Summary Document
This document is a summary of message from the Purdue University Cytometry Laboratories Cytometry Discussion list.
http://www.cyto.purdue.edu/hmarchiv/index.htm
Initial message sent:
I would like to get some feedback about reporting clinical flow cytometry results for bone marrow samples. Does everyone report the percentage of abnormal and normal cell populations present. If so, what value do you use to express the result - do you report each population as a percentage of CD45 positive cells, or as a percentage of total nucleated cells in the sample.
My reason for asking - we have a bone marrow sample from a patient with erythroleukaemia. The morphology differential shows 89% erythroid precursors and 4% blasts. The flow cytometry results show 20% erythroblasts (bright CD71/glycophorin A) and a smaller number of myeloblasts. The percentage of erythroblasts between the two techniques is markedly different. I have tried to explain to our Haematologist this is most likely due to the red cell lysing technique used to prepare the sample which can result in selective loss of red cell precursors (we are using ammonium chloride).
Does anyone else have experience with flow cytometry analysis of erythroleukaemia ?. How do you report the results when the erythroid cell percentages are so different between the two techniques.
Replies:
We routinely report BM differential counts on all our clinical reports as a proportion of total nucleated cell counts. This avoids the confusing situation you mentioned.
We always report a percentage of cells in each subset (lymphocytes, monocytes, granulocytes and abnormal cells) as a percentage of total nucleated cells. Therefore, the difference in results in the case like your can be easily explained to the hematologist. When the difference in % of abnormal cells is such remarkable like in erythroleukemia cases and obviously flow cytometry -based number is less accurate in this case (just because we loose some of these cells through lysis), we prefer to do an additional immunostain and report the % from morphological analysis and immunophenotype from flow.
You are completely right, nucleated red cells are lysed selectively, probably depending on their degree of haemoglobinisation. We have definately seen the degree of loss that you describe, especially in the pure erythroleukamia.
We have had a number of erythroleukaemias over the years, and we report on the blast cell population, as a proportion of nucleated cells. I would gate on CD45 if I thought there was a significant amount of debris present. It doesn't really matter as long as you define what the denominator is in your desription eg. X blasts as a proportion of non- erythroid cells. You could give both on your report if you feel it would help, and also a comment regarding "significant loss of RBC precursors due to processing and lysis".
Alternatively you could say a small blast cell population was detected in the sample, with the following phenotype... , since when using a lysis reagent on samples with lots of NRC, your precision for the % of blasts might not be that great; you could always check each tube.
The number of blasts is really irrelevant, rather, from the phenotyping point of view we are trying to define the linage of the blasts. The degree of infiltration should be taken from the assessment of the bone marrow smear (as per the WHO criteria for AL.)
If you are concerned the number of blasts being assessed is small, you can count more events to ensure you see enough blasts in your gate.
I don't have the WHO book at home but is nicely describes the types of erythroid lineage leukaemias. I always look it up when we find one, so I get it right.ie blasts with myeloid features, vs those which are part of the erythoid lineage.
At the Ohio State University Medical Center in Columbus, Ohio, we report the percentage of blasts (or CD45 dim/negative) and the percentage of lymphocytes as a percent of total events analyzed. We routinely apply a forward scatter discriminator to exclude sub cellular particles from this analysis. In addition to the discriminator we employ a "non-debris" gate in the forward scatter vs. side scatter histogram to further eliminate non-cellular material from the analysis. We frequently observe loss of erythroblasts (as well as plasma cells, large cell lymphoma, and others) due to sampling/processing artifacts (being associated with the Arthur James Cancer Hospital we see hundreds of leukemia cases). As is always the case with Flow Cytometry, the biggest challenge is correlating flow results with morphologic review.
When performing a bone marrow study we describe the abnormal population (therefore, we report what the blasts are showing and not relative to the total nucleated cell population) We report what percentage blasts we are seeing by flow and then what that population shows. Though we have talked ourselves blue in the face trying to get clinicians comfortable with the idea that we will drop the % positive values and report only whether we consider the marker positive or negative, and if positive, what the intensity of the staining is, they are still choking and gagging on the concept. Statisticians here also cannot let go of decades of % positive analyses. Thus, we report three values, positive or negative, intensity, and % positive in hopes that we can get them comfortable with the first two measures and drop the third.
Because aspirates are often hemodilute, only the manual differential count from the bone marrow aspirate smear slide should be used to determine percentages of the various cell populations. In a manual differential you can control where you are counting and you can avoid counting in hemodiluted areas. In the flow report I still report the percentages (as a % of total nucleated cells) with a statement that percentages may not be accurate due to hemodilution.
We analyze our BM specimens 3 ways when we believe there is AML-M6. First we perform CD45FITC vs propidium iodide [DNA] and obtain the CD45[-] percent of all nucleated events. If this population is aneuploid compared to the lymphocyte cluster [CD45+] and is greater than 50% of nucleated cells then AML is strongly suspected. We then perform a manual BM differential of the core biopsy and aspirate to correlate the percentages. We then perform immunophenotyping of the aspirate using density gradient removal of RBCs with Histopaque. We use Syto13 to gate on nucleated cells and then analyze for CD71, CD34, HLA-DR, and Glycophorin-A. This method gives the most accurate evaluation of the aspirate and the percentages of myeloblasts vs erythroblasts. RBC lytic methods are always suspect but if you examine the percent of lymphocytes in each method, you will get an idea of selective loss of non-lymphoid elements. Fin ally, AML-M6 must meet the W.H.O. criteria which is based upon morphology of nucleated cells. Correlation of morphology and flow will allow one to report the flow percentages
Our lab reports findings based on CD45 gating. In cases of erythroid or megakaryocytic anomalies, we report the percentage of CD45 dim cells, with their corresponding phenotype. Your reported percentage of blasts sounds reasonable for a finding consistent with the diagnosis of erythroleukemia. ( for confirmation: Are the blasts CD117 or CD34/Glyc A positive?)
The academic question is why? the difference in percentages.
Your logic is reasonable regarding the discrepancies for differing blast percentages of flow versus smear morphology. Any type of centrifugation, and/or washing and lysing are sure to take a toll on a cell suspension.
We have found differences similar to this in a few cases of marrow aspirates from myeloma patients. The smear will show 80% myeloma cells but flow will confirm only 30% CD38bright/CD117/CD56 positive cells.
One way of reporting results on leukemia flow cytometry is to give
percentage positivity of individual markers and then add a line in the
report that states "Above percentages are on a blast gate defined on a
FSC/SSC plot OR SSC/CD45 plot (whichever was used) that comprises
.........% of all events.
A good leukemia flow report - in my opinion - should include the
following :
Patient details: Name, Age, Sex, Referring doctor etc.
Sample details: Sample Id, Sample type, date/time of collection etc.
Markers used
(here one gives the percentage positivity and the details of
gating strategy used)
Markers summary
Positive:
Dim:
Negative:
Final Impression: eg Precursor B-lineage ALL (Type 1) etc.
Some labs choose not to give details of percentage positivity and give
only marker summary in the reports. This is more of a lab policy issue.
However if one does decide to give percentages, it is better to include
the details of gating strategy as well.
Flow BM samples can sometimes get diluted with peripheral blood and thus
result in discrepant blast percentage than what is observed on BM touch
smears. So, I recommend to restrict the of use flow for characterising the
blast phenotype and rely on bone marrow touch smear for evaluating the
exact percentage of blasts.
