[1028] PAX8 and WT1 Are Superior to PAX2 and BRST2 in Distinguishing Mullerian Tract Tumors from Breast Carcinomas.

Michael A DiMaio, Andrew H Beck, Kelli D Montgomery, Robert B West, Teri A Longacre, Christina S Kong. Stanford University, CA

Background: Determining primary site of metastatic carcinoma remains challenging, especially in women with breast and gynecologic cancers. To determine the best panel of immunohistochemical (IHC) markers to identify primary site, we evaluated WT1, BRST2, PAX2 and PAX8 in a large series of primary and metastatic mullerian and breast carcinomas.
Design: IHC for PAX2, PAX8, WT1, and BRST2 were performed on tissue microarrays (TMA) (see table for case breakdown). The 249 uterine carcinomas in the TMA consisted of 207 endometrial and 42 endocervical carcinomas. 10 whole sections of mullerian metastases were also evaluated. TMAs were scored as negative, weak or strong. Decision trees were constructed using Rpart in "R".
Results: The best performing classifier included Pax8 and WT1, achieving an accuracy on the study set of 78%. Accuracy was not significantly improved by adding PAX2 and/or BRST2. Due to the relatively small number of ovarian non-serous cases, this classifier was not useful for discriminating these cases from the other groups but showed 94% accuracy when the analysis was limited to uterine, ovarian serous carcinomas, and breast.

IHC Expression in Mullerian and Breast Carcinomas
All uterine carcinomas (n=249)69 (27.7%)232 (93.2%)6 (2.4%)19 (7.6%)
Ovarian, serous carcinoma (n=146)44 (30.1%)138 (94.5%)134 (91.8%)4 (2.7%)
Ovarian, non-serous carcinoma (n=66)8 (12.1%)48 (72.7%)14 (21.2%)4 (6.1%)
Mullerian metastases (n=10)7 (70.0%)10 (100.0%)6 (60.0%)0 (0%)
Endometriosis (n=5)3 (60%)5 (100%)4 (80%)0 (0%)
Breast carcinoma (n=210)0 (0%)0 (0%)4 (1.9%)165 (78.6%)
Breast metastases (n=87)3 (3.4%)9 (10.3%)2 (2.3%)67 (77.0%)

Conclusions: A combination of PAX8 and WT1 is 94% accurate in differentiating breast, uterine, and ovarian serous carcinomas. However, PAX8 and WT1 negative ovarian non-serous carcinomas are misclassified as breast carcinomas by this algorithm; these cases consist of 28% mucinous, 50% endometrioid, and 22% clear cell carcinomas. Since addition of BRST-2 or PAX2 provides little additional information, identification of primary site by IHC may be more problematic for these histologic subtypes. Only strong nuclear PAX8 staining should be considered positive in the clinical diagnostic setting because weak PAX8 can be seen in 10% of breast metastases, 56% of which are also BRST2(-); weak PAX8(+)/BRST2(-)/WT1(+) phenotype may also rarely occur.
Category: Gynecologic & Obstetrics

Monday, February 28, 2011 9:30 AM

Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 142, Monday Morning


Close Window