Absolute Cell Counts

[CRStevens]

Hi all,

I was wondering if anyone has attempted generating absolute cell counts on the CyTOF?
In theory we should be able to?
My thought was using a known amount of a metal labeled bead spiked into a sample. If anyone has tried this or has some wisdom would love to discuss. Seems like an important output.

-Chad

[mleipold]

Hi Chad,

Are you thinking about spiking the beads in at the *beginning* of the protocol (like TruCount for flow)?


Over the years, there's been some discussion about this.

The biggest issue that keeps coming up is the cell recovery on the instrument: the best recoveries are usually 40-50% on a Helios or with a Supersampler, but inversely that means that 50-60% of your sample is missing (and that's not counting washing-based losses along the steps of the protocol).

Even under optimal circumstances:
1. All cell types going "missing" at the same rate to avoid cell ratio distortions (generally a decent assumption as far as I can tell)
2. Low number of bead+cell doublets which might necessitate low cell concentration and therefore long acquisition time)
3. Well-behaved non-clumpy sample (therefore uniform data acquisition)
4. Beads behaving like cells and being "lost" in washes just like cells (might be reasonably true?)

that still gives you really big error bars on absolute counting (rather than relative counting). At the very least, I think there would be a lot of preparatory experiments to characterize your instrument's recovery (ie, that your recovery variance of CyTOF vs TC20 or whatever is tightly 44-46% rather than lower 35-40% or broader 40-55%)


Looking at the BD TruCount instructions
https://www.bdbiosciences.com/en-us/pro ... ivd.340334
https://www.bdbiosciences.com/content/d ... -22402.pdf

it looks like there are several important aspects:
1. Commercially prepared tubes with known amounts of beads.
- at least within your own project, you could probably make a "lot" of tubes that would be fairly internally consistent for your purposes.

2. Addition of precise amount of sample.

3. Lyse-no-wash acquisition (or at the very least, I don't see any centrifugation steps or buffer washes in the TruCount protocol).
* I'm not sure how you would do this in CyTOF without having tire tracks in your acquisition. How much super-stringent Ab titration (including longer overnight staining like in Whyte et al https://doi.org/10.1002/cpz1.589 ) could help, I'm not sure. But I don't know that it would fix your Ir intercalator streaking.

4. Not explicitly mentioned, but implied: the much higher cell recovery on a flow cytometer.


I've never used TruCount on flow, so anyone with hands-on experience, please chime in and comment on how they would see a CyTOF version working!


Mike

[CRStevens]

Thanks Mike,

I have used the Truecount previously, but we used it on samples that were already processed. With this in mind I was wondering if spiking in a metal-labeled quantified bead into your final sample just prior to acquisition so there would be no centrifugation. I know it is a bit of a pipedream, but wanted a discussion to think about possibilities.

My thought was spike xuL of known beads into 1mL of sample and run it dry.
Then you would know (reasonably minus the capilary volume) the amount of sample run, the number of beads/uL.

The efficiency is the biggest issue I agree. We have run internal data that I can't disclose that shows there is even loss across all populations with efficiency. This means if you lose 50% it is 50% across the board for all populations (based on flow comparisons). Would you be able to calculate efficiency on a run by run basis and use this as a factor to multiply your. It is a bit circular, though as you would need to have a way to count to know your efficiency, which i guess would then have to be done with a cell counter or flow...?

[rgrenfell]

Would the EQ beads not work for this?

You might have do work out the absolute bead number, but it should work from then?

Richard