Reverse phase protein array (RPPA) combined with computational analysis to unravel relevant prognostic factors in non- small cell lung cancer (NSCLC): a pilot study

Abstract

In this work high throughput technology and computational analysis were used to study two stage IV lung adenocarcinoma patients treated with standard chemotherapy with markedly different survival (128 months vs 6 months, respectively) and whose tumor samples exhibit a dissimilar protein activation pattern of the signal transduction. Tumor samples of the two patients were subjected to Reverse Phase Protein Microarray (RPPA) analysis to explore the expression/activation levels of 51 signaling proteins. We selected the most divergent proteins based on the ratio of their RPPA values in the two patients with short (s-OS) and long (l-OS) overall survival (OS) and tested them against a EGFR-IGF1R mathematical model. The model with RPPA data showed that the activation levels of 19 proteins were different in the two patients. The four proteins that most distinguished the two patients were BADS155/136 and c-KITY703/719 having a higher activation level in the patient with short survival and p70S6S371/T389 and b-RAFS445 that had a lower activation level in the s-OS patient. The final model describes the interactions between the MAPK and PI3K-mTOR pathways, including 21 nodes. According to our model mTOR and ERK activation levels were predicted to be lower in the s-OS patient than the l-OS patient, while the AMPK activation level was higher in the s-OS patient. Moreover, KRAS activation was predicted to be higher in the l-OS KRAS-mutated patient. In accordance with their different biological properties, the Moment Independent Robustness Indicator in s-OS and l-OS predicted the interaction of MAPK and mTOR and the crosstalk AKT with p90RSK as candidates to be prognostic factors and drug targets.