Performance of endoscopic ultrasound in staging rectal adenocarcinoma appropriate for primary surgical resection.

ABSTRACT

BACKGROUND & AIMS: Endoscopic ultrasound (EUS) often is used to stage rectal cancer and thereby guide treatment. Prior assessments of its accuracy have been limited by small sets of data collected from tumors of varying stages. We aimed to characterize the diagnostic performance of EUS analysis of rectal cancer, paying particular attention to determining whether patients should undergo primary surgical resection. METHODS: We performed a retrospective observational study using procedural databases and electronic medical records from 4 academic tertiary-care hospitals, collecting data on EUS analyses from 2000 through 2012. Data were analyzed from 86 patients with rectal cancer initially staged as T2N0 by EUS. The negative predictive value (NPV) was calculated by comparing initial stages determined by EUS with those determined by pathology analysis of surgical samples. Logistic regression models were used to assess variation in diagnostic performance with case attributes. RESULTS: EUS excluded advanced tumor depth with an NPV of 0.837 (95% confidence interval [CI], 0.742-0.908), nodal metastasis with an NPV of 0.872 (95% CI, 0.783-0.934), and both together with an NPV of 0.767 (95% CI, 0.664-0.852) compared with pathology analysis. Incorrect staging by EUS affected treatment decision making for 20 of 86 patients (23.3%). Patient age at time of the procedure correlated with the NPV for metastasis to lymph node, but no other patient features were associated significantly with diagnostic performance. CONCLUSIONS: Based on a multicenter retrospective study, EUS staging of rectal cancer as T2N0 excludes advanced tumor depth and nodal metastasis, respectively, with an approximate NPV of 85%, similar to that of other modalities. EUS has an error rate of approximately 23% in identifying disease appropriate for surgical resection, which is lower than previously reported.