Endometrial Adenocarcinoma with Clear Cell Morphology: Interobserver Variability and Immunohistochemical Analysis Using Hepatocyte Nuclear Factor -1β
G Han, D Delair, CB Gilks, M Kobel, PB Clement, AD Tabrizi, RA Soslow. Memorial Sloan-Kettering Cancer Center, New York; University of British Columbia, Vancouver, Canada; Tabriz University of Medical Science, Tabriz, Islamic Republic of Iran
Background: Accurate subtyping of endometrial adenocarcinoma remains a challenge. We investigated interobserver variability in diagnosing endometrial carcinoma with clear cell morphology and the usefulness of HNF-1β for diagnosis of endometrial clear cell carcinoma (CC).
Design: 44 cases of endometrial adenocarcinoma originally diagnosed as CC, mixed clear cell carcinoma (MEC), or non-clear cell carcinoma (NCC) with clear cell changes were reviewed by one gynecologic pathologist (the reviewer) and reclassified into 4 groups: CC (n=12), NCC (n=22), MEC (n=2), and not sure (NS, n=8). Cases were then reviewed by 5 pathologists (the panel) using WHO criteria, without knowing original or reviewer diagnoses. Interobserver reproducibility was evaluated by kappa analysis. Immunohistochemistry for HNF-1β was performed. Results were recorded as the percentage of positive tumor cells multiplied by stain intensity (0-absent, 1-weak, 2-moderate, 3-strong).
Results: There was only moderate agreement among panelists in 3 of 4 diagnostic subgroups (kappa=0.56 for CC, 0.57 for NCC, 0.53 for NS, 0.65 for MEC). 10 cases (91%) with an HNF-1β score > 150 were diagnosed as CC by the reviewer or by a majority of panelists (5 of these cases were diagnosed as CC by both the reviewer and a majority of panelists). One case (9%) was diagnosed as NS by the reviewer and NCC by a majority of panelists. Of 31 cases with an HNF-1β score <150, 1 case (3%) was diagnosed as CC by the reviewer and a majority of panelists; 3 cases (10%) were diagnosed as NS by the reviewer and CC by a majority of panelists; 19 cases (61%) were diagnosed as NCC by both the reviewer and a majority of panelists; 8 cases (26%) were diagnosed as NCC by either the reviewer or a majority of panelists. The clear cell and non-clear cell areas in two MEC cases showed different HNF-1β scores, 160 and 0, respectively, in one case, and 290 and 90, respectively, in the other.
Conclusions: Only moderate interobserver agreement was noted in the evaluation of endometrial CC, unlike ovarian CC where interobserver agreement is reportedly substantial to almost perfect. Despite that, diffuse HNF-1β staining defines a homogeneous group of carcinomas with clear cells that with only rare exceptions represent pure CC. Weak or negative HNF-1β staining is encountered predominantly in NCCs.
Category: Gynecologic & Obstetrics
Wednesday, March 24, 2010 9:30 AM
Poster Session V # 126, Wednesday Morning