Microbiome dysbiosis and chemotherapy resistance in acute myeloid leukemia (AML)

Abstract

AML often relapses due to chemotherapy resistance, increasingly linked to gut microbiome dysbiosis. Microbial drug modification, immune modulation, and metabolite-driven survival/epigenetic changes (e.g., SCFAs, kynurenine) promote resistance. Clinical data associate reduced diversity, loss of Faecalibacterium, and Enterococcus overgrowth with poorer outcomes. Microbiome interventions (FMT, probiotics, diet) show promise; priorities are standardizing methods and defining microbe–metabolite mechanisms to guide trials.

Introduction

Acute myeloid leukemia (AML), the most common acute leukemia in adults, is characterized by the aggressive proliferation of abnormal myeloid cells due to complex genetic mutations. Outcomes remain poor, especially in older adults (median diagnosis age 68), with five-year survival rates of only 15–32%, and less than 10–15% in patients over 60^1–6. While standard “7 + 3” chemotherapy can cure 30–40% of younger patients and induce remission in up to 60% overall, its effectiveness and tolerability decrease markedly with age and comorbidities, leading to early mortality rates of 10–25%. Targeted treatments like FLT3 and IDH1/2 inhibitors, and especially venetoclax-based agents provide alternatives for vulnerable patients, but aggressive nature of AML and comorbidities continue to limit their efficacy^1,7–13.

Despite new therapies, treatment failure remains a major problem in AML. Drug resistance remains a major obstacle in AML management: primary resistance, often due to genetic factors, impedes initial remission, while secondary resistance drives relapse through emerging leukemic clones^14,15. Fewer than one-third of adults achieve sustained remission, highlighting the need for new strategies to prevent resistance and relapse^4,7,14,16. While combination therapies are increasingly favored to improve efficacy in AML, they also increase risks of toxicity, drug resistance, and complications, including damage to the gut lining and microbiome that may further compromise patient health^17–20. Therapy-induced dysbiosis is particularly significant, as chemotherapy often depletes beneficial bacteria like Faecalibacterium prausnitzii and Bifidobacterium spp., with these changes persisting after treatment and heightening infection susceptibility. Lower microbial diversity and dominance by pathogenic or antibiotic-resistant species are linked to higher infection and mortality rates during induction chemotherapy and post–stem cell transplantation, while greater diversity and specific beneficial taxa correlate with improved survival, reduced graft-versus-host disease (GVHD), and better outcomes^21–24. Consequently, this review focuses on how gut microbiome alterations specifically influence chemotherapy resistance in AML, synthesizing mechanistic evidence and clinical correlations and highlighting microbiome-targeted interventions that may restore chemosensitivity.

Clinical and preclinical studies report reduced gut microbial diversity in AML; complementary human cohort and fecal-metabolomics studies further link specific taxonomic shifts and amino-acid/SCFA changes to AML pathophysiology, while murine work shows that restoration of butyrate-producing taxa can improve barrier integrity^25–27. Notably, microbiota-focused interventions are showing promise for restoring microbial balance, lowering infection risks, and improving clinical outcomes in AML patients²⁸. Overall, the dynamic interplay between genetic, environmental, and microbial factors not only shapes the gut microbiota but also influences leukemia treatment and prognosis, presenting the microbiome as a novel target for prevention, diagnosis, and therapy in hematologic malignancies²⁹.

These findings position the gut microbiome as both a mediator of AML treatment outcomes and a promising therapeutic target, underscoring the urgent need for innovative clinical strategies that incorporate microbial considerations into personalized leukemia care. Accordingly, our review aims to comprehensively examine the interplay between the gut microbiota and AML therapies, highlight emerging interventions to restore microbial balance, and discuss how these advances could enhance patient recovery and optimize the efficacy of chemotherapy and targeted treatments.

Mechanisms of microbiome-driven chemoresistance in AML

Microbial modulation of drug metabolism

The vast enzymatic repertoire of the gut microbiota can significantly impact the pharmacokinetics and pharmacodynamics of various chemotherapeutic agents used in AML treatment³⁰. This microbial influence can lead to either drug activation or inactivation, ultimately affecting their therapeutic efficacy and contributing to resistance³¹.

Daunorubicin, an anthracycline antibiotic, is a cornerstone of the standard “7 + 3” induction regimen for AML. Its efficacy and toxicity are intricately linked to its metabolism, a process historically attributed to host enzymes, particularly the cytochrome P450 (CYP450) superfamily located primarily in the liver³². However, the gut microbiota itself encodes a diverse repertoire of bacterial cytochrome P450 enzymes (BacCYPs) that catalyze a wide range of oxidative transformations on endogenous and xenobiotic substrates. Although BacCYPs are widespread and catalytically versatile, direct evidence for BacCYP-mediated metabolism of anthracyclines in the human gut remains limited; most reported microbial transformations of anthracyclines involve reductive deglycosylation rather than P450-mediated oxidation^33,34. While direct evidence for microbial P450-mediated metabolism of daunorubicin in the human gut is still emerging, compelling data from related compounds provides a strong basis for this mechanism. Gut bacteria possess analogous reductases capable of performing a reaction known as reductive deglycosidation. This process cleaves the daunosamine sugar from the anthracycline’s core structure, rendering the drug inactive³⁵. For instance, studies have shown that certain gut bacteria, including Raoultella planticola, Klebsiella pneumoniae, and strains of Escherichia coli, can metabolize the structurally similar anthracycline doxorubicin³⁶. This biotransformation often occurs via reductive deglycosylation, a process that cleaves the daunosamine sugar moiety from the aglycone core, yielding metabolites with significantly reduced toxicity and anti-cancer activity³⁷.

Irinotecan (a topoisomerase I inhibitor) is sometimes used in relapsed or refractory AML. Its interactions with the gut microbiome are therefore clinically relevant. The case of irinotecan provides a powerful example of how microbial drug metabolism can induce a state of “toxic resistance.” The liver converts irinotecan to the active metabolite SN-38. The host then conjugates SN-38 with glucuronic acid to form the inactive SN-38G, facilitating excretion³⁸. However, this detoxification process is reversed by the gut microbiota. Numerous gut bacteria, including species from the phyla Firmicutes, Bacteroidetes, and Proteobacteria, produce the enzyme β-glucuronidase (GUS)³⁹. In the gut, bacterial GUS enzymes cleave the glucuronide moiety from SN-38G, regenerating the highly toxic SN-38 directly within the intestinal lumen⁴⁰. This localized reactivation of SN-38 is the primary driver of the severe, dose-limiting gastrointestinal toxicities associated with irinotecan, including debilitating diarrhea and mucositis³⁸.

The gut is a major hub of folate metabolism, with various bacterial species both producing and consuming folate, creating a complex metabolic network⁴¹. The dysbiotic state characteristic of AML often involves an expansion of the phylum Proteobacteria, which includes Escherichia coli⁴². Many strains of E. coli are intrinsically resistant to methotrexate, largely due to the presence of the highly efficient AcrAB-TolC multidrug resistance efflux pump, which actively exports the drug from the bacterial cell⁴³.

Immune system interactions

Depletion of immunomodulatory taxa (e.g., Faecalibacterium, Akkermansia)

Dysbiosis in AML often features loss of butyrate producers such as Faecalibacterium prausnitzii. These bacteria maintain gut homeostasis and support normal immune responses⁴². For instance, F. prausnitzii is a major producer of butyrate, a SCFA with potent anti-inflammatory properties that helps support the function of regulatory T-cells (Tregs), which are essential for preventing excessive inflammation⁴⁴. In AML patients, the depletion of butyrate-producers like Faecalibacterium leads to a breakdown of this barrier, a condition known as “leaky gut”²⁴. This allows bacterial components, most notably lipopolysaccharide (LPS), to translocate into the circulation⁴⁵. Preclinical murine data indicate that microbiome-driven increases in circulating LPS can accelerate leukemic progression; human microbiome profiling and metabolomics studies corroborate that AML patients show consistent shifts in taxa and metabolites that would promote barrier dysfunction and systemic inflammation^25–27. The role of other bacteria, like Akkermansia muciniphila, is more complex. While often beneficial, its pre-chemotherapy expansion in AML patients has been paradoxically linked to an increased risk of neutropenic fever, possibly by degrading the protective mucus layer of an already inflamed gut²³.

Overgrowth of pathobionts (e.g., Enterococcus) promoting T-cell exhaustion

Multiple cohorts report Enterococcus expansion and reduced butyrate-producers in AML; these human observations align with immunophenotyping showing T-cell dysfunction in patients and with mechanistic data linking dysbiosis to immune exhaustion^26,42,46. At diagnosis, circulating CD8 + T-cells from AML patients exhibit clear signs of exhaustion, a state characterized by the high expression of multiple inhibitory receptors (e.g., PD-1, Tim-3, 2B4) and senescence markers (e.g., CD57), coupled with reduced proliferative capacity and effector function. This exhausted phenotype is a strong predictor of poor response to chemotherapy and is only reversed in patients who achieve a deep and durable remission^46,47. Fusobacterium species, particularly F. nucleatum, are also frequently elevated in AML patients, including in pediatric AML cohorts, and have been associated with disease relapse⁴⁸. These bacteria can induce pro-inflammatory signaling pathways and, importantly, contribute to chemoresistance by promoting autophagy in cancer cells. Autophagy is a cellular process that can help cancer cells survive stressful conditions, including chemotherapy-induced damage^49,50. For clinical associations linking these mechanisms to outcomes in AML patients, see Section 3.

Cytokine dysregulation: reduced IL-17/IL-22 (critical for mucosal immunity)

The integrity of the gut mucosal barrier is paramount for preventing the systemic dissemination of microbial products that drive inflammation and immunosuppression. This barrier is actively maintained by the mucosal immune system, which relies on specific cytokine signals, notably Interleukin-17 (IL-17) and IL-22⁵¹. Notably, some commensal bacteria metabolize tryptophan into indole derivatives that activate AhR and potently induce IL-22 production by ILC3s and T cells, thereby promoting antimicrobial peptide production, epithelial regeneration and barrier function, a microbiota-dependent pathway that directly links tryptophan metabolism to the IL-22, mediated mucosal defence described above^52,53. These cytokines are produced by specialized immune cells, including T helper 17 (Th17) cells and type 3 innate lymphoid cells (ILC3s), in response to signals from the commensal microbiota. IL-17 and IL-22 act on intestinal epithelial cells to stimulate the production of antimicrobial peptides (e.g., RegIIIγ), enhance the mucus layer, and reinforce the tight junctions that seal the barrier. In AML, the profound dysbiosis and the direct cytotoxic effects of chemotherapy disrupt this delicate balance, leading to a vicious cycle of mucosal collapse. The decimation of the commensal bacteria that are required to stimulate Th17 and ILC3 cells leads to a reduced capacity to produce protective IL-17 and IL-22. This cytokine deficiency compromises the integrity of the intestinal barrier, resulting in a “leaky gut” phenotype⁵⁴. A compromised barrier allows for increased translocation of pro-inflammatory microbial components, such as LPS from Gram-negative bacteria, from the gut lumen into the systemic circulation. This flood of microbial products fuels the systemic inflammation that drives T-cell exhaustion and creates a pro-survival, pro-inflammatory tumor microenvironment that directly supports chemoresistance. This self-perpetuating feedback loop, where treatment-induced dysbiosis leads to mucosal immune failure, which in turn fuels the systemic inflammation that promotes resistance, is a central mechanism by which the microbiome undermines the efficacy of AML therapy²⁷.

Metabolite-mediated resistance

SCFAs, acetate, propionate, and butyrate, are produced by bacterial fermentation of dietary fiber. Butyrate, made by Faecalibacterium and Roseburia, is depleted in many AML patients and can act as a natural HDAC inhibitor with pro-apoptotic effects on myeloid cells. These observations suggest SCFA loss may affect therapy response^{24,25,27,42,55}. Butyrate is a natural HDAC inhibitor, a class of compounds known for their anti-cancer effects. By inhibiting HDACs, butyrate triggers programmed cell death (apoptosis) in AML cells by shutting down the critical pro-survival Akt signaling pathway⁵⁶. In stark contrast, acetate, another major SCFA, appears to play a pro-survival role. Recent studies have revealed a remarkable metabolic crosstalk wherein AML cells can co-opt bone marrow stromal cells, prompting them to secrete acetate⁵⁷. The AML cells then take up this acetate and use it as an alternative fuel source to feed the tricarboxylic acid (TCA) cycle and support lipid biosynthesis, thereby enhancing their metabolic fitness and survival⁵⁸. Butyrate is primarily produced by bacterial fermentation of dietary fiber by obligate anaerobic commensals (for example, Faecalibacterium prausnitzii and Roseburia spp.). In addition to serving as the main energy source for colonic epithelial cells, butyrate functions as a natural HDAC inhibitor. Preclinical studies indicate that butyrate can induce apoptosis in myeloid leukemia cell, partly via HDAC inhibition and downstream modulation of pro-survival signaling (for example, Akt), and that maintenance of butyrate-producing taxa is associated with improved gut barrier integrity and reduced systemic inflammation, factors that may limit leukemic progression^27,42,56. By contrast, acetate, also produced by microbial fermentation but which can additionally arise from metabolic crosstalk with bone marrow stromal cells, can be co-opted by AML cells as an alternative carbon source. Stromal cell–derived acetate is taken up by leukemia cells and can be metabolized in the TCA cycle or used for lipid biosynthesis, supporting cellular fitness under nutrient or therapeutic stress. This metabolic flexibility implicates acetate metabolism (and enzymes such as ACSS1/2) as potential therapeutic targets in AML^57,58.

Tryptophan, an essential amino acid, is a key substrate for both host and microbial metabolism, and its metabolic fate is a critical checkpoint in immune regulation⁵⁹. A major route of tryptophan catabolism is the kynurenine pathway, which is initiated by the enzymes indoleamine 2,3-dioxygenase 1 (IDO1) and tryptophan 2,3-dioxygenase (TDO)⁶⁰. The gut microbiota is a major site of tryptophan metabolism and can significantly influence the kynurenine pathway⁵⁹. Kynurenine is a potent endogenous ligand for the Aryl Hydrocarbon Receptor (AhR), a transcription factor expressed in various immune cells, including T-cells. Activation of AhR in T-cells by kynurenine is profoundly immunosuppressive; it promotes the differentiation of naive T-cells into immunosuppressive Tregs, inhibits the function of cytotoxic T-cells, and contributes to T-cell exhaustion⁶¹. In addition to the host-driven kynurenine pathway, the gut microbiota directly converts tryptophan into several indole derivatives (for example, indole-3-aldehyde, indole-3-acetic acid, indole-3-propionic acid and related compounds) that also act as ligands for AhR. Microbiota-derived indoles (notably indole-3-aldehyde produced by certain Lactobacilli) activate AhR in ILCs and T cells, promoting IL-22 transcription and epithelial repair programs that preserve mucosal barrier integrity. Thus, microbial indoles provide a complementary, microbe-derived AhR → IL-22 axis that supports mucosal homeostasis and can counterbalance immunosuppressive or pro-tumorigenic AhR signalling from kynurenine in certain contexts (see 2.2.3)^53,59,62. This pathway is a well-established mechanism of tumor immune evasion across many cancer types⁶³. Kynurenine, produced through host and microbiota-influenced catabolism of tryptophan (via IDO1 and TDO pathways), is an endogenous ligand of AhR. AhR activation by kynurenine in T cells promotes regulatory T-cell differentiation and contributes to T-cell dysfunction and exhaustion, thereby fostering an immunosuppressive microenvironment that can facilitate AML immune escape. Targeting the kynurenine–AhR axis (for example, with IDO inhibitors or by modulating microbial tryptophan metabolism) therefore represents a potential therapeutic strategy^{59–61,63,64}. Bile acids, synthesized in the liver from cholesterol, undergo extensive⁴² modification by the gut microbiota. Host-synthesized primary bile acids (e.g., cholic acid) are converted into secondary bile acids (SBAs), such as deoxycholic acid (DCA) and lithocholic acid (LCA), exclusively by bacterial enzymes, primarily from Clostridia species⁶⁵. These SBAs are not inert waste products but are potent signaling molecules that can modulate host physiology and pathophysiology⁶⁶. Certain SBAs, particularly DCA, have been shown to promote pro-inflammatory and pro-survival signaling pathways, most notably the Nuclear Factor-kappa B (NF-κB) pathway⁶⁶. The NF-κB signaling cascade is a master regulator of inflammation, immunity, and cell survival. Critically, the NF-κB pathway is constitutively activated in approximately 40% of AML patients and is a key driver of leukemic cell proliferation, evasion of apoptosis, and chemoresistance⁶⁷.

Epigenetic and genomic influences

Epigenetic plasticity, the ability to change gene expression without altering the DNA sequence, is a hallmark of cancer that allows malignant cells to adapt to therapeutic pressures and acquire resistance⁶⁸. One of the key mechanisms of epigenetic regulation is histone acetylation, which is controlled by the opposing activities of histone acetyltransferases (HATs) and HDACs. Butyrate, in particular, plays a complex and paradoxical role in cancer, capable of both fighting and fostering the disease. The SCFA butyrate is a well-characterized natural inhibitor of class I and II HDACs⁶⁹. By inhibiting HDACs, butyrate promotes histone hyperacetylation, which generally leads to a more open chromatin structure and can alter the expression of genes involved in cell cycle arrest, differentiation, and apoptosis. Indeed, studies have demonstrated that treatment with sodium butyrate can induce apoptosis in human myeloid leukemia cell lines, highlighting its potential as an anti-leukemic agent⁷⁰. This function casts the dysbiosis of AML in a new light. As established, AML is characterized by a profound depletion of butyrate-producing bacteria and, consequently, lower levels of butyrate in the gut⁴². However, this same epigenetic reprogramming harbors a dark side: the “chemoresistance paradox.” While forcing some cells toward death, butyrate-induced HDAC inhibition can also grant other cells a dangerous level of “epigenetic plasticity”⁷¹. Instead of dying when exposed to chemotherapy, a subset of AML cells can enter a dormant, drug-tolerant persister (DTP) state⁷². These DTP cells are not genetically resistant but are phenotypically primed to endure the treatment, eventually leading to relapse⁷³. The ultimate engine of acquired chemoresistance is clonal evolution, a process of Darwinian selection driven by the continuous acquisition of new genetic mutations⁶⁸. Factors that increase the rate of mutagenesis can therefore accelerate the emergence of resistant clones. AML cells are often intrinsically genomically unstable. For example, common mutations like FLT3-ITD lead to the overproduction of intracellular reactive oxygen species (ROS), which are potent DNA-damaging agents. This high intrinsic ROS burden, combined with defects in DNA damage repair pathways, creates a hypermutable state that fuels clonal evolution⁷⁴. The dysbiotic gut microbiome adds a powerful extrinsic layer to this process. The chronic inflammation and microbial translocation associated with dysbiosis are potent sources of systemic ROS³⁹. These microbially-generated ROS can directly cause DNA damage in host cells⁶⁵. This leads to a “synergistic genotoxicity” hypothesis for chemoresistance in AML. The multifaceted mechanisms linking gut dysbiosis to chemoresistance in AML, spanning drug metabolism, immune evasion, and metabolite-mediated pathways, are summarized in Table 1. A conceptual overview of microbiome-driven mechanisms in healthy versus dysbiotic gut states during AML is depicted in Fig. 1.

Table 1.

Mechanisms of microbiome-driven chemoresistance in AML

Mechanism	Key microbial factors	Impact on chemoresistance	Supporting evidence
Drug metabolism	Raoultella, Klebsiella, E. coli	Inactivate daunorubicin via reductive deglycosidation	Preclinical models (anthracycline inactivation)35–37
	Bacterial β-glucuronidase (GUS)	Reactivate SN-38 (irinotecan), causing toxicity	Clinical GI toxicity data38–40
Immune dysregulation	Depletion of Faecalibacterium	Reduced butyrate → impaired Treg function → leaky gut	AML mouse models (LPS translocation)27,42,45
	Overgrowth of Enterococcus/Fusobacterium	T-cell exhaustion, autophagy induction	Human AML T-cell profiling46,48
Metabolite shifts	↓ Butyrate (HDAC inhibition)	Loss of AML cell apoptosis	In vitro leukemia cell studies27,56
	↑ Acetate (stromal crosstalk)	Fuels TCA cycle/lipid synthesis in AML cells	Metabolic flux analysis57,58
	Kynurenine (tryptophan pathway)	AhR activation → T-cell exhaustion/Treg differentiation	Immunosuppression studies61,63,64

Fig. 1. Dysbiotic gut microbiota promotes chemoresistance in AML.

Left panel (Healthy gut): Butyrate-producing bacteria (e.g., Faecalibacterium prausnitzii, Clostridia) maintain intestinal barrier integrity, inhibit HDAC, and induce apoptosis in AML cells. Immunomodulatory taxa (e.g., Akkermansia muciniphila) support Treg and IL-22/IL-17-dependent mucosal immunity, while CD8⁺T cells remain functional. Right panel (Dysbiotic gut in AML): Loss of butyrate producers and overgrowth of pathobionts (e.g., Enterococcus, E. coli, Fusobacterium nucleatum) drive chemoresistance through: (i) Drug metabolism: Enzymatic inactivation of anthracyclines (e.g., daunorubicin) and reactivation of toxic metabolites (e.g., SN-38 from irinotecan) via bacterial β-glucuronidase (GUS); efflux pumps (AcrAB-TolC) confer methotrexate resistance. (ii) Immune dysfunction: Depletion of SCFAs and increased LPS translocation promote “leaky gut,” NF-κB activation, T-cell exhaustion ( ↑ PD-1/Tim-3/CD57), and autophagy in AML cells. (iii) Metabolite shifts: Accumulation of acetate (fueling TCA cycle/lipid biosynthesis in AML cells) and kynurenine (activating AhR→Treg differentiation). These pathways collectively enhance AML survival and relapse risk.

Clinical evidence linking dysbiosis to chemoresistance

Clinical cohort and interventional studies in AML document that reduced gut microbial alpha-diversity and depletion of SCFA-producing taxa are associated with higher infectious complications, lower remission rates, and shorter event-free and overall survival. Representative clinical reports include large cohort analyses linking microbiome signatures to infectious risk (e.g., Galloway-Peña et al., 2020) and interventional/autologous-FMT studies that restore diversity after intensive chemotherapy (e.g., Malard et al., 2021). For mechanistic details (barrier dysfunction, LPS translocation, metabolite and immune pathways) see Section 2. Low abundance of gut bacteria in chemotherapy-treated AML patients is linked to a poor response to therapy and a higher risk of infection⁴⁵. Chemotherapy remission rates fall as they get less diverse. Indeed, this dysbiosis can contribute to stimulating the progrowth of those opportunist pathogenic bacteria that sensitize the patient toward infections, sepsis, and mucositis⁷⁵. Section 3 focuses on AML-patient clinical associations between gut dysbiosis and outcomes. AML-specific clinical signals include decreased alpha diversity and depletion of SCFA producers associated with gut barrier dysfunction and weight loss in AML patients⁴⁵. Correlations between microbiome composition at diagnosis and infection risk / event-free survival have been reported in pediatric AML cohorts⁴⁸. Interventional strategies (FMT, probiotics, pre/pro/postbiotics) are discussed in Section 4. Clinical correlations between dysbiosis features and AML outcomes, alongside emerging interventional data, are consolidated in Table 2.

Table 2.

Clinical evidence in AML patients and immediate hematologic contexts linking dysbiosis to outcomes

Dysbiosis feature	Clinical association	Intervention studies	Key findings
↓ Microbial diversity	↑ Infection risk, ↓ remission rates45,169	FMT	Restored diversity; ↓ infections170–172
↑ Proteobacteria	↑ Inflammation, relapse risk45,48	Probiotics (Lactobacillus rhamnosus GG)	↓ Mucositis, ↑ treatment tolerance85–87,173
↓ SCFA producers	Barrier dysfunction, weight loss45,174	Prebiotics (resistant starch)	Enhanced butyrate; improved barrier integrity175
Antibiotic exposure	↑ Neutropenia, ↓ OS/EFS45,80,162	Antibiotic stewardship	Preservation of keystone taxa; ↓ resistance76,84

AML cohorts consistently report expansion of Proteobacteria (including Enterobacterales such as E. coli) during induction chemotherapy and antibiotic exposure; these blooms correlate clinically with barrier dysfunction markers, bloodstream infection risk, and worse short-term outcomes in several cohort studies. (See cohort examples cited in Table 2 and Pötgens et al., 2023; Galloway-Peña et al., 2020.) For mechanistic pathways linking these changes to chemoresistance, see Section 2. Broad-spectrum antibiotic exposure during induction chemotherapy correlates with reduced microbiome diversity and worse clinical outcomes (increase infections, prolonged neutropenia, shorter EFS/OS) in multiple AML cohorts. This resistance makes it difficult to control the infections and also results in a fatal condition⁷⁶. Patients with extreme neutropenia, a consequence of cytotoxic chemotherapy characterized by fever and leukopenia, face a greater chance of bacteremia. Carbapenem usually comes via intravenous injection. Conversely, the extended running of these antibiotics may exacerbate neutropenia and heighten an individual’s susceptibility to infections from other bacterial strains⁷⁷. Using too many broad-spectrum antibiotics can put pressure on the bacteria in our bodies, which may help create and spread antimicrobial resistance (AMR). This means that the bacteria become less responsive to these antibiotics, making the infections they cause harder to treat. The ultimate outcome is that the bacteria demonstrate reduced sensitivity to these antibiotics, rendering the disorders they induce ever more challenging to treat⁷⁸. Moreover, antibiotic-resistant bacteria may result in significant or even fatal infections, including sepsis and invasive candidiasis. One of the most worrying threats, especially for patients with hematological malignancies, is the emergence of carbapenem-resistant Enterobacterales (CRE)⁷⁹.

The consequence of antibiotic acquaintance during induction chemotherapy on overall survival (OS) and event free survival (EFS) has been of concern, both for hematologic and solid organ malignancies⁸⁰. Neutropenia stirring secondary to induction chemotherapy can apartment the patient at risk for neutropenic states and the potential for infection. The conventional method to contract with such complications is prophylactic or therapeutic antibiotic treatment. But antibiotics can destroy the normal gut microflora, which is vital to the immune system⁸¹. Changes to the microbiome could change the approach the body responds to cancer treatment, with implications on outcomes. Several clinical cohorts report that early or prolonged exposure to broad-spectrum antibiotics (notably carbapenems) correlates with reduced microbiome diversity and with worse outcomes (prolonged neutropenia, higher infectious complications and shorter EFS/OS), findings that align with experimental data showing antibiotics can impair the anti-tumor activity of cytotoxic agents through immune and metabolic mechanisms^82,83. These lines clinically provide the idea of superinfections to inform the antibiotic treatment of cancer patients with febrile neutropenia, advocating for the enlarging of antibiotic spectra by identifying etiologies. This approach, particularly in immunocompetent children, should serve as an alternative to broad-spectrum antibiotic therapies. Further research is essential to clarify the interaction between antibiotic medication, microbiota composition, and treatment outcomes in chemotherapy. Research is investigating methods to retain the microbiome while simultaneously reducing the risk of infections⁸⁴.

Numerous case reports have revealed significant advantages of probiotics for patients with hematologic malignancies previous to the commencement of chemotherapy. One patient with AML who consumed probiotics throughout chemotherapy exhibited a reduction in gastrointestinal infections and improved tolerance to treatment, including diarrhea and other digestive issues related to chemotherapy⁸⁵. In another case report, a patient with AML was also treated with probiotics and chemotherapy, and the gut flora diversity was significantly restored. It was also linked to better clinical results, such as more immunologic function recovery and less febrile neutropenia. The impact of probiotic supplementation on these parameters as well as on the patients’ quality of life and clinical outcome was assessed in a randomized controlled study during the period of aplasia in patients having AML after induction chemotherapy The initial findings showed that participants treated with antibiotics and who received probiotics developed significantly less antibiotic associated diarrhea and were significantly less possible to suffer an infection than those who did not take probiotics⁸⁶. Another study confirmed the immunomodulatory properties of probiotics in patients with different types of cancer undergoing chemotherapy, including AML. The alteration of immunologic index and gut microbiota composition were clearly observed, and treatment with probiotics could promoting the increased numbers of beneficial bacteria and enhancing immunity function that are associated with beneficial effects on treatment outcomes⁶⁰. A group of patients with hematologic malignancies receiving chemotherapy were given specific species of probiotics to investigate the potential of probiotics to decrease chemotherapy associated complications such as mucositis and infections. Patients received probiotics, there were fewer complications, and they were better able to tolerate chemotherapy. Mechanisms of action of probiotics include the restoration of the gut microbiota disorders caused by chemotherapy, increased immune function (Antitumor activity of probiotics in cancer and construction of a tissue) leading to an improved ability of the body to fight infections and cancer, and attenuation of gut inflammation, helping to ameliorate the gastrointestinal toxicity of chemotherapy⁸⁷.

Therapeutic strategies to overcome microbiome-mediated resistance

Microbiome restoration

Disordered gut microbiota alters the absorption and metabolism of chemotherapeutic agents, which may cause elevated toxicity and reduced drug efficacy. Optimizing the microbiota can enhance treatment tolerance, improve bone marrow function, and support normal hematopoiesis^82,88. Therefore, any approach aimed at microbiome restoration could increase chemotherapy response²⁹.

Fecal microbiota transplantation (FMT)

Most clinical FMT studies focus on allo-HSCT because GVHD is a major problem. However, FMT may also help reverse chemotherapy-induced dysbiosis and reduce microbiome-driven chemoresistance in AML. Intensive cytotoxic regimens and broad-spectrum antibiotics cause profound loss of microbial diversity and depletion of butyrate-producing taxa; these changes promote local mucosal injury, systemic inflammation, epigenetic alterations (HDAC-related), and expansion of pathobionts that express drug-modifying enzymes (e.g., GUSs, reductases) or multidrug efflux systems, all mechanisms that can reduce chemosensitivity. By restoring diversity and replenishing key SCFA-producing organisms, FMT administered after intensive chemotherapy has the potential to (i) reconstitute butyrate levels and their HDAC-modulatory effects that favour leukemia cell apoptosis, (ii) reduce blooms of bacteria that inactivate or re-activate drug metabolites, and (iii) improve mucosal barrier function and systemic immune recovery, mechanisms linked to improved drug response in preclinical models and early clinical work^27,28,37,38. Early interventional reports (including small phase I/II and single-arm studies) suggest that FMT can restore microbial diversity and reduce infectious complications after intensive chemotherapy; however, systematic reviews and meta-analyses in the allo-HSCT and immunocompromised settings emphasize that the current evidence is heterogeneous (variable donor selection, route and formulation, small and frequently non-randomized cohorts) and that robust randomized data with prespecified clinical endpoints (relapse, progression-free and overall survival, and safety in severely immunocompromised hosts) are still lacking, therefore FMT remains a promising but not yet proven strategy to prevent microbiome-mediated chemoresistance^89–91. The aim of FMT in allo-HSCT patients includes eliminating antibiotic-resistant bacteria, restoring microbiota diversity, and prophylaxis and treatment of acute graft versus host disease (aGVHD)⁹². Occurring in 40–60%, one of the main obstacles to allo-HSCT success is aGvHD⁹³. A long list of factors influences aGVHD occurrence, and the impaired microbiome and microbial metabolites are among the most significant factors⁹⁴. Conditioning regimen, supportive medications, especially broad-spectrum antimicrobial therapy, are the most important reasons for developing dysbiosis and subsequently aGVHD⁹⁵. Table 3 shows clinical trials on the FMT in the GVHD setting.

Table 3.

Clinical trials on the FMT in the GVHD setting

Trial identifier	Phase / Status	Patient population	GVHD setting	FMT Type / Route	Comparator	Primary outcomes	Key findings / Notes	Related article
NCT06026371 (USA)	Phase I / Ongoing	Early post-transplant patients	Prophylaxis	Oral encapsulated FMT	None (single-arm)	Microbiota engraftment; severe GVHD incidence	Donor-dependent efficacy observed	98
NCT04935684 (USA)	Phase II / Ongoing	Adults post-allo-HSCT	Prophylaxis	Healthy donor FMT / enema	Control: no FMT	GVHD-free, relapse-free survival	First randomized prophylactic FMT trial	176
NCT05067595 (USA)	Phase I / Ongoing	Gut GVHD (steroid-responsive/dependent)	Treatment	Oral or colonoscopic ± fiber	Randomized	Safety, microbiota diversity	First study combining FMT and fiber	—
NCT04139577 (USA)	Phase I / Completed	Treatment-naïve or steroid-refractory GI GVHD	Treatment	Oral capsules	None	Feasibility; ORR; safety	70% CR; donor microbiota expanded	94
NCT03148743 (China)	Phase I / Ongoing	10–60 y/o post-HSCT	Treatment	Not specified	Observational	Stool changes	Tracking efficacy and safety	177
NCT03812705 (China)	Phase II / Completed	Steroid-resistant/dependent	Treatment	Colonoscopy or gastroscopy	None	ORR and time to response	Repeated cycles evaluated	—
NCT04014413 (China)	Pilot / Ongoing	Multiple dysbiosis conditions incl. GVHD	Treatment	Not specified	Open-label	Symptom improvement	No GVHD-specific data yet	—
NCT04269850 (Russia)	Phase I/II / Terminated	Grade III–IV GI GVHD	Treatment-naïve (FMT + ruxolitinib + steroids)	Oral capsules	None	ORR; OS; AEs	Terminated early	96,97
NCT03678493 (USA)	Phase I / Completed	Steroid-refractory acute GI GVHD	Treatment	Donor-derived / enema	None	Safety; clinical response	65% achieved CR at day 28	—
NCT03359980 (France)	Phase IIa / Completed	Grade III–IV steroid-refractory	Treatment	Frozen pooled donor/enema	None	GI response at day 28	29% achieved CR + VGPR	100
NCT03819803 (USA)	Phase I / Ongoing	Refractory GVHD post-allo-HSCT	Steroid-refractory	Colonoscopic/terminal ileum	Observational	Remission at day 90	44.4% sustained response	-
NCT06938165 (China)	Pilot / Completed	Steroid-refractory GVHD	Treatment	GI route not specified	None	ORR at 12 weeks	Awaiting data	—

AE adverse event; aGVHD acute graft-versus-host disease; allo-HSCT allogeneic hematopoietic stem cell transplantation; CR complete response; FMT fecal microbiota transplantation; G-CSF granulocyte colony-stimulating factor; GVHD graft-versus-host disease; IM intestinal microbiota; LGI lower gastrointestinal; OS overall survival; PFS progression-free survival; RCT randomized controlled trial; SCFA short-chain fatty acids; SDI Shannon diversity index; TRM transplant-related mortality.

This table includes only completed or ongoing interventional trials with publicly registered protocols. Trials with a status of “Withdrawn” or “Unknown” on ClinicalTrials.gov were excluded.

Combination therapy of FMT with different treatments has also been studied. Early clinical experiences combining FMT with ruxolitinib in steroid-refractory intestinal aGVHD, from small case series to 10 patients, suggest timing may influence outcomes. Despite some toxicities, this approach shows promising response rates and microbiome restoration^96,97. The feasibility and safety of early FMT along with systemic corticosteroids were demonstrated for high-risk aGVHD. Although dysbiosis recovery was observed, the small population and concurrent administration of corticosteroids hinder confidence in the efficacy of FMT in this trial⁹⁴.

Several factors come into play in FMT outcome, including taxonomic composition of donor, Microbiota-inherent engraftment ability, Fecal specimen storage, and nutritional condition^98,99. It was reported that FMT from AML patients to AML mice resulted in leukemia progression²⁷. Therefore, it is crucial to carefully screen donors for any conditions, including inflammatory diseases, autoimmunity, and cancers. In a Phase 2 clinical trial of FMT for steroid-refractory-aGvHD, it was revealed that pooling donors is safe, feasible, and creates higher taxonomic richness compared to a single donor¹⁰⁰. We summarize the GVHD literature (Table 3) because the allo-HSCT experience supplies the largest body of clinical and safety data for FMT; systematic reviews and recent pooled analyses indicate generally high response rates in steroid-refractory GVHD but also underline substantive heterogeneity in protocols and follow-up, and the majority of evidence comes from small, sometimes non-randomized series, consequently, lessons from the transplant setting are informative but not directly generalizable to non-transplant AML chemotherapy, and high-quality randomized trials (with relapse/GVL endpoints and standardized donor/processing criteria) are required before routine adoption^89,101,102.

Due to the susceptibility of AML patients to a wide range of complications, it is difficult to precisely determine whether the short and long-term adverse events are associated with FMT or not¹⁰³. Moreover, one of the key dynamics in transplantation is the delicate balance between GVHD and the graft-versus-leukemia (GVL) effect. Several studies have reported an association between these two effects. For example, a recent study showed that moderate grade 2 GVHD has been linked to a lower risk of relapse, likely due to its immune-mediated anti-leukemic activity¹⁰⁴. This prompts an important question regarding interventions such as FMT, which lessen GVHD by boosting anti-inflammatory responses¹⁰⁵. Might reducing GVHD via FMT also diminish the advantageous GVL effect and increase relapse risk? Exploring the association between the intestinal microbiome and transplantation outcome, Eubacterium limosum was significantly associated with a reduced relapse rate. Interestingly, this association was strongest after a T-cell-replete graft, indicating the probable effect of the microbiome on graft versus tumor¹⁰⁶. So FMT may not only weaken the GVL, but also potentially enhance this effect. Overall, GVHD is a devastating and, in higher grades, fatal disease. Therefore, GVHD management, despite the increased risk of relapse, is urgent; long-term studies with mature follow-up patients are required for this.

A 2022 systematic review reported a promising overall response rate of 82.4% after FMT in patients suffering from aGVHD¹⁰³. FMT has also been administered as palliative therapy to reduce the toxicity of immunotherapy in 12 cancers, including one hematological cancer¹⁰⁷. Overall, this microbiome intervention method could be considered for alleviating immunotherapy in leukemia settings in future investigations.

Importantly, donor selection and timing are critical: preclinical data indicate that FMT from AML patients can accelerate leukemia in murine models (a cautionary finding), and transplant-setting FMT may modulate the balance between GVHD and GVL. Therefore, FMT as an adjunct to chemotherapy should be tested in controlled trials with careful assessment of relapse risk, engraftment of beneficial taxa (e.g., butyrate producers), and functional metabolomic endpoints.

Probiotics/Prebiotics

Probiotics refer to live microorganisms that support host health when consumed in the proper dose^88,108. Prebiotics are natural food components that support the growth of probiotics and boost the immune system¹⁰⁹. Postbiotics are defined as preparations made from non-living microorganisms or their components that still provide health benefits to the host¹¹⁰. Synbiotics refer to the combination of prebiotics and probiotics for therapeutic goals in humans and animals¹⁰⁹. Recent systematic reviews and meta-analyses show that selected probiotic formulations reduce the incidence and severity of cancer therapy–related oral mucositis and therapy-associated diarrhea, and may lower antibiotic-associated diarrhea in some cancer cohorts; nevertheless, randomized trials are heterogeneous in strain composition, dosing, timing, and patient selection, and many excluded severely immunocompromised participants, limiting direct applicability to AML patients undergoing intensive induction. Importantly, safety concerns, including rare cases of probiotic-associated bacteremia in neutropenic patients, mean that recommendations for probiotic use in AML should be cautious and individualized pending larger trials in this population^111–113. Taken together, meta-analytic evidence supports symptomatic benefits of probiotics for some therapy-related toxicities, but there is currently insufficient high-quality randomized evidence to conclude that probiotics (or other microbiome supplements) reduce relapse or chemotherapy resistance in AML; targeted, strain-specific RCTs that include immunocompromised patients are needed^112,113. The effectiveness of pro-, pre-, and postbiotics as a combination and supportive therapy has been mentioned. They can induce apoptosis in cancer cells, reduce the colonization of pro-cancer microorganisms, and lessen damage to healthy cells caused by oxidative stress¹¹⁴.

In a single-blinded pilot study including 9 AML and 51 ALL pediatric patients, a probiotic supplement was evaluated in cases with normal neutrophil counts. The use of the probiotic Lactobacillus rhamnosus GG was effective in minimizing the gastrointestinal manifestations¹¹⁵. The restoration of Lactobacillus species as oral probiotics was effective in the management of leukemia-induced muscle atrophy in an acute leukemia mouse model¹¹⁶. Evaluating the effect of Lactobacillus casei on drug-resistant Candida albicans from AML patients revealed a time-dependent beneficial property. The antifungal effect of Lactobacillus casei was significantly stronger on fluconazole-resistant species than on amphotericin B-resistant ones¹¹⁷.In a double-blind, placebo-controlled trial on Pediatric Patients with hematologic malignancies, including two AML, lactoferrin was found to be safe and effective in improving dysbiosis and potentially reducing pathobionts, such as Enterococcus. Therefore, compounds with prebiotic-like effects could reverse the intestinal barrier damage^118,119.

Lactobacillus plantarum KLDS1.0318 could balance the intestinal immunity and metabolic profile in BALB/C mice treated with Cyclophosphamide. The mechanism of this probiotic could be considered cytokine secretion stimulation and Th1/Th2 regulation. L. plantarum KLDS1.0318 increases the richness of Bifidobacterium and Lactobacillus while reducing Escherichia and Enterococcus¹²⁰. Multiple studies have addressed the effect of different probiotics in mice treated with Cyclophosphamide^121,122. Nevertheless, there is a clear need for further prospective studies on other chemotherapeutic agents.

Considering a report from an AML patient suffering from a severe oral infection caused by Lactobacillus rhamnosus, which was linked to dairy intake, careful administration of probiotics is critical¹²³. While probiotics hold promise in cancer care, several challenges still need to be addressed, including accurate strain identification, antibiotic resistance gene transfer, bacterial translocation risks, inconsistent mucosal colonization, and the need for personalized approaches based on host microbiota⁹⁹.

Other agents may also indirectly balance the microbiota hemostasis. Employing transgenic mice with the potential to secrete a mucus-strengthening protein (Tg222), these animal models showed improved mucosal barrier repair, higher citrulline levels, preserved microbial diversity, and reduced bacterial translocation post-chemotherapy compared to wild-type controls¹²⁴. Evaluating a xenograft AML model revealed that curcumin strengthens intestinal integrity through enrichment of Acidophilus, B. bifidum, and L. reuteri. This microbiome effect of curcumin increases cytarabine sensitivity indirectly^125–127. Faecaliibacterium administration to AML mice didn’t result in any improvement of outcome; however, gavage with butyrate could delay the relapse of disease^27,128. Butyrate has been shown to induce T-regulatory cells, suggesting a local effect of this metabolite rather than systemic modification of the immune system¹²⁹. Therefore, gavage of metabolite cocktails may be a beneficial microbiome-based approach for future studies⁹².

Dietary interventions

Considering the evidence linking chemotherapy resistance with body mass index (BMI) and obesity, it is reasonable to conclude that diet also plays a crucial role in disease outcomes. The impact of diet extends beyond BMI, obesity, and nutritional antioxidant status; it also influences the structure and abundance of the microbiota. A microbial fingerprint is generated from multiple sources, particularly dietary¹³⁰. A nicotinamide (NAM) rich diet also may induce resistance to APO866, a prototypic Nicotinamide Phosphoribosyltransferase (NAMPT) inhibitor on AML cell lines¹³¹. This suggests dietary interventions may indirectly influence leukemia resistance. The most important Dietary interventions include Intermittent fasting, fasting-mimicking diets (FMD), ketogenic diets, and fiber-rich Mediterranean diets, which have been shown to have the potential for balancing gut microbiome^132,133.

A pilot study on a high-fat ketogenic diet before and during induction therapy revealed that this nutritional intervention is not only safe but also increases DNA damage in blast cells and decreases non-leukemic lymphocytes from senescence¹³⁴. Moreover, it was revealed that the diffuse large B cell lymphoma mouse model receiving CD19 CAR-T cells and a ketogenic diet showed enhanced OS. Mechanistically, β-hydroxybutyrate, a fundamental metabolite in the ketogenic diet, supports CAR T cell expansion and subsequently leads to superior tumor control¹³⁵.

The focus of dietary restriction experiments has been mainly on ALL rather than AML. A mouse model study showed that fasting inhibits autophagy and increases the effectiveness of vincristine¹³⁶. In another study, fasting induces differentiation and selectively inhibits leukemia only in ALL mice, without showing beneficial effects on AML, indicating that the effect depends on the cancer type¹³⁷. The scenario could be complex since there is a growing body of autophagy-inducing effects of nutritional restriction^138,139. On the other hand, autophagy may be a survival mechanism causing drug resistance in AML¹⁴⁰. Therefore, future studies may provide awareness about fasting as an adjuvant therapy. Moreover, a comprehensive picture of the network between dietary interventions, gut microbiome, and autophagy requires future studies.

Pharmacomicrobiomics

Different aspects of microbiome in treatment, including pharmacokinetics, pharmacodynamics, and pharmacological toxicity, are called Pharmacomicrobiomics¹³². Gut microbes can interact with chemotherapeutic drugs via several ways, including transport, immunomodulation, metabolism, enzymatic degradation, reduction of microbial diversity, and ecological leapfrogging⁸⁶. Mechanisms of chemoresistance involve several aspects, such as the influence of specific microbiota on drug inactivation, tumor microenvironment reshaping, and immune response modulation^141–143. Several microorganisms play a significant role in immune system regulation. For example, Bacteroides fragilis interacts with TLRs and FOXP3; therefore, it affects immune tolerance¹⁰³. So, it is reasonable to consider dose escalation in patients showing an overall immune-inhibitory microbiome.

In 2014, an interesting finding from a mouse model revealed an undeniable role of the microbiome in tumor immune response. Successful systemic immunity after cyclophosphamide is highly associated with gut microbiota. Cyclophosphamide induces the selective translocation of several gram-positive bacteria to secondary lymphoid organs and polarization toward a Th1 and Th17. Next, these bacteria stimulate pathogenic Th17, indicating the significant role of microbiota in anticancer immune response¹⁴⁴. A preclinical study demonstrated an association between cyclophosphamide and microbiome by supporting the idea that microbiome-activated T cells are responsive to similar tumor antigens. Cyclophosphamide has been shown to trigger the translocation of Enterococcus hirae from the intestinal lumen to secondary lymphoid tissues, where it drives the activation of TMP1-specific CD8⁺ T cells and promotes IFN-γ production. This microbiota-dependent immune stimulation appears to be essential for the drug’s full antitumor activity. Interestingly, this effect is lost when broad-spectrum antibiotics are administered but can be restored by reintroducing E. hirae strains that express the TMP1 epitope. Notably, tumors engineered with mutations in the PSMB4 antigen, disrupting its cross-reactivity with TMP1, escape immune recognition, leading to resistance against therapy¹⁴⁵.

One of the challenges of leukemia treatment is that drugs or their metabolites may be potential substrates for bacterial enzymes and can be altered into toxic agents. This undesirable alteration not only damages the epithelium, causing adverse events, but also hinders dose intensification and limits drug efficiency. Because select commensals contribute to drug metabolism and to host anti-tumor immunity, indiscriminate eradication with broad-spectrum antibiotics may be counterproductive; where possible, strategies that spare key commensals (or protect them using microbiome-sparing adjuncts) should be explored as alternatives to blind depletion^146,147. Other supportive therapies, such as Opioids, worsen the toxicity¹⁴⁸. However, inhibiting a specific enzyme responsible for producing toxic metabolites could be a better strategy to prevent damage to the epithelium. It was shown that an oral inhibitor of microbial GUSs could reduce the toxic effect of CPT-11 in a mouse model^149,150. However, many factors need to be evaluated carefully, including the effect of inhibitors on the other symbiotic microbiota and mammalian corresponding enzymes, the optimal dose of inhibitors, and detailed pharmacokinetics studies. Evaluating 9 AML cell lines, it was shown that intestinal bacteria are responsible for APO866 resistance. Two factors are essential for the reduced leukemic cell death, including NAPRT expression and bacterial access to NAM. The concurrent inhibition of NAMPT and NAPRT or the inhibition of NAMPT and oral antibiotics could restore the anti-cancer activity of APO866¹³¹.

Some studies indicate that restoring microbial diversity and SCFA production might be important for developing dendritic cells (DC)-based immunotherapies⁴². Evaluating a dendritic cell line and mouse bone marrow-derived DCs revealed that SCFAs strengthen the antigen uptake and presentation in DCs¹⁵¹. In contrast, studying the microbiota composition of 14 healthy donors along with their DC properties revealed a negative effect of microbiome diversity and SCFA-producing taxa on the generation of immunogenic DCs¹⁵². Another example is that cross-reactivity between a tumor antigen and an intestinal bacteriophage prolongs the anti-tumor response of PD1 inhibitors¹⁴⁵. Therefore, microbiota regulation in order to optimize patients’ anti-leukemia response may be an undeniable stride to personalized medicine¹⁵³.

Antibiotic exposure, stewardship, and direct links to chemoresistance

Prophylactic and empiric antibiotics remain essential in AML because febrile neutropenia and sepsis can be fatal. However, growing evidence shows that antibiotics can do more than cause dysbiosis, they can also impair chemotherapy efficacy through several defined mechanisms. Antibiotics frequently cause loss of microbial diversity and selective overgrowth of pathobionts (for example Enterococcus spp.), with three pathways linking this ecological shift to chemoresistance in AML: (i) Loss of beneficial taxa and metabolites. Broad-spectrum antibiotics deplete butyrate-producing commensals (e.g., Faecalibacterium, Roseburia) and thereby reduce microbiome-derived HDAC-inhibitory and immunoregulatory signals that sensitize leukemic cells to apoptosis and support effective anti-tumor immunity; loss of these signals can promote drug tolerance^82,147. (ii) Selection for bacteria that modify drugs or carry resistance mechanisms. Antibiotic-selected blooms of Proteobacteria and Enterococcus increase the frequency of bacteria encoding drug-modifying enzymes and efflux systems (and expand the resistome), raising the risk that bacterial metabolism or sequestration will alter drug bioavailability or re-activate toxic/less effective metabolites^45,146. (iii) Disruption of microbiome-dependent anti-tumor immunity. Animal models and human correlative studies show that antibiotic treatment can abrogate microbiota-dependent immune priming (for example, the microbiota required for maximal activity of cyclophosphamide and other agents), leading to weaker anti-leukemia immune responses and diminished therapeutic effect^83,154. Clinically, prolonged or early exposure to broad-spectrum agents (notably carbapenems in some AML cohorts) is associated with reduced gut diversity, higher rates of bloodstream infection and colonization by resistant organisms, prolonged neutropenia, and worse short-term outcomes; these clinical observations align with the mechanistic pathways above and suggest that antibiotic exposure is a modifiable contributor to poorer treatment responses^45,82.

Practical implications and priorities. Based on current evidence, stewardship strategies in AML should be framed not only to reduce infections and AMR, but to preserve microbiome-dependent mechanisms that support chemosensitivity. Priority actions to test in trials include: (1) minimize duration of broad-spectrum therapy when clinically safe and preferentially use targeted agents guided by cultures/rapid diagnostics; (2) evaluate microbiome-sparing adjuncts (for example, orally administered beta-lactamases or targeted enzyme inhibitors that protect gut commensals) and selective microbial enzyme inhibitors where appropriate; (3) incorporate microbiome monitoring into AML clinical trials and stratify outcome analyses by antibiotic exposure; and (4) investigate planned microbiome restoration (for example autologous FMT after infection control) as an adjunct to preserve or restore chemosensitizing taxa. These measures are meant to preserve the mechanistic links (metabolic, immunologic, epigenetic) by which the microbiota modulates chemosensitivity rather than simply “prevent dysbiosis”^146,155,156.

Recent syntheses of the literature extend these observations: multiple reviews demonstrate that antibiotic-induced dysbiosis correlates with impaired responses to some anti-cancer therapies (including immunotherapies) and with worse clinical outcomes in selected cohorts, but confounding by indication and heterogeneity in antibiotic exposures complicate causal inference. These systematic assessments support a stronger emphasis on antibiotic stewardship trials that measure microbiome and oncologic endpoints, and on testing targeted or microbiome-sparing approaches in randomized designs^17,157.

Given that the microbiome varies not only by region and dietary habits, there are also intra-individual differences, suggesting that any therapeutic approach to restore microbiome regulation could produce a unique response in each patient. In line with this, an in vitro investigation on a gut model and chemotherapy demonstrated that the extent of dysbiosis is less than that reported in in vivo evaluations¹⁵³. Therefore, the patient’s response is a critical factor, highlighting the significance of personalized medicine in microbiota-modulating therapeutic strategies.

Challenges and future directions

The burgeoning field of microbiome research in AML confronts significant methodological and biological challenges that must be addressed to translate findings into clinical practice. A primary obstacle is the substantial heterogeneity in study designs, patient populations, and sequencing methodologies, which complicates the identification of consistent microbial signatures predictive of chemotherapy outcomes across diverse cohorts^24,132. For instance, while several studies report reduced alpha diversity and depletion of SCFA-producing taxa like Faecalibacterium and Roseburia at diagnosis, the specific pathogenic bacteria enriched (e.g., Enterococcus vs. Streptococcus) vary considerably between geographic and institutional cohorts^24,42,132. This variability underscores the complex interplay between host genetics, environmental exposures, and baseline microbiota, necessitating large-scale, multi-center studies with standardized protocols for sample processing, sequencing depth, and bioinformatic analysis to establish robust microbiome-based biomarkers^24,132. While this narrative review synthesizes mechanistic and clinical data linking gut dysbiosis to chemotherapy resistance in AML, several important limitations must be acknowledged. The clinical literature is still small and highly heterogeneous (single-center cohorts, variable sampling time-points and sequencing platforms, and inconsistent metadata capture), which limits causal inference and the generalizability of specific taxonomic associations across regions and care settings^24,158. Many mechanistic claims remain rooted in preclinical models (germ-free or antibiotic-treated mice and ex vivo assays) whose translational fidelity to human AML is imperfect. In addition, supportive-care confounders (broad-spectrum antibiotics, antifungals, opioids, and nutritional interventions) substantially influence the microbiome and are variably reported across studies, increasing residual confounding^24,82,158.

Additionally, the mechanistic links connecting specific bacterial taxa or their metabolites to chemoresistance pathways remain incompletely elucidated. Although preclinical models suggest that microbiota-derived metabolites like butyrate modulate drug efficacy through HDAC inhibition or immune activation, validating these mechanisms in human AML requires sophisticated multi-omics integration and functional assays using gnotobiotic models^45,159,160. The confounding effects of supportive therapies, particularly broad-spectrum antibiotics, further obscure causal relationships. Antibiotics induce dysbiosis characterized by Enterococcus domination and reduce microbial diversity, which correlates with prolonged neutropenia and increased infection risk, factors directly impacting chemotherapy tolerance and dose intensity^45,161,162. Developing strategies to preserve microbial resilience during antibiotic exposure, such as beta-lactamase inhibitors to protect commensals or targeted narrow-spectrum agents, represents an urgent clinical need^45,161. Moreover, the impact of novel targeted therapies (e.g., venetoclax, FLT3 inhibitors) on gut microbiota and vice versa remains largely unexplored, creating knowledge gaps in pharmacomicrobiomics for emerging AML treatment regimens^132,160,163. Table 4 summarizes key bacterial taxa and representative mechanistic/clinical citations discussed above (see Sections 2.2.1–2.2.2). It is included here as a concise reference for clinicians/researchers and to list potential microbiome-targeted interventions that follow from the findings described earlier.

Table 4.

Key bacterial Taxa implicated in AML chemoresistance and their mechanisms

Bacterial taxon (representative refs)	Abundance in AML	Functional role	Impact on chemotherapy / clinical outcome	Potential interventions (examples)
.Enterococcus spp42,82,106	Increased	Virulence factors; intrinsic antibiotic resistance; domination after broad-spectrum antibiotics	Associated with prolonged neutropenia and bloodstream infection (BSI) risk; linked to worse outcomes after dysbiosis.	Targeted narrow-spectrum therapy; phage therapy; antibiotic stewardship.
Eubacterium limosum24,106	Decreased	Butyrate/SCFA production; supports enterocyte function and metabolic homeostasis	Higher abundance associated with reduced relapse after allo-HCT; correlates with muscle strength and nutritional status.	Prebiotic/dietary support; precision probiotics; FMT in selected settings.
Faecalibacterium spp. (F. prausnitzii)27,42,44	Decreased	Major butyrate producer; anti-inflammatory; supports barrier/immune regulation	Lower abundance linked to increased infection risk and barrier dysfunction.	Prebiotics (inulin, FOS); dietary fiber; FMT to restore SCFA producers.
Blautia spp178.	Variable (cohort-dependent)	SCFA production; redox and metabolic modulation	Associations with anorexia/cachexia and with hematopoietic recovery in some cohorts.	Dietary modulation (fiber/polyphenols); targeted probiotics.
Akkermansia muciniphila23,44	Decreased (but context-dependent)	Mucin degradation; promotes barrier integrity and metabolic health	Often anti-inflammatory, but pre-chemotherapy expansion can be contextually linked to neutropenic fever; effects are cohort-dependent.	Dietary polyphenols; prebiotics to support beneficial growth; careful monitoring in neutropenia.

Pressing research gaps and prioritized next steps

To accelerate translation of microbiome science into interventions that reduce chemotherapy resistance in AML, we propose five immediate, high-priority research gaps and corresponding actions: Standardize microbiome measurement and metadata reporting. Harmonize pre-analytic (sample timing, collection device, storage temperature, freeze-thaw handling), analytic (16S vs. shotgun, library preparation, sequencing depth, negative/positive controls), and bioinformatic parameters (taxonomic databases, filtering thresholds, normalization). Adoption of reporting checklists and minimum metadata (e.g., STORMS; MIxS) will improve comparability and meta-analysis across cohorts and trials¹⁵⁸. Define mechanisms linking microbiota to chemotherapy pharmacology and immunomodulation. Prioritize mechanistic preclinical studies (gnotobiotic mice, microbial mono- or defined consortia, and ex vivo organoid/microbiome co-culture systems) that test whether specific taxa or microbial enzymes (for example β-glucuronidases, reductases, or acetate-generating pathways) alter the pharmacokinetics /pharmacodynamics of AML drugs and modulate anti-leukemia immune responses. Validation in human longitudinal multi-omics (metagenomics + metabolomics + host transcriptomics/immune profiling) should follow¹⁶⁴. Standardize intervention protocols (FMT, probiotics, pre/pro/postbiotics) and donor selection. For FMT and donor-derived products, establish consensus donor-screening criteria, product formulation (fresh vs frozen vs encapsulated), timing relative to chemotherapy, and safety surveillance, drawing on transplant and stool-bank experience. Early harmonized protocols will allow pooled analyses and safety signal detection¹⁶⁵. Harmonize clinical trial design and core endpoints. Design randomized trials with prespecified microbiome endpoints (diversity metrics, engraftment, key metabolites) and clinically meaningful oncologic endpoints (response depth, relapse-free survival, infection rates). Standardized antibiotic/ supportive-care reporting is essential because these strongly confound microbiome–outcome relationships¹⁶⁴. Build open, interoperable data resources and apply best-practice analytic pipelines. Deposit raw sequence data with complete MIxS-compatible metadata and use reproducible bioinformatic workflows to enable pooled, machine-learning-enabled discovery and external validation of microbial biomarkers of chemoresistance¹⁶⁴.

Addressing these gaps will (i) improve reproducibility across centers, (ii) enable mechanistic, hypothesis-driven trials, and (iii) accelerate regulatory- and clinician-adoption of safe, effective microbiome-directed strategies in AML.

With the gaps above addressed, subsequent interventional studies should be hypothesis-driven, use harmonized sample/endpoint protocols, and integrate mechanistic multi-omic substudies to permit causal inference and biomarker development¹⁶⁴. Promising approaches include precision probiotics containing Eubacterium eligens or Clostridium butyricum, which correlate with improved muscle strength and reduced mucositis in preliminary studies, and FMT to restore microbial diversity post-chemotherapy^24,45,161. Early clinical trials demonstrate that autologous FMT can reconstitute microbial diversity after intensive chemotherapy, potentially reducing infectious complications, but long-term impacts on relapse and survival require further investigation^24,45. Bacterial metabolites also offer therapeutic targets; exogenous administration of butyrate or propionate could reverse SCFA depletion observed in AML patients, thereby enhancing epithelial barrier integrity and reducing systemic inflammation that compromises chemotherapy efficacy^24,42,159. Similarly, inhibiting bacterial beta-glucuronidase may prevent chemotherapeutic drug inactivation in the gut lumen, a mechanism implicated in reduced efficacy of irinotecan and potentially anthracyclines¹⁶⁰. Advanced technologies such as multi-omics integration (metagenomics, metabolomics, transcriptomics) and machine learning algorithms will be crucial for deciphering host-microbe-drug interactions and developing predictive models of treatment response^42,45,161. For example, longitudinal metabolomic profiling could identify microbial-derived metabolites like phenylacetate or hippurate as early biomarkers of dysbiosis-induced chemoresistance^24,45. Furthermore, exploring intratumoral microbiota in bone marrow niches may reveal localized microbial influences on leukemic stem cell survival and drug penetration, an understudied aspect of AML microenvironment biology¹⁶⁰. Finally, addressing ecological resilience through dietary interventions (e.g., high-fiber diets to promote Akkermansia growth) or prebiotics tailored to nourish keystone taxa could sustain microbial homeostasis during treatment, mitigating dysbiosis and enhancing chemosensitivity^159,161. As the field advances, rigorously designed clinical trials incorporating microbiome monitoring as secondary endpoints will be essential to validate these approaches and usher in an era of microbiome-informed precision oncology for AML. Actionable recommendations to operationalize microbiome-guided, personalized AML therapy. To accelerate clinical translation, we recommend that future studies explicitly integrate gut microbiome profiling with established clinical and molecular risk factors (cytogenetics, driver mutations [FLT3, NPM1, IDH1/2], and measurable residual disease [MRD]) into combined predictive models. Practically, this requires (i) standardizing timing and methods for specimen collection (baseline prior to induction, at nadir, post-remission, and at relapse), sample processing (shotgun metagenomics where possible plus targeted 16S as backup), and core metadata capture (antibiotic exposures, diet, BMI, transfusion history, and exposure to targeted agents); (ii) prospectively powering multi-center cohorts to develop multivariable prognostic models and machine-learning classifiers that combine microbiome features (taxa, resistome, and metabolite signatures) with clinical/molecular covariates and MRD status^166–168; (iii) pre-specifying clinically meaningful endpoints (MRD conversion, event-free survival, relapse rate) and including intermediate functional readouts (fecal metabolomics, systemic LPS/kynurenine levels) as mechanistic secondary endpoints; and (iv) validating models in independent external cohorts and by biological functional assays (ex vivo drug sensitivity with paired fecal transplants or gnotobiotic mouse rescue experiments). Trial designs should include stratified randomization based on combined risk scores (for example, high molecular risk + dysbiotic microbiome vs. low risk + eubiotic microbiome) to test microbiome-directed interventions (microbiome-sparing antibiotic strategies, precision probiotics or defined consortia, autologous or selected donor FMT, and targeted microbial enzyme inhibitors). Finally, data harmonization, biobanking for longitudinal multi-omics, and transparent reporting standards will be essential so predictive models can be shared, compared, and prospectively validated across institutions.

Conclusion

Chemotherapy resistance and relapse continue to undermine AML management, necessitating innovative strategies beyond conventional genetics. This review underscores the gut microbiome as a pivotal player in AML chemoresistance, where dysbiosis, induced by chemotherapy, antibiotics, or the disease itself, compromises drug efficacy via metabolic interference, immune suppression, and epigenetic dysregulation. Clinical evidence consistently links low microbial diversity, loss of SCFA producers (e.g., Faecalibacterium), and pathogenic blooms (e.g., Enterococcus) to poor outcomes, including infection susceptibility and relapse. Interventions such as FMT, probiotics, and dietary adjustments demonstrate potential to restore microbial balance, enhance treatment response, and reduce complications. Notably, while meta-analytic syntheses indicate symptomatic benefits of some probiotic regimens and encouraging early signals for FMT in the transplant/GVHD setting, the overall evidence is heterogenous, often underpowered, and frequently excludes the most immunocompromised patient, factors that limit immediate generalizability to AML induction settings. In light of the limitations described above, we propose concrete, testable steps that can be enacted now: (i) embed standardized microbiome sampling and reporting into AML trials, (ii) conduct randomized studies of stewardship-driven antibiotic algorithms and microbiome-sparing adjuncts, (iii) evaluate autologous or precision FMT and defined consortia in controlled settings with rigorous safety endpoints, and (iv) prioritize interventions that restore SCFA producers and barrier function (diet/prebiotics/postbiotics) while avoiding unselected live-microbe strategies in profoundly neutropenic patients. These coordinated efforts, linked to multi-omic mechanistic substudies and harmonized endpoints, offer the fastest and safest path to determine whether microbiome-informed care can reduce dysbiosis-driven chemoresistance and improve AML outcomes. Integrating microbiome strategies into AML care could help overcome resistance and reduce relapse.

Author contributions

S.S. Alemohammad, M.F. Manesh, F. Tavakoli, A. Shams, and A. Zeynalabadi wrote and carried out investigation and visualization for the manuscript. H. Soleimani Samarkhazan contributed conceptualization. M. Tian revised the manuscript. All authors reviewed and approved the final version.

Data availability

No datasets were generated or analysed during the current study.

Competing interests

The authors declare no competing interests.