Main

Extramedullary hematopoiesis (EMH) in the spleen is a characteristic feature of the chronic myeloproliferative disorders (CMPD). EMH may also occur less frequently in patients with reactive conditions or other neoplastic disorders such as myelodysplastic syndrome (MDS) or chronic myelomonocytic leukemia (CMML). The normal adult spleen virtually always contains a very small number of mature hematopoietic elements, but the origin of the precursor cells giving rise to EMH and the molecular mechanisms underlying their development are uncertain. It has been hypothesized that EMH cells are derived from either circulating neoplastic stem cells, which undergo terminal differentiation within the spleen as part of the neoplastic process or reactivated embryonic remnants of fetal hematopoiesis as compensation for marrow failure.1, 2, 3, 4, 5 The former concept is more strongly favored, as hematopoietic elements in EMH have histomorphological and molecular findings similar to those of the corresponding marrow biopsies.1, 2, 3, 4, 5 However, the clonal relationship between splenic EMH cells and neoplastic bone marrow cells remains unconfirmed.

Recently, the V617F mutation in the Janus kinase 2 gene (JAK2V617F) was shown to be frequently and preferentially identified in bone marrow and peripheral blood cells of CMPD patients. The JAK2V617F allele, present in all hematopoietic cells arising from the neoplastic clone, has been detected in the majority of cases of polycythemia vera (PV), essential thrombocythemia (ET), chronic idiopathic myelofibrosis (CIMF) and acute leukemia transformed from a preexisting CMPD, but it is rarely identified in healthy controls or patients with other myeloid disorders.6, 7, 8, 9, 10, 11, 12, 13 Functioning as a constitutively activated cytoplasmic tyrosine kinase, the mutated JAK2V617F protein probably mediates myeoloproliferation via dysregulation of various JAK2/STAT5 signaling pathways. Targeted genes, including the antiapoptotic protein Bcl-xL, have been implicated in vital cellular processes such as proliferation and survival.13, 14, 15, 16, 17 Furthermore, retroviral transduction of JAK2V617F into murine hematopoietic stem cells was recently shown to result in the development of a polycythemia-like phenotype.7, 18

The role of JAK2V617F in splenic EMH has not been previously studied. We used laser-capture microdissection (LCM), real-time PCR melting curve analysis and immunohistochemistry to characterize the hematopoietic precursor cells found in spleen and bone marrow specimens of patients with significant EMH. These results demonstrate that the JAK2V617F mutation is frequently present in splenic EMH cells associated with CMPD. Thus, the precursor cells that lead to extramedullary hematopoietic expansion most likely originate from the transformed bone marrow clone. In addition, the overexpression of Bcl-xL in JAK2V617F-positive megakaryocytes suggests that the JAK2/STAT5/Bcl-xL pathway may be important to the myeloproliferative process, at least partially, in splenic EMH.

Materials and methods

Patient Samples

Retrospective review of pathology and hematology files from our respective institutions identified 47 splenectomy specimens demonstrating significant EMH from patients previously diagnosed with CMPD, MDS, CMPD/MDS, chronic myelogenous leukemia (CML) and acute myeloid leukemia (AML). Biopsies were performed between 1993 and 2005, and patient ages ranged from 37 to 90 years with a median age of 64 years. The morphologic diagnosis of EMH was confirmed, and the pattern of EMH was classified as nodular (macronodular proliferation of EMH), diffuse (diffuse pattern of EMH with trilineage myeloid involvement), granulocyte predominant (immature granulocyte predominance) or scant (few hematopoietic cells present in no distinct pattern).19 Twenty cases had adequate bone marrow specimens available for correlative molecular studies; the time interval between splenectomy and bone marrow biopsy was less than 12 months for 17 cases and less than 33 months for the remaining three cases. None of the cases had additional biopsies from other anatomical sites that typically demonstrate EMH. Results of a pilot study evaluating JAK2 mutational status on a small subset of these cases using a different methodology and different criteria is also currently under review (Table 1). This study was approved by the Institutional Review Board.

Table 1 Molecular, histological and clinical results from patients with splenic extramedullary hematopoiesis (EMH)
Full size table

Laser-Capture Microdissection

LCM was performed on 10 unstained splenectomy sections using a PixCell II Laser Capture Microdissection system (Arcturus Engineering, Mountain View, CA, USA) to enrich cell fractions for either EMH or non-EMH cells.20, 21

DNA Extraction and Real-Time PCR Melting Curve Analysis for JAK2 Genotype

DNA was extracted from formalin-fixed paraffin-embedded spleen sections, bone marrow aspirates and LCM-enriched cell fractions as described previously.22 Real-time PCR melting curve analysis for the JAK2 wild-type and V617F mutant allele was performed as described previously.22 Homozygous mutant human erythroleukemia (HEL) and homozygous wild-type multiple myeloma (RPMI8226) cell lines were used as positive and negative controls, respectively. As previously reported, the analytical sensitivity of our real-time PCR melting curve assay is capable of detecting one homozygous V617F mutant cell in 20 total cells (5% lower limit of mutant allele frequency detection).22 Furthermore, the peak height (dF/dT vs T) of each JAK2 allele curve (JAK2V617F vs wild-type) proportionately increases or decreases with its relative frequency, providing a semiquantitative measurement (Figure 1). The relative proportion of wild-type and mutant JAK2 alleles was scored as either mutant only (M), mixed clonality (M&W) for cases with similar peak heights of both JAK2V617F and wild-type alleles, mutant predominant (M>W) for cases with a large peak of mutant alleles and a smaller peak of wild-type alleles, wild-type predominant (W>M) for cases with a large peak of wild-type alleles and a smaller peak of mutant alleles, or wild-type only (W).

Figure 1
figure 1

JAK2 real-time PCR melting curve. A representative JAK2 real-time PCR melting curve is shown. Serial dilution of a homozygous JAK2V617F human erythroid leukemia cell line (HEL) with a homozygous JAK2 wild-type multiple myeloma cell line (RPMI8226) demonstrates the semiquantitative measurement of allele frequency by comparison of relative peak heights (47°C for JAK2V617F compared with 55°C for wild-type). JAK2V617F to wild-type ratio (HEL:RPMI8226) is indicated by light blue=1:0, light green=1:1, red=1:4, orange=1:9, brown=1:19, dark green=1:39 and dark blue=0:1. The negative control (water only) is shown in gray. As demonstrated in the brown curve with a barely observable JAK2V617F peak, the lower limit of detection for this assay is one homozygous mutant cell in total 20 cells. The relative peak height of the JAK2V617F allele decreases as the ratio of mutant to wild-type cells decreases, providing a semiquantitative measurement of allele frequency.

Full size image

Bcl-xL Immunohistochemical Staining

Bcl-xL expression in splenectomy specimens was studied by immunohistochemical staining using standard methods.23, 24 Briefly, antigen retrieval was performed at 100°C in a steamer for 30 min (Tris, pH9 buffer), BCL-xL antibody (2H12, Zymed Laboratories Inc., San Francisco, CA, USA) was applied at 1:100, and DAB was used for signal detection. Bcl-xL expression was scored as percentage of positive megakaryocytes.

Results

The JAK2 V617F mutant allele was detected in 22 of the 47 splenectomy specimens, including 11/17 CIMF (65%), 7/7 PV (100%), 1/1 ET (100%), 1/3 CMPD unclassifiable (33%), 1/5 CMML (20%), 0/5 CML (0%), 1/3 MDS (33%) and 0/6 AML (0%) cases, whereas only the JAK2 wild-type allele was detected in the other 25 specimens (Table 1). Twenty cases had corresponding bone marrow biopsies available for JAK2 genotyping. Nineteen (95%) demonstrated a concordant genotype with the spleen, including eight showing both mutant and wild-type alleles and 12 showing only wild-type alleles (Table 1). The single case (case 11) with a discordant JAK2 genotype showed both mutant and wild-type alleles in the spleen but only wild-type alleles in the bone marrow. Of note, another case (case 6) showed a mutant allele predominance in the spleen and a wild-type allele predominance in the bone marrow.

To enhance the sensitivity of JAK2V617F detection and determine whether JAK2V617F alleles specifically originated from the splenic EMH cells, 10 cases were further analyzed by LCM of either EMH or non-EMH cells. JAK2 genotyping results from the EMH-enriched cell fractions were concordant with the whole-spleen studies, and they demonstrated that LCM significantly increases the mutant allele proportion (Table 1). For example, cases 14, 15, 19 and 26 demonstrated a significantly increased mutant allele proportion in the LCM EMH-enriched splenic cell fraction compared with the whole-spleen samples. In comparison, the non-EMH cell fractions showed enrichment for the wild-type allele. Only case 8 revealed similar amounts of mutant and wild-type JAK2 alleles in the EMH and non-EMH cell fractions. This finding suggests that the JAK2V617F alleles were predominantly possessed by the hematopoietic elements of splenic EMH rather than being derived from contaminating bone marrow hematopoietic cells in circulation.

The JAK2 genotype in splenic EMH cells was then correlated with other clinicopathologic parameters. Among all cases, the weight of spleens with JAK2V617F alleles tended to be higher than those with only wild-type alleles (2461±1348 g for JAK2V617F-positive compared with 1837±1142 g for JAK2 wild-type only, student's t-test, P=0.096) (Table 2). Similarly, spleen weights among only the non-CML CMPD cases (PV, ET and CIMF) were slightly higher in JAK2V617F-positive cases (2627±1371 g for JAK2V617F-positive compared with 2390±1269 g for JAK2 wild-type only, student's t-test, P=0.713). There were no statistically significant differences between the hemoglobin concentration, platelet count or white blood cell count in cases with or without JAK2V617F alleles in splenic EMH cells at the time of splenectomy among all cases studied or among the non-CML CMPD cases (Table 3). There was also no correlation between presence of JAK2 mutant alleles and histological pattern of splenic EMH cells (Table 1).

Table 2 Mean and standard deviation of spleen weight, hemoglobin concentration, platelet count and white blood cell count in all cases
Full size table
Table 3 Mean and standard deviation of spleen weight, hemoglobin concentration, platelet count and white blood cell count in non-CML CMPD cases
Full size table

Expression of Bcl-xL, a well-described antiapoptotic protein and downstream target of JAK2, was studied in splenic EMH cells by immunohistochemistry. In the 26 cases available for further morphologic evaluation, the percentage of megakaryocytes expressing Bcl-xL was significantly higher in spleens with the JAK2V617F allele (62±22% for JAK2V617F-positive compared with 27±20% for JAK2 wild-type only, student's t-test, P=0.0002) (Table 1 and Figure 2). Furthermore, expression of Bcl-xL in more than 50% of megakaryocytes was observed in 10/13 JAK2V617F-positive cases (77%) but only 1/13 JAK2V617F-negative cases (8%) (Fisher's exact test, P=0.001). This finding suggests that JAK2V617F upregulates Bcl-xL expression and may contribute to the extramedullary expansion of EMH clones with mutant alleles.

Figure 2
figure 2

Bcl-xL immunohistochemistry. Representative micrographs of positive cytoplasmic Bcl-xL immunohistochemical staining of EMH megakaryocytes in a JAK2V617F-positive spleen (left panel, case 11, × 200) and negative Bcl-xL immunohistochemical staining in a JAK2V617F-negative spleen (right panel, case 23, × 200).

Full size image

Discussion

In this study, we demonstrated that the JAK2V617F allele is frequently present in the splenic EMH cells of CMPD patients but rarely present in the splenic EMH cells of patients with other myeloid disorders. This frequency of JAK2V617F positivity in splenic EMH cases is similar to those previously reported using peripheral blood and bone marrow samples from patients with PV, CIMF and ET.6, 7, 8, 9, 12, 13, 25, 26, 27 These findings confirm that JAK2V617F is predominantly associated with the hematopoiesis of non-CML CMPD clones, even in the setting of splenic EMH. Of interest, although our sample size was relatively small, the frequency of JAK2V617F in splenic EMH cases associated with MDS and CMML appears to be slightly higher than those previously described in peripheral blood or bone marrow studies.26, 27, 28, 29 This suggests that patients with MDS or CMML carrying the JAK2 V617F mutation may be more likely to develop splenic EMH, a feature more commonly observed in CMPD than MDS or CMML. Further studies are needed to confirm this observation.

Our results support the hypothesis that hematopoietic precursors giving rise to splenic EMH in patients with neoplastic myeloid disorders may be derived from the transformed bone marrow clone. In virtually all cases of this series, the genotype of the splenic EMH cells was concordant with the corresponding bone marrow specimen. LCM-enriched EMH fractions had more JAK2 V617F mutant alleles than the non-EMH fractions, and one case had JAK2V617F only in spleen, suggesting that the mutant allele is exclusive to the splenic EMH cells. The LCM procedure eliminates the possibility that JAK2V617F alleles were contributed by contaminating neoplastic bone marrow cells in circulation. We hypothesize that the clonal myeloid cells may enter the bone marrow sinuses via intrasinusoidal hematopoiesis, a morphological finding commonly observed in CIMF.30 These precursor cells may then circulate to and become trapped within the spleen, leading to a clonal extramedullary hematopoietic expansion. O'Malley et al3, 4 recently reported that EMH cases associated with neoplastic myeloid disorders showed concordant loss-of-heterozygosity (LOH) lesions and X-chromosome inactivation patterns in the respective bone marrow and spleen specimens. Furthermore, additional LOH abnormalities were identified in the spleen, suggesting evolution of the original clone. In addition, the spleen is a common site of blast transformation for patients with neoplastic myeloid disorders.19 It would be of great interest to further study our cohort of cases once assays using paraffin-embedded tissues become commercially available to detect other mutations associated with CMPD or acute leukemia, such as MPL-W515L, MPL-W515K, JAK2-L611S, JAK2-K607N and JAK2-T875N.31, 32, 33, 34, 35, 36

The finding of increased Bcl-xL expression in megakaryocytes of JAK2V617F-positive spleen specimens suggests that the JAK2 V617F mutation may induce myeloproliferation via the STAT5/Bcl-xL pathway, at least partially, in splenic EMH. In support, sustained high levels of Bcl-xL were recently demonstrated to promote erythropoietin independent colony formation in vitro.14 However, regardless of JAK2 genotype, Bcl-xL expression was not observed in the erythroid or myeloid lineages of the splenic EMH cells in our series. This is most likely due to the limited sensitivity of immunohistochemistry for detecting low levels of Bcl-xL expression. Furthermore, the observed tendency for larger spleens to occur in JAK2V617F-positive cases agrees with a previous report of 166 CMPD patients (43 PV, 111 ET, 12 CIMF) demonstrating the JAK2 V617F mutation to confer a higher probability of splenomegaly.37 Thus, the various JAK2/STAT5 signal transduction pathways, including activated expression of Bcl-xL, may be important to the extramedullary expansion of neoplastic myeloid cells.

CMPD patients with the JAK2 V617F mutation were previously shown to have higher hemoglobulin/hemocrit levels and leukocyte counts at the time of diagnosis.38, 39, 40 However, we did not observe such differences in this series of patients with established disease at the time of splenectomy. This may be due to the effect of splenomegaly and/or EMH in modifying these hematologic parameters. Alternatively, although less likely, it could be secondary to a relatively smaller sample size.

In summary, our results suggest that the JAK2V617F-positive cells in splenic EMH originate from the transformed bone marrow clone and the mutation plays a significant role in the initial development or subsequent progression of splenic EMH in JAK2V617F-positive patients with CMPD, CMML and MDS. Thus, the JAK2/STAT5 pathway inhibitors currently undergoing clinical trials may have therapeutic efficacy as a pharmaceutical alternative to surgical intervention for JAK2V617F-positive CMPD patients with significant splenomegaly.