Many phenotypes in both acute lymphoblastic leukemia (ALL) and AML are highly associated with characteristic cytogenetic abnormalities.⁹ In ALL, these include cases associated with mixed lineage leukemia rearrangements, TEL-AML1 or E2A-PBX1. In AML, the most important link is in promyelocytic leukemia with the t(15;17) translocation. Whereas lack of human leukocyte antigen-DR subregion (HLA-DR) is the best-known abnormality, only about half of cases of DR-negative AML turn out to be acute promyelocytic leukemia, and other combinations of marker expression are much more sensitive and specific for acute promyelocytic leukemia.¹⁰ AML associated with the t(8;21) translocation also shows a characteristic phenotype.¹¹

Minimal Residual Disease Detection in Acute Leukemia

Several studies have demonstrated that the presence of residual leukemic cells in the marrow of patients who are in clinical and morphologic remission is a very strong adverse prognostic factor.12–20 Although the most extensive data exist in cases of childhood ALL,^13–17,19 the principle has also been shown to apply to persons with adult ALL^21,22 and to persons with AML.^18,20

Both molecular and flow-based methods have been used to detect minimal residual disease (MRD). Flow-based assays of MRD are based on the principle that nearly all leukemias show a pattern of expression of antigens that is aberrant compared with the pattern seen in normal differentiation.15,23–29 This aberrancy can take several forms. Some leukemic cells can abnormally express antigens of a different lineage or show loss of expression of a normal lineage marker. A more common finding is expression of normal differentiation antigens, but at an intensity that is different from that expected for a particular stage of differentiation. This latter attribute makes flow MRD analysis applicable to most cases of leukemia. Nonetheless, recognizing these deviations requires a clear understanding of patterns of maturation in normal differentiation, including marrow regeneration, as viewed in multiparameter space.

The pattern of antigen acquisition and loss during B-cell maturation in the bone marrow has been very well characterized, and certain markers are particularly useful for distinguishing normal and leukemic maturation. Consideration of markers including intensity of CD45, CD34, CD10, CD58, and CD38 or aberrant coexpression of myeloid or other unexpected antigens can allow detection of as few as 1 in 10⁴ leukemic cells, even when normal B-cell precursors are present in significant numbers (Fig. 15-2). Marrow T-cell acute lymphoblastic leukemia (T-ALL) also can be distinguished from normal T cells, in part because T-ALL will typically coexpress cytoplasmic CD3 and TdT, which is never seen on any normal cell.¹⁵ However, after treatment with steroid-based therapies, T-cell leukemias may undergo therapy-induced maturation and lose expression of TdT and other markers useful for distinguishing immature and mature T cells.³⁰ Thus T-ALL MRD detection often relies on detecting the same type of phenotypic abnormalities as seen in B-cell acute lymphoblastic leukemia (B-ALL). Detection of myeloid MRD usually is a more elaborate process because of the greater phenotypic heterogeneity in AML. Certain aberrant combinations, including coexpression of CD34 and CD56, CD117 and CD15, or CD7 and myeloid antigens occur with sufficient frequency to be useful in a large number of cases.^24,28,31 Achieving 10⁻⁴ sensitivity on a consistent basis is difficult, and sensitivities are more typically closer to 10⁻³. Because abnormal populations at diagnosis may not persist at recurrence, monitoring patients with acute leukemia is best done by looking for phenotypic differences from normal, rather than for a specific abnormal phenotype present at diagnosis.

Figure 15-2 Minimal residual disease detection in acute lymphoblastic leukemia. Normal B cells have a fixed pattern of expression of different antigens as they mature. Leukemic cells depart from this normal pattern. In this case, an end induction marrow was stained by 6-color flow cytometry with the combination CD20-FITC/CD10-PE/CD38-PerCPCy5.5/CD58-APC/CD19-PE-Cy7/CD45-APCCy7, and the abnormal population was revealed by sequential gating. Only selected correlated antigens are displayed. A, CD19+ B cells are gated, and thus the great majority of the 750,000 events collected are not displayed (B, which only shows B cells). CD45-low, CD10+ cells are gated and displayed in the next panel. C, Normal B cell precursors (*green*) are recognized because they have relatively bright CD38 and relatively dim CD58, whereas the leukemic cells (*purple*) are bright CD58 and dim CD38. D, The few leukemic cells are superimposed on a CD45 versus side scatter display of the entire sample. Minimal residual disease accounted for 0.02% of the total cells.

Myelodysplasia and Chronic Myeloproliferative Neoplasms

Flow cytometry can detect abnormalities in marrow stem cell disorders other than acute leukemia. Patients with chronic myeloproliferative neoplasms (MPNs) invariably show abnormalities in either blast phenotype or maturation pattern,³² although routine clinical phenotyping of MPNs is not performed in most cases. Interest in using flow cytometry as a primary diagnostic modality in myelodysplastic syndromes is increasing. In addition to finding phenotypically abnormal blasts, many studies have demonstrated characteristic phenotypic abnormalities in either myeloid or erythroid maturation in persons with myelodysplastic syndrome (MDS), as well as describing abnormalities in granulocyte light scatter that correspond to the morphologic finding of hypogranularity.^33–41 A scoring system has been proposed to use flow cytometric characteristics to diagnose MDS, and this approach has been shown to have value in assessing the prognosis of patients.^33,37 However, consensus documents for the diagnosis and classification of MDS do not yet use flow cytometry results as a primary diagnostic criterion for MDS; rather, flow results are considered as one of several “co-criteria” to be used in patients with clinical suspicion of MDS who do not yet fulfill criteria for diagnosis using standard methods.^39,41

Chronic Lymphoproliferative Disorders and Lymphoma

Lymphoid Tissue Analysis

Flow-cytometric immunophenotyping plays a major role in the evaluation of patients with lymphoma (see Box 15-1).^44–44 Lymphoid tissue first must be disaggregated to produce a cell suspension suitable for flow-cytometric analysis. Although this task is readily accomplished in most low-grade lymphomas, high-grade lymphomas may give a nonrepresentative sample because of the greater fragility of the neoplastic cells compared with residual normal cells. Fibrosis also may make disaggregation difficult or result in disruption of neoplastic cells. Consequently, flow-cytometric analysis of lymphoma requires careful gating to ensure that the cells of interest are analyzed.

In contrast to the situation with acute leukemias, no single antibody serves as a useful surrogate for the neoplastic population. However, in B-cell malignancies, neoplastic B cells usually greatly outnumber normal cells (although there may be normal T cells), and thus use of a pan–B-cell antibody such as CD19 or CD20 can help isolate the neoplastic cells for analysis. In addition, forward scatter, as a marker of cell size, may be a very useful parameter to help distinguish frequently larger neoplastic cells from smaller normal counterparts.

Although some laboratories prefer tissue immunohistochemistry to flow cytometry for phenotyping lymphomas, the two techniques are complementary. The principle advantages of flow cytometry relate to its speed and its ability to identify and phenotype precisely the neoplastic elements in a heterogeneous sample. A complete phenotypic characterization and, in many cases, a diagnosis can be achieved within a few hours from the time of biopsy, rather than the several days needed for histopathological examination and immunostaining. Flow cytometry also is far better than immunohistochemistry for demonstrating light-chain restriction, and therefore clonality in B-cell neoplasms, and can also demonstrate phenotypic aberrancies that allow distinction between normal and neoplastic B cells.

The disadvantages of flow cytometry come from the architectural disruption created by making a cell suspension and the aforementioned possible loss of cells of interest. Hodgkin lymphoma, in particular, is not diagnosed by routine flow cytometry, although careful attention to the procedures used for phenotyping can overcome this drawback.⁴⁵ Even in non-Hodgkin lymphoma, grading may be difficult from flow-cytometric immunophenotyping alone, and precise classification can be accomplished only rarely; lymphoma classification still relies heavily on morphology, which must be correlated with flow-cytometric information.

Classification of lymphoma and grading, in particular, can be improved by assessment of DNA content by flow cytometry (as discussed later in this chapter), because S-phase fraction is closely correlated with grade. Burkitt lymphoma, which has the highest S fraction of any neoplasm, usually can be reliably identified by flow cytometry. However, in lesions such as follicular lymphoma, grading systems still are based on morphologic criteria, and although there may be reason to expect that flow-based methods might be more objective, these methods have not been validated in large series to the point at which they are generally accepted.

One of the most fruitful applications of flow cytometry in the diagnosis of lymphoma is in the analysis of specimens from fine-needle aspiration (FNA).48–48 Cells obtained by FNA already are in suspension, and flow-cytometric analysis contributes significantly to the cytopathological diagnosis of lymphoma, which is notoriously difficult. Although FNA has been used for years in the evaluation of recurrent adenopathy in patients with known lymphoma, it can serve as a primary diagnostic modality in lymphoma when combined with flow cytometry. Most non-Hodgkin lymphomas can be recognized and graded, and, importantly, lymphomas also can usually be excluded. Optimal application of this technology, however, requires close collaboration between the cytopathologist and flow cytometrist, a clear understanding of the strengths and limitations of the two techniques, and a willingness to revert to open biopsy in ambiguous cases.

Blood and Marrow Analysis

Limitations attendant on generating cell suspensions do not apply to patients with blood or marrow involvement by chronic lymphoproliferative disorders. Subclassification of these lesions relies heavily on flow cytometry.42,44,49–51 Chronic lymphocytic leukemia (CLL) is essentially defined by its immunophenotypic characteristics. Other B-cell lymphoproliferative disorders have characteristic, if not always absolutely specific, phenotypes. Mantle cell lymphoma often can be recognized in its leukemic phase, although the most specific marker, cyclin D1, is very difficult to detect by flow with current techniques. Hairy cell leukemia, conversely, has not only a characteristic but also a highly specific phenotype^42,50,52; occasionally, patients with unexplained pancytopenia can be determined to have hairy cell leukemia when only a tiny number of blood cells with the classic phenotype are identified⁵² (Fig. 15-3).

Figure 15-3 Flow-cytometric detection of hairy cell leukemia. A small population (*arrows*) accounts for less than 1% of this peripheral blood sample and can readily be recognized. Expression of CD19 and CD20 is brighter than the background normal B cells (*blue*), and the abnormal cells also express CD103 and CD25.

Flow cytometry can also detect prognostic factors in persons with CLL. Recognizing that CLL can be subdivided into a poor-prognosis type associated with nonmutated immunoglobulin V region genes and a better-prognosis mutated phenotype, several studies have attempted to identify flow surrogates. The first marker found to be prognostic was CD38, although the correlation with either mutational status or prognosis is imperfect.53,54 Zeta-chain–associated protein kinase 70 (ZAP 70) appears to be a better surrogate marker, although it too is not perfect; however, it does provide prognostic information.^57–57 However, even though this assay is widely available, particularly in commercial reference laboratories, cutoffs between positive and negative may be arbitrary, and thus significant variability in results can exist among laboratories.^58,59 Attempts have been made to improve reproducibility,⁶⁰ but it is not always certain that a particular assay will give the same results as those in the published clinical literature. Other attempts have been made to define prognostically significant subsets of patients with CLL using other combinations of markers,⁶¹ but these combinations have not been adopted in routine practice.

The primary value of flow cytometry in assessing chronic leukemias is its ability to demonstrate clonality in B-cell populations based on restricted expression of one type of immunoglobulin light chain. This feature is particularly valuable in the evaluation of patients with unexplained lymphocytosis 62,63 and also is useful for determining the stage of patients with B-cell non-Hodgkin lymphoma,^64,65 because the presence of as little as 0.5% to 1% of a clonal B-cell population, or even less, can be demonstrated in the marrow or blood of some patients. However, the high sensitivity of flow cytometry for detecting clonal populations has demonstrated small clonal B-cell populations in some patients without obvious lymphoma or leukemia.⁶⁶ Thus the finding of a small marrow clone should be interpreted with caution in the absence of other evidence of lymphoma, analogous to finding a monoclonal gammopathy in a patient without evidence of myeloma.

Flow cytometry has also been used extensively in the study of patients with myeloma. One limitation of flow cytometry is that plasma cells are underrepresented on marrow aspirates studied by flow compared with the prevalence of cells on films or in biopsies; as a result, this technique cannot be used to quantify myeloma cells and thus cannot substitute for morphology when applying traditional diagnostic criteria. However, plasma cells have a characteristic phenotype, including very bright expression of CD38 and expression of CD138,67,68 and thus are easy to recognize. With the use of membrane-permeabilization techniques, it is easy to demonstrate cytoplasmic immunoglobulin (Ig) light-chain restriction.⁶⁷ It is of more significance that neoplastic plasma cells usually have abnormal phenotypes. Although no single specific phenotypic abnormality permits distinction between benign monoclonal gammopathy and myeloma, flow cytometry can still be helpful to distinguish myeloma from monoclonal gammopathy of undetermined significance, or to predict progression of monoclonal gammopathy of undetermined significance or smoldering myeloma.^71–71 Moreover, flow cytometric assessment of abnormal plasma cells at the time of diagnosis of myeloma has been shown to have prognostic significance.⁷² In addition, detection of circulating plasma cells or persistence of an abnormal phenotype after therapy, including detection of minimal residual disease, is predictive of outcome in patients with myeloma.^{69,71,73–76}

Clonality of T-cell populations also can be demonstrated by flow cytometry, although the method is more complex and is not as widely available as that used for B cells. This technique is based on demonstration of restriction of V-beta gene use in T-cell leukemias.79–79 T-cell malignancies also often show abnormal T-cell phenotypes, which most often are characterized by loss of a normal pan-T antigen or expression of a T-cell antigen at abnormal intensity.^44,80,81 Because certain small, unusual, and even clonal T-cell populations may be seen in small numbers in non-neoplastic conditions, this technique has more limited sensitivity than demonstration of a clonal B-cell population. However, when abnormal T cells account for more than a very small percentage of cells, multiparameter flow cytometry easily demonstrates them and plays a significant role in categorizing these uncommon tumors.

Residual Disease Detection in Chronic Lymphoproliferative Disorders

As with acute leukemia, MRD can be detected by flow cytometry in chronic lymphoproliferative disorders. The most work has been done with CLL. Although detection of light-chain–restricted clones is difficult when the level is much below 0.1%, aberrant phenotypes that allow detection with a sensitivity of at least 1 in 10⁻⁴ can be detected in most cases of CLL,84–84 and standardized approaches to MRD detection in persons with CLL have been suggested.⁸⁵ Just as in acute leukemia, the presence of MRD in patients considered to be in remission by standard criteria is associated with an adverse prognosis,⁸³ whereas clearance of MRD after therapy is associated with a good outcome.⁸⁶

Solid Tumors: Analysis for DNA Content

Numerous fluorescent dyes bind stoichiometrically to cellular DNA and thus can provide an accurate assessment of DNA content of cells. In a single-parameter fluorescence histogram, tumors with abnormal numbers of chromosomes show a distinct peak separate from the normal G₀/G₁ peak of diploid cells. Such tumors are referred to as aneuploid. Moreover, cells progressing through the cell cycle show incrementally increased levels of DNA while they are in S phase, and those in G₂ or in mitosis show exactly twice the amount of DNA as that represented by the G₀/G₁ population. Thus integration of the area under the curve between the G₀/G₁ and G₂/M peaks gives the proportion of cells in S phase, commonly referred to as S-phase fraction. A variety of software packages are available to calculate S-phase fraction from these histograms.

DNA content analysis in tumors was one of the earliest clinical applications of flow cytometry. Many studies have attempted to define the prognostic significance of either ploidy or S-phase fraction in a large number of different solid tumors. The finding that fixed, paraffin-embedded tissue sections could be used for DNA analysis by flow made possible many retrospective studies on archival material for which the outcome was already known.⁸⁷ However, the literature is confusing and contradictory, and the early promise of this measurement as an important diagnostic and prognostic marker in cancer has not been realized. Although some studies have demonstrated prognostic significance to measurements of ploidy, and especially S-phase fraction, in a number of tumors—most specifically bladder, prostate, and breast cancer—many studies conflict, and as a result, this technology has not been widely embraced in clinical oncology. Although some practices use S-phase fraction to help manage patients with early-stage, node-negative breast cancer, DNA flow cytometry has largely been replaced by molecular prognostic markers.⁸⁸ One reason flow cytometry had difficulty making inroads into the clinic was the very poor interlaboratory reproducibility of DNA measurements, particularly S-phase fraction.⁸⁹ Interestingly, some studies that used highly sophisticated analytic methods to reanalyze old data suggested that in breast cancer, at least, both ploidy and S-phase fraction are powerful independent prognostic markers,⁹⁰ but it is unlikely that these techniques will be adopted in routine clinical practice. A detailed summary of all the controversies regarding DNA cytometry is outside the scope of this chapter, but several reviews have been published.^91,92