ClearLLab LS Webinar

What should we think about when designing an antibody panel? At least three aspects should be considered:

• Dye brightness vs antigen density
• Spillover/spreading
• Co-expression patterns

Of course, we need to ensure a good match of dye brightness with antigen density, which is the old-school world of fluorochrome selection. However, for 10-color cytometry, we must also include the impact of spillover and antigen co-expression patterns in our consideration, and during the next paragraphs I would like to illustrate how these factors influence our results.

Following this more theoretical part, finally I would like to show you how the panel design of the ClearLLab LS screening reagent practically reflects these panel design rules.

Brightness & Antigen Density

Figure 1. Matching brightness and antigen density.

Figure 2. Here is an example for applying this rule. The weakly expressed antigen scaled on the Y axis was labelled with a strong dye and the lineage antigen on the X axis, labeled with a less bright dye.

For weakly expressed antigens we need to select bright dyes, while strongly expressed antigens can be detected using any dye, says the most basic rule of panel design, which every flow cytometry user may recall from his first training sessions.

I would like to suggest a modification of the second part of this rule, as it also allows for selecting a bright dye for a strongly expressed antigen. This could result in a very bright signal that may, under certain conditions, impact other detection channels through spillover. We will have a more detailed look at this undesirable effect a few paragraphs later.

For now, the takeaway message would be that we want to avoid bright signals whenever possible, and rather prefer moderate intensities.

Consequently, strong antigens should be detected within our medium bright dyes instead of being detected with any dye, which would also include bright dyes.

Brighter Is Not Always Better

Figure 3. Bright emissions can cause strong spillover to other channels resulting in a loss of sensitivity through data spreading.

To demonstrate the undesirable effect of combining strong antigens with bright dyes, I have arranged in the first row of this illustration some plots from a bright dye staining of a strongly expressed antigen. In the second row, data from staining of the same strongly expressed antigen with a less-bright dye is shown with the same channel on the right. Why the separation of the positive population is much weaker in the second row we can well imagine without knowing the precise numbers here and we still get the same positive portion as seen in the first row.

However, in the first row we observed a substantial spreading of the bright lineage-positive population in three out of the four channels on the Y axis. This loss of resolution, and hence the loss of sensitivity, is a consequence of the inevitable spillover from the bright dye in channel 2 into longer wavelengths that’s detected in channels 3, 4 and 5. Due to the dim lineage signal in the second row, spillover spreading is absent. Spreading obviously corresponds to spillover intensity. We also observed comparable spreading in channels 3 and 4 in the first row, even though channel 4 is more spectrally distant to channel 2 than channel 3. Obviously, more factors influence spillover spreading than spillover intensity.

Spreading Depends on Bandpass Wavelength

Figure 4. Similar spillover from ECD to PE-channel (tandem leakage) and ECD to far-red PC7-channel.

Figure 5. Spillover causes much more spreading in the dark red channel resulting from a decreasing photon-electron-coversion rate (i.e., decreasing sensitivity).

In order to illustrate further how spillover impacts sensitivity, we may have a look at these two data plots.

This comparable spillover from ECD to PE tandem dye, either into the green yellow PE channel in the left plot, or into the near infrared PC channel 7 on the right plot. You may verify the similar amount of spillover by comparing the values of the ECD positive populations on the Y axis of the two plots, or just look at the compensation values that you would need to apply for correct data display, as seen in the upper left corner of the plots.

Applying these compensation values, you may be surprised that a similar amount of spillover light may result in different degrees of spreading, depending on the wavelength of the affected channel. The reason for this difference is that the sensitivity of photomultiplier detectors decreases as the wavelength increases. Longer wavelength photons have a lower probability of being converted into signals, thus creating a larger measurement error. You can imagine that this wavelength dependency of spillover spreading is very important for sensitive panel design.

Spreading Map

Figure 6. Data spreading pattern on Navios with standard filter configuration. Numbers represent LOG(loss of sensitivity) / LOG(intensity of primary signal).

If we run the same spillover spreading assessment across a 10-color Navios we can map out the effect on resolution for each dye in combination with each channel. In the metrics shown, the dyes are represented by columns and the detectors by rows. Please do not confuse these metrics with the compensation metrics, as comparable compensation factors may be related to different spillover spreading, as we saw in the previous figure. The different shades of grey represent the extent of spreading that occurs upon spillover. For example, we can see that PC 5.5 and the 6th row from the left have dark grey fields together with the PC 7 row, empty APC Alexa 700 row.

The example data on the right gives an impression of the spillover spreading represented by the different shades of grey in the matrix. For example, PC-7 PC 5.5 spreading into an APC Alexa 700 and PC-7 detection channels can be assumed to cause substantial loss of sensitivity. If not, fluorochromes typically have spectral tailing toward longer wavelengths while there is little spillover to shorter wavelengths. You can find this also reflected in the spreading map of the Navios, as seen here, as we go down a column in the matrix.

Spreading Corresponds to Spillover

In order to emphasize that spreading and compensation are profoundly different parameters, you may inspect the three plots below. Data from the same sample has been acquired with different detector amplification settings, resulting in different compensation values.

Figure 7. Increase of fluorescence background is independent of PMT settings but does depend on the amount of spectral overlap.

Obviously, spreading does not correspond to the different sizes of these compensation values, as spreading as the compensation is only computational correction of the amplified signal from the photomultiplier detectors.

Loss of Sensitivity Due to Spillover

To remind ourselves why we should consider spillover spreading to be just as important as matching antigen density and dye brightness, I've chosen the example below, so both plots show the same combination of antigens but with a different pairing of the fluorochromes to each marker.

Figure 8. Sensitivity of detection for an Ag in multicolor cytometry is a function of the entire antibody panel on a chosen instrument configuration.

Only the antibody conjugate shown on the Y axis is identical between the two plots. It is obvious that the antibody panel on the left provides much better resolution of the dimly positive population than the panel on the right, due to less spreading on the background population of the left. However, we have used the same fluorochromes for both datasets – only the pairing of the fluorochromes with the antibodies was different.

Impact of Antigen Expression Characteristics & Co-expression

Figure 9. Matching brightness and antigen density.

If the same set of fluorochromes were used for both cases, why is the background spreading so dramatically different?

The answer to this question can be found in the antigen expression patterns, as these determine the co-labeling of cells with the respective fluorochromes. In the example we can see how two fluorochromes A&B, with mutual spillover spreading, impact each other. We're looking at a hypothetical expression pattern that is modulated between a high positive B-negative, A&B high positive and A-negative B-high positive.

The grey triangles indicate areas in the plot where dim positive events cannot be distinguished from negative events. Either for a weak, A-co-labeling of cells of the strong B expression, or for weak, co-labeling of cells with a strong A expression.

Antigen Expression Characteristics / Patterns

Figure 10. Consider Correlations, Inverse Correlations, Loss of Ag Patterns.

Let's change our assumptions now, such that an aberrant antigen would be expressed on all cells, except the black dots and the descendant antigen that is expressed only on a portion of the population – and consequently does not occur without the parent antigen being expressed. As fluorochromes to select from, we now assume a fluorochrome A without spilling onto B, so no grey triangle is seen here. Then fluorochrome B with a considerable spillover, spreading into A, as indicated by the grey triangle.

We have now obtained two completely different situations depending on how we pair the fluorochromes with the antigens. The combination of parent B ascendant A that we see on the left will cause spillover spreading and loss of sensitivity, while swapping the fluorochromes, as seen on the plot on the right, we will see no loss of sensitivity at all.

Panel Design Interdependencies

Figure 11. Panel Design Independencies.

Taking together antigen density, fluorochrome brightness, instrument spreading behavior and the expected expression patterns, here depicted as a family tree in the upper section of the diagram, we end up with a complex set of independent parameters. This is characteristic of multicolor flow cytometry.

ClearLLab LS: The Dyes

How is this set of interdependent parameters reflected in practical panel design? Let's have a look at the implications for the ClearLLab LS. The set of dyes used for the ClearLLab LS screening regent is optimally suited for Navios and Navios EX flow cytometers.

Figure 12. ClearLLab LS: The dyes.

Five dyes are in the bright range that you see here on the right. Three dyes in the medium bright range and the two dyes on the left, upper row in the dim range. Please see also the overlay plot showing dyes from each of the groups conjugated to the same CD8 clone for illustration of the relative brightness.

ClearLLab LS: The Antigens

Figure 13. ClearLLab LS: The antigens.

This figure shows the ClearLLab antigen family. CD45 is the master parent antigen in the mononucleocyte fraction on top with direct descendant, CD3 CD56, CD34 and CD19 representing lineages that exclude each other, indicated with the red lines. Depending on the expression characteristics, some antigens may occur more than once in the tree. As seen here for CD56 as in K cell image antigen but also as descendant antigen from CD3 and for CD5 which is a descendant principal T-cell antigen from CD3 but also with different expression characteristics in descendant antigen of CD19.

The color-coded squares beneath the antigens indicate the typical levels of expression for this antigen within a certain branch of the tree. CD34 myeloid might co-express some of the lymphoid markers shown here. However, characterization of these abnormal entities should be subject to a separate, dedicated phenotyping panel, as ClearLLab LS was designed only for screening purposes.

ClearLLab LS Implications I

Figure 14. Densities: Matching weak expression with bright dye, strong expression with less bright dye.

CD56, CD10 and CD34 have weak to medium-strong expression densities often showing modulated expression densities. Therefore, these antigens are labeled with strong dyes. CD3, CD45, CD8 and CD20 are densely expressed antigens, consequently, the less bright eyes, Chrome Orange, Pacific Blue and medium, bright dyes, FITC and APC Alexa 750 are suitable labels. But why do the strong antigen CD4 and CD5 carry bright labels? Would this not affect the resolution or other challenges we have seen on the figures before?

ClearLLab LS Implications II

Figure 15. Exclusions: Allowing for positions with strong mutual spillover.

Looking at the exclusive expression patterns indicated by the colored frames in this figure, we can see that CD4 is not co-labeled with the ECD-labeled antigen CD19, so we will not experience sensitivity issues for these excluded antigens.

The same situation applies to CD19 in ECD and CD56 PC 5.5, Kappa Lambda, CD8 and CD4. The latter antigens can be co-expressed in some hematolymphoid malignancies. However, the spreading spillover of FITC into PE will be only moderate due to the short wavelength of the PE bandpass filter. Furthermore, CD34 positivity would not be seen on normal CD20 positive B-cells or bright CD5 on T-cells would not be associated with B-cell CD20.

ClearLLab LS Implications III

Figure 16. Parents: Allowing for strong spillover from descendant antigens.

The red arrows indicate spillover from their respective fluorochrome neighbors. However, the receiving channels host parents’ antigens, whose detection will not be compromised by spillover from descendent antigens, as we have seen earlier.

ClearLLab LS Implications IV

Figure 17. CD19 Co-expressions: Minimize spillover between co-expressed antigens.

For co-expressed antigens, it was made sure that a considerable spectral distance lies between them, so that spillover spreading would not impair sensitivity. The principal is shown here for CD19 co-expressions, master parent CD45 was placed on the spectral island with minimal or no spillover into any other channel.

ClearLLab LS Implications V

Figure 18. CD3 Co-expressions: Minimize spillover between co-expressed antigens.

In this figure we find the same angle of hue, seen for CD19 previously and now applied to CD3 co-expression. CD5 occupies a spectral island just as CD3 and CD45 do, for weak CD56 expression densities on CD4. However, you would still have to expect a moderate loss of resolution due to the long wavelength of the PC 5.5 band pass.

We all know that major flow cytometry clinical applications include determination of immunologically defined leukocyte subpopulations, and the immunophenotyping of hematological malignancies. Both applications can be presented through the ClearLLab lymphoid screening tube.

Regarding the first clinical application, the determination of immunologically defined leukocyte subpopulations will emphasize the populations, their precise absolute number and percentage determination needed, and the reagents used are CE-IVD. But when we do immunophenotyping of hematological malignancies, the goal of the analysis is to identify an abnormal population within an “unknown” immunophenotype. These populations could be more than one, and the answer of course, is to use multiparameter flow cytometry.

Immunophenotyping of hematological malignancies, factors affecting results

Then we must discriminate an abnormal, possibly malignant population from the normal cells, characterize the phenotype of abnormal cells and then enumerate with different populations whether they are abnormal or normal. We should find the number of the abnormal cells, the degree of heterogeneity of antigen expression, the intensity of expression of the markers, the ability to discriminate all the populations, and, of course, we should have a correct selection of marker combinations. We should find the precise enumeration of the blasts, which is the cell line.

Immunophenotyping of hematological malignancies: selection of MABs

The selection of monoclonal antibodies is crucial. But in our reports, we should give an estimation of abnormal cell percentages, description of the complex immunophenotype, and possibly a compatible diagnosis.

Hematological malignancies indications for referral and follow-up

When the sample arrives at the lab it usually has some indications:

Also, the sample can come for follow-up of malignancies to evaluate response to treatment or to see if relapse is occurring, and for the follow-up of HIV or other infections.

The lab then decides how to proceed. To do this we need to determine if there is an abnormal B-cell population, and if this population is mature or immature. If we identify an abnormal B-cell population, then we will proceed with further clarification. We should also determine if there is an abnormal T-cell population and if it is mature or immature. If so, we proceed with further clarification. Are there any immature CD34+? If so, we should proceed with further clarification of those cells. If no abnormalities are detected, we should have the answer about the lymphocyte main populations ready for the patient. All these answers can be achieved using the ClearLLab lymphoid screening tube.

B Cells

We have a way to discriminate singlets, and, of course, we can see the CD44-positive cells, so we have all the leukocytes. We can see the percentage of CD3 cells on total leukocytes, the percentage of CD19 B-cells on total leukocytes, as well as the percentages of T-cell and B-cell only lymphocytes.

T Cells

Regarding T-cells, if we gate on CD3 cells we can find the subpopulations of CD4 and CD8 cells, and the CD3 cells that express in the CD56. We can also find in this dot plot the NK cells, and the cells that are CD56 positive or CD3 negative. Also, if we gate only on T-cells, we can see how many of these cells are CD3 positive and CD5 positive, and if they are in the lymphocyte gate, some cells that are only CD3 positive or only CD5 positive.

Healthy Donor

Regarding the B-cell, we see the CD19-positive cells, the B cells; are all of them CD20 positive but mature B-cells? Do they express CD10 or CD5? Are they polyclonal, polyclonal Kappa and Lambda or monoclonal?

Immature cells

And of course, we can determine if there are CD34+ immature cells, and if these cells are CD10+. To do that, you should gate on the blasts and not only lymphocytes as is depicted in this picture.

Advantages ClearLLab lymphoid screen

ClearLLab LS provides information on:

• T, B and NK cells
• CD4+, CD8+, CD5+ and CD56+ T-cells
• Kappa+ Lambda+, CD20+ and CD10+ B-cells
• CD5+ B-cells, monoclonal or not, for B-lymphoproliferative diseases
• CD34+, CD10+/- blasts for acute leukemias

Compare this to the traditional approach where we were going step-by-step to find out what was wrong. We now have the advantage of being able to detect most B-cell abnormalities that could have gone unnoticed if we used the old approach. We can detect most B-cell abnormalities and immature cells that could have gone unnoticed. This tube is also CE-IVD, so it is already validated and ready for accreditation.

Disadvantages of a ClearLLab lymphoid screen

To balance out a comparison to the traditional approach, we must consider the disadvantages -- which can be overcome if we use our expertise.

Some T-cell abnormalities may not be detected e.g., CD3-CD4+ cells or CD3 weakCD4+ cells, but this can be corrected with additional gating and checks.

More gates could be checked, like the double-positive CD4+CD8+ cells or the double- positive CD8+CD56+ cells.

It is difficult to use the old, easy way of the single platform to determine an absolute number, which is obligatory according to international guidelines. In addition, we cannot use Q-Prep for the lysis of red blood cells, as this leads to a loss of time and automation, but it is necessary for many colors.

Hematological malignancies healthy donor T-Cells

Here's an example from the lab of a healthy donor. You see this CD45-positive cell. You see the percentage of T-cells and B-cells on lymphocytes. You see the double-positive CD4+CD8+ cells, the CD4 cells and the CD8 cells. You see the cells at the double- positive for CD3 and CD5 themselves, and those only positive for CD3 or CD5. You see the NK cells, CD56 positive CD3 negative, and the T-cells that express CD56.

Hematological malignancies healthy donor B-Cells

Regarding the B-cells: the cells that are double positive for CD19+ CD20+ mature B-cells but do not express CD10, that are not monoclonal but polyclonal Kappa and Lambda, and that do not express CD5 or CD10.

Hematological malignancies CLL

Here we see that we have a large percentage of CD19 B-cells on lymphocytes, but these cells are absolutely monoclonal, they are Lambda positive. They are double positive for CD20, and at the same time they are positive for CD5. A lot of T-cells are not CD3 positive. CD5 cells are not CD3 positive. So, we have a CD5positive beta lymphoproliferative disease that is not a CLL.

Here we have a case where the B-cells are CD20 negative, CD5 positive, and do not express Kappa or Lambda chains, or they could express very dimly one of these populations. So, we have a beta lymphoproliferative disease which could be CLL or aberrant.

Here is another case of the CLL; we see again that there is no expression of Kappa Lambda chains, but the CD20 is negative and CD5 is positive. But if we gate on the few CD20 positive B-cells, we can see that these cells are polyclonal, and they are Kappa and Lambda positive.

In the same case, we then studied together in this tube the T-cells. We see that there is an interesting population of CD3 positive and CD5 negative but also seems to be CD56 positive B-cells CD8 positive. So, we can draw some conclusions about T-cells being the big key for lymphoproliferative disease.

Here is another case where CD19 seems weak. The cells do not express Kappa and Lambda chains. They are positive for CD20 and positive for CD5. And here we have the case where the percentage of CD19 is only 16%, but when we look at the Kappa Lambda expression, we can see that we clearly have a monoclonal population that is CD20 positive, CD5 negative. Here the old approach could miss this population because we could say that the B-cells are not right, so why check for them?

Here is a case with a very highSS for the B-cells, and a very high expression of Lambda on these cells with highSS, high expression of CD20 and high expression of CD5.

Malignant B-cell clones

Here is a case where we found two malignant B-cell clones. We find that there are some CD5 positive cells, CD3 negative, B-cells are B-cells. You can see they are double positive for CD5 and CD19, and B-cells do not express Kappa Lambda chains, so it is a CLL like clone.

At the same time, we see that there are some B-cells, but CD20 positive CD5 negative, but are Kappa positive monoclonal. So, we have two malignant B-cell clones that can be easily detected with the ClearLLab lymphoid screen tube.

CSF sample

Here is a CSF sample, and if we gate on the lymphocytes, we see that there is a population that is CD5 positive and CD3 negative. This population could be missed if we gate it only on the T-cells. On the CD3 positive cells we missed you will see we had this abnormal population. For this population, if we gate on it and not on the CD3 positive cells, it is double positive for CD4 and CD8, and at the same time, with this tube we can see that it is also positive for CD10. This is helpful for the correct identification of this population.

Bone marrow sample

Here is a picture of some blasts, CD34 positive, CD10 positive in a bone marrow, that can be detected, and their percentage can be determined with the ClearLLab lymphoid screen tube, so afterward we should find out what kind of blasts they are.

When using a ClearLLab LS tube you follow the CE-IVD guidelines, but you can also check all gating possibilities, use your expertise and check any suspected abnormals. In flow cytometry, a diagnosis is approached by “knowing how” (millions of pictures in our brains) and not by “knowing what.”

The ClearLLab LS tube, and the harmonization approach, could place the “knowing what” before the “knowing how.” But still, you should always remember, every single case is an individual one.