PURPOSE: Prognostic factors in cancer of the cervix for patients treated with external beam irradiation (EBR) and low-dose-rate (LDR) brachytherapy have been characterized. However, despite the increasing use of high-dose-rate (HDR) intracavitary placements (ICP), few studies with adequate follow-up have analyzed prognostic factors. This study investigates pretreatment and treatment factors for their correlation with treatment outcome after EBR and HDR-ICP. METHODS AND MATERIALS: Between September 1985 and December 1994, 181 patients with carcinoma of the cervix FIGO stages IB-IV received EBR and HDR brachytherapy. Hemoglobin (Hb) levels were maintained above a level of 11 g/dl during the treatment by transfusion. Patient age ranged from 34 to 84 years (median: 66). The median follow-up time for patients at risk is 69 months (range: 23-140). Pretreatment and treatment parameters analyzed to determine their prognostic value included age, FIGO stage, tumor size, tumor type and grade, pretreatment Hb level, number of HDR-ICP, total dose from HDR-ICP, overall dose to point A, and overall treatment time. Also evaluated was the prognostic value of enlarged lymph nodes noted on pretreatment CAT scan of the abdomen and pelvis. Endpoints studied in uni- and multivariate analyses were disease-specific survival (DSS), freedom from disease (FFD), pelvic control (PC), and probability of distant metastases (DM). RESULTS: At 5 years the DSS, FFD, and PC rates for all patients were 60%, 58%, and 67%, respectively. The 5-year FFD by stage was: IB: 94%; II: 63%; IIIB: 43%; and IV: 0%. The PC rates were 94%, 66%, 59%, and 0%, respectively. In univariate analysis the prognostic factors identified for FFD were FIGO stage, tumor size, initial Hb level, and enlarged pelvic and/or paraaortic nodes (all: p < 0.0001). Age was inversely correlated with outcome (p = 0.0081). The 5-year FFD rates for tumors (< 3, > or = 3 < 6, > or = 6 cm) were 97%, 65%, and 24%; patients with initial Hb levels < or = 11g/dl had a FFD of 26% versus 69% for patients with levels > 11g/dl; and those with pelvic and/or paraaortal nodes > or = 1 cm had a survival of 32% versus 68% in patients with negative readings. The same factors were also prognostically significant for DSS, PC, and DM. Patients with persistent disease or pelvic failures had a significantly higher incidence of DM than patients in whom pelvic disease was controlled (p < 0.0001).Histological and treatment parameters including overall treatment time were not of prognostic significance for any of the endpoints studied. In multivariate analysis tumor size was the most powerful parameter for DSS, FFD, PC (p < 0.0001) and DM (p = 0.0001), followed by low initial Hb level (DSS: p = 0.0004, FFD: p = 0.0009, PC: p = 0.0012, DM: p = 0.0265), and enlarged pelvic and/or paraaortic nodes which were predictive for DSS (p = 0.0210) and DM (p = 0.0011). CONCLUSION: This study confirms that prognostic factors for patients treated with HDR brachytherapy are similar to those reported in previous series that employed LDR brachytherapy. The significance of tumor size, pretreatment Hb level, and enlarged pelvic and/or paraaortic lymph nodes on CAT scan over FIGO stage of disease were demonstrated. Future prospective trials should be undertaken to confirm the validity of these factors and to elucidate their therapeutic implications.