Despite constant improvement in existing therapeutic efforts, the overall survival rate of lung cancer patients remains low. Enzyme activities may identify new therapeutically targetable biomarkers and overcome the marked lack of correlation between cellular abundance of translated proteins and corresponding mRNA expression levels. We analysed tyrosine kinase activities to classify lung adenocarcinoma (LuAdCa) resection specimens based on their underlying changes in cellular processes and pathways that are agents of or result from malignant transformation. We characterised 71 same-patient pairs of early-stage LuAdCa and non-neoplastic LuAdCa resection specimen lysates in the presence or absence of a tyrosine kinase inhibitor. We performed ex vivo multiplex tyrosine phosphorylation assays using 144 selected microarrayed kinase substrates. The obtained 76 selected phosphotyrosine signature peptides were subsequently analysed in terms of follow-up treatments and outcomes recorded in the patient files. For tumour, node, metastasis (TNM) stage 1 LuAdCa patients, we noticed a larger tyrosine kinase inhibitor-induced decrease in tyrosine phosphorylation for long-term as opposed to short-term disease survivors, for which 26 of 76 selected peptides were significantly (p < 0.01, FDR < 3%) more inhibited in the long-term survivors. Using statistical class prediction analysis, we obtained a 'prognostic-signature' for long- versus short-term disease survivors and correctly predicted the survival status of 73% of our patients. Our translational approach may assist clinical disease management after surgical resection and may help to direct patients for an optimal treatment strategy.