[1246] Interobserver Agreement on High-Grade Endometrial Carcinoma and Correlation with ER, p53 and p16 Expression

Davinder Sidhu, Maire Duggan, Phil Clement, Carol Ewanowich, Jocelyne Arseneau, Matthew Cesari, Martin Kobel, Guangming Han. University of Calgary/Calgary Laboratory Services, Calgary, Canada; Vancouver General Hospital, Vancouver, Canada; Royal Alexandra Hospital, Edmonton, Canada; McGill University Health Centre, Montreal, Canada

Background: High grade endometrial carcinomas (HGEC) are a heterogeneous group. Inclusion criteria for clinical trials such as PORTEC 3 include high-grade tumor with FIGO grade 3 endometrioid (EC3), serous (SC) or clear cell (CCC) histology. We aim to assess interobserver reproducibility in diagnosing HGEC histologic types and whether routine immunohistochemical marker may aid diagnosis.
Design: 116 cases of HGEC were identified from gynecologic oncology consultation files at Tom Baker Cancer Centre. Representative slides (1-3 /case) were independently reviewed by 5 pathologists according to practice. The diagnoses were summarized into categories: endometrioid low-grade (ECLG), EC3, SC, CCC and other. Interobserver reproducibility was assessed by kappa analysis. p16 (0<80%, 1≥80%), p53 (0<50%, 1≥50%) and ER(0<50%, 1≥50%) expressions were assessed using tissue microarray.
Results: A consensus diagnosis (≥ 80% agreement) was established in 78 cases (67%) with the reviewers' submitted diagnosis, and in 85 cases (73%) with the categorized diagnosis. The kappa values are shown in Table 1. Of the 31 cases with no consensus agreement, 14 (12%) had major disagreement (high-grade vs. low-grade), 17 (15%) had minor disagreement in high-grade tumor type. Among the major disagreement cases, the main problems were ECLG vs. EC3 and ECLG vs. SC. A panel of ER, p16, p53 correlated with histologic diagnosis of EC3 vs. SC in 89% of cases (Figure 1). Six cases with major disagreement showed an immunophenotype suggestive of endometrioid type. The other 8 cases were triple positive, a combination that is unlikely for ECLG.

Kappa Value for Each Pair of Reviewers
ReviewerSubmitted DiagnosisCategorized Diagnosis

Conclusions: 1. Interobserver disagreement exists in 1/3 to 1/4 of cases in grade and type. 2. Major disagreement on high-grade vs. low-grade is seen in 12% of cases with potentially significant clinical consequences. 3. A panel of ER, p16, and p53 may help to assign endometrioid or serous type.
Category: Gynecologic & Obstetrics

Wednesday, March 21, 2012 1:00 PM

Poster Session VI # 198, Wednesday Afternoon


Close Window