Introduction Following colorectal cancer resection, high-risk tumour pathology guides provision of adjuvant therapy and follow-up. Currently, routinely assessed features include local invasion, nodal status,venous invasion (VI), serosal invasion, differentiation and margin status. It was recently reported that elastica tissue stains increase frequency of detection of VI to > 50%, reducing interobserver variation, increasing its prognostic value (Roxburgh 2010, Kirsch 2013). Given recurrence is either local, systemic or both, we sought to examine the role of routinely assessed pathological criteria including VI (detected using elastica) in determining recurrence following resection for colorectal cancer.

Methods From an institutional database 555 patients undergoing curative resection between 1997–2009 were identified with recurrence and follow-up data available. Pathology data was taken from reports issued at the time. VI was assessed prospectively with routine elastica staining for 417 patients and retrospectively in 138. Analysis was performed with binary logistic regression. Due to a high number of comparisons, to enter the multivariate model a significance level of < 0.01 was used.

Results Of 555 patients, 141 (25%) developed recurrence. Frequency of VI detection was 54%. On logistic regression, elastica detected VI, T stage, lymph node involvement, serosal involvement, margin involvement, tumour perforation, peri-tumoural inflammation (Klintrup grade) and tumour necrosis were predictors of recurrence (any site, all P < 0.05). Differentiation was not. On multivariate analysis VI (OR 3.27, P < 0.001), lymph node involvement (OR 2.34, P = 0.005), serosal invasion (OR 2.38, P = 0.005), Klintrup grade (OR 0.68, P = 0.037) were independently predictors of recurrence. Most recurrence was systemic (75%). The same features predicted systemic recurrence as did overall recurrence but on multivariate analysis, only VI (OR 2.90, P = 0.004), lymph node involvement (OR 2.27, P = 0.012) and necrosis (OR 1.63, P = 0.013)were independent predictors of systemic recurrence. In the 35 cases of local recurrence VI, T stage, lymph node involvement, serosal involvement and margin involvement were significantly related (P < 0.05). On multivariate analysis, only VI (OR 2.28, P = 0.057) and T stage (OR 2.53, P = 0.003) were independent predictors.

Conclusion Whilst several pathological features predict local and systemic recurrence after surgery, VI detected at increased frequency (54%) with elastica stains was the only consistent, independent predictor of recurrence, at least as important as nodal spread. These results support implementation of routine measures such as elastica staining to optimise reporting of VI.

Disclosure of Interest None Declared