Abstract
BACKGROUND: The survival of non-small cell lung cancer (NSCLC) patients who receive potentially curative surgical treatment is hampered in part because of the lack of effective strategies for selecting cases with fatal prognosis.
METHODS: Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry (MALDI MS) was here used to analyze protein profiles of more than lung (NL) specimens. MS peaks associated with relapse-related death were selected using three distinct statistical methods, and used to build an individualized weighted-voting–based prediction classifier of prognosis in a training cohort. We further applied multiple proteomic approaches to identify a set of discriminatory proteins that might be associated with clinical behavior of NSCLC.
RESULTS: We based our bioinformatic analyses upon more than 2000 MS signals obtained from frozen lung specimens in a training cohort. The resulting individualized classifier based on the expression profile could accurately predict both relapse-free and overall survival in the training cohort, and, more importantly, its utility could be validated in an independent blinded test cohort.
CONCLUSIONS: We have discovered a relapse-related death signature that may distinguish NSCLC patients with a high risk of relapse and death from those with a favorable prognosis, judging from our findings with both training and independent test cohorts. The present individualized prediction classifier should have an impact on choice of optimal clinical management of NSCLC.
