Advances in nanotechnology are introducing the exciting possibility of cancer identification at early stages via analysis of the personalized biomolecular corona (BC), i.e. the dynamic "halo" of proteins that adsorbs onto NPs following exposure to patients' plasma. In this study, we develop a blood test for early cancer detection based on the characterization of the BC that forms around Graphene Oxide (GO) nanoflakes. Among its elective properties, GO binds low amounts of albumin, the most abundant protein in the blood and one of the most enriched proteins in the BC of many nanomaterials. This unique property of GO allows strong adsorption of poorly concentrated plasma proteins without abundant protein depletion. In our study, GO nanometric flakes have been used to analyze BCs from 50 subjects, half of them diagnosed with pancreatic cancer and half of them being healthy volunteers. Pancreatic cancer was chosen as the model of a high mortality disease with poor survival rates due to its delayed diagnosis. The receiver operating characteristic (ROC) curve analysis was applied to measure the diagnostic accuracy of the BC-based test. We obtained an area under the curve (AUC) of 0.96 and the test discriminated cancer patients from healthy subjects with a sensitivity of 92%. Finally, a double-blind validation was made using a second test dataset (10 healthy subjects + 10 pancreatic cancer patients) and it confirmed the results obtained on the first training dataset. Being highly accurate, fast, inexpensive and easy to perform, we believe that the BC-enabled blood test has the potential to become a turning point in early detection of cancer and other diseases.