Methods are provided for the detection of and determining prognosis of metastatic breast,
lung,
prostate, and / or
pancreatic cancer using various genetic markers, including markers for
gene clusters linked by Esx. In one method,
breast cancer micrometastases and non-
small cell lung cancer metastases or micrometastases are detected in a patient by determining whether the AGR2 or TFF1 genes are overexpressed in a
cell sample compared to control
lymph node tissue cells. In a further method, the likelihood that a patient diagnosed with
breast cancer will respond to hormonal therapy is predicted by determining a higher expression level of the AGR2
gene compared to a control
gene. In a further method, a decreased probability of survival for a patient diagnosed with early stage non-
small cell lung cancer is predicted by determining a higher expression level of the AGR2 gene compared to a control gene. Kits for practicing the methods of the invention are further provided. Methods are also provided for the identification of markers for which overexpression is indicative of the presence of micrometastatic
disease.