VIewPoINT

microbial effects and contributions and to predict patient-specific responses to therapy The cancer while avoiding unwanted toxicities. Predictive machine-learning and Eran Elinav , Wendy S. Garrett, Giorgio Trinchieri and Jennifer Wargo artificial intelligence algorithms may greatly contribute to the ability of clinicians to Abstract | Collectively known as the microbiota, the commensal and other deliver personalized patient care. These microorganisms that colonize the epithelial surfaces of our body have been shown algorithms can unbiasedly process an to produce small molecules and metabolites that have both local and systemic enormous amount of patient data as input, including complex variables such as the effects on cancer onset, progression and therapy response. To date, most studies microbiota composition and function, focusing on the microbiome have used traditional preclinical mouse models and and predict patient-specific physiological identified correlative relationships between microbial species and cancer or clinical outcomes such as glycaemic phenotypes. Now, the profound influence of the microbiota on the efficacy of response to different foods5 and, in the cancer treatments, such as immunotherapies, has begun to be extensively context of cancer, can be used to predict characterized in humans. Paramount to the development of microbiota-based positive responders to particular treatments. Algorithms of this sort are expected to play a therapeutics, the next challenge in microbiome research will be to identify growing role in the field of oncology. individual microbial species that causally affect cancer phenotypes and unravel the underlying mechanisms. In this Viewpoint article, we asked four scientists Wendy S. Garrett. Ideally, we would model working on the cancer microbiome for their opinions on the current state of the the effects of the microbiota on cancer onset, field, where the research is heading and how we can advance our understanding progression and therapy with virtual avatars. Virtual avatars or digital twins are widely used to rationally design microbial-based therapeutics to transform treatment strategies in the aerospace industry. If jet engines have for patients with cancer. digital twins, why not people6? The virtual avatar is core to the achievement of bona fide How should we best model the of immune checkpoint inhibitor therapy personalized precision medicine. However, microbiota effects on cancer onset, in transplanted mice, whereas FMT from data scientists have some exciting challenges progression and therapy response and improve non-responding patients did not1. Similar to tackle before this becomes a reality. upon current preclinical models? findings were also observed by other In the near term, there are some bridging groups2, some integrating multi-omics with approaches that warrant further development Eran Elinav. The effect of the gut microbiota preclinical models to enable the identification and effort. Organoids represent one approach on cancer depends on an interaction between of several commensal members suggested to for many solid tumour malignancies4,7. three extremely complex, constantly evolving, contribute to the clinical effect. There is an increasing number of biobanks, biological entities — the microbiota, the Human-compatible preclinical models of both academic and commercial, that offer tumour and the . Unravelling cancer are becoming more sophisticated and human-derived organoids with excellent this intricacy throughout the clinical course will certainly contribute to the understanding genotyping, and organoid propagation is of malignancies is best modelled by an of the role of the microbiome in cancer. now within the financial reach of many labs. integrated approach combining microbiome For example, in vivo patient-derived Genetic manipulation of many different and host multi-omic characterization xenografts, which involve the implantation types of organoids is now achievable owing (for example, by 16S ribosomal RNA of human tumours in living mice, are to CRISPR–Cas technology8. An increasing amplicon sequencing, shotgun increasingly used for high-throughput number of types (for example, immune, metagenomic sequencing and metabolomic drug screening to discover patient-specific stromal and microbial) are being integrated characterization of the microbiome, and effective treatments3. In vitro models into organoid cultures, making these single cell RNA sequencing of the host) such as organ-chips are 3D organ-level systems better models of the tumour coupled with the use of preclinical models structures that simulate in vivo cancer microenvironment. Additionally, the ability to (that is, germ-free mice, patient-derived pathophysiology while conserving some flip organoids inside-out offers opportunities xenografts and organoids) and advanced aspects of its microenvironment4. This to model exposures (to microbial products or computational tools. This multimodal allows experimentalists to study tumour microbial metabolites) that influence cancer approach was recently utilized by some expansion, invasion and metastasis in onset and progression as well as more easily researchers in cancer studies with notable a highly controlled environment that is evaluate the effects of therapeutics that are success. For example, faecal microbiome otherwise unamenable for research with delivered luminally. transplantation (FMT) of stool samples in vivo models. Implanting a patient’s Mice remain an outstanding but far from from patients who responded to therapy individual microbiome and tumour cells in perfect in vivo model for cancer studies. into germ-free mice improved the effect such an in vitro system may help to integrate Site-specific recombinase, CRISPR–Cas and

Nature Reviews | CancEr Viewpoint

inducible expression (small molecule and Giorgio Trinchieri. Most of the studies Colonization of mice with human viral) technologies enable the more efficient investigating the effect of the microbiota or commensals is a common procedure to capture of human within mice in specific bacterial species on cancer onset and evaluate the effect of the human microbiota a time-tunable and tissue-specific fashion. progression have utilized genetically induced on carcinogenesis or therapy. However, The technologies that humanize mice are or chemical carcinogenesis mouse models, human commensal bacteria do not always astounding, and some are rather simple. while the studies of cancer therapy have colonize the mouse in the same way as they Gnotobiotics (specialized husbandry been largely based on mouse transplantable do humans. Also, the response of the mouse practices that allow for experimental tumours. Indeed, the preclinical models mucosal and immune systems to introduced that are microorganism-free or that allowed the development of immune commensals is not always identical to that colonized with specific microbial consortia) checkpoint inhibition therapy mostly used of humans. Studies using monocolonization is really rather ‘low tech’ yet enables mice transplantable tumours. These models are are also hampered by the concern that, in to harbour human microbial communities often criticized as not being representative of the absence of an ecologically balanced via inoculation from tumours themselves human tumours. However, in studying the microbiome, the results obtained may or, more commonly, stool samples. effect of the microbiota on cancer therapy, not be representative of the role of that This husbandry practice, while far from genetic cancer models present difficulties individual bacterial species in human new, is enabling a tremendous amount including limited numbers of mutations pathology. Robust approaches are those of cancer microbiome studies that are and reduced immunogenicity compared using colonization of mice with a complete shedding light on why immuno-oncology with many human tumours. Transfection human microbiome with or without the therapeutics work or fail in patients, and efficiencies of the viral vectors used in those specific species or strain being investigated. such knowledge is being translated to models and the tumour-promoting effect A more accurate mechanistic analysis benefit patients1,9,10. Intrinsic to husbandry of inflammation may be affected by the can be obtained by using an ecologically is, of course, diet and other environmental presence and composition of the microbiota. and metabolically balanced consortium exposures (for example, light exposures This may be a source of confusion not of a small number of human commensal and xenobiotics). Microbiome and only in cancer therapy studies but also in bacteria. Genetic modification of the strains population health studies are revealing evaluating the role of the microbiota on being studied or the use of bacterial strains how important the environmental factors cancer onset and progression. Important with partially different may help in of food and pharmaceuticals are for caveats when using transplantable tumours identifying mechanisms and molecules by health, and increasingly, these factors are are that most cell lines originating from which the microbiome and defined strains being considered in making mice a more B6 mice have been infected by rearranged affect carcinogenesis or therapy. effective model for human cancer. Beyond replicating endogenous murine leukaemia In cancer therapy studies, the main humanizing mice via editing, (MLVs) and that MLV antigens are objectives are to identify the microbiota gnotobiotics, and diet and environmental tumour-associated rejection antigens12. Also, composition that favours therapy response and perturbations, there are also means to the common use of transfected artificial to devise methods to target the microbiome to humanize the mouse immune system antigens in transplantable tumours may improve therapy effectiveness. To study via adoptive transfer of human CD34+ generate mimicry with bacterial antigens, the role of human commensal bacteria, stem cells with autologous human creating resistance mechanisms that may not gnotobiotic mice (that is, germ-free mice) thymic grafts11. be real in autologous human tumours. reconstituted with from healthy donors, patients or human commensal-defined The contributors consortia have been used. Clinical trials using faecal microbiota transplant aiming to eran elinav is a professor in the immunology Department, weizmann institute of science, israel, improve the success of immunotherapy have and the Director of the Cancer-Microbiome Division, Deutsches Krebsforschungszentrum been initiated13. Gnotobiotic mice associated (DKFZ), Germany. His research focuses on deciphering the molecular basis of host–microbiome with the transplant donor microbiota may interactions and their effects on health and disease, with a goal of personalizing medicine and nutrition. ideally help to predict whether the transplant may be effective in improving the response Wendy S. Garrett is a professor of immunology and infectious diseases at Harvard and a medical to therapy. Most studies used gnotobiotic oncologist in the Gastrointestinal Cancer Center at Dana-Farber Cancer Institute. She is mice that were associated with patients’ faecal co-director of the Harvard Chan Center for the Microbiome in Public Health. Her laboratory seeks to carry out mechanistic microbiome studies focused on cancer, inflammatory bowel disease and microbiota shortly before being tested for 14,15 immune and epithelial cell development and function. tumour growth and response to therapy . However, the immune system of germ-free Giorgio Trinchieri is a US National Institutes of Health (NIH) Distinguished Investigator and Director of the Cancer and Inflammation Program, Center for Cancer Research (CCR), National mice does not develop normally, and the Cancer Institute (NCI), NIH. His research has focused for many years on the interplay between sudden exposure to commensal bacteria inflammation, innate resistance and adaptive immunity and on the role of pro-inflammatory induces inflammatory and immune stress cytokines and interferons in the regulation of haematopoiesis, innate resistance and immunity that may alter the response to therapy. against infections and tumours. The present focus of his laboratory is on the role of inflammation, The use of mice treated with antibiotics innate resistance, immunity and the commensal microbiota in carcinogenesis, cancer progression instead of germ-free mice adds complications and prevention or therapy of cancer. owing to the effects of antibiotics on host cells Jennifer Wargo is an Associate Professor of Surgical Oncology and Genomic Medicine at the and possibly the competition of the remaining University of Texas MD Anderson Cancer Center and is a co-leader of the Melanoma Moon Shot host commensals with the transferred human programme. Her primary interests are in response and resistance to targeted therapy for microbiome. An alternative approach is to melanoma and other cancers, neoadjuvant strategies and the influence of the gut and tumour associate breeding pairs of germ-free mice microbiome on immunity and antitumour immune responses. with the human microbiome and then test

www.nature.com/nrc Viewpoint

their progeny that have been exposed to the involved in carcinogenesis. A classic example appropriate preventive and therapeutic human microbiota from birth. More studies is seen in Helicobacter pylori, a World Health strategies for individuals, a primary are needed to analyse whether the faecal Organization (WHO) class I carcinogen, and second prevention measure and an transplant used for creation of gnotobiotic which causes gastric cancer. Other bacteria adjuvant therapeutic — as both a target mice reproduces faithfully the microbiota such as Fusobacterium spp. are associated and a treatment. One key challenge of the human donors and how stable a with colorectal adenocarcinoma, and is the execution of the appropriate human-type microbiota is in mice through patients with colon cancer have an increased population-health-scale studies for the time and generations. abundance of Escherichia coli21,22. microbiome in cancer. We desperately Further to being involved in cancer need studies of the microbiome both across Jennifer Wargo. Now that we have causation, the microbiota may also the cancer continuum and across cancer established that the microbiota can impact contribute to responsiveness or resistance to types on a greater scale — thousands and cancer onset, progression and therapy chemotherapy treatment regimens. Exciting tens of thousands of subjects rather than response, we need to better understand new antibody-based immune checkpoint hundreds. Hand in hand with the need the mechanisms behind this via analysis inhibitors show variable efficacies, with for population-health-scale studies is the of additional patient cohorts and through treatment success suggested to be influenced continued commitment to mechanistic optimized in vitro and in vivo models. by host factors and gut microbiota microbiome studies to move beyond Integrating data from existing and composition23. Furthermore, microbiota taxa correlation, pinpoint mechanism — to prospective cohorts of patients with cancer residing within tumours have been found the extent to which one can in preclinical is critical to this effort. However, this is to confer tumour chemo-resistance brought models — and validate host–microbiome difficult, as a wide range of techniques have about by microbial drug metabolism24. targets using multiple complementary been used in these studies (for example, with Overall, I envision the future of cancer assays. Also, it is important to point out differences in how samples were collected, care as involving a holistic treatment that the within the human stored and processed and in the sequencing approach personalized to patient genetic and microbiome, or more precisely those within methods and the pipelines used for analysis). microbiome characteristics. Involvement the microbiota, are not the only microbial An important and necessary first step will of gut microbiota species in carcinogenesis taxa that live within and on the human body. be to harmonize and standardize such or in modulation of treatment efficacy may The human microbiome also encompasses approaches so that data may be more easily also pave the way towards new interventions the proteins and metabolites produced by compared between groups16. altering microbiota composition and individual members of the community, In such studies, it is important to function. For example, prebiotic or by larger networks within the microbial remember that these gut microorganisms personalized nutritional approaches community and by humans in concert with are not sole influencers of cancer and may alter the microbiome configuration the microbiota (for example, cometabolites). that numerous other factors are at play — towards one that favours cancer treatment including host genomics, host immunity, responsiveness. Patient-tailored probiotics G.T. Assaying microbiome composition environmental exposures (such as may supplement commensals crucial for cancer diagnosis has been proposed for ultraviolet radiation and smoking) and for cancer treatment success. ‘Postbiotic’ a few types of human cancers; however, tumour-intrinsic factors (such as alterations interventions, consisting of molecules the most promising results are those based in the genome, epigenome and tumour generated or modified by commensal on the identification of certain strains of microenvironment)17,18. Thus, these factors bacteria, may enable the supplementation Fusobacterium spp. as an independent must be taken into account as we assess data or inhibition of microbiome-derived small diagnostic assay for colon cancer25. The from new studies and as we optimize existing molecules, thereby impacting the human approach has generated interest because and novel preclinical models. The interaction host while bypassing the variable microbial of the low invasiveness of the test but has of microorganisms with each of these other ecosystem itself. In cases in which bacterial not yet reached a high level of accuracy, factors must also be carefully considered, as elimination is a need, novel approaches and it would not detect colon cancer bidirectional influences have been noted19. such as phage cocktail treatment may help associated with bacteria other than With regard to preclinical models, to eliminate cancer-promoting bacteria Fusobacterium spp.25. germ-free mouse models are an important while avoiding disadvantageous alterations Because certain bacteria, when resource; however, critical efforts are needed to the microbiota as a whole. Impacting the administered systemically, tend to to optimize novel models for microbiome host side of the host–microbiome interface accumulate and proliferate selectively in the research. This includes models such as the may enable gut barrier function to be anaerobic microenvironment of tumours, ‘gut-on-a-chip’20. relaxed, thereby allowing better influx of genetically modified bacterial strains have chemotherapeutic drugs, or alternatively been proposed to be used in cancer therapy How do you envisage our understanding the barrier to be tightened, thereby avoiding in a therapeutic approach that is promising of the microbiome being used in cancer microbial influx inducing infectious and and worth pursuing26. diagnostics and therapeutics, and what are the inflammatory adverse effects. Collectively, Recent data in experimental animals key challenges for ? I envision these modalities to be used in and to some extent in patients showed combinations in various patient-specific, that the composition of the gut microbiota E.E. Understanding the contribution of cancer-specific and symptom-specific modulates the efficacy of cancer the microbiome to cancer development, contexts in optimizing cancer patient care. chemotherapy and immunotherapy and progression and treatment responsiveness that targeting the microbiota could lead could revolutionize patient management W.S.G. There is tremendous opportunity to an increased immunotherapy success strategies. It is now established that some for the microbiome as a prognostic rate14,15,27,28. Several roadblocks, however, distinct members of the microbiota are biomarker, a guide for the selection of still exist. Colonization of mice with

Nature Reviews | CancEr Viewpoint

patients’ microbiomes has been used to enhance the therapeutic efficacy of immune In addition to its role as a potential characterize the mechanisms by which checkpoint inhibitors9. On the basis of these biomarker, there is intense interest and certain microbiota compositions enhance results, a similar consortium is planned to ongoing efforts to target gut microorganisms the response to immunotherapy14,15. be tested in patients with cancer treated with within the microbiota therapeutically to However, as discussed above, the human anti-PD130. Trials have also been initiated impact a number of medical conditions microbiome transferred into mice does using oral treatment with a monoclonal — including cancer. Such efforts have not always perfectly reproduce the donor microbial product (a single strain of demonstrated marked success in diseases such microbiome; it may be unstable, and Bifidobacterium spp.) that was found to as Clostridium difficile colitis — a condition the response of the mice to the human be associated with favourable response characterized by profound dysbiosis and microbiome may not be identical to that of to anti-PD1 and anti-PD1 ligand 1 (PDL1) overgrowth of a specific bacterial species the patients. Studies in anti-programmed in mice and in patients2,31. in the gut. Refractory cases can be treated cell death 1 (PD1)-treated patients via modulation of the gut microbiota via identified bacterial species that appear J.W. The microbiome is emerging as a FMT — demonstrating proof of principle for to correlate with successful response, potential biomarker as well as a tractable such an approach34. Strategies to modulate but unfortunately, each study identified therapeutic target in improving responses to the gut microbiota are now being used in completely different and unrelated species15. cancer therapy. Perhaps the most compelling the treatment of cancer; however, there are Thus, a challenge remains to identify clinical data to date are in the setting extensive considerations in using such an reliable microbiome-related biomarkers for of treatment with immune checkpoint approach with respect to the type of strategy prediction of response. These controversial blockade, where several studies have now to use, conditioning regimens and numerous results may be due to the heterogeneity demonstrated differential signatures in other considerations35. Examples include of the human microbiome between gut microorganisms of responders versus use of FMT from healthy individuals versus individuals and in different geographical non-responders (with cohorts of patients from patients with cancer who experienced areas. Particularly instructive in this respect with several different cancer types treated a complete response after being treated with is a recent study in southern China29. with monoclonal antibodies targeting immune checkpoint blockade therapies13,36, as In that study, the district in which the cytotoxic T lymphocyte antigen 4 (CTLA4) well as use of defined bacterial consortia (with individuals lived was shown to be the major and/or PD1)1,2,10,32,33. or without pre-conditioning regimens31,37) on determinant of microbiome diversity, and The development of the microbiome as the basis of insights gained from preclinical the microbiome composition was predictive a biomarker is appealing, as several studies and clinical studies. Certainly, we must work of susceptibility to metabolic disease within have demonstrated strong associations of together as a global community as we move each district but not across districts29. Also, specific gut microbiome signatures with forward with such approaches to learn how the individual bacterial species identified response to immune checkpoint blockade1,2 best to use them. in different clinical studies correlating with — with signatures in the gut microbiota Beyond the gut microbiome, the tumour response may just be the tip of an iceberg of outperforming other known biomarkers in microbiome and microbiota at other more complex ecological changes. selected studies2. However, we are clearly sites (such as the skin, aerodigestive tract Faecal transplant trials have been planned in the early stages of the development of these and other sites) must also be taken into or initiated to treat patients undergoing signatures as biomarkers, and complexities consideration given their potential influence immunotherapy13. Because we are still with such an approach certainly exist. and impact38. unable to define a favourable microbiome, Specifically, regarding gut microorganisms, these clinical protocols have been based it is unclear which metrics are most How do we advance towards showing on the transfer of faecal microbiomes from important (diversity of the microbiota, causal rather than correlative patients with cancer that have successfully relative abundance of specific bacterial taxa relationships between microbiota species and responded to anti-PD1 therapy into patients or functional status of the microorganisms). cancer phenotypes? who have failed therapy13. When we are able Furthermore, numerous methods exist to to define a favourable microbiome for cancer profile the microbiota (including PCR-based E.E. Indeed, while the association between therapy, it would be preferable to utilize approaches, 16 S sequencing, metagenomic commensal microorganisms and various balanced faecal microbiomes from healthy sequencing, metabolomic profiling, features of cancer has been shown in donors rather than dysbiotic microbiomes culturomics (culturing specific taxa from many studies and in different cancer from sick patients. It will also be important samples) and other strategies), and it is types, a mechanistic proof of causality to characterize the effect of diet in unclear at present which approach should be constitutes a major challenge of the field. improving therapy efficacy by modifying the used (in the short term, as well as in the long Different strategies are being utilized composition of the microbiota. term). Several factors should be taken into in demonstrating causal roles of whole Rather than faecal transplant that may consideration when contemplating using microbiome signatures or of distinct also transfer pathogens or pathobionts, such approaches in patients, such as the bacterial strains in cancer treatment. an important goal remains to identify the length of turnaround for a particular assay These include FMT (replacing one mechanisms shared by different bacteria and the positive and negative predictive microbiome with another), administration that enhance therapy response and to value. Ultimately, the gut microbiome of broad-spectrum antibiotics (depleting design ecologically balanced consortia of should be used in conjunction with other the microbiome) and transfer of whole commensal bacteria that could enhance known and novel biomarkers (optimally microbiomes, microbial consortia or isolated therapy response in any clinical setting. via an integrated approach) to improve strains into cancer-harbouring germ-free A consortium of human commensals was diagnostic accuracy — though strategies mice. For example, Sethi et al.39, reported shown in mice to induce interferon-γ using integrative biomarkers are somewhat that gut microbiota depletion, using a (IFNγ)-secreting CD8+ T cells and to under-developed at present. broad-spectrum cocktail of oral antibiotics,

www.nature.com/nrc Viewpoint

induced a significantly reduced melanoma This includes in the setting of treatment Likewise, bacteriophages may hold a burden in mice. This phenotype was related with immune checkpoint blockade, where predator–prey relationship with the bacterial to an altered immune cell balance in the studies showing associations in patient microbiome, therefore potentially driving tumour microenvironment, that is, increased cohorts were bolstered by data in mouse cancer-associated bacterial expansion, which numbers of antitumour IFNγ-secreting models demonstrating that responder may impact tumorigenesis or the response to T cells (T helper 1 cells and cytotoxic T cells), and non-responder phenotypes could therapy. Of note, microbiomes other than the and decreased numbers of immune be recapitulated by FMT into germ-free gut microbiome, including the skin, mouth, populations secreting both the pro-tumour mouse models1,2,10. However, despite these genitourinary and respiratory microbiomes, interleukin-17A (IL-17A) (IL-17A+CD3+) publications showing specific bacterial taxa may also contribute to local cancer formation, and IL-10 (IL-10+CD4+CD3+)39. In another associated with response versus non-response progression, metastasis and response to study, FMT from patients (responder or among the cohorts, there was little to no treatment but are currently much less studied non-responder to immunotherapy) into overlap between each of the cohorts. This has than the dense gut microbiome. Collectively, germ-free mice transferred the phenotype been associated with some angst in the field, these ecosystems will constitute exciting new to recipient mice1. Mice that received and careful analysis from expert investigators frontiers in cancer microbiome research in responder FMT showed a higher density suggests that functional approaches (such as the next decade. of CD8+ T cells than those that received analysis of microbial gene expression using non-responder FMT, consistent with human RNA sequencing or metabolomics) will likely W.S.G. Thinking of the microbiota as just data showing a higher density of CD8+ be necessary to help identify the fundamental bacteria is regretfully narrow regarding T cells in baseline samples of responders mechanisms mediating different phenotypes15. the multi-faceted roles of the microbiota versus non-responders. Ongoing and future analyses should take in cancer. There are viral sequences (phage these findings into consideration and include and non-phage) in tumoural and stool W.S.G. Multidisciplinary team science that standardized approaches to sequencing as samples from patients with cancer, and they involves ‘wet lab’ experiments in robust well as functional assays to gain better insight. represent a fascinating signal that warrants and reproducible preclinical models (more Key to these efforts are ongoing and focus and investigation. Beyond eukaryotic than one model should be used to validate planned clinical trials, of which there are organisms such as fungi (mycobiota), there observations whenever possible) is necessary many35 — including efforts to modulate are long-standing associations between and essential for moving from identifying the gut microbiota using FMT and human cancer and non-human eukaryotic hypothesis-generating correlations to laying administration of bacterial consortia or organisms, for example, the trematode the groundwork for mechanisms of causality. probiotics, as well as dietary modulation. Schistosoma haematobium and bladder These trials should include intense biomarker cancer and the trematodes Opisthorchis G.T. There is still much that can be done to assessment, ideally with standardization and viverrini and Clonorchis sinensis and gall optimize the interpretation of the clinical data harmonization of profiling techniques and bladder and bile duct cancers43. Beyond and to obtain solid results unambiguously collection of metadata. cancer onset, progression and therapy supportive of correlation and possibly hinting Efforts to modulate the microorganisms response, microorganisms can also at causative effects. Strict standardization at tumour sites or other sites in the body contribute to the morbidity and mortality for sample collection, bacterial lysis, DNA are also underway, though these efforts are of patients with cancer through infections. purification sequencing, bioinformatics less advanced than those targeting the gut Cancer treatment can put patients at risk and statistical analysis should be applied microbiota. of typical and atypical infections following to all the clinical studies to improve the treatment-related immune compromise, ability to compare the results in different Beyond bacteria, could there be a role hospital-acquired infections and multidrug- trials and clinical centres. For the analysis for other microbial groups, such as resistant microbial infections and second to and interpretation of clinical studies using the virome and mycobiome, in cancer onset, complex clinical exposures and history. in-depth multiomics investigation of both progression and therapy response? host cells and commensal microorganisms, G.T. Except for the role of H. pylori in stomach the most advanced systems analysis and E.E. Indeed, the gut virome, parasitome cancer, all the microorganisms officially machine-learning approaches should be and fungome (mycobiome) are potentially recognized as human carcinogens are viruses utilized. Evidence of causal relationships will important microbiome components and parasites44. Evidence is now emerging also be inferred from the results of clinical that are much less studied than the for a role of fungi in upper gastrointestinal trials using faecal microbiome transplant bacterial microbiome. Some infectious neoplasia45. The role of components of the or defined bacterial consortia. Eventually, non-commensal DNA and RNA viruses microbiome other than bacteria, including precise mechanistic studies will have to rely (that is, human papilloma , human bacteriophages, in modulating cancer therapy on the use of gnotobiotic mice, with the herpesvirus 8, Epstein–Barr virus, still needs to be fully investigated. caveat of differences between human and cytomegalovirus, hepatitis C virus and human mouse physiology. T lymphotropic virus 1) are widely known J.W. Although many studies have focused to be oncogenic, and these may play a role mainly on bacteria and their roles in cancer J.W. Although much of the published data in tumour pathogenesis in certain contexts. onset, progression and therapy response, regarding microbiota species and cancer Mycobiome alterations were suggested to other microorganisms (such as viruses, or therapy response phenotypes show only be associated with acute graft versus host protozoa and fungi) may also play an correlative relationships, causal relationships disease40, squamous cell carcinoma of the important role. These entities are somewhat are beginning to be demonstrated in some tongue41 and colorectal cancer42. However, less well studied in the current literature, cases — with insights into mechanisms a clear causative connection is yet to be where many of the studies utilized 16S being gained. proved for these interesting associations. sequencing and focused solely on bacteria

Nature Reviews | CancEr Viewpoint

taxa without taking other microorganisms 6. Bruynseels, K., Santoni de Sio, F. & van den Hoven, J. 34. Wang, Y. et al. Fecal microbiota transplantation for Digital twins in health care: ethical implications of an refractory immune checkpoint inhibitor-associated into consideration. An example of these emerging engineering paradigm. Front. Genet. 9, 31 colitis. Nat. Med. 24, 1804–1808 (2018). other microorganisms is bacteriophages, (2018). 35. McQuade, J. L., Daniel, C. R., Helmink, B. A. & 7. Drost, J. & Clevers, H. Organoids in cancer research. Wargo, J. A. Modulating the microbiome to improve which are numerically more abundant than Nat. Rev. Cancer 18, 407–418 (2018). therapeutic response in cancer. Lancet Oncol. 20, any other class of microorganism in the gut 8. Fujii, M., Clevers, H. & Sato, T. Modeling human e77–e91 (2019). digestive diseases with CRISPR-Cas9-modified 36. US National Library of Medicine. ClinicalTrials.gov 46 microbiome and more diverse . There is organoids. Gastroenterology 156, 562–576 (2019). https://clinicaltrials.gov/ct2/show/NCT03353402 (2019). evidence that these viruses can play a major 9. Tanoue, T. et al. A defined commensal consortium 37. US National Library of Medicine. ClinicalTrials.gov elicits CD8 T cells and anti-cancer immunity. Nature https://clinicaltrials.gov/ct2/show/NCT03817125 (2019). role in mediating the therapeutic efficacy of 565, 600–605 (2019). 38. Byrd, A. L., Belkaid, Y. & Segre, J. A. The human skin strategies to modulate the gut microbiota, 10. Routy, B. et al. Gut microbiome influences efficacy of microbiome. Nat. Rev. Microbiol. 16, 143–155 (2018). PD-1-based immunotherapy against epithelial tumors. 39. Sethi, V. et al. Gut microbiota promotes tumor with data from a study demonstrating Science 359, 91–97 (2018). growth in mice by modulating immune response. that sterile faecal filtrate transfer (filtering 11. Wege, A. K., Melkus, M. W., Denton, P. W., Estes, J. D. Gastroenterology 155, 33–37 (2018). & Garcia, J. V. Functional and phenotypic 40. van der Velden, W. J. et al. Role of the mycobiome in out live bacteria) is effective in treating characterization of the humanized BLT mouse model. human acute graft-versus-host disease. Biol. Blood patients with C. difficile colitis47. Other Curr. Top. Microbiol. Immunol. 324, 149–165 (2008). Marrow Transplant. 19, 329–332 (2013). 12. Young, G. R. et al. Resurrection of endogenous 41. Mukherjee, P. K. et al. Bacteriome and mycobiome microorganisms have not been well in antibody-deficient mice. Nature 491, associations in oral tongue cancer. Oncotarget 8, studied in the tumour, gut and other 774–778 (2012). 97273–97289 (2017). 13. US National Library of Medicine. ClinicalTrials.gov 42. Coker, O. O. et al. Enteric fungal microbiota dysbiosis sites in patients with cancer — however, https://clinicaltrials.gov/ct2/show/NCT03341143 and ecological alterations in colorectal cancer. Gut 68, efforts are underway to do so. Such efforts (2019). 654–662 (2018). 14. Vetizou, M. et al. Anticancer immunotherapy by 43. Brindley, P. J., Costa, J. & Sripa, B. Why does infection will require a movement towards more CTLA-4 blockade relies on the gut microbiota. Science with some helminths cause cancer? Trends Cancer 1, comprehensive means of profiling the 350, 1079–1084 (2015). 174–182 (2015). 15. Gharaibeh, R. Z. & Jobin, C. Microbiota and cancer 44. IARC Working Group on the Evaluation of Carcinogenic microbiota (using metagenomic sequencing immunotherapy: in search of microbial signals. Gut Risks to Humans. Biological agents Volume 100 B. and other strategies), as well as optimization https://doi.org/10.1136/gutjnl-2018-317220 (2018). A review of human carcinogens. IARC Monogr. Eval. 16. Knight, R. et al. Best practices for analysing Carcinog. Risks Hum. 100, 1–441 (2012). of reference databases to fully characterize microbiomes. Nat. Rev. Microbiol. 16, 410–422 (2018). 45. Zhu, F. et al. Autoreactive T cells and chronic fungal these microorganisms. Nonetheless, this 17. Blank, C. U., Haanen, J. B., Ribas, A. & infection drive esophageal carcinogenesis. Cell Host Schumacher, T. N. The “cancer immunogram”. Microbe 21, 478–493 (2017). will certainly add substantially to our Science 352, 658–660 (2016). 46. Draper, L. A. et al. Long-term colonisation with donor understanding of the impact and diagnostic 18. Cogdill, A. P., Andrews, M. C. & Wargo, J. A. bacteriophages following successful faecal microbial Hallmarks of response to immune checkpoint transplantation. Microbiome 6, 220 (2018). and/or therapeutic potential of the blockade. Br. J. Cancer 117, 1–7 (2017). 47. Ott, S. J. et al. Efficacy of sterile fecal filtrate transfer microbiota in cancer. 19. Zmora, N., Soffer, E. & Elinav, E. Transforming for treating patients with Clostridium difficile infection. medicine with the microbiome. Sci. Transl Med. 11, Gastroenterology 152, 799–811 (2017). Eran Elinav 1,2*, Wendy S. Garrett3,4,5*, eaaw1815 (2019). Giorgio Trinchieri 6* and Jennifer Wargo 7* 20. Shin, W. & Kim, H. J. Intestinal barrier dysfunction Acknowledgements orchestrates the onset of inflammatory host- E.E. thanks all the students, postdocs and employees at the 1 Immunology Department, Weizmann Institute microbiome cross-talk in a human gut Elinav laboratory at Weizmann Institute of Science and of Science, Rehovot, Israel. inflammation-on-a-chip. Proc. Natl Acad. Sci. USA Deutsches Krebsforschungszentrum (DKFZ) for their continued 115, E10539–E10547 (2018). partnership. E.E. is a senior fellow of the Canadian Institute of 2Cancer-Microbiome Division, Deutsches 21. Rubinstein, M. R. et al. Fusobacterium nucleatum Advanced Research (CIFAR) and an international scholar of the Krebsforschungszentrum (DKFZ), Neuenheimer Feld promotes colorectal carcinogenesis by modulating Bill and Melinda Gates Foundation and Howard Hughes 280, Heidelberg, Germany. E-cadherin/beta-catenin signaling via its FadA adhesin. Medical Institute (HHMI). W.S.G. thanks members of her lab- Cell Host Microbe 14 3 , 195–206 (2013). oratory and collaborators for stimulating discussions as well Departments of Immunology and Infectious Diseases 22. Tahara, T. et al. Fusobacterium in colonic flora and as the US National Institutes of Health (NIH) National Cancer and Genetics and Complex Diseases, Harvard T. H. molecular features of colorectal carcinoma. Cancer Institute (NCI) and Cancer Research UK for research support. Chan School of Public Health, Boston, MA, USA. Res. 74, 1311–1318 (2014). J.W. thanks the patients who contributed to these studies, as 23. Bashiardes, S., Tuganbaev, T., Federici, S. & Elinav, E. well as the faculty, staff, students and postdoctoral fellows at 4 Broad Institute of Harvard and MIT, Cambridge, The microbiome in anti-cancer therapy. Semin. the University of Texas MD Anderson Cancer Center who con- MA, USA. Immunol. 32, 74–81 (2017). tributed to this work (which was supported by the Melanoma 24 5Department and Division of Medical Oncology, . Geller, L. T. et al. Potential role of intratumor bacteria Moon Shot programme). J.W. also thanks worldwide collabo- in mediating tumor resistance to the chemotherapeutic rators for providing critical insight and opportunities to learn Dana-Farber Cancer Institute and Harvard Medical drug gemcitabine. Science 357, 1156–1160 (2017). together. J.W. is a member investigator of the Parker Institute School, Boston, MA, USA. 25. Zhang, X. et al. Fecal fusobacterium nucleatum for the for Cancer Immunotherapy and is also an awardee of an diagnosis of colorectal tumor: a systematic review and Innovative Research Grant from Stand Up to Cancer–American 6Cancer and Inflammation Program, Center for Cancer meta-analysis. Cancer Med. 8, 480–491 (2019). Association for Cancer Research. J.W.’s research efforts are Research, National Cancer Institute, National Institutes 26. Forbes, N. S. et al. White paper on microbial also supported by a Sabin Family Fellowship, the Melanoma of Health, Bethesda, MD, USA. anti-cancer therapy and prevention. J. Immunother. Research Alliance and the NIH (1R011CA219896-01A1). Cancer 6, 78 (2018). 7Department of Surgical Oncology and Department 27. Iida, N. et al. Commensal bacteria control cancer Competing interests of Genomic Medicine, The University of Texas MD response to therapy by modulating the tumor E.E. is a paid consultant at DayTwo and BiomX. E.E. states that Anderson Cancer Center, Houston, TX, USA. microenvironment. Science 342, 967–970 (2013). none of the issues covered in this piece are related to, funded 28. Viaud, S. et al. The intestinal microbiota modulates or endorsed by, shared or discussed with or licensed to these *e-mail: [email protected]; e.elinav@ the anticancer immune effects of cyclophosphamide. commercial entities. W.S.G. is a member of the scientific advi- dkfz-heidelberg.de; [email protected]; Science 342, 971–976 (2013). sory boards for Evelo Biosciences, Kintai Therapeutics and [email protected]; [email protected] 29. He, Y. et al. Regional variation limits applications of Leap Therapeutics. G.T. declares no competing financial inter- healthy gut microbiome reference ranges and disease ests. J.W. is an advisory board member and consultant and https://doi.org/10.1038/s41568-019-0155-3 models. Nat. Med. 24, 1532–1535 (2018). adviser for AstraZeneca, Bristol-Myers Squibb, F. Hoffman–La Published online xx xx xxxx 30. Vedanta Biosciences. Bristol-Myers Squibb Roche–Genentech, GlaxoSmithKline, Merck Sharp & Dohme, and Vedanta Biosciences announce a new clinical MicrobiomeDX, Biothera Pharmaceuticals and Novartis 1. Gopalakrishnan, V. et al. Gut microbiome modulates collaboration to evaluate OPDIVO® (nivolumab) and Pharmaceuticals Corporation, receives grant, clinical and response to anti-PD-1 immunotherapy in melanoma VE800 in patients with advanced or metastatic research support from Bristol-Myers Squibb, F. Hoffman–La patients. Science 359, 97–103 (2018). cancers. Vedanta Biosciences https://www.vedantabio. Roche–Genentech, GlaxoSmithKline and Novartis Pharma­ 2. Matson, V. et al. The commensal microbiome is com/news-media/press-releases/detail/2492 (2018). ceuticals Corporation, is a speakers bureau participant with associated with anti-PD-1 efficacy in metastatic 31. US National Library of Medicine. ClinicalTrials.gov Bristol-Myers Squibb, Dava Oncology, Gilead Sciences, melanoma patients. Science 359, 104–108 https://clinicaltrials.gov/ct2/show/NCT03595683 (2018). Illumina, Imedex, MedImmune and Omniprex and receives (2018). 32. Frankel, A. E. et al. Metagenomic shotgun sequencing other financial or material support from a US patent applica- 3. Pauli, C. et al. Personalized in vitro and in vivo cancer and unbiased metabolomic profiling identify specific tion (PCT/US17/53.717) submitted by the University of Texas models to guide precision medicine. Cancer Discov. 7, human gut microbiota and metabolites associated MD Anderson Cancer Center that covers methods to enhance 462–477 (2017). with immune checkpoint therapy efficacy in immune checkpoint blockade responses by modulating 4. Sontheimer-Phelps, A., Hassell, B. A. & Ingber, D. E. melanoma patients. Neoplasia 19, 848–855 (2017). the microbiome. Modelling cancer in microfluidic human organs-on- 33. Chaput, N. et al. Baseline gut microbiota predicts chips. Nat. Rev. Cancer 19, 65–81 (2019). clinical response and colitis in metastatic melanoma Publisher’s note 5. Zeevi, D. et al. Personalized nutrition by prediction of patients treated with ipilimumab. Ann. Oncol. 28, Springer Nature remains neutral with regard to jurisdictional glycemic responses. Cell 163, 1079–1094 (2015). 1368–1379 (2017). claims in published maps and institutional affiliations.

www.nature.com/nrc