Antibiotic Bowel Prep Before Colon Cancer Surgery May Be Selecting for the Bacteria That Cause Recurrence
Antibiotic bowel prep before colon cancer surgery is meant to prevent infection; but in patients whose cancer recurs, the resistant bacteria in their gut are unfazed, and on the day of surgery their total bacterial load is 23 times higher than in non-recurring patients. The prep may be precisely what selects for future recurrence.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Bowel prep cannot wash away the microbial signature of recurrence
The cohort is indeed small: 19 patients, 6 of whom recurred about 10 months after surgery, roughly 30%, with sites mostly in the lung, liver, and anastomosis. Fecal samples taken on the day of surgery captured the key feature: after mechanical bowel prep, oral neomycin and metronidazole, plus intravenous cefoxitin, the originally clear microbial differences between the left and right colon and the rectum were all flattened, but the compositional difference between the 'will recur' and 'will not recur' groups remained. All these patients received standard bowel preparation, meaning that by the time the surgeon closed, the gut microbiota of the recurring patients had already diverged from that of the non-recurring ones.
Bowel prep is nearly ineffective against resistant microbiota
Each patient served as their own control: after bowel prep, the total bacterial load in the non-recurrence group dropped about 10-fold; in the recurrence group, the bacterial load was almost unaffected. Thus, on the day of surgery, the total bacteria in the stool of recurring patients was 23 times higher than in non-recurring patients (calculated by 16S rRNA gene copy number). This is not because the prep was done incorrectly—oral neomycin plus metronidazole, laxatives, and an intravenous push of cefoxitin 30 minutes before skin incision were all strictly followed. The problem is that this perioperative regimen at best clears susceptible bacteria, but has almost no effect on the bacteria in the recurring patients' guts that can withstand all three drugs.
Bowel prep amounts to an in vivo selection screen
Resistance gene sequencing provides the material basis for the 'selection' hypothesis: the total number of resistance genes in the gut of recurring patients, as well as the number of resistance genes against the three bowel-prep drugs—neomycin, metronidazole, and cefoxitin—were higher than in non-recurring patients. Petra Levin pointed out on the show that this is not a vague 'selective pressure' but the most standard selection in a microbiology lab—like screening for successful transformants on an ampicillin plate, except here the entire gut microbiota is being screened. But genes and phenotype do not fully overlap: the authors did not observe functional resistance to neomycin, and they acknowledged that the small cohort can only yield directional conclusions.
The pathogenic phenotype only appears after bowel prep
Before surgery, collagenase gene abundance did not differ between the recurrence and non-recurrence groups; only by the day of surgery did the difference emerge. The recurrence group had higher relative abundance and absolute counts of collagenase genes, as well as higher collagen-degrading activity of the microbiota; among the 13 patients who did not recur, 12 had no collagenolytic colonies with clear zones. The authors put it directly: antibiotic bowel prep is a prerequisite for revealing this pathogenic phenotype. Before prep, enzyme-producing bacteria are inconspicuous mixed in the intact community; after prep clears susceptible bacteria, the remaining resistant and enzyme-producing taxa gain the upper hand in the residual community.
Enzyme-producing enterococci promote cancer cell metastasis
Functional experiments distinguish correlation from causation. Adding the supernatant from the day-of-surgery microbiota of recurring patients to the upper chamber of a transwell significantly enhanced the migration of human and mouse colon cancer cells, as well as invasion through a collagen-coated layer; supernatant from the non-recurrence group had no effect. From the 72 colonies picked from recurring patients' samples, Enterococcus was the dominant genus; when cancer cells were treated separately with collagenase-producing and non-producing strains, only collagenase-positive strains induced migration and invasion, while negative strains had no effect. Whole-genome comparison between enzyme-producing and non-producing strains revealed 8,961 variant alleles, and the enzyme-producing strains also carried more virulence-related genes such as adhesins and type IV secretion systems.
Phage uses size to pull off bacterial pili
The second story in this episode is unrelated to cancer, but the mechanism is counterintuitive. The single-stranded RNA phage uses type IV pili as its receptor, and its MAT protein is responsible for binding the pilus; experiments showed that the N-terminal 200 amino acid residues of the MAT protein are sufficient to induce pilus shedding, and the amount of shedding is proportional to the number of viral particles. To rule out enzymatic cleavage, the researchers attached a large fluorescent microsphere to this protein segment—it also caused pilus shedding. This shows that the effect is due to size, not catalytic activity: when the pilus is about to retract into the cell, it gets stuck by this 'big head,' and the connection is mechanically broken. Since type IV pili are involved in motility, adhesion, and biofilm formation, the truncated MAT protein could become an anti-virulence tool that does not kill bacteria.
In their own words · checked verbatim
the total number of bacteria was 23 times greater in the stool collected on the day of surgery compared to those who did not develop a recurrence.
This is classic selective pressure, folks.
when you transform a plasmid into a strain and you select on ampicillin, it's basically an experiment in a human.
this suggests that the antibiotic bowel prep is required to uncover this virulent phenotype.
there were no collagenolytic colonies in the 12 of the 13 patients who never developed a recurrence.
this mechanical biology model of phage infection where force generation and structural physics becomes a central player in the entry of the virus.
Figures
| New colorectal cancer cases in the US in 2024 | 153,000 cases | 4:17 |
| Recurrence rate after surgical resection | 30% | 4:17 |
| Number of recurrences in this study cohort | 6/19, about 30%, about 10 months | 9:32 |
| Fold difference in total bacteria on surgery day between recurrence and non-recurrence groups | 23 times | 13:41 |
| Fold decrease in total bacteria after bowel prep in non-recurrence group | 10 times | 13:41 |
| Proportion of non-recurring patients without collagenolytic colonies | 12/13 | 19:53 |
| Number of colonies picked from recurring patients' samples | 72 | 25:00 |
| Number of variant alleles between enzyme-producing and non-producing Enterococcus strains | 8,961 | 26:05 |
| Common sites of recurrence | lung, liver, anastomosis | 9:32 |
Glossary
- bowel prep
- Cleansing of the bowel before colorectal surgery to reduce infection risk, including oral antibiotics and laxatives.
- collagenase
- An enzyme that degrades collagen; bacteria use it to invade tissue, and cancer cells also use it to pave the way for metastasis.
- type IV pilus
- A retractable filamentous 'grapple' on the bacterial surface, involved in twitching motility, adhesion, and biofilm formation.
- MAT protein
- The maturation-associated protein of single-stranded RNA phages; the part protruding from the capsid is responsible for catching and breaking off pili.
How to listen
Colorectal surgeons and oncologists, gut microbiome researchers, and any clinician who must decide whether to continue antibiotic bowel prep before surgery.
If you only care about cancer, you can skip the RNA phage mechanism study after 32:30.