The microbiome of pseudomyxoma peritonei: a scoping review

Rationale

Pseudomyxoma peritonei (PMP) is a rare phenomenon characterised by the progressive accumulation of mucin within the peritoneum. Although PMP has been associated with mucinous tumours of different origins [1], 2], the majority of cases results from a ruptured appendiceal mucinous neoplasm [3]. The tumour cells that are released into the abdominal cavity continue to proliferate and exude mucin, resulting in mucinous ascites and tissue deposits, which can develop insidiously over time. As the volume of mucin increases, so does the intra-abdominal pressure, leading to the compression of adjacent structures and eventually bowel obstruction [4].

Despite only affecting two to four per million annually [5], 6], the prognosis can be poor, depending largely on the histological subtype [7], 8]. Low-grade PMP [9], previously known as disseminated peritoneal adenomucinosis, has five-year survival rates of 84 %. On the other hand, high-grade PMP, previously known as peritoneal mucinous carcinomatosis, is associated with a 10–40 % five-year survival rate [10], 11].

Cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) is the mainstay treatment for PMP. However, the rate of surgical complications and associated morbidity is as high as 24–50 % and the recurrence rate is 25 % [7], 12], 13]. Given the paucity of treatment options for patients with PMP, new insights into the pathophysiology of the disease are essential to explore alternative potential therapeutic avenues.

The gut microbiome (the collective microbiota found within the gastrointestinal tract) has been linked to chronic diseases such as type 2 diabetes, hypertension and inflammatory bowel disease [14], and there is a growing interest in the role of pathogens in carcinogenesis and tumour progression [15], [16], [17]. Several pathogens including Fusobacterium nucleatum [18], Peptostreptococcus anaerobius [19], Bacteroides fragilis [20] and Escherichia coli [21], for example, are thought to contribute to carcinogenesis in colorectal cancer through a variety of molecular mechanisms [22]. Conversely, a growing body of evidence supports the protective effects of bacteria, particularly of healthy commensal gut microbiota, against inflammation and cancer. Short-chain fatty acids produced by microbial fermentation of dietary fibre not only have anti-inflammatory properties but also facilitate apoptosis of colonic malignant cells by inhibiting histone deacetylases [23]. The commensal microbiome of the digestive system also preserves the epithelial integrity and prevents pathogen overgrowth, reducing inflammation and colitis-associated cancers [24].

In PMP, one theory is that bacteria translocate from the intestinal lumen to the peritoneum at the time of appendiceal perforation, but few studies have examined the microorganisms that colonise PMP tissue.

Objectives

A scoping review was carried out to systematically collect and synthesise the research done in this area and map it to the framework detailed above, to emphasise the importance of this topic and to highlight the current gaps in knowledge.

The aims of this scoping review were to determine what micro-organisms have been identified in PMP samples and explore what role, if any, they may play in disease outcomes.

Eligibility criteria

Our search aimed to identify peer-reviewed journal papers that examined microorganisms collected from human PMP tissue or mucin samples. PMP from all reported tumour origins were included to maximise the search results. All study types were included given the highly specialised area of research, to allow a comprehensive review of all existing evidence. Papers outside of the scope of the research question were excluded. Review articles that did not provide novel data were also excluded.

Information sources

PubMed, EMBASE and Scopus databases were used to identify potentially relevant studies.

Search

An initial search was performed to identify common keywords (e.g., pseudomyxoma peritonei, peritoneal cancer, microbiome, microbiota, Pseudomonas, H. pylori, cytoreductive surgery, mucin). Keywords outside the scope of our research question (e.g., cytoreductive surgery, mucin) were excluded to ensure focused results. Other keywords such as ‘PMP’ ‘peritoneal cancer’, ‘appendiceal cancer’, ‘peritoneal neoplasms’, ‘Pseudomonas’ and ‘H. pylori’ were also excluded from the search as they lacked specificity and were likely to generate a large volume of insignificant results. Synonyms, truncations and MeSH terms were used in the PubMed search (Figure 1).

Figure 1:

Search strategies used in PubMed, Embase and Scopus.

Selection of sources of evidence

All search results were exported to Rayyan [26]. Duplicate results were removed. Titles and abstracts were assessed to determine relevance, according to the eligibility criteria described above. All studies that both histologically confirmed a diagnosis of PMP and performed microbiological analysis of the tumour tissue or mucin were selected for further screening to confirm relevance. Studies investigating the aetiology of infection after CRS/HIPEC in patients with PMP, for example, were excluded. Additionally, all animal studies were excluded. Two reviewers performed the screening process sequentially to identify relevant publications and reduce selection bias. Disagreements were discussed and resolved.

Data charting process

Two reviewers independently extracted relevant data from all the selected manuscripts. The data was manually extracted and subsequently charted in an excel spreadsheet containing headings for each selected variable. The results tables generated by each reviewer were compared to identify and resolve any disagreements, and then merged.

Data items

The variables recorded included: manuscript title; authors; year of publication; study population (i.e. primary site of PMP, histological grading); type of specimen collected (i.e. tissue vs. mucin); methodology (i.e. detection methods, taxonomic rank); main results; and limitations of the study.

Synthesis of results

A two-stage analysis was performed. The primary analysis involved the identification and documentation of all microorganisms identified from PMP tissue and mucin samples, at the various taxonomic ranks. Subsequently, the literature was further examined to determine the possible clinical significance of the identified microorganisms.

Selection of sources of evidence

A total of 107 results were generated from the searches. PubMed generated 48 results, Embase generated 43 and Scopus generated 16. Duplicates were identified and removed. The remaining 74 unique studies were assessed to determine relevance. After excluding all studies that did not satisfy our eligibility criteria, nine were selected for review (Figure 2).

Figure 2:

Database screening process and study selection.

Characteristics of sources of evidence

Eight studies isolated, identified, and characterised microorganisms from PMP specimens collected intra-operatively (Table 1). There was significant heterogeneity in the type of study and methodology used. The relevant studies included two case reports, an announcement, two retrospective laboratory studies performing analysis on archived specimens, and three prospective laboratory studies. The microbial detection methods also varied amongst the different studies, including traditional culturing, in-situ hybridisation (ISH), V6 polymerase chain reaction (PCR) and whole genome sequencing.

Villarejo-Campos et al. and García-Olmo et al. conducted gene sequencing on fresh mucin samples prospectively collected from patients with PMP to identify and classify organisms at different taxonomic levels [27], 28]. Gilbreath et al. analysed archived formalin-fixed paraffin-embedded (FFPE) tissue specimens and mucin samples [29]. Several case studies identified previously unrelated or novel microorganisms from PMP samples. Gatalica et al. reported PMP associated with HPV-positive cervical cancer [30], whilst Lo et al. isolated a novel bacterial species, later re-classified by Lawson et al. [31], 32]. Another two studies examined the density of typed and non-culturable bacteria (TNCB), rather than identifying or classifying the organisms. Semino-Mora et al. compared the bacterial density of PMP samples from patients with no pre-operative antibiotic exposure to the density of non-neoplastic non-perforated appendix samples [33]. In a later study, Semino-Mora et al. prospectively examined the correlation between pre-operative antibiotics and the bacterial density of PMP samples, focussing on H. pylori [34].

In addition to the retrospective microbial analysis of PMP samples Gilbreath et al. also conducted a prospective case-control study comparing the survival of PMP patients that did and did not receive pre-operative antibiotics. The outcomes of these groups were updated five years later (Table 2) [35]. Whilst these two studies didn’t directly examine the PMP microbiome, they were developed to provide clinical context and guidance regarding the possible implications of their earlier results, and were therefore included for additional breadth.

Synthesis of results

Summary of evidence

Several organisms have been identified from PMP tissue or mucin samples (Table 3). Only two studies, Villarejo-Campos et al. and García-Olmo et al. used fresh tissue. A range of microbial detection and identification methods were used, including genomic sequencing, ISH and culture. At the phylum level, proteobacteria, was consistently detected in greatest relative abundance (73–82.86 %) in both PMP tumour tissue and both cellular and acellular mucin (Table 4). Other less abundant phyla included actinobacteriota, bacteroidetes/bacteroidota, firmicutes, verrucomicrobia and acidobacteria. Three studies detected Pseudomonas at the genus level, which alongside Escherichia, Shigella, Methylobacterium and Variovorax represented the most prevalent Proteobacteria.

Interestingly, in the healthy human gut a higher proportion of firmicutes, bacteroidetes and actinobacteria is seen and proteobacteria represent only a small minority of the total metagenomic species [36], [37], [38]. Whereas in inflammatory bowel disease, a reduction in firmicutes and rise in proteobacteria has been reported [39], 40], which is more in-keeping with the microbiome of PMP specimens. The difference in oxygen levels may contribute to this shift in phylogenetic profile [41], 42].

Some of the genera identified in PMP specimens, like Stenotrophomonas, Escherichia, Shigella, Moraxella, Pseudomonas, Streptococcus and Helicobacter, include species that exist on the spectrum of microbe-host relationships, from commensal organisms to pathogens. The relevance of these organisms in the progression or outcome of PMP is uncertain. Others, such as Saccharopolyspora, Sphingobium, Methylotenera, Variovorax and Acidovorax, have seldom been associated with human disease are more commonly seen in environments like soil, marine sediments, and plants, thus are more likely to represent contaminants.

The histological grade of PMP, on the other hand, is a strong prognostic factor, with high-grade PMP conferring worse survival than low-grade PMP [43]. High-grade specimens showed significantly higher bacterial density, H. pylori inclusive, than low-grade specimens and non-neoplastic non-perforated appendix specimens [33], 34]. However, the bacterial densities from the latter two cohorts were not significantly different. It is unclear whether the increased bacterial density is contributing towards the increased cytological atypia and mitotic activity seen in high-grade PMP or whether they simply thrive better in that environment. Therefore, histological grade may be a major confounding factor in the effect of bacterial density on disease outcome.

Two interesting relationships which may explain this association have been highlighted. One is between bacterial density and MUC2 expression and the other between bacterial density and β-catenin expression.

The MUC genes transcribe the glycoprotein that form mucin. MUC2 is expressed by healthy goblet cells in the intestine but is overexpressed in gastric and colorectal cancers [44], 45]. Semino-Mora et al. [33] observed that both bacterial density and MUC2 expression were significantly higher in high-grade PMP than in low-grade PMP. Previous work had evidenced the transcriptional activation of mucin by Pseudomonas aeruginosa lipopolysaccharide in cystic fibrosis [46] and PMP [47], suggesting higher bacterial densities, particularly of species such as P. aeruginosa, may be contributing to increased MUC2 expression and therefore volume of peritoneal mucin. While there may be a link between certain bacteria and MUC2 overexpression in addition to an independent correlation between the histological grade of PMP and total MUC2 expression, it is difficult to draw a link between all three factors. This is due to evidence suggesting that the increased MUC2 expression in high-grade PMP is due to a greater number of mucinous tumour cells rather than gene overexpression at a cellular level [47]. Perhaps then, the bacterial density could be attributed to the cellularity of the mucin, which is greater in high-grade PMP than low-grade PMP, resulting a richer environment for growth, although there is currently little evidence to support this theory.

It has also been suggested that bacteria may modulate β-catenin signalling pathways [48], 49], which are known to regulate apoptosis as well as the transcription of oncogenes. Semino-Mora et al. observed that β-catenin expression was higher in high-grade PMP specimens compared to the low-grade PMP specimens, although the difference was not statistically significant [34]. The authors also demonstrated that pre-operative antibiotics were associated with significantly lower β-catenin expression in high-grade PMP specimens, but not in low-grade specimens. Whilst they claimed that the antibiotics were responsible for this decrease, they didn’t perform a longitudinal analysis before and after antibiotic treatment, nor did they account for other confounders. Although the correlation between bacterial density, β-catenin and histological grade was observed in this small sample size, a causative effect has not been proven.

Even though the data shows an association between the density of TNCB and H. pylori and the degree of mucin expression, particularly in higher histological grades of PMP, the evidence is insufficient to determine if the bacteria play a pathological role in this.

The impact of pre-operative antibiotics survival on PMP patients was also investigated. Initially, Semino-mora et al. demonstrated that antimicrobial treatment was associated with reduced H. pylori and TNCB densities [34]. Subsequently, they followed-up the patient from the antibiotic cohort over a ten-year period and reported their outcomes. The relevance of these results is severely limited by our inability to compare them to the non-antibiotic cohort, as well as the small sample size. Furthermore, the authors did not record or acknowledge confounding factors which may have influence survival, nor did they perform any statistical analysis, particularly multivariate, which would have increased the interpretability of their results. More robust survival studies are required to consolidate our understanding of the potential role of antibiotics in the management of PMP.

Limitations of existing evidence

A significant limitation of all studies included was the small cohorts of PMP patients included, which is unsurprising given the rarity of the condition. Additionally, there was considerable variation in the type of specimen analysed (i.e. fresh tissue, FFPE, mucin), the molecular detection methods, and PMP tumour origin. This hindered our ability to compare the results and draw meaningful conclusion.

Furthermore, there was often no comment on confounding factors, such as unintentional pre-operative antibiotic exposure or evidence of contamination during specimen collection. Nor was there rigorous use of controls. Experts have raised concerns about the impact of contamination, not only during clinical collection procedures but also during the extraction and sequencing process, and cautioned about the risk of that confounding the results [50]. In fact, contamination could be responsible for the detection of bacteria commonly associated with nosocomial infections, such as Sphingomonas and Bosea, in several of the studies discussed in this review [27], 29].

Better understanding of the limitations of detection methods is also required to avoid drawing erroneous conclusions. While detection methods such as genome sequencing has greatly enhanced the ability to detect microorganisms, they cannot discern viability cross-sectionally [51], 52]. Traditionally, culture and isolation has been used to identify viable organisms, although more recent strategies like viability PCR have been developed [53], 54]. Only two studies attempted to do this, one through culture [29], and another using a polyclonal anti-H. pylori antibody to discern the viability of the identified H. pylori [34]. Without evidence of organism viability, the ability to determine their role in this disease is significantly limited [29].

Suggested direction for future study

Multicentre studies, involving specialist centres, may be able to overcome the issue of small sample size by pooling together their patients to compile larger cohorts. Clear description of the tumour origin and histology, type of specimen collected, and possible confounding factors such as antibiotic exposure or evidence of contamination during the collection process, is crucial for greater comparability of results.

Current studies lack species-level resolution, which can be difficult to achieve when intratumoral microbial biomass is low. Optimising methodology will be key to pinpoint potentially oncogenic organisms or clusters. This can be achieved through the use of controls to assess contaminant DNA and microbial detection methods that provide higher resolution profiling [55], [56], [57].

Ultraclean DNA extraction kit and concurrent sequencing of negative control samples, for example, may reduce contamination by the ‘kitome’ [58], [59], [60]. The use of paraffin controls in addition to FFPE tissues would also enable authors to discount bacterial communities originating from the paraffin itself [61]. Furthermore, comparing the tumour bacterial profile with the patient’s own gut microbiome would lend valuable insight, by highlighting differences in bacterial profiles or conversely providing further evidence of surgical contamination.

Using detection methods that are capable of discerning viable organisms from non-viable ones, such as viability PCR and culture, would clarify the role of microorganism in carcinogenesis or disease progression. However, this can only be confirmed with evidence of the exact mechanism of action, given the wide spectrum of microbe-host interactions. Examining PMP tumour cells or cellular mucin in vitro alongside specific microorganism in a stromal environment, such as with MUC2-secreting human intestinal cell lines, could provide a new useful research model.

Review limitations

Only three databases were used as part of our search strategy. Although a large number of results were generated, only a few of these were relevant. Given the specificity of our search question, we could have performed forward and backward reference searching to maximise relevant results. Our review was also limited by the small number of screeners and reviewers (n=2), resulting in higher risk of selection bias. There was some missing data identified during the extraction process, including tumour origin and tissue type, which affected the interpretation and comparability of results.

Conclusions

This review aimed to identify what organisms have been detected in PMP specimens, and correlate that with evidence of pathogenicity and effect on disease outcome.

Several bacteria were identified from tumour and mucin specimens, at different taxonomic ranks. The relative proportion of different phyla more closely resembles the microbiome seen in inflammatory bowel disease than in the healthy gut, with a higher relative abundance of Proteobacteria compared to Firmicutes and Bacteroidetes. Only Pseudomonas plecoglossicida and a novel bacterium, P. muciniphila gen. nov., sp. nov., were identified at the species level. Most of the organisms described were identified through molecular sequencing, so their viability is questionable. However, bacteria from the genera Bosea, Corynebacterium, Propionibacterium and Niastella were successfully cultured from PMP samples. Whilst there may be a tenable association between bacterial density and histological grade, the nature of this relationship is unclear.

There is insufficient evidence to establish a causal link between the microbial alterations observed and any tumour behaviour or clinical outcomes. No definitive conclusions can be made regarding the microbiota’s role in PMP progression.

Significant methodological challenges remain in this field of study. Ongoing research, using considered and cautious methodology, has the potential to catapult our understanding of this malignancy and change the landscape of its clinical management.