al. (2008) performed a meta-analysis of >1.5 million individuals and 40,000 fatal and nonfatal events. They found that greater adherence to the Mediterranean diet was significantly associated with reduced risk of overall mortality and incidence of Parkinson’s and Alzheimer’s diseases. Given the Mediterranean diet’s composition and the health benefits it supports, it may also improve response rates to immunotherapy (Gamrath et al., 2025). Well-planned clinical studies are required to validate this hypothesis.

Any dietary manipulation concomitant or before therapy would be reliant on the patient’s compliance and motivation to adhere to the required diet plan. For example, U.S. guidelines recommend 14 g of fiber per 1,000 calories consumed (U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2020). Increasing the recommendation to 50 g daily should meet the requirements necessary to manipulate gut microbial composition, but this is two to three times the amount consumed today by an average Western person (Gamrath et al., 2025). Increasing fiber intake to this extent is not an easy goal to achieve for people not used to such diets; 50 g of dietary fiber is equivalent to about 11 medium-sized apples or ∼3 cups (225 g) of cooked lentils. Furthermore, in clinical practice, any dietary intervention must consider the patient’s needs. This includes tailoring the diet according to the patient’s age and tolerability, potential allergen susceptibility, and any cultural, religious, or ethical concerns they may have (Kraeuter et al., 2020). Nevertheless, while it can be challenging to adopt lifestyle and dietary changes (especially after receiving emotionally charged news of a cancer diagnosis), many patients still want to make the effort, if only to achieve a sense of control over their situations (Hoedjes et al., 2022). Patients can be motivated to change, and encouragement from family, friends, and healthcare providers can be significant motivating factors (Gamrath et al., 2025).

FMT is being investigated to both boost ICB success and alleviate ICB-induced irAEs. The procedure aims to quickly and dramatically change a patient’s gut ecology by transferring stool from either a healthy donor or a patient with an exceptional response to treatment to the recipient. Notably, the therapeutic use of fecal matter dates back over 2,000 years to ancient China (Davidovics and Hyams, 2013), fell into obscurity, and was later revived in the mid-20th century for treating severe colitis (Eiseman et al., 1958; Jamal et al., 2023; van Nood et al., 2013). More recently, its role in boosting ICB response was first demonstrated in mice when the gut microbiome’s influence on ICB success was suddenly recognized (Routy et al., 2018b, 2023; Vétizou et al., 2015). Comprehensive reviews written by Jamal et al. (2023) and Porcari et al. (2023) provide in-depth explorations of FMT and its emerging role in immunotherapy, and we encourage readers to consult them for additional insights. FMT was first employed as a potential therapeutic option in immuno-oncology practice for refractory ICB–induced colitis, one of the most common irAEs caused by ICB (Marin-Acevedo et al., 2019; Postow et al., 2018; Wang et al., 2018). Traditional treatment strategies typically involve high-dose corticosteroids and biologics like infliximab or vedolizumab, yet FMT has demonstrated promise for patients who were unresponsive to these interventions. For instance, Wang et al. (2018) reported significant symptom resolution in two patients with refractory colitis following FMT, accompanied by reconstitution of the gut microbiome, increased microbial diversity, and a rise in Treg cell populations in the colonic mucosa. Additional studies have corroborated these findings, with patients achieving rapid and sustained symptom relief after FMT, sometimes within 10 days (Dai and Liu, 2022; Groenewegen et al., 2023).

Pioneering studies have already shown the effectiveness of FMT in promoting and restoring the efficacy of ICB in melanoma. For instance, studies have demonstrated that transferring gut microbiota from ICB responders to nonresponders can improve response rates to ICB therapies and overcome resistance. Baruch et al. (2021) reported that about one third of metastatic melanoma patients who were refractory to anti-PD-1 therapy exhibited clinical benefits after FMT from donor patients achieving objective responses. Similarly, Davar et al. (2021) found that combining FMT with anti-PD-1 therapy in refractory melanoma patients led to meaningful changes in gut microbial composition, increased activation of CD8⁺ natural killer-like T cells, recirculation of mucosa-associated invariant T cells, and distinct metabolic shifts associated with enhanced antitumor responses. Lastly, Routy et al. (2023) extended these observations in Canada by investigating FMT from healthy donors rather than from ICB responders. In their phase I trial of treatment-naïve advanced melanoma patients (NCT03772899), a single dose of FMT administered via oral capsules was combined with anti-PD-1 therapy. The treatment proved safe and achieved an objective response rate of 65%, including multiple complete responses. Responders experienced more robust donor microbiota engraftment over time and displayed enriched immunogenic taxa (Routy et al., 2023). This study also underscores the potential of leveraging healthy donor stool (i.e., if Santé Canada regulations forbid the transfer of fecal material from patient to patient) to bolster antitumor immunity in patients naïve to immunotherapy. Collectively, these three studies demonstrate that microbiota modulation through FMT could optimize host immune fitness, making it a powerful adjunct treatment to ICB therapies.

Clinical trials for other indications besides melanoma are currently underway, and their results are eagerly anticipated (Table 1). For example, the PERFORM phase I study (NCT04163289) is evaluating the safety of combining FMT with immunotherapy in treatment-naïve metastatic renal cell carcinoma (mRCC) patients. Preliminary findings from this study indicate that 93.3% of planned FMTs were administered without dose-limiting toxicities, and an objective response rate of 44% was observed, though irAEs occurred in 80% of patients (Fernandes et al., 2022). Notably, the TACITO trial (NCT04758507), the first randomized FMT clinical trial, has provided initial evidence that FMT can enhance the efficacy of targeted therapies combined with ICB in patients with mRCC. In this study, patients receiving FMT alongside axitinib and pembrolizumab showed a significantly higher 1-year PFS (66.7% versus 35.0%, P = 0.036) and improved overall response rates (54% versus 28%) compared with placebo, with a median PFS of 14.2 versus 9.2 mo (Ciccarese et al., 2024). FMT-LUMINate (NCT04951583) is a phase II study evaluating the combination of FMT and ICB in patients with advanced NSCLC and melanoma. Preliminary results have already shown an objective response rate of over 75% in both diseases with limited adverse effects (Elkrief et al., 2024a), supporting the potential for FMT to enhance outcomes in other solid cancers. These findings underscore the broader potential of this MCI to optimize immune responses across different cancer types, laying the groundwork for future clinical applications.

Despite the promising results from these pioneering studies and the trials currently underway, challenges remain. First, the mechanism of action of FMT in immuno-oncology remains only partially understood, despite strong indications that the transfer of immunostimulatory commensals can restore antitumor responses (Derosa and Zitvogel, 2021). The therapeutic mechanism of FMT appears multifaceted, involving not only the restoration of microbial richness but also immune modulation, as evidenced by reductions in CD8⁺ cytotoxic T cells and shifts in the Treg-to–cytotoxic T cell ratio within affected colonic tissues (Halsey et al., 2023). A related challenge concerns quality control and donor screening, with the risk of transmitting multi-drug–resistant organisms or other pathogens (Jamal et al., 2023; Zipkin, 2021). Novel approaches—such as using bioreactors to create “fecal-like” ecosystems or encapsulating specific beneficial bacterial consortia—are being explored to refine MCIs, minimize associated risks, and standardize their efficacy (Zipkin, 2021; Zitvogel and Kroemer, 2024). For example, MET-4 is a consortium of 30 bacterial species that have been co-administered with ICB in patients with advanced solid cancers (NCT03686202) (Spreafico et al., 2023). Although the trial did not reach statistical significance, it demonstrated that MET-4 was well tolerated, with a response rate of ∼35% (6/17 patients) compared with 14% (1/7 patients) in the control arm. Additionally, 9/17 (53%) patients experienced clinical benefit, compared with only 1/5 (20%) in the control group.

Lastly, factors underlying the enterofecal compatibility code between donors and recipients are poorly understood (Elkrief and Routy, 2021). Variables like antibiotic pretreatment (Baruch et al., 2021; Jamal et al., 2023), administration route (D’Haens and Jobin, 2019), donor–recipient genetic variation (Borgers et al., 2022; Wang et al., 2016), possibility of long-term microbial engraftment (Derosa and Zitvogel, 2021), and preexisting immune memory or unintended immunization against gut commensals in the recipient all require systematic investigation.

Probiotics are beneficial live microorganisms that can potentially confer health benefits (Mackowiak, 2013). Some bacterial strains and species have already been found to enhance ICB efficacy. For example, by feeding Bifidobacterium species to mice, Sivan et al. (2015) increased tumor-specific T cells in the periphery and potentiated the accumulation of CD8⁺ T cells in the tumor bed. Additionally, a multicenter analysis in Japan concluded that supplementing patients diagnosed with advanced or recurrent NSCLC with probiotics (specifically Clostridium butyricum, which is expected to generate a dominance of Bifidobacterium spp.) alongside anti-PD-1 monotherapy significantly improved disease control and PFS, and more so in patients who took antibiotics (Takada et al., 2021). Another promising probiotic species is A. muciniphila. In preclinical studies, anti-PD-1 resistance in mice engrafted with FMT lacking A. muciniphila was reversed by its oral supplementation (Derosa et al., 2022). A phase I clinical trial (NCT05865730) is currently underway that is evaluating the daily administration of Oncobax-AK—a capsule containing live Akkermansia massiliensis (strain p2261, SGB9228), which shares overlapping health-related functions with A. muciniphila—in advanced kidney cancer patients who are Akkermansia-deficient in first-line therapy.

Among the probiotics investigated in combination with ICB therapy, C. butyricum MIYAIRI 588 (CBM588) is arguably the most extensively studied in clinical settings and widely used in Asia to improve symptoms related to dysbiosis, like diarrhea or constipation (Seki et al., 2003; Shimbo et al., 2005). Two clinical trials have reported the potential of CBM588 supplementation to enhance response rates in randomized cohorts of mRCC patients receiving nivolumab, either in combination with tyrosine kinase inhibitors (NCT05122546) or as part of dual therapy with ipilimumab (NCT03829111). Ebrahimi et al. (2024) found that supplementation with CBM588 improved the overall response rate (74% versus 20%) and the 6-month PFS rate (84% versus 60%). Interestingly, CBM588 supplementation was associated with increased abundance of Ruminococcaceae, which has been linked to ICB response and metabolomic changes (Ebrahimi et al., 2024). Furthermore, Dizman et al. (2022) found that patients supplemented with CBM588 had significantly longer PFS. CBM588 has also been associated with metabolomic modulation and immune tone regulation, including increased levels of IL-1β, IL-10, and granulocyte-colony stimulating factor. Notably, retrospective epidemiological studies showed an association between CBM588 uptake and restoration of clinical efficacy of ICB in NSCLC patients previously treated with antibiotics or PPIs (Tomita et al., 2020, 2022). In preclinical mouse studies, CBM588 improved intestinal barrier function despite antibiotic-induced dysbiosis and promoted the expansion of IL-17A–producing γδ T cells and CD4⁺ T cells in the colonic lamina propria (Hagihara et al., 2020).

As stated above, the efficacy of probiotics can be mediated through modifications of the metabolome. For instance, while the exact mechanism is unclear, A. muciniphila has been found to elevate concentrations of polyamines, SCFAs, secondary bile acids, and metabolites associated with antiaging and anticancer properties (Grajeda-Iglesias et al., 2021). In melanoma, Lactobacillus reuteri enhances the release of tryptophan catabolites, such as indole-3-aldehyde, which promotes IFN-γ–producing CD8⁺ T cells (Bender et al., 2023). In other mouse studies, Lactobacillus gallinarum improved anti-PD-1 efficacy in colorectal carcinoma models by its derived metabolite indole-3-carboxylic acid (Fong et al., 2023), and Lactobacillus johnsonii enhanced effector CD8⁺ tumor-infiltrating T cells by producing indole-3-propionic acid (another tryptophan metabolic by-product) in multiple cancer models (Jia et al., 2024). Lastly, CBM588 is well known for its production of SCFAs, particularly butyrate, which may play a role in enhancing the efficacy of anti-PD-1 therapy (Tomita et al., 2022).

Probiotics have the potential not only to improve ICB response rates but also to circumvent dysbiosis-mediated ICB resistance or alleviate ICB-induced irAEs. For example, mouse studies have shown that B. fragilis can restore anti-CTLA-4 efficacy after antibiotic treatment or in germ-free mice (Vétizou et al., 2015). In a CT-26 colorectal cancer model, B. longum and Bifidobacterium adolescentis administration reduced antibiotics-induced resistance to ICB (Funayama et al., 2025). Other probiotics like F. prausnitzii (Bredon et al., 2024) or Lactobacillus rhamnosus (Gao et al., 2021) restored ICB responses in the context of antibiotic-induced microbiota perturbation. Probiotics may also be used to alleviate irAEs. For instance, in clinical trials conducted on metastatic colorectal cancer patients treated with chemotherapy or ICB, administration of live Bifidobacterium spp. reduced treatment complications such as nausea and vomiting, loss of appetite, bloating, or diarrhea (Wang et al., 2025). Lastly, oral administration of Coprobacillus cateniformis was able to overcome antibiotic-associated resistance to anti-PD-L1 therapy in mouse models by downregulating PD-L2 expression on dendritic cells, thereby disrupting the inhibitory PD-L2-RGMb (repulsive guidance molecule b) signaling pathway (Park et al., 2023).

Engineered probiotics seem to be a viable alternative to naturally derived microorganisms. In 2020, Gurbatri et al. (2020) engineered an Escherichia coli Nissle 1917 probiotic bacterial strain to release nanobodies targeting PD-L1 and CTLA-4 for local therapeutic activity in preclinical models. Furthermore, Yue et al. (2024) engineered a novel L. reuteri strain designed to selectively target tumors and release PD-L1 nanobodies in a breast cancer mouse model. By combining the two treatments, they decreased the adverse events of ICB treatment by alleviating intestinal inflammation. Lastly, Yang et al. (2024) modified E. coli to enforce murine decoy-resistant IL-18 mutein expression, which induced CD8⁺ T cell– and natural killer cell–dependent immune responses and long-term antitumor responses against bulky cancers.

Despite these encouraging results, further clinical validation studies are required. Safety needs to be primarily considered as some studies have demonstrated an association with an increased risk of bacteremia in intensive care units (Yelin et al., 2019) or tumor growth (Arthur et al., 2013) following probiotic supplementation. Furthermore, at least one study has explicitly demonstrated the potential negative effects of probiotics on treatment efficacy in a retrospective cohort of stage IV melanoma patients treated with ICB (Spencer et al., 2021). These effects were confirmed in the study’s murine experiments; probiotics such as Bifidobacterium or Lactobacillus attenuated the anti-PD-1 response compared to supplementation with a high-fiber diet. Lastly, there is no consensus about which microbes can serve as viable probiotics. Thus, more research is needed before they can be widely adopted to boost ICB efficacy.

Prebiotics are a class of compounds that mainly consist of oligosaccharides, inulin, mucin, starch, and naturally sourced products that support the growth of beneficial bacteria in the gut and are a potentially simple means of modulating the immune response. For instance, a mouse study showed that orally administered inulin induced memory T cell responses and improved responses to anti-PD-1 therapy (Han et al., 2021). In contrast, another study tested inulin and mucin in association with mitogen-activated protein kinase/ERK kinase inhibitors in melanoma mouse models and found both compounds induced antitumor immune responses and reconfigured gut microbiota composition (Li et al., 2020). Notably, these effects were abolished in germ-free mice—confirming the critical role of gut bacterial metabolism—and did not enhance the antitumor effect when co-administered with anti-PD-1 therapy. Thus, more research is required to reconcile these disparate findings.

Other studies have highlighted the potential of naturally derived products as prebiotics. For example, supplementation with castalagin, a polyphenol and the active compound in the camu-camu berry (Myrciaria dubia), enhanced antitumor immune responses and improved anti-PD-1 efficacy in preclinical models (Messaoudene et al., 2022). Castalagin was also found to enrich Ruminococcaceae and A. muciniphila in the gut microbiota. Based on these findings, a clinical trial is currently underway to evaluate camu-camu as a prebiotic in NSCLC and melanoma patients receiving ICB (NCT05303493). Another promising clinical trial is also underway for patients with unresectable melanoma (NCT06466434). This study aims to evaluate the impact of a prebiotic food–enriched diet on the response to first-line ICB therapy and its effect on the abundance of Faecalibacterium species in the gut microbiota. A new randomized study (NCT06049576) being conducted at the City of Hope Medical Centre in California, USA, is investigating camu-camu in combination with ipilimumab and nivolumab in first-line mRCC (Barragan-Carrillo et al., 2024).

Finally, synbiotics are a “mixture comprising live microorganisms and substrate(s) selectively utilized by host microorganisms that confer a health benefit on the host” (Swanson et al., 2020). Synbiotics typically include strains such as Lactobacillus, Bifidobacterium spp., Saccharomyces boulardii, and Bacillus coagulans (Yan et al., 2024). One example is MS-20, a fermentation product generated by a consortium of Lactobacillus and yeast, enriched with secondary metabolites (Lee et al., 2024). When administered alongside anti-PD-1 therapy in mouse models, MS-20 induced shifts in gut microbial composition, increasing members of the Clostridiaceae and Ruminococcaceae families, while also modulating the immune response and enhancing IgA production (Lee et al., 2024). However, further research is needed to optimize synbiotics to boost ICB efficacy.

The gut microbiome profoundly influences ICB, but challenges remain to be addressed before successfully implementing MCIs to enhance their efficacy. A primary issue is the lack of a universally accepted definition of dysbiosis (Brüssow, 2020). As explained above, “dysbiosis” remains a loosely defined concept—an imbalance within the microbiota that may either result from or drive disease, while the precise nature of this imbalance (e.g., altered taxonomic composition vs. disrupted host regulatory controls) is still debated.

Next, we need to better clarify how gut bacteria influence ICB success. Several hypotheses have emerged in the literature. Metabolic by-products like butyrate are essential for T cell memory (Bachem et al., 2019) and may enhance ICB efficacy, though, paradoxically, butyrate supplementation has sometimes diminished anti-CTLA-4 efficacy (Coutzac et al., 2020). Additionally, ICB appears to promote bacterial translocation from the gut to lymphoid organs via dendritic cells (Choi et al., 2023), suggesting that bacteria—possibly mobilized by ICB-induced toxicities (Almonte et al., 2021)—could disseminate to distant tumors and boost antitumor immunity. Moreover, the gut microbiome’s vast antigen reservoir may facilitate cross-reactivity between bacterial and tumor antigens, thereby enhancing antitumor immune responses (Fluckiger et al., 2020; Naghavian et al., 2023).

There is also an urgent need for more judicious antibiotic guidelines for cancer patients and high-risk individuals. Beyond moderation and avoidance of overuse, targeted use of narrow-spectrum antibiotics against specific bacterial strains may mitigate adverse effects on the gut microbiota, particularly since antibiotics given near ICB initiation are especially detrimental (Derosa et al., 2021; Elkrief et al., 2024b; Routy et al., 2018b). Delaying immunotherapy when feasible and evaluating ICB suitability in patients prone to recurrent infections are critical. Moreover, given the TOPOSCORE’s predictive value, choosing antibiotics that spare beneficial SIG2 bacteria may offer therapeutic advantages. When antibiotic use is unavoidable, supplementing with pro- or prebiotics—or employing strategies like the oral adsorbent DAV132 (Messaoudene et al., 2024)—can help preserve gut ecology. More selective alternatives to antibiotics, such as targeted phage therapy or monoclonal antibodies against specific bacteria (Gaborieau et al., 2024; Wang et al., 2022), may also minimize dysbiosis and sustain ICB efficacy.

Lastly, we cannot overlook the rapid advancements in artificial intelligence (AI) technology and its potential to greatly improve our understanding of the microbiome–ICB axis. AI enables researchers to navigate the complexity of multi-omics data, uncovering how microbial composition, metabolic signatures, and host immune responses influence ICB outcomes (Abavisani et al., 2024; Li et al., 2022). AI-driven analyses have the potential to better refine and predict patient responsiveness to ICB through advanced pattern recognition, enabling more personalized treatment approaches. Furthermore, computational tools offer deeper insights into how gut microbiota modulate ICB efficacy, identifying key microbial and metabolic patterns and interactions. Despite these transformative capabilities, AI solutions demand standardized datasets (including standardized sample collection and processing), transparent reporting, and interpretable models to mitigate overfitting risks and address ethical concerns (Probul et al., 2024). Integrating AI into ICB and microbiome research could reshape clinical care, guiding more effective, patient-specific immunotherapy strategies.

The gut microbiome is an important determinant of ICB efficacy, influencing both therapeutic outcomes and the incidence of irAEs. Evidence from clinical and preclinical studies underscores the role of specific microbial taxa and their metabolites in shaping antitumor immunity, highlighting the potential for MCIs to enhance ICB responses. Strategies like dietary modifications, probiotics, prebiotics, and FMT offer promising avenues for improving patient outcomes by modulating gut microbial composition. However, translating these findings into clinical practice requires a deeper understanding of how microbial communities interact with host immunity and how dysbiosis contributes to ICB resistance.

Addressing these challenges will necessitate standardized methods for defining and measuring dysbiosis, as well as biomarkers that reliably predict treatment response. Large-scale clinical trials are needed to validate the efficacy of MCIs, particularly in identifying which patients may benefit most from these interventions. Additionally, the complex interplay between gut microbes, systemic immunity, and tumor microenvironments must be further explored to develop more precise therapeutic strategies. Combining microbiome modulation with other adjunct therapies, such as chemotherapy, targeted agents, or novel immunomodulators, could further enhance ICB efficacy while mitigating treatment-associated toxicities.

Harnessing the gut microbiome’s therapeutic potential represents a transformative step in cancer immunotherapy. As research continues to unravel the intricate connections between microbial ecosystems and immune function, personalized treatment approaches that integrate microbiome modulation could redefine cancer care. By incorporating microbiome-based strategies into immunotherapy regimens, clinicians may improve response rates, extend patient survival, and reduce treatment-associated complications, paving the way for more effective and personalized oncology interventions.

Author contributions: A.A. Almonte: conceptualization, investigation, visualization, and writing—original draft, review, and editing. S. Thomas: conceptualization, investigation, visualization, and writing—original draft, review, and editing. L. Zitvogel: conceptualization, resources, and writing—review and editing.