Diagnostic accuracy of 1,000 endorectal ultrasounds before transanal endoscopic microsurgery for rectal neoplastic lesions.

INTRODUCTION

Endorectal ultrasound (EUS) is an essential tool for local staging of rectal neoplasia; however, its diagnostic accuracy in distinguishing non-invasive from invasive lesions before transanal endoscopic microsurgery (TEM) remains a matter of debate.

METHODS

A retrospective analysis of 1,000 consecutive EUS examinations performed before TEM between 1993 and 2025 was conducted using a prospectively maintained database. EUS levels (0-3) were correlated with the final histopathological outcome. Lesions were categorised as non-invasive (LGD, HGD, Tis) or invasive (pT1-pT3). Diagnostic metrics-sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy-were calculated overall and across three chronological periods. Separate analyses were performed for post-neoadjuvant (ypT03) and post-endoscopic resection groups.

RESULTS

Among 883 evaluable EUS studies for the primary analysis (non-invasive vs invasive pT1-pT3), overall sensitivity was 86.4%, specificity 64.9%, PPV 81.5%, NPV 72.7%, and accuracy 78.7%. All indices improved over time, with accuracy rising from 62.9% in early cases to 73.9% in the most recent period. In the post-neoadjuvant group (n = 47), sensitivity remained high (89.7%), but specificity was low (33.3%), likely due to overstaging related to fibrosis. In the post-endoscopic resection group (n = 70), the apparent accuracy was 44.3%, suggesting a high rate of false-positive invasion predictions.

CONCLUSIONS

EUS before TEM shows good overall accuracy and excellent reliability for excluding deep invasion, with progressive improvement over the past 3 decades. While overstaging remains a limitation in post-treatment and non-dysplastic lesions, EUS continues to play a pivotal role in selecting candidates for organ-preserving rectal surgery. In post-endoscopic resection scars and post-neoadjuvant rectum, EUS findings should be interpreted cautiously and integrated with MRI/endoscopic morphology.