Introduction

Antibiotic therapy has produced unquestionable advances in the management of patients with cancer, a population with intrinsically higher risk of bacterial infection as a result of malignancy or treatment-related immune suppression.

While antimicrobial therapy has markedly reduced morbidity and mortality stemming from infection, the effects of broad-spectrum antibiotics on commensal, non-pathogenic bacterial species have remained for a long time an under-appreciated effect of this therapeutic class of drugs.

The gut microbiota, source of over 100 trillion bacteria, exists in a condition of mutually beneficial relationship with the host. Commensal bacteria are provided with a niche to colonise the host in return for their participation in the digestion of nutrients and xenobiotics, protection from pathogens and shaping of the host’s immune system subsets. Derangement of this delicate relationship has been increasingly well-characterised in the context of tumour-specific immune tolerogenesis [1].

Multiple levels of evidence now support the link between sensitivity to immunotherapy, taxonomic diversity and enrichment in specific gut bacterial taxa, suggesting that some species or species consortia provide intrinsic immune-modulating properties. The landmark study by Gopalakrishnan [2] demonstrated how broader stool bacterial diversity and higher representation of Ruminococcaceae communities including Faecalibacterium positively influences patients’ survival following ICPI by promoting a strongly immune-reactive microenvironment and lower systemic release of pro-inflammatory cytokines [3]. Many other commensal bacteria have subsequently been recognised to play a similar role including Bifidobacteria spp., a saccarolytic Gram-positive genus highly represented within the gut that facilitates dendritic cell maturation and increased accumulation of antigen-specific T-cells within the tumour microenvironment [4]. Similarly, the presence of the anaerobic commensal Akkermansia Muciniphila is more common in responders to ICPI, who display higher peripheral CD4 and CD8 memory T-cell responses to this bacterium [5].

Antibiotic (ATB) therapy imposes profound and protracted changes to the taxonomic diversity of the host microbial ecosystem, affecting the composition of up to 30% of the bacterial species in the gut microbiome [6], consequently leading to loss of microbial functions that are protective for the host. Such changes in gut microbial communities are rapid and pervasive, occurring within days from the first antibiotic dose [7] and persisting for up to several months after completion of therapy [8].

Mounting evidence from epidemiological studies has underscored the detrimental role of antibiotics in ICPI outcome, with exposure to antibiotics having been linked to shortened progression-free, overall survival and reduced response rates in patients receiving ICPI as part of clinical trials and in routine practice (Table 1). In a previous study, we demonstrated time-dependence of antibiotics exposure as a strong, tumour-agnostic determinant of outcome in ICPI recipients, confirming prior, but not concurrent antibiotic therapy as doubling the risk of primary progression to immunotherapy and leading to a > 20-months shortening in patients’ survival independent of established prognostic factors and corticosteroid use [10]. Whilst mirroring pre-clinical evidence, where antibiotic pre-conditioning ahead of tumour implantation leads to impaired responses to ICPI in mice [26, 27], the expanding body of clinical studies has so far painted an incomplete picture as to the mechanistic foundations underlying the relationship between ATB and immunotherapy, a point of greater consequence given the potential practice-influencing implications of ATB prescribing in the clinic.

Table 1 The relationship between antibiotic exposure and outcomes from immune checkpoint inhibitor therapy
Full size table

Most of the studies highlighting the importance of a healthy gut microbial environment as a pre-requisite for ICPI response were unfortunately characterised by insufficient data on preceding or concomitant antibiotic exposure, making it impossible to disentangle the role of antibiotic-induced perturbation of the gut ecosystem in influencing clinically meaningful outcomes in these patients [3].

Mechanistically, the breadth and depth of downstream effects produced by antibiotics within the cancer-immune synapse are an important challenge in studying this prognostically adverse relationship. On one hand, the direct bacteriostatic/bactericidal effect of antibiotics can cause selective pressure within the host microbial ecosystem and instigate an alternative microbiota state characterised, amongst other traits, by downregulation of major histocompatibility complex (MHC) class I/II genes and impaired effector T-cell responses, immunologic traits implicated in reduced responsiveness to ICPI [28].

ATB-induced depletion of gut bacteria can also shift the repertoire of microbial-associated molecular patterns (MAMPS). These molecules signal through mucosal innate immune cells primarily via toll-like receptors (TLRs) and NOD1 [29] to influence neutrophil priming, reduce local cytokine release and prime adaptive immunity by influencing the expression of MHC genes within the intestinal mucosa and reduce immunoglobulin secretion [30]. Antibiotic treatment impairs TH1/TH17 responses in tumour-bearing mice through direct pre-conditioning of the gut microbiota, reducing the efficacy of cyclophosphamide-mediate immune-rejection of the tumour [31]. In addition, antibiotics can also reduce the capacity of adoptively transferred CD8+ T-cells to mediate a tumour-specific response through altered LPS/TLR4 signaling in lymphodepleted mice [32].

By disrupting the gut ecosystem, antibiotics instigate downstream metabolic alterations within the microenvironment with complex repercussions to the tumour-host-microbe interface. Amongst them, changes in the availability of short-chain fatty acids produced by Akkermansia, Faecalibacteria and Enterococcus from the catabolism of non-digestible carbohydrates and the conversion of primary bile acids to secondary bile acids (including deoxycholate) mediated by Clostridiales can significantly alter gut homeostasis and lead to profound and clinically meaningful immune-modulatory consequences [33]. The immune-metabolic repercussions secondary to gut dysbiosis, potentially reversible by oral Akkermansia supplementation [34], might explain the influence of body mass index in determining response to ICPI [35, 36].

With improved characterization of immune-microbiologic underpinning of the relationship between antibiotics and ICPI outcome, a key question now is whether disruption of a well-equilibrated gut bacterial ecosystem is truly causal in this relationship, and thus whether reversal of antibiotic-mediated gut dysbiosis might prove beneficial in restoring full sensitivity to ICPI. Whether a favourable gut microbiota is a reflection of an otherwise healthy host rather than the primum movens of clinically meaningful anti-cancer immune responses is still the subject of intense debate [13]. To this end, appreciating how antibiotics might dynamically affect such a strong immune-microbiologic correlate of response to checkpoint inhibition is of key importance to pave the way for strategies that could restore or protect the integrity of this important phenotypic correlate of response. To address the multiplicity of mechanisms that are likely to underscore this complex and bi-directional relationship, the coordinated study of a number of fundamental pathophysiologic processes including bacterial translocation, immune-modulation, an altered metabolome, enzymatic degradation and reduced diversity of the gut microbiome has been proposed as an overarching framework [37].

Gaining sufficient insight as to the mode of action by which bacteria might work as biotherapeutic agents is not just important for patient prognostication, but is in fact key to a successful, rational development of microbiome-modulating therapies which improve patient’s outcome with ICPI. With antibiotic use now having been validated as an important and dynamic factor influencing outcome from immunotherapy, concerted efforts should be aimed at characterizing the candidate taxonomic features in the gut microbiota that are associated with worse outcome from ICPI in the context of preceding and concomitant antibiotic exposure and evaluate them in conjunction with the concomitant prescription of proton pump inhibitors, corticosteroids and vaccines, all of which have been postulated to influence ICPI response [38].

Recognising these changes is expected to facilitate the clinical development of diverse biotherapeutic approaches to induce microbiome reprogramming including dietary interventions with pre-biotics, therapeutic administration of single or multiple types of bacterial species or their metabolites, selective antibiotic therapy or faecal microbial transplantation, all of which are currently at the focus of intense clinical research efforts [26].