mcanton at wcube.fr
Wed Mar 4 06:51:43 EST 1998
This is going to be a rather long answer or comments to the question posted
by Kathy Carswell.
This answer is absolutely not directed against any of the existing beads or
quantification system. However we thought that our (BioCytex) long
experience of quantitative flow cytometry QFCM (more than 15 years) allowed
us to contribute to this discussion and to share with all of you some of our
First, the only thing you need in order to create your own regression curves
with any type of QFCM calibration beads (DAKO Qifikit, Quantum series,
Quantibrite, or others) is a spreadsheet such as EXCEL or equivalent.
The goal is to compute log MFI versus log Ab molecules (or fluorochrome
molecules or ABC), i.e., values provided for each bead.
Thus if you get MFI linearized values (from 1 to 10,000 a.u. on a FACS, or
from 0.1 to 10,000 on any Coulter instrument), then compute:
log (Ab molecules) = a * log (MFI) + b
The slope a should be very close to 1 (0.95 to 1.05); b lies between 2 and
3, depending on apparatus sensitivity and settings. This means that when MFI
= 1.0 (1st channel), i.e., the lowest possible sensitivity level at the used
setting, the background is equivalent to 100 to 1,000 antibody molecules:
log (Ab molecules) = 1 * log 1 (=0) + b
For any unknown sample (x), the computer can then calculate:
log Ab(x) molecules = a * log MFI(x) + b
Ab(x) molecules = 10 * (a * logMFI(x) + b)
A graphical representation of the regression line shows two 4-log decades
coordinates: the abscissa x = log(MFI) is equivalent to what you see on the
FCM histogram, i.e., from 1 to 10,000 a.u.; the ordinate y = log (Ab
molecules) goes from 100 to 1,000,000 molecules.
A good example of such a representation can be seen in Poncelet P, et al
(1996), Eur.J. Histochem., 40 (suppl. 1, 15-32)
in the DAKO QIFIKIT flyer (ask your closest DAKO representative in the USA
or in Europe or in Japan)
in the technical data sheet of the APOCYT Fas kit (ask ALEXIS CORPORATION,
San Diego, CA).
In case your FCM results are still provided as channel number (old
versions), the abscissa from channel 1 to 256 is already equivalent to log MFI.
log (Ab molecules) = a * channel No + b
and proceed as above for unknown samples.
In that case, a is different from 1, but b generally still lies between 2 and 3.
A common error to avoid is to simply compute:
Ab molecules = alpha * MFI + beta
These calibration beads cocktails are finely tuned for log amplification and
doing this will provide unreliable values in the low MFI range. In fact in
lin/lin coordinates, only the highest level beads have much influence on the
Now if you do not want to bother at all with logs and exponentials, TALLYCAL
(DAKO) is a specific tool which is devoted to doing that for you. It can
handle any type of bead cocktail and any type of MFI expression (linear
values, FACS or Coulter, channel number).
David L. Havilar was doubtful about Quantibrite being a means to
quantitatively determine the number of CD? molecules per cell. We can not
speak specifically on QuantiBrite (We ordered it in January 98 and still ...
did not get it!). However for our own products (QIFIKIT/DAKO, APOCYT
Fas/ALEXIS Corp, PLATELET line/ALEXIS Corp, CYTOQUANTline/DIAGNOSTICA
STAGO), the technologies are developed so that the results agree with
radiolabelled MAb binding results (Poncelet and Carrayon (1985),
J.Immunol.Methods, 85, 65-75).
More recently, data generated with QIFIKIT were again cross-checked against
Scatchard plots by one user who wanted to measure MDR-related P-gP (Ferrand
et al (1996), Cytometry, 23, 120-125).
As an answer to David's request, i.e., ligand binding assays, the paper from
Carayon et al (1996), Blood, 87, 8, 3170-3178, (1996) , describes the
equivalence between the QIFI approach and a titrated ligand binding assay
for measuring a mitochondria-associated intra-cytoplasmic antigen
(peripheral benzodiazepin receptor).
Quantitation of platelet surface glycoproteins has often been performed by
radio-binding and the values for GpIIb/IIIa, GpIb and GMP140 (CD62) measured
using our Platelet Gp kit, Platelet GpIb/IX/V kit and Platelet GpIIb/IIIa
Occupancy kit, give values perfectly matching the published radio-binding
Thus we would like to emphasize that at least some quantitative FCM
application kits correlate well with more traditional assays ans thus can be
used to determine the number of Mab molecules per cell.
Life is not that bad!
In some occasions, however, Mab binding can give a false estimate of Ag
sites per cell. This can come from a monoclonal antibody binding bivalently,
as for GpIIb/IIIa (approx. 50,000 Mabs/platelet, approx 50,000 fibrinogen
receptors/platelet, ... but approx. 100,000 GpIIb/IIIa molecules per
platelet) ref: Jordan et al, Blood (1996), whereas one would expect tehm to
most often bind monovalently when used at saturating concentration.
If we must admit that the choice of Mab influences the standardization of
the data, we now must recognize that the fluorochrome conjugation is another
source of variability.
That PE-conjugates could be the best choice still temains to be established.
We always trusted the use of un-modified and un-conjugated Mabs. The old
battle about direct versus indirect immunofluorescence FCM techniques is now
totally obsolete. Indeed the indirect approach has been dramatically
improved so that no-wash whole blood application kits allowing two- or
three- or more color counter staining are now available.
Besides that last statement we and many other individuals using the so
called indirect approach have obtained consistent and standardized results
for more 10 years (see Porwitt-McDonald et al., (1996) Blood, 87, 1162-1169.
We apologize for being quite long. We do not talk a lot but since
quantitative FCM is our know-how, we just feel that sometime we should
Hope this helps
Philippe Poncelet, PhD
Michel Canton, PharmD
More information about the Cytometry