A predictive model for lymph node yield in colon cancer resection specimens.

Objective: To develop a predictive model of lymph node yield in a series of colon cancer resection specimens with detailed anatomic and surgical technique data.

Background: Lymph node yield in colon resection specimens has been associated with accuracy of staging and cancer outcomes. We hypothesized that lymph node yield is associated with multiple factors including patient, tumor,and surgical variables.

Methods: The pathology specimens from 152 elective colon neoplasm resections were prepared so that the lymph nodes were separated according to their anatomic relationship to the vascular pedicles and to the tumor. Prior to dissection, the specimen was measured. A linear regression analysis of a priori identified predictors and confounders of lymph node quantity was performed. Potential predictors in the model were age, gender, tumor stage, size, location,and differentiation, presence of lymphovascular or perineural invasion,mucinous histology, number of vascular pedicles, and use of endoscopic tattoo. Potential confounders were American Society of Anesthesiologists class, body mass index, count of lymph node metastasis, and specimen length.

Results: Tumor size, tumor location, number of resected pedicles, and use of tattoo had a significant linear or quadratic relationship with lymph node yield when controlling other variables. 23% of the variation in lymph node count was explained by the 15 variables in the model. A model with the 4 significant variables explained 19% of the variation.

Conclusion: Multiple tumor and surgical factors are associated with lymph node yields in colon specimens. A standard minimum of lymph nodes may not be applicable to all colon cancer resections.