Role of the microbiome in occurrence, development and treatment of pancreatic cancer

The oral cavity is a large reservoir of microbes including more than 700 types of bacteria, viruses and fungi [16, 32]. When conditions change, the commensal microbiomes can become pathogenic and lead diseases including PC [16]. Periodontal disease, an inflammation caused by oral microbes, has been regarded as a risk factor for PC. For example, Maisonneuve et al. conducted a meta-analysis of eight studies and suggested a significant link between periodontal disease and increased risk of PC [33]. The relative risks for PC in individuals with periodontitis and edentulism were 1.74 (95% confidence interval, CI 1.41–2.15) and 1.54 (95%CI 1.16–2.05), respectively [33]. Farrell et al. conducted an analysis using human oral microbiological microarrays to study variation in salivary microbiomes among 10 patients with resectable PC and 10 matched healthy controls. They identified 410 bacterial taxa [11], including 16 (3.9%) organisms such as Neisseria elongata and Streptococcus mitis whose frequencies differed significantly between the two groups. This study revealed significant changes in the microbial composition of PC patients and healthy controls [11]. In addition, Porphyromonas gingivalis, a commonly identified bacterium in patients with periodontal disease, is thought to increase risk of developing PC [17, 34]. To understand the connection between oral microbiomes and PC, Michaud et al. measured antibodies against oral bacteria in pre-diagnostic blood samples from 405 PC patients and 416 matched controls nested in the European Prospective Investigation into Cancer and Nutrition (EPIC) study [35]. The results showed that individuals with high levels of antibodies against P. gingivalis ATCC 53978 were at two-fold increased risk of PC compared with individuals with lower levels of these antibodies [35]. In addition, the authors found that individuals with consistently high levels of antibodies to common oral bacteria were at 45% lower risk for PC compared to those with lower antibody levels [35]. Similarly, other studies using 16S rRNA sequencing suggested that the presence of Haemophilus, Porphyromonas, Leptotrichia and Fusobacterium species in the oral cavity was also associated with increased risk of PC in humans [9, 18, 36].

As one of star oral microbiomes associated with PC, P. gingivalis has been extensively studied. Although several studies have shown that P. gingivalis is a new risk factor for PC, further researches are needed to explore the specific mechanisms leading to PC. Some researchers hypothesized that one potential mechanism might be that a peptidyl-arginine deiminase enzyme secreted by P. gingivalis leads to p53 and K-ras mutations following degradation of arginine [19], while others focus on the effect of oral bacteria on the systemic immune response, including IL-1β, IFNγ, and TNF [2, 37]. How, then, can the oral microbiome affect the pancreas? Gaiser et al. found higher loads of oral bacterial DNA in the cyst fluid of intraductal papillary mucinous neoplasms (a condition which can progress to PC), providing support for the relationship between oral and pancreatic microbes [20]. By feeding wild-type mice fluorescently-labeled Enterococcus faecalis or Escherichia coli, Pushalkar et al. found these bacteria accrued in the pancreas of mice, demonstrating that microbes can migrate to the pancreas and directly affect the pancreatic microenvironment [4]. However, no clear evidence has indicated which channels microbes use to reach the pancreas. In fact, although the pancreas belongs to distal organ of the digestive tract, it is still connected to the digestive tract via duodenum. Therefore, oral microbes are likely to enter the pancreas through the digestive tract. In addition, it is well known that oral microorganisms easily enter the blood, leading to bacteremia. Therefore, oral microorganisms may also enter the pancreas through blood circulation. In summary, the exact channel of oral microbes into the pancreas requires further researches to determine.

Helicobacter pylori, a well-known bacterium that colonizes the human stomach, has been the subject of increasing attention over the last 30 years [38]. Many previous studies have attempted to correlate the presence of H. pylori with PC using serologic and culture-based methods. However, the vast majority of commensal microbes cannot be cultured, affecting the objective investigation of their role in pancreatic diseases [21, 39‐42]. New techniques, such as next-generation sequencing and metagenomics, have provided a more representative assessment of the microbial community in health and disease and the dynamic interactions between microbiomes and their human hosts [22]. These techniques may help in understanding the association between H. pylori and PC. To determine whether H. pylori infection was associated with PC, Raderer et al. used enzyme-linked immunosorbent assays to analyze IgG antibodies against H. pylori in blood samples from 92 patients with PC, 30 patients with gastric cancer, 35 patients with colorectal cancer, and 27 healthy volunteers [39]. The results showed that 65% of patients with PC and 69% of patients with gastric cancer were seropositive, compared with only 45% of the other individuals [39]. Mei et al. found that H. pylori could be detected in the duodenum at higher frequencies in PC patients than in healthy controls [23]. Similarly, other studies used meta-analysis to confirm that H. pylori was associated with increased risk of PC in humans [24‐26, 43, 44]. The probable mechanism of microbiome transmission involves changes in the intestinal microbial environment, which can alter the composition of the intestinal microbiome, increase intestinal permeability and permit microbial access to the bloodstream and distant organs [27, 45]. H. pylori may promote the development of PC by causing chronic mucosal inflammation as well as changes in cell proliferation and differentiation [28].

However, several other studies concluded that H. pylori was not associated with PC [25, 29, 46, 47]. One of the potential explanations for this discrepancy is that H. pylori has several variants, of which cytotoxin-associated gene A (Cag-A) positive strains have been the best studied. Cag-A has been linked to multiple diseases such as gastric inflammation and ulceration, gastric cancer and PC [48, 49]. However, it remains controversial whether Cag-A-positive or Cag-A-negative strains are associated with PC [50, 51]. An effect modification by ABO blood type was reported in a large case-control study: the association between PC and Cag-A-negative H. pylori was evident only in individuals with non-O blood types [41]. This finding may be explained by differences in the terminal antigens of gastrointestinal mucins in individuals with non-O blood types, which affects binding by H. pylori [2].

No single clear mechanism has been widely accepted to explain associations between H. pylori and PC. One hypothesis suggests that H. pylori infection leads to hyperchlorhydria and enhanced release of secretin, promoting pancreatic hyperplasia [52]. Conversely, another hypothesis suggested that H. pylori infection led to atrophic gastritis and hypochlorhydria, resulting in bacterial overgrowth and overproduction of N-nitrosamines [53]. In summary, the role of H. pylori in PC remains unclear. Further studies are required to consider other potential confounding risk factors and conclusively explore whether H. pylori is truly associated with the occurrence and development of PC.

Traditionally, the pancreas has not been considered to have its own microbiome. Li et al. performed bacterial 16S rRNA gene-specific PCR to analyze the microbial constituents in the pancreatic cyst fluids, where Bacteroides, Escherichia/Shigella, and Acidaminococcus were predominant [30]. They reflected the local microbiota in the pancreas, and prove that pancreatic cyst fluid is a very important sample for microbial identification. Now, in addition to P. gingivalis and H. pylori, other microbes have been identified in PC tissues. Using 16S rRNA gene sequencing, Pushalkar et al. found high proportions of Proteobacteria (45%), Bacteroidetes (31%), and Firmicutes (22%) species in PC tissues [4]. Interestingly, they concluded that the microbiome proportions in PC tissue were quite different from those of normal pancreatic tissue. Some recent research also showed similar results [31, 54]. Thus, the pancreas is not sterile and has its own microbial environment which may affect the occurrence and development of PC. More complex mechanisms involving a large number of factors may alter the pancreatic microbiome. These alterations may occur via natural and non-natural channels.

Some studies have linked fungi and viruses to the development of PC. For example, a prospective cohort study conducted in Sweden showed that Candida infection in the oral cavity was associated with development of PC [55]. Similarly, another population-based cohort study in Taiwan showed that risk of PC was significantly higher among Candida-infected individuals [56]. However, the relationships between fungal infections and PC require further study. The roles of hepatitis viruses in the development of hepatocellular carcinoma are relatively clear. However, some evidence suggested that hepatitis viruses may also be associated with PC. Katakura et al. found increased serum levels of pancreatic enzymes in viral hepatitis patients [57], while Jin et al. identified a link between hepatitis B virus and chronic pancreatitis [58]. These studies link chronic hepatitis, chronic pancreatitis and PC and demonstrate that a potential role of viruses in PC cannot be ignored.

Microbial infections often lead to inflammation [59]. Sometimes inflammation is a protective response to factors such as pro-inflammatory mediators, environmental toxins, or chronic infection [60‐62]. However, inflammation is also a risk factor for development of many cancers. For example, Dejea et al. demonstrated that patients with familial adenomatous polyposis had higher frequencies of E. coli and Bacteroides fragilis cells in the colonic mucosa compared with healthy individuals. Moreover, tumor-prone mice had higher interleukin-17 (IL-17) and IL-23 levels and developed tumors faster upon stimulation by microbes [63, 64]. Inflammation of the pancreas also increases the incidence of PC. Patients with hereditary autoimmune pancreatitis are estimated to carry a lifetime risk of 40% of developing PC and patients with chronic pancreatitis, a recognized risk factor for PC, have a 13-fold higher risk of PC than other individuals [65, 66]. Multiple cohort studies have shown that acute pancreatitis is also associated with the development and progression of PC [67‐69]. However, acute pancreatitis is not a direct factor, but the chronic prolongation of inflammation leads to the occurrence and development of PC [68]. Although there are no identified pathogens, this type of chronic inflammation can also be caused by microbial infections [70]. Microbial-induced inflammation leads to tumorigenesis through activation of tumor-related inflammatory signaling pathways.

The occurrence and development of tumors are closely related to the immune system, and the impact of microbiomes on the immune system is a very hot topic. Recently, Riquelme et al. used human-into-mice fecal microbiota transplantation experiments from short-term survival, long-term survival, or control donors, and they found modulating the tumor microbiome can affect tumor growth as well as tumor immune infiltration [15]. Different studies have highlighted the various roles of microbiomes in the immune system, including effects on immune maturation and immune suppression.

The gut microbiomes and the immune system can affect one another in the gut lamina propria, and similar effects have recently been described at extraintestinal sites [94, 95]. Round et al. found that the immune systems of germ-free mice were deficient, with hypoplastic lymphoid organs and impaired immune cells. However, the immune systems of these mice matured after gut microbiome transplantation from specific-pathogen-free mice [96]. Mechanistically, microbiomes can act as antigens and activate the immune system. Damage to the intestinal mucosa allows microbes to enter the gut lamina propria and migrate to distant lymphoid organs, resulting in activation of the immune system. However, in the absence of intestinal microflora, the immune system cannot be activated [94, 97]. Responses to gut microbiomes mediated by IL22⁺ innate lymphoid cells, Th17 cells and regulatory T cells occurred in mice deficient in adaptive immunity, indicating that the gut microbiome can promote innate immunity [98]. Several studies showed that specific microbes, such B. fragilis and Bifidobacterium species, may be important factors for maturation of the immune system [99‐101]. In lung cancer, bladder cancer, kidney cancer and melanoma, similar results have been reported [102‐104].

However, other studies showed the opposite result, finding that immune systems became activated during antibiotic treatment in models of liver cancer, colon cancer and melanoma [105‐107]. Immune cells are essential in the PC microenvironment, where they promote tumorigenesis along with related inflammatory factors, and thus immune cell infiltration has further effects on disease progression [95, 108]. Several studies have focused on immune cell infiltration in PC. Infiltration by different leukocyte subsets can have different effects on tumorigenesis and progression, either promoting tumor growth or inhibiting tumor progression [109]. Some studies have shown that Th1-polarized CD4⁺ and CD8⁺ T cells inhibit pancreatic tumor growth in a mouse model and are associated with prolonged survival in human PC [94, 106, 110]. In contrast, antigen-specific Th2-polarized CD4⁺ T cells can promote progression of PC in mice [106, 111], and are associated with shorter survival in human PC [112] (Fig. 2). Another study found that FOXP3⁺ regulatory T cells can promote immune escape in PC [113]. However, differentiation of T cells may be regulated by microbiome composition. For example, the number of pancreas-infiltrating CD45 immune cells was reduced in antibiotic-treated NOD/SCID (Non-obese diabetic/severe combined immunodeficient) mice [107]. In KC mice (K-ras^LSL.G12D; Pdx1-Cre mice) and KPC mice (K-ras^LSL.G12D; p53R^172H/+; Pdx1-Cre mice), myeloid-derived suppressor cell infiltration was reduced during antibiotic treatment. Moreover, Th1 polarization of CD4⁺ T cells and cytotoxicity of CD8⁺ T cells were enhanced as shown by high T-BET, tumor necrosis factor (TNF)-α and interferon-γ expression [4] (Fig. 2).

Microbe-mediated immune suppression was associated with pattern recognition receptors, and inhibition of these receptors slowed tumor development [114]. As mentioned above, following interaction with LPS, TLR4 can activate the downstream NK-κB/MAPK pathway and lead to development of PC (Fig. 2). Several TLRs (TLR2, TLR4, TLR5 and TLR7) are associated with suppression of innate and adaptive immunity to promote development of PC [4, 111, 115]. Mechanistically, activation of TLRs results in inhibition of interactions between macrophages and lymphocytes, which are abrogated in the absence of TLRs signaling [4].

These opposite results may indicate that different microbiomes play different roles in immunity or in different tumor models. It is clear that further studies are needed to target the PC-associated microbiome to enhance immunotherapy. Moreover, the composition of the gut microbiome can divide patients into responders and nonresponders for immunotherapy, which demonstrates the significance of exploring specific microbial features as diagnostic markers in PC.

Microbiomes play an important role in the body, especially in the metabolism of sugars, amino acids and fats as well as synthesis of vitamins or other nutrients [116, 117]. Microbiomes cause changes in human metabolism, contributing to a variety of metabolic diseases such as obesity and diabetes. Obesity and diabetes are also important factors in the development of PC. Papamichael et al. reported that colonization by H. pylori, which is a potential independent risk factor for PC, may be associated with obesity and diabetes [118]. Therefore, microbiomes also affect the development of PC via changes in metabolism.

Obesity is a risk factor for PC in both men and women [119‐122]. Obesity affects the progression of pancreatic tumors by modifying the interactions between adipocytokines [123, 124], adiponectin [125‐127], deoxycholic acid [128], and many other molecules. Furthermore, the gut microbiome is also believed to play an important role in connecting obesity and PC. Donohoe et al. found that the body mass indices of lean mice could be increased by transplantation of gut microbiomes from obese animals, which were able to digest more nutrients [129]. Therefore, we speculate that microbiomes may participate in the occurrence and development of cancer through some metabolic mechanisms. In addition to changes in microbial diversity [130], some microbial metabolites may be associated with the development of obesity. For example, short-chain fatty acids (SCFAs), which are enriched in obese individual, can activate the MAPK signaling pathway through G-protein-coupled receptors and lead to cancer cell proliferation [86]. In addition, obesity can promote release of LPS from the gut microbiome and therefore lead to endotoxemia [131]. Ren et al. demonstrated that PC patients had more LPS-producing bacteria than healthy controls, supporting a potential relationship between endotoxemia and PC [8]. In fact, obesity is also a type of inflammatory state. As mentioned above, microbiomes can influence the development and progression of PC through different mechanisms, and the NF-κB pathway is a common pathway in both inflammation and cancer. Pagliari et al. suggested that obesity is associated with the release of various pro-inflammatory cytokines, such as IL-6 and TNF, which activate the NF-κB pathway and regulate downstream cancer-associated signals [132].

Diabetes is also a risk factor for PC. In diabetic patients, the ratio of Firmicutes to Bacteroidetes species is relatively high, affecting metabolism of carbohydrates and production of SCFAs [133]. Perry et al. demonstrated that increased levels of acetate in the blood led to insulin resistance [134], while Devaraj et al. showed that decreased levels of butyrate in the intestine promoted low-level inflammation and caused insulin resistance [135]. Decreased levels of butyrate also impaired epithelial tight junctions in the intestinal mucosa and promoted entry of bacterial endotoxins into the blood [136]. This mechanism could link diabetes and PC through endotoxemia. This effect of butyrate level was similar in individuals of different races, and metformin was able to adjust the level of butyrate effectively [137]. However, more experimental evidence is needed to confirm connections between the microbiome, diabetes and PC.

Currently, the relationship between metabolic diseases and PC has been extensively studied. However, the relationships between microbiomes and metabolic diseases are not currently sufficient to draw firm conclusions. Promisingly, we still emphasize the importance of metabolic disorders associated with microbial diversity and microbial metabolites, which are worth further exploration.

Despite many studies suggesting an association between oral microbial dysbiosis and PC, no convincing evidence has indicated whether oral microbial dysbiosis is causally related to or merely an effect of early PC [138]. However, further studies of bacterial markers of periodontal disease such as P. gingivalis and changes in microbial diversity may suggest non-invasive screening biomarkers for PC. Recent developments suggest that salivary RNA markers can be used to identify oral bacteria by high-throughput sequencing of bacterial small subunit ribosomal RNA (16S rRNA) genes [7, 139, 140]. Therefore, saliva testing, a non-invasive test of oral biomarkers, may become a convenient strategy to screen for PC in the future. However, the existing results must be confirmed in larger multicenter prospective studies [140].

In addition, other body fluids may contain diagnostic markers of the microbiome relevant to other cancers. For example, feces can be used as a biomarker for colorectal cancer while urine may contain biomarkers of bladder cancer [141, 142]. Therefore, other body fluids such as feces, blood and pancreatic juice may also provide diagnostic markers for PC. All of these biomarkers require more study to demonstrate their potential value.

In Fusobacterium-associated colorectal cancer, metronidazole treatment could reduce not only the Fusobacterium load, but also cancer cell proliferation and patient-derived xenograft tumor growth [143]. Similarly, if there are a variety of microbes located in or associated with PC, these microbes could also become future therapeutic targets for PC. In this part, we will discuss the role of microbiomes in gemcitabine therapy, PD-1 targeted therapy, and antibiotics therapy, aiming to emphasize that some microbiomes can be seen as therapeutic targets in PC.