The review concluded that anastomotic leakage had significant adverse oncological outcomes and was associated with an increased local recurrence rate and reduced survival after surgery for colorectal cancer. The uncertain quality of the evidence base and variation in the local recurrence results mean that a degree of caution is warranted when interpreting the authors’ results.

To examine the long-term oncological impact of anastomotic leakage after restorative surgery for colorectal cancer for the outcomes of local recurrence, distant recurrence and survival.

MEDLINE, EMBASE, CINAHL, and The Cochrane Library were searched up to August 2009 for articles published in English. Search terms were reported. Reference lists of relevant articles were handsearched.

Studies of the impact of anastomotic leakage on locoregional recurrence, distant recurrence and cancer specific survival after surgery for colorectal cancer were eligible for inclusion. Studies where non-restorative resections were not excluded or separable from the data, or that studied postoperative sepsis per se and not anastomotic leakage alone were excluded.

Included studies considered anastomotic leakage in the colon and/or the rectum. The studies were published from 1984 to 2009, and were conducted in a variety of countries including the UK, Japan, Italy and Germany. The use of stoma was not reported in most studies. Radiotherapy or chemotherapy use was not reported in most studies.

Two reviewers independently performed study selection; disagreements were resolved by consensus.

The authors stated that studies were assessed for methodological quality and validity, but no further details were reported.

Data were extracted on locoregional or distant recurrence and on cancer specific survival, and used to calculate odds ratios (ORs), with 95% confidence intervals (CIs). Studies were categorized on the basis of the type of anastomosis (colonic and rectal). Study authors were contacted for missing data.

Two reviewers independently extracted data; disagreements were resolved by consensus.

DerSimonian-Laird random-effects meta-analysis was undertaken to calculate pooled odds ratios and 95% confidence intervals. Statistical heterogeneity was assessed using Ι² and the Q statistic. Where analyses were deemed homogenous, fixed-effect meta-analysis was used.

Sensitivity analysis was conducted excluding poor quality studies, outlier studies, and studies which pre-dated total mesorectal excision.

Publication bias was assessed using Egger’s test and funnel plots.

Twenty-one studies were included in the review (21,902 patients), including one prospective randomised trial, 13 prospective cohort studies, and seven retrospective cohort studies. The study sample size ranged from 53 to 2,480 patients. The average length of follow-up ranged from 24 to 90 months, where reported.

There was a statistically significantly increased risk of developing a local recurrence of cancer when there was anastomotic leakage for studies describing rectal anastomoses (OR 2.05 95% CI 1.51 to 2.8; Ι²=61%; 13 studies). There was a statistically significantly increased risk of developing a local recurrence when there was anastomotic leakage for studies describing rectal and colon anastomoses (OR 2.90, 95% CI 1.78 to 4.71; Ι²=70%; nine studies). There was no significant difference in local recurrence in three studies in colon anastomoses. There was significant heterogeneity and some evidence of publication bias for local recurrence outcomes.

There was no significant increase in distant recurrence (seven studies).

In an analysis that excluded patients who died in the early postoperative phase, long-term cancer-specific mortality was significantly higher after anastomotic leakage (OR 1.75, 95% CI 1.47 to 2.1; Ι²=36%; 11 studies).

Sensitivity analysis did not significantly alter results.

Anastomotic leakage had significant adverse oncological outcomes and was associated with an increased local recurrence rate and reduced survival after surgery for colorectal cancer.

The inclusion criteria for the review were clearly defined. Several relevant data sources were searched. There was potential for language bias, as only English language studies were included. Publication bias was assessed and detected in some analyses. Attempts were made to reduce reviewer error and bias throughout the review.

Details of quality assessment criteria and their results were not reported, which made it difficult to determine the quality of the evidence base. The authors noted that there were differences across the studies for the diagnosis of anastomotic leakage and disease recurrence. Only one of the studies was a randomised controlled trial, the rest were observational studies. Studies were combined using standard statistical techniques. Statistical heterogeneity was assessed, which was appropriate. Significant heterogeneity was detected in the local recurrence outcomes, which may indicate that the data was not suitable for pooling.

This, coupled with the uncertain quality of the evidence base, mean that a degree of caution is warranted when interpreting the authors’ results.

Practice: The authors stated that meticulous surgical techniques to minimize anastomotic leakage were needed.

Research: The authors stated that further trials were unlikely due to the large number of participants required and likely low number of events.

Cancer Research UK.