Microbiota-based approaches to mitigate infectious complications of intensive chemotherapy in patients with acute leukemia

Abstract

Despite advances in antimicrobial treatments, infection remains a common complication of intensive chemotherapy in patients with acute leukemia. It has become progressively apparent that the current antimicrobial focus has shortcomings that result from disruption of the commensal microbial communities of the gut. These effects, collectively known as dysbiosis, have been increasingly associated worldwide with growing complications such as Clostridioides difficile infection, systemic infections, and antibiotic resistance. A revision of the current practice is overdue. Several innovative concepts have been proposed and tested in animal models and humans, with the overarching goal of preventing damage to the microbiota and facilitating its recovery. In this review, we discuss these approaches, examine critical knowledge gaps, and explore how they may be filled in future research.

INTRODUCTION

Intensive chemotherapy is a treatment modality used in patients with aggressive malignancies, such as acute leukemia and high-grade lymphoma. Intensive chemotherapy is typically used with curative intent but is associated with significant toxicity.¹ Common complications of intensive chemotherapy include organ toxicity and infection. The main reasons for the host’s high vulnerability to infection during intensive chemotherapy are immunosuppression and damage to the mucosal barriers that separate the host from billions of microorganisms. Tissues affected most severely by intensive chemotherapy are those with the highest turnover rate (ie, hematopoietic cells and epithelial cells of the gastrointestinal tract). As a result, neutropenia and mucositis are common among intensively treated patients. Damage to the mucosal barriers facilitates microbial translocation to the bloodstream,^2,3 which given the host’s immunocompromised state, can result in sepsis and death. The interaction between the gut mucosal barrier and microbiota influences the magnitude of gut mucosal damage during chemotherapy and its repair following chemotherapy. The gut microbiota provides tonic stimulation to the gut epithelium⁴; thus, loss of this stimulation from the microbiota after antibiotics can make the gut mucosal barrier even more vulnerable to chemotherapy toxicity. In addition, microbiota effects on the gut barrier depend on taxon specific functional interactions with epithelial cells. Expansion of mucolytic bacteria such as Akkermansia⁵ can result in thinning of the mucin layer covering the epithelium, thereby enhancing access of predominantly luminal bacteria to the epithelium.⁶ In contrast, many mucin-adherent bacteria such as Clostridium cluster XIVa are capable of producing gut barrier-protective and anti-inflammatory short-chain fatty acids⁷ and can protect against bacterial translocation.⁸ Collectively, homeostasis of the ecosystem composed of the host’s immune system, mucosal barrier, and microbiota is critical for protection against bacterial translocation to the bloodstream. This homeostasis is severely impaired during antileukemia chemotherapy.

Neutropenic fever (NF) is a typical early sign of impending sepsis⁹ and occurs in >80% of patients with hematologic malignancies who develop neutropenia.¹⁰ The main tool to prevent and treat infections during intensive chemotherapy has been antibiotics.⁹ Antibacterial antibiotics, especially fluoroquinolones, are almost universally used to prevent NF in patients treated with intensive antileukemia chemotherapy.¹¹ If NF occurs, these antibiotics are empirically escalated to target potential uncovered bacteria. Common initial regimens include third (or higher) generation cephalosporins, piperacillintazobactam, and carbapenems. Institutional antimicrobial sensitivities are often used to guide empirical antibiotic choices, but they are generally not tailored to the patient’s colonizing flora except for specific pathogens like vancomycin-resistant enterococci (VRE), methicillin-resistant Staphylococcus aureus, or extended spectrum beta lactamase-producing gram negative bacilli. There are no universally promoted or accepted guidelines on the duration of escalated antibiotics if the microbiological work-up remains negative and no obvious clinical infection is diagnosed. Some physicians continue the escalated regimen until neutrophil recovery, while others continue until defervescence, and others prioritize hemodynamic stability as the main reason to initiate antibiotic de-escalation.

Collectively, patients undergoing intensive chemotherapy experience high antibiotic “pressure.” In the following sections, we will discuss the efficacy and shortcomings of this exclusively antimicrobial approach and some of the potential alternative or complementary strategies that may be implemented to mitigate infectious complications of intensive chemotherapy.

CAUSES OF DYSBIOSIS IN PATIENTS WITH ACUTE LEUKEMIA

Intensive chemotherapy and antibiotic prophylaxis are used in a variety of clinical settings, including acute leukemia, aggressive lymphomas, and solid tumors. In patients with acute leukemia, however, the myelosuppressive effect of chemotherapy is more intense and long-lasting, the intestinal epithelial injury is more pronounced, and as a consequence, the infectious complications are more frequent.

Antibiotics.

Antibiotics are the main initiators of dysbiosis in patients with acute leukemia. A 2007 meta-analysis showed that fluoroquinolone prophylaxis was associated with trends to higher rates for colonization with resistant bacteria.¹² Levofloxacin has been associated with increased risk for Clostridioides difficile infection (CDI),¹³ Candida colonization,¹⁴ and breakthrough bacteremia with multidrug-resistant organisms.¹⁵ In a recent single-center analysis,¹⁶ the effect of levofloxacin (250 mg per day) prophylaxis on the microbiota during intensive therapy was limited to 2 anaerobic commensal genera Parabacteroides (decreased) and Blautia (increased). No effect was found on microbiota diversity. Considering the overall lesser disruption of the microbiota after levofloxacin exposure compared with broad-spectrum beta-lactam antibiotics,¹⁷ levofloxacin is probably our best current choice for antibacterial prophylaxis during intensive therapy. In addition, the practice of noprophylaxis is discouraged given the existing evidence for clinical benefit from large randomized trials.

Nutrition.

Although antibiotics are the key players in causing dysbiosis in acute leukemia patients, they are likely not the only etiology. Mucositis and chemotherapy-induced nausea make enteral nutrition challenging in many patients undergoing intensive therapy. Poor enteral nutrition depletes the gut of dietary fiber and other important nutrients that commensal bacteria need. In addition, enteral nutrition improves gut barrier integrity¹⁸ and may be important in reducing bacterial translocation. The evidence for a benefit (or lack thereof) of parenteral nutrition in acute leukemia patients undergoing intensive chemotherapy is poor, with no reported successful prospective study. In hematopoietic cell transplantation (HCT) recipients, however, the evidence is mixed with one randomized trial showing a survival benefit with prophylactic total parenteral nutrition¹⁹ and another one showing no benefit.²⁰ Neither study evaluated the microbiota. A recent trial randomized patients to receive enteral nutrition via a nasogastric tube starting day 1 post-HCT vs standard of care.²¹ Standard of care was oral nutrition until oral intake declined to ≤60% of requirements for 3 days and appeared unlikely to improve for another week, at which point parenteral nutrition was started. Antibiotic prophylaxis was similar between the groups. The gut microbiome at day 30 post-transplant was compared between the 2 groups. The overall composition of the microbiome and its diversity were not different between the groups. However, enteral nutrition was associated with a higher abundance of Bacteroidetes and lower abundance of Proteobacteria and Enterococcus. Gram-negative rods, many of which belong to the phylum Proteobacteria, are the cause of the most dangerous bacteremias in immunocompromised patients. Finally, a small nonrandomized study evaluated 20 pediatric patients who received enteral vs parenteral nutrition post-HCT. Patients receiving enteral, but not parenteral, nutrition had quick recovery of microbiome diversity, composition, and short-chain fatty acid production after transplant.²²

In a recent murine syngeneic HCT study, antibioticmediated microbiota depletion impaired dietary energy uptake and hematopoietic recovery; this hematopoietic effect was prevented by enteral supplementation of sucrose (a simple carbohydrate absorbed directly by the host).²³ These findings suggest that enteral dietary supplementation may counter the detrimental effects of dysbiosis on hematopoiesis. Although the effects of enteral nutrition on the gut microbiota appear to be overall beneficial, specific nutrients may promote pathogen expansion in the gut. In a recent murine allogeneic HCT study, dietary lactose drove expansion of Enterococcus in the gut and dietary lactose depletion attenuated Enterococcus outgrowth.²⁴ Therefore, specific enteral nutritional adjustments may result in mitigation of dysbiosis.

Chemotherapy and other nonantibiotic drugs.

Chemotherapy-induced cytotoxic damage to the gut barrier can alter the gut microbiota, although direct evidence in acute leukemia patients is lacking due to technical difficulties of studying the gut barrier in this vulnerable patient population. Alpha defensin-5 and Reg3α are the most abundant Paneth cell antimicrobial peptides and modulate the gut microbiota.^25–28 Specifically, alpha defensin-5 maintained Bacteroidetes numerical dominance over Firmicutes in murine experiments.²⁵ Reg3γ-knockout mice had increased levels of mucosaassociated Gram-positive bacteria, consistent with the activity of Reg3 against Gram-positive bacteria.²⁸

Antibiotics are not the only group of medications with antimicrobial activity. A recent analysis showed that a large variety of nonantibiotic drugs also influence the gut microbiota; this list includes proton pump inhibitors used to suppress gastric acid secretion and antipsychotic medications used to control nausea in patients receiving chemotherapy.²⁹ Specifically, proton pump inhibitor use has been associated with lower microbial diversity and an increase in Enterococcus, Streptococcus, and Staphylococcus.^30,31 Atypical antipsychotic drug use has been associated with increased Lachnospiraceae and decreased Akkermansia.³² Among chemotherapeutic drugs commonly used in leukemia induction regimens, antimetabolites²⁹ and cyclophosphamide³³ are 2 examples with well-established antibacterial activity. In addition, there is in vitro evidence for the activity of daunorubicin and etoposide against both aerobic and anaerobic bacteria.³⁴ High-dose chemotherapy in patients with non-Hodgkin lymphoma was associated with a reduction in the relative abundance of Firmicutes and Actinobacteria and a rise in the relative abundance of Proteobacteria.³⁵ Although patients did not receive antibiotics during chemotherapy, most had received penicillin V, trimethoprim-sulfamethoxazole, or both until the day of admission. Thus, some of the observed microbiome changes were likely related to recent antibiotic use.

Together, strong evidence from human studies indicates that excessive antibiotic pressure during intensive chemotherapy severely injures the microbiota. Direct evidence for a beneficial effect of enteral nutrition on the microbiota in acute leukemia patients is lacking, but the subject has not been adequately investigated. Finally, the available evidence suggests that the gut microbiota changes as a direct effect of chemotherapeutic and other nonantibiotic drugs. Acute leukemia patients typically receive a large number of nonantibiotic medications during intensive chemotherapy. The cumulative effect of these drugs on the gut microbiota may be major and possibly worsen antibiotic-induced dysbiosis.

EFFICACY OF ANTIBIOTICS IN PREVENTING INFECTIONS

A seminal randomized placebo-controlled clinical trial in 2005 established levofloxacin as a standard antibacterial prophylaxis in adult patients with cancer in whom neutropenia was expected to occur for more than 7 days.¹¹ Levofloxacin (500 mg daily) decreased the risk of NF by 20% (95% confidence interval [95%CI]: 14%–26%) compared to placebo. Although the efficacy of levofloxacin compared to placebo was remarkable, 65% of patients in the levofloxacin group still developed NF and hence were treated with escalated antibiotics. Levofloxacin decreased the risk of microbiologically documented infections, bacteremia, and single-agent Gram-negative bacteremia by 7%–17%. The effect of levofloxacin prophylaxis on the total duration of exposure to antibiotics was not reported; however, the total cost of antibiotics was significantly lower in the levofloxacin group. The largest randomized trial in children undergoing intensive chemotherapy comparing prophylactic levofloxacin with placebo showed an 11% (95%CI: 4%–18%) lower risk of NF among patients with acute leukemia in the levofloxacin group and 22% (95%CI: 9%–34%) lower risk of bacteremia in the levofloxacin group when patients with acute leukemia and HCT recipients were combined.³⁶ Similar to the adult study, 71% of patients in the levofloxacin group still developed NF and hence were treated with escalated antibiotics. In this study, levofloxacin prophylaxis decreased the total duration of exposure to escalated antibiotics for the treatment of NF. This reduced exposure to antibiotics likely explains the observed lower rates of CDI in the levofloxacin group (8% vs 14% in the placebo group). A major contributor to bacteremia episodes in the levofloxacin group were Viridans group streptococci, suggesting that levofloxacin may have limitations in preventing bacteremia of oropharyngeal origin.

The survival benefit of antibiotic prophylaxis in the current practice is questionable. A recent large meta-analysis (2 randomized-controlled trials and 12 observational studies) showed approximately 70% lower risk of NF and 40% lower risk of bacteremia, but no change in overall mortality.³⁷ This lack of survival benefit may be due to the effectiveness of antibiotic treatment of infections that occur with or without preceding antimicrobial prophylaxis.

The appropriate endpoint for measuring the value of escalated antibiotics for the treatment of NF is not clear. Because fever is only a sign, rather than a diagnosis, the value of defervescence as the main endpoint is questionable and may not even be achievable with a stronger antimicrobial attack. A large randomized trial, for example, compared ceftazidime alone with a combination of cephalothin, gentamicin, and carbenicillin as frontline empirical treatment of NF. Only about 60% of patients in both groups defervesced within 72 hours of starting antibiotics.³⁸ However, in both groups, almost all patients eventually became afebrile. A 2005 meta-analysis showed that addition of a glycopeptide antibiotic (eg, vancomycin to improve coverage against Gram-positive bacteria) to the initial empirical regimen for NF did not change time to defervescence, but was associated with more toxicity.³⁹ These results suggest that lack of early defervescence may not be a poor prognostic sign and, if the patient is clinically stable, should not be the primary goal of therapy. For most patients with clinical stability, stepwise escalation of antibiotics as needed could be preferred over a combination empirical therapy upfront, but this has not been clinically rigorously tested.

Together, current antimicrobial practice is only partially effective in preventing NF, but is highly effective in permitting eventual clinical recovery.⁴⁰

FUNCTIONS OF THE GUT MICROBIOTA RELEVANT TO INTENSIVE CHEMOTHERAPY

Not only the host, but also the microbial communities within the host are profoundly influenced during intensive chemotherapy. Microbiota of primary relevance during intensive chemotherapy are oropharyngeal and intestinal microbiota, collectively referred to as the gut microbiota in this review. The gut microbiota is separated from the host by an epithelial barrier, which is severely compromised during intensive chemotherapy. Two critical functions of normal microbiota in their mutualistic relationship with the host⁴¹ are disrupted by antibiotics: gut barrier support and pathogen colonization resistance.

Gut barrier support.

The innate immunity receptors, such as toll-like receptors (TLRs), on the basolateral membrane of epithelial cells and tissue-resident antigen-presenting cells recognize commensal microbiota; this recognition provides tonic stimulation to the gut barrier.⁴ Although the effect of microbiota support for the gut barrier during chemotherapy is difficult to study in humans, animal studies have indicated that it mitigates cytotoxic damage and promotes faster recovery. Specifically, the commensal microbiota activates TLR signaling via the MyD88 pathway to promote production of tissue-protective factors and confer resistance to chemical and radiation-induced intestinal injury in mice.⁴ It is not clear whether specific groups of microorganisms or a community of microorganisms are important for this tissue-protective function. It has been well established, however, that short-chain fatty acids, especially butyrate, enhance the function of intestinal barrier tight junctions and serve as a nutrient for colonocytes.^42,43 Butyrate is primarily produced from fermentation of dietary fiber by Gram-positive bacteria in the Firmicutes phylum (mostly Clostridium clusters IV and XIVa).⁴⁴

Pathogen colonization resistance.

A healthy gut microbiota prevents pathogen colonization, a concept that was shown more than 50 years ago.⁴⁵ The unprecedented success of fecal microbiota transplantation (FMT) in eradicating CDI and its promising ability to decolonize antibiotic-resistant bacteria in the gut highlight the efficacy of a balanced, well-composed microbial community in preventing the entry and expansion of pathogens.^46–49 The mechanisms of colonization resistance have been partially elucidated and include both direct and indirect pathways. The direct pathway includes contact-mediated mechanisms and killing via secreted antimicrobial compounds (eg bacteriocins). One of the best examples for the bacteriocin-mediated direct antagonism is thuricin CD produced by Bacillus thuringiensis. Thuricin CD has potent antibacterial activity against C. difficile.⁵⁰ One of the best examples for cell contact-mediated killing is the type VI secretion system of various Proteobacteria, which enables them to translocate specific toxins into their target cells.⁵¹

Several examples for the indirect pathway of colonization resistance through the host immunity have been described. As an example, Bacteroides thetaiotaomicron enhances resistance to viral infections by stimulating host secretion of type I IFN-induced GTPases and to Gram-positive bacterial colonization by stimulating secretion of antimicrobial peptide, regenerating islet-derived protein 3 (Reg3), from intestinal epithelial and Paneth cells.⁵² Another indirect mechanism for colonization resistance is inter-bacterial competition for resources; for example, B. thetaiotaomicron protects against Citrobacter rodentium colonization by consuming monosaccharides that are required for the pathogen’s growth.⁵³ Bacterial stimulation of colonization resistance through the host may be triggered by factors shared among different bacterial groups. For example, recognition of flagellin, the primary protein in the locomotor machinery of almost all motile bacteria, by the host’s intestinal dendritic cells enhances small intestinal epithelial Reg3 secretion and potent mucosal immune responses against luminal bacteria.⁵⁴

VRE and C. difficile commonly overgrow and colonize the intestinal tract of patients undergoing intensive chemotherapy patients.^55–57 The overall burden of clinical infections with these 2 organisms is very large. C. difficile is the most frequently reported nosocomial pathogen in the US, with about half a million cases in 2011 and 29,000 deaths.⁵⁸ It was estimated in a 2013 article that C. difficile increases US annual expenditures by about 1.5 billion dollars.⁵⁹ Similarly, VRE infections are growing worldwide; VRE is now among the top 3 causes of bloodstream infection.⁶⁰ A European study in 2018 estimated the median attributable costs per VRE infection to be over EUR 13,000.⁶¹ In patients with acute leukemia undergoing intensive chemotherapy, C. difficile and VRE infections each occur in approximately 10%–20% of patients.^55–57 In addition, VRE colonization is a strong risk factor for subsequent VRE bloodstream infection in these patients.^62,63 Some of the obligate anaerobic commensal microbiota play a crucial role in preventing gut colonization by VRE and C. difficile. For example, cooperation between Clostridium cluster XIVa species Blautia producta and Clostridium bolteae cleared VRE from the intestines of mice and restored colonization resistance against this pathogen.⁶⁴ Blautia producta exerted this protective effect by secreting a strain-specific lantibiotic.⁶⁵ In addition, the host’s innate immunity may be involved. In murine experiments, bacterial flagellin stimulated the TLR5-MyD88 pathway and induced Reg3γ expression in intestinal epithelial cells and Paneth cells, which in turn killed Gram-positive bacteria such as VRE.⁶⁶ Bacterial lipopolysaccharide-induced stimulation of the TLR4-MyD88 pathway had the same effect on Reg3γ expression and resistance against VRE colonization.⁶⁷ Antibiotic-induced depletion of bacteria may contribute to loss of VRE colonization resistance via these and other mechanisms.

Low microbial diversity is a common feature of stool samples from patients with CDI⁶⁸ and a loss of diversity frequently precedes clinical infections.⁶⁹ At a taxonomic level, a reduction in Clostridiales Incertae Sedis XI was found to precede CDI in one study.⁶⁹ In other studies, enrichment for Enterococcaceae and Lactobacillaceae and depletion of Ruminococcaceae and Lachnospiraceae have been found to correlate with CDI.^70,71 The precise mechanisms by which certain consortia of bacteria may protect the gut against C. difficile colonization and infection remain to be elucidated.

Together, the gut microbiota strengthens the gut barrier and limits the emergence and spread of pathogens, including antibiotic-resistant pathogens (eg, VRE), within the gut. These functions are important under physiological conditions, and become critical determinants of the risk of infectious complications during intensive chemotherapy when both the host’s immunity and gut barrier are severely compromised.

DYSBIOSIS AND ITS RELATION TO INFECTION IN ACUTE LEUKEMIA PATIENTS

Several studies have characterized the emergence, progression, and resolution of dysbiosis in acute leukemia patients and how specific changes are related to infectious complications. Current evidence indicates that microbial diversity and composition are 2 microbiota characteristics clearly affected during intensive chemotherapy.

Gut microbial diversity declines during intensive chemotherapy, and such species as Lactobacillus and Enterococcus become dominant^72–74 at the cost of such species as Clostridia and Blautia.^34,75 A second course of intensive chemotherapy makes the microbial communities even more unstable, with higher vulnerability to enterococcal expansion and departure from the baseline community structure.⁷⁶ Departures from baseline are often at least partially reversible (microbiome “resilience”), but in some cases the microbiome settles down into a completely new steady state, with potential implications for the host.⁷⁷ Lower diversity of the stool microbiome at baseline and/or a more marked decline in diversity during chemotherapy predicted higher rates of subsequent infections in several adult AML studies,^72,75 though not in a large pediatric ALL study.⁷⁴ In addition, a higher relative abundance of Porphyromonadaceae (a family under phylum Bacteroidetes and containing Porphyromonas and Dysgonomonas) at baseline seemed to protect against subsequent infections in an adult AML study.⁷⁵ The diversity of the microbiome not only decreases with time during intensive chemotherapy, but also loses stability. This instability, both in the oral and intestinal microbiome, was associated with a higher rate of infection in a previous analysis.⁷⁸ The observed instability in diversity was associated with longer exposure to antibiotics and relative expansion of potentially pathogenic species such as Staphylococcus and Streptococcus.

Poorly understood changes in the overall composition of the gut microbiome are associated with altered risk of infection. A higher relative abundance of Faecalibacterium prausnitzii was found in one pediatric ALL study to be correlated with lower risk for infectious complications.⁷⁹ In another study of pediatric ALL patients, those with a higher baseline relative abundance of Proteobacteria had a higher rate of NF. Microbiota domination by Enterococcaceae or Streptococcaceae at any time during chemotherapy predicted infection in subsequent phases of chemotherapy.⁷⁴ A pediatric ALL study suggested an expansion of Ruminococcus torques and R. gnavus, 2 mucolytic gram-positive anaerobic bacteria, during chemotherapy.⁸⁰ Expansion of mucolytic bacteria may cause or worsen mucositis and increase the risk of bacterial translocation. A recent analysis⁸ suggested that the effect of the microbiota on the risk of bloodstream infection may be time-dependent, with such mucolytic taxa as Akkermansia⁵ increasing the risk within the next few days and such mucoprotective taxa as butyrogenic species belonging to Clostridium cluster XIVa⁷ reducing the immediate risk. Collectively, specific compositional changes in the microbiome, largely caused by antibiotics and often manifesting as domination events, predict the risk of systemic infection. The mechanistic link between gut dysbiosis and systemic (including distant) infection is a fertile area for future research.

Together, the gut microbiota loses diversity and becomes vulnerable to pathogen domination during intensive chemotherapy. Some dominant bacteria directly cause bloodstream infection via gut translocation, whereas others weaken the gut barrier and facilitate translocation of other bacteria. Table I summarizes the available literature on the relationship between gut dysbiosis and infection in patients with acute leukemia.

Table I.

Summary of the available literature on the relationship between gut dysbiosis and infection in patients with acute leukemia

Study, country, age group	Acute leukemia, treatment	Microbiota-antibiotic association	Microbiota-infection association
Galloway-Peña et al. 2016,72 USA, Adults	AML, newly diagnosed, intensive induction	Carbapenem >72 h ~ Lower alpha diversity	Baseline alpha diversity ~ Infection during induction
Galloway-Peña et al. 2019,75 USA, Adults	AML, newly diagnosed, intensive induction	Total days on antibiotics before neutrophil recovery ~ Lower alpha diversity at neutrophil recovery	Lower baseline alpha diversity ~ Infection before neutrophil recovery Baseline relative abundance of Porphyromonadaceae ~ Lower rate of infection before neutrophil recovery
		Carbapenem >72 h ~ Lower alpha diversity	More rapid decline in alpha diversity ~ Infection within 3 mo after neutrophil recovery
Hakimetal.2018,74 USA, Pediatrics	ALL, newly diagnosed, intensive induction (multiphase)	No associations	No association between baseline alpha diversity and neutropenic fever Baseline relative abundance of Proteobacteria ~ Neutropenic fever Dominance of Enterococcaceae or Streptococcaceae ~ Neutropenic fever and infection throughout therapy
Nearingetal. 2019,79 Canada, Pediatrics	ALL, newly diagnosed, intensive induction (multiphase)	Vancomycin ~ Lower alpha diversity	Lower alpha diversity ~ Bloodstream infection Faecalibacterium prausnitzii ~ lower chance for infection
Galloway-Peña et al. 2017,78 USA, Adults	AML, newly diagnosed, intensive induction	No associations	Temporal variability (ie, instability) of alpha diversity ~ Infection within 3 mo after neutrophil recovery Stenotrophomonas relative abundance ~ Infection within 3 mo after neutrophil recovery

Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia.

Only studies reporting data on intestinal microbiota-infection associations in patients with acute leukemia are shown.

MICROBIOTA-ORIENTED STRATEGIES TO PREVENT AND CORRECT DYSBIOSIS

Various therapeutic approaches have aimed to protect or restore the microbiota during intensive chemotherapy. Table II summarizes the approaches used to prevent microbiota injury and strategies to facilitate microbiota repair.

Table II.

Summary of microbiota-oriented approaches to mitigate infectious complications in acute leukemia patients treated with intensive chemotherapy

Approach	Concept	Advantages	Challenges
Antibiotic stewardship	Minimize antibiotic exposure by rapid de-escalation and/or discontinuation	Simple intervention Numerous supportive clinical trials in patient setting	Unknown best timing Lack of reliable biomarkers to assure safety Unclear best clinical endpoint Possible reluctance amongst clinicians
Nonselective luminal adsorbents	Adsorb the luminal fraction of antibiotics and prevent exposure of microbiota	Simple intervention	Adsorption of critical nonantibiotic drugs Difficult to implement during chemotherapy No published data in patient setting
Selective luminal antibiotic degraders	Degrade the luminal fraction of antibiotics and prevent exposure of microbiota	Simple intervention Encouraging efficacy in patient setting (respiratory tract infection)	Degraders of most antibiotic classes are not available
Prebiotics	Help the beneficial microbes by stimulating their growth	Simple intervention High likelihood of safety	Questionable history of efficacy May not fix large microbiota injuries No published data supporting efficacy in patient setting
Probiotics	Restore a critical component of the microbiota	Simple intervention Commercial products available Safety data in patient setting	May not fix large microbiota injuries Risk of microbial translocation Minimal published data supporting efficacy in patient setting
Fecal microbiota transplantation	Restore entire microbial community	Comprehensive microbiota restoration Strong clinical efficacy (C. difficile infection)	Risk of microbial translocation Unknown effects of a new microbial community in a new host (ie, microbiota-host “mismatch”) Donor involvement in allogeneic products Technical challenges in production, storage, and administration Limitation of autologous products in patient settings because of possible baseline alterations

Antibiotic de-escalation.

Antibiotics are the major culprit in dysbiosis during intensive chemotherapy; hence, rapid de-escalation of broad-spectrum antibiotics initiated for empirical treatment of NF may reduce the severity of microbiome injury and prevent infectious complications. Carbapenem exposure increases the risk for VRE colonization, a strong risk factor for VRE BSI.⁸¹ In addition, carbapenem exposure for >72 hours during intensive AML chemotherapy was associated with higher risk of subsequent all-cause infection in the 90 days after hematopoietic recovery.⁷⁵ A reduction in carbapenem use as empirical therapy for NF can reduce VRE colonization, VRE bloodstream infection, hospitalization length, and costs.^82,83 Controlled implementation of strict antibiotic stewardship in one study reduced carbapenem use, without worsening clinical outcomes such as intensive care unit transfers, bacteremia, and mortality.⁸⁴

An observational study of antibiotic de-escalation (either deleting one of the empirical antibiotics from a combined regimen or using a beta-lactam antibiotic with a narrower spectrum of activity) in the ICU setting for patients with neutropenia and sepsis did not yield evidence for increased mortality within 30 days or 1 year.⁸⁵ The best evidence supporting an early de-escalation/discontinuation strategy in patients with NF comes from a large randomized study in patients with hematological malignancies or HCT recipients with NF. Empirical antibiotics were stopped in hemodynamically stable patients in the experimental group after 72 hours of defervescence, while treatment was continued in the control group until neutrophil recovery. The study met its primary endpoint. The experimental group achieved a mean 2.5-day increase in empirical antibiotic-free duration, with no compromise in clinical outcomes. In an older randomized study, children with cancer-related NF were evaluated, and those deemed low-risk (no identifiable focus of bacterial infection, hemodynamic stability, negative blood cultures, and lower serum C-reactive protein) were enrolled. Empirical antibiotics were stopped on day 3 in the experimental group (regardless of defervescence or lack thereof) and continued until neutrophil recovery in the control group. The groups had similar clinical outcomes.⁸⁶

Together, there is fairly strong evidence to support an early de-escalation/discontinuation strategy for escalated antibiotics used to treat NF when the etiology remains unknown, but patients remain clinically stable. The optimal timing of such a strategy has not been determined, but could be based on defervescence, calendar days irrespective of fever, or biomarkers of ongoing infection or inflammation.

Oral activated charcoal.

Oral administration of nonspecific adsorbents such as activated charcoal can protect the gut microbiota by irreversibly adsorbing the luminal fraction of antibiotics. Such products will not influence systemic levels of intravenously administered antibiotics unless the antibiotic (eg, metronidazole)⁸⁷ undergoes significant enterohepatic recycling. For orally administered antibiotics with desired systemic effects, the site of action of the ideal adsorbent would be distal to the absorption site for the target antibiotic. DAV132 (Da Volterra, Paris, France) consists of encapsulated activated charcoal.⁸⁸ Targeted delivery of DAV132 to the terminal ileum (and more distal sites) permits sequestration of the unabsorbed fraction of antibiotics that reaches the terminal ileum, and before this fraction can injure the colonic microbiota. Targeted delivery of DAV132 was accomplished by external coating with a pH dependent enteric polymer.⁸⁸ An initial proof-of-concept trial in 18 healthy subjects who received a single oral dose of amoxicillin (absorbed in the proximal intestine) and sulfasalazine (metabolized and absorbed in the cecum) indicated that DAV132 adsorbed substances in the proximal colon without interfering with drug absorption in the small intestine.⁸⁸ These encouraging results led to a subsequent randomized controlled trial in 44 healthy subjects who received a 5-day course of moxifloxacin (a fluoroquinolone) with (n = 14) or without (n = 14) DAV132 and 2 control groups (8 subjects each) receiving DAV132 alone or an inactive substitute. DAV132 was administered orally 3 times daily before meals for 7 days. In the group receiving DAV132 and moxifloxacin, both drugs were given on days 1–5, followed by 2 days of DAV132 alone. DAV132 was highly effective in sequestering moxifloxacin, decreasing its free fecal concentration by 99%, with no effect on plasma levels. DAV132 protected the richness of the gut microbiota in antibiotic-treated individuals; 93% of the metagenomics species affected by moxifloxacin were at least partially protected by DAV132. No clinical adverse effects were observed. In an ex vivo analysis, DAV132 adsorbed at least 90% of the 14 tested antibiotics from 6 categories.⁸⁹

A randomized phase 2 clinical trial (NCT03710694) evaluated the safety and efficacy of DAV132 in hospitalized adult patients at high risk for CDI and who received fluoroquinolones for the treatment of acute infections or for prophylaxis in patients with neutropenia. Patients in the experimental arm received 15 grams per day of oral activated charcoal for the duration of fluoroquinolone plus an additional 2 days, whereas the control group received fluoroquinolones only. Patient recruitment has been completed, but the results of this trial have not been reported. DAV132 is under investigational new drug by the FDA and has a Conformit e Europ eenne mark in Europe.

Currently, safety and efficacy data for the use of activated charcoal in cancer patients are lacking. Intensively treated patients with cancer frequently receive several supportive care oral medications such as anti-emetic drugs, antidiarrheal drugs, and medications to prevent transfusion reactions. Because of the nonselective, high adsorptive potency of activated charcoal, demonstrating safety in highly vulnerable patients undergoing intensive chemotherapy is critical before an efficacy trial can be attempted. One safe way to ensure lack of undesired adsorption of beneficial medications is using their intravenous formulations (if available) or alternative intravenous medications. However, this may be logistically and financially difficult. Substantially more work needs to be done before nonselective adsorbents such as activated charcoal can enter clinical trials in intensively treated cancer patients.

Beta-lactamases and metallo-beta-lactamases.

Another strategy to counter antibiotics in the intestinal lumen is by antibiotic-specific or antibiotic class-specific enzymatic elimination. Ribaxamase (SYN-004) is an oral beta-lactamase designed to degrade intravenous beta-lactam antibiotics (penicillins and cephalosporine beta-lactams) in the upper small intestine (because of enteric coating) and prevent dysbiosis. In a double-blind, placebo-controlled clinical trial, patients older than 50 years and receiving inpatient ceftriaxone for at least 5 days for a lower respiratory tract infection received ribaxamase vs placebo until 72 hours after the last antibiotic dose. A statistically significant 2.4% absolute risk reduction for C. difficile infection (primary endpoint) was observed in the experimental group.⁹⁰ Although this reduction was only 2.4%, it represented ~60% risk reduction from the 3.4% actual infection rate in the control group. In a subsequent canine study, ribaxamase was enteric coated to be protected against gastric acid and to be released in the upper small intestine. This strategy poses a problem against ribaxamase use with oral beta-lactam antibiotics because the absorption of beta-lactams in the upper small intestinal will be compromised in the presence of ribaxamase. Therefore, a delayed release formulation of ribaxamase (SYN-007) was engineered to allow enzyme release in the lower small intestine, distal to the site of oral antibiotic absorption. In a canine study, ribaxamase did not affect serum levels of amoxicillin (administered orally for 5 days), but prevented dysbiosis characterized by diversity loss and emergent antibiotic resistance.⁹¹ SYN-006 is an oral metallo-beta-lactamase with additional carbapenemase activity, isolated from Bacillus cereus and enteric coated; the drug was tested in pigs receiving intravenous ertapenem for 4 days. SYN-006 mitigated antibiotic-induced dysbiosis (emergence of various antibiotic resistance genes) without altering systemic levels of ertapenem.⁹² Considering that cefepime and carbapenems are among the most common broad-spectrum antibiotics used in patients with acute leukemia, these novel agents hold promise for future clinical application.

Prebiotics.

Prebiotics are substrates that are selectively utilized by host microorganisms conferring a health benefit.⁹³ Although the focus in this review is the gastrointestinal tract, prebiotics can also be administered directly to other colonized sites such as skin. Major groups of prebiotics include conjugated linoleic acid, polyunsaturated fatty acid, human milk and other oligosaccharides, phenolics and phytochemicals, and readily fermentable dietary fibers. There are also nonprebiotic substances that may positively affect the microbiome; examples in this category include vita-mins and minerals.⁹³ In a pilot study, unmodified potato starch induced butyrate production from the microbiota; however, responses were highly varied and seemed to depend on the relative abundance of the butyrogenic species Eubacterium rectale.⁹⁴ A randomized controlled trial of oral resistant starch in prevention of radiation-induced proctitis in women with cervical cancer did not find a difference between the experimental and control arms in the severity of clinical or functional proctitis, diarrhea, or stool short-chain fatty acid concentrations.⁹⁵ An ongoing randomized phase 2 trial (NCT02763033) is investigating whether short-term (~100 days) administration of potato-starch can increase butyrate concentrations in the gut and reduce the rates of acute graft vs host disease (GVHD) after allogeneic HCT. Open questions include the correct amount and schedule of prebiotic supplementation, correct timing of the intervention, and the possible dependence of response on the initial composition of the microbiota. The authors are not aware of any completed study of prebiotics in patients with acute leukemia.

Probiotics.

Probiotics are live microorganisms that, when administered in adequate amounts, confer a health benefit on the host.⁹⁶ Our focus in this review is on probiotics ingestible via the oral route. The evidence for a beneficial role for probiotics to alleviate chemo-therapy-induced gastrointestinal side effects is weak. A meta-analysis in 2018 of all randomized trials of probiotics as prophylaxis against cancer therapy-induced diarrhea did not find an effect. Data from any type of cancer therapy study (cytotoxic, targeted, and immuno-therapies) was included in this meta-analysis.⁹⁷ In a pilot study of pediatric patients with acute leukemia undergoing intensive chemotherapy, patients were randomized to receive Lactobacillus rhamnosus GG twice daily for 7 days or no probiotics. The study met its primary endpoint of reducing the rate of gastrointestinal side effects, particularly nausea, vomiting, and abdominal distension.⁹⁸ One rationale for using probiotics is competitive inhibition of pathogenic bacteria in the gut and thereby limiting their translocation. In a study of 14 intensively treated patients with AML, patients received the probiotic strain Enterococcus faecium M-74 enriched with selenium, starting between days −2 and +2 and continuing until neutrophil recovery. The study did not meet its primary endpoint of reducing the incidence of NF; all patients developed NF, none of which were caused by the probiotic.⁹⁹

A conceptual challenge is that the entire microbiome or a large part of it, rather than a few species of it, may be essential for optimal physiological function. Restoring one or a few specific bacterial taxa, by a probiotic for example, may not restore the function that a complex, interconnected microbial community would serve for the host. In fact, manipulating only one component of the microbiome may have unpredictable and even adverse effects on the host. In one major murine and human study on this topic, the recovery of postantibiotic microbiota (composition, diversity, and function) was incomplete and significantly delayed after administration of commercially available probiotics.¹⁰⁰ In addition, bacterial transmission originating from the probiotic is a potential concern, especially in patients receiving cytotoxic therapy with toxicity to the gut barrier. The transmissibility of probiotic bacteria to the bloodstream has been well established in critically ill patients, in some cases with de novo mutations conferring antibiotic resistance.¹⁰¹ There is at least one reported case of septic shock due to probiotics after using probiotic-enriched yogurt in a patient with AML who underwent high-dose chemotherapy and autologous stem cell rescue.¹⁰²

Besides conventional probiotics such as lactic acid bacteria and bifidobacteria species, there is growing interest in “next-generation” probiotics including members from Clostridium clusters IV, XIVa, and XVIII; Faecalibacterium prausnitzii (with anti-inflammatory effects); Akkermansia muciniphila (desirable effects on glucose and fat metabolism); Bacteroides uniformis and fragilis (immune homeostatic effects); and Eubacterium hallii (short-chain fatty acid production).¹⁰³ The safety and efficacy of these probiotics have not been tested in patients with acute leukemia.

Fecal microbiota transplantation (FMT).

The premise of transferring large consortia of gut microbes to a patient from a donor (allogeneic) or the patient themselves (autologous) is to bring an entire community of micro-organisms, rather than components of it, to the host. Engraftment of FMT is defined based on achievement of progressive similarity of the reconstituting community to the donor’s fecal microbiome and is often, but not always, a required measure of success.^104,105 Reconstitution of the new community can restore the physiology of the original microbiome lost because of antibiotics and other insults. FMT can reconstitute colonization resistance and potentially even eliminate multidrug resistant pathogens. The source of FMT may be the patient (ideally, before dysbiosis occurs) or a healthy donor. The route of administration may be via enema, colonoscopy, nasojejunal/nasoduodenal tube, or oral (encapsulated forms). The clinical efficacy of FMT has been best demonstrated in eradicating recurrent C. difficile infection.^46,47

In the setting of intensive therapy, most of the published evidence for safety and efficacy of FMT comes from HCT studies. The gut microbiome of patients undergoing allogeneic HCT is disrupted by antibiotics used during conditioning and before engraftment.¹⁰⁶ Gut dysbiosis has been linked with acute GVHD in several studies,¹⁰⁷ and there is tremendous interest in restoring the microbiota to prevent and/or treat acute GVHD. The first FMT study in the allogeneic HCT setting was a case series of 3 patients with refractory acute GVHD who responded to FMT.¹⁰⁸ Another study of 13 allogeneic HCT recipients showed the efficacy of allogeneic FMT in restoring microbial diversity.¹⁰⁹ The efficacy of autologous FMT in restoring microbial diversity after allogeneic HCT was suggested in the preliminary analysis of a randomized trial of FMT vs placebo to prevent C. difficile infection.¹¹⁰ Another retrospective study suggested the efficacy of FMT in eradicating gut colonization with multidrug-resistant pathogens before or after HCT.⁴⁸ The completed, single-arm Odyssee study (NCT02928523; 20 adults) examined the effect of autologous FMT in patients with AML or high-grade myelodysplastic syndromes receiving intensive chemotherapy. The 2 primary endpoints were microbiota diversity recovery and eradication of multidrug-resistant bacteria on stool culture. The results of this study have not been published, but an abstract reported >90% diversity recovery 10 days after FMT, reduction of antibiotic resistance genes after FMT, and no serious adverse events within 30 days after FMT.¹¹¹ An ongoing randomized phase 2 trial (NCT03678493) of intensively treated AML patients receiving allogeneic oral encapsulated FMT vs placebo after each antibacterial antibiotic exposure will provide more definitive insight as to whether FMT reduces the rates of all-cause (primary endpoint) or specific infectious complications. There are numerous open questions about FMT in the setting of intensive therapy, including the optimal source, timing, frequency, and mode of administration, as well as the patient subgroups more likely to benefit. Pathogen transmission leading to bloodstream infection is a risk with this procedure,¹¹² and strict screening for microbial pathogens should be practiced in FMT product selection and preparation.

Bacteriophage therapy.

Targeting specific bacterial species in the gut by bacteriophages is a novel method that allows for precise manipulation of the microbiome. Phages can be selected and genetically modified to infect specific pathogenic bacteria, such as antibiotic-resistant species or those promoting mucosal inflammation.¹¹³ Bacteriophages targeting cytolysin-positive E. faecalis decreased cytolysin in the liver and attenuated alcoholic liver disease in humanized mice.¹¹⁴ One of the challenges in phage therapy is that because phages contain DNA/RNA, they can result in genome evolution in both the microbiome and host, with unpredictable consequences.¹¹⁵ As an example, although a lytic phage may directly eliminate only the specific susceptible bacteria, this effect can result in alterations of inter-bacterial interactions and community-level dynamics, eventually influencing the host.¹¹⁶ Perhaps most relevant to patients with acute leukemia is a study using a murine model of chemotherapy-induced neutropenia with bacterial translocation and bacteremia by a highly virulent multidrug resistant E. coli. Phage therapy reduced the systemic burden of the target bacteria.¹¹⁷

CHALLENGES AND FUTURE DIRECTION

The shortcomings of an exclusively antimicrobial practice to prevent and treat infections has become increasingly apparent; it is therefore time to incorporate the intestinal microbiota into practice to combat dysbiosis and its associated consequences.¹¹⁸ NF is a central event during intensive chemotherapy. It has been the primary event to prevent using antibiotics and the primary sign to trigger the use of even more antibiotics. Therefore, a nonantibiotic or supplemental alternative approach added to prophylactic antibiotics to limit NF may substantially reduce exposures to escalated antibiotics and, hence, dysbiosis and dysbiosis-related infections. Such an approach would likely incorporate measures to mitigate gut barrier damage. Unfortunately, there has been little progress in this area, which is a needed focus of future research. In addition, alternative approaches to minimize microbiota exposures to antibiotics are currently being tested in clinical trials, including nonselective adsorbents, selective antibiotic degraders, prebiotics, probiotics, and FMT. Nonselective adsorbents such as oral activated charcoal will likely need modifications to prevent their unintentional adsorption of multiple other nonantibiotic oral medications. In contrast, planned co-administration of nonabsorbable selective degraders of antibiotics such as beta-lactamases may be particularly suited in acute leukemia patients because of their very frequent exposure to beta-lactams. Among microbiota-restorative strategies, probiotics may benefit the host by partly reconstituting the microbiota, but their salutary effect is likely less than bringing the entire community of microbes back to the gut via FMT. However, transferring a carefully selected subset of the microbiota might be safer than an extremely complex microbiota community, especially from an allogeneic donor. These advantages and disadvantages must be considered when designing future clinical trials. Pathogen transmission can be life-threatening in the immunocompromised host and requires special precautionary assessment of the transferred microbial material. Importantly, evidence-based antibiotic stewardship should be expanded and strongly emphasized in inpatient and outpatient settings to minimize unnecessary exposures.

Several challenges need to be overcome. First, the mechanisms of the association between dysbiosis and clinical infections need to be more clearly understood. This is critical because the microbiota and host constitute an ecosystem that could be fundamentally, and sometimes permanently, altered if important interconnections are not understood when treating individual components. Second, because different animals have different microbiota, interspecies generalization of the results are likely confounded. Finally, suitable animal models of intestinal mucositis are limited and not universally agreed upon.¹¹⁹ This makes mechanistic studies of mucositis challenging because the intestinal tract is not directly and safely accessible in fragile acute leukemia patients at risk for bleeding and infectious complications.

Decades of clinical investigation to mitigate infections in acute leukemia patients undergoing intensive therapy have focused on antimicrobial strategies. The coming years will witness a shift in focus and incorporation of alternative, multifaceted approaches that incorporate host and microbiota factors as well as their mutual interactions in the management of vulnerable patients with acute leukemia.