Should I use cells or beads? This question is at the center of many discussions nowadays in the flow cytometry world. Beads were once lauded as the easy alternative to cells to prepare your compensation controls. But as more and more researchers look at their spectral flow data, it is becoming clear that there is an issue: there are persistent problems with the unmixing matrix when using beads! And the worst part is that the issues vary with the different combinations of beads and fluorophores, making the whole thing a giant mess.

We’ve seen many publications and posters dealing with this problem (here, here). Dr. Debajit Bhowmick, Flow Cytometry Facility Director at the ECU, just placed the latest stepping stone with his paper ‘Side-by-Side Comparison of Compensation Beads Used in Polychromatic Flow Cytometry‘. An impressive list of fluorophores and beads were tested together in order to evaluate the accuracy of the compensation and unmixing. And the results are somewhat frightening.

Compensation beads fail to properly correct cell data. PBMCs were corrected by the same cells or compensation beads, also stained as in Fig. 1. The beads that passed bright and brighter were included. Unmixing the cell data using the same cell data are shown in the firs row. Rows 2–7 include the same cell data (columnwise) but were unmixed using different beads as represented on the left. Values between −125 and 125 are indicated in green (acceptable), values between −170 and −125 and 125 and 170 are in orange (partly acceptable), and values beyond −170 and 170 are in red (not acceptable). All data were acquired on the Cytek Aurora and the BD FACSAria Fusion (n = 3). (text from the paper)

The highlights of the paper, as I see it, are as follow:

• The spectral signature of fluorophores on the instrument will vary depending on the type of beads, and will vary over time. Repeated acquisition of the same beads/fluorophore sample does not present the same variability.
• The unmixing issues appear to be fairly random: they occur on both tandem and non-tandem dyes, on a different combination of beads and fluorophores, and vary at times between two repeats of an experiment. This opens the door for a wide array of reasons explaining the discrepancies – suggestions includes sample prep variations, some chemical interactions between the beads and the dyes.
• Cells seem to be the most reproducible solution for spectral correction.

There is a lot of work to do to clarify the situation. The solution likely resides in a combination of approaches that will alleviate the issue. Maybe the composition of the beads can be tweaked to reduce interactions with the dyes? Maybe a better software approach to calculate the compensation/unmixing matrix can be developed? (AutoSpill has not been used on the dataset, so it would be interesting to see what it can do about it). On that note, SONY has been advocating for using beads all the time on their ID7000 spectral flow platform. Is the ID7000 better at dealing with the small variations in signatures and providing accurate unmixing? If you know, drop me a line!

The question of what to do once you are done with your analysis and are still dealing with unmixing errors remains. The Purity School of Thoughts tells us to get back to the bench and find better controls. But the Realists are there to tell us that this may not be an acceptable solution, and we’ll need to devise ways to make the corrections on the fly, or find proper controls to deal with the situation. It seems like an FMO would do the trick, but I could be wrong. And I’m totally making up these opposing schools of flow cytometry ideology.

For now, I think the very general guideline that the CAT provides to our users remains the most valuable (if somewhat difficult to achieve in practice): spend a good amount of time characterizing as wide a variety of controls as you can, and decide which ones will be best for your assay. It’s tedious work, but will allow you to deal properly with your spectral correction.

How do you deal with your compensation and unmixing issues? Let us know below!