Deadly tumor-initiating cells (TIC) seed metastases throughout the body and cause relapses in patients. Whether these tumor-initiating cells can also be referred to as stem cells, specifically, cancer stem cells, has been up for debate. 

"Our research establishes for the first time the relationship between the normal stem cell program and the cancer stem cell program, albeit in the context of the mammary gland," says the lead author. Tumor-initiating cells are in fact cancer stem cells, but cancer stem cells do not arise from normal stem cells." 
Cancer stem cells emerge after undergoing an epithelial-to-mesenchymal transition (EMT), which endows the cells with the motility and flexibility required for seeding new tumors.

In the Nature paper, authors used a mouse model to show which cells within the normal and cancerous mammary gland express the related master regulators Snail and Slug, both of which confer stem-like traits on mammary cells.

Slug, which regulates gland-reconstituting activity in breast tissue, is expressed at higher levels in normal stem cells found in the 'basal layer" of mammary ducts. Snail, a factor first discovered in the context fruit fly embryonic development, is expressed by tumor-initiating cells in the luminal layer of cells in these ducts. Snail has the power to confer aggressive traits on cancer cells that Slug is incapable of doing when it is expressed at normal levels.

"Normal stem cells reside in one layer in the mammary duct; cancer stem cells arise in another. What that means is that cancer stem cells do not arise from normal stem cells. This has been a point of much discussion, but now there is evidence--finally!" says the lead author.

http://wi.mit.edu/news/archive/2015/variations-cell-programs-control-cancer-and-normal-stem-cells