Re: interpretation and validity of small populations of cells

Rick,

there is no "general cut off".	Many people have wondered what to  
interpret as "real"... but the bottom line is that every single event  
is "real"; the only question is whether it is biologically  
meaningful.  To assess that, one must do the same thing as we do for  
every other kind of experiment -- to determine the range of values  
that are measured experimentally.

Thus, if you have a particular staining panel, and in a particular  
gate you find zero cells in all 50 controls, then finding even one  
cell in that gate for your experimental sample becomes "real" and  
biologically meaningful.  On the other hand, if you find an average of	
1000 events in this particular gate in all controls, and you find 950  
events in your experimental sample, then the 950 events are still  
real, but they don't comprise a difference from the control.

So there is no absolute threshold that is meaningful; it is different  
for each experiment.  If you have 0.05% in your gate, whether it's 2  
cells or 2000 cells, I would want to see the same backup data to  
demonstrate that this is "biologically meaningful".

Of course... the PRECISION of a measurement will be far different  
depending on how many events are counted.  If you count only 2 events,	
then the precision is about +/- 70% (square root of 2 divided by 2).   
If you count 1000 events, the precision is about +/- 3% (square root  
of 1000 divided by 1000).  So when you have few events, that doesn't  
mean they aren't believable, it just means that you don't have as good	
a handle on the true fraction that they represent as you would with  
many more events.  Note that the precision for a rare populations is  
ONLY dependent on the number of events in that population that were  
counted -- i.e., the precision on 100 events is the same whether the  
overall sample had 100,000 events (0.1%) or 100 million events	
(0.0001%)!

mr

On Dec 12, 2007, at 4:10 PM, <Rick.Schlichtemeier@UCHSC.edu>
<Rick.Schlichtemeier@UCHSC.edu 
 > wrote:

> Hi,
> I was wondering if anyone had any information as to how to interpret	
> low numbers of cells of interest.
> Is there a general cut off and number of cells that is generally  
> accepted?  I have heard that it is okay
> to go down to 0.05  as long as 50 cells are counted.	But I wanted  
> to be sure about this.
> Thanks very much
> Rick Schlichtemeier
> Research assistant
> Univ of Colorado Health Sciences Center
>