Flow Cytometric Principles in Hematopathology

The Flow Cytometer

The flow cytometer consists of three main components—fluidics, optics, and electronics—in association with a desktop computer. The fluidic system includes the pumps used to aspirate the cell suspension into the cytometer and the tubing through which cells are propelled through the cytometer. After being aspirated into the cytometer, the specimen is surrounded by a stream of buffered saline (sheath fluid) introduced into the instrument at higher pressure than the specimen, such that the cells in the specimen assume a roughly single-file position because of the phenomenon of hydrodynamic focusing. The focused stream then reaches the flow cell, a quartz cuvette where the cells are illuminated by light from one or more lasers. Ultimately, the fluid stream is directed to a waste receptacle.

The optical system includes (1) the lasers used to excite the fluorescent dyes conjugated to the antibodies used in the assay, (2) the system for conveying the laser light to the flow cell, (3) the system for conveying the emitted fluorescent light from the cells to specific detectors, and (4) the detectors themselves. The detectors are usually photomultiplier tubes (PMTs) or photodiode arrays (PDAs) that convert photons to electrical impulses. The light emitted by fluorescently labeled cells is conveyed to the detectors via a combination of dichroic mirrors that allow light of defined wavelengths to pass while reflecting light of other wavelengths and optical filters that further narrow the wavelengths reaching a detector. In newer cytometers, fiber optic cables may help to convey the emitted light to the appropriate detector clusters, whereas the light travels in air in older cytometers.

The electronics system measures the electrical impulses generated by the detectors and converts these measurements to digital information that is gathered and interpreted by the analysis software. The associated computer system directly interfaces with the flow cytometer and controls its functions. In most of the newer cytometer systems, data analysis can be done either on the computer connected to the flow cytometer or on other computers that access the data via a central server (i.e., off-line analysis).

Clinical Indications for Flow Cytometry

There are four major clinical indications for FC in the evaluation of hematolymphoid neoplasia, as defined in the 2006 Bethesda International Consensus Recommendations on the Flow Cytometric Immunophenotypic Analysis of Hematolymphoid Neoplasia. First, FC is central to the diagnosis and classification of hematopoietic neoplasms, by defining cell lineage in the chronic lymphoproliferative disorders and acute leukemias, and by becoming the gold standard for identifying paroxysmal nocturnal hemoglobinuria. The 2006 consensus recommendations also recognized FC as a useful adjunct in the workup for myelodysplastic syndromes and myeloproliferative neoplasms, including chronic myelomonocytic leukemia. Second, as antigens associated with disease prognosis in hematopoietic neoplasms are identified (e.g., ZAP-70 and CD49d expression in chronic lymphocytic leukemia/small lymphocytic lymphoma), FC has been used to measure such antigens. Third, with the advent of therapies targeted at specific antigens in hematopoietic neoplasms (e.g., anti-CD20 therapy in most B-cell lymphomas/leukemias, anti-CD52 therapy in some T-cell lymphomas/leukemias), FC has been used to identify potential therapeutic targets. Fourth, FC has been used extensively to look for and quantify residual disease after therapy including minimal residual disease, which usually is defined as less than 0.1% involvement by neoplastic cells.

Specimen Requirements for Flow Cytometry

In the clinical laboratory, specimens amenable to FC include fresh, unfixed peripheral blood, bone marrow, body fluids, and finely minced solid tissue such as lymph node biopsy or spleen. For tissue and cerebrospinal fluid (CSF) specimens, immediate immersion into a tissue culture medium, such as RPMI 1640 supplemented with 10% fetal bovine serum, 1% glutamine, and 1% penicillin-streptomycin antibiotics, is required to maximize cell viability. If the tissue or CSF is likely to reach the FC laboratory in 24 hours or less, the cells may be maintained at room temperature in such medium. If delivery to the FC laboratory is likely to take more than 24 hours, or if the specimen may be exposed to relatively high ambient temperatures as during the summer months, the sample container should be shipped with a packet of wet ice inside. Cell suspensions should never be subjected to extremes of temperature (<0°C or >37°C) and should therefore never be transported on dry ice.

Peripheral blood and bone marrow aspirates must be anticoagulated before transport to the FC laboratory. The most common anticoagulants include ethylenediamine tetraacetic acid (EDTA) and heparin, both of which allow high-quality FC evaluation, although other anticoagulants such as acid citrate dextrose (ACD) and citrate do not preclude FC. During the validation of each FC assay, it is important to confirm appropriate assay performance for all anticoagulants anticipated once the assay is brought on-line. Peripheral blood, bone marrow, and non-CSF body fluid specimens are typically transported and stored at room temperature.

Specimen Processing

Universal biohazard precautions should be taken with all specimens in the FC laboratory. Tissue that is minced to create cell suspensions should be filtered through a 40-µM or 60-µM cell strainer to remove large particles that may clog the cytometer tubing and/or bind fluorescently-labeled antibodies non-specifically. After specimens have been incubated with antibodies and washed, but before they are analyzed on the flow cytometer, they should be fixed in a dilute solution of formaldehyde or paraformaldehyde, which stabilizes antigen-antibody interactions by introducing cross-links, and also inactivates infectious agents. Adding bleach to the instrument's waste container at a final concentration of 10% and daily purging of the fluidic system with 10% bleach at the time of instrument shutdown further minimize biohazard risks.

A zero-tolerance policy should exist for specimen mix-ups in the FC laboratory to minimize the chance that incorrect patient data are reported. Procedures to minimize the possibility of mix-ups can include fastidious labeling of specimen containers and associated paperwork, double-checking of all specimens at multiple points during processing, and separate processing of specimens to minimize the risk of cross-contamination.

In any specimen containing a large amount of peripheral blood, the erythrocytes should be removed before introducing the specimen into the flow cytometer. Most clinical FC laboratories lyse the red blood cells at some point during specimen processing, using either a commercially available reagent or a homemade ammonium chloride solution. In many laboratories, antibodies are incubated with the cells in the presence of erythrocytes (the so-called whole blood technique). After incubation, the erythrocytes are lysed and the bound antibodies are fixed to the cells of interest at the end of processing, just before the cell suspension is evaluated on the flow cytometer. An alternative method is up-front bulk lysis of the red blood cells, which is less time efficient than the whole blood method and has the potential to lyse some leukocytes in addition to erythrocytes. Up-front bulk lysis is best avoided in specimens with small numbers of leukocytes, such as CSF and scanty tissue biopsy specimens. Note that erythrocyte lysis techniques, when applied to bone marrow specimens, destroy the great majority of the nucleated erythroid precursors, compromising the ability to evaluate these cells.

The majority of antigens currently evaluated in clinical FC are cell surface associated, but nuclear antigens (e.g., terminal deoxynucleotidyl transferase [TdT] in lymphoblasts) or cytoplasmic antigens (e.g., myeloperoxidase in myeloid blasts) also can be evaluated by FC. Table 23.1 contains a list of antigens commonly evaluated in the workup for hematolymphoid neoplasia. When both cell surface and cytoplasmic antigens are evaluated in the same assay, the surface staining is performed first. The cells are then fixed and permeabilized with reagents for these purposes and then stained for the cytoplasmic antigens.

Instrument Configuration and Quality Control

The type of flow cytometer in the laboratory dictates the number of antigens evaluated simultaneously. Single-laser flow cytometers can evaluate three to five antigens simultaneously in addition to the generic light-scatter properties of forward scatter (proportional to cell size) and side scatter (also known as orthogonal or 90-degree light scatter and proportional to cytoplasmic abundance or granularity). Two-laser instruments can usually evaluate six to eight antigens simultaneously, depending on the detector configuration, whereas three-laser instruments are typically required to evaluate nine or ten antigens simultaneously. The feasibility of nine- and ten-color FC for leukemia–lymphoma immunophenotyping has been demonstrated, and this technology is currently being used by a growing number of clinical laboratories. The simultaneous assessment of such a large number of antigens minimizes the number of tubes of cells and antibodies that must be set up, and it is of particular benefit for the analysis of scanty specimens. In addition to saving technologist time, which is typically the most valuable commodity in the FC laboratory, nine- and ten-color FC offers additional cost savings by minimizing redundancy of antibody and other reagent usage across tubes and by maximizing the efficiency of flow cytometer use.

Regardless of the type of flow cytometer, quality control (QC) measures must be performed and documented on a daily, weekly, and monthly basis to ensure optimal instrument performance. In our FC laboratory, daily QC usually employs brightly fluorescent, 4- to 6-µM plastic microbeads to help ensure that the voltages allotted to the individual detectors are adequate to detect the expected level of fluorescence and to confirm adequate laser power and alignment. The overriding principle for optimizing detector voltage is to maximize the signal-to-noise ratio for that detector. A second daily QC function in our laboratory is a standard nine-color T-cell analysis assay on a commercially available, well-characterized, stabilized whole blood preparation, and confirmation that the proportions of the various cell populations fall within the published ranges for the preparation. Less frequent QC measures include (1) confirming linearity of detectable fluorescence by using a series of microbeads having known fluorescence properties ranging from negative to very bright; (2) confirming the reproducibility of fluorescence in multiple replicate assays of a single specimen; and (3) confirming an acceptably low level of specimen carryover from one tube to the next. Less than 0.1% carryover should be sought for standard FC assays, and less than 0.01% carryover for minimal residual disease assays, which is usually achievable if an adequate amount of blank sheath fluid is run through the cytometer between the collection of individual tubes of cells and antibodies. Annual or semiannual preventive maintenance by a service representative of the cytometer manufacturer should be performed and documented.

Antibodies: Compensation and Panel Design

Each of the fluorochromes used in FC has a well-characterized absorption and emission spectrum that extends over a range of wavelengths ( Fig. 23.1 ). The simultaneous use of multiple antibodies conjugated to different fluorochromes therefore results in some degree of spillover, in which a portion of the fluorescence from a given fluorochrome is detected by a detector targeted for a different fluorochrome. As a result of spillover, the fluorescence detected by each detector actually represents the sum of the fluorescence from multiple fluorochromes. The majority of detected fluorescence almost always comes from the fluorochrome the detector was designed to detect, but significant contributions may come from other fluorochromes because of spillover. To adjust for spillover, a mathematical correction known as compensation (or color compensation ) is applied routinely to all multiparametric FC data. As more antibodies are used in individual tubes (e.g., 6- to 10-color analysis), the potential for compensation artifacts increases. A detailed discussion of compensation is beyond the scope of this chapter, but it is important to recognize that proper compensation is another critical QC function in the FC laboratory.

In clinical FC, antibodies are typically used in defined combinations, or panels, to answer specific questions about specific cell populations. Most laboratories performing leukemia–lymphoma immunophenotyping have an acute leukemia panel to distinguish acute myeloid leukemia from acute lymphoid leukemia, and a lymphoma panel to distinguish benign from malignant lymphoid proliferations. At a minimum, antibody panels should measure sufficient antigens to distinguish normal–benign from abnormal–neoplastic cell populations with a high degree of sensitivity and specificity. See Table 23.2 for a list of antigens important in the evaluation of specific hematopoietic cell populations, many of which are included in the published recommendations from the 2006 Bethesda International Consensus Conference. For detailed discussion on FC assay validation, see the 2013 series of consensus documents published jointly by the International Council for Standardization of Haematology and the International Clinical Cytometry Society.

With the expanded utilization of targeted antibody therapy against cell lineage-associated surface antigens, it is important to remember that therapeutic antibodies often prevent diagnostic FC antibodies from recognizing their cognate antigens. For example, is it very common for clinical FC laboratories to receive samples from B cell non-Hodgkin lymphoma patients previously treated with anti-CD20 antibody therapy, such as rituximab. This prior therapy creates the possibility that a neoplastic B cell population will not be identified by an anti-CD20 FC antibody due to masking of the antigen by the therapeutic antigen, or due to selection of a CD20-negative neoplastic subclone. When one designs a new FC assay, it is therefore imperative to be aware of potential targeted therapies that may increase the likelihood of a false-negative result, and to design the assay to minimize the possibility of a false-negative, e.g., by adding anti-CD19 and anti-CD22 antibodies to a B cell-directed assay.

The quality of FC data not only depends on the specific antigens evaluated, but also on the specific fluorochromes conjugated to specific antibodies and on the ways in which antibodies are used together. For example, when a weakly expressed antigen is sought, such as an aberrantly expressed lymphoid antigen on myeloid blasts, conjugation of the relevant antibody to a bright fluorochrome, such as phycoerythrin, can maximize the chance of detecting the antigen. Conversely, it is often unwise to use a bright fluorochrome to detect a strongly expressed antigen, because the bright fluorescence emission from this strategy will likely create compensation problems resulting from spillover.

Clinical FC Test Ordering

When a specimen is received in the FC laboratory, the pathologist or FC technologist must consider the underlying clinical question in deciding which antibodies and panels to use in the evaluation. When the clinical question is limited, such as whether a bone marrow aspirate from a patient with known B-cell non-Hodgkin lymphoma has evidence of disease, then it is appropriate to perform a limited study focused primarily on the cell population in question, such as the mature B cells. If such a limited evaluation is to be performed, it is important to have an antibody to the leukocyte common antigen CD45 in the assay, so that unexpected expansions of cell populations warranting additional FC evaluation, such as blasts or monocytes, will be recognized. When the clinical question is much broader, for example, ruling out a hematolymphoid neoplasm in the bone marrow of a patient with pancytopenia and no known malignancy, then a broader evaluation of the myeloid, lymphoid, and plasmacytic lineages is likely to be appropriate. If there is clinical concern for a lymphoproliferative disorder in a patient without a prior diagnosis, it is prudent to evaluate the B, T, and natural killer (NK) cells whenever possible. However, because B-cell lymphomas are much more common than T-cell or NK-cell lymphomas (particularly in Western patient populations), it is reasonable to rule out a B-cell malignancy before evaluating the T cells if a limited amount of specimen is available.

Data Acquisition

The overall data collection process by which antibody-stained cells are propelled through the flow cytometer, illuminated by the lasers, and detected by the optical system is known as acquisition . The number of cells (also known as events ) required for evaluation depends on both the purpose of the flow cytometric assay and on the nature of the specimen. For example, in the evaluation of a lymph node that is replaced by lymphoma, a relatively low number of viable cells (e.g., 10,000) can be acquired per aliquot (or tube) of cells and antibodies, because this relatively low number of cells will allow adequate characterization of the neoplastic population. In contrast, the characterization of a low-frequency cell population (e.g., myeloid blasts in most bone marrow specimens) requires a much larger number of viable cells (e.g., 100,000) to be collected. In our FC laboratory, if adequate viable leukocytes are available, then we routinely collect at least 100,000 viable leukocytes per tube for initial diagnostic specimens. On the other hand, when we look for minimal residual disease in blood or bone marrow, we make every effort to collect 500,000 to 1,000,000 viable leukocytes depending on the specific assay, to increase the sensitivity of the assay. Assuming that the presence of 50 neoplastic cells in an aliquot will enable confident identification of this population, the evaluation of 500,000 total viable cells offers the ability to detect the neoplastic population at a frequency of 1 in 10,000 cells, or 0.01%.

Data Analysis

FC data analysis requires specialized computer programs designed for this purpose. Such programs are available for both PC and Macintosh platforms, and they ideally permit adequate evaluation of both QC and specimen data. At a minimum, such programs should be able to perform compensation, generate two-dimensional histograms (also known as two-parameter dot plots or scatter plots ) of the data, and enumerate the various cell populations of interest for reporting purposes.

The process of targeting the analysis to the cell populations of interest is known as gating , described in detail in Figure 23.2 . In our own laboratory, the gating of FC data always proceeds as follows. First, assess the quality of specimen acquisition via the time parameter ( Fig. 23.2A ). Second, exclude multicell aggregates (two-cell doublets and higher order aggregates) from the analysis ( Fig. 23.2B ), because flow cytometric evaluation must be restricted to individual leukocytes (i.e., singlets). Third, exclude nonviable cells from the analysis. Because nonviable cells typically have a marked decrease in forward scatter (FS) because of membrane damage from cellular degeneration, FS-versus–side scatter (SS) gating (see Fig. 23.2C ) is a simple and effective way to exclude nonviable cells. FS-versus-SS gating also excludes many nonlysed erythrocytes because of their small size. An alternative method for removing nonviable cells from the analysis is the addition of a DNA-binding dye, such as 7-amino-actinomycin D (i.e., 7-AAD, which is well excited by 488-nm lasers) or 4′,6-diamidino-2-phenylindole, dihydrochloride (i.e., DAPI, which is well excited by 405-nm lasers), to the cell suspension just before acquisition on the flow cytometer. These dyes penetrate the damaged plasma membranes of nonviable cells but are excluded from viable cells. Cells demonstrating the characteristic fluorescence of these dyes are presumed to be dead or dying and are excluded from further evaluation by gating ( Fig. 23.2D ). Fourth, separate the cells according to broad cell type. Because viable lymphocytes, monocytes, and granulocytes typically show reproducible differences in their combined FS and SS characteristics, FS-versus-SS gating can be used effectively to separate these three cell populations in peripheral blood. This gating strategy becomes less effective when applied to bone marrow, because one particularly important bone marrow population—the blasts/progenitors—has FS-versus-SS characteristics that overlap both the lymphocytes and monocytes. As a result, many laboratories use CD45-versus-SS gating to separate the various bone marrow cell populations, including the blasts/progenitors (see Fig. 23.2E ). The fifth gating step, which is particularly useful when looking for a mature B-cell or T-cell lymphoproliferative disorder, is lineage-specific gating (see Fig. 23.2F and G ). In this gating strategy, evaluation of a pan–B-cell antigen, such as CD19, or a pan–T-cell antigen, such as CD3, restricts the analysis to these lymphoid populations. An inherent limitation of lineage-specific gating is that B- or T-cell neoplasms with aberrant loss of CD19 or CD3, respectively, will not be identified by this gating procedure. Therefore a more generic gating step, such as CD45-versus-SS gating to identify all the lymphocytes, which can then be evaluated for aberrant antigenic loss, should always be used in conjunction with lineage-specific gating. A sixth potential gating step, which we always perform on the data from our two 10-color FC tubes focussed on myeloid maturation, is to exclude unlysed erythroid precursors from the analysis ( Fig. 23.2H ). This step is particularly useful in the setting of significant erythroid hyperplasia in the marrow, in which invariably incomplete lysis of erythroid precursors during specimen processing can results in large numbers of early erythroid precursors, particularly proerythroblasts, contaminating the blast/progenitor gate.

A variety of antigenic abnormalities may be observed during the FC evaluation of malignant hematopoietic cell populations. These abnormalities typically contrast with the highly regular and reproducible patterns of antigen expression seen in benign hematopoietic cell populations and include: (1) abnormal increases or decreases in the levels of expression of antigens normally on the cells of interest, including complete loss of expression (e.g., aberrant loss of CD7 on the neoplastic CD4 ⁺ T cells of mycosis fungoides–Sézary syndrome); (2) abnormally homogeneous expression of antigens that normally show coordinate variation in expression in a population of interest (e.g., abnormally homogeneous expression of CD34 and CD38 on myeloid blasts in myelodysplasia or acute myeloid leukemia); (3) asynchronous antigen expression in which the timing of antigen expression during a maturational process is abnormal (e.g., asynchronous expression of CD13 and CD16 during neutrophil maturation in myelodysplasia); and (4) aberrant expression of nonlineage antigens (e.g., aberrant expression of the T-cell–associated antigen CD7 on leukemic myeloid blasts). To appreciate these abnormalities, the flow cytometrist must thoroughly understand the normal patterns of antigen expression in the cell populations of interest.

Thought should be given to the numeric scale used to display each FC parameter. FS is typically presented on a linear scale. SS may be presented on a linear or a log scale, with the latter offering a relatively compressed view of the data. Antigen-associated fluorescence may be presented on a traditional four-logarithm scale or, as some authors have recommended in recent years, on a modified log scale (so-called logicle display) that does not alter the actual data values or summary statistics computed from the data. Functionally, logicle display employs a hyperbolic sine function that transforms the appearance of the compensated FC data by centering antigen-negative and low-positive populations on a linear scale around a “0” point on the relevant axis while retaining logarithmic scaling for antigen-positive populations with moderate- to high-level positivity. In many cases, logicle scaling improves one's certainty about the nature of negative- and low-positive populations.

Data Reporting

Clinical leukemia/lymphoma/myeloma immunophenotyping data are typically reported in one of two ways. The less informative way is to report the percentage of cells in the population of interest that is considered positive for each of the antigens evaluated. Because such lists of antigens do not provide a unifying description of the abnormal cell populations in the specimen, a much more useful way to report FC data is to describe the detailed immunophenotype of each abnormal cell population in a free-text format in the FC report, including the proportion each population represents of the total viable cells. In describing levels of antigen expression associated with abnormal cell populations, our preferred terms are high-level and low-level instead of bright and dim, because the latter two terms describe levels of fluorescence associated with bound antibodies, which are only surrogates for actual levels of antigen expression.

In addition to describing the detailed immunophenotype of any abnormal cell populations, it is useful for clinical FC laboratories to report the following basic parameters for each specimen: (1) a CD45-versus-SS-based different count of the relevant viable leukocyte populations in the specimen (i.e., the proportions of mature lymphocytes, monocytes, granulocytes, blasts [including both immature B cells and myeloid blasts], and plasma cells); (2) among the mature lymphocytes, the proportions of B cell, T cells, and NK cells; (3) among the mature B-lymphoid cells, the kappa-to-lambda ratio, which is usually a surface kappa-to-lambda ratio for mature B cells and a cytoplasmic kappa-to-lambda ratio for plasma cells; and (4) among the T cells, the CD4-to-CD8 ratio.

Introduction

B-lymphoid neoplasms are best understood in the context of normal B-cell maturation in the marrow ( Fig. 23.3 ). Precursor B-cell neoplasms correspond to the pro–B-cell and pre–B-cell, or B-lymphoblast, stages. Mature B-cell neoplasms correspond to the early–naive (pregerminal center), germinal center, and postgerminal center B-cell stages. Plasma-cell neoplasms correspond to the terminal stage of B-cell maturation.

Precursor B-Lymphoid Neoplasms

According to the 2008 WHO classification (and 2016 update), precursor B-cell neoplasms are referred to generically as B-lymphoblastic leukemias–lymphomas (B-LBL); under the 2001 WHO classification, these were called precursor B-lymphoblastic leukemias–lymphomas . Most cases develop as leukemias involving the bone marrow and blood, and lymphomatous presentation is rare. B-LBL subtypes with a favorable prognosis include those bearing the t(12;21)(p12;q22) or hyperdiploidy with greater than 50 chromosomes. Subtypes with an unfavorable prognosis include cases having the t(9;22)(q34;q11), the t(1;19)(q23;p13), or the t(4;11)(q21;q23). Virtually all B-LBLs express the B-cell–associated antigen CD19, in addition to human leukocyte antigen (HLA)-DR and the lymphoblast-associated antigen TdT ( Fig. 23.4 ). Most B-LBLs express CD10 (the common acute lymphoblastic leukemia antigen, or CALLA) and the blast-associated antigen CD34, and most lack surface light chains. CD45 expression is typically low to occasionally negative.

In addition to yielding prognostic information, certain karyotypes have characteristic immunophenotypes. The t(12;21) is associated with relatively low-level CD9 and no expression of the mature B-cell antigen CD20. Hyperdiploid cases frequently lack CD45. The t(9;22) is associated with aberrant expression of myeloid antigens, such as CD13 and CD33, and occasional lack of CD45. The t(1;19)(q23;p13) is associated with a lack of CD34. The t(4;11) is associated with loss of CD10 expression (CALLA-negative pre–B-ALL), lack of CD20, and frequent aberrant expression of the myeloid-associated antigen CD15.