Changes in the gut bacterial communities in colon cancer surgery patients: an observational study

Surgical site infections (SSI) are a leading cause of healthcare-associated infections [1]. Among surgical procedures, colorectal surgery is associated with the highest incidence of SSI, despite widespread implementation of evidence-based preventive practices [2]. Indeed, in Switzerland, the average incidence of SSI in colon surgery is 14.4% [3].

Patient-level risk factors for SSI in colorectal surgery include age [4, 5], obesity [5‐8], and diabetes [4]. Intraoperative risk factors include emergency surgery [9, 10], contaminated or dirty surgery [5, 10], duration of surgery, and creation of an ostomy [7, 9]; laparoscopic surgery is a protective factor [5, 11].

In the field of colorectal surgery, there is a body of indirect evidence suggesting that the microbiome plays an important role in the postoperative outcome, notably in the incidence of SSI, including anastomotic leak [12]. The current paradigm surrounding management of the high bacterial load prior to surgery is to maximise decontamination, either by perioperative antibiotic prophylaxis alone (one of the most effective measures) or in combination with bowel cleansing (mechanical bowel preparation) and/or topical oral antibiotics [13]. However, this decontamination indiscriminately affects both beneficial and potentially pathogenic bacteria.

There is a real gap in the literature regarding adequate description of the modulation in the intestinal microbiota in patients undergoing colon surgery using state-of-the-art next-generation sequencing techniques.

Therefore, we performed a prospective observational study to evaluate changes in the gut microbiota in adult patients undergoing perioperative antibiotic prophylaxis regimens and elective colon surgery, and to assess differences between preoperative (rectal) and intraoperative (luminal) microbiome. Because all patients received perioperative antibiotic prophylaxis, we believe that the changes in the gut microbiota would be due to the impact of antibiotics.

We performed a single-centre prospective observational cohort study including adult patients who underwent elective colon surgery at Geneva University Hospitals, Switzerland, which is a regional tertiary centre.

Eligible patients were identified from the operating programme and were included if they were adults (age ≥ 18 years), and were scheduled to undergo elective colon surgery for oncological reasons. Informed consent was obtained during the preoperative surgical visit. Included patients have also been enrolled in a prospective cohort in a national surveillance system of SSI (Swissnoso) [3]. Exclusion criteria were emergency colon surgery, rectal surgery, receiving any topical oral or systemic oral/parenteral antibiotic within 30 days preceding the colon surgery, and necessity of mechanical bowel preparation. Patients who were unable or unwilling to provide informed consent or for whom 30-day follow-up was difficult/impossible (e.g., residing overseas) were also excluded.

According to institutional perioperative prophylaxis guidelines, patients receive a second-generation cephalosporin (cefuroxime) plus metronidazole within 60 min preceding the incision, except those known to have intestinal colonisation with ESBL-E who receive ertapenem, as non-targeted prophylaxis is associated with an increased risk of SSI [14].

Patients in our study were all included in our institutional Fast-track protocol [15]. Briefly, nutritional status was assessed one month prior to surgery. In case of impaired nutritional status (as testified by low serum pre-albumin levels and a Nutritional Risk Score (NRS) ≥ 3 [16], patients received protein-enriched oral supplements. No dietary restrictions were enforced before surgery. Non-diabetic patients were given a hyperglycaemic oral solution (ProvideXtra®, Fresenius Kabi, Germany) the night before (400 ml) and two hours (200 ml) prior to surgery. A low-fibre diet was introduced at postoperative day 1, and was continued for approximately two weeks. Protein-enriched oral supplements were given between postoperative days 2 and 7.

Between March 27, 2019 and September 17, 2019 28 adult patients were scheduled to undergo elective colorectal cancer surgery and were therefore considered as eligible for the study. One patient was excluded because the postoperative sample was not collected, leaving 27 patients in the final analysis. The median age was 63.6 years and 44% were female. One patient was operated for a non-resectable polyp, two were operated for neuroendocrine tumours, and the remaining 24 patients had adenocarcinoma. Table 1 summarises other patient characteristics.

Most (81.5%) patients received standard perioperative antimicrobial prophylaxis as they were not ESBL-E carriers (Table 2). Four patients (14.8%) received ertapenem due to ESBL-E carriage, although 1 patient was subsequently found to have a negative screening swab. One patient received meropenem due to a history of carriage of carbapenem resistant Enterobacterales (Escherichia coli harbouring the OXA-48 gene, phenotypically susceptible to meropenem). Two patients developed SSI: one patient who received ertapenem for ESBL E. coli carriage developed superficial SSI with Streptococcus anginosus, Staphylococcus lugdunensis, and Enterococcus faecalis, as revealed by routine culture; another patient, who received standard prophylaxis, developed organ/space SSI and the recovered intraoperative microorganisms were S. anginosus, E. faecalis, and Colinsella aerofaciens, as revealed by routine culture.

This prospective hypothesis-generating pilot study evaluated changes in the rectal microbiota of colon cancer patients requiring a surgical intervention. The results suggest that surgery, including perioperative antimicrobial prophylaxis, induces rapid changes in the gut microbiota.

There is a dearth of studies evaluating perioperative changes in the rectal microbiome of colon cancer surgery patients. A prospective study on 54 US patients showed that both preoperative and postoperative composition of the rectal microbiome, in particular higher proportions of Proteobacteria and Bacteroidetes and lower levels of Actinobacteria and Firmicutes, were associated with postoperative ileus, but not SSI [30]. These results are not directly comparable to ours for several reasons: first, Shogan et al. [12] included both colon and rectal surgery patients. Second, they included patients with diverse surgical indications: 61% were operated for non-cancer related diseases such as inflammatory bowel disease, which may be associated both with altered rectal microbiome and development of postoperative complications. This is reflected also by the fact that 14% patients experienced postoperative ileus. Third, 82% of their patients underwent mechanical bowel preparation, which is also known to alter the microbiome. In contrast, our study was a more homogenous population, with minimal disruption of the microbiome apart from surgery and perioperative antimicrobial prophylaxis.

Van Praagh et al. published a cohort study evaluating the association between the local mucosal tissue microbiome at the site of anastomosis with the risk of anastomotic leakage of 123 colorectal surgery patients in the Netherlands, most of whom (95%) were operated for cancer [31]. Anastomotic leakage was associated with increased relative abundance of Lachnospiraceae and Bacteroidaceae. Again, patients in their cohort underwent mechanical bowel preparation [32], and thus these results do not reflect the “native” microbiome.

The study by Jin et al. assessed the ability of the intraoperative tissue microbiome composition to predict postoperative ileus in colorectal cancer patients in China [33]. Their results suggest decreased relative abundance of Faecalibacterium in patients who would later develop postoperative ileus, consistent with the results by Shogan et al. [30].

Another observational study based on targeted PCR assays to evaluate changes in bacterial counts before and after colon surgery, showed decreases in beneficial obligate anaerobes and increases in pathogenic bacteria; this study did not have the potential to provide a comprehensive overview of the microbiome because it did not use next-generation sequencing-based techniques [34].

Several interventional human studies suggest that preoperative administration of probiotics or synbiotics to colorectal surgery patients may be associated with postoperative complications, including infection; these studies, however, are fraught with methodological shortcomings, and the level of uncertainty on the effectiveness of this intervention [35]. One study using 16S rRNA gene-based metataxonomics was a single-centre randomized clinical trial which showed that perioperative administration of probiotics (Bifidobacterium longum) in patients undergoing colorectal surgery increases the proportions of Actinobacteria; there were, however, no statistically significant differences between the intervention group and control group either in terms of composition of the microbiota nor in terms of clinical outcome, probably due to the small size of the study [36].

Animal models show that colon surgery and perioperative measures (perioperative antibiotic prophylaxis, fasting, mechanical bowel preparation) alter the microbiome [12]. In rats, colectomy is associated with important increases in the amount of pathogenic bacteria (Enterococcus, Escherichia) [37].

Strengths of our study include state-of-the-art next-generation sequencing and bioinformatics techniques, a well-defined homogenous study population, as well as robust statistical analyses. Ours was a prospective study, decreasing the risk of bias associated with retrospective studies.

The main limitation of this study was the sample size, which was small, as was the number of events we had initially planned to evaluate (SSI). Despite this, we present interesting and novel findings, which will need to be confirmed in larger studies. Another limitation is the short duration of follow-up of the rectal/gut microbiome; indeed, it would have been interesting to study the dynamics of recovery of the different taxa, and investigate factors associated with this recovery. Finally, we only focused on the intestinal microbiome, whereas a more comprehensive approach also including the skin and nares microbiome should be adopted since the members of these communities can cause SSI [38, 39].

Understanding changes in the gut microbiome of colon cancer surgery patients is important, yet understudied [40]. Further, well-designed prospective studies with homogenous populations should be performed, in combination with other explanatory factors in order to obtain significant insights [41]. The research agenda should also include the changes of the gut microbiome specifically associated with mechanical bowel preparation and, more importantly, non-absorbable oral antimicrobials [42].

In conclusion, our results showed rapid changes in the rectal microbiota following colon surgery. We also observed differences in microbiota structure between stoma and rectal swabs, whereas previous studies showed good similarity between rectal swab and stool microbiota [43, 44]. Our findings highlight that results obtained by using rectal swab (and possibly faecal matter) microbiota, as a proxy for colonic microbiota, should be interpreted with caution due to the differences between rectal and luminal colonic bacterial communities, notably in the relative abundance of anaerobic gram-positive cocci and gram-negative rods.

Sequencing data were submitted to the European Nucleotide Archive (ENA; www.​ebi.​ac.​uk/​ena; study number: PRJEB44214).