Background: We evaluated the prognostic value of pretreatment serum biomarkers in predicting outcomes in cervical cancer patients subjected to treatment.
Methods: Serum samples collected from 60 cervical cancer patients and 60 age-matched healthy individuals were used for the detection of 22 biomarkers, prior to therapy. Cox multivariate analysis and classification and regression tree analysis (CART) were performed to evaluate the prognostic factors.
Results: Cox multivariate analysis disclosed that carbohydrate antigen 153 (CA153), squamous cell carcinoma antigen (SCC) and tumor necrosis factor-α (TNF-α) are associated with prognosis in cervical cancer. CART analysis led to the stratification of patients into 3 groups: (1) serum concentrations of CA153 ≥17.60 μg/l, (2) serum concentrations of CA153 <17.60 μg/l and TNF-α ≥10.60 pg/ml, and (3) serum concentrations of CA153 <17.60 μg/l and TNF-α <10.60 pg/ml. The 2-y overall survival rates for Groups 1, 2 and 3 were 33.3%, 60.0% and 93.9%, respectively.
Conclusions: Higher serum concentrations of TNF-α, SCC and CA153 before therapy are independently associated with poor prognosis in patients with stage I and II disease. Combined usage of these three biomarkers allows efficient evaluation of outcomes in cervical cancer patients.
Keywords: Cox regression model; Kaplan–Meier curve; Prognosis; Uterine cervical neoplasm.
© 2013 Elsevier B.V. All rights reserved.