Identificación y validación de firmas pronósticas basadas en proteína para estadios iniciales de cáncer de pulmón no microcítico

Resumen

BACKGROUND Lung cancer remains the leading cause of cancer deaths worldwide. Nowadays, the tumour-node-metastasis (TNM) staging system is the current standard for predicting prognosis in NSCLC patients. However, TNM stage does not capture the complexity of the disease since heterogeneous clinical outcomes with identical stage are commonly observed. Thus, the identification of additional prognostic markers that improve the current staging system is mandatory. Over the last decade, mRNA microarray has been used to obtain gene expression signatures that could predict clinical outcome in NSCLC. However, the lack of reproducibility, and the expensive technology have limited the use of these signatures in the routine practice. Thus, the goal of this study was to develop a protein based prognostic signature for early stage NSCLC by using immunohistochemistry (ICQ) in formalin-fixed paraffin-embedded (FFPE) samples, an affordable technology widely available in most hospitals. HYPOTHESIS The identification of a robust protein-based signature that accurately classifies early stage NSCLC patients into high or low risk of developing post-resection recurrence will improve the prognostic performance of the TNM staging, allowing a better clinical management of these patients. GENERAL OBJECTIVE Identification of a robust protein-based signature capable of predicting outcome in early NSCLC, patients using FFPE resected tumor samples. METHODS First, we tested the antibody specificity using Western Blotting, ICQ and siRNA technology in NSCLC cell lines. FFPE samples from two different cohorts were analysed by ICQ. The prognostic discrimination was assessed by Cox proportional hazard analysis using recurrence or death as primary outcome. The prognostic model was internally validated by using bootstrapping method. We used siRNA technology to analyse the functional relevance in the proliferation of the selected genes. We developed an automatic image analysis based on the Fiji software. RESULTS The total amount of selected genes through the application of a robust algorithm was 18. The combination of bootstrapping and Cox proportional Hazard analysis allowed us to identify the more relevant proteins and define four specific prognostic algorithms using recurrence-free survival or overall survival as primary outcome in both histologic subtypes. In a subsequent multivariate analysis, the prognostic indexes derived from the four signatures were independent risk factors for the outcome. To assess the medical applicability of the models, we analyzed the benefit of combining the pathological stage with the molecular prognostic models. The combination of the molecular models and pathological stage were a better indicator of lung cancer outcome than pathological stage. Moreover, we validated the prognostic value of the molecular models and the combining models in an independent cohort of patients. Furthermore, we analyzed the clinical utility, with the goal of identifying which patients in stage IB could obtain a benefit from the adjuvant chemotherapy. In the case of ADC patients who received platinum-based chemotherapy after surgery, an association between higher prognostic index and longer overall survival was found (p=0.003). In vitro downregulation of some genes included in these signatures promoted the reduction of the proliferation ratio. Finally, we developed and validated an automatic image analysis to quantify the cytoplasmic immunostaining in NSCLC tumors. CONCLUSION. The combination of different protein-based indexes is a valuable tool for the selection of NSCLC that will take advantage of post-surgery treatment or a tight follow up.