Hi all,

I am study a cell type (MDSC) that are known to suppress the proliferation of immune cells (mainly T cells). In order to test for suppression, we do what we call a 'suppression assay'. In this assay, we use 3H-thymidine incorporation to measure (a) proliferation of stimulated T cells with control cells and (b) proliferation of stimulated T cells with the MDSCs. Both 'a' and 'b' are tested in triplicate, resulting in both a mean and stdev for each. We then use these numbers (reported in CPM) to determine the percent suppression by the MDSCs (% suppression = 100% - b/a). Currently, I am working to determine if contact is necessary for suppression in the context of our system. I am doing this using a transwell system. I would like to compare the percent suppression (calculated using the CPMs) with and without the transwell insert.

However, because of the design of the insert, the 'control' wells (those with T cells and control cells, no MDSCs, represented above by the letter 'a') give different values with and without the insert. Therefore, when comparing suppression, I cannot simply compare the CPMS with MDSCs ('b'). How do I perform statistics to compare the % suppression? My instinct would be to use a T-test, but I am concerned that because the data being analyzed is a ratio of two CPMs, I may not be able to use this test. Can I transform this data so that I can use a t-test? I have read that the arc-sine transformation may work, but this seems controversial and appears to be being used less and less. If transforming is not the answer, can I use a non-parametric test (ie the Mann-Whitney U test)?

Thank you so much for your help!