A simple risk model to predict survival in patients with carcinoma of unknown primary origin

Carcinoma of unknown primary origin (CUP) is characterized by diverse histological subtypes and clinical presentations, ranging from clinically indolent to frankly aggressive behaviors. This study aimed to identify prognostic factors of CUP and to develop a simple risk model to predict survival in a cohort of Asian patients. We retrospectively reviewed 190 patients diagnosed with CUP between 2007 and 2012 at a single medical center in Taiwan. The clinicopathological parameters and outcomes of our cohort were analyzed. A risk model was developed using multivariate logistic regression and a prognostic score was generated. The prognostic score was calculated based on 3 independent prognostic variables: the Eastern Cooperative Oncology Group (ECOG) scale (0 points if the score was 1, 2 points if it was 2-4), visceral organ involvement (0 points if no involvement, 1 point if involved), and the neutrophil-to-lymphocyte ratio (0 points if ≤3, 1 point if >3). Patients were stratified into good (score 0), intermediate (score 1-2), and poor (score 3-4) prognostic groups based on the risk model. The median survival (95% confidence interval) was 1086 days (500-1617, n=42), 305 days (237-372, n=75), and 64 days (44-84, n=73) for the good, intermediate, and poor prognostic groups, respectively. The c-statistics using the risk model and ECOG scale for the outcome of 1-year mortality were 0.80 and 0.70 (P=0.038), respectively. In this study, we developed a simple risk model that accurately predicted survival in patients with CUP. This scoring system may be used to help patients and clinicians determine appropriate treatments.