Discovery and validation of protein biomarkers for lung cancer

Abstract

Lung cancer is a disease characterized by high incidence and mortality rates as well as by a remarkable resistance to common treatment strategies. Reliable and validated biomarkers could represent a possible breakthrough in the management of this tumor type, by facilitating diagnosis, refining prognosis and providing guidance towards the choice of the most appropriate therapy.

In this thesis we approach the field of lung cancer biomarkers using proteomics technologies on tumor and plasma samples from a large local institutional biobank. In paper I a protocol to prepare enriched tumor cell suspensions from freshly collected surgical samples is developed. Eight cases are analyzed using a bottom-up proteomics approach. Compared with the direct lysis of fresh frozen specimens, with this sample preparation we were able to identify more than twice as many proteins. In addition, we could demonstrate an effective removal of contaminants deriving from plasma, stroma and red blood cells.

In paper II we analyze with a top-down proteomics approach tumor lysates from 39 lung cancer cases. We could observe a characteristic 3-peak profile which corresponded to the post-translational modification pattern of the calcium-binding protein S100A6. We performed an early clinical validation of the expression of this protein on 103 lung cancer samples using immunohistochemistry on specimens loaded on a tissue microarray. In cases with a negative p53 expression, the positive tumor expression of S100A6 was associated with a significant survival improvement, confirming on clinical material the pro-apoptotic role proposed for S100A6 by in vitro investigations and suggesting a functional connection with the p53 pathway.

An extensive validation of the findings obtained in paper II is performed in paper III. The expression of S100A6 and other proteins from the same protein family, namely S100A2, S100A4, S100A7, S100A8, S100A9, S100A10 and S100A11, was assessed on a large and well characterized tumor cohort consisting of 494 stage I lung cancer cases. S100A2 and S100A6 were shown to be positive prognostic factors of progression-free survival in p53- negative adenocarcinomas, whereas S100A10 was a negative prognostic factor of progression-free survival in p53-positive squamous-cell tumors.

Finally, in paper IV we evaluate the diagnostic and prognostic role of plasma level of two forms of cytokeratin 18, an indicator of cell death, in comparison with the well established biomarker Cyfra 21.1 (cytokeratin 19 fragment). The study was conducted on 179 lung cancer cases, 113 subjects with benign lung diseases and 200 healthy blood donors. Cyfra 21.1 was a more reliable diagnostic biomarker than cytokeratin 18 (diagnostic accuracy of 94% vs 56%, respectively). However, in the total lung cancer cohort and in a subset of 78 patients with stage III-IV lung cancer receiving either combination chemoradiotherapy or 1st-line palliative chemotherapy, cytokeratin 18 was the only biomarker which retained an independent prognostic potential on multivariate survival analysis.

In conclusions, the works presented in this thesis explore possibilities and limitations of biomarker research on lung cancer, highlighting the importance of the quality of the samples and of the accuracy and completeness of clinical data in order to obtain reproducible results.