Multivariable analysis of prognostic factors for toxicity and survival for patients enrolled in phase I clinical trials

Background: Patients with advanced solid tumors may be included in phase I clinical trials. In such studies, the benefit expected is generally lower than the likelihood of toxicity and may even be non-existent if the patient's life expectancy is too short. This study was performed to identify prognostic variables for toxicity and survival in patients who participate in phase I clinical trials. Patients and methods: One hundred fifty-four patients treated on a phase I clinical trial in our institute were evaluated retrospectively. Univariable and multivariable analyses of patients' characteristics were undertaken to determine their effects on the probability of grade 3 and 4 toxicity and on survival. Results: Grade 3 or 4 toxicity was experienced by 56 patients (36%): dosage level at entry (P < 0.001) and age over 65 years (P = 0.03) were independently associated with the risk of toxicity. Median overall survival was 5 months. The multivariable analysis identified performance status 2 or 3 (P < 0.001) and lactate dehydrogenase levels greater than 600 UI (P < 0.001) as independent adverse prognostic variables for overall survival. Using these two parameters, we determined a prognostic index which allowed us to discriminate three risk groups of patients with an observed median survival of 8.5, 4.5 and 1.5 months, respectively. Conclusions: Subgroups with different survival expectancy can be identified among patients who are eligible for phase I clinical trials. If confirmed, the proposed prognostic model may be useful for therapeutic decision making in palliative oncology.