Development of a nomogram for predicting the risk of anastomotic leakage after a gastrectomy for gastric cancer.

AIMS

To investigate the incidence of and factors associated with anastomotic leakage (AL) following gastrectomy for gastric cancer.

METHODS

We retrospectively analyzed 3632 patients who underwent a laparoscopic gastrectomy or open gastrectomy for gastric cancer. A logistic regression model was used to identify the determinant variables, and a nomogram for AL was developed.

RESULTS

A total of 3632 patients were included in the study, 50 of whom (1.4%) developed AL. Postoperative deaths occurred in 6 (0.2%) patients with AL failure to rescue. Esophagojejunal AL (27/50) and Billroth I AL (20/50) were the most common types of AL. Gastrografin swallow was the main diagnostic method. The diagnosis was made a median of 9 days postoperatively. The median healing time for AL was 34.5 days. Of the 50 AL patients, 56% of patients could be managed nonsurgically, whereas 28% of patients required percutaneous radiologic drainage, 6% of patients were treated by endoscopy, and 10% of patients required a second surgery. A multivariate analysis showed the following adverse risk factors for AL: age ≥65 years, hemoglobin ≤8.0 g/dL and malnourishment. A multivariable model for AL showed a strong optimism-adjusted discrimination (concordance index, 0.675). The 5-year overall survival rates for patients without or with AL were 59.4% and 67.4%, respectively (p = 0.354).

CONCLUSIONS

AL was infrequent but was more prevalent in patients with age ≥65 years, hemoglobin ≤8.0 g/dL and malnourishment. We created a novel nomogram that can provide individualized prediction of AL in patients after a gastrectomy for gastric cancer, which may help clinicians in making treatment decisions.