Uterine Involvement by Colorectal Adenocarcinoma: Morphologic and Immunohistochemical Patterns That May Mimic Primary Endometrial, Endocervical or Vaginal Adenocarcinoma
A Naujokas, JT Rabban. UCSF, San Francisco, CA
Background: Although spread of colorectal adenocarcinoma to the ovaries is well recognized, spread to the uterus/vagina is rare and may potentially be misclassified as primary gynecologic in origin; this may also lead to under-recognition of Lynch syndrome. This study reports the morphology and immunohistochemistry of colorectal cancer involving the uterus/vagina, emphasizing features that may be confused with those of primary endometrial, endocervical or vaginal adenocarcinoma.
Design: Cases of colorectal adenocarcinoma involving the uterus/vagina were identified from our institutional cancer registry (1985 to 2009). Clinical history was obtained from electronic records; slides were reviewed and confirmatory immunostaining was performed with CK7, CK20, CDX2, estrogen receptor, and p16. MLH1, PMS2, MSH2 and MSH6 staining was performed in a subset. Tumor growth pattern was classified by whether the classic features of metastatic colon cancer were present: dirty necrosis, segmental necrosis and garlanding. Tumor was also classified by whether the growth pattern simulated a primary endometrial, endocervical or vaginal adenocarcinoma.
Results: Among 1,734 women with colon cancer, 15 (age 28-75 years) had pathologically diagnosed involvement of uterine body, uterine cervix or vagina. History of primary colon origin was known in 14/15 prior to undergoing uterine surgery and 12/15 had prior chemotherapy and or radiotherapy. Tumor distribution included: endometrium(6), myometrium(6), endocervix(7), and/or vagina(3). Tumor type was conventional adenocarcinoma(11), pure mucinous adenocarcinoma(4) or a mixed(1). Histology revealed low grade architecture in 12/15, low grade cytology in 7/15, goblet cells in 4/15, dirty necrosis in 10/15, segmental necrosis in 10/15, garlanding in 6/15; none had squamous differentiation. All were CDX2 positive but 9/11 were p16 positive (7 strong/diffuse, 2 weak/patchy); 11/12 were CK20+CK7-. H&E morphology was clearly colonic in origin in only 5/15; the remainder had growth simulating primary uterine/vaginal cancer. Mismatch repair was defective in 2/5 cases. One patient had Lynch syndrome.
Conclusions: Uterine/vaginal involvement by colorectal cancer may mimic primary gynecologic cancer for several reasons: 1.) classic colonic morphology is not always present 2.) p16 expression may be present 3.) goblet cells may mimic intestinal type endocervical cancer and 4.) growth patterns may resemble primary endometrial, endocervical or vaginal cancer.
Category: Gynecologic & Obstetrics
Wednesday, March 24, 2010 9:30 AM
Poster Session V # 145, Wednesday Morning