The microbiome and surgery: breakthrough or just hype?

In 1980, the number of studies including the word “microbiome” was around eleven, today using microbiome as a search word in pubmed yields over 100,000 entries. For surgeons, the relevance of the gut microbiome lies in its promise to explain disease pathogenesis (i.e cancer, appendicitis, diverticulitis,  surgical site infections) and treatment effects (antibiotic prophylaxis, bowel preparation, etc). Yet because the data output of a typical microbiome analysis can be vast, determining what is signal versus noise has become problematic. Similar to the early days when human gene chips became available, displays of massive datasets indicating that a patient in group A is “different” from patients in group B leaves readers skeptical. For example, when the human gene chip became available, the transcriptome (mRNA expression of nearly 20,000 protein coding genes) of human samples could be compared between patient samples. Yet these initial screens only described “differences” between groups of patients and failed to identify actionable items. The descriptive nature of these studies has forced some, for example, to completely question the genetic basis of cancer. Are we falling into the same trap with microbiome studies?

Why microbiome studies are different. Claims that sequencing of the human genome was going to lead to major cures of complex diseases such as cancer have indeed been disappointing. First we were told that cancer is a genetic disease; once “junk” DNA turned out not to be junk, and once it became clear that gene-environment interactions (via histone modification?) played an important regulatory role in gene expression, the role of “lifestyle” became the new hype¹. So where does that leave the microbiome in all of this? Issues such as how indoor and outdoor air quality, smoking, alcohol consumption, dietary choices, etc., influence one’s microbiome and then how in turn, its metabolites change host genetics is now under investigation.

Perhaps one of the most striking examples of the power of microbiome analyses is a study examining the gut microbiome of 34 monozygotic twins discordant for multiple sclerosis; one twin suffered from the disease while the other did not². Deep analysis of faecal samples from the discordant twins demonstrated clear differences; yet when samples were transferred into germ-free mice, only samples from the affected twin produced an encephalomyelitis-like picture whereas unaffected twin samples did not. The neurotoxic metabolites from the gut microbiome that play a role in this effect are now coming to light³. The fact that monozyotic twins are born with different fingerprints and the genomic identity in their microbiomes is highly variable should diminish our enthusiasm for interrogating host genes only⁴.  Animals are holobionts, consisting of both host and microbial genes, each interacting with one another and with the environment.  At the individual patient level, this presents major challenges to understand disease pathogenesis and its treatment.  

What is now emerging as centrally important to human health is diet, and the gut microbiome is centre stage in this regard. Relevant to surgical sciences and outcome studies is the role of obesity, smoking, diet and medication use on surgical outcome. As many of such factors are modifiable, surgeons are particularly interested in how such factors can be manipulated to decrease complications.  While antibiotic exposures, smoking, and substance abuse are clearly areas that can be preoperatively addressed, both the quality and quantity of an individual’s diet is emerging as most relevant and certainly the most modifiable. For example, mice that consume a western diet are not only more susceptible to anastomotic leak, they can also develop surgical site infections from gut microbiota via the Trojan Horse Hypothesis⁵. Conversely, patients who consume a diet rich in fibre following curative colon cancer surgery are less likely to develop a recurrence^6,7. Taken together, nutritional assessment beyond conventional markers such as serum visceral protein status (i.e., albumin, pre-albumin) and muscle mass (sacropenia, frailty) may involve a comprehensive dietary history, the use of food logs and dietary prehabilitation prior to surgery.

Surgeon beware. Surgeons tend to make changes in their practice when the proposed intervention has a strong scientific premise behind it and in some way just seems to “make sense.”  Most clinical studies that simply compare one intervention group (i.e bowel prep, symbiotics, probiotics) to another fall short in convincing experts in the field to make a major change in clinical practice. One reason for this is the failure to adequately measure reliable “readouts” that can explain the variance within the group of treated individuals. Comparing mean values only “between treatment groups” without reconciling the differences in outcome “within groups” remains a problem⁸.  Just as it is not helpful to demonstrate that older, sicker patient with more advanced cancer, on average (i.e., mean values comparing between groups) have worse survival compared to younger, healthier patients with less invasive cancer, it is not helpful to present massive datasets from a microbiome study that demonstrate that the microbiomes among those patients with complications are different compared to the microbiomes of patients without complications. Causal inference at the molecular level is needed and within-group variability must be explained. 

In summary, microbiome sciences are indeed the next big thing as they are able to explain phenomena at the individual patient level, especially when samples can transferred to mice that phenocopy the disease of interest. How our environment, our life history and the disease process itself affect all genes, be they microbial or host derived, is what matters most. Surgeons recognize that they perturb multiple systems when they operate on patients. As a result, they remain eager to understand, at the individual patient level, how to disentangle the molecular mechanisms that explain why one patient recovers uneventfully while another does not.