Endometrial cancer is the most common gynecological cancer worldwide. In women from developed countries, endometrial cancer is usually detected after the age of 60 years. Nearly 90% of the diagnosed cases are sporadic, while 10% are attributed to genetic factors. Thus, it is necessary to develop a diagnostic platform. Currently, the lymph node status is a major prognostic factor in assessing endometrial cancer. However, despite its high sensitivity, it is still in debate. In order to increase its selectivity and specificity, in-depth proteomic study has been performed on patient’s sentinel lymph node (SNL) samples and on grade I to grade III endometrial carcinoma (EC) samples. We corrrlated with TcGA data provided by the pathological Atlas on endometrial cancer. We identified specific markers of bad diagnosis that clearly correlate SNL to EC i.e. PRSS3, PTX3, ASS1, YBX2, RBM25, MUC5B. These markers have also been detected in TCGA and validated by IHC using TMA. We also detected good prognosis markers for the overall survival such as CD74, GOLM1, SLC9A3R2, SP100 proteins. In order to establish a robust classification for diagnosis and prognosis, we also identify specific mutated proteins and ghost proteins issued from non-coding region of mRNA or from Non-coding RNA. Thus, it is now possible to stratify patients based on those markers. Markers for EC and SNL cancer-grading have also been identified and validated. Taken together, it is now possible to propose a novel approach for molecular diagnosis, prognosis and patient’s stratification in EC which will be of excellent utility in therapeutic decision.