Purpose: We aimed to identify candidate proteins for tumor markers to predict the response to gefitinib treatment.
Experimental design: We did two-dimensional difference gel electrophoresis to create the protein expression profile of lung adenocarcinoma tissues from patients who showed a different response to gefitinib treatment. We used a support vector machine algorithm to select the proteins that best distinguished 31 responders from 16 nonresponders. The prediction performance of the selected spots was validated by an external sample set, including six responders and eight nonresponders. The results were validated using specific antibodies.
Results: We selected nine proteins that distinguish responders from nonresponders. The predictive performance of the nine proteins was validated examining an additional six responders and eight nonresponders, resulting in positive and negative predictive values of 100% (six of six) and 87.5% (seven of eight), respectively. The differential expression of one of the nine proteins, heart-type fatty acid-binding protein, was successfully validated by ELISA. We also identified 12 proteins as a signature to distinguish tumors based on their epidermal growth factor receptor gene mutation status.
Conclusions: Study of these proteins may contribute to the development of personalized therapy for lung cancer patients.