CD34/QBEND10 Immunostaining in the Bone Marrow Trephine Biopsy A Study of CD34-Positive Mononuclear Cells and Megakaryocytes Goran Torlakovic, MD; Ruth Langholm, MD; Emina Torlakovic, MD Context. The immunohistochemical detection of CD34 protein using QBEND10 antibody in bone marrow trephine biopsies was shown recently to be a precise method for quantitation of blasts and a possibly useful approach in diagnosis and classification of myelodysplastic syndrome. Objectives. To evaluate CD34 cells in bone marrow biopsies with various diagnoses and to assess how counts obtained using this method correlate with blast counts obtained by traditional morphologic evaluation of bone marrow smears. Design. Bone marrow trephine biopsies from 108 adult patients were evaluated by immunohistochemistry using anti-cd34 antibody (QBEND10). CD34 mononuclear cells were counted and compared with the blast counts in the bone marrow aspirate smears or imprints. CD34 mononuclear cell clusters and CD34 megakaryocytes were also recorded. The type of positivity (membranous vs cytoplasmic) and the percentage of CD34 megakaryocytes were evaluated because the presence of CD34 megakaryocytes was recently suggested to be present in myelodysplastic syndrome, but not in myeloproliferative disease or nonneoplastic bone marrow. Results. Six of 24 biopsies with partial involvement by non-hodgkin lymphoma and 5 of 60 biopsies with reactive changes had 5% to 10% CD34 mononuclear cells and were associated with lymphocytosis and increased hematogones. The CD34 mononuclear cell clusters were found only in myelodysplastic syndrome and myeloproliferative disease. The CD34 megakaryocytes were present in all diagnostic groups. Conclusion. The number of CD34 mononuclear cells was often slightly higher than the number of myeloid blasts in the bone marrow smears, probably due to increased hematogones. The presence and the number of CD34 megakaryocytes do not appear to have diagnostic value, but this finding should be further investigated in relation to clinical parameters. (Arch Pathol Lab Med. 2002;126:823 828) The alteration of the process of cell maturation and an increased percentage of myeloid blasts are essential criteria for the diagnosis of acute myeloid leukemia, as well as for classification of myelodysplastic syndrome (MDS). 1,2 CD34 is a surface glycophosphoprotein expressed on developmentally early lymphohematopoietic stem and progenitor cells, small-vessel endothelial cells, and embryonic fibroblasts. 3 Although the precise function of CD34 protein remains largely unknown, studies identifying L-selectin as a ligand for CD34 and CD34 overexpression experiments in hematopoietic cells indicate a role for CD34 in cell adhesion and inhibition of hematopoiesis. 4,5 QBEND10 is a monoclonal anti-cd34 antibody raised against human placental endothelial cells. 6 The immunohistochemical detection of CD34 protein using QBEND10 antibody is a relatively novel method for detection of myeloid blasts in MDS or acute myeloid leukemia, as well as for detection of lymphoblasts in acute lymphoblastic leukemia in bone marrow trephine sections. 7,8 Accepted for publication February 14, 2002. From the Department of Pathology, The Norwegian Radium Hospital, Oslo, Norway. Reprints: Emina Emilia Torlakovic, MD, Department of Pathology,The Norwegian Radium Hospital, Montebello, 0310 Oslo, Norway (e-mail: emina.torlakovic@labmed.uio.no). Horny and coauthors 7 have used a quantitative determination of CD34 cells per high-power field and semiquantitative evaluation using a scale from 0 to 4. However, a recent article by Baur and coauthors 9 showed that precise quantitation of C34 mononuclear cells (MNCs) in 500 or more nucleated cells is possible and useful in diagnosis and classification of MDS. Since the beginning of our practice to count CD34 MNCs in bone marrow trephine biopsies, we have noticed that there are reactive bone marrows and marrows involved by non-hodgkin lymphoma (NHL) that have CD34 MNC counts greater than 5%. The implication of this finding would be that it is not possible to use CD34 MNC counts for the classification of MDS. This study was designed to count CD34 MNCs in the trephine sections with a focus on conditions that should not have an increase of myeloid blasts. Our study confirms that if an optimized immunohistochemical method is used, it is possible to reproducibly count CD34 MNCs, as was previously suggested, 7 9 but the counts may be higher than 5% in bone marrow biopsies of patients with either reactive or malignant lymphocytosis. We also counted megakaryocytes expressing CD34. METHODS Case Selection One hundred eight bone marrow biopsies were retrieved from the files of the Department of Pathology, The Norwegian Radium Arch Pathol Lab Med Vol 126, July 2002 CD34/QBEND10 Immunostaining in Bone Marrow Biopsies Torlakovic et al 823
Hospital, Oslo, Norway. The selection was designed to include various diagnoses in which no increase in the number of myeloid blasts is expected. Included were 60 cases of reactive bone marrow with various diagnoses (infection, staging for Hodgkin disease and NHL, autoimmune disorders, and others), 24 cases of bone marrow biopsy partially involved with NHL (various diagnoses, most frequently follicular lymphoma and chronic lymphocytic leukemia), 8 cases of myeloproliferative disease, and 16 cases of MDS. Bone marrow smears or imprints were available in 41 (44%) of the 108 cases. The patients ages ranged from 5 to 89 years (mean 58 years, median 60 years). Two of the authors (G.T. and E.T.) reviewed all the cases. Immunohistochemistry B5-fixed trephine biopsies were decalcified, embedded in paraffin, cut at 4 m, and manually immunostained using anti-cd34 antibody (dilution 1:200; QBEND10; Monosan, Uden, Netherlands) after epitope unmasking by heating in EDTA at ph 8.9 in a 1000-W microwave oven for 20 minutes (4 5 minutes). Thereafter, the EnVision method (Dako, Golstrup, Denmark) was used according to the manufacturer s instructions. The sections were counterstained with hematoxylin. All cases showed very strong positivity in endothelial cells, which were used as an internal control. Anti terminal deoxynucleotidyl transferase (TdT) polyclonal antibody (dilution 1:50; Supertechs, Bethesda, Md) was used in selected cases. Epitope unmasking was performed using heatinduced epitope retrieval (as for QBEND10). Immunostaining was performed using Ventana Medical Systems Basic DAB Detection Kit 250-001 (Ventana Medical Systems Inc, Tucson, Ariz) according to the manufacturer s instructions. The staining was performed in a gen II automated immunostainer (Ventana). Counting and Identification of CD34 MNCs and TdT MNC Clusters Counting was performed at 1000 magnification using immersion oil. The counts were recorded using a cell counter (Assistant Counter, AC-8, Karl Hecht GmbH & Co, Sondheim/Rhön, Germany). The percentages of positive MNCs were recorded using 1 decimal point. Two counts of CD34 MNCs were performed for each biopsy, a 500-cell count and an independent 1000-cell count. The MNCs recorded as positive fulfilled 2 conditions: (1) the cells had a size consistent with myeloid blasts, and (2) no evidence of a tendency to form vascular spaces was found. Membranous, cytoplasmic, and Golgi localization of immunostaining was noted. Mononuclear cells with any of these patterns and the appearance of blasts were included in the count. The percentage of CD34 megakaryocytes was determined in at least 100 counted megakaryocytes in each case. Cells positive for TdT were counted in 1000 nucleated bone marrow cells. The presence of clusters of CD34 MNCs and TdT MNCs, as defined by Rimsza and coauthors 10 (ie, a contiguous group of at least 6 cells in a group or linear arrangement), was also recorded. Myeloid blasts were identified and counted as usual in the bone marrow aspirate smears routinely stained with Wright- Giemsa. Statistical Methods A Kruskal-Wallis test was used to study associations between different CD34 counts and the type of diagnosis. P values less than.05 were considered significant. Figure 1. A, Correlation of the 500-cell and 1000-cell counts. The counts were performed at 1000 magnifications. Random fields were chosen for counting. B, Correlation of the CD34 mononuclear cell (MNC) counts on trephine sections and myeloid blast counts on bone marrow aspirate smears or imprints. RESULTS CD34-Positive MNCs Comparison of the 500- and 1000-cell counts is shown in Figure 1, a. Comparison of the myeloid blast count in bone marrow smears and the CD34 MNC count in the trephine biopsies is shown in Figure 1, b. A trend for the CD34 MNC count to be slightly larger than the myeloid blast count was noted, but the difference was not statistically significant. The Table shows the association of CD34 MNC counts and the type of diagnosis. Twenty-five percent of bone marrow biopsies involved by NHL had CD34 MNC counts between 5% and 10%. All of the cases with CD34 MNC counts greater than 5%, including 5 (8.3%) of 60 reactive bone marrow specimens and 6 (33.3%) of 18 bone marrow specimens involved by NHL, also showed increased TdT counts (range, 4.0% 8.9%). This finding was also associated with either more than 10% lymphocytes in the differential count (in all of the cases in which a bone marrow smear was available) or mild to moderate lymphocytosis in the trephine sections (as shown by morphologic examination and CD3 and CD20 immunostaining [results not shown]). When reactive bone marrow specimens were analyzed separately, 24 (40%) of 60 had lymphocytosis, and 5 (20.8%) of those 24 were associated with increased CD34 MNC counts between 5% and 10%. These findings indicate that the reason 824 Arch Pathol Lab Med Vol 126, July 2002 CD34/QBEND10 Immunostaining in Bone Marrow Biopsies Torlakovic et al
CD34 MNCs Association of CD34-Positive Mononuclear Cell Counts and Diagnosis* Diagnosis, No. (%) Reactive NHL MPD MDS Total No. (%) Up to 5% 6% 10% 11% 20% 20% 55 (92) 5 (8) 18 (75) 6 (25) 5 (63) 2 (25) 1 (12) 3 (19) 10 (62) 3 (19) 78 (72) 16 (15) 11 (10) 3 (3) Total 60 (100) 24 (100) 8 (100) 16 (100) 108 (100) * MNCs indicates mononuclear cells; NHL, non-hodgkin lymphoma involving bone marrow; MPD, myeloproliferative disease; and MDS, myelodysplastic syndrome. When all diagnoses were studied, the difference between categories was significant (P.001, Kruskal-Wallis test). When MPD and MDS are excluded, the difference between reactive bone marrow and NHL-involved bone marrow was not significant (P.63). The percentage of TdT cells was between 4.0% and 8.9% in these 11 cases. Figure 2. The number of cases gradually decreases with the increasing number of CD34 megakaryocytes. Most of the cases with positive megakaryocytes represent bone marrow biopsies with reactive changes. for increased CD34 MNCs in the trephine biopsy specimens may be lymphocytosis associated with increased hematogones. That hematogones are causing increased CD34 MNCs is also suggested in Figure 1, b. Even though the difference in the counts was not statistically significant, a trend to have lower counts on the bone marrow smears is evident. Although at least a subpopulation of hematogones is CD34 and as such is included in the CD34 MNC count, they are not included in the myeloid blast count obtained by morphologic examination of the bone marrow smears. The finding of increased TdT cells in the bone marrow samples with increased CD34 MNCs (between 5% and 10%) supports this contention. The mean age of the patients with increased CD34 MNCs between 5% and 10% and increased TdT MNCs was 48 years, and the median was 50 years. No evidence of clusters of CD34 MNCs was found in any of the reactive bone marrow biopsies or in those partially involved by NHL, but at least 1 such cluster was found in all cases of MDS that had increased CD34 MNCs. Clustering of TdT MNCs was also not present. CD34-Positive Megakaryocytes The number of CD34 megakaryocytes and the intensity of immunostaining varied significantly from biopsy to biopsy and from one megakaryocyte to another in the same biopsy, respectively. Figure 2 shows the relation of the percentage of CD34 megakaryocytes and the number of biopsies. About one third of all cases showed low-level expression of CD34 in megakaryocytes. As the number of CD34 megakaryocytes increased, the number of cases proportionally decreased. However, no association between the number of CD34 megakaryocytes and the type of diagnosis or the number of CD34 MNCs was found. Twenty-seven (52%) of 52 reactive bone marrow biopsies had between 5% and 30% CD34 megakaryocytes, and 14 (27%) of 52 had between 30% and 70%. Generally, some smaller forms expressed membranous CD34 only rarely (Figures 2 and 3). Golgi-type and diffuse cytoplasmic positivities were noted in most megakaryocytes that were immunoreactive for CD34. Occasionally, megakaryocytes with naked nuclei also expressed CD34, but the locali- Arch Pathol Lab Med Vol 126, July 2002 CD34/QBEND10 Immunostaining in Bone Marrow Biopsies Torlakovic et al 825
Figure 3. Bone marrow trephine sections immunostained by QBEND10. A, CD34 megakaryocytes (arrow, full line) and CD34 mononuclear cells showing granular cytoplasmic positivity (arrowheads) and Golgi apparatus, diffuse cytoplasmic, and membranous positivity (arrow, dotted line) (original magnification 1000). B, Blasts with CD34 positivity localized to Golgi apparatus (arrow, dotted line) (original magnification 1000). C, Megakaryocytes: cytoplasmic and membranous expression of CD34 (original magnifications 500 [left upper corner, right upper corner, and left lower corner] and 1100 [right lower corner]). D, Megakaryocytes with lower nuclear-cytoplasmic ratios negative for CD34 (arrows) (original magnification 1100). zation of the signals was not possible due to high nuclearcytoplasmic ratios. In addition, no association was found between the content or the distribution of megakaryocytes and the number of CD34 megakaryocytes. Specifically, 3 of 8 cases of myeloproliferative disease with a diagnosis of essential thrombocythemia and increased megakaryocytes had low numbers of CD34 megakaryocytes (1%, 2%, and 11%). COMMENT Determination of the number of myeloid blasts is one of the most important diagnostic tasks in the morphologic 826 Arch Pathol Lab Med Vol 126, July 2002 CD34/QBEND10 Immunostaining in Bone Marrow Biopsies Torlakovic et al
study of bone marrow specimens in the evaluation of MDS. 2,11 Any increase greater than 5% myeloid blasts in bone marrow aspirate smears or imprints is accepted as evidence of excess blasts and is associated with an increased risk of progression as an independent prognostic variable. According to the French-American-British Morphology Group classification, a diagnosis of refractory anemia (RA) can be used only if the number of myeloid blasts is less than 5%, RA with excess blasts is appropriate if the number of blasts is between 5% and 20%, and the diagnosis is RA with excess blasts in transformation if the number of myeloid blasts is between 21% and 30%. 2 Despite phenomenal scientific advances that have provided new insights into the biology and genetics of these disorders, the exact enumeration of myeloid blasts remains essential in the evaluation of MDS and has diagnostic and prognostic importance. 12 The blast percent in MDS carries a risk score of 0.5 for blast percentages of 5% to 10%, 1.5 for blast percentages of 11% to 20%, and 2.0 for blast percentages of 21% to 30%. 11 This rule could not be applied for the cases of RA studied by Baur and coauthors 9 because they found no evidence of increased risk of progression in their RA cases that had more than 5% CD34 MNCs. Our study shows that a finding of 5% to 10% CD34 MNCs is not unusual in reactive bone marrow biopsies with lymphocytosis or in bone marrow biopsies involved by NHL. We also showed that the number of myeloid blasts in bone marrow smears is generally less than the number of CD34 MNCs. The difference in counts is explained in our study by simultaneous detection of increased counts of TdT MNCs. Hematogones are B-lymphocyte progenitor cells, which coexpress CD34 and TdT during their first stage of differentiation. 13 Our study also shows that hematogones may appear in adults even without significant lymphocytosis (already at 10% lymphocytes in the bone marrow). We suggest that increased hematogones should be ruled out even in an adult population before a diagnosis of excess blasts is made based on CD34 MNC counts on the trephine biopsy. Even though no such increase of CD34 MNCs was found by Horny and coauthors, 7 their study is not comparable to ours because of the lower sensitivity of their methods of immunostaining. Almost half of their bone marrow biopsies designated as normal and reactive did not contain any CD34 MNCs. The authors performed their immunostains with the avidin-biotin-peroxidase complex (ABC) method and did not use antigen-retrieval methods. That the number of CD34 MNCs does not necessarily equal the number of myeloid blasts was also shown by Oertel and coauthors, 14 who found in their study of bone marrow aspirates from healthy donors that other MNCs that can be positive for CD34 are promonocytes and lymphocytes (hematogones?). In accordance with the findings of Rimsza et al, 10 we found no evidence of clusters of CD34 MNCs in any of the reactive bone marrow specimens. Our results additionally support the conclusion from their study that the finding of even a single cluster of 6 or more CD34 MNCs is a significant finding and also suggest that the significance of finding clusters of CD34 MNCs is probably the same in acute lymphoblastic leukemia and MDS/acute myeloid leukemia. Baur and coauthors 9 also found that it was the presence of clusters of CD34 MNCs (defined as 3 positive cells), rather than the absolute number larger than 5%, which was significantly associated with progression of RA. We have determined that a 500-cell count is representative and yields results identical to a 1000-cell count. Even in a diagnostic setting, performing a 500-cell count using a cell counter takes very little time, because its sole purpose is to record positive MNCs. The clusters may not be readily appreciated using 400 magnification. The blasts with weaker expression of CD34, regardless of its cellular localization (membranous, cytoplasmic, Golgi), could be difficult to identify at that power of magnification, but should also be accounted for. The detection of cytoplasmic CD34 is not unexpected. There are intracellular stores of CD34 protein that can be quickly translocated to the plasma membrane in response to extracellular signals, which enable rapid up-regulation of CD34 surface expression in normal human bone marrow cells. 15 Proliferation of small vessels does not prevent precise counting, because it is possible to differentiate mononuclear blastlike cells from endothelial cells. While generally there is no need to perform CD34 MNC counts on trephine biopsies if good bone marrow aspirate smears or imprints are available, we would recommend such counts if the smear quality is not optimal and/or the blast count is difficult, as well as when no smears are available. The latter is often the case in some European countries, where this type of blast quantitation could be both useful and necessary. Our study shows that the CD34 MNC count on the trephine biopsy gives at least a good approximation of the absolute number of myeloid blasts in the bone marrow. Great caution is necessary in the interpretation of the counts between 5% and 10% CD34 MNCs, with particular consideration for hematogones. Megakaryocyte progenitors express the CD34 antigen, but the precise stage along the megakaryocytic differentiation at which the CD34 is turned off is not known. Debili and coauthors 16 found that most double-stained (CD34,GPIb ) cells were polyploid in their cell culture study, proving that CD34 is still expressed on polyploid transitional immature megakaryocytes. Their findings also suggested that the expression of CD34 is related to the ability of the megakaryocytic precursors to accomplish DNA synthesis. However, only surface CD34 expression was addressed since the authors used flow cytometry to study CD34 expression. Only rare smaller megakaryocytes were found to express membranous CD34 in our study (Figure 3, c). The great majority of the CD34 megakaryocytes showed only Golgi or diffuse cytoplasmic localization of CD34. Expression of CD34 by mature megakaryocytes has been reported previously by Oertel and coauthors, 14 who found from 1% to 11% positive megakaryocytes in the CD34 fraction of bone marrow aspirates from 6 healthy donors. Machhi and coauthors 17 considered a staining level of 20% or more of the megakaryocytes to be significant and associated with MDS. We found no association between the percentage of CD34 megakaryocytes and any diagnostic category. Two of 4 cases with the largest number of positive megakaryocytes were reactive bone marrow biopsies; one was performed for the staging of classical Hodgkin disease, but was not involved by Hodgkin disease, and the other involved a patient with thrombocytopenia secondary to hypersplenism. The other 2 cases included RA with excess blasts and acute myeloid leukemia-m7. Overall, about 50% of the biopsies with various diagnoses had more than 20% CD34 megakaryocytes. However, we did find one rule about CD34 immunoreactivity in megakaryocytes: The Arch Pathol Lab Med Vol 126, July 2002 CD34/QBEND10 Immunostaining in Bone Marrow Biopsies Torlakovic et al 827
frequency of the cases decreases with the increasing percentage of CD34 megakaryocytes. The meaning and the significance of the cytoplasmic CD34 immunoreactivity with QBEND10 in mature megakaryocytes are not clear and should be addressed in future studies. References 1. Bennett JM, Catovsky D, Daniel MT, et al. Proposals for the classification of the acute leukaemias (FAB) co-operative group. Br J Haematol. 1976;33:451 458. 2. Bennett JM, Catovsky D, Daniel MT, et al. Proposals for the classification of the myelodysplastic syndromes. Br J Haematol. 1982;51:189 199. 3. Krause DS, Fackler MJ, Civin CI, et al. CD34: structure, biology, and clinical utility. Blood. 1996;87:1 13. 4. Baumheter S, Singer MS, Henzel W, et al. Binding of L-selectin to the vascular sialomucin CD34. Science. 1993;262:436 438. 5. Fackler M, Krause D, Smith O, et al. Full-length but not truncated CD34 inhibits hematopoietic cell differentiation of M1 cells. Blood. 1995;85:3040 3047. 6. Fina L, Molgaard HV, Robertson D, et al. Expression of the CD34 gene in vascular endothelial cells. Blood. 1990;75:2417 2426. 7. Horny H-P, Wehrmann M, Schlicker HUH, et al. QBEND10 for the diagnosis of myelodysplastic syndrome in routinely processed bone marrow biopsy specimens. J Clin Pathol. 1995;48:291 294. 8. Toth B, Wehrmann M, Kaiserling E, et al. Immunophenotyping of acute lymphoblastic leukaemia in routinely processed bone marrow biopsy specimens. J Clin Pathol. 1999;52:688 692. 9. Baur AS, Meugé-Moraw C, Schmidt P-M, et al. CD34/QBEND10 immunostaining in bone marrow biopsies: an additional parameter for the diagnosis and classification of myelodysplastic syndromes. Eur J Haematol. 2000;64:71 79. 10. Rimsza LM, Viswanatha DS, Winter SS, et al. The presence of CD34 cell clusters predicts impending relapse in children with acute lymphoblastic leukemia receiving maintenance chemotherapy. Am J Clin Pathol. 1998;110:313 320. 11. Greenberg P, Cox C, LeBeau MM, et al. International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood. 1997;89:2079 2088. 12. Willman CL. Acute leukemias: a paradigm for the integration of new technologies in diagnosis and classification. Mod Pathol. 1999;12:218 228. 13. Longacre TA, Foucar K, Crago S, et al. Hematogones: a multiparameter analysis of bone marrow precursor cells. Blood. 1989;73:543 552. 14. Oertel J, Oertel B, Schleicher J, et al. Immunotyping of blasts in human bone marrow. Ann Hematol. 1996;72:125 129. 15. Fackler MJ, Civin CI, May WS. Up-regulation of surface CD34 is associated with protein kinase C-mediated hyperphosphorylation of CD34. J Biol Chem. 1992;267:17540 17546. 16. Debili N, Issaad C, Masse JM, et al. Expression of CD34 and platelet glycoproteins during human megakaryocytic differentiation. Blood. 1992;80:3022 3035. 17. Machhi J, Bunyi-Teopengco E, Chang C, et al. CD34 expression in megakaryocytes favors myelodysplasia [abstract]. Mod Pathol. 2001;14:171A. 828 Arch Pathol Lab Med Vol 126, July 2002 CD34/QBEND10 Immunostaining in Bone Marrow Biopsies Torlakovic et al