Glassy cell features (GCF) were identified as composing a predominant pattern (more than 85% of histology) in six cases and a focal pattern (33-85% of histology) in 10 cases of a series of 53 adenocarcinomas (AC) and adenosquamous carcinomas (ADSQ) of the uterine cervix. In three cases examined ultrastructurally, GCFs correlated with many cytoplasmic polyribosomes and abundant rough endoplasmic reticulum, but Golgi complexes and tonofilaments were scant and intracytoplasmic lumina were absent. Intracellular mucin was present in the areas showing GCFs of four ADSQs with predominant GCFs and six ADSQs with focal GCFs. Two of three cases examined ultrastructurally showed intracellular electron-dense material that corresponded to mucin secretory material. Immunohistochemical studies of the six ADSQs with predominant GCF cases showed the following pattern of reactivity: monoclonal carcinoembryonic antigen (CEA), 2 of 6 cases; polyclonal CEA, 3 of 6; CA 125, 0 of 6; CA 19-9, 0 of 6; placental alkaline phosphatase, 0 of 6; and vimentin, 1 of 6. Focal GCF areas showed monoclonal CEA, 4 of 9 cases; polyclonal CEA, 3 of 9; vimentin, 4 of 9; while CA 125, CA 19-9, and placental alkaline phosphatase were negative in areas of GCFs. One of three patients with ADSQ with predominant GCFs and five of nine patients with ADSQ with focal GCFs with at least 1 year of follow-up were disease free. No association between GCFs (combined focal and predominant) and recurrent disease was present when compared to the other 29 AC and ADSQ patients with follow-up. Recurrent disease in our series of AC and ADSQ was only associated with stage III or IV disease at presentation (P less than 0.001). There was no association with adenosquamous histology, histologic grade, lymphatic invasion, or age. There were insufficient cases of ADSQ with predominant GCFs with follow-up to evaluate fully prognostic significance of this subgroup. Our study suggests that GCFs are part of the spectrum of differentiation in ADSQ of the cervix rather than a distinct histologic type of carcinoma with unique clinical significance.

Eighteen uterine adenosquamous carcinomas that showed focal glassy cell features (33% to 85% of tumor histology) or predominant glassy cell features (greater than 85% of tumor histology) were studied by in situ hybridization for human papillomavirus (HPV). Viral DNA was present in neoplastic cells in five cases: type 18 in four cases (two cervical adenosquamous carcinomas with predominant glassy cell features, two cervical adenosquamous carcinomas with focal glassy cell features) and type 16 in one case (cervical adenosquamous carcinoma with predominant glassy cell features). Positive intranuclear staining for HPV DNA was present within areas of squamous and glandular differentiation and within areas with glassy cell features. The mean age of HPV(+) patients was less than HPV(-) patients (mean, 57 years, compared to 67 years). No significant association between HPV status and prognosis or glassy cell features was detected. Human papillomavirus types 16 and 18 are associated with adenosquamous carcinoma with predominant glassy cell features or focal glassy cell features, "glassy cell carcinoma." Automated colorimetric in situ hybridization is an effective method to detect HPV DNA.

A series of 53 carcinomas of the uterine cervix with a component of glandular differentiation were identified and included 29 pure adenocarcinomas and 24 adenosquamous carcinomas. Cervical adenosquamous carcinomas were defined as glandular carcinomas mixed with a squamous carcinoma component. Cervical pure adenocarcinomas were classified into various Mullerian subtypes analogous to other portions of the female genital tract yielding 14 mucinous/endocervical, 11 serous, 2 clear cell, and 2 endometrioid adenocarcinomas. A panel of immunostains including monoclonal carcinoembryonic antigen (CEA-M), polyclonal carcinoembryonic antigen (CEA-P), CA 125, CA 19-9, placental alkaline phosphatase, and vimentin showed no association with histologic differentiation except for mucinous/endocervical subtype (7 of 11 CEA-M or CEA-P positive and 7 of 11 CA 19-9 positive). Recurrent disease in adenocarcinoma and adenosquamous carcinoma was associated with stage III or IV disease at presentation (P < .001), serous histology (P < .05), absence of strong CEA-M staining (P < .025), absence of strong CEA-P staining (P < 05), and presence of vimentin staining (P < .05). No association was found between survival and other histologic subtypes of adenocarcinoma (mucinous/endocervical, endometrioid, or clear cell), adenosquamous carcinoma, histologic grade, lymphatic invasion, age, or immunohistochemical staining for CA 125, CA 19-9, or placental alkaline phosphatase. When only stage I and II disease was considered, there was no correlation between histology or immunohistochemistry and outcome. Int J Surg Pathol 1(3): [181][182][183][184][185][186][187][188][189][190] 1994

This study correlated the histologic and immunohistochemical features of cervical and endometrial glandular carcinomas (adenosquamous carcinoma [ADENSQ] and adenocarcinoma [AC] ) with clinical outcome. A series of 87 uterine glandular carcinomas (53 cervical, 33 endometrial, and 1 arising in both cervix and endometrium) were histologically classified into mullerian subtypes: 28 ADENSQ, 19 serous AC, 19 mucinous AC, 15 endometrioid AC, and 6 clear cell AC. Utilizing both nuclear and architectural features, 66 glandular carcinomas were high histologic grade (3) and 21 were low histologic grade (1 or 2). Immunohistochemical studies performed on 83 of the cases showed: 33 + for monoclonal carcinoembryonic antigen (CEA-M); 38 + for polyclonal CEA (CEA-P); 26 + for placental alkaline phosphatase; 18 + for CA 125; 29 + for CA 19-9; 24 + for vimentin; 60 + for cytokeratin CAM 5.2; and 81 + for cytokeratin AE 1 : 3. The following significant correlations were identified. ADENSQ histology was associated with CEA-M staining (P < .025), and mucinous histology was associated with CA 19-9 staining (P < .025). Cervical primary site was associated with ADENSQ histology (P < .001) and staining with CEA-M (P < .025) and CEA-P (P < .05). Endometrial primary site was associated with endometrioid histology (P < .001). Forty-five patients had recurrent disease, 30 patients were disease-free for more than 1 year, and 12 patients had insufficient follow-up evaluation. Recurrent disease was associated with stage III or IV tumors (P < .001), grade 3 histology (P < .001), serous differentiation (P < .001), invasion to at least the middle third of the myometrium (P < .001) and large size of residual tumor at hysterectomy (mean 3.9 cm versus 1.3 cm, P < .005). Disease-free survival was associated with endometrioid differentiation (P < .05), strong CEA-M staining (P < .001), CEA-P staining (P < .025), and CA 19-9 staining (P < .05). Considering only stage 1 and 2 patients, grade 3 histology ( P < .025), deep myometrial invasion (P < .01), and size (P < .05) were still associated with recurrence and strong CEA-M staining (P < .025) was still associated with disease-free survival. However, strong CEA-M staining, deep myometrial invasion, and size of tumor after hysterectomy were all associated with histologic grade. Considering just histologic grade 3 carcinomas in stage 1 and 2 patients, absence of strong CEA-M staining, deep myometrial invasion, and size of tumor was no longer associated with recurrent disease. Histologic grade was the only independent predictor of prognosis in stage I and II patients. Int J Surg Pathol 1 (1): [13][14][15][16][17][18][19][20][21][22][23][24] 1993