Tumour proliferation and apoptosis in human uterine cervix carcinoma II: Correlations with clinical outcome

Purpose: The prognostic value of tumour proliferation and apoptosis measurements were studied prospectively in patients with carcinoma of the uterine cervix, relative to other established clinical factors. Materials and methods: The labelling index (LI) for bromodeoxyuridine was determined by flow cytometry (fc) and also by immunohistochemistry. Apoptosis was assessed histologically using morphological criteria. Patients were treated with radical radiation therapy (RT). Results: The median/mean LT-fc were 6.7%/7.9% (range 1.52-3.9%). The median/mean apoptosis index (AI) were 1.0%/1.6% (range 0-6.8 %). To date, 27 patients have died of disease, and the median follow-up for alive patients is 3.2 years (range 0.4-6.0 years). Among 64 patients who completely responded to treatment, 25 patients have relapsed (six pelvic, 17 distant and two pelvic and distant). In univariate analysis, the most significant factors for disease-free survival (DFS) were large tumour size (P = 0.0001), low haemoglobin (P = 0.01), LI-fc (DFS 67% for LI < 7%, 33% for LI greater than or equal to 7%, P = 0.03), and T-pot (DFS 66% for T-pot > 5 days, 35% for T-pot less than or equal to 5 days, P = 0.04) Stage, overall treatment time (OTT), S-phase fraction, ploidy, T-s, LI by histology, mitotic index, and AI were not significant. Multivariate analysis (Cox's model) showed that the only significant prognostic factors for DFS were tumour size and OTT. However, for small tumours (diameter < 6 cm), either a high LI-fc ( greater than or equal to 7%) or a high AI ( > 1%) was associated with poorer DFS. whereas patients with larger tumours (diameter greater than or equal to 6 cm) fared poorly regardless of LI-fc and AI. Conclusions: Tumour size was the most important prognostic factor in cervix carcinoma. Although none of the biologic parameters have independent prognostic significance when the effect of initial tumour size was taken into account, our data suggests that LI and AI may be useful in discriminating outcome for patients with smaller tumours when managed by radical RT. These findings support the hypothesis that rapidly proliferating tumours are less likely to be controlled with a conventional course of RT. (C) 1999 Elsevier Science Ireland Ltd. All rights reserved.