The CDs and their combinations help to enrich for CSCs, but DO NOT identify them; i.e., most of the selected cells are not CSCs, but a few may be. It is critical to recognize that when cells are injected in a suspension, the number of cell injected follow a Poisson distribution. So, when a single CSC is injected, 37% will recieve no cell and hence develop no cancer, while 63% will receive greater than or equal to 1 CSC and so develop a cancer. Thus, when breast cancer cells were sorted by flowcytometery (or magnetic bead separations) 1 - 2% of the initial tumor cells were CD24 neg CD44 pos and CD326 (ESA or EpCam) positive. However, injection of 100 such cells into the "humanized" breast fat pads of NOD/SCID (now would use NSG) mice, yielding about 63% of mice getting the tumor, showed that only 1% of the cells in this 2% of the tumor were CSCs, i.e., 1-2 per 10,000 of the original tumor cells were CSCs.
The point is that a functional assay is needed to measure the number of CSCs. One such option is our patented Hybrid Spheroid Assay.
I agree with Dr. Lange, you need to demonstrate functionality. As for marker selection, we have had very good correlation when using SmartFlare to sort Nanog/Sox2 positive cells from the fresh tumor. See that paper in my profile. But you still need to show function!