Immunophenotyping is complementary to morphology in the contemporary practice of marrow cell identification. Flow cytometers use similar principles to the automated hematology analyzers discussed in Chap. 2, with the notable difference that fluorescence-labeled monoclonal antibodies directed toward cluster of differentiation (CD) antigens are the primary diagnostic tool. As described in the World Health Organization classification of hematologic malignancies,72 immunophenotypic data (expression of cell surface, intracytoplasmic, and nuclear antigens) are key determinants of diagnosis and classification of hematopoietic malignancies. The principle of immunophenotyping is to diagnose and follow neoplastic cell populations by virtue of differential patterns of protein expression. Only the basic principles of flow cytometry analysis are described in this chapter, so that the reader has the basis for understanding the phenotypic characteristics associated with the hematopoietic disorders described in greater detail in other chapters of this book.
Flow cytometers are automated hematology analyzers that use principles of light scatter and fluorescence to define cellular populations in which to analyze expression of proteins typically identified by fluorescent tagged antibodies. A single-cell suspension is aspirated into a laminar flow of isotonic diluent that passes in front of one or more laser beams. Light scatter and fluorescence data are collected using specific photomultiplier tubes with appropriate filtration to collect scattered light (same wavelength as the incident laser light) or fluorescence emitted light (at a longer wavelength determined by the dye used). Multiple detectors with different filtration coupled with single or multiple lasers are used to collect highly multiplexed data. As with automated hematology analyzers, light scatter information is collected at a low angle (correlates with cell size) and 90-degree angle (correlates with cellular granularity and nuclear complexity; Chap. 2, Fig. 2–1). The latter measurement is especially useful in separating developing myeloid progenitors, monocytes, and mature granulocytes from lymphoid cells and blasts.
Immunophenotyping can be achieved by using monoclonal antibodies specific to certain cell surface proteins, most of which have CD designations as defined by international workshops. A primary requirement for flow cytometry analysis is that cells must be viable and in single-cell suspension prior to staining, which is why this method is used largely for hematopoietic malignancies and immunologic disorders, and not for analysis of solid tumors. This consideration also explains differences in results between flow cytometry and morphologic or immunohistochemical observations when samples with highly adherent neoplastic cells are analyzed. For instance, in multiple myeloma or large cell lymphoma the proportion of malignant cells is typically lower (or absent) by flow cytometry compared with marrow biopsy. In a well-equipped and appropriately staffed clinical laboratory, preliminary information often can be provided within 3 to 4 hours after the initial sample collection, thereby facilitating institution of appropriate therapy (e.g., in the case of newly diagnosed acute leukemias).
Clinical laboratories typically use four- to six-color analysis, plus side and forward light scatter, for routine diagnostic panels. For research studies, simultaneous analysis of up to 20 simultaneous fluorochromes is possible, by excitation with up to 5 lasers and separate collection of the emitted light produced by interaction with each laser. At present, routine use of that many simultaneously measured markers is not necessary for clinical diagnosis. An important consideration is that analysis with more simultaneous colors places greater demand on resources for development, maintenance and ongoing quality assurance. Most clinically important phenotypic markers are analyzed as cell surface proteins by directly adding conjugated antibodies to cell suspension, followed by washing and lysis of red cells.73 Assessment of intracytoplasmic and nuclear-associated proteins is accomplished after staining for surface makers by then fixing cells in suspension and adding the relevant antibodies in conjunction with a membrane-permeabilizing agent. Some lineage-specific markers (CD3 in precursor T cells; CD79a and CD22 in B cells; myeloperoxidase in granulocyte lineage; cyclin D1 in mantle cell lymphoma) are expressed only in the cytoplasm at certain stages of development. Fluorescence and light scatter data are stored electronically as list mode data files that can be archived and later reanalyzed using appropriate software. As the number of parameters collected on individual cells increases, standard ways of looking at multiple two-parameter histograms of gated cell populations become more difficult. Data analysis techniques and automation appropriate to discovery and interpretation of multidimensional data sets such as those generated by various “-omics” analyses may become part of the multiparameter flow cytometry workflow.74,75 Computational methods for identifying cell populations in highly multidimensional data sets have been shown to be more effective in reliable and consistent identification of clinically relevant cell populations in multicolor flow cytometry than manual gating and analysis,76 particularly in the context of a consensus approach using an ensemble of algorithms, as is commonly done today in weather forecasting.
In heterogeneous specimens such as marrow, in which the relevant clinical population (such as blasts) may be a minor population overall, a strategy for specifically identifying the population(s) of interest is necessary. As discussed in Chap. 2, this is accomplished for blood cells by very complex cluster analysis using multiple physical parameters. Because the flow cytometer has a much more sophisticated analytical capability at the back end with the fluorescent markers, the front-end selection of cells is not intended to be definitive, but should include the cells of interest and exclude nonrelevant cells, particularly those that may create an interpretive problem if included in the analysis. This process, referred to as gating, is typically accomplished by a combination of CD45 (common leukocyte antigen) and 90-degree light scatter (side scatter). As shown in Figure 3–4, lymphocytes, monocytes, myeloid precursors, and blast cells can be reasonably distinguished in marrow using this method. It is important to exclude monocytes, if they are not the cells one wishes to phenotype, as they express high-affinity Fc receptors that nonspecifically bind antibodies and may cause false-positive fluorescence signals. Individual lineages, such as eosinophils, basophils, and neutrophils, or stages of neutrophilic maturation, are not distinguished as automated hematology analyzers do for blood, but this is not necessary for the diagnostic questions usually asked by flow cytometry. The “blast gate,” defined by dim CD45 expression and low to intermediate side scatter, is a helpful region within which to identify and phenotype blast cells using more specific markers (only a minority of cell in this gate may be blasts,77 but many cells with confounding immunophenotypes are excluded). Care must be taken to look for cells with unusual light scatter patterns not fitting in the usual “gates” to make sure the abnormal cells are not “hiding” in these regions. In particularly complex clinical circumstances, several fluorescent markers can be used just to identify a rare or subtly defined neoplastic subset, which can then be more definitively phenotyped in additional tubes containing those “backbone” markers to define the cells of interest plus additional markers to phenotype them. This strategy benefits from the ability to simultaneously measure up to 8 fluorescent markers in currently available clinical systems.75 For samples with low cell viability, gating strategies based on light scatter and/or vital exclusion dyes, such as 7α-actinomycin-D, to limit analysis to the viable cell population only, may be used.73 Strategies are commonly used to exclude cell doublets, for instance, based on the relationship of the pulse width (duration of signal) to pulse height of the forward light scatter signal. Immunocytochemistry of a marrow biopsy specimen is the preferred method for phenotyping solid tumors and is highly complementary to flow cytometry in diagnosis of lymphoid infiltrates.
Flow cytometry examples: A. Normal marrow showing CD45 versus side scatter, which identifies major cell populations as indicated. B. Acute lymphoid leukemia, in which an expanded blast population is evident in the CD45 versus side scatter histogram (shown in green). Those cells with dim CD45 and negative side scatter (green) are then gated, so that expression of cell markers on this population only can be analyzed, as shown in the three histograms to the right, where the population is shown to be CD19+/CD79a+ (B cell), terminal deoxynucleotidyl transferase (TdT)+ (immature lymphoid), and CD3− (not T cell), hence B-precursor lymphoblastic leukemia. C. Chronic lymphocytic leukemia (CLL), in which an expanded lymphocyte population is evident on the CD45 versus side scatter histogram (shown in red), with coexpression of CD5 and CD19 (consistent with CLL), and expression of only surface immunoglobulin light-chain κ isotype on the CD5+ cells, showing that the population is monoclonal.
COMMON FLOW CYTOMETRY APPLICATIONS IN HEMATOLOGY
Often the diagnostic question involves characterizing an expanded blast population or detection and analysis of a clonal lymphoid population (see Fig. 3–4). These determinations can be achieved by examining lineage-specific or maturation stage-specific markers. For instance, in marrow, immature cell populations can be identified by expression of antigens such as CD34 and CD117. In some instances, the stage of differentiation can be determined using combinations of markers that are expressed only during certain phases of differentiation (e.g., dual expression of CD4 and CD8 in an immature T-precursor population). Stage-specific phenotypes are sometimes valuable clues to clinically relevant diagnoses, such as the characteristic lack of human leukocyte antigen-D related (HLA-DR) expression in promyelocytic leukemia, a phenotype that mimics the normal loss of this antigen in promyelocytes. Among lymphoid leukemia/lymphomas, chronic lymphocytic leukemia (CLL)/small cell lymphoma, mantle cell lymphoma, hairy cell leukemia, and B- or T-precursor lymphoblastic leukemia, among others, have distinctive immunophenotypes. Methodologic advances such as multicolor analysis (6 or more simultaneous fluorescence colors) have allowed flow cytometers to detect and analyze diagnostically important rare subpopulations such as Reed-Sternberg cells in classical Hodgkin lymphoma.78 Aberrant phenotype combinations suggestive of malignancy, such as coexpression of high levels of CD56 or CD117 on plasma cells in plasma cell myeloma, or loss of the T-cell markers CD7 or CD26 in a mature phenotype T cell in T-cell lymphoma, are diagnostically useful. Expression intensity is a frequent clue to diagnosis; for example, the weak surface immunoglobulin and weak CD20 expression in CLL. In reporting flow cytometry immunophenotyping results, the summary immunophenotype of the relevant population(s) should be described, rather than simply a listing of the percentage of cells positive for each marker, with subpopulations noted as observed.
Immunophenotyping of marrow by flow cytometry is a useful adjunct to established morphologic and cytogenetic criteria in diagnosing MDS,79 and is predictive of later development of overt MDS in the diagnostically challenging group of cytopenic patients with initially morphologically equivocal marrow findings.80 Abnormalities in MDS marrow include an increase in the percentages of CD34+ cells even when blasts are not morphologically increased; decreased CD34+ B progenitors; aberrant antigen expression by myeloid progenitors, mature granulocytic cells, or monocytes; and decreased side scatter of mature granulocytic cells. A simplified scoring system (so-called Ogata score) has been validated for use in an interlaboratory study,81 and international guidelines for a more extensive scoring panel have been published by the International/European Leukemia Net Working Group for Flow Cytometry in MDS.82 Immunophenotyping may provide therapeutically relevant prognostic information independent of existing risk factors incorporated in the commonly used (and recently revised) International Prognosis Scoring System (IPSS) score.83
Clonality of immunoglobulin-expressing B-cell malignancies involving marrow (e.g., CLL and lymphoplasmacytic lymphoma) can be determined by simultaneous assessment of surface κ and λ immunoglobulin light-chain expression on the surface of B cells, often in combination with a characteristic neoplastic immunophenotype, such as CD5 expression in CLL B cells. Technical considerations are important to minimize nonspecific binding of serum monoclonal immunoglobulins to the surface of lymphocytes. Cytoplasmic κ and λ identification, in conjunction with aberrant surface immunophenotype, can also be useful in establishing the clonality of plasma cell neoplasms in marrow. T-cell clonality is not as easily demonstrated, because there are dozens of Vβ specificities expressed by T cells, and antibodies exist only to identify approximately 70 percent of these. Analysis of Vβ repertoire of the αβ T-cell antigen receptor in cells with an atypical immunophenotype, can identify clonal populations of T cells at diagnosis and posttherapy,84 but this methodology is less-routinely used in clinical laboratories.
Presence of minimal residual disease (MRD) measured by either detection of a molecular target using PCR or aberrant immunophenotype using multicolor flow cytometry is increasingly used as a prognostic marker, often in the setting of clinical trials, in a variety of hematopoietic neoplasms, including acute leukemia, plasma cell myeloma, and CLL. In some settings, such as chronic phase CML, MRD monitoring by molecular assay is standard practice. In others, both flow cytometry and molecular methods for MRD detection are used successfully, each having their own advantages and limitations.85 Flow cytometry assays for MRD are considerably more complex than standard diagnostic phenotyping and need to be designed for this specific purpose,86 with collection of large numbers of events, multicolor strategies for detecting different types of subtle aberrant phenotypes, consistent data interpretation protocols, and confirmation of absence of “background” cells in noninvolved (but otherwise comparable; e.g., posttherapy) marrows expressing each aberrant phenotype to be tested. Immunophenotypic evidence of MRD can be sought by detection of specific leukemia associated immunophenotype, or by observation of a population in multidimensional space shifted significantly from any corresponding normal population (“different from normal” approach). Immunophenotypic shifts can occur during treatment, so identification of more than one aberrant phenotype to be tested is advisable when possible. Interpretation of MRD assays is dependent on type of neoplasia, timing of testing, treatment regimen, and standardization of assay methods. Standardization of flow cytometry MRD assays is a challenge that will need to be resolved as these assays move into more routine clinical practice. In acute leukemia, where MRD testing by flow cytometry has been most intensively studied, the assays provide posttreatment prognostic information, but the translation of this information into therapeutic decision making based on MRD risk stratification is further advanced in pediatric leukemia and acute promyelocytic than in adult acute myelogenous leukemia (AML).87 MRD detection by either flow cytometry or PCR is standard clinical practice in pediatric acute lymphocytic leukemia (ALL),85,86 where nearly all patients have leukemia associated targets suitable for either molecular or flow cytometry MRD assays. Detection of MRD in AML is more challenging because half of patients lack a molecular target suitable for MRD testing, but the majority of AML patients have a leukemia-associated phenotype that can be detected at the 0.1 percent level or below by multicolor flow cytometry. MRD detection in chronic lymphocytic leukemia (CLL)88,89 and plasma cell myeloma90 is also possible by flow cytometry and molecular techniques, and is becoming a relevant clinical issue now that therapeutic options in these malignancies are rapidly improving.
Flow cytometry is used to enumerate CD34+ progenitors when evaluating the adequacy of blood stem cell collections (Chaps. 23 and 28), with several routine assay kits available.91 Flow cytometry offers the potential to incorporate other markers defining clinically relevant progenitor and stem cell subpopulations.92 Lymphocyte subset quantitation is diagnostically important in acquired and congenital immunodeficiency states. Flow cytometry analysis of glycosylphosphatidylinositol (GPI) linked proteins in multiple blood cell types is the gold standard for diagnosis of paroxysmal nocturnal hemoglobinuria (Chap. 40).93 Flow cytometry detection of paroxysmal nocturnal hemoglobinuria (PNH) clones is facilitated by using FLAER (fluorescently labeled inactive variant of the bacterial protein aerolysin), which binds to all (GPI) linked structures thus sensitively detecting GPI-linked protein expression in multiple cell lineages. Guidelines for standardized PNH flow assays have been published94 and these assays are readily set up in clinical laboratories.