Below is my views on "Mean Fluorescence Index" (MFI), a question raised by
Dr. Dawson, and I wish to hear more comments.
As the way MFI is calculated, it means fluorescence intensity of a given
sub-population normalized against the cell number (percentage, frequency).
In other words, this index is supposed to tell us QUANTITATIVELY how much
the given sub-population contributes to expression of the marker within the
whole cell population flowing through the laser focus. The most important
underlying assumption is that: (1) fluorescence intensity of a single cell
is a function of the antigen (marker) quantity in the single cell, I = Func
[Qa] (I: Intensity, Func: Function, Qa: Antigen quantity of a single cell);
(2) this function MUST stay the same from experiment to experiment.
The second assumption is not always true. Just name a few factors that can
alter this Func.
First of all, as the cytometer is recording the channel number, instead of
absolute fluorescence intensity for a single cell, instrument settings such
as voltage and amplification do matter significantly (I can artificially
make a "negative" cell appear in 3rd or even 4th log). Therefore, in order
to compare data involving fluorescence intensity from day to day, it's
better to leave the instrument settings unaltered. Second, for the fact of
channel-recording cytometers, the last channel cannot be included in any
intensity or MFI calculation, because signals beyond that channel will be
recorded as if they fall into that channel. An example, in calculation of
MFI, one cell in the last channel of a 4-log amplifier will equal to no less
than 1,000 cells within first log. Third, as fluorescence signals come from
conjugated antibodies, the F/P ratio of an antibody does matter. F/P ratio
is a lot-specific parameter, different from company to company and from vial
to vial, and thus it is very hard to find another vial of antibody with the
same F/P ratio as the one you just used up. Just an example, FITC-labeled Ab
can have F/P ratio from 3 to 7. That is, you can have two-fold difference in
your intensity and MFI data out of the same cell, just because you opened up
a new vial of Ab. Fourth, as antibodies can be associated with each other in
many ways, one antigen/epitope in a single cell might be linked with many Ab
molecules. This seems particular true in intracellular staining. It is
obvious if overdose Ab is allowed to interact with Ag, we can end up with
over-estimated Ag amount. On the other hand, even when Qa is unchanged, but
the accessibility of antigen to Ab is changed, then the Func is changed.
By and large, I feel it is not easy and sometimes extremely hard to use
quantitative cytometry in a justifiable way.
However, as far as I understand, it is relatively easy to interpret data if
we are: (1) to compare a tiny population producing something in huge
quantity on the single-cell basis, to a huge population producing little or
nothing, and support the tiny population is significant; or (2) to
quantitatively measure and express a shift of whole population on the
histogram when we cannot find a distinctive "positive" sub-population. The
reason is that these can be done within one experiment (everything is
internally controlled), and the conclusion is qualitative, instead of
quantitative.
Just an example: in comparing CD4 and CD8 cells in terms of gamma Interferon
production, we may end up with the same number of positive cells from CD4
and CD8, but we find every CD4 cell gives higher F intensity. Then, using
MFI we can say within the studied cell population CD4 is more significant in
terms of IFN production than CD8 is. This conclusion is not so easy to draw
just looking at dot plots without sort of calculation, if the mean
fluorescence (for interferon) is not so sharply different between CD4 and
CD8 cells.
Going back to Dr. Dawson's question, I believe the acceptability of data
reported as MFI must be discussed case by case. As today's flow cytometric
application and the biomedical research project behind it can be so
complicated, MFI can definitely not be applied everywhere. Nevertheless, it
is so desirable to have reliable quantitative flow cytometry: at least I can
avoid 3-day ELISA. I wish cytometry experts can resolve those problems soon.
Hai Qi M.D.
Department of Pathology, UTMB
WHO Collaborating Center of Tropical Diseases
301 Univ. Blvd., Keiller Bldg., Rm. 1.104
Galveston, Texas 77555-0609
Tel: 409-7728163
Fax: 409-7476869
============== Original Message ===============
Dear Flowers,
I have come across a term "mean fluorescence index" in several
recent papers. This term is used to describe the total fluorescent
capacity of a population of cells and is derived from the percentage of
cells staining positive for a marker multiplied by the fluorescent
intensity of the population expressing the marker. Does anybody have an
idea about the acceptability of this calculation to express flow data ?
--
Sincerely,
Harry D. Dawson, Ph.D.
Laboratory of Immunology
National Institute on Aging, NIH
5600 Nathan Shock Dr.
Baltimore, MD. 21224-6825
Phone: (410) 558-8300 ext. 7660
Fax: (410) 558-8284
E-mail: dawsonh@grc.nia.nih.gov
This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:53:22 EST