Introduction
Gastrointestinal surgical pathology represents a growing challenge to healthcare systems around the world,
1,2 with both developing and industrialised societies shouldering this burden.
1,3‐6 In the UK alone, 1.3 m general surgical interventions take place annually.
7 Unfortunately, major gastrointestinal surgery is accompanied by complication rates of up to 24%.
8 Some 220,000 hospital admissions were due to complications from surgery in 2017–2018 in the UK alone.
9 The commonest post-operative complication is infection, but others include cardiac arrhythmias and acute kidney injury, as well as anastomotic leakage and wound dehiscence.
10 Although these have been ameliorated by advances in critical care, minimally invasive techniques, subspecialisation and centralisation of expertise, improved risk stratification may inform future interventions to reduce risk. Quantifying these risks accurately is ever more important as we face decisions to operate on a population that is increasingly older with significant co-morbidities.
Current perioperative risk management involves well-established tools such as P-POSSUM and ASA grades.
11 However, these tools involve subjective judgement, and there is evidence that P-POSSUM scoring in particular overestimates risk.
12‐14 The genome’s contribution to post-operative complications is under-researched and may better explain variance which is unaccounted for by clinical assessment tools. For instance, there is convincing research linking genetic variation with the host response to sepsis and trauma, but there is limited work investigating this in surgery.
15‐18 Single nucleotide polymorphisms (SNPs) are the commonest form of genetic variation,
19 and many have been linked to disease.
20 For instance, polymorphisms associated with high
TNF-α have been associated with sepsis.
18,21TNF-α is a critical cytokine regulating acute inflammation via activation of immune cells, release of large amounts of inflammatory mediators, downstream signalling and endocrine effects.
22
The association between particular SNPs and risk of developing post-operative complications has been explored in cardiac surgery,
23 with one group reporting three SNPs associated with post-operative myocardial infarction.
24 There have been several genetic association studies taking a similar approach in gastrointestinal surgery, which have aimed to identify whether particular polymorphisms are associated with poor outcome. These studies are challenging to interpret and were limited by small sample sizes, incomplete reporting of results, population stratification and implausible effects of individual SNPs, amongst others. To our knowledge, this is the first review examining these publications in a systematic manner. Such work may identify high-risk genes, which if externally validated as part of a clinical tool, could allow clinicians to offer improved risk prediction as part of the increasingly affordable ambition of personalised medicine. Addressing these risks perioperatively, with prehabilitation for example, could reduce future surgical morbidity and mortality.
Methods
Objectives
To systematically review studies in patients who have undergone major gastrointestinal surgery, and whether any single nucleotide polymorphisms are associated with post-operative complications.
Protocol and Registration
We followed the guidelines of the PRISMA statement
25 in preparing this review and registered with the international prospective register of systematic reviews, PROSPERO (CRD42019122342) on 15 February 2019.
26
Search Strategy
We searched the electronic databases MEDLINE (using the Ovid platform) and Web of Science for papers published from January 1990 to January 2019. The following free-text terms were used: “surgical procedures”, “gastrointestinal surgery”, “genetic polymorphism”, “single nucleotide polymorphism”, “post-operative complications” and “post-operative outcomes”. For MEDLINE searches, we used the MeSH terms “digestive system surgical procedures” AND “polymorphism, genetic” AND “postoperative complications”. Searches involving MeSH terms included index subheadings. We also reviewed reference lists of both identified articles and review articles for additional relevant studies. We did not have any language restrictions.
Study Selection, Inclusion and Exclusion Criteria
Studies were independently screened by title and abstract by JB and SL. Both authors subsequently performed full-text review. Any disagreement was resolved by discussion with a third author. We considered a study for inclusion if it met the following criteria: (1) patients underwent major oesophagogastric, hepatobiliary and colorectal resections (elective or emergency), (2) the variation studied was a single nucleotide polymorphism, and (3) post-operative complications were investigated in the 90-day post-operative period. We excluded studies in less major operations, including cholecystectomy, appendectomy and hernia repair. We also excluded outcomes which were not strictly complications, such as analgesic requirement and risk of cancer recurrence. Participants with particular conditions including inflammatory bowel disease, transplant patients and immunocompromised individuals were also excluded.
Data Extraction and Synthesis
For each eligible study after full-text review, we extracted year of publication, study population, sample size, patient demographics, post-operative outcomes, outcome measure, statistical analyses and statistic presented. Genetic information extracted included frequency of alleles, SNPs of interest, SNP reference identifier (where published), genotype method and locus.
Risk of Bias and Quality Assessment
Each study was assessed for quality using the STrengthening the REporting of Genetic Association Studies (STREGA) addendum to the Strengthening Reporting of Observational Studies (STROBE) statement.
27
Statistical Analysis
Where multiple studies (defined as ≥ 3) reported on the same SNP, a meta-analysis was conducted. The analysis, as well as tests of heterogeneity, and Hardy-Weinberg calculations were performed using the MetaGenyo: Meta-Analysis of Genetic Association Studies tool.
28 Our
p value threshold for Hardy-Weinberg equilibrium was 0.05. If the included studies demonstrated significant heterogeneity, a random effects model was applied. Heterogeneity was assessed with a
I2 statistic with values of 25% considered low, 50% moderate and 75% high. Egger’s test assessed for publication bias.
Discussion
Key Results
This review found that research examining SNPs and post-operative complications has focussed almost exclusively on immunity and infection as clinical outcomes. We identified four polymorphisms of immune cytokine genes were associated with post-operative complications. We also conducted a meta-analysis of the only SNP which was examined by three or more studies,
TNF-α-308. This polymorphism is involved in production of
TNF-α, pro-inflammatory cytokine. Our meta-analysis examined the SNP in the context of gastrointestinal surgery. One study reported patients with the
TNF-α-308 GG genotype had fourfold higher odds of a post-operative infection: OR 4.2 (CI 1.7–10.5),
39 whilst another reported that it was the AA genotype which was associated with infection with OR 4.17 (CI 1.5–11.48).
40 We can compare this with the results of a previous meta-analysis which examined
TNF-α-308 in sepsis more generally. This found a much more conservative effect size with OR = 1.32 (
p < 0.001) in the A vs G allele model in a much larger data set (
n = 9373).
45 The differing results could conceivably be explained by statistical error, for the reasons outlined below.
Limitations
We found that the variety of SNPs identified have not been researched in an organised fashion. Despite the number of SNPs examined (22), there was little overlap between papers in terms of which SNPs were studied. There was poor consistency between findings and results were not replicated. These findings are thus highly vulnerable to type 1 and 2 error. The problem is compounded by the very small size of relevant published research. We only identified five relevant papers in total, and only three which examined non-infectious complications. As Table
1 shows, there was also variability in how infection was defined in these studies, which may have contributed to the heterogeneity of results.
Our work has demonstrated highly variable outcome measures, populations, genotyping methods and conclusions. These methodological issues are common in genetic epidemiology.
46 Results suitable for extensive meta-analysis are essential in genetic association studies but only a minority of positive gene-association results are ever reproducible.
47,48 Large, well-powered samples are required to examine the effect of single polymorphisms on complex multifactorial outcomes,
47,49 particularly when the true effect size may be small. The evidence presented here shows that further meta-analysis is essential because of small sample sizes, but also challenging because too few papers examined the same SNPs.
The scattershot approach taken in the existing literature is unlikely to yield meaningful results. These papers have all been genetic association studies with SNPs pre-selected. This makes the findings vulnerable to confirmation bias.
Interpretation and Biological Mechanisms
Despite the difficulties we have highlighted, it is biologically plausible that genetic variation leading to a dysregulated immune system could result in post-operative infection or sepsis. Sepsis is a complex entity involving aberrant host responses to an infectious pathogen.
50 It involves a broad range of pro- and anti- inflammatory signals with mediation at the organ, tissue, cellular and molecular levels.
51,52
TNF-α is a cytokine and pyrogen with a multitude of complex inflammatory effects, produced predominantly by macrophages but also by B and T lymphocytes.
53,54 It induces downstream inflammatory pathways via nuclear factor κΒ activation and arachidonic acid formation.
55 It has a synergistic effect with IL-1β and also has roles in apoptotic cell death, macrophage differentiation, IL-6 induction, fibroblast production, B cell proliferation and direct antiviral activity.
53 It is also thought to have both pro- and anti-inflammatory roles depending on the phase of the acute insult.
53,55 The
TNF-α-308 functional polymorphism has been shown to correlate with variability in TNF cytokine production,
56 and individuals who are genetically predisposed to over-express
TNF-α suffer more pronounced inflammatory reactions.
55 This would be consistent with published meta-analyses linking
TNF-α polymorphisms to sepsis generally.
18,45
Despite the promising explanatory framework for why a cytokine polymorphism may contribute to post-operative complications, our work shows that the existing data in major gastrointestinal surgery are not clear or strong enough to draw clinically meaningful conclusions at present.
Future Work
There are no genome-wide association studies (GWAS) published in this area. We suggest that an efficient route for answers would be for this type of study to be performed in the first instance, identifying candidate SNPs. These could then be tested in multiple large-scale genetic association studies, which could subsequently be meta-analysed.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.