February 2012
Feature Story

William Check, PhD

Every tissue in the body is capable of trickery. Every specimen that slides across the microscope stage or flows through the cytometer could be masquerading as something it’s not—benign cells imitating a neoplasm or cancerous cells playing the innocent. When it comes to chicanery, bone marrow is no exception: As the CAP ’11 course titled “Practical Challenges in Bone Marrow Evaluation” made clear, bone marrow has more than its share of disguises.

Explaining the overall purpose of the course, Carla S. Wilson, MD, PhD, a professor in the Department of Pathology at the University of New Mexico, who organized the session, told CAP TODAY, “We wanted to key in on some of the most important differential diagnoses that we deal with in bone marrow pathology.”

One of the goals of the course was to teach attendees to recognize benign entities that mimic neoplasms, such as dysplasia and marrow fibrosis. Of her dysplasia presentation, Dr. Wilson, who is also medical director of flow cytometry at UNM and Tricore Laboratories, said, “I tried to highlight some of the new considerations you have to make in dysplasia, such as copper deficiency, which has been overcalled a myelodysplastic syndrome.”

Marrow fibrosis, too, wears a double face. “Reticulin myelofibrosis may be seen in both neoplasia and non-neoplastic processes,” Kaaren K. Reichard, MD, told the session’s attendees. She was at UNM and Tricore at the time of the meeting and is now senior associate consultant in the Division of Hematopathology, Department of Laboratory Medicine and Pathology, Mayo Clinic.

Speakers also discussed cases illustrating marrow hypocellularity and lymphoid aggregates, both of which may or may not signal neoplasms. Cases of peripheral pancytopenia plus central marrow hypocellularity “can be very tricky to diagnose accurately, because they can be caused by a number of conditions,” David Czuchlewski, MD, assistant professor of pathology at UNM and associate director of the molecular diagnostics laboratory at TriCore, said during the course. In these situations, it is important to consider inherited causes of bone marrow failure, as well as to distinguish a myelodysplastic syndrome (MDS) from aplastic anemia.

Addressing lymphoid aggregates, Dr. Reichard listed five criteria that can help discriminate non-neoplastic from neoplastic processes.

Finally, Dr. Wilson discussed a plasma cell neoplasm, which has a rich and complex differential. “The basic criteria for calling myeloma have not changed for many years,” Dr. Wilson told CAP TODAY. “In 2008 they did change, but a lot of people don’t know why or how to use the new criteria. I wanted to show how the new classification can help pathologists make better diagnoses.”

A general goal of the course was to illustrate the value of ancillary testing—flow cytometry, immuno-stains, FISH, cytogenetics, and molecular analysis—to help resolve challenging bone marrow presentations. “Even though each presenter focused on specific topics, we all emphasized how important it is to integrate findings obtained from various testing modalities to arrive at a final diagnosis,” Dr. Reichard said in an interview. In the case of bone marrow fibrosis that she discussed, she showed that a negative flow cytometric analysis doesn’t necessarily mean that one is dealing with a non-neoplastic bone marrow. In this situation, immunohistochemistry may provide evidence for a malignant verdict. “To get a good interpretation, you need to know when one piece of data doesn’t match and you need to do an additional test,” Dr. Reichard said.

Dr. Wilson first presented two cases to illustrate the importance of considering copper deficiency in the differential diagnosis of MDS. In the first case, a 77-year-old woman was treated for 24 months with cyanocobalamin for anemia thought to be due to vitamin B₁₂ deficiency. Her family said she had worsening short-term memory, as well as problems with gait and appetite loss, even though her iron and vitamin B₁₂ levels were normal. She had general cytopenia with normochromic normocytic anemia and a “striking” leukopenia, with neutrophils decreased to 27 percent.

“Our key finding on bone marrow evaluation,” Dr. Wilson said, “was cytoplasmic vacuoles in precursor cells of both myeloid and erythroid lineages” (Fig. 1 and Fig. 2). Flow cytometry showed polytypic B cells with a population of hematogones (reactive B cell precursors).

Dr. Wilson listed possible causes of cytoplasmic vacuoles. One type of etiology is acute alcohol intoxication or chronic alcoholism, but in both of these situations only erythroid precursors are affected. The patient had no history of drug toxicity or pancreatic dysfunction, or features suggesting a myeloid neoplasm.

Copper deficiency is also in the differential. Direct testing showed decreased serum copper and increased serum zinc. Over the last decade denture adhesive cream with high zinc levels has been recognized as a likely cause of this combination of features (Barton AL, et al. Ann Clin Biochem. 2011;48: 383–385). Copper deficiency mimics both the abnormal marrow and neurodegenerative symptoms of vitamin B12 deficiency.

In this patient the diagnosis of zinc-induced copper deficiency (“zinc overload syndrome”) led to treatment with oral copper and cessation of zinc-containing denture fixative. After two months the woman’s CBC and differential counts normalized and her short-term memory improved.

Dr. Wilson’s second case featured ring sideroblasts, another sign of dyserythropoiesis that can be caused by copper deficiency. In this case a 63-year-old woman had been diagnosed as having MDS and she had been treated with four cycles of decitabine and G-CSF. When no clinical improvement was seen after four months, she was referred for further workup. Cytogenetics, flow cytometry, and FISH were negative for signs of MDS. On the original slides Dr. Wilson saw changes associated with copper deficiency, including ring sideroblasts and vacuoles in precursors of myeloid and erythroid lineages.

Direct testing showed decreased copper. Copper supplementation led to a resolution of the symptoms.

“The important lesson from this case is that dysplasia does not necessarily indicate a clonal disorder,” Dr. Wilson emphasized.

Recognition of dysplasia is poorly reproducible among pathologists, Dr. Wilson noted. “We are good at counting blasts,” she said in an interview. However, the correlation coefficients between pathologists for diagnosing myeloid dysplasia and erythroid dysplasia are poor—0.45 and 0.27, respectively, in one study. To complicate things, erythroid atypia is common with stress erythropoiesis or regeneration. “It can be difficult to differentiate malignant from reactive changes, especially in the erythroid lineage,” Dr. Wilson says. “At what point do you call atypia dysplasia?”

An important criterion is that MDS is more likely than copper deficiency when dysplastic cells make up more than 10 percent of a lineage. Increased hematogones, on the other hand, tip the scales in favor of copper deficiency. If these criteria are not definitive, MDS may be diagnosed by the identification of recurrent MDS-related chromosomal abnormalities on cytogenetic testing.

Dr. Czuchlewski presented a case of bone marrow failure, which he said fits under the course umbrella of “emerging concepts of pathogenesis that have given rise to new diagnostic possibilities.” Since bone marrow failure syndromes are associated with hypocellularity, he chose a case combining these two features—a 47-year-old man with hypocellular marrow in the setting of peripheral cytopenia (absolute neutrophil count of 0.3 × 109/L) (Fig. 3).

Dr. Czuchlewski said the first task in a case like this is to rule out a subcortical bone specimen, which is often hypocellular. For accurate determination of cellularity, consensus guidelines recommend a bone marrow core biopsy greater than or equal to 2 cm in length. “If it is not adequate, consider telling the clinician to go back and get an adequate specimen,” Dr. Czuchlewski said. An inadequate bone marrow biopsy is not common but should be kept in mind.

The next step is to rule out unusual (hypoplastic) presentations of various leukemias. “Leukemias are usually hypercellular, but they can be hypocellular,” Dr. Czu-chlew-ski said. Markers can be assessed by immunostaining, because flow cytometry can be misleading due to marrow hypocellularity. Helpful markers for leukemias that may be hypocellular are: hairy cell leukemia, CD20; acute myelogenous leukemia, CD34 and CD117; acute lymphoblastic leukemia, CD34, CD10, and TdT; and T-cell large granular lymphocytic leukemia, CD3. As a screening panel, Dr. Czuchlewski recommended CD20, CD34, and CD3.

Inherited bone marrow failure syndromes, which Dr. Czuchlewski called “an often overlooked diagnostic possibility,” are considered next. “Don’t exclude these in older children,” he advised. “Their incidence does not go to zero even in adulthood.” Physical abnormalities and a family history of bone marrow failure or malignancies are helpful, but some inherited bone marrow failure cases may show none of these features.

Two important multilineage inherited bone marrow failure syndromes are Fanconi anemia and dyskeratosis congenita, in which impaired telomerase function leads to short telomeres. Key tests are chromosome breakage analysis for Fanconi anemia and, in dyskeratosis congenita, flow cytometry FISH for telomere length in subsets of leukocytes. This latter test is still uncommon. In Fanconi anemia, Dr. Czu-chlewski said, “Starting first with DNA sequencing is an inefficient approach,” because so many genes can be affected.

After inherited bone marrow failure syndromes come the acquired syndromes, chiefly aplastic anemia.

“Aplastic anemia can be difficult to distinguish from hypocellular MDS,” Dr. Czuchlewski cautioned. Some cases have overlapping features—erythroid dysplasia (“stress” dyserythropoiesis), macrocytic anemia, and cytogenetic abnormalities. “Historically people have used the presence of a clonal abnormality on cytogenetics to move in the direction of calling myelodysplastic syndrome,” Dr. Czuchlewski told CAP TODAY. More recently, it has been recognized that up to 12 percent of cases of aplastic anemia show abnormal cytogenetic clones in the absence of myelodysplastic syndrome. Monosomy 7 in aplastic anemia is associated with a high risk of progression to MDS or acute myeloid leukemia.

Helpful distinguishing features are non-erythroid dysplasia and increased blasts, both of which can be present in MDS but should not be seen in aplastic anemia (Fig. 4). Increased CD34-positive blasts are more consistent with an interpretation of hypocellular MDS. However, Dr. Czuchlewski says, “Only a subset of cases of myelodysplastic syndrome present with a hypocellular appearance.”

He called the association between bone marrow failure and paroxysmal nocturnal hemoglobinuria (PNH) “very tricky.” The classic presentation of PNH is hemolysis, anemia, cytopenias, and increased risk of thrombosis. Dr. Czuchlewski noted a newer way of thinking about PNH. “Using a very sensitive method, you can demonstrate the presence of small populations of clones with a PNH phenotype in some cases of aplastic anemia, as well as in some cases of myelodysplastic syndrome, especially the hypocellular type.” In this view there is overlap among all of these entities, even classic overt PNH, which includes some degree of bone marrow failure. Further, it is now known that aplastic anemia can evolve into hypocellular MDS or hemolytic PNH, and PNH can evolve into aplastic anemia. All of this overlap complicates diagnosis.

The underlying cause of PNH is a mutation that decreases expression of the protein GPI, which anchors many standard cell surface proteins, including CD markers such as CD59, CD14, and CD16 detected on flow cytometry. To find PNH clones with very high sensitivity, a new reagent—FLAER, or fluorescent aerolysin, which binds directly to the GPI anchor—is used in conjunction with standard flow cytometry on peripheral blood (Fig. 5). Absence of FLAER reactivity indicates lack of GPI. Standard flow cytometry for CD antigens detects GPI with a sensitivity of one percent and identifies classic hemolytic PNH. Flow cytometry with FLAER increases the sensitivity for PNH clones to 0.01 percent and detects small clones associated with bone marrow failure. If PNH clones are present at a subclinical frequency in a case of aplastic anemia, it is important to monitor for progression to classic hemolytic PNH.

High-sensitivity flow cytometry for PNH clones can be considered in cases of hemolytic myelodysplastic syndrome, Dr. Czuchlewski said. Some specialists also suggest high-sensitivity PNH flow cytometry in some other subsets of MDS cases. Dr. Czuchlewski called this approach “controversial,” and added, “I suggest you discuss that with your clinician.”

Dr. Reichard described the case of a 45-year-old HIV-positive man with a history of colon carcinoma and an unexplained ongoing anemia. A bone marrow biopsy showed large dense lymphoid aggregates. “However,” Dr. Reichard told CAP TODAY, “this pathology was not related to the anemia. Between the time we got the bone marrow specimen and the time we examined it, we learned that the man had had a parvovirus infection.” In this situation the lymphoid aggregates reflected a morphologic finding often seen in HIV-positive individuals.

In cases where there is not such an obvious cause for lymphoid aggregates, the differential diagnosis revolves around non-neoplastic versus neoplastic (leukemia/lymphoma). Five morphologic criteria can help distinguish these two classes. Generally, in non-neoplastic conditions, lymphoid aggregates are few, small, well-circumscribed, non-paratrabecular, and composed of mostly small mature lymphocytes with some histiocytes and plasma cells (Fig. 6). In neoplastic conditions, on the other hand, one sees large multiple aggregates with infiltrative borders, cytologic atypia, and often monotonous cellular composition.

A few well-known morphologic variations on non-neoplastic lymphoid aggregates include: in HIV-positive persons, polymorphous, often large, poorly circumscribed aggregates that contain large cells with cytologic atypia and that can look very abnormal; after rituximab therapy, lymphoid aggregates with essentially all T cells, requiring staining with CD79a or CD19 to distinguish non-neoplastic lymphoid aggregates from residual lymphoma; and regressed germinal centers in Castleman disease.

Neoplastic lymphoid aggregates may show partial and/or diffuse effacement—morphologically normal bone marrow adjacent to abnormal bone marrow. “This is what most of us pathologists see the most,” Dr. Reichard told CAP TODAY. However, lymphoid aggregates may form a sinusoidal pattern, which Dr. Reichard said is “really sneaky. Even experienced pathologists can miss it, particularly if it is made up of small cells.” With the sinusoidal pattern, a morphologically discrete lesion is typically not evident.

Use of flow cytometry in the workup of lymphoid aggregates is “useful but requires knowledge of testing caveats,” Dr. Reichard cautioned. It can be useful in the workup of morphologically occult lymphoid aggregates, such as a sinusoidal pattern, when the aspirate is adequate. If flow cytometry is normal, immunohistochemistry can be helpful.

Dr. Reichard noted that morphologically obvious lymphoid aggregates—large nodules that hug the bone—are easy (Fig. 7). However, even in obvious cases IHC can be helpful, since you may encounter false-negative flow cytometry results.

If flow cytometric analysis is performed up front, use of IHC is conditioned by the initial flow cytometric findings. If flow is negative, morphology is consulted. If morphology is within normal limits and there is no pertinent clinical history or unexplained cytopenias, IHC is not indicated. If morphology is “suspicious,” IHC is likely indicated.

Regarding immunohistochemistry, Dr. Reichard said, “CD34, CD20, and CD3 are your friends.” In an interview she explained: “In any case where you have an unexplained cytopenia or the aspirate is inadequate or there is fibrosis, immunostains are cheap, easy, and informative.” In non-neoplastic lymphoid aggregates, where T cells usually predominate, there is typically a core of CD20-positive B cells surrounded by CD3-positive T cells.

Flow cytometry and IHC are complementary, Dr. Reichard noted. In flow cytometry you lose tissue architecture, but you get a nice multiparametric immunophenotype. With IHC you can’t look at many markers simultaneously, but you can see tumor cells and patterns. “For each technique there are antibodies that are readily available and useful,” she told CAP TODAY. With IHC, stains such as cyclin D1, annexin 1, and EBER (EBV-encoded RNA) can help when prior testing is inconclusive. For instance, cyclin D1 is positive in mantle cell lymphoma. “Flow cytometry can get you far; then immunohistochemistry can get you farther,” Dr Reichard said.

In this case the final diagnosis was benign lymphoid aggregates in an HIV-positive person. “With morphology, immunostains, and flow you can pretty much always get to the right diagnosis,” Dr. Reichard said. “In this case the clinical knowledge helped us to put it all together.”

Dr. Reichard’s second case was a 48-year-old woman with a long history of systemic lupus erythematosus and progressive cytopenias. She now presented with mild pancytopenia. A workup for infectious causes was negative.

Attempted bone marrow sampling yielded a dry tap, so a touch preparation was made. Dr. Reichard noted that touch preps are typically helpful, but sometimes cells remain adherent to the core biopsy. In this case it was productive and showed intact trilineage hematopoiesis with normal morphology—no dysplasia.

A bone marrow biopsy showed hypercellularity, which Dr. Reichard called “unexpected,” as well as streaming effect. Staining for reticulin fibers showed mild reticulin fibrosis.

In the differential diagnosis for reticulin fibrosis are both non-neoplastic and neoplastic conditions. Among the former are chronic infection and autoimmune myelofibrosis; neoplastic causes include lymphoid malignancy, myeloproliferative neoplasms, and (rarely) MDS.

Dr. Reichard showed an example of a non-neoplastic disease that can be associated with reticulin fibrosis—gray platelet syndrome. “The whole point of this case is to demonstrate that fibrosis may be a result of non-neoplastic disorders,” Dr. Reichard said in an interview. “Just because you see fibrosis, that doesn’t necessarily mean malignancy or a primary bone marrow neoplasm. A non-clonal disorder can lead to fibrosis that progresses over time.” To arrive at an accurate diagnosis, it is necessary to do a complete pathologic examination covering the whole differential.

Reticulin fibrosis is assessed with silver stain and evaluated by the number of fibers stained and their thickness (Fig. 8). A grading system has been established, with zero being a normal amount and thickness (Thiele J, et al. Haematologica. 2005;90:1128–1132). Reticulin fibrosis is generally considered reversible.

In contrast, collagen fibrosis, which is detected by trichrome stain (Fig. 9), is not normally present in bone marrow so it is essentially always pathologic. It is generally irreversible.

In a workup for bone marrow fibrosis, flow cytometry may yield false-negative results due to fibrosis, hemodilution, or both. Blasts and abnormal lymphoid populations (paratrabecular or fibrotic lesions, classical Hodgkin lymphoma) may be missed. An antibody against CD34 is useful to assess for blasts (though not all blasts are CD34-positive). Also, antibodies against CD20 and CD3 are helpful in the initial detection of B- and T-cells, respectively.

In this case there was mild reticulin fibrosis, and CD34, CD20, and CD3 were within normal limits, ruling out MDS (in conjunction with the normal touch preparation morphology) and a lymphoid neoplasm. Remaining possibilities included collagen vascular disease/autoimmune condition and perhaps myeloproliferative neoplasms. Autoimmune myelofibrosis and myeloproliferative neoplasms have overlapping features, such as hepatosplenomegaly, hypercellularity, reticulin fibrosis, small megakaryocytes with dark nuclei, and lymphoid infiltrates (Bass RD, et al. Am J Clin Pathol. 2001;116: 211–216). They can usually be distinguished because autoimmune myelofibrosis has cytopenias, rather than cytoses; no eosinophilia or basophilia; and lack of megakaryocyte clustering or bizarre morphology.

In this patient the final diagnosis was increased reticulin myelofibrosis associated with underlying systemic lupus erythematosus.

As her take-home messages, Dr. Reichard reiterated that reticulin myelofibrosis may be seen in both neoplastic and non-neoplastic processes and that myelofibrosis often results in a dry tap, making touch preps useful for cytology (and potentially genetic—for example, fluorescence in situ hybridization— studies). Finally, on the core biopsy she advises CD34, CD20, and CD3 staining to detect blasts and lymphoid cells.

Dr. Wilson presented a case that illustrates the complexity of the diagnosis of plasma cell neoplasms. She chose this case to emphasize plasma cell myeloma, also called multiple myeloma, because it is the second most common type of hematopoietic malignancy (10 percent to 15 percent) and one of the most extensively studied.

“This case came to us in consultation because the outside pathologist was worried about lymphoma,” Dr. Wilson told CAP TODAY. The outside laboratory’s immunostains were confusing to the pathologist. In particular, CD20—a B cell marker—was positive. Dr. Wilson noted that 15 percent to 20 percent of myelomas turn out to be positive for CD20. “This was tricky because the cells looked like lymphoma,” she added. The specimen was also positive for cyclin D1 (Fig. 10).

To complicate matters, a monoclonal IgG protein (M component) was seen on serum protein electrophoresis. “The quantity of serum monoclonal protein detected has been part of the criteria for calling myeloma,” Dr. Wilson told CAP TODAY.

The outside pathologist wondered whether this was mantle cell lymphoma, a cyclin D1-positive B-cell lymphoma. However, the sample had the wrong pattern of infiltration for mantle cell lymphoma and the serum wouldn’t contain a significant M component. Hairy cell leukemia, which is also CD20-positive, would typically be only weakly positive for cyclin D1.

Plasma cell myeloma would fit the CD20-positive, cyclin D1-positive, M component-positive case under study. Although the abnormal cells didn’t look like plasma cells, “Myelomas can look like different cell types, including lymphoid cells.” (Fig. 11).

She tested the sample for expression of CD138, which stains only plasma cells among normal bone marrow cells; 15 percent of the total bone marrow cells were positive. These cells were also kappa light chain restricted by in situ hybridization, confirming a clonal plasma cell population.

“Now we know that it’s a plasma cell neoplasm,” Dr. Wilson concluded. But “plasma cell neoplasm” merely means an expanded monoclonal plasma cell population. At the early end of the spectrum of plasma cell neoplasm lies monoclonal gammopathy of undetermined significance (MGUS), which is 100 times more common than plasma cell myeloma. MGUS progresses to symptomatic myeloma at the rate of one percent per year. Asymptomatic myeloma progresses at a higher rate to symptomatic myeloma. “Where on the spectrum does [our case] fall?” Dr. Wilson asked. “Do we have sufficient evidence to diagnose plasma cell myeloma in 2011?”

One diagnostic problem concerns the plasma cell fraction. Previously 30 percent was the cutoff. However, one-third of those who have plasma cell myeloma don’t meet this criterion at initial evaluation. Moreover, sampling errors can arise if there is considerable bone marrow sparing by the neoplastic process. And the fraction of plasma cells can be underestimated on flow cytometry because plasma cells are “sticky.” In the 2008 revision, the cutoff for myeloma was made 10 percent clonal plasma cells, or M protein of ≥30 g/dL. Symptomatic myeloma is present if the patient has clonal plasmacytosis and M protein plus any of the four “CRAB” signs—calcium elevated, renal dysfunction, anemia, or bone lesions (lytic lesions or osteoporosis with fracture). In this patient, the criteria for plasma cell myeloma were satisfied.

“Adequate bone marrow sampling is critical” to a good diagnosis of plasma cell neoplasms, Dr. Wilson emphasized. Aspirates and touch imprints of trephine biopsies are helpful.

“Myeloma cells have complex genetic aberrations,” she said. Conventional cytogenetics is informative in 30 percent to 40 percent of newly diagnosed cases of plasma cell myeloma. According to Dr. Wilson, gene expression profiling is more informative, but it is expensive and more technically demanding.

“We will be changing our discussion of myeloma next year,” she predicts.