Intestinal-Type Cervical Adenocarcinoma In Situ and Adenocarcinoma Exhibit a Partial Enteric Immunophenotype With Consistent Expression of CDX2

Most cases of cervical adenocarcinoma in situ (AIS) and adenocarcinoma are of the usual or endocervical type. However, intestinal types of AIS and adenocarcinoma exist. With an intestinal-type adenocarcinoma in the cervix, the question may arise as to whether one is dealing with a primary cervical neoplasm or direct or secondary spread from an intestinal adenocarcinoma. In organs such as the ovary, urinary bladder, esophagus, and gallbladder, intestinal-type glandular epithelium often expresses enteric markers, but this has hardly been studied in the cervix. The purpose of this study was to investigate whether intestinal-type AIS and adenocarcinoma in the cervix express enteric markers and to ascertain whether these antibodies are of value in the distinction from a metastatic intestinal adenocarcinoma. We compared the immunophenotype of these lesions with that of usual-type AIS and adenocarcinomain the cervix. Cases included were AIS of usual type (n = 6), primary cervical adenocarcinoma of usual type (n = 6), AIS of intestinal type (n = 21), primary cervical adenocarcinoma of intestinal type (n = 3), primary cervical adenocarcinoma with signet ring cells (n = 2), and colorectal adenocarcinoma involving the cervix (n = 5). All cases were stained with cytokeratin (CK) 7, CK20, monoclonal carcinoembryonic antigen (CEA), p16, and CDX2. Staining was categorized as negative, focally positive (<50% cells), or diffusely positive (50% or more cells). Usual-type AIS was always diffusely CK7 positive, typically diffusely CEA and p16 positive, and always CK20 negative. CDX2 was positive in 1 case. All usual cervical adenocarcinomas were diffusely CK7 and p16 positive, and all were immunoreactive with CEA. Five and 2 cases were CK20 and CDX2 positive, respectively. Intestinal-type AIS was diffusely CK7 positive (all cases) and typically CK20 negative and diffusely CEA and p16 positive. All but 1 case exhibited diffuse nuclear positivity with CDX2. In addition, usual-type AIS adjacent to intestinal type was CDX2 positive in 13 of 21 cases. The 3 cases of primary cervical intestinal-type adenocarcinoma were diffusely CK7 positive, focally or diffusely positive with CK20 and CDX2, and focally positive with CEA. One case was diffusely p16 positive, 1 focal and 1 negative. The foci of signet ring cells in the 2 primary cervical adenocarcinomas were diffusely CK7 and p16 positive and negative with CK20 and CDX2. Colorectal adenocarcinomas involving the cervix were typically diffusely positive with CK20, CEA, and CDX2; negative with CK7; and negative or focally positive with p16. Intestinal types of cervical AIS and adenocarcinoma exhibit a partial enteric immunophenotype, usually with diffuse expression of CDX2 and, in some cases, staining with CK20. They maintain their CK7 immunoreactivity and are usually p16 positive. Although there is immunophenotypic overlap, focal staining with CK20 together with diffuse CK7 and sometimes p16 positivity helps to distinguish intestinal types of cervical adenocarcinoma from involvement by a colorectal adenocarcinoma; CEA and CDX2 are of no value in this regard. CDX2 positivity in usual-type AIS adjacent to intestinal type and in occasional cases of pure usual-type AIS may be a reflection of early intestinal differentiation before this is morphologically apparent. Using a set of cases of AIS diagnosed in a single institution over a 7-year period (77 usual type; 13 intestinal type), intestinal type was more likely to be associated with early invasive adenocarcinoma than usual type (31% vs 17%), suggesting that intestinal differentiation may be a risk factor for invasion in premalignant cervical glandular lesions.