摘要

Background: Collagen changes in the extracellular matrix caused by neoadjuvant chemoradiotherapy are a potential mechanism of anastomotic leakage. We aimed to construct a fully quantitative collagen score to describe collagen structure changes in the extracellular matrix and then develop and validate a prediction model to identify patients who are at a high risk of postoperative anastomotic leakage. Methods: This is a retrospective study in which 372 patients were enrolled, and their baseline clinicopathological characteristics were collected. Anastomotic distal and proximal "doughnut" specimens underwent second harmonic generation imaging, and collagen features were extracted. A LASSO regression was used to select significant predictors, and the collagen score was constructed. A prediction model based on collagen score was developed and internally and externally validated. Results: The primary cohort included 214 consecutive patients, and the anastomotic leakage rate was 8.9%. The validation cohort comprised 158 consecutive patients, and the anastomotic leakage rate was 10.1%. The collagen score was significantly related to anastomotic leakage in both cohorts (P < .001). Multivariate analysis revealed that tumor location, preoperative albumin, and collagen score were independent predictors of anastomotic leakage. These 3 predictors were incorporated into the prediction model, and a nomogram was established. The model showed good discrimination in the primary (area under the curve: 0.954) and validation (area under the curve: 0.928) cohorts. Decision curve analysis demonstrated that the nomogram was clinically useful. Conclusion: The collagen score is associated with anastomotic leakage, and the collagen nomogram based on the collagen score is useful for individualized prediction of anastomotic leakage in rectal cancer patients with neoadjuvant chemoradiotherapy after surgery.

  • 单位
    中山大学; 南方医科大学