Compensation in Clinical Lab
roederer at drmr.com
Tue Dec 11 13:05:53 EST 2007
Let me begin by stating that I don't advocate compensating once a
day. No, it would be better to compensate every experiment.
If only everyone's instrumentation and reagents could be as well
qualified and consistent as yours! Yes, in an ideal world,
compensation would only have to happen once.
Unfortunately, this is not the case for a vast majority of
laboratories, particularly non-clinical laboratories. There are
multiple issues at work. One is the reproducibility of instrument
setup (and, incidentally, monitoring of PMT performance and laser
performance, alignment, etc, all of which can influence compensation
without triggering many QC measures). Most laboratories do not use an
instrumentation setup which assures the same sensitivity from day to
day. Indeed, any laboratory which uses the same settings (for PMT
voltage etc) is guaranteed not to have the same sensitivity on all
days; the only way to assure equal sensitivity is to calibrate based
on fluorescence targets.
Perhaps the most critical variable, however, is the reagents. Many
reagents have different fluorescence characteristics (spectra)
depending on such variables as (1) lot number; (2) manufacturer; (3)
fixation conditions [time, temperature, buffer]; (4) exposure to
light; (5) time since staining. While these variables are
considerably less different for fluors like FITC, Alexa488, PE, they
are extremely variable for tandem dyes (Cy5PE, Cy7PE, etc.). Much to
my surprise, we recently discovered that APC reagents from one
manufacturer are slightly different than APC from others in that it
required more compensation into the Alexa700 channel! For 20 years
I've assumed that APC spectra are identical.... no more.
Staining conditions, as noted above, can also affect compensation.
So, you are right, in a clinical laboratory where you have extremely
rigorously controlled instrument setup (and I presume you calibrate
for sensitivity), where you use the same reagents from day to day (and
do a bridging study to ensure that different lots of fluorochrome are
identical), where the panels are relatively simple (3-4 color, for
example, and not utilizing any tandem reagents), and where you have
rigorously controlled the staining procedure, then there is no need to
do compensation for every experiment.
I am concerned that you only checked compensation settings after your
recent service, however. I would in fact expect the compensation
settings to return to "normal" after servicing -- that's part of the
idea of servicing the instrument. How do you know that in the weeks
preceding service that it wasn't drifting? Note that simply viewing
compensated data is no way to judge whether or not the data is
properly compensated -- I can give you tons of samples which are
either properly compensated or not properly compensated, and no one
can judge by looking at them which they are.
The best way to determine how often you need to do compensation
controls is to do them regularly, and see how much variation you get
in the compensation matrix. Any time the values are different by more
than about 0.5 (in units of percent), then you will have significant
This is why I recommend, for 99% of the laboratories out there, to
perform not just daily compensation, but EVERY EXPERIMENT that uses
unique reagents. (Daily is acceptable if you use all the same
reagents and if you are sure that your staining conditions are
identical). I should note that I am very proud of the system that
Steve Perfetto and his crew have put in place here at the VRC for
QC'ing our seven different instruments -- they are extremely
reproducible. And yet, I still insist that everyone do their own comps
for every experiment. Why? Because everyone is using different
reagents, different staining conditions, different fixatives,
different exposures to light. Nearly every example that has been
brought to me of "bad compensation" arose because someone didn't
prepare their compensation tubes that day for that experiment.
Finally, note that if you are doing experiments where you are
measuring MFI, then you really have to compensate every single time.
Here, very small variations in compensation (much smaller than we
could ever detect visually, even if we know what we're doing) can have
enormous impact on the output variable.
On Dec 11, 2007, at 10:35 AM, Mark Simmerson wrote:
> Hi Flowers
> I'm a bit puzzled by the advocating of doing compensation as
> frequently as once a day.
> As long as one can demonstrate that your flow cytometer is stable
> via regular QC to
> ascertain optimum laser alignment, instrument standardisation of
> light scatter intensity,
> fluorescence intensity and hydrodynamic focussing, plus instrument
> linearity, then why do
> you need to perform time consuming, costly (and unnecessary)
> compensation experiments.
> I would suggest that if you keep the same panels, instrument
> settings, etc, then the only
> time one need recheck compensation is when the instrument may have
> been altered - eg
> after a service.
> We installed a FC500 about 18 months ago and once the initial
> compensation settings were
> ascertained, we have not needed to alter them. I checked the
> settings after our recent
> service and found that the settings were so similar to the original
> that no action was
> This doesn't mean that one can safely ignore compensation, but
> surely a good flow
> cytometrist always examines the data to ensure that data is not -
> for example - smeared
> along either the x or y axis.
> Mark Simmerson
> Lead BMS Flow Cytometry
> Sheffield Children's Hospital
>>>> "Irina Grigorieva PhD" <Irina.Grigoriena at northside.com> 11/29/07
>>>> 9:29 pm >>>
> Hello Richard,
> Our lab is using FACSCanto. We have 2 six-color instruments and we
> them. When using Canto software (for lymphocyte typing pane or CD34
> enumeration), you should not even have problems with compensation,
> run 7-color beads setup and do the automated compensation.
> For immunophenotyping you use CST beads for daily setup (Diva 6
> software), so you have to perform the compensation exercise. But
> CST setup, you do not have to do this exercise daily (I am not even
> talking about every run). First, you establish your compensation
> and we do it by using CD8 in every color for a normal donor. Then you
> compare your results for every antibody you use in your panels with
> (ex. compare CD8-PEcy7 with CD33-PECy7, if you use CD33 in this
> Then you run your cocktails (or whatever format you use for your test
> antibodies) and verify compensation settings. When initially
> establishing compensation values we always additionally check them
> on CD
> Chex, just to confirm the values. We call this procedure 'Establishing
> and validating compensation settings'.
> Now you have a different task of monitoring your compensation settings
> and this procedure does not necessary include running compensation
> exercise. We run daily normal controls and we have established our
> normal ranges for each antibody. Therefore, we can use these data to
> overall assess our daily compensation, comparing the patterns and
> values. And then each lab is deciding for themselves when the patterns
> do not look consistent or numbers do not match exactly. One day
> prior to
> that happening is date of your compensation exercise. Depending on the
> lab it may be 1 day, 1 week, 1 month.
> We have started doing it every day 2 years ago, and then changed our
> schedule according to the patterns we saw overtime. The instrument is
> very reliable, settings do not require a lot of adjustments even at
> time of scheduled compensation exercise. Therefore, my lab policy
> states: After initial establishing and validating compensation
> the verification of the use of correct settings is performed weekly.
> Additionally the same compensation exercise is performed in a case of
> any questionable results, changing the reagents, and also when laser,
> PMT optical filter components or settings are modified.
>> From my personal experience there are a lot of myths created around
> flow cytometry: danger of using tandems, risk of using stored samples
> for acquisition, constant failure of PMT and compensation settings,
> All this forces clinical flow people to perform a lot of excessive
> not necessary needed. Since we run ASR tests (beauty and trouble of
> clinical flow), we have to do an extra work to create the system with
> proper parameters, normal ranges, controls and standards. Then you can
> use your own system to validate all your procedures, from storing
> conditions to number of compensation runs per unit of your time.
> Overall I truly believe that with modern equipment, very capable
> software and reliable reagents our clinical flow life should become
> complicated, although Current Protocols recommendations do not support
> my theory.
> Regards. Irina
> Irina Grigorieva, PhD
> Director, Flow Cytometry Laboratory
> Northside Hospital, Atlanta, GA
> (404)- 851-6541
> e-mail: irina.grigorieva at northside.com
> From: Richard D. Schretzenmair [mailto:rds at mail.med.upenn.edu]
> Sent: Tuesday, November 27, 2007 7:10 PM
> To: cyto-inbox
> Subject: Compensation in Clinical Lab
> Currently, we are validating a FACSCanto for clinical use. In trying
> develop standard protocols for the operation of this instrument, a
> question arose about how often compensation is required? We had
> to compensate daily, but Current Protocols says that compensation is
> required for every run (in the compensation myths section). When you
> compensating 6 colors, then include each antibody conjugated to a
> dye, your could easily exceed a dozen tubes for each compensation run.
> In the era of tight hospital budgets, the cost of antibodies,
> compensation beads (or compensation samples) and labor needs to be
> It is my understanding that compensation is dictated by the instrument
> setup (filters, mirrors, PMT Voltages, lasers) and the specific
> fluorochromes used. This machine is QC'ed using DiVa 6's CST software,
> which does a maniacally thorough job at QC. So, if the instrument
> is not changed, CST QC is passed and PMT voltages held constant, how
> often does compensation really have to be performed? Samples are run
> immediately after staining. It seems that doing compensation with each
> run (several times a day), would be more likely to introduce human
> error, than to correct for any true changes in compensation.
> Considering the factors effecting compensation, is it really required
> for every run, or even daily, given the above circumstances?
> Richard D. Schretzenmair
> Flow Cytometry and Cell Sorting Shared Resource
> Abramson Cancer Center
> University of Pennsylvania School of Medicine
> 297 John Morgan Bldg.
> 3620 Hamilton Walk
> Philadelphia, PA 19104-6082
> Phone: 215-898-3528
> FAX: 215-898-4227
> rds at mail.med.upenn.edu
> CONFIDENTIALITY NOTICE: This electronic mail transmission has been
> sent by Northside
> Hospital. It may contain information that is confidential,
> privileged, proprietary, or
> otherwise legally exempt from disclosure. If you are not the
> intended recipient, you are
> hereby notified that you are not authorized to read, print, retain,
> copy or disseminate
> this message, any part of it, or any attachments. If you have
> received this message in
> error, please delete this message and any attachments from your
> system without reading
> the content and notify the sender immediately of the inadvertent
> transmission. There is
> no intent on the part of the sender to waive any privilege.
More information about the Cytometry