A survival analysis using physique-adjusted tumor size of non-small cell lung cancer.

Abstract

Differences in individual body sizes have not been well considered when analyzing the survival of patients with non-small cell lung cancer (NSCLC). We hypothesized that physique-adjusted tumor size is superior to actual tumor size in predicting the prognosis. Eight hundred and forty-two patients who underwent R0 resection of NSCLC between 2005 and 2012 were retrospectively reviewed, and overall survival (OS) was evaluated. The physique-adjusted tumor size was defined as: x-adjusted tumor size=tumor size×mean value of x/individual value of x [x=height, weight, body surface area (BSA), or body mass index (BMI)]. Tumor size category was defined as≤2, 2-3, 3-5, 5-7, and >7cm. The separation index (SEP), which is the weighted mean of the absolute value of estimated regression coefficients over the subgroups with respect to a reference group, was used to measure the separation of subgroups. The mean values of height, weight, BSA, and BMI were 160.7cm, 57.6kg, 1.59m2, and 22.2kg/m2, respectively. The 5-year survival rates ranged from 88-59% in the non-adjusted tumor size model (SEP 1.937), from 90-57% in the height-adjusted model (SEP 2.236), from 91-52% in the weight-adjusted model (SEP 2.146), from 90-56% in the BSA-adjusted model (SEP 2.077), and from 91-51% in the BMI-adjusted model (SEP 2.169). The physique-adjusted tumor size can separate the survival better than the actual tumor size.