Page 1 of 1

titration intepretation

PostPosted: Wed Feb 10, 2021 3:42 pm
by GregHopkins
I'm curious to find out how everyone interprets their antibody titration data. I would imagine there is a diverse set of methods everyone uses to interpret the data.


We've been using several different methods for a number of years usually involving looking at staining intensity and background reduction, but have had intentions of using something more mathematically derived.

Re: titration intepretation

PostPosted: Wed Feb 10, 2021 6:10 pm
by mleipold
Hi Greg,

I'm not sure what you're asking. Are you looking for something like the Average Overlap Frequency ( https://doi.org/10.1016/j.jim.2017.08.011 )?

Also, are you talking about titration of a new prep of an already-characterized conjugate (ie, you ran out of a prep, and just made another conjugate of the same clone to the same polymer labeled with the same metal)? Or are you also talking about characterizing a new marker/clone?


Mike

Re: titration intepretation

PostPosted: Wed Feb 10, 2021 6:15 pm
by GregHopkins
Hi Mike,

Not looking to necessarily change our methods, per se, but I am trying to get a sense of the methods that all the users use to intepret titrations of new targets as well as new lots of validated targets. We have a system we've been using for a while that has been more observational, and its been pretty consistent in getting equal staining from lot to lot, but definitely seeking to take the eye of the observer aspect out of it a bit more.

Really, its reaching out to see how all the users in the forum go about analyzing their titrations, to see if there are things that we could use to improve our method.

Greg

Re: titration intepretation

PostPosted: Wed Feb 10, 2021 6:45 pm
by mleipold
Hi Greg,

New prep of a Historical Conjugate:
The main thing that we do to validate new preps of historical conjugates is running them against a "known" sample. For example, we use cryoPBMC controls on most of our plates; we get those controls by processing an LRS chamber, so we get high tens to low hundreds of vials from the single draw/LRS. Therefore, we typically have "historical" data from that marker-conjugate's previous batch(es) from previous plates. If there are stims involved then obviously you need to stim, etc.

But in general this has worked for us. I'll admit to being a bit lazy and not often running full titration curves on new preps: 9/10 times, the titer for the new prep is the same as the old prep. 9/10 of the remaining 1/10 time, it's within a factor of 2, and I test that in a subsequent experiment if the staining seems low, high, or streaky. But I don't run customer samples until I'm happy with the titer (including occasionally tossing and starting over with a conjugate with really off-titer signal or background). And, from Daniel's comments here ( viewtopic.php?f=4&t=1414 ) on how very tight titering (along with stronger fixation) has resolved Sinai's intermittent signal intensity issues, I might change my plans.


Development of a New Conjugate:
This is a trickier question. Hopefully you have some idea of where to start: previous flow cytometry datasets, previous papers, etc. So, you'll know whether you have to Stim (and how) to find the marker; whether sample processing has depleted (ie, few/no Grans in PBMCs), etc.

Hopefully you also know which cell types the marker *should* be on, and just as importantly, which ones the marker should *not* be on (negative control beyond MMO). This is particularly important in the case of dimmer and/or rarer markers. For example, CD85j (ILT2/LILRB1) staining on CD8+ and NK cells is often dim and highly donor-dependent. However, it's expressed on many/most Monocytes and B cells, and virtually no CD4+ cells (especially in Healthy people, but even most disease states I've examined). Therefore, I can feel confident in a shoulder or dim peak on CD8+ and NK cells, because of having inherent Positive (B cells, Monocytes) and Negative (CD4+) controls inherently in the sample and exposed to the same staining conditions.

There's also the ugly/wonderful matter of biological variability between donors. This can be both in terms of Frequency (my current Healthy Donor PBMC control has virtually no CD57+ CD8+ cells, but does have CD57+ NK cells) as well as Expression/Marker Intensity (1-2% of of the donors I've seen have CD33dim Monocytes that are nicely CD14+). This is one reason why I try to stain at least 3 different donors when doing development work: you could still get misled, but it's increasingly unlikely that you'll get 3 weird donors for a given marker.

As an example: when I first joing the HIMC back in 2010, my first job was to adapt the HIPC Lyo panel to the CyTOF platform. A few clones had to be changed along the way. In particular, I had problems with CD28 at that time. I finally found a clone that gave good staining on T cells, and little or no staining on biologically negative cells. The T cell staining matched the fluorescent flow staining on the same Healthy Control donor that a labmate performed. So, I thought I was done.

Unbeknownst to me, that Healthy Control donor was weird in having a very high percentage of CD28+ CD8+ cells. And my limited immunology knowledge at the time didn't raise a flag when most of the donors in the plates I started running *also* showed a high Freq of CD28+ CD8+. Someone from a different lab analyzing the data noticed the issue, and I had to go back and change the CD28 clone again. I then talked with the manufacturer about that CD28 clone (how it would only stain T cells, but would give a false positive on most CD8+); they then admitted that other people had reported that problem.

Had I tried 3 different donors and compared them to the flow experiments my labmates were doing, I would have caught this earlier.


In summary:
1. Check multiple donors.
2. Check all major cell populations for false positives/false negatives.


Mike