Source of errors in the evaluation of early rectal cancer by endoluminal ultrasonography

Dis Colon Rectum. 2001 Sep;44(9):1302-9. doi: 10.1007/BF02234788.

Abstract

Purpose: Although preoperative evaluation of early rectal cancers can be done by endoluminal sonography and by means of colonoscopic findings, it is still controversial whether endoluminal sonography can effectively discriminate mucosal from submucosal lesions. This study was performed to verify objective causes of errors in the evaluation of early rectal cancer (T0/1) using a review of videotaped endoluminal sonography images.

Methods: Eighty-nine patients with suspected early rectal cancer on endoluminal sonography were included. Two different scanners with appropriate probes were used according to tumor location, i.e., transrectal ultrasonography was used to scan up to 8 cm of the rectum above the anal verge, whereas endoscopic ultrasonography was used to assess higher lesions. Endoluminal sonography images were correlated with histologic infiltration and were reevaluated carefully to identify sources of errors.

Results: Sensitivity and specificity were 83.1 and 96.5 percent, respectively, for tumor staging, whereas sensitivity was very low compared with specificity (16.7 vs. 90.2 percent) for metastatic lymph nodes. Endoluminal sonography images showed irregularity of the underlying tumor border (P < 0.01) and hypoechoic blurring or cutoff of the inner and outer hypoechoic layers (P < 0.001), all of which closely correlated with histologic infiltration of tumor cells. Overstaging occurred more than twice as often as understaging in tumor reevaluation (14 vs. 5 occurrences). In contrast to tumors, lymph nodes showed a similar amount of both overstaging (four cases) and understaging (five cases). The sources of errors were summarized as five types: false instrumentation, interpretive errors, anatomic defects, imaging failure, and inevitable errors.

Conclusions: Because false instrumentation, interpretive errors, and anatomic defects were considered preventable, 23 (82.1 percent) of the 28 errors might have been avoided. Therefore, a clear image by endoluminal sonography can effectively distinguish mucosal from submucosal lesions in early rectal cancer.

Publication types

  • Evaluation Study

MeSH terms

  • Diagnosis, Differential
  • Endosonography* / methods
  • Endosonography* / standards
  • False Negative Reactions
  • Humans
  • Neoplasm Metastasis
  • Neoplasm Staging
  • Observer Variation
  • Rectal Neoplasms / diagnosis
  • Rectal Neoplasms / diagnostic imaging*
  • Sensitivity and Specificity