[632] Interobserver Variability in the Recognition of Desmoplasia in Colorectal Carcinoma

John V Groth, Caleb W Cheng, Lisa Rooper, Sanjiv Prabhu, Zachary Lewis, Cameron Dowlatshahi, Betty Chung, Wei Song, Chelestes Grace, Andrew Bandy, Steven Garzon, Odile David, Hajwa H Kim, Andre Kajdacsy-Balla, Robert J Cabay. University of Illinois Hospital & Health Sciences System, Chicago, IL; Johns Hopkins University School of Medicine, Baltimore, MD

Background: Desmoplasia is a descriptive term applied to the presence of excessive connective tissue deposition at the advancing front of invasive carcinoma. Desmoplasia may also provide some clues regarding the potential of metastatic disease for some tumors. The histologic appearance of desmoplasia is variable, and the assessment of the microscopic features of desmoplasia is not part of common pathology practice. A categorization of desmoplasia as mature, intermediate or immature has been proposed in advanced rectal cancer (Ueno et al. Gut 2004;53:581-586). We questioned whether this classification scheme can be easily utilized and reproduced in daily practice.
Design: Five pathologists and five residents at our institution determined the absence or presence of desmoplasia and, when appropriate, after reading the above cited article, applied the categorization scheme to two selected areas in each of 20 colorectal carcinoma cases. No training sessions were given. Interobserver agreement was measured using the categorization scores each observer assigned.
Results:

Interobserver Agreement
DesmoplasiaObserversKappaStandard ErrorProb>Z
Presence/absencePathologists + Residents0.1240.0240.000
 Pathologists0.1570.0500.001
 Residents0.0940.0500.030
CategorizationPathologists + Residents0.1870.0140.000
 Pathologists0.2540.0310.000
 Residents0.1660.0300.000


Desmoplasia categorization scores: 0 = No desmoplasia, 1 = Immature, 2 = Intermediate, 3 = Mature
Conclusions: Pathologist, resident and overall kappa scores showed substantial interboserver variability in both the identification and characterization of desmoplasia. This demonstrates the difficulty in achieving agreement regarding the microscopic nature of desmoplasia and identifies the need for additional training measures. Extra educational experiences, such as live or computer-based training sessions, may lead to reduced interobserver variability and increase the likelihood of adding the assessment of desmoplasia to routine clinical practice should subsequent analyses of associated outcome data warrant it.
Category: Gastrointestinal

Monday, March 4, 2013 9:30 AM

Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 97, Monday Morning

 

Close Window