Human Pathology (2015) xx, xxx–xxx
www.elsevier.com/locate/humpath
Original contribution
Nuclear factor-erythroid 2, nerve growth factor receptor, and CD34–microvessel density are differentially expressed in primary myelofibrosis, polycythemia vera, and essential thrombocythemia☆,☆☆,★ Nuri Yigit MD a,b,⁎,1 , Shannon Covey MD a , Sharon Barouk-Fox MA a , Turker Turker MD b , Julia Turbiner Geyer MD a , Attilio Orazi MD a a
Department of Pathology and Laboratory Medicine, Weill Cornell Medical College/New York–Presbyterian Hospital, New York, NY 10065 b Department of Pathology, Gulhane Military Medical Academy and School of Medicine, Ankara 06010, Turkey Received 24 February 2015; revised 5 May 2015; accepted 7 May 2015
Keywords: Myeloproliferative neoplasms; NF-E2; NGFR; Microvessel density; Marrow stromal cells
Summary Because of the presence of various overlapping findings, the discrimination of polycythemia vera (PV) from prefibrotic/fibrotic primary myelofibrosis (PF/F-PMF) and essential thrombocythemia (ET) may be challenging, particularly in suboptimal bone marrow biopsy specimens. In this study, we assessed whether differences in the expression of nuclear factor-erythroid 2 (NF-E2), nerve growth factor receptor (NGFR; CD271), CD34, CD68, p53, CD3, CD20, and CD138 by immunohistochemistry could be useful in separating among them. Higher frequencies of nuclear positive erythroblasts with NF-E2 were observed in ET and PV cases (50% ± 13.3% and 41.5% ± 9.4%, respectively) when compared with both PF-PMF (21% ± 11.7%) and F-PMF (28.5% ± 10.8%). We found that with a cutoff level of at least 30% nuclear staining for NF-E2 in erythroblasts, we could reliably exclude the possibility of PMF. Conversely, NGFR+ stromal cells per highpower field (HPF) was significantly increased in F-PMF (53.5 ± 19.1/HPF) and PF-PMF (13.5 ± 3.8/HPF) compared with ET (4.4 ± 2.2/HPF) and PV (6.6 ± 3.3/HPF). Similarly, differences in CD34–microvessel density was remarkable in F-PMF and PF-PMF cases in comparison with PV and ET (49.9 ± 12.1/HPF, 29.3 ± 12.4/HPF, 13.7 ± 4.6/HPF, and 11.9 ± 5.1/HPF, respectively). Thus, the assessment of NF-E2 and NGFR expression and the evaluation of CD34–microvessel density may provide additional support in reaching a correct diagnosis in these cases of myeloproliferative neoplasms. © 2015 Elsevier Inc. All rights reserved.
☆
Disclosures: The authors declare no conflict of interest and no relevant financial or nonfinancial relationship. Abstract of this material was accepted as a poster presentation at United States and Canadian Academy of Pathology 2015 annual meeting in Boston, MA, USA. ★ Conflict of interest statement: Authors declare no conflict of interest. ⁎ Corresponding author at: Department of Pathology and Laboratory Medicine, Weill Cornell Medical College/New York–Presbyterian Hospital, 525 East 68th St, Starr Pavilion, 715 New York, NY 10065. E-mail addresses: [email protected], [email protected] (N. Yigit), [email protected] (S. Covey), [email protected] (S. Barouk-Fox), [email protected] (T. Turker), [email protected] (J. T. Geyer), [email protected] (A. Orazi). 1 Permanent address: Department of Pathology, Gulhane Military Medical Academy and School of Medicine, Etlik Kecioren, Ankara 06010, Turkey. ☆☆
http://dx.doi.org/10.1016/j.humpath.2015.05.004 0046-8177/© 2015 Elsevier Inc. All rights reserved.
2
1. Introduction The diagnosis of polycythemia vera (PV), prefibrotic/fibrotic primary myelofibrosis (PF/F-PMF), and essential thrombocythemia (ET), the main subtypes of Ph′ chromosome– negative/BCR-ABL1–negative myeloproliferative neoplasms (MPNs), is currently based on the combination of histologic, clinical, and laboratory parameters [1]. Selected clinical (eg, presence/absence of splenomegaly), laboratory (lactate dehydrogenase [LDH] and erythropoietin [EPO] levels), and a number of histologic features of the bone marrow (BM) such as size, pleomorphism, nuclear complexity and nuclear shape of megakaryocytes, status of iron storage in marrow, and certain peripheral blood findings like leukoerythroblastosis may help in distinguishing among them [2,3]. However, especially in the early stages of disease and/or in the presence of morphologically suboptimal BM specimens, their differential diagnosis is frequently difficult. This is largely due to their partially overlapping morphologic features (eg, similar degree of marrow cellularity and megakaryocytic proliferation) and molecular findings, and/or the lack of adequate clinical information. In such cases, a diagnosis of MPN, unclassifiable, is often rendered [4]. A precise diagnosis of MPN subtype is crucial because of the distinct clinical courses and outcomes of these entities [5]. This ambiguity has stimulated efforts to search for more reliable methods for confirming a specific diagnosis. Among these efforts, differences in STAT5 phosphorylation levels or in the expression frequency of thrombopoietin receptor in megakaryocytes have been proposed as useful diagnostic tools [6,7]. Goerttler et al [8] demonstrated that overexpression of nuclear factor-erythroid 2 (NF-E2), a transcription factor that has a regulating function on erythroid and megakaryocytic maturation, is a typical finding in PV and that NFE2 gene alterations are able to cause erythrocytosis and thrombocytosis in a murine model. More recently, a study from the same group described potential diagnostic usefulness of immunostaining for NF-E2 in separating subtypes of MPN using BM biopsies [9]. In addition to investigating the characteristics of the neoplastic cells potentially useful to separate these diseases, efforts have also been focused on differences in stromal components collectively known as BM microenvironment such as the histologic assessment of neoangiogenesis by microvessel density (MVD) counting. MVD had been previously investigated in MPN cases and found to be consistently higher in PMF cases than in other subtypes [10]. However, variable results in ET and PV have been obtained in other studies [11,12]. The expression of nerve growth factor receptor (NGFR), also known as CD271, has been proposed as a useful marker to identify BM reticular stromal cells in BM biopsies [13–15]. The NGFR-positive cells have a close anatomical interaction with nearly all types of hematopoietic cellular elements and play an essential role in regulation of hematopoiesis. NGFR
N. Yigit et al. expression has been recently correlated with other BM features including the degree of fibrosis in patients with primary immune thrombocytopenia, treated with a recombinant thrombopoietin agent [16]. p53 overexpression is a known marker of aggressiveness both in myeloid and in lymphoid neoplasm, but there is no information in relation to a possible differences in the degree of positivity in subtypes of MPN [17,18]. Similarly, there is only limited information in relation to possible differences in the quantity of “inflammatory cells” such as reactive B and T lymphocytes, histiocytes, and plasma cells found in the cellular background of these MPN subtypes. In this study, we investigated NF-E2 expression in erythroblasts, p53 overexpression in BM cells, NGFR expression, and degree of MVD in marrow stroma, as well as the frequency of lymphocytes, plasma cells, and histiocytes in the reactive cellular background.
2. Materials and methods 2.1. Case selection We retrospectively selected 50 cases in total, including 10 cases of PV, PF-PMF, F-PMF, ET, and control groups each, among the files of the Department of Pathology and Laboratory Medicine of Weill Cornell Medical College/ New York–Presbyterian Hospital in New York City. Clinical, radiologic, and laboratory data including LDH levels, complete blood counts, presence of splenomegaly, and conventional/molecular cytogenetics from all patients have also been retrieved from the hospital information system. All biopsies were the initial diagnostic samples in the chronic phase at the disease outset without any history of treatment. The control group was composed of staging BM biopsies, all of which were uninvolved. BM core biopsy and aspirate samples were routinely obtained from the iliac spine. All were fixed in Bouin solution, decalcified by using a hydrochloric acid/ethylenediaminetetraacetic acid (EDTA) combination, processed routinely, and embedded in paraffin. Biopsies were stained with hematoxylin and eosin, Gomori silver impregnation, and Masson trichrome, whereas BM aspirates and peripheral blood smears were Giemsa stained. We also had clot biopsies of 10 ET, 4 PF-PMF, 5 F-PMF, and 4 PV cases, all of which were fixed in formalin, along with the core biopsies that were processed routinely. The cases diagnosed before 2008 were reevaluated and reclassified according to the 2008 World Health Organization classification criteria [19]. The degree of fibrosis was graded on a scale of MF-0 to MF-3 in accordance with the European Consensus Grading System [20]. The design of this study was approved by the institutional review board of Weill Cornell Medical College.
NF-E2, NGFR, and CD34–MVD in MPNs
3
2.2. Immunohistochemistry BM core biopsies were immunostained for NF-E2, NGFR, CD34, CD68, p53, CD3, CD20, and CD138 markers by using an autostainer (Bond-III; Leica Microsystems, Buffalo Grove, IL) following the manufacturer's protocol. We also performed double staining with NF-E2 and CD71 in selected cases of each group to verify whether NF-E2 expression was restricted to erythroid lineage or not. Appropriate staining was confirmed using external positive and negative controls for each antibody. The characteristics of the antibodies are summarized in Table 1.
2.3. Immunohistochemical evaluation All the stained slides have independently been evaluated by 3 blinded reviewers (N. Y., J. T. G., and A. O.). NF-E2 nuclear-stained erythroblasts were assessed as the percentage of the total erythroblasts, whereas the number of CD3, CD20, and CD138+ cells was calculated as the percentage of the total marrow cellularity. The mean values of NGFR, p53, and CD68+ cells were counted per high-power field (HPF; 0.238 mm2: 22-mm ocular lens and ×400 total magnification). For NGFR and CD68, only the highlighted bodies of the cells with a visible nucleus were counted, excluding the enlarged cytoplasmic dendritic structures. To assess MVD, the mean number of thin-walled vessels with or without a lumen and sinusoids visualized by CD34+ endothelial cells was counted per HPF. Arterioles and myeloid/megakaryocytic precursors, which are normally costained with CD34, were excluded during the evaluation. Also, the consecutively or discontinuously appearing branches of the same vessel or tortuous sinusoid on the section plane were ignored in order to avoid the bias of repeated counting. For all cases, the areas of highest vascularization (hot spots), which were appreciated visually, were selected for quantifying BM vascularity.
2.4. Statistical analysis All groups were tested using the nonparametric Kruskal-Wallis test to compare continuous variables among Table 1
groups, with Bonferroni correction. A P value less than .05 was considered statistically significant. Receiver operator characteristic analysis and the value of area under the curve were constructed to assign a cutoff level for staining percentages, which can be used in discriminating each diagnostic category from others. All analyses were performed using SPSS for Windows version 17.0 statistical software package (SPSS, Chicago, IL).
3. Results All patients were in the chronic phase of their diseases. The mean age of the groups was 60 years (range, 18-91 years) with a male-to-female ratio of 0.81. Cytogenetic analysis demonstrated a normal karyotype in all ET cases; 2 del20q, 1 del9q, and 1 trisomy 8 in PF-PMF; 3 del20q, 1 delXq, and 1 monosomy 18 in F-PMF; and 1 trisomy 9 in PV cases, respectively. Fibrosis scores, JAK2 status, and immunohistochemical staining characteristics are summarized in Table 2. Calculated mean and median values of the stains in all groups were correlated with minor differences. Statistically significant values were identified for NF-E2, NGFR, and MVD (Fig. 1). In core biopsies of all cases, NF-E2 positivity was seen predominantly in a nuclear localization for early stage erythroblasts and cytoplasmic in late stage. All erythrocytes were strongly positive. ET and PV cases had higher frequencies of nuclear positive erythroblasts (50% ± 13.3% and 41.5% ± 9.4%, respectively) when compared with both PF-PMF (21% ± 11.7%) and F-PMF (28.5% ± 10.8%; P b .001; Fig. 2). We have confirmed the specificity of NF-E2 to erythroid lineage by using double immunohistochemical staining, which was composed of NF-E2 plus CD71. We have observed a consistent staining pattern with NF-E2, overlapping to the single-stained cores (Fig. 3A-C). In the receiver operator characteristic analysis for establishing a cutoff value for differentiating ET and PV from PF/F-PMF, the calculated area under the curve value was found to be 0.94 (95% confidence interval, 0.853-1.027). Thus, when 30% of staining was accepted as the cutoff level, the sensitivity and specificity
Characteristics of the antibodies
Marker
Company
Clone
Host
Dilution
Antigen retrieval method and time (min)
NF-E2, St. Louis, MO NGFR, Cambridge, MA CD34, Buffalo Grove, IL CD68, Carpinteria, CA p53, Fremont, CA CD3, Buffalo Grove, IL CD20, Carpinteria, CA CD138, Raleigh, NC CD71, Buffalo Grove, IL
Sigma Abcam Leica Dako Biogenex Leica Dako AbD Serotec Leica
Polyclonal ME20.4 QBEnd/10 PG-M1 1801 PS1 L26 B-A38 10F11
Rabbit Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse
1:50 1:1600 1:100 1:300 1:200 1:100 1:200 1:25 1:100
H2 (20) E1 (10) H2 (20) H1 (30) H2 (20) H2 (30) H1 (30) H1 (30) H1 (30)
Abbreviations: H1, Bond Epitope Retrieval Solution 1: a citrate-based pH 6.0 solution; E1, Bond Enzyme Pretreatment Kit: 1 drop Bond enzyme concentrate (proteolytic enzyme) with 7 mL of Bond enzyme diluent (Tris-buffered saline); H2, Bond Epitope Retrieval Solution 2: an EDTA-based pH 9.0 solution.
5:5 7:3
3:7
PF-PMF (n = 10) 61 (26-91) F-PMF (n = 10) 72 (29-90)
ET (n = 10)
6/10 55 49 (18-68)
5/10 4/10 78 84
Abbreviations: PV, polycythemia vera; PF-PMF, prefibrotic stage of primary myelofibrosis; F-PMF, fibrotic stage of primary myelofibrosis; ET, essential thrombocythemia; NF-E2, nuclear factor-erythroid 2; NGFR, nerve growth factor receptor; HPF, high-power field. a The value was statistically significant in comparison with PF/F-PMF cases (P b .001). b The difference was statistically significant compared with PV and ET cases (P b .001). c Statistically significant relationship was observed between ET and PV cases (P = .021).
13.6 ± 3.8 c 5.8 ± 3.4 5.5 ± 2.5 0 24.4 ± 8.5 11.9 ± 5.1 4.4 ± 2.2
3.1 ± 2.3 5.7 ± 2.2 3.4 ± 2.3 4.4 ± 2.4 9±4 9.1 ± 3.3 0 0 13.5 ± 3.8 b 29.3 ± 12.4 b 21.8 ± 5.6 53.5 ± 19.1 b 49.9 ± 12.1 b 18.9 ± 10.5
3.8 ± 1.3 8.8 ± 2.6 0 22.7 ± 5.9 13.7 ± 4.6 6.6 ± 3.3
= 8) 41.5 ± 9.4 a = 2) = 10) 21 ± 11.7 = 6) 28.5 ± 10.8 = 4) = 10) 50 ± 13.3 a 83
10/10
MF-0 MF-1 MF-1 MF-2 MF-3 MF-0 3:7 58 (44-78) PV (n = 10)
(n (n (n (n (n (n
CD20 (%) CD68 TP53 CD3 (%) (PG-M1; /HPF) (/HPF) CD34+ microvessels (/HPF) NF-E2 (nuclear NGFR staining in (/HPF) erythroblasts; %) Mean age Sex Mean JAK2+ Fibrosis and range (y) (male/ cellularity cases female) (%) Cases
Cases' characteristics and staining profiles of BM core biopsies with mean values Table 2
5 ± 1.9
N. Yigit et al. CD138 (%)
4
were calculated as 85% and 90%, respectively. In the ET and PMF groups, NF-E2 overexpression was not correlated with the status of JAK2 mutation or allele burden, which also appeared unrelated for PV cases. Only rare myeloid precursors and a variable proportion of megakaryocytes in the sections were highlighted as seen in one of previously published study [9]. The megakaryocytes had the similar staining patterns among all the groups. In addition, plasma cells had marked perinuclear localization of NF-E2 with a granular quality in all groups. The number of NGFR+ stromal cells was significantly increased in F-PMF (53.5 ± 19.1/HPF) and PF-PMF (13.5 ± 3.8/HPF) when compared with ET and PV (P b .001; Fig. 4A and B). Also, a statistically meaningful difference was observed between F-PMF and PF-PMF (P b .001). The dendritic structures of stromal cells were thicker and more continuous in all PMF cases. In addition, conspicuous peritrabecular osteoblastic rimming highlighted by NGFR was noted around the newly developing osteosclerotic foci. Two PV cases having MF-1 showed NGFR density similar to that of PF-PMF. MVD was assessed in F-PMF, PF-PMF, PV, and ET cases, at 49.9 ± 12.1/HPF, 29.3 ± 12.4/HPF, 13.7 ± 4.6/HPF, and 11.9 ± 5.1/HPF, respectively. Statistically, increased MVD was detected in PF/F-PMF cases in comparison with PV and ET (P b .001; Fig. 5A and B). Four of 10 F-PMF cases showed microvascular hot spots as well as apparently hypovascular regions in some intertrabecular spaces. Although we have also observed groups of T cells highlighted by CD3 in PMF cases, a statistically meaningful difference of its expression has been detected only with ET and PVs (more T cells in ET cases than PVs; P = .021). Variable amounts of CD20, CD138, and CD68+ inflammatory cells were observed in different MPN cases, and they did not show any statistically significant differences among disease groups. None of the biopsies were positive with p53 (all b1%). Stromal and inflammatory cells were generally dispersed within the marrow, except with some lymphoid aggregates in a few cases. Four F-PMF cases had newly vascularized hot spots, accompanied by increased numbers of NGFR+, CD3+, and CD20+ cells, as well as decreased CD68+ histiocytes. The remaining vessel-poor areas on the same biopsy sections showed opposite immunoexpression profiles for these markers when compared with the newly vascularized hot spots. We have found the lowest levels of expression of NF-E2, NGFR, and CD68 in the control groups, when compared with PV, PMF, and ET. Likewise, MVD was much lower in control cases. (NF-E2 was positive in 12% ± 4.8% of erythroblasts; NGFR+ stromal cells, 3.8 ± 0.8/HPF; CD68+ cells, 7.2 ± 1.4/HPF; and MVD, 4.6 ± 1/HPF.) In clot biopsies, staining profiles of all antibodies demonstrated quite similar characteristics to the cores. Unfortunately, statistical analysis could not be performed because of the limited number of clot samples.
NF-E2, NGFR, and CD34–MVD in MPNs
Fig. 1
5
Median values and ranges of staining for NF-E2, NGFR, and MVD in all groups.
Fig. 2 A and B, BM morphologic features with hematoxylin and eosin stain and nuclear NF-E2 staining in most of the erythroblasts of ET. C and D, NF-E2 similarly marks many erythroblasts of PV. E and F, Prefibrotic primary myelofibrosis showing decreased numbers of nuclear stained erythroblasts compared with ET and PV. A, C and E, hematoxylin and eosin stain, original magnification ×200; B, D and F immunoperoxidase, ×400.
6
N. Yigit et al.
Fig. 3 A, By double immunohistochemical staining with NF-E2 (brown chromogen) and CD71 (red chromogen), normal BM biopsy shows that NF-E2 is mainly restricted to the erythroblasts (orange arrow), although some are not marked (violet arrow). Only rare myeloid precursors are positive for NF-E2 (black arrow). B, Many erythroblasts of PV showing NF-E2 positivity. Megakaryocytes are also nuclearly positive for NF-E2. C, In prefibrotic primary myelofibrosis, the NF-E2 labels significantly less erythroblasts than in ET. A-C, immunoperoxidase, original magnification ×1000.
4. Discussion Separating PV from ET or PMF particularly in patients seen early in the course of disease can be challenging due to their partially overlapping features. In some of these patients, clinical findings and laboratory results may be inconclusive. The problem is more typically seen in cases of early phase PV and in PMF, both of which may be confused with ET. In early PV, the diagnostic cutoff for establishing erythrocytosis is often not met initially, but only later in the course of the disease. These cases have been recently referred to as “masked PV” [21,22]. In these cases, only BM histology and red cell mass determination allow for separation from ET [22]. Similarly, PMF patients can develop splenomegaly, a minor criterion for diagnosing PMF, only months to years after their initial presentation. Elevated LDH levels, another minor parameter for the diagnosis of PMF, cannot be detected or might not be available in some institutions. Although differences in the frequency of JAK2, MPL, or CALR gene mutations can be useful in separating among these subtypes of MPN, most patients share in common JAK2 V617F. These issues underline the importance of
finding new reliable tools that can be used to separate PV and PMF from ET. NFE2 gene encodes NF-E2, which is a hematopoietic transcription factor and binds conserved functional elements present in globins and many other erythroid/megakaryocytic genes [23]. NF-E2 displays an essential role for regulating erythroid and megakaryocytic maturation. Its overexpression was detected in most patients with PV, ET, and PMF [8,24]. Bogeska and Pahl [25] investigated NF-E2 role on pathogenesis of PV and found that elevated NF-E2 expression caused aberrant expansion of the stem and progenitor cell compartments. Conversely, decreased NF-E2 levels resulted in diminished numbers of hematopoietic stem cells as well as multipotent, common myeloid, and granulocyte-monocyte progenitors, but megakaryocyte-erythroid progenitors were nearly not affected. Hence, they concluded that elevation of NF-E2 expression is required and responsible for the EPO-independent erythroid maturation, which is a pathognomonic hallmark of PV. Mutschler et al [26] stated that NF-E2 overexpression delays the early phase of erythroid maturation resulting in an expansion of erythroid progenitors and overproduction of erythrocytes derived from each progenitor cell.
Fig. 4 A, Extensive NGFR+ stromal cells in prefibrotic primary myelofibrosis (immunoperoxidase, original magnification ×200). B, Rare and dispersed NGFR+ stromal cells in ET (immunoperoxidase, ×400).
NF-E2, NGFR, and CD34–MVD in MPNs
7
Fig. 5 Comparison of MVD between prefibrotic primary myelofibrosis cases showing intensely stained microvessels as well as enlarged sinusoids (A) and ET with sparse microvessels (B). Immunoperoxidase, original magnification both ×200).
In addition, Kapralova et al [27] found elevated NFE2 transcripts in different primary and secondary polycythemias, which can be caused both by neoplastic and by nonneoplastic disorders. Therefore, they concluded that NF-E2 overexpression is not specific to MPNs. Interestingly, some studies also showed that the leukemogenic fusion proteins can repress the production of transcription factors involved in control of lineage differentiation, which results in tumorigenesis [28]. In this manner, expression level of NF-E2 was detected to be repressed by AML/ETO and PML/RAR fusion proteins, which was considered as a crucial step in leukemogenesis [29]. Kaufmann et al [30] generated mice overexpressing NF-E2 and observed many features of MPN, including thrombocytosis, leukocytosis, EPO-independent colony formation, and characteristic BM histology. Conversely, loss of NF-E2 was shown to result in anemia and severe thrombocytopenia [31]. In accordance with these studies, we observed immunohistochemical NF-E2 overexpression in erythroblasts in all 3 MPN entities. In addition, similar to most published studies, we did not find a correlation between NF-E2 overexpression and JAK2 status. The highest degree of nuclear overexpression (N30% of erythroblasts) was found exclusively in PV and ET, but not in PF/F-PMF. Aumann et al [9] investigated immunohistochemical expression levels of NF-E2 in MPNs. According to their findings, the highest nuclear staining was found in PMF (33.7% ± 10.7%) followed by PV (23.9% ± 10%) and ET (16.3% ± 4.1%), respectively. Consequently, they proposed using a cutoff level of 20% to discriminate PMF from ET. However, our findings failed to confirm this suggestion and the highest NF-E2 nuclear expression levels were seen in erythroblasts of ET patients in our series. This discrepancy may be caused by the usage of different fixative solutions. Our samples were all fixed in Bouin solution, whereas in a previous study we referenced, either formalin or calcium-glutaraldehydeformaldehyde was used [9]. Also, decalcifying agents may have some effect on the tissues which may result in such discordance on immunohistochemistry. We prefer a combina-
tion of hydrochloric acid and EDTA for decalcification. Instead, Aumann and colleagues used a mixture of EDTA disodium salt and Tris(hydroxymethyl)aminomethane for their cases. However, the results obtained using corresponding formalin-fixed clot sections, which were available in a proportion of our cases, showed comparable results raising the possibility that differences in the diagnostic criteria applied to these cases may have been responsible for the discrepancy. Previously published evidence showed that NGFR is a specific marker for BM stromal cells. These cells are components of the marrow microenvironment together with histiocytes, adipocytes, and endothelial cells and are believed to have a fibroblast-like function. They are hypothesized to have a regulatory role in the homing of stem cells, precursor cell differentiation, and releasing of mature cells to the circulation [32]. The NGFR-positive cells are characterized by oval-shaped nuclei, scant cytoplasm, and numerous elongated dendritic-like processes that surround hematopoietic cells, adipocytes, vessels, and sinusoids [14]. A recent study claimed that stromal cells lose their ability to support hematopoiesis and increase their matrix remodeling potential in MPN cases, which contributes to the development of myelofibrosis [33]. Our results showed a correlated increase in stromal cells, fibrosis, and vessels in PMF cases. These parameters were particularly high in F-PMF. It is also assumed that the dendritic structures of stromal cells cover the endosteal surfaces of the bone trabeculae [13]. Similar to these observations, NGFR in our study showed thick dendritic layering around the newly developing immature osteosclerotic foci. Although the thick, continuous, and bulbous meshwork of stromal cell projections was observed to intermingle with various marrow cells in PMF cases, ET showed mostly thin and interrupted projections of stromal cells, similar to the control group. According to the literature, the highest MVD was seen in PMF cases in accordance with the worst degree of fibrosis in MPN cases. Anemia, one of the minor criteria for the diagnosis of PMF, was hypothesized to be an agent to activate BM
8 neovascularization as a response against tissue hypoxia [34]. In the previous studies, high MVD in MPNs was found to be well correlated with hypercatabolic symptoms, vascular endothelial growth factor expression, and marked splenomegaly [10,11]. There was no significant correlation between MVD and the severity of anemia or degree of cellularity in our series, perhaps due to the limited number of studied cases [35]. We did not detect a correlation between MVD and degree of cellularity, anemia, LDH level, or splenomegaly among PMF cases, as in some previously reported studies [12]. Although CD3+ T cells were present in higher amounts in ET than PV, the difference was not significant when compared with PMF cases. On the other hand, benign inflammatory cells can be regulated by various disease-specific or nonspecific mechanisms during the course of disease due to alterations of mature blood cell counts. Thus, T-cell counting may be useful for discriminating ET from PV, when other required criteria fit the diagnosis of ET.
5. Conclusions To summarize, assessment of NF-E2, NGFR, and MVD may provide a novel way to evaluate biopsies from patients with MPNs. A cutoff level of more than 30% nuclear-stained NF-E2+ erythroblasts favors the diagnosis of ET or PV and can exclude PMF. Conversely, high levels of NGFR+ stromal cells and MVD are highly suggestive of PMF. When the morphologic, clinical, laboratory, and molecular findings are consistent with an early stage of MPN, but not able to point out a specific entity, NF-E2, NGFR, and MVD assessment by immunohistochemistry can be beneficial.
Acknowledgments We thank Joanne Israel and Felicia Chaviano for their excellent research assistance.
References [1] Vardiman JW, Bain BJ, Thiele J, Horny HP, Kvasnicka HM. Myeloproliferative neoplasms. In: Swerdlow SH, Campo E, Harris NL, et al, editors. WHO classification of tumors of haematopoietic and lymphoid tissues. 4th ed. Lyon, France: IARC Press; 2008. p. 32-65. [2] Thiele J. Philadelphia chromosome-negative chronic myeloproliferative disease. Am J Clin Pathol 2009;132:261-80. [3] Barbui T, Thiele J, Vannucchi AM, Tefferi A. Problems and pitfalls regarding WHO-defined diagnosis of early/prefibrotic primary myelofibrosis versus essential thrombocythemia. Leukemia 2013;27:1953-8. [4] Buhr T, Hebeda K, Kaloutsi V, Porwit A, Van der Walt J, Kreipe H. European Bone Marrow Working Group trial on reproducibility of World Health Organization criteria to discriminate essential thrombocythemia from prefibrotic primary myelofibrosis. Haematologica 2012;97:360-5.
N. Yigit et al. [5] Tefferi A. How I treat myelofibrosis. Blood 2011;117:3494-504. [6] Risum M, Madelung A, Bondo H, et al. The JAK2V617F allele burden and STAT3- and STAT5 phosphorylation in myeloproliferative neoplasms: early prefibrotic myelofibrosis compared with essential thrombocythemia, polycythemia vera and myelofibrosis. APMIS 2011; 119:498-504. [7] Ponce CC, Chauffaille Mde L, Ihara SS, Silva MR. MPL immunohistochemical expression as a novel marker for essential thrombocythemia and primary myelofibrosis differential diagnosis. Leuk Res 2012;36:93-7. [8] Goerttler PS, Kreutz C, Donauer J, et al. Gene expression profiling in polycythaemia vera: overexpression of transcription factor NF-E2. Br J Haematol 2005;129:138-50. [9] Aumann K, Frey AV, May AM, et al. Subcellular mislocalization of the transcription factor NF-E2 in erythroid cells discriminates prefibrotic primary myelofibrosis from essential thrombocythemia. Blood 2013;122:93-9. [10] Mesa RA, Hanson CA, Rajkumar V, Schroeder G, Tefferi A. Evaluation and clinical correlations of bone marrow angiogenesis in myelofibrosis with myeloid metaplasia. Blood 2000;96:3374-80. [11] Lundberg LG, Lerner R, Sundelin P, Rogers R, Folkman J, Palmblad J. Bone marrow in polycythemia vera, chronic myelocytic leukemia, and myelofibrosis has an increased vascularity. Am J Pathol 2000;157:15-9. [12] Boveri E, Passamonti F, Rumi E, et al. Bone marrow microvessel density in chronic myeloproliferative disorders: a study of 115 patients with clinicopathological and molecular correlations. Br J Haematol 2008;140:162-8. [13] Cattoretti G, Schiró R, Orazi A, Soligo D, Colombo MP. Bone marrow stroma in humans: anti-nerve growth factor receptor antibodies selectively stain reticular cells in vivo and in vitro. Blood 1993;81:1726-38. [14] Caneva L, Soligo D, Cattoretti G, De Harven E, Deliliers GL. Immunoelectron microscopy characterization of human bone marrow stromal cells with anti-NGFR antibodies. Blood Cells Mol Dis 1995;21:73-85. [15] O'Malley DP, Sen J, Juliar BE, Orazi A. Evaluation of stroma in human immunodeficiency virus/acquired immunodeficiency syndrome–affected bone marrows and correlation with CD4 counts. Arch Pathol Lab Med 2005;129:1137-40. [16] Boiocchi L, Orazi A, Ghanima W, Arabadjief M, Bussel JB, Geyer JT. Thrombopoietin receptor agonist therapy in primary immune thrombocytopenia is associated with bone marrow hypercellularity and mild reticulin fibrosis but no other stromal abnormalities. Mod Pathol 2012; 25:65-74. [17] Orazi A, Kahsai M, John K, Neiman RS. p53 overexpression in myeloid leukemic disorders is associated with increased apoptosis of hematopoietic marrow cells and ineffective hematopoiesis. Mod Pathol 1996;9:48-52. [18] Chatzitolios A, Venizelos I, Tripsiannis G, Anastassopoulos G, Papadopoulos N. Prognostic significance of CD95, P53, and BCL2 expression in extranodal non–Hodgkin's lymphoma. Ann Hematol 2010;89:889-96. [19] Swerdlow SH, Campo E, Harris NL, et al. WHO classification of tumors of haematopoietic and lymphoid tissues. 4th ed. Lyon, France: IARC Press; 2008. [20] Thiele J, Kvasnicka HM, Facchetti F, Franco V, van der Walt J, Orazi A. European consensus on grading bone marrow fibrosis and assessment of cellularity. Haematologica 2005;90:1128-32. [21] Barbui T, Thiele J, Carobbio A, et al. Masked polycythemia vera diagnosed according to WHO and BCSH classification. Am J Hematol 2014;89:199-202. [22] Silver RT, Chow W, Orazi A, Arles SP, Goldsmith SJ. Evaluation of WHO criteria for diagnosis of polycythemia vera: a prospective analysis. Blood 2013;122:1881-6. [23] Andrews NC. The NF-E2 transcription factor. Int J Biochem Cell Biol 1998;30:429-32. [24] Wang W, Schwemmers S, Hexner EO, Pahl HL. AML1 is overexpressed in patients with myeloproliferative neoplasms and mediates JAK2V617Findependent overexpression of NF-E2. Blood 2010;116:254-66.
NF-E2, NGFR, and CD34–MVD in MPNs [25] Bogeska R, Pahl HL. Elevated nuclear factor erythroid-2 levels promote epo-independent erythroid maturation and recapitulate the hematopoietic stem cell and common myeloid progenitor expansion observed in polycythemia vera patients. Stem Cells Transl Med 2013;2:112-7. [26] Mutschler M, Magin AS, Buerge M, et al. NF-E2 overexpression delays erythroid maturation and increases erythrocyte production. Br J Haematol 2009;146:203-17. [27] Kapralova K, Lanikova L, Lorenzo F, et al. RUNX1 and NF-E2 upregulation is not specific for MPNs, but is seen in polycythemic disorders with augmented HIF signaling. Blood 2014;123:391-4. [28] Westendorf JJ, Yamamoto CM, Lenny N, Downing JR, Selsted ME, Hiebert SW. The t(8;21) fusion product, AML-1-ETO, associates with C/EBP-alpha, inhibits C/EBP-alpha–dependent transcription, and blocks granulocytic differentiation. Mol Cell Biol 1998;18:322-33. [29] Alcalay M, Meani N, Gelmetti V, et al. Acute myeloid leukemia fusion proteins deregulate genes involved in stem cell maintenance and DNA repair. J Clin Invest 2003;112:1751-61.
9 [30] Kaufmann KB, Gründer A, Hadlich T, et al. A novel murine model of myeloproliferative disorders generated by overexpression of the transcription factor NF-E2. J Exp Med 2012;209:35-50. [31] Levin J, Peng JP, Baker GR, et al. Pathophysiology of thrombocytopenia and anemia in mice lacking transcription factor NF-E2. Blood 1999;94:3037-47. [32] Weiss L, Geduldig U. Barrier cells: stromal regulation of hematopoiesis and blood cell release in normal and stressed murine bone marrow. Blood 1991;78:975-90. [33] Schneider RK, Leisten I, Ziegler S, et al. Sustained alterations in bone marrow stromal cells from patients with myeloproliferative neoplasms (MPN) contribute to remodelling of the bone marrow microenvironment prior to the manifestation of myelofibrosis. Blood 2013;122:4102-4102. [34] Pugh CW, Ratcliffe PJ. Regulation of angiogenesis by hypoxia: role of the HIF system. Nat Med 2003;9:677-84. [35] Panteli K, Zagorianakou N, Bai M, Katsaraki A, Agnantis NJ, Bourantas K. Angiogenesis in chronic myeloproliferative diseases detected by CD34 expression. Eur J Haematol 2004;72:410-5.
www.elsevier.com/locate/humpath
Original contribution
Nuclear factor-erythroid 2, nerve growth factor receptor, and CD34–microvessel density are differentially expressed in primary myelofibrosis, polycythemia vera, and essential thrombocythemia☆,☆☆,★ Nuri Yigit MD a,b,⁎,1 , Shannon Covey MD a , Sharon Barouk-Fox MA a , Turker Turker MD b , Julia Turbiner Geyer MD a , Attilio Orazi MD a a
Department of Pathology and Laboratory Medicine, Weill Cornell Medical College/New York–Presbyterian Hospital, New York, NY 10065 b Department of Pathology, Gulhane Military Medical Academy and School of Medicine, Ankara 06010, Turkey Received 24 February 2015; revised 5 May 2015; accepted 7 May 2015
Keywords: Myeloproliferative neoplasms; NF-E2; NGFR; Microvessel density; Marrow stromal cells
Summary Because of the presence of various overlapping findings, the discrimination of polycythemia vera (PV) from prefibrotic/fibrotic primary myelofibrosis (PF/F-PMF) and essential thrombocythemia (ET) may be challenging, particularly in suboptimal bone marrow biopsy specimens. In this study, we assessed whether differences in the expression of nuclear factor-erythroid 2 (NF-E2), nerve growth factor receptor (NGFR; CD271), CD34, CD68, p53, CD3, CD20, and CD138 by immunohistochemistry could be useful in separating among them. Higher frequencies of nuclear positive erythroblasts with NF-E2 were observed in ET and PV cases (50% ± 13.3% and 41.5% ± 9.4%, respectively) when compared with both PF-PMF (21% ± 11.7%) and F-PMF (28.5% ± 10.8%). We found that with a cutoff level of at least 30% nuclear staining for NF-E2 in erythroblasts, we could reliably exclude the possibility of PMF. Conversely, NGFR+ stromal cells per highpower field (HPF) was significantly increased in F-PMF (53.5 ± 19.1/HPF) and PF-PMF (13.5 ± 3.8/HPF) compared with ET (4.4 ± 2.2/HPF) and PV (6.6 ± 3.3/HPF). Similarly, differences in CD34–microvessel density was remarkable in F-PMF and PF-PMF cases in comparison with PV and ET (49.9 ± 12.1/HPF, 29.3 ± 12.4/HPF, 13.7 ± 4.6/HPF, and 11.9 ± 5.1/HPF, respectively). Thus, the assessment of NF-E2 and NGFR expression and the evaluation of CD34–microvessel density may provide additional support in reaching a correct diagnosis in these cases of myeloproliferative neoplasms. © 2015 Elsevier Inc. All rights reserved.
☆
Disclosures: The authors declare no conflict of interest and no relevant financial or nonfinancial relationship. Abstract of this material was accepted as a poster presentation at United States and Canadian Academy of Pathology 2015 annual meeting in Boston, MA, USA. ★ Conflict of interest statement: Authors declare no conflict of interest. ⁎ Corresponding author at: Department of Pathology and Laboratory Medicine, Weill Cornell Medical College/New York–Presbyterian Hospital, 525 East 68th St, Starr Pavilion, 715 New York, NY 10065. E-mail addresses: [email protected], [email protected] (N. Yigit), [email protected] (S. Covey), [email protected] (S. Barouk-Fox), [email protected] (T. Turker), [email protected] (J. T. Geyer), [email protected] (A. Orazi). 1 Permanent address: Department of Pathology, Gulhane Military Medical Academy and School of Medicine, Etlik Kecioren, Ankara 06010, Turkey. ☆☆
http://dx.doi.org/10.1016/j.humpath.2015.05.004 0046-8177/© 2015 Elsevier Inc. All rights reserved.
2
1. Introduction The diagnosis of polycythemia vera (PV), prefibrotic/fibrotic primary myelofibrosis (PF/F-PMF), and essential thrombocythemia (ET), the main subtypes of Ph′ chromosome– negative/BCR-ABL1–negative myeloproliferative neoplasms (MPNs), is currently based on the combination of histologic, clinical, and laboratory parameters [1]. Selected clinical (eg, presence/absence of splenomegaly), laboratory (lactate dehydrogenase [LDH] and erythropoietin [EPO] levels), and a number of histologic features of the bone marrow (BM) such as size, pleomorphism, nuclear complexity and nuclear shape of megakaryocytes, status of iron storage in marrow, and certain peripheral blood findings like leukoerythroblastosis may help in distinguishing among them [2,3]. However, especially in the early stages of disease and/or in the presence of morphologically suboptimal BM specimens, their differential diagnosis is frequently difficult. This is largely due to their partially overlapping morphologic features (eg, similar degree of marrow cellularity and megakaryocytic proliferation) and molecular findings, and/or the lack of adequate clinical information. In such cases, a diagnosis of MPN, unclassifiable, is often rendered [4]. A precise diagnosis of MPN subtype is crucial because of the distinct clinical courses and outcomes of these entities [5]. This ambiguity has stimulated efforts to search for more reliable methods for confirming a specific diagnosis. Among these efforts, differences in STAT5 phosphorylation levels or in the expression frequency of thrombopoietin receptor in megakaryocytes have been proposed as useful diagnostic tools [6,7]. Goerttler et al [8] demonstrated that overexpression of nuclear factor-erythroid 2 (NF-E2), a transcription factor that has a regulating function on erythroid and megakaryocytic maturation, is a typical finding in PV and that NFE2 gene alterations are able to cause erythrocytosis and thrombocytosis in a murine model. More recently, a study from the same group described potential diagnostic usefulness of immunostaining for NF-E2 in separating subtypes of MPN using BM biopsies [9]. In addition to investigating the characteristics of the neoplastic cells potentially useful to separate these diseases, efforts have also been focused on differences in stromal components collectively known as BM microenvironment such as the histologic assessment of neoangiogenesis by microvessel density (MVD) counting. MVD had been previously investigated in MPN cases and found to be consistently higher in PMF cases than in other subtypes [10]. However, variable results in ET and PV have been obtained in other studies [11,12]. The expression of nerve growth factor receptor (NGFR), also known as CD271, has been proposed as a useful marker to identify BM reticular stromal cells in BM biopsies [13–15]. The NGFR-positive cells have a close anatomical interaction with nearly all types of hematopoietic cellular elements and play an essential role in regulation of hematopoiesis. NGFR
N. Yigit et al. expression has been recently correlated with other BM features including the degree of fibrosis in patients with primary immune thrombocytopenia, treated with a recombinant thrombopoietin agent [16]. p53 overexpression is a known marker of aggressiveness both in myeloid and in lymphoid neoplasm, but there is no information in relation to a possible differences in the degree of positivity in subtypes of MPN [17,18]. Similarly, there is only limited information in relation to possible differences in the quantity of “inflammatory cells” such as reactive B and T lymphocytes, histiocytes, and plasma cells found in the cellular background of these MPN subtypes. In this study, we investigated NF-E2 expression in erythroblasts, p53 overexpression in BM cells, NGFR expression, and degree of MVD in marrow stroma, as well as the frequency of lymphocytes, plasma cells, and histiocytes in the reactive cellular background.
2. Materials and methods 2.1. Case selection We retrospectively selected 50 cases in total, including 10 cases of PV, PF-PMF, F-PMF, ET, and control groups each, among the files of the Department of Pathology and Laboratory Medicine of Weill Cornell Medical College/ New York–Presbyterian Hospital in New York City. Clinical, radiologic, and laboratory data including LDH levels, complete blood counts, presence of splenomegaly, and conventional/molecular cytogenetics from all patients have also been retrieved from the hospital information system. All biopsies were the initial diagnostic samples in the chronic phase at the disease outset without any history of treatment. The control group was composed of staging BM biopsies, all of which were uninvolved. BM core biopsy and aspirate samples were routinely obtained from the iliac spine. All were fixed in Bouin solution, decalcified by using a hydrochloric acid/ethylenediaminetetraacetic acid (EDTA) combination, processed routinely, and embedded in paraffin. Biopsies were stained with hematoxylin and eosin, Gomori silver impregnation, and Masson trichrome, whereas BM aspirates and peripheral blood smears were Giemsa stained. We also had clot biopsies of 10 ET, 4 PF-PMF, 5 F-PMF, and 4 PV cases, all of which were fixed in formalin, along with the core biopsies that were processed routinely. The cases diagnosed before 2008 were reevaluated and reclassified according to the 2008 World Health Organization classification criteria [19]. The degree of fibrosis was graded on a scale of MF-0 to MF-3 in accordance with the European Consensus Grading System [20]. The design of this study was approved by the institutional review board of Weill Cornell Medical College.
NF-E2, NGFR, and CD34–MVD in MPNs
3
2.2. Immunohistochemistry BM core biopsies were immunostained for NF-E2, NGFR, CD34, CD68, p53, CD3, CD20, and CD138 markers by using an autostainer (Bond-III; Leica Microsystems, Buffalo Grove, IL) following the manufacturer's protocol. We also performed double staining with NF-E2 and CD71 in selected cases of each group to verify whether NF-E2 expression was restricted to erythroid lineage or not. Appropriate staining was confirmed using external positive and negative controls for each antibody. The characteristics of the antibodies are summarized in Table 1.
2.3. Immunohistochemical evaluation All the stained slides have independently been evaluated by 3 blinded reviewers (N. Y., J. T. G., and A. O.). NF-E2 nuclear-stained erythroblasts were assessed as the percentage of the total erythroblasts, whereas the number of CD3, CD20, and CD138+ cells was calculated as the percentage of the total marrow cellularity. The mean values of NGFR, p53, and CD68+ cells were counted per high-power field (HPF; 0.238 mm2: 22-mm ocular lens and ×400 total magnification). For NGFR and CD68, only the highlighted bodies of the cells with a visible nucleus were counted, excluding the enlarged cytoplasmic dendritic structures. To assess MVD, the mean number of thin-walled vessels with or without a lumen and sinusoids visualized by CD34+ endothelial cells was counted per HPF. Arterioles and myeloid/megakaryocytic precursors, which are normally costained with CD34, were excluded during the evaluation. Also, the consecutively or discontinuously appearing branches of the same vessel or tortuous sinusoid on the section plane were ignored in order to avoid the bias of repeated counting. For all cases, the areas of highest vascularization (hot spots), which were appreciated visually, were selected for quantifying BM vascularity.
2.4. Statistical analysis All groups were tested using the nonparametric Kruskal-Wallis test to compare continuous variables among Table 1
groups, with Bonferroni correction. A P value less than .05 was considered statistically significant. Receiver operator characteristic analysis and the value of area under the curve were constructed to assign a cutoff level for staining percentages, which can be used in discriminating each diagnostic category from others. All analyses were performed using SPSS for Windows version 17.0 statistical software package (SPSS, Chicago, IL).
3. Results All patients were in the chronic phase of their diseases. The mean age of the groups was 60 years (range, 18-91 years) with a male-to-female ratio of 0.81. Cytogenetic analysis demonstrated a normal karyotype in all ET cases; 2 del20q, 1 del9q, and 1 trisomy 8 in PF-PMF; 3 del20q, 1 delXq, and 1 monosomy 18 in F-PMF; and 1 trisomy 9 in PV cases, respectively. Fibrosis scores, JAK2 status, and immunohistochemical staining characteristics are summarized in Table 2. Calculated mean and median values of the stains in all groups were correlated with minor differences. Statistically significant values were identified for NF-E2, NGFR, and MVD (Fig. 1). In core biopsies of all cases, NF-E2 positivity was seen predominantly in a nuclear localization for early stage erythroblasts and cytoplasmic in late stage. All erythrocytes were strongly positive. ET and PV cases had higher frequencies of nuclear positive erythroblasts (50% ± 13.3% and 41.5% ± 9.4%, respectively) when compared with both PF-PMF (21% ± 11.7%) and F-PMF (28.5% ± 10.8%; P b .001; Fig. 2). We have confirmed the specificity of NF-E2 to erythroid lineage by using double immunohistochemical staining, which was composed of NF-E2 plus CD71. We have observed a consistent staining pattern with NF-E2, overlapping to the single-stained cores (Fig. 3A-C). In the receiver operator characteristic analysis for establishing a cutoff value for differentiating ET and PV from PF/F-PMF, the calculated area under the curve value was found to be 0.94 (95% confidence interval, 0.853-1.027). Thus, when 30% of staining was accepted as the cutoff level, the sensitivity and specificity
Characteristics of the antibodies
Marker
Company
Clone
Host
Dilution
Antigen retrieval method and time (min)
NF-E2, St. Louis, MO NGFR, Cambridge, MA CD34, Buffalo Grove, IL CD68, Carpinteria, CA p53, Fremont, CA CD3, Buffalo Grove, IL CD20, Carpinteria, CA CD138, Raleigh, NC CD71, Buffalo Grove, IL
Sigma Abcam Leica Dako Biogenex Leica Dako AbD Serotec Leica
Polyclonal ME20.4 QBEnd/10 PG-M1 1801 PS1 L26 B-A38 10F11
Rabbit Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse
1:50 1:1600 1:100 1:300 1:200 1:100 1:200 1:25 1:100
H2 (20) E1 (10) H2 (20) H1 (30) H2 (20) H2 (30) H1 (30) H1 (30) H1 (30)
Abbreviations: H1, Bond Epitope Retrieval Solution 1: a citrate-based pH 6.0 solution; E1, Bond Enzyme Pretreatment Kit: 1 drop Bond enzyme concentrate (proteolytic enzyme) with 7 mL of Bond enzyme diluent (Tris-buffered saline); H2, Bond Epitope Retrieval Solution 2: an EDTA-based pH 9.0 solution.
5:5 7:3
3:7
PF-PMF (n = 10) 61 (26-91) F-PMF (n = 10) 72 (29-90)
ET (n = 10)
6/10 55 49 (18-68)
5/10 4/10 78 84
Abbreviations: PV, polycythemia vera; PF-PMF, prefibrotic stage of primary myelofibrosis; F-PMF, fibrotic stage of primary myelofibrosis; ET, essential thrombocythemia; NF-E2, nuclear factor-erythroid 2; NGFR, nerve growth factor receptor; HPF, high-power field. a The value was statistically significant in comparison with PF/F-PMF cases (P b .001). b The difference was statistically significant compared with PV and ET cases (P b .001). c Statistically significant relationship was observed between ET and PV cases (P = .021).
13.6 ± 3.8 c 5.8 ± 3.4 5.5 ± 2.5 0 24.4 ± 8.5 11.9 ± 5.1 4.4 ± 2.2
3.1 ± 2.3 5.7 ± 2.2 3.4 ± 2.3 4.4 ± 2.4 9±4 9.1 ± 3.3 0 0 13.5 ± 3.8 b 29.3 ± 12.4 b 21.8 ± 5.6 53.5 ± 19.1 b 49.9 ± 12.1 b 18.9 ± 10.5
3.8 ± 1.3 8.8 ± 2.6 0 22.7 ± 5.9 13.7 ± 4.6 6.6 ± 3.3
= 8) 41.5 ± 9.4 a = 2) = 10) 21 ± 11.7 = 6) 28.5 ± 10.8 = 4) = 10) 50 ± 13.3 a 83
10/10
MF-0 MF-1 MF-1 MF-2 MF-3 MF-0 3:7 58 (44-78) PV (n = 10)
(n (n (n (n (n (n
CD20 (%) CD68 TP53 CD3 (%) (PG-M1; /HPF) (/HPF) CD34+ microvessels (/HPF) NF-E2 (nuclear NGFR staining in (/HPF) erythroblasts; %) Mean age Sex Mean JAK2+ Fibrosis and range (y) (male/ cellularity cases female) (%) Cases
Cases' characteristics and staining profiles of BM core biopsies with mean values Table 2
5 ± 1.9
N. Yigit et al. CD138 (%)
4
were calculated as 85% and 90%, respectively. In the ET and PMF groups, NF-E2 overexpression was not correlated with the status of JAK2 mutation or allele burden, which also appeared unrelated for PV cases. Only rare myeloid precursors and a variable proportion of megakaryocytes in the sections were highlighted as seen in one of previously published study [9]. The megakaryocytes had the similar staining patterns among all the groups. In addition, plasma cells had marked perinuclear localization of NF-E2 with a granular quality in all groups. The number of NGFR+ stromal cells was significantly increased in F-PMF (53.5 ± 19.1/HPF) and PF-PMF (13.5 ± 3.8/HPF) when compared with ET and PV (P b .001; Fig. 4A and B). Also, a statistically meaningful difference was observed between F-PMF and PF-PMF (P b .001). The dendritic structures of stromal cells were thicker and more continuous in all PMF cases. In addition, conspicuous peritrabecular osteoblastic rimming highlighted by NGFR was noted around the newly developing osteosclerotic foci. Two PV cases having MF-1 showed NGFR density similar to that of PF-PMF. MVD was assessed in F-PMF, PF-PMF, PV, and ET cases, at 49.9 ± 12.1/HPF, 29.3 ± 12.4/HPF, 13.7 ± 4.6/HPF, and 11.9 ± 5.1/HPF, respectively. Statistically, increased MVD was detected in PF/F-PMF cases in comparison with PV and ET (P b .001; Fig. 5A and B). Four of 10 F-PMF cases showed microvascular hot spots as well as apparently hypovascular regions in some intertrabecular spaces. Although we have also observed groups of T cells highlighted by CD3 in PMF cases, a statistically meaningful difference of its expression has been detected only with ET and PVs (more T cells in ET cases than PVs; P = .021). Variable amounts of CD20, CD138, and CD68+ inflammatory cells were observed in different MPN cases, and they did not show any statistically significant differences among disease groups. None of the biopsies were positive with p53 (all b1%). Stromal and inflammatory cells were generally dispersed within the marrow, except with some lymphoid aggregates in a few cases. Four F-PMF cases had newly vascularized hot spots, accompanied by increased numbers of NGFR+, CD3+, and CD20+ cells, as well as decreased CD68+ histiocytes. The remaining vessel-poor areas on the same biopsy sections showed opposite immunoexpression profiles for these markers when compared with the newly vascularized hot spots. We have found the lowest levels of expression of NF-E2, NGFR, and CD68 in the control groups, when compared with PV, PMF, and ET. Likewise, MVD was much lower in control cases. (NF-E2 was positive in 12% ± 4.8% of erythroblasts; NGFR+ stromal cells, 3.8 ± 0.8/HPF; CD68+ cells, 7.2 ± 1.4/HPF; and MVD, 4.6 ± 1/HPF.) In clot biopsies, staining profiles of all antibodies demonstrated quite similar characteristics to the cores. Unfortunately, statistical analysis could not be performed because of the limited number of clot samples.
NF-E2, NGFR, and CD34–MVD in MPNs
Fig. 1
5
Median values and ranges of staining for NF-E2, NGFR, and MVD in all groups.
Fig. 2 A and B, BM morphologic features with hematoxylin and eosin stain and nuclear NF-E2 staining in most of the erythroblasts of ET. C and D, NF-E2 similarly marks many erythroblasts of PV. E and F, Prefibrotic primary myelofibrosis showing decreased numbers of nuclear stained erythroblasts compared with ET and PV. A, C and E, hematoxylin and eosin stain, original magnification ×200; B, D and F immunoperoxidase, ×400.
6
N. Yigit et al.
Fig. 3 A, By double immunohistochemical staining with NF-E2 (brown chromogen) and CD71 (red chromogen), normal BM biopsy shows that NF-E2 is mainly restricted to the erythroblasts (orange arrow), although some are not marked (violet arrow). Only rare myeloid precursors are positive for NF-E2 (black arrow). B, Many erythroblasts of PV showing NF-E2 positivity. Megakaryocytes are also nuclearly positive for NF-E2. C, In prefibrotic primary myelofibrosis, the NF-E2 labels significantly less erythroblasts than in ET. A-C, immunoperoxidase, original magnification ×1000.
4. Discussion Separating PV from ET or PMF particularly in patients seen early in the course of disease can be challenging due to their partially overlapping features. In some of these patients, clinical findings and laboratory results may be inconclusive. The problem is more typically seen in cases of early phase PV and in PMF, both of which may be confused with ET. In early PV, the diagnostic cutoff for establishing erythrocytosis is often not met initially, but only later in the course of the disease. These cases have been recently referred to as “masked PV” [21,22]. In these cases, only BM histology and red cell mass determination allow for separation from ET [22]. Similarly, PMF patients can develop splenomegaly, a minor criterion for diagnosing PMF, only months to years after their initial presentation. Elevated LDH levels, another minor parameter for the diagnosis of PMF, cannot be detected or might not be available in some institutions. Although differences in the frequency of JAK2, MPL, or CALR gene mutations can be useful in separating among these subtypes of MPN, most patients share in common JAK2 V617F. These issues underline the importance of
finding new reliable tools that can be used to separate PV and PMF from ET. NFE2 gene encodes NF-E2, which is a hematopoietic transcription factor and binds conserved functional elements present in globins and many other erythroid/megakaryocytic genes [23]. NF-E2 displays an essential role for regulating erythroid and megakaryocytic maturation. Its overexpression was detected in most patients with PV, ET, and PMF [8,24]. Bogeska and Pahl [25] investigated NF-E2 role on pathogenesis of PV and found that elevated NF-E2 expression caused aberrant expansion of the stem and progenitor cell compartments. Conversely, decreased NF-E2 levels resulted in diminished numbers of hematopoietic stem cells as well as multipotent, common myeloid, and granulocyte-monocyte progenitors, but megakaryocyte-erythroid progenitors were nearly not affected. Hence, they concluded that elevation of NF-E2 expression is required and responsible for the EPO-independent erythroid maturation, which is a pathognomonic hallmark of PV. Mutschler et al [26] stated that NF-E2 overexpression delays the early phase of erythroid maturation resulting in an expansion of erythroid progenitors and overproduction of erythrocytes derived from each progenitor cell.
Fig. 4 A, Extensive NGFR+ stromal cells in prefibrotic primary myelofibrosis (immunoperoxidase, original magnification ×200). B, Rare and dispersed NGFR+ stromal cells in ET (immunoperoxidase, ×400).
NF-E2, NGFR, and CD34–MVD in MPNs
7
Fig. 5 Comparison of MVD between prefibrotic primary myelofibrosis cases showing intensely stained microvessels as well as enlarged sinusoids (A) and ET with sparse microvessels (B). Immunoperoxidase, original magnification both ×200).
In addition, Kapralova et al [27] found elevated NFE2 transcripts in different primary and secondary polycythemias, which can be caused both by neoplastic and by nonneoplastic disorders. Therefore, they concluded that NF-E2 overexpression is not specific to MPNs. Interestingly, some studies also showed that the leukemogenic fusion proteins can repress the production of transcription factors involved in control of lineage differentiation, which results in tumorigenesis [28]. In this manner, expression level of NF-E2 was detected to be repressed by AML/ETO and PML/RAR fusion proteins, which was considered as a crucial step in leukemogenesis [29]. Kaufmann et al [30] generated mice overexpressing NF-E2 and observed many features of MPN, including thrombocytosis, leukocytosis, EPO-independent colony formation, and characteristic BM histology. Conversely, loss of NF-E2 was shown to result in anemia and severe thrombocytopenia [31]. In accordance with these studies, we observed immunohistochemical NF-E2 overexpression in erythroblasts in all 3 MPN entities. In addition, similar to most published studies, we did not find a correlation between NF-E2 overexpression and JAK2 status. The highest degree of nuclear overexpression (N30% of erythroblasts) was found exclusively in PV and ET, but not in PF/F-PMF. Aumann et al [9] investigated immunohistochemical expression levels of NF-E2 in MPNs. According to their findings, the highest nuclear staining was found in PMF (33.7% ± 10.7%) followed by PV (23.9% ± 10%) and ET (16.3% ± 4.1%), respectively. Consequently, they proposed using a cutoff level of 20% to discriminate PMF from ET. However, our findings failed to confirm this suggestion and the highest NF-E2 nuclear expression levels were seen in erythroblasts of ET patients in our series. This discrepancy may be caused by the usage of different fixative solutions. Our samples were all fixed in Bouin solution, whereas in a previous study we referenced, either formalin or calcium-glutaraldehydeformaldehyde was used [9]. Also, decalcifying agents may have some effect on the tissues which may result in such discordance on immunohistochemistry. We prefer a combina-
tion of hydrochloric acid and EDTA for decalcification. Instead, Aumann and colleagues used a mixture of EDTA disodium salt and Tris(hydroxymethyl)aminomethane for their cases. However, the results obtained using corresponding formalin-fixed clot sections, which were available in a proportion of our cases, showed comparable results raising the possibility that differences in the diagnostic criteria applied to these cases may have been responsible for the discrepancy. Previously published evidence showed that NGFR is a specific marker for BM stromal cells. These cells are components of the marrow microenvironment together with histiocytes, adipocytes, and endothelial cells and are believed to have a fibroblast-like function. They are hypothesized to have a regulatory role in the homing of stem cells, precursor cell differentiation, and releasing of mature cells to the circulation [32]. The NGFR-positive cells are characterized by oval-shaped nuclei, scant cytoplasm, and numerous elongated dendritic-like processes that surround hematopoietic cells, adipocytes, vessels, and sinusoids [14]. A recent study claimed that stromal cells lose their ability to support hematopoiesis and increase their matrix remodeling potential in MPN cases, which contributes to the development of myelofibrosis [33]. Our results showed a correlated increase in stromal cells, fibrosis, and vessels in PMF cases. These parameters were particularly high in F-PMF. It is also assumed that the dendritic structures of stromal cells cover the endosteal surfaces of the bone trabeculae [13]. Similar to these observations, NGFR in our study showed thick dendritic layering around the newly developing immature osteosclerotic foci. Although the thick, continuous, and bulbous meshwork of stromal cell projections was observed to intermingle with various marrow cells in PMF cases, ET showed mostly thin and interrupted projections of stromal cells, similar to the control group. According to the literature, the highest MVD was seen in PMF cases in accordance with the worst degree of fibrosis in MPN cases. Anemia, one of the minor criteria for the diagnosis of PMF, was hypothesized to be an agent to activate BM
8 neovascularization as a response against tissue hypoxia [34]. In the previous studies, high MVD in MPNs was found to be well correlated with hypercatabolic symptoms, vascular endothelial growth factor expression, and marked splenomegaly [10,11]. There was no significant correlation between MVD and the severity of anemia or degree of cellularity in our series, perhaps due to the limited number of studied cases [35]. We did not detect a correlation between MVD and degree of cellularity, anemia, LDH level, or splenomegaly among PMF cases, as in some previously reported studies [12]. Although CD3+ T cells were present in higher amounts in ET than PV, the difference was not significant when compared with PMF cases. On the other hand, benign inflammatory cells can be regulated by various disease-specific or nonspecific mechanisms during the course of disease due to alterations of mature blood cell counts. Thus, T-cell counting may be useful for discriminating ET from PV, when other required criteria fit the diagnosis of ET.
5. Conclusions To summarize, assessment of NF-E2, NGFR, and MVD may provide a novel way to evaluate biopsies from patients with MPNs. A cutoff level of more than 30% nuclear-stained NF-E2+ erythroblasts favors the diagnosis of ET or PV and can exclude PMF. Conversely, high levels of NGFR+ stromal cells and MVD are highly suggestive of PMF. When the morphologic, clinical, laboratory, and molecular findings are consistent with an early stage of MPN, but not able to point out a specific entity, NF-E2, NGFR, and MVD assessment by immunohistochemistry can be beneficial.
Acknowledgments We thank Joanne Israel and Felicia Chaviano for their excellent research assistance.
References [1] Vardiman JW, Bain BJ, Thiele J, Horny HP, Kvasnicka HM. Myeloproliferative neoplasms. In: Swerdlow SH, Campo E, Harris NL, et al, editors. WHO classification of tumors of haematopoietic and lymphoid tissues. 4th ed. Lyon, France: IARC Press; 2008. p. 32-65. [2] Thiele J. Philadelphia chromosome-negative chronic myeloproliferative disease. Am J Clin Pathol 2009;132:261-80. [3] Barbui T, Thiele J, Vannucchi AM, Tefferi A. Problems and pitfalls regarding WHO-defined diagnosis of early/prefibrotic primary myelofibrosis versus essential thrombocythemia. Leukemia 2013;27:1953-8. [4] Buhr T, Hebeda K, Kaloutsi V, Porwit A, Van der Walt J, Kreipe H. European Bone Marrow Working Group trial on reproducibility of World Health Organization criteria to discriminate essential thrombocythemia from prefibrotic primary myelofibrosis. Haematologica 2012;97:360-5.
N. Yigit et al. [5] Tefferi A. How I treat myelofibrosis. Blood 2011;117:3494-504. [6] Risum M, Madelung A, Bondo H, et al. The JAK2V617F allele burden and STAT3- and STAT5 phosphorylation in myeloproliferative neoplasms: early prefibrotic myelofibrosis compared with essential thrombocythemia, polycythemia vera and myelofibrosis. APMIS 2011; 119:498-504. [7] Ponce CC, Chauffaille Mde L, Ihara SS, Silva MR. MPL immunohistochemical expression as a novel marker for essential thrombocythemia and primary myelofibrosis differential diagnosis. Leuk Res 2012;36:93-7. [8] Goerttler PS, Kreutz C, Donauer J, et al. Gene expression profiling in polycythaemia vera: overexpression of transcription factor NF-E2. Br J Haematol 2005;129:138-50. [9] Aumann K, Frey AV, May AM, et al. Subcellular mislocalization of the transcription factor NF-E2 in erythroid cells discriminates prefibrotic primary myelofibrosis from essential thrombocythemia. Blood 2013;122:93-9. [10] Mesa RA, Hanson CA, Rajkumar V, Schroeder G, Tefferi A. Evaluation and clinical correlations of bone marrow angiogenesis in myelofibrosis with myeloid metaplasia. Blood 2000;96:3374-80. [11] Lundberg LG, Lerner R, Sundelin P, Rogers R, Folkman J, Palmblad J. Bone marrow in polycythemia vera, chronic myelocytic leukemia, and myelofibrosis has an increased vascularity. Am J Pathol 2000;157:15-9. [12] Boveri E, Passamonti F, Rumi E, et al. Bone marrow microvessel density in chronic myeloproliferative disorders: a study of 115 patients with clinicopathological and molecular correlations. Br J Haematol 2008;140:162-8. [13] Cattoretti G, Schiró R, Orazi A, Soligo D, Colombo MP. Bone marrow stroma in humans: anti-nerve growth factor receptor antibodies selectively stain reticular cells in vivo and in vitro. Blood 1993;81:1726-38. [14] Caneva L, Soligo D, Cattoretti G, De Harven E, Deliliers GL. Immunoelectron microscopy characterization of human bone marrow stromal cells with anti-NGFR antibodies. Blood Cells Mol Dis 1995;21:73-85. [15] O'Malley DP, Sen J, Juliar BE, Orazi A. Evaluation of stroma in human immunodeficiency virus/acquired immunodeficiency syndrome–affected bone marrows and correlation with CD4 counts. Arch Pathol Lab Med 2005;129:1137-40. [16] Boiocchi L, Orazi A, Ghanima W, Arabadjief M, Bussel JB, Geyer JT. Thrombopoietin receptor agonist therapy in primary immune thrombocytopenia is associated with bone marrow hypercellularity and mild reticulin fibrosis but no other stromal abnormalities. Mod Pathol 2012; 25:65-74. [17] Orazi A, Kahsai M, John K, Neiman RS. p53 overexpression in myeloid leukemic disorders is associated with increased apoptosis of hematopoietic marrow cells and ineffective hematopoiesis. Mod Pathol 1996;9:48-52. [18] Chatzitolios A, Venizelos I, Tripsiannis G, Anastassopoulos G, Papadopoulos N. Prognostic significance of CD95, P53, and BCL2 expression in extranodal non–Hodgkin's lymphoma. Ann Hematol 2010;89:889-96. [19] Swerdlow SH, Campo E, Harris NL, et al. WHO classification of tumors of haematopoietic and lymphoid tissues. 4th ed. Lyon, France: IARC Press; 2008. [20] Thiele J, Kvasnicka HM, Facchetti F, Franco V, van der Walt J, Orazi A. European consensus on grading bone marrow fibrosis and assessment of cellularity. Haematologica 2005;90:1128-32. [21] Barbui T, Thiele J, Carobbio A, et al. Masked polycythemia vera diagnosed according to WHO and BCSH classification. Am J Hematol 2014;89:199-202. [22] Silver RT, Chow W, Orazi A, Arles SP, Goldsmith SJ. Evaluation of WHO criteria for diagnosis of polycythemia vera: a prospective analysis. Blood 2013;122:1881-6. [23] Andrews NC. The NF-E2 transcription factor. Int J Biochem Cell Biol 1998;30:429-32. [24] Wang W, Schwemmers S, Hexner EO, Pahl HL. AML1 is overexpressed in patients with myeloproliferative neoplasms and mediates JAK2V617Findependent overexpression of NF-E2. Blood 2010;116:254-66.
NF-E2, NGFR, and CD34–MVD in MPNs [25] Bogeska R, Pahl HL. Elevated nuclear factor erythroid-2 levels promote epo-independent erythroid maturation and recapitulate the hematopoietic stem cell and common myeloid progenitor expansion observed in polycythemia vera patients. Stem Cells Transl Med 2013;2:112-7. [26] Mutschler M, Magin AS, Buerge M, et al. NF-E2 overexpression delays erythroid maturation and increases erythrocyte production. Br J Haematol 2009;146:203-17. [27] Kapralova K, Lanikova L, Lorenzo F, et al. RUNX1 and NF-E2 upregulation is not specific for MPNs, but is seen in polycythemic disorders with augmented HIF signaling. Blood 2014;123:391-4. [28] Westendorf JJ, Yamamoto CM, Lenny N, Downing JR, Selsted ME, Hiebert SW. The t(8;21) fusion product, AML-1-ETO, associates with C/EBP-alpha, inhibits C/EBP-alpha–dependent transcription, and blocks granulocytic differentiation. Mol Cell Biol 1998;18:322-33. [29] Alcalay M, Meani N, Gelmetti V, et al. Acute myeloid leukemia fusion proteins deregulate genes involved in stem cell maintenance and DNA repair. J Clin Invest 2003;112:1751-61.
9 [30] Kaufmann KB, Gründer A, Hadlich T, et al. A novel murine model of myeloproliferative disorders generated by overexpression of the transcription factor NF-E2. J Exp Med 2012;209:35-50. [31] Levin J, Peng JP, Baker GR, et al. Pathophysiology of thrombocytopenia and anemia in mice lacking transcription factor NF-E2. Blood 1999;94:3037-47. [32] Weiss L, Geduldig U. Barrier cells: stromal regulation of hematopoiesis and blood cell release in normal and stressed murine bone marrow. Blood 1991;78:975-90. [33] Schneider RK, Leisten I, Ziegler S, et al. Sustained alterations in bone marrow stromal cells from patients with myeloproliferative neoplasms (MPN) contribute to remodelling of the bone marrow microenvironment prior to the manifestation of myelofibrosis. Blood 2013;122:4102-4102. [34] Pugh CW, Ratcliffe PJ. Regulation of angiogenesis by hypoxia: role of the HIF system. Nat Med 2003;9:677-84. [35] Panteli K, Zagorianakou N, Bai M, Katsaraki A, Agnantis NJ, Bourantas K. Angiogenesis in chronic myeloproliferative diseases detected by CD34 expression. Eur J Haematol 2004;72:410-5.
