Abstract
Delay in diagnosing oral squamous cell carcinoma (OSCC) can be still identified as a major cause of its high morbidity and mortality. To date, the early diagnosis for OSCC is mainly based on clinical oral examination and there is an urgent need for reliable markers; thus, advancements in molecular technologies has set the stage for investigating new markers, as well as new diagnostic matrices. The aim of the present study is to investigate the presence of proteomic signatures of OSCC in saliva and their use as potential biomarkers for early and non-invasive diagnosis. Saliva from 45 OSCC patients and 30 healthy controls was analysed by SELDI-TOF mass spectrometry and ProteinChip® technology. A supervised multivariate statistical analysis (Classification and Regression Tree - CART) was used to build models for discriminating between OSCC and controls, and between early (ES-OSCC) and late stage (LS-OSCC) cancers. The peptide with 8041 Da mass was 22-fold more expressed in OSCC, thus being a suitable potential biomarker. Classification and regression analysis allowed to build a model that was capable of correctly classifying all cancers and controls in an independent testing set, using the 8041 m/z peak as splitter. Eleven peaks were also differently expressed between ES-OSCCand LS-OSCC, but, basing on these differences, it was not possible to build an algorithm to predict tumour staging. These findings confirm that saliva proteome in OSCC patients is different from healthy controls and these variations might reflect different stages of disease progression and are worthy of further validation as diagnostic and prognostic biomarkers.