Cervical carcinoma with glandular differentiation: histological evaluation predicts disease recurrence in clinical...

Pathologists confront questions concerning the clinical implications of the more complex, evolving histopathologic classification in cervical carcinoma with glandular differentiation (CCGD) and the associated precursor intraepithelial lesions. Pseudoneoplastic pitfalls, such as microglandular hyperplasia, constitute the subject of recent reports, but the extent of misinterpretation for CCGD is unknown. To address these issues, we retrospectively reviewed all the histopathologic material for 67 patients treated for early clinical stage (I or II) CCGD. Two patients (3%) had pseudoneoplastic glandular lesions (two microglandular hyperplasias). The remaining 65 CCGDs included 35 pure adenocarcinomas (18 mucinous, six serous, five endometrioid, five clear cell, and one adenoid cystic), 26 adenosquamous carcinomas (17 showed > or = 50% and nine showed > 10% but < 50% squamous differentiation-all nonkeratinizing; four were predominantly glassy cell type, and the others showed the following adenocarcinoma component differentiation: 11 mucinous, eight serous, and three endometrioid) and four villoglandular papillary adenocarcinomas (all four were mucinous). In situ carcinoma was identified in 54%. The two patients with pseudoneoplastic lesions were disease free (after 96 and 108 months). Twenty-one patients with CCGD had recurrent disease at 4 to 144 months (mean, 45; median, 18) including three local recurrences, 10 with distant metastasis, and eight with both. Thirty-five patients with CCGD were disease free at 12 to 216 months follow-up (mean, 80.6; median, 65). Adenosquamous (P < .0002, predictive value [PV] = .68) and serous differentiation (P < .05, PV = .61) were the only histological types associated with disease recurrence. Vascular space invasion (P < .0002, PV = .7), deeper invasion (P < .0005), nuclear grade (P = .002, PV = .51), larger tumors on clinical exam (P < .01) or pathological evaluation (P < .01), and presence of pelvic lymph node metastasis at surgery (P < .05, PV = .7) are additional features associated with recurrent disease. A combination of adenosquamous or serous differentiation and vascular space invasion maximized PV for recurrent disease at a level of .75. Mucinous, endometrioid, or clear cell histological types, architectural grade, or the distinction between clinical stages I and II were not associated with recurrent disease. None of the four patients with villoglandular papillary adenocarcinoma exhibited recurrent disease, but confirmation of this histological subtype's prognostic value was hindered by the small number of cases identified (P = .16). Adenosquamous and serous differentiation, nuclear grading, pathological evaluation of vascular space and lymph node involvement, and recognition of pseudoneoplastic glandular lesions helped predict recurrent disease in low clinical stage CCGD in this retrospective study.