Statistics from Altmetric.com
Colonic adenocarcinomas composed predominantly or exclusively of cells with clear cytoplasm are extremely rare.1,2 Considerable diagnostic difficulties can arise in distinguishing primary colonic clear cell adenocarcinoma and metastatic carcinoma from sites such as ovary or kidney. Here, we describe a case of primary colonic clear cell adenocarcinoma that probably arose in endometriosis. The possible presence of endometriosis was only appreciated on review and after the examination of multiple levels and extra histological sections.
A 65 year old woman presented with crampy lower abdominal pain and the passage of blood and mucus from the rectum. Barium enema showed an apparently malignant stricture of the rectosigmoid and she underwent an anterior resection. Preoperative serum CA125 was not measured. At surgery, the clinical impression was of a primary colorectal tumour. Small haemorrhagic nodules were present on the pelvic and abdominal peritoneum, suggestive of endometriosis. There were multiple metastatic lesions within the liver. Both ovaries and kidneys appeared normal.
The surgical specimen consisted of a 30 cm length of colon. A polypoid ulcerated tumour involved the mucosa and infiltrated through the full thickness of the colonic wall.
Histology of the tumour showed an ulcerated surface. The tumour was composed entirely of cells with abundant clear cytoplasm and prominent cell membranes (fig 1A). Several growth patterns were present. Much of the tumour had a pronounced papillary pattern, with hyalinised cores covered by tumour cells (fig 1A). Tubular and solid areas were also identified. There was moderate nuclear pleomorphism and low mitotic activity, with a formal mitotic count revealing 1–2 mitoses/10 high power fields. Areas of necrosis were present and there was extensive lymphovascular permeation, both within the tumour and within submucosal and serosal lymphatics away from the tumour. Calcified psammoma bodies and intracytoplasmic periodic acid Schiff (PAS) positive eosinophilic hyaline inclusions were also present. The adjacent colonic mucosa showed no dysplastic features. The tumour infiltrated through the full thickness of the colonic wall into the surrounding fat.
Situated within the fat, on the external surface of the tumour, a cystic structure was present. This had an epithelial lining, which focally consisted of a single layer of plump cells with abundant eosinophilic cytoplasm (fig 1B). These cells merged with a single layer of cells with abundant clear cytoplasm, similar to those seen within the main tumour. Surrounding the cyst a fibrous stroma was present but no definite endometrial type stroma was identified. Histology of a liver biopsy taken at the time of laparotomy showed metastatic clear cell carcinoma.
Immunohistochemical staining showed diffuse strong positive membrane staining of tumour cells with CA125 (fig 2A) (CIS Bio International, High Wycombe, UK ). There was also diffuse strong positivity for cytokeratin 7 (CK7; Dako, Ely, UK) (fig 2B), but no staining for CK20 (Dako), which stained adjacent normal colonic mucosa. Staining for type IV collagen (Dako) and laminin (Dako) showed positivity of the hyalinised cores within the papillary areas. The cells lining the cystic structure stained strongly with Ber-EP4 (Dako).
We consider it probable that the colonic clear cell carcinoma in this case arose in endometriosis. This is based on the presence of a cystic structure at the deep aspect of the tumour, which was lined by cells with eosinophilic cytoplasm. Although endometrial type stroma was not identified, the morphological findings are similar to those that can be seen in long standing endometriosis. In addition, at laparotomy, there was a clinical impression of endometriosis surrounding the tumour with multiple small haemorrhagic pelvic and abdominal peritoneal nodules. The possible importance of this cystic structure was only appreciated after review of the case and examination of multiple levels and extra histological sections. A possible transition was seen within the epithelial lining of the cyst from cells with eosinophilic cytoplasm, suggestive of endometriosis, to cells with abundant clear cytoplasm, similar to those seen within the main tumour. One of us (WGM) has previously observed similar features in ovarian clear cell carcinoma arising in endometriosis. It was thought possible that the cystic structure could have been a mesothelial lined cyst, but this was excluded by strong positivity of the lining cells for Ber-EP4.
Malignant transformation in endometriosis was first described by Sampson in 1925,3 who recommended that three criteria be met for a definitive diagnosis, namely: (1) there should be histological evidence of endometriosis in close proximity to the tumour; (2) no other primary site of malignancy should be identified; and (3) the histological appearance of the tumour should be compatible with an origin in endometriosis. In our patient, only the second and third of these criteria were fully satisfied. However, these criteria are restrictive because in many cases the tumour may completely obliterate pre-existing endometriosis, making it impossible to confirm its presence unequivocally. Tumours that can arise in endometriosis include endometrioid adenocarcinoma, clear cell carcinoma, squamous carcinoma, endometrioid stromal sarcoma, adenosarcoma, and carcinosarcoma.4,5
Clear cell adenocarcinoma of the ovary is associated with pelvic endometriosis in 50–70% of cases and a quarter of ovarian clear cell carcinomas can be shown to arise in endometriotic cysts. It should therefore be no surprise if occasionally a clear cell carcinoma of ovarian type should arise in extraovarian endometriosis, and several such cases have been reported.6–8 Endometrioid type adenocarcinoma has occasionally been described arising in colonic endometriosis,9 and we are aware of a single previous report of clear cell carcinoma arising in endometriosis of the sigmoid colon.8
In our patient, the strong positivity of tumour cells with CA125 provides evidence of Mullerian derivation. Although focal immunoreactivity can be present in primary colonic carcinoma, positivity to this extent is unusual. The strong immunoreactivity for CK7, combined with CK20 negativity, is also in keeping with an ovarian type primary, the converse pattern of staining being expected in a primary colonic neoplasm.10 A further histological pointer to an ovarian type tumour was the presence of calcified psammoma bodies. The ovaries and kidneys appeared grossly normal at laparotomy, helping to exclude the possibility of a colonic metastasis from an ovarian or renal primary.
In summary, we describe an unusual case of primary colonic clear cell adenocarcinoma that has probably arisen in extraovarian endometriosis. When confronted with an extraovarian tumour with the histological appearances described above, pathologists should consider a primary of ovarian type and an origin in endometriosis. The demonstration of endometriosis might require the examination of multiple levels and extra histological sections. Even then, residual endometriosis might not be definitely demonstrated because it may be completely obliterated by tumour. Confirmation that a tumour is of ovarian type is of clinical importance, because chemotherapeutic regimens will differ from those administered for a typical colonic adenocarcinoma.