Going with the flow in
  hematopathology

November 2004
Cover Story

William Check, PhD

New information is continually emerging about hematopoietic neoplasms, information that is quickly incorporated into tests for diagnosis, classification, prognosis, and residual disease monitoring. “There has been not so much an increase in the total volume of testing, but a wider palette of tests to choose from,” says Brent Wood, MD, PhD, director of the hematopathology laboratory and associate professor of laboratory medicine at the University of Washington. Dr. Woods metaphor suggests that a pathologist is a visual artist, skillfully blending morphology, flow cytometry, immunohistochemistry, karyotyping, FISH, and molecular methods to produce a clinical picture. With current four- and five-color flow instruments, as well as IHC stains, multi-dye FISH kits, and SYBR green and orange for real-time PCR, laboratory work on hematopoietic neoplasms does produce many colorful images.

One might also conceive the pathologist in charge of working up a suspected leukemia or lymphoma as an orchestra conductor, directing each laboratory instrument to make its contribution at the right time to produce the desired result. “If you understand the performance characteristics of different assays and the specificity of phenotypic findings, you have a wider variety of tools at your disposal,” says Dr. Wood, whose laboratory does all of these tests except cytogenetics.

Like a musical maestro, the pathologist—whether it is a hematopathologist for a blood or bone marrow sample or an anatomic pathologist in the case of a lymph node—has responsibility for the outcome. “In a lot of places every laboratory issues a separate report and the burden of combining them may fall on the hematologist,” says Daniel Arber, MD, director of clinical hematology and flow cytometry and associate director of molecular pathology at Stanford University. "But interpretation of laboratory data really should be done by a pathologist."

Interpreting multiple test results demands expertise and experience. Karen Mann, MD, PhD, associate professor of pathology and director of molecular hematopathology at Emory University, describes one way of handling this task. "At Emory there is a pool of us who sign out molecular, flow, and morphology, so we have access to all the data," she says. Combining these results will usually provide a diagnosis. Hematopathology training programs incorporate all these modalities, Dr. Mann notes.

Among available tests, flow cytometry’s role might be likened to the string section—the foundation of the orchestra. Bruce Davis, MD, president of Trillium Diagnostics and associate director of clinical research at Maine Medical Center Research Institute, Scarborough, expresses the prevailing view: "Flow cytometry is an important adjunct to morphologic diagnosis in the classification of any hematolymphoid malignancy." Dr. Wood adds, "Flow cytometry is really useful in all categories [of hematopoietic neoplasms] for diagnosis and for therapeutic monitoring."

John Carey, MD, vice chair of clinical pathology at Henry Ford Hospital and Health Care System, Detroit, seconds Dr. Davis’ call for morphology as a prerequisite for efficient flow cytometry. "Flow cytometry should not be routinely done as a single panel to detect a broad range of mature and immature hematolymphoid neoplasias," Dr. Carey cautions. "It’s essential to do morphologic triage first."

Dr. Wood points to another prerequisite to doing optimal flow cytometry—clinical knowledge. "For every specimen, the panel is based on the clinical picture," he says. For a patient with a history of B-cell lymphoma, Dr. Wood might run a fairly short panel, perhaps eight antibodies. For an anemia of unknown cause, on the other hand, he would run a comprehensive panel that might contain 26 antibodies. "We encourage clinicians to provide some idea of what type of test they want to pursue," he says, "but also a narrative about what clinical question they are trying to answer."

Even for the same question, different cytometrists might use different panels. "There is a huge variety in how many and which antibodies people use," Dr. Wood says. Charles Caldwell, MD, PhD, CRC chair in cancer research, professor of pathology and anatomical sciences, and director, Ellis Fischel Cancer Center, University of Missouri, Columbia, adds: "Pathologists use different stains to look at tissues. A pathologist might like how one particular histological stain performs. It’s not much different with monoclonal antibodies."

There are many good and appropriate ways to practice flow cytometry, says Irene Check, PhD, director of the Division of Clinical Pathology and of the immunology and flow cytometry laboratory at Evanston (Ill.) Northwestern Healthcare and professor of pathology at the Northwestern Feinberg School of Medicine, Chicago. In addition to knowing basic principles, she says, "you have to understand the clinical question the clinician is asking and integrate it with other data." Dr. Check had a recent case in which a blood sample arrived with no clinical information. On CBC there was only a slightly elevated leukocyte count and mild anemia. Looking in the computerized medical record revealed a history of myelodysplasia. "So the question the clinician was asking was whether this elevated leukocyte count reflected transformation to acute leukemia," she says. "Once I knew that, I could select appropriate antibodies. Unfortunately, it’s not always self-evident what the clinician is thinking."

Dr. Davis notes that, in a regional or reference laboratory for flow cytometry, "you can develop a great degree of expertise, both in technical processing of the specimen as well as interpretation." Even so, he agrees that the approach is dependent on the clinical question. "Not so much that it is valuable to know what specific disease a clinician may be thinking about, but rather the spectrum of signs and symptoms," he says. Helpful terms might be "lymphocytosis," "leukocytosis," "rule out lymphoma because of lymphadenopathy," or "monoclonal gammopathy." Having select clinical information also helps ensure that the technology will be used cost-effectively. "Many of us feel that if there is no clinical information, the only resort is to analyze for all possible blood diseases," he says.

At a symposium led by the CAP Diagnostic Immunology Resource Committee, held in September at the CAP ’04 annual meeting, clinical indications for flow cytometry were reviewed in light of the 1995 flow cytometry consensus recommendations and the 2001 WHO classification of hematopoietic neoplasms. "We are several years out now from the original consensus conference on immunophenotyping of leukemias and lymphomas," says Dr. Caldwell, committee chair. "In this workshop we wanted to revisit some of the original recommendations, then use evidence-based medicine to update some newer approaches." Other workshops may follow, and the committee may draft a white paper and circulate it to other interested groups, such as the American Society for Clinical Pathology, the Clinical Cytometry Society, and the American Society of Hematology.

Speaking of the different applications for which flow can be used, Dr. Carey says, "Diagnosis is most important. Prognosis and disease monitoring are increasing, but there is more controversy." Technology is improving, he notes. "Much more robust instruments and software are now available. And CAP Surveys results argue for four and greater color analysis."

Dr. Wood calls the WHO classification different because it moved beyond morphology, which previous schemes had largely relied on, to integrate clinical information, morphology, immunophenotyping by flow or IHC, cytogenetics, and molecular data. "Flow cytometry is only one piece of the evidence," Dr. Wood says. In acute leukemia, for instance, the cornerstones of diagnosis are blast percentage, obtained by morphology, and lineage (myeloid vs. lymphoid), which is provided by flow. However, cytogenetic or molecular abnormalities play an increasingly important role, he says. Diagnosis of hematopoietic neoplasms now requires a multi-technique approach, except for a few disorders. Hairy cell leukemia, for instance, can be recognized by morphology and immunophenotyping.

The workshop focused on flow cytometry for immunophenotyping leukemia and lymphoma. "In this context," Dr. Caldwell says, "it is a very central element. But it is only one element in the total picture." He cites as an example acute promyelocytic leukemia, or APL, which usually has a specific rearrangement of chromosomes 15 and 17. Identifying this subclass is important because it has a specific therapy. "There are multiple ways of coming up with the correct answer in APL," Dr. Caldwell says. Classic cytogenetics can show the chromosomal rearrangement directly. That chromosomal rearrangement produces a rearrangement of two genes, PML and RAR-alpha, and PCR probes are now available to find those gene rearrangements. Flow cytometry presents a third approach, which would show a myeloid leukemia. However, Dr. Caldwell says, "Flow isn’t always as specific as genetic methods in this case, so we might use flow cytometry and one genetic test to refine the diagnosis." Having determined the diagnosis, molecular genetics might be the best tool for monitoring residual disease. "So I see these methods as complementary rather than as competing," Dr. Caldwell says.

Selecting the best tests depends on the clinical entity suspected, says Dr. Arber. In acute leukemia, molecular genetics and cytogenetics provide a lot of prognostic information and are significant in the WHO classification, so cytogenetics is usually done. "However, immunophenotyping is still the first-line test and is done on every case," Dr. Arber says. In particular, immunophenotyping gives information about lineage. "Using a combination of IPT and morphology, you can often suggest one of the recurring cytogenetic abnormalities," Dr. Arber says. "If regular karyotyping doesn’t pick it up, you can do molecular testing."

Karyotype analysis is done on all leukemias to provide a marker that may help identify the subclass and to monitor therapy or recurrence by PCR or FISH. "Karyotyping is especially important in acute myeloid leukemia [AML]," Dr. Arber says. WHO now has four categories of AML with recurring cytogenetic abnormalities, which have prognostic significance. Picking them up with routine karyotyping or molecular methods is important for classification.

Dr. Wood says that, with AML, immunophenotyping can "flow you in the right direction." Some immunophenotyping findings can suggest the presence of underlying genetic abnormalities. Since cell surface markers result from the genetic program, a disturbance of that genetic program can cause changes in expressed cell surface proteins. As a result, a single genetic abnormality, such as a translocation, can be recognized by immunophenotyping. "If you looked for these abnormalities genetically," Dr. Wood says, "you would have to run a fairly large panel of FISH probes or PCR primers."

Dr. Mann agrees that, with acute leukemia, the mainstay of diagnosis is a combination of morphology, flow cytometry, and cytogenetics. "Typically, the clinician who is performing the bone marrow aspirate and biopsy is the one who initiates that workup," she says. As an adjunct at initial diagnosis, FISH is sometimes used, specifically in cases where a translocation is not detectable or is difficult to pick up by conventional cytogenetics or where they did not get good metaphase spreads. "The most common pediatric acute lymphoblastic leukemia [ALL]-associated translocation is t(12;21)(p13;q22)," Dr. Mann says, "and it is not able to be detected by conventional cytogenetics." Similarly, in some adult AMLs an inversion of chromosome 16 can be difficult to pick up by cytogenetics.

Flow cytometry is less important for chronic myelogenous leukemia, or CML, Dr. Mann says, unless the patient is progressing to accelerated phase or blast crisis. "The majority demonstrate a myeloid phenotype. However, about 20 percent of cases go into lymphoid blast crisis, and flow cytometry is important in these cases," she says. "In CML, we are doing a lot of disease monitoring," she adds, "where quantitative RT-PCR gets into the mix. This doesn’t mean, however, there is not an ongoing role for cytogenetics."

For the most common type of chronic leukemia, chronic lymphocytic leukemia, or CLL, a specimen might come directly for flow cytometry, Dr. Check says. It can be diagnosed on blood, but also sometimes on bone marrow and on tissue. (One of the advantages of flow is that it can be done on a wide range of sample types.) A clinician might have found an elevated lymphocyte count on a CBC. For the right age group and clinical setting, CLL is part of the differential diagnosis of lymphocytosis. Or a routine CBC done by the hematology laboratory might show a very high lymphocyte count, which is flagged for manual differential review by the hematopathologist. It might show monomorphic lymphocytes consistent with CLL, and flow might be ordered to phenotype.

Suspicion for acute leukemia might also come from a routine CBC on which blasts are seen in the peripheral blood. Such a sample would almost automatically be sent for flow cytometry phenotyping. "Some leukemias don’t have much in the way of blasts in the peripheral blood," Dr. Check says. "For these, the clinician would do bone marrow biopsy and aspirate, and specimens would be sent to flow cytometry and to cytogenetics."

Dr. Carey agrees that, for CLL, immunophenotyping of blood by flow cytometry is recommended for adults with unexplained persistent absolute lymphocytosis prior to bone marrow or lymph node biopsy. "It provides a reliable means of separating CLL and mantle cell and hairy cell leukemia from other subtypes of chronic lymphoproliferative disorders," he says.

In lymphoma, as in leukemia, immunophenotyping is the first-line test. Cytogenetics isn’t as prognostically significant, but it can help classify the disease. Some cases fall between lymphoma and atypical, and finding a genetic abnormality pushes the diagnosis toward lymphoma. Finding a genetic abnormality can also help to subclassify lymphoma. "But most people don’t do genetics on every suspected lymphoma," Dr. Arber says. "Certainly we don’t do karyotypes on every case. And we only use molecular tests when they are needed to clarify specific situations."

Typically the ordering clinician knows from the clinical story that lymphoma is part of the differential diagnosis, Dr. Check says. "Usually with a suspected lymphoma, the sample comes right to flow cytometry and we do a preliminary panel for T subsets and/or monoclonal B cells," she says. "We might stop there or do additional flow studies." If there is a clear resolution, molecular testing to look for clonal T or B cells would not be done. An abnormal pattern in T cell antigens that is suspicious for a T cell lymphoma is typically confirmed with T cell receptor gene rearrangement studies. T cell receptor gene rearrangement studies might also be done despite normal T cell flow cytometry results if there is a high index of suspicion of T cell lymphoma on clinical or morphologic grounds. For a B cell lymphoma, "99 percent of the time flow cytometry will be clear," Dr. Check says. "So the only time we might do an immunoglobulin gene rearrangement study would be if flow cytometry is equivocal." Also, the pathologist might order determination of cyclin D1 or other markers to subclassify a lymphoma as, for instance, mantle cell.

For Hodgkin lymphoma, the picture is not so clear. "Flow cytometry is generally unable to detect neoplastic cells of Hodgkin lymphoma at this point," says Dr. Wood. "So if that is the only clinical suspicion, flow is not indicated." If one is looking also for conditions that might mimic Hodgkin, such as diffuse large B cell lymphoma, flow would be an appropriate test to do. Dr. Check expresses another view. "I don’t think anyone says specifically, ’Rule out Hodgkin lymphoma,’" she says. "They say, ’Work up lymphoma.’" Another consideration is that on followup Hodgkin sometimes turns into non-Hodgkin lymphoma.

In a cutting-edge application in lymphoma, Dr. Wood notes that flow cytometry can be used to determine clonality using antibodies against the V-beta subunit of the T cell receptor. This method is much faster than PCR. "In this situation, you are looking for the predominance of one of the V-beta isoforms," Dr. Wood says. Flow allows you to identify clonality on a specific abnormal cell subpopulation, whereas in PCR you look at the overall population.

Another emerging application of flow cytometry is to help diagnose myelodysplastic syndrome. "Just as the morphologic changes of myelodysplasia differ from normal myeloid morphology, so does the antigenic composition of the cells," Dr. Davis says. "Finding an abnormal clone can be a great adjunct to support a morphologic suspicion." Different combinations of monoclonal antibodies for this complicated task have been published (Elghetany MT. Hematologica. 1998; 83:1104-1115; Kussick SJ, Wood BL. Arch Pathol Lab Med. 2003; 127: 1140-1147).

Dr. Davis says flow cytometry can also be used to identify protein targets of therapeutic monoclonal antibodies, such as Rituxan (against CD20) and Mylotarg (against CD33), as well as prognostic factors, such as ZAP-70 in CLL.

A further laboratory contribution to managing hematopoietic malignancies is using abnormalities identified during initial workup to monitor for minimal residual disease to evaluate response to therapy and to detect relapse. Monitoring for minimal residual disease, or MRD, is becoming more important for a variety of diseases, says Dr. Wood. "Data are still being acquired as to what role it will play in directing therapy," he says, "but there is already good information that it helps with long-term prognosis and survival, particularly for acute leukemias." Dr. Carey adds, "Flow cytometry is a sensitive method to detect residual disease in chronic lymphoproliferative diseases, but the clinical implications have not yet been established."

Flow cytometry is used to monitor MRD in lymphomas, says Dr. Arber, because flow can pick up fairly small populations of abnormal cells. "We can also use flow cytometry for monitoring residual disease in leukemias, such as for detecting CD7 on myeloid cells," he adds. "There is a long list of aberrant phenotypes that people follow."

A newer alternative for monitoring MRD is molecular testing for gene rearrangements, particularly in lymphomas. A complete set of PCR primers for this purpose has been developed and validated by a European working group and is now being marketed by In Vivo Scribe in the U.S., says Dr. Arber (see, for example: Bruggeman M, et al. Leukemia. 2004;18:709-719; van der Velden VH, et al. Leukemia. 2004;18:884-886). There are other multiplex packages to look for common translocations in leukemias and lymphomas. "For leukemias, these packages will pick up the most common things, but will miss a lot of leukemias and uncommon abnormalities," Dr. Arber says. "But in lymphoma, we are really looking for B and T cell-associated gene rearrangements and a handful of translocations, so you can detect just about every lymphoma using the Biomed 2 approach." However, he cautions, it is not ideal for MRD testing because it employs consensus primers and won’t pick up low levels, such as one in 100,000 cells.

Choosing between flow and molecular for this purpose depends on sensitivity, specificity, cost, and turnaround time, says Dr. Davis, as well as the individual disease entity involved. In acute myeloid leukemia, for instance, there are currently very few molecular markers, whereas 80 to 85 percent of AML cases have a phenotypic abnormality that one can use for MRD detection. Typically the sensitivity of molecular methods is perhaps 10-fold greater, he concedes. But "if you use several colors and collect lots of cells," Dr. Davis says, "you can get to a sensitivity of 1:100,000 with flow" (see, for example: Malec M, et al. Leukemia. 2004; 18: 1630- 1636; Bott cher S, et al. Leukemia. 2004; 18: 1637-1645).

Moreover, it is not clear what level of sensitivity is clinically relevant. "Therapeutic threshold is part of what is now being determined in clinical trials," says Dr. Wood. In acute leukemia in children, molecular is at least a lot more sensitive than flow. "But the actual level needed for therapeutic decision-making may be achievable by flow," he says. "And flow is more rapid and less expensive."

Dr. Mann considers the two major issues at this point to be the higher cost of the labeled PCR primers and probes and the need to do enough of any PCR assay for MRD to maintain expertise.

At this point, even some high-volume flow centers don’t do MRD monitoring. "We don’t do it because our clinicians haven’t asked for it," Dr. Check says. "Our clinicians are not yet ready to act on that information." For determining relapse, microscopic examination is used. "At our institution, morphology trumps flow cytometry," she says.

Laboratorians must make another choice—immunophenotyping can be done by either flow or IHC. "Each place does it their own way," Dr. Check says. For instance, at Evanston Northwestern Healthcare cyclin D1 in the context of lymphoma is done by IHC. IHC and flow each have advantages. Cytometry is fast, providing an answer in hours. And with flow it is possible to use multiple colors to look for characteristic combinations of antibodies on individual cells. Surface kappa or lambda are better done by flow. An advantage of IHC is that it is possible to see marker patterns in the context of cell architecture. And it can be done on fixed tissues. With flow, on the other hand, the sooner the sample is analyzed, the better. "We prefer to start with flow and fill in what is needed with IHC," Dr. Check says.

Dr. Carey tends to favor IHC more often than most other laboratorians. Speaking at the CAP workshop, he said, "You will get a variety of opinions here. I tend to be the most conservative."

Says Dr. Wood, "I think Dr. Carey makes some valid points." It used to be that antibodies for IHC and tissue sections were relatively limited. Advances have now produced a large variety of antibodies for IHC, Dr. Wood says, so in many circumstances IHC can make a definitive diagnosis. "Flow cytometry is another way of approaching a similar type of problem," he says, "It’s usually more rapid and sensitive. These days one can use either technique to get to the same point in many cases that are based in lymph nodes. For peripheral blood or bone marrow, however, flow has advantages."

In the best of all worlds, Dr. Caldwell says, the hematologist and the pathologist would communicate often and extensively. "But in the end, all data should come together at one focal point and the pathologist will make the final diagnosis," he adds. While it is reasonable to expect this to happen at a tertiary center, where tools and expertise are available on site, what about other settings? In small community hospitals, some techniques may be present, but specimens may also be sent out for diagnosis. "Those hospitals frequently do not have medical oncologists on hand to be doing certain kinds of induction therapy," Dr. Caldwell says, "so mostly if a diagnosis of acute leukemia is suspected, patients should be referred to a specialty center for both diagnosis and treatment."

When specimens are sent to commercial laboratories for testing, who puts all the information together to make a diagnosis? One option is that the local pathologist will do this. But, Dr. Caldwell says, "one could argue that whoever is signing out in a commercial laboratory should also look at histology. It’s dangerous to look at just one component without the others." He adds, "Doing a test in isolation can sometimes give you not just no answer, but the wrong answer."

Limited sample analysis bears on a recently devised practice called "technical-only flow cytometry," in which a commercial laboratory provides the technical aspect of flow cytometry, then sends a portion of those results to the referring physician for local interpretation. The report does not contain full listmode data, which is what an experienced flow cytometrist would use to make an interpretation, but only selected percentages derived from the listmode. Because the referral laboratory does not do the interpretation, the local pathologist can do Part B or professional billing. "This practice has many of us in the field very concerned," Dr. Davis says. "We wonder, are patients benefiting from this practice?" Dr. Davis characterizes technical-only flow cytometry as "the blind leading the blind: One party receives a specimen knowing less than full clinical background, and the other party receives a portion of the flow results, in particular a printout of results gated by another individual, then tries to make an interpretation." In Dr. Davis’ view, "There is a real potential for missing low-frequency abnormal populations."

Dr. Check can’t even envision how technical-only flow cytometry can be interpreted. "My understanding of how it works," she says, "is that you send a sample to the laboratory, which does the markers—how would they pick markers?—and sends you back percentage values without any interpretation. How a pathologist could then put that data into the right context I think would be very difficult." An experienced flow cytometrist would want to know such things as, of all the cells in this specimen, how many were looked at? Did they look at lymphocytes only? Did they look for blasts? "Interpretation is related to a pattern that one sees on flow," Dr. Check says. "We are not just looking at a number."

Several trends can be seen in the field of laboratory analysis of hematologic malignancies. One is a growing trend toward more fine-needle aspirates, as opposed to tissue biopsies, says Dr. Check. These include fine-needle specimens of what she calls "deep innards—abdominal masses, places that would in past years have been accessible only through open biopsy." A fine-needle aspirate does not present as much tissue, so the laboratory has to use it wisely. Also, tissue architecture is lacking. "So I think with fine-needle aspirates that flow cytometry becomes much more important to confirm the pathologist’s morphologic and cytologic impressions," she says.

Dr. Mann predicts an increase in minimal residual disease monitoring, especially in leukemias. MRD monitoring is now clearly standard of care in CML, whether treated with Gleevec or bone marrow transplantation. "We have that test online," Dr. Mann says. "However, our clinicians are sending for similar tests to reference laboratories for the 15;17 translocation in APL and the inversion 16 translocation in a subtype AML." She sees MRD monitoring being done by molecular tests, which are more sensitive than cytogenetics.

Another clear trend is an explosion of information about expression patterns of genes in hematopoietic neoplasms, Dr. Wood says. "We need a more detailed understanding of what the molecular abnormalities are in each neoplasm and what the consequences are for the various cell types," he says. With such information, clinical scientists will be able to develop better techniques to identify neoplasms and therapeutic targets and prognostic information about how those neoplasms will behave. "Whether those abnormalities will be found by flow cytometry or molecular tests or cytogenetics or IHC remains to be determined," Dr. Wood says. "The choice will largely be based on sensitivity and specificity, but also on cost and the rapidity with which each assay can be performed." There may even be new assays, he speculates.

Of one thing Dr. Wood is sure: “It will be an exciting time.”

William Check is a medical writer in Wilmette, Ill.