Exploring the impact of gut microbiota and diet on breast cancer risk and progression

Abstract There is emerging evidence that resident microbiota communities, that is, the microbiota, play a key role in cancer outcomes and anticancer responses. Although this has been relatively well studied in colorectal cancer and melanoma, other cancers, such as breast cancer (BrCa), have been largely overlooked to date. Importantly, many of the environmental factors associated with BrCa incidence and progression are also known to impact the microbiota, for example, diet and antibiotics. Here, we explore BrCa risk factors from large epidemiology studies and microbiota associations, and more recent studies that have directly profiled BrCa patients' gut microbiotas. We also discuss how in vivo studies have begun to unravel the immune mechanisms whereby the microbiota may influence BrCa responses, and finally we examine how diet and specific nutrients are also linked to BrCa outcomes. We also consider future research avenues and important considerations with respect to study design and implementation, and we highlight some of the important unresolved questions, which currently limit our overall understanding of the mechanisms underpinning microbiota‐BrCa responses.


| INTRODUCTION
Annually, breast cancer (BrCa) is predicted to affect over 2 million new patients, with more than 600 000 BrCa-related deaths worldwide, second only to lung cancer in incidence and mortality. 1 The financial implications of the disease on patients and health services are equally staggering with average treatment costs ranging from £22 000 to £115 000 for a single patient depending on disease stage. 2 Moreover, while patients diagnosed in early stages usually have good prognostic outcomes, those diagnosed in late stages of the disease have very poor 5-year survival rates, less than 30%. 2  which are classified according to the expression of several proteins. [3][4][5][6] Crucially, expression of these proteins determines which therapies clinicians employ to treat the disease such as neoadjuvant hormone therapies, chemotherapies and/or radiotherapies. 7,8 Successful treatment outcomes rely on efficient activation of anticancer responses, therefore delineating factors that beneficially or negatively impact associated immune responses are key. One such emerging modulator of BrCa aetiology is the gut microbiota, which may represent a viable therapeutic target for altering the course of the disease.

| BrCa RISK FACTORS: IS THE MICROBIOTA A MISSING LINK?
Alongside known genetic factors, all cancers are considered to have an "environmental" element associated with increased risk and disease progression. These associations are often uncovered in large human epidemiological studies that correlate lifestyle factors (eg, smoking, drug exposure and diet) with cancer onset and clinical outcomes. [9][10][11] As many of these factors are also known to alter the gut microbiota, population-based association studies at least intimate that the gut microbiota dictates cancer outcomes. 9,12,13 A stable and diverse microbial ecosystem (in adults) is considered optimal for health, although the exact taxa that confer beneficial effects is an ever changing conundrum, and is dependent on age, diet, medications, host genetics and numerous other external factors. In many cases, a reduction in alpha and beta diversity is associated with increased disease risk for many conditions, but correlations with specific microbial taxa may change between patient cohorts. In terms of "beneficial" bacterial members, those best studied include some of the following: Bifidobacterium, Lactobacillus, Akkermanisa, Ruminococcus, Bacteroides and Faecalibacterium; however, species and strain-level differences are key considerations. More recent studies indicate that overall function of the total microbial community may allow more "healthy" signatures to be found, as recently described for colorectal cancer (CRC) and choline degradation. 14 For reviews on this subject, see References 15 and 16. The gut-tumour connection includes sites known to have direct cross talk between the host and the gut microbiota (eg, CRC), but also in sites distal to the gut (eg, the skin, liver and breast). It is likely that, what applies in one cancer setting is by no means universal, and BrCa, by its extreme heterogeneity and relative low incidence of genetic predisposition, is particularly unique. Consequently, there are many large studies focused on understanding how different environmental factors influence BrCa, and each of these factors influence (and are influenced by) the microbiota. These include the following: 1. Diet: In the 1990s, several groups investigated the association between diet and BrCa risk. For example, a low-fat diet elicited a lower risk of relapse after tumour resection. 15 Recent metaanalyses of cohort studies continue to correlate dietary patterns with BrCa risk. 16 This topic is covered in more detail in Section 3 of this review.

2.
Obesity: Complementary to a low-fat diet, obesity is associated with increased risk of developing postmenopausal BrCa with a worse clinical outcome. Meta-analysis of nine studies showed increased BrCa risk with increased body mass index (BMI). 17 Associations between obesity and postmenopausal BrCa may be due to adipose tissue catalysing the formation of oestrogen after menopause, thereby increasing circulating oestrogen levels 17,18 ; see Point (4).

Alcohol consumption:
Excessive alcohol intake is also recognised as a risk factor for BrCa. 19 Although the specific molecular mechanisms driving this correlation remain unknown, ethanol may (a) induce molecular damage in mammary cells; (b) inhibit oestrogen-metabolising enzymes in the liver; (c) increase aromatase activity in the liver, which has been reported to facilitate the conversion of testosterone to oestrogen 20 ; see Point (4). 4. Changes in circulating hormonal levels: Alongside uterine, ovarian and prostate cancers, some forms of BrCa are oestrogen driven.
Both a late menarche and an early menopause decrease the risk of developing BrCa. 11 For a recent review on this subject, see Reference 21.
5. Antibiotic exposure: Use of antibiotics is becoming increasingly controversial, with unexpected adverse effects being reported in several disease contexts. [22][23][24]  One commonality between each of these risk factors is that they significantly alter the profile of the gut microbiota (see Table 2), suggesting a strong link between microbiota make-up and BrCa development. This suggests that a perturbed microbiota and therefore However, this is dependent on the baseline microbiota of individuals, and in many cases we do not understand the impact of these factors on specific microbial communities, species and strains, which may play a key and oversized beneficial role. Further studies are needed in this area, including within BrCa patient cohorts.

| GROWING EVIDENCE LINKING THE GUT MICROBIOTA AND BrCa
Changes in the abundance (ie, levels) of particular microbes have been associated with several cancers. A well-known example is Helicobacter pylori, which initiates gastric inflammation and the formation of precancerous lesions. 38  • Members of the microbiota can digest otherwise indigestible components of our diet (eg, dietary fibre) • Dietary fibre constituents can (a) boost nutritional intake, (b) act as a substrate for other microbiota members to colonise and (c) act as a metabolite 27 Short-chain fatty acids (SCFA), a constituent of metabolised dietary fibre, can module host immune responses Bioactive compounds, a constituent of metabolised polyphenols, encourage growth of beneficial bacteria for example, Bifidobacterium and Lactobacillus and production of SCFA 28,29 Obesity • Gut microbiota profiles differ among obese and lean patients and between those with metabolic syndrome 30 • In mice, studies showed that an obese microbiota profile had a greater nutritional intake capacity 31 Members of an obese microbiota profile encoded enzymes that could more efficiently degrade polysaccharides Alcohol • Perturbations of the gut microbiota profile was observed in alcoholics vs nonalcoholics 32 This resulted in lower abundance of Bacteroidetes and higher abundance of Proteobacteria Alcoholics also had higher levels of serum endotoxin • Alcoholics tended to have greater gut permeability, which could lead to a local inflammatory state and disease, for example, alcohol-related liver disease 33 • It can be hypothesised that changes in microbiota members due to alcoholism alter the metabolites available by the host to use for other physiological processes including gut barrier function

Hormones
• In 1998, a group observed that germ-free mice, which do not have a gut microbiota, regained normal oestrous levels upon accidental bacterial contamination This suggested a link between gut bacteria and reproductive capacity 34 • Microbiota members possess ß-glucoronidase, which can deconjugate already metabolised oestrogen Thereby increasing levels of systemic oestrogen, increasing the risk of ER+ breast cancer 35 • A population-based study demonstrated an association between oestrogen metabolism and phylogenetic diversity of the gut microbiota, suggesting a link between the gut bacteria and circulating reproductive hormones 34

Antibiotics
• Antibiotics severely impact the gut microbiota, most notably they reduce microbial diversity After depletion due to antibiotics it became easier for pathogenic bacteria, for example, Salmonella to colonise due to lack of competitive exclusion 36 The change in microbiota members consequently influenced the availability of metabolites used by the host, which could influence, for example, host immune responses 36 classes Clostridiaceae and Ruminococacceae and in Faecalibacterium, and a relative decrease in abundance of Dorea and Lachnospiraceae. 42 In short, these studies suggest that postmenopausal BrCa patients do have an altered microbiota signature, which was reported in the Twin UK cohort study. 40

| ANIMAL STUDIES OF THE LINKS BETWEEN THE GUT MICROBIOTA AND BrCa
To date, there is still a relatively limited body of research exploring the role of the microbiota and different in vivo BrCa models, although interest is growing, and lessons learned from other cancers may be applicable in this underresearched area. Crucially, studies to date have indicated that microbiota modulation of the immune system may represent a key cross-roads determining disease and treatment outcomes. A distinct advantage of using preclinical models is the ability to explore the impact of microbiota on different BrCa subtypes, which is key given the heterogeneous nature of this cancer type (see Table 1).

| Evidence of microbiota involvement in non-BrCa disease
Understanding mechanistic links between the gut microbiota and BrCa is in its infancy. Thus, to better gauge the potential for the microbiota to influence BrCa occurrence and progression, it is important to con-  46 suggesting a protumorigenic influence of the bacteria.
Microbes have also been observed to play antitumorigenic roles, which in many cases appears to be via education of the host immune system. One of the seminal studies in the microbiome cancer field identified an association between Bifidobacterium abundance and reduced melanoma tumorigenesis in a subcutaneous allograft B16. PyMT-BO1 and EO771 tumours from antibiotic-treated animals. Notably, inhibition of mast-cell activation confirmed this immune population to be driving enhanced primary tumour growth after antibiotic-induced microbiota disturbances. Importantly, the same study also demonstrated that using a clinically relevant cephalosporin antibiotic promoted the same increase in primary tumour growth and was associated with similar increases in tumour mast cells. 55 Based on the similarity of the orthotopic models used in these two studies, it is likely that the differences in primary tumour growth between them were due to the differences in treatment regimen.
Buchta-Rosean et al allowed for a bacterial recolonisation period of 4 days, which likely aided in slowing primary tumour growth, possibly through an immunological re-priming. 24  The metabolome of the microbiota is also known to play a key role in host health, influencing an array of biological pathways including cellular proliferation, metabolism and immunity. Although studies focusing on microbial metabolites in BrCa models are very limited, a previous study determined that cadaverine (produced during microbial breakdown of animal tissue) supplementation reduced both primary and metastatic burden in an orthotopic 4T1 triple-negative-like BrCa model in BALB/c mice. 56 As highlighted previously, short-chain fatty acid (SCFAs, particularly butyrate) also have links to cancer and are known to promote an invasive/aggressive phenotype in BrCa in vitro. 57 Crucially, these microbial metabolites are derived after fermentation/metabolism of dietary components and therefore diet may act as a key overriding factor that impacts the microbiota, their metabolites and subsequent host interactions, leading to differential BrCa outcomes.

| DIET, THE GUT MICROBIOTA AND BrCa
As already mentioned earlier (and in Table 2), extensive epidemiological studies have laid the groundwork for understanding that diet has a major role on cancer risk and progression. [58][59][60] One of the key roles played by the microbiota is breakdown of complex dietary substrates into their constituent bioactive compounds; therefore, there is growing interest in understanding functional outcomes and the underlying mechanisms governing diet-microbe interactions with respect to cancer.

| Diet and BrCa
Although the correlations between BrCa risk and dietary intake have been intensively studied, the underlying associations or effector mechanisms remain poorly understood. Historically, increased risk of BrCa has been tied to high intake of red meat and animal fat, 61,62 with decreased risk being concurrently linked to fruit and vegetables consumption. 63  "healthy" diet effects are only significant in premenopausal women, but across receptor-positive and receptor-negative tumours. 16 Alcohol intake is also a significant risk factor, with high consumption linked with disease recurrence and reduced survival. 65

| Diet and the gut microbiota
There is a strong evolutionary relationship between the gut microbiota and diet. Certain members genomically encode enzymes such as glycoside hydrolases, which allow poly-and/or oligosaccharide carbohydrates to be metabolised. 68 Some "generalist" microbes, such as members of the Bacteroides phyla (eg, Bacteroides thetaiotaomicron), degrade a wide array of carbohydrates, while other "specialist" gut microbes (eg, Rosaburia intestinalis) degrade specific oligosaccharides. 69 Ingestion of dietary fibre elicits a dynamic response from communities of these metabolising microbes through extensive primary and secondary degradation. Here, primary degraders (eg, B. thetaiotaomicron) convert polysaccharides into oligosaccharides and secondary metabolites (eg, SCFAs), which can then be utilised by secondary degraders (eg, Eubacterium rectale) to further enhance nutrient bioavailability (eg, breakdown to monosaccharides) and support community colonisation. 70 This so-called "cross-feeding" is a determinant of gut population dynamics, as increased metabolism often affords selective advantages, which increase microbe abundance and subsequent digestion efficiency. 69 Previous work in CRC has indicated that microbial-derived SCFAs, particularly butyrate, have anticancer effects (demonstrated in cancer cell cultures 71,72 and animal models 73 ); however, clinical supplementation studies have proved difficult due to issues with bioavailability and toxicity. 74 Therefore, a more nuanced microbiota and defined diet approach may represent a more realistic avenue to improve cancer outcomes. This highlights the interlinking relationship between microbiota composition and metabolite production, with both factors requiring consideration if we are to realistically improve cancer outcomes.
More recently, it is now appreciated that specific components from fruit and vegetables (eg, polyphenols) may also be processed and influence the gut microbiota by acting as prebiotics (defined as dietary substrates, which can be utilised by host microorganisms to confer a health benefit). 75 Dietary polyphenols have well-documented effects on the host, which has been reviewed elsewhere. 28 Studies have shown that increased polyphenol intake is associated with higher levels of beneficial bacteria (such as Bifidobacterium and Lactobacillus) and SCFAs in humans, while also decreasing levels of bacteria that have been associated with disease, so-called pathobionts. 29

| Mechanistic links between diet and BrCa
The majority of in vivo mechanistic knowledge is confined to tumours One such consideration relates to the differential risks between BrCa subtypes (see Table 1), including treatment-resistant (eg, TNBC) tumours. To date this fundamental BrCa factor has been somewhat ignored, therefore future microbiota profiling studies of BrCa should consider clear stratification of patients by histological and molecular subtypes, which may allow comparison between groups and associated pathological outcomes. Age will also need to be carefully considered in these groups as the microbiota diversity does decrease in elderly populations. 91 Previous studies have indicated that healthy individuals living at home have a more "robust" microbiota compared to those living in long-term residential care (which may be linked to diet). There are also alterations in microbiome composition associated with frailty indices and inflammatory status. 92 This is obviously important for BrCa as, the older a woman is, the more likely she is to get BrCa; rates are highest in women over 70. 10  An important caveat to any resulting findings, however, is that mice of the same genetic background housed at different animal facilities will likely have different microbiota profiles, which may influence experimental cancer outcomes. 98 Overall, more comprehensive preclinical findings will better inform and facilitate design of translational human studies looking at the impact of defined dietary components on the gut microbiota and BrCa outcomes, as well as profiling responders vs nonresponders to treatments in the context of diet.