Clinical Significance of Nm23-H1 Proteins Expressed on Cell Surface

Home , H1 (classification)

Leukemia (2003) 17, 196–202  2003 Nature Publishing Group All rights reserved 0887-6924/03 $25.00 www.nature.com/leu Clinical signiﬁcance of nm23-H1 proteins expressed on cell surface in non-Hodgkin’s lymphoma N Niitsu1, Y Honma2, K Iijima3, T Takagi4, M Higashihara1, U Sawada5 and J Okabe-Kado2

1Department of Hematology and Internal Medicine IV, Kitasato University School of Medicine, Kanagawa, Japan; 2Saitama Cancer Center Research Institute, Saitama, Japan; 3First Department of Internal Medicine, Toho University School of Medicine, Tokyo, Japan; 4Hematology- Oncology Division, Chiba, Cancer Center Hospital, Chiba, Japan; and 5First Department of Internal Medicine, Nihon University School of Medicine, Tokyo, Japan

The nm23 gene was isolated as a metastasis suppressor gene in both aggressive NHL and AML.7–9 In a collaborative study that exhibits low expression in high-level metastatic cancer between many institutions involving a large number of cells. Its gene is related to the prognosis of acute myelogenous leukemia (AML) and non-Hodgkin’s lymphoma (NHL). In this patients, we found that serum nm23-H1 was an important study, we examined the expression of nm23-H1 protein on the independent prognostic factor in NHL for diffuse large B cell lymphoma cell surface of NHL. In 28 of 108 cases (25.9%), we lymphoma (DLBCL) and peripheral T cell lymphoma (PTCL).8 observed Ն20% of cell surface nm23-H1 protein expression We also examined four distinct risk groups defined by the and expression was especially high in peripheral T cell lym- widely used clinical international prognostic index (IPI) and phomas and extranodal NK/T cell lymphomas. We also demonstrated that when serum nm23-H1 protein levels are observed a significant correlation between serum nm23-H1 level and cell surface nm23-H1 expression levels. In patients high, prognosis is poor in all four risk groups. Given these with high levels of cell surface nm23-H1 expression, overall and results, measurement of nm23-H1 protein may be useful in progression-free survival rates were significantly lower than the treatment method selection process. Whether such serum those in patients with low surface nm23-H1 expression levels. nm23-H1 protein is directly produced by lymphoma cells is When surface nm23-H1 and serum nm23-H1 were combined, unclear, but its production appears to reflect the prognosis of patients with high levels of both exhibited a poorer prognosis lymphoma. In this study, to examine the clinical significance than patients with a high level of one or the other. These results indicate that in addition to serum nm23-H1, cell surface nm23- of nm23-H1 protein on the surface of lymphoma cells, we H1 may be used as a prognostic factor in planning a treatment studied the relationships of surface nm23-H1 to serum nm23- strategy. The nm23-H1 protein appears to be intimately related H1 and other cell surface markers by flow cytometry measure- to biological aggressiveness of lymphoma and, therefore, ment and tracked the subsequent survival rates. might be a molecular target of NHL treatment. Leukemia (2003) 17, 196–202. doi:10.1038/sj.leu.2402699 Keywords: nm23-H1 proteins; cell surface; non-Hodgkin’s lymphoma; prognostic factor Materials and methods Patients Introduction We measured cell surface nm23-H1 expression from February 1999 to March 2000 in a consecutive series of 108 patients The first nm23 gene was isolated on the basis of its reduced with untreated NHL (62 men and 46 women; median age, 68 expression in highly metastatic murine melanoma cell lines, years; age range, 22–78 years) whose clinical course was as compared with their nonmetastatic counterparts, and has being managed by the lymphoma treatment study group in therefore been proposed as a metastatic suppressor gene.1 Japan. The cases were classified morphologically according Subsequently, high homology between nm23 proteins and to the REAL schema10 as defined elsewhere with or without nucleoside diphosphate kinases (NDPK) has been identified in adjunctive immunohistochemical and molecular studies. Of a number of species.2 We found that a differentiation inhibi- these 108 patients with NHL, 79 presented with B cell neo- tory factor purified from conditioned medium of a differen- plasms and 29 with T cell neoplasms. Of the 79 patients with tiation-resistant mouse myeloid leukemia cell line was ident- B cell neoplasms, three had B cell chronic lymphocytic leuke- ical to the nm23 protein.3,4 In humans, eight nm23 isotypes mia (CLL), three had mucosa-associated lymphoid tissue (nm23-H1, nm23-H2, DR-nm23, nm23-H4, nm23-H5, nm23- (MALT), five had mantle cell lymphoma (MCL), 13 had follicu- H6, nm23-H7 and nm23-H8) have been identified to date.5 lar lymphoma (FL), and 55 had DLBCL. Of the 29 patients Immunohistochemical studies have reported that patients with with T cell neoplasms, 18 had PTCL, two had anaplastic large high-grade non-Hodgkin’s lymphoma (NHL) and Hodgkin’s cell lymphoma (ALCL), six had extranodal NK/T cell lym- lymphoma exhibit significantly higher levels of nm23-H1 phoma (NK/T), and three had adult T cell leukemia/lymphoma expression than do those with low-grade NHL.6 (ATLL). Clinical staging was performed according to the Ann Recently, we established an enzyme-linked immunosorbent Arbor classification system.11 Evaluation included a complete assay (ELISA) technique to determine the serum level of nm23- history and physical examination, chest roentgenography, H1 protein. We previously reported that serum levels of bone marrow aspiration and biopsy, computed tomography nm23-H1 in aggressive NHL and acute myelogenous leuke- of the chest, abdomen and pelvis, hemogram and differential mia (AML) were significantly higher than those in controls, counts and routine biochemistry tests. Laparotomy was not and that high nm23-H1 levels correlated with poor prognosis performed for staging. The IPI was determined in all NHL cases.12 Patients with aggressive NHL and disseminated disease were treated with cyclophosphamide, vincristine, predni- Correspondence: N Niitsu, Department of Hematology and Internal Medicine IV, Kitasato University School of Medicine, 1-15-1 Kitasato, solone, bleomycin, doxorubicin and procarbazine (COP- 13 Sagamihara-shi, Kanagawa 228-8555, Japan; Fax: +81-42-778-9978 BLAM), cyclophosphamide, doxorubicin, vincristine and Received 4 February 2002; accepted 19 June 2002 prednisolone (CHOP),14 or another anthracycline-containing Cell surface nm23-H1 and NHL N Niitsu et al 197 combination chemotherapy regimen. Indolent lymphoma was strate and an alkaline phosphatase-detecting kit (BioRad Lab). in most cases treated with a single agent or cyclophospham- The reaction was stopped with 50 ␮l of 0.4 N NaOH. ide, vincristine and prednisone (COP). Patients were re-evalu- Absorbance was measured at 405–415 nm with a correction ated every few months, undergoing physical examination, wavelength of 620–630 nm using a microplate reader. hemogram and differential counts, biochemistry tests, and computed tomography of the chest, abdomen and pelvis. The median follow-up time was 21 months (range, 15–25 months), Statistical analysis and the final follow-up examination was carried out in April 2001. Approval of the study protocol was obtained from the The patients’ characteristics were compared using chi-square institutional review board at each participating institute. Infor- tests. Judgment criteria used for the analysis were progression- med consent was obtained from patients according to the free survival (PFS) and overall survival (OS). Progression was Declaration of Helsinki. defined as a progression of the lymphoma in nonresponding patients or partial response patients, a relapse for complete response patients, or death from any cause without pro- Flow cytometry for human cell surface nm23-H1 gression. PFS was calculated as the time from randomization to the first progression. Patients who did not experience a pro- The compound 7-amino-actinomycin D (7AAD) was pur- gression at the time of analysis were censored at the most chased from Sigma Chemical (St Louis, MO, USA). The drug recent date of disease assessment or at the stop date if the was diluted in phosphate-buffered saline (PBS) as a 25 ␮g/ml ° most recent date was later. OS was measured from the date stock solution, which was stored at 4 C. FITC-conjugated of randomization to the date of death, regardless of cause. monoclonal anti-nm23-H1 antibody was purchased from Sei- Patients who were still living at the time of analysis were cen- kagaku Kougyo (Tokyo, Japan). PE-conjugated anti-Leu4 sored at the most recent date of disease assessment or at the (CD3), Leu3a (CD4), Leu2a (CD8), LeuM5 (CD11c), Leu11a stop date if the most recent date was later. Survival analysis (CD16), and Leu12 (CD19) were purchased from Becton was performed according to the Kaplan–Meier method.15 The Dickinson (Cowley, UK) and CD5, CD10 and CD56 were pur- statistical significance of differences in survival was determ- chased from BD PharMingen (San Diego, CA, USA). FITC- and ined by the log-rank and generalized Wilcoxon’s tests.16 Dif- PE-conjugated murine IgG monoclonal antibodies of unre- ferences between groups were evaluated by the Mann–Whit- lated specificities were always used as controls. ney U test (nonparametric analysis),17 and P < 0.05 was taken Fresh tissues were transported immediately in RPMI 1640 to indicate significance. solution to the flow cytometry laboratory. Lymphocytes were disaggregated and released from solid tissue by injection of RPMI 1640 solution. Up to 1 × 106 cells were pelleted, incubated with 5 ␮l of 7AAD solution (final 1 ␮g/ml), 50 ␮lof Results FITC-conjugated monoclonal antibody, and PE-conjugated monoclonal antibody for 30 min at 4°C while protected from Correlation between nm23-H1 expression and serum light, and then pelleted by centrifugation. The supernatant was parameters or cell surface markers removed, and the cells were resuspended in Cell FIX solution (Becton Dickinson). Two-color flow cytometric immunophen- Figure 1 shows a case of DLBCL, in which both CD19 and otyping was performed on a FACS Caliber instrument (Becton nm23-H1 markers were positive but markers CD3, CD5, Dickinson) by collecting 10 000 ungated list-mode events, sel- CD16 and CD56 were negative. The results are the represen- ecting an appropriate lymphocyte gate on the combination of tative patterns of a case of surface nm23-H1-positive DLBCL, forward and side scatter, and analyzing cells with the most and we also obtained similar results from another three appropriated lymphocyte gate. The 7AAD solution was used patients (data not shown). Figure 2 shows a case of extranodal to eliminate dead cells. Percentage surface nm23-H1 corre- NK/T cell lymphoma in which both CD56 and nm23-H1, and sponds to % of cells expressing nm23 at their surface. CD16 and nm23-H1 markers were positive but markers CD3, CD5 and CD19 were negative. The results are the representative patterns of a case of surface nm23-H1 positive extranodal ELISA for human nm23-H1 NK/T cell lymphoma, and we also obtained similar results from another three patients (data not shown). In 28 of 108 We previously established an ELISA procedure to determine patients (25.9%), Ն20% expression of cell surface nm23-H1 nm23-H1 protein levels in the serum.7–9 Briefly, 96-well plates was observed. On the other hand, the nm23-H1 expression (Corning, Corning, NY, USA) were coated with a monoclonal was observed in 12 of 55 patients with DLBCL (21.8%), and in anti-nm23-H1 antibody (Seikagakukougyo, Tokyo, Japan), eight of 18 patients with PTCL (44.4%). We found a significant washed four times with PBS, and incubated with 25% Block correlation between serum nm23-H1 levels and cell surface Ace solution (Dainihon Seiyaku, Osaka, Japan). Serum nm23-H1 expression (r = 0.531, P < 0.0001) (Figure 3a). A samples were diluted two-fold with PBS and 50 ␮l aliquots slightly correlation was observed between serum LDH level were added to the wells. After incubation at room temperature and cell surface nm23-H1 expression (Figure 3b), although a for 1 h, the wells were washed four times with PBS containing significant correlation was found between serum nm23-H1 0.05% Tween 20 (T-PBS). Samples were then incubated at and LDH levels (r = 0.464, P < 0.0001) (data not shown). No room temperature for 1 h with a polyclonal anti-nm23-H1 correlation was observed with soluble interleukin-2 receptor antibody (Santa Cruz Biotechnologies, Santa Cruz, CA, USA), (sIL-2R). No significant correlation was found between cell washed four times with T-PBS, and incubated with alkaline surface nm23-H1 expression and CD19 (Figure 4a) or CD3 phosphatase-conjugated anti-rabbit IgG (BioRad Lab, Rich- (Figure 4b), while significant correlations were seen with mond, CA, USA). After four washes with T-PBS, alkaline phos- CD16 (r ϭ 0.389, P = 0.000031) (Figure 4c) and CD56 (r = phatase activity was detected with diethanolamine as a sub- 0.37, P = 0.00008) (Figure 4d). We also examined CD4, CD5,

Leukemia Cell surface nm23-H1 and NHL N Niitsu et al 198

Figure 1 Cell surface expression of CD3 (b), CD19 (c), CD56 (d) Figure 2 Cell surface expression of CD3 (b), CD19 (c), CD56 (d) and CD16 (e) with nm23-H1 with diffuse large B cell lymphoma and CD16 (e) in patients with extranodal NK/T cell lymphoma by two- (DLBCL) by two-color flow cytometry. Fresh tissues were transported color flow cytometry. Same as the legend of Figure 1. The results immediately in RPMI 1640 solution to the flow cytometry laboratory. are the representative patterns of a case of surface nm23-H1 positive Lymphocytes were disaggregated and released from solid tissue by extranodal NK/T cell lymphoma patient, and we also obtained similar injection of RPMI 1640 solution. To eliminate dead cells, cells were results from another three patients (data not shown). incubated with 7AAD solution, and they were incubated with FITC- conjugated monoclonal anti-nm23-H1 antibody, and PE-conjugated anti-CD3 (b), anti-CD19 (c), anti-CD56 (d), or anti-CD16 (e) mono- nificantly higher in PTCL (26.9 ± 21.02%, P < 0.01) and NK/T clonal antibody. Two-color flow cytometric immunophenotyping was cell lymphoma (46.04 ± 37.77%, P < 0.001). Thus, nm23-H1 performed on a FACS Caliber instrument by collecting 10 000 ungated list-mode events, selecting an appropriate lymphocyte gate (a region appears to be related to the biological aggressiveness of lym- in a) on the combination of forward (FSC) and side scatter (SSC). The phoma. results are the representative patterns of a case of surface nm23-H1- positive DLBCL, and we also obtained similar results from another three patients (data not shown). Relationship between cell surface nm23-H1 protein levels and clinicopathological data in aggressive lymphoma CD8, CD10 and CD23 but did not find any significant correlation with cell surface nm23-H1 (data not shown). The relationship between cell surface nm23-H1 expression levels and clinicopathological factors was investigated in 81 patients with aggressive NHL (Table 2). When the median Studies of cell surface nm23-H1 protein expression nm23-H1 expression was used as the threshold value, a high level in each histopathological type expression level of nm23-H1 before treatment was strongly associated with poor prognostic features such as elevated As shown in Table 1, the cell surface nm23-H1 expression serum LDH level and Ann Arbor stage (P < 0.05 for all com- level (mean ± s.d. was 2.96 ± 1.14% in reactive lymphoaden- parisons, Mann–Whitney U test or Kruskal–Wallis test). The opathy (including viral infection and systemic lupus assessment of each risk group classified according to IPI erythematosis) (n = 7). In addition, we examined cell surface revealed significantly higher expression of nm23-H1 in the nm23-H1 expression level for each histopathological type. high–intermediate (H–I) and high (H) risk groups than in the Compared with reactive lymphoadenopathy, no statistically low (L) and low–intermediate (L–I) risk groups (P = 0.01, Krus- significant difference was found for all B cell lymphomas; kal–Wallis test). Complete remission was achieved in 65 however, among T cell lymphomas, the expression was sig- (80.2%) of the 81 patients, and the median surface level of

Leukemia Cell surface nm23-H1 and NHL N Niitsu et al 199 Correlation of cell surface nm23-H1 with OS and PFS

The 81 patients with aggressive NHL were divided into two groups on the basis of cell surface nm23-H1 expression levels, with a threshold of 40%. Then the OS rate was calculated for patients falling into each 10% interval from an nm23-H1 expression of 20% upward. OS rates were significantly different for patients having nm23-H1 expression levels higher than 40% compared with those having levels of 40% or lower. Consequently, an nm23-H1 level of 40% was taken as the threshold value. The 2-year OS rate for the high (n = 8) nm23- H1 group was 20.0% and for the low (n = 73) nm23-H1 group was 76.2% (P = 0.0008 for the log-rank test and P = 0.0014 for generalized Wilcoxon’s test) (Figure 5a), with PFS values of 23.4% for the high group and 46.0% for the low group (P = 0.02 for the log-rank test and P = 0.0068 for generalized Wilcoxon’s test) (Figure. 5b). This indicated that the prognosis of aggressive lymphoma was poor when the cell surface nm23-H1 level was 20% or higher. We investigated the possibility of obtaining a more accurate prognosis by combining cell surface nm23-H1 and serum nm23-H1 measurements. The combination of cell surface nm23-H1 (40%) and serum nm23-H1 (80 ng/ml) is shown in Figure 6. We divided the 81 patients into three groups on the basis of both cell surface nm23-H1 and serum nm23-H1: group A (n = 64) had low levels of both factors, group B (n = 11) had a low level of only one factor, and group C (n = 6) had no low levels. The 2-year OS rate for group A was 78.6%, group B was 49.7%, and group C was 0% (Figure 6a). An analysis of survival probability among these groups showed Figure 3 Relationship between serum nm23-H1 and surface that group C had a significantly shorter survival period than nm23-H1 expression (a) and between LDH and surface nm23-H1 = expression (b). the other groups (P 0.0001). The 2-year PFS rate for group A was 50.1%, group B was 29.5%, and group C was 0%, and the prognosis for group C was again significantly worse than nm23-H1 in these patients was 11.3% at the time of diagnosis. those for the other groups (P = 0.0001 for the log-rank test This value was significantly lower than that in patients who and P = 0.003 for generalized Wilcoxon’s test) (Figure 6b). showed a partial remission or failed to respond (18.5%, P = These results indicate that the combination of cell surface 0.0326), suggesting a close relationship between the surface nm23-H1 and serum nm23-H1 may be a useful prognostic level of nm23-H1 and therapeutic responsiveness. factor.

Figure 4 Relationship between CD19 expression and surface nm23-H1 expression (a), CD3 expression and surface nm23-H1 expression (b), CD16 expression and surface nm23-H1 expression (c), and CD56 expression and surface nm23-H1 expression (d).

Leukemia Cell surface nm23-H1 and NHL N Niitsu et al 200 Table 1 Surface nm23-H1 protein expression in non-Hodgkin’s lymphoma

No. of Surface nm23-H1 patients mean ± s.d. (range)

B cell neoplasms B cell chronic lymphocytic leukemia 3 9.03 ± 7.70 (0.4–23.2) Mucosa-associated lymphoid tissue 3 3.73 ± 1.87 (0.2–9.6) Mantle cell lymphoma 5 14.80 ± 5.56 (0.9–40.2) Follicular lymphoma 13 4.79 ± 2.16 (0.4–42.1) Diffuse large B cell lymphoma 55 15.81 ± 11.96 (0.8–50.2) T cell and NK cell neoplasms Peripheral T cell lymphoma 18 26.90 ± 21.02* (8.6–70.4) Anaplastic large cell lymphoma 2 8.25 ± 0.50 (8.2–8.4) Extranodal NK/T cell lymphoma 6 46.04 ± 37.77*** (14.8–92.1) Adult T cell lymphoma/leukemia 3 20.77 ± 15.46 (5.2–40.1) Reactive hyperplasia 7 2.96 ± 1.14 (0–5.1)

*P < 0.01, ***P < 0.001.

Table 2 Correlation of surface nm23-H1 with clinical and prognostic factors in aggressive lymphoma

Characteristic No. of Median surface P value patients nm23-H1 (%)

Age (yr) >60 42 13.1 0.4105 Ͻ60 39 12.3 Serum LDH level Normal 25 10.6 0.0104 ϾNormal 56 14.3 Ann Arbor stage I–II 40 10.6 0.0367 III–IV 41 15.3 IPI L + L–I 47 10.6 0.0103 HI + H 34 16.9 Immunophenotype T 20 13.8 0.0836 B 55 11.3 NK 6 40.8 Response CR 65 11.3 0.0326 PR + NR 16 18.5

Discussion Figure 5 Overall survival (a) and progression-free survival (b) As shown in the Introduction section, eight distinctly different curves of patients with aggressive lymphoma. The eight patients with isotypes of nm23 gene family have now been identiﬁed in high surface nm23-H1 expression had a worse prognosis than the 73 humans.5 Among this family, only nm23-H1 and nm23-H2 patients with low surface nm23-H1 expression. have been extensively studied in human cancers. The proteins encoded by nm23-H1 and nm23-H2 show 88% amino acid sequence homology and the genes locate on the same region in various hematological neoplasms including AML, acute of chromosome 17q21 in tandem.18–21 Its gene products lymphoblastic leukemia, myelodysplastic syndrome, and encode NDPK which belongs to a family of genes highly con- chronic myelogenous leukemia (CML) in blastic crisis, than in served in eukaryotes. The nm23-H1 and nm23-H2 are com- normal blood cells.24 The progression of CML was pletely identical to NDPK-A and NDPK-B, respectively.21 accompanied by the overexpression of nm23-H1 and nm23- There is an inverse relationship between nm23-H1 expression H2 mRNA. This increase in nm23-H1 was observed not only and metastasis in cancers of the breast, ovaries, cervix, gastric in leukemia, but also in malignant lymphoma. It has been epithelia, hepatocellular carcinoma and melanomas.2 How- reported that high-grade NHL and Hodgkin’s lymphoma exhi- ever, overexpression of nm23-H1 is indicative of poor patient bited signiﬁcantly higher levels of nm23-H1 expression than prognosis in neuroblastomas, and mutations of nm23-H1 pro- low-grade NHL.6 We previously reported that the prognosis teins are detected in advanced neuroblastoma.22,23 Therefore, of AML cases in which nm23-H1 mRNA is highly expressed nm23-H1 is implicated in the regulation of metastasis in a was poor.24 However, measurement of nm23-H1 mRNA variety of human cancers. The nm23-H1 was overexpressed before treatment is time-consuming and we thus sought a

Leukemia Cell surface nm23-H1 and NHL N Niitsu et al 201 certain other prognostic factors, namely LDH or sIL-2R. There- fore, serum nm23-H1 might be produced not only as a result of nm23-H1 separating from the cell surface, but also with the aid of T cells or NK cells. However, the molecular mechanism by which nm23 protein is expressed on the cell surface remains unclear. This protein might be expressed via intra- cellular transportation after its synthesis or shed from cells and then attached to the cell surface. Moreover, as nm23 pos- sesses an RGD sequence, recognized to be characteristic of the integrin family, it may possess a secretory function, similar to cytokinins, in addition to a function as an adhesion ligand.27 However, it is more likely that cytoplasmic nm23- H1 is released from cells (after lyses) and is then captured by other cells at their surface, since nm23-H1 displays no leader peptide. Moreover, it would also prove very interesting to know whether nm23-H1, expressed on cell surfaces, is intact (full-length) or fragmented. Further studies could clarify this point. In clinical observation studies and nm23-H1 cell surface studies, surface nm23-H1 was found to be related to LDH level and stage in a manner similar to serum nm23-H1, suggesting that surface nm23-H1 may also be related to lymphoma dissemination or tumor burden. In the cases of complete remission, expression was low, possibly related to treatment reactivity. Contemporary NHL treatment includes assessment with the IPI, proposed in 1993 as a prognosis fore- casting model.12 The IPI abstracts five prognostic factors: age Figure 6 Classification of patients with aggressive non-Hodgkin’s (>60 years old), clinical stage (III, IV), performance status (2– lymphoma according to the two important independent prognostic 4), serum LDH (>normal), and number of extranodal lesions factors. Overall survival (a) and progression-free survival (b) curves (>1). Depending on the number of factors, IPI then divides are shown. Group A (n = 64): low serum nm23-H1 (<80 ng/ml) and cases into four groups: L risk (0–1), L–I risk (2), H–I risk (3) low surface nm23-H1 expression (<40%). Group B (n = 11): low and H risk (4–5). Recently, many biological prognostic factor serum nm23-H1 (<80 ng/ml) and high surface nm23-H1 expression studies have been performed to seek a replacement for IPI or (>40%) or high serum nm23-H1 (>80 ng/ml) and low surface nm23- H1 expression (<40%). Group C (n = 6): high serum nm23-H1 (>80 another indicator that may be used in combination with IPI. ng/ml) and high surface nm23-H1 expression (>40%). In one such study, the genetic expression of DLBCL was sys- tematically analyzed using cDNA microarrays, and two DLBCL types that were clearly distinguishable at the molecu- more rapid method that would require only a small blood lar level were identified (one with germinal center B-like RNA sample, and established a system in which nm23-H1 protein expression and the other with activated B-like RNA in blood is measured by ELISA. In a previous study, we meas- expression). Comparison of survival rates revealed that the ured the serum nm23-H1 protein levels by ELISA in 102 prognosis of activated B-like DLBCL was significantly poorer patients with AML and found that serum nm23-H1 protein than that of germinal center B-like DLBCL; the same result levels were high in patients with AML compared with con- was obtained when the comparison was restricted to low-risk trols, and that patients with high levels exhibited a poor prog- clinical patients.28 Shipp and her colleagues, who identified nosis.9 We had previously reported that serum nm23-H1 lev- a B-aggressive lymphoma (BAL) gene in DLBCL, found that els in patients with NHL were also significantly high the level of expression of this gene among high-risk DLBCL compared with controls and that the prognosis for high-level patients was significantly higher than that among low-risk cases was poor.7 However, no proof exists that serum nm23- DLBCL patients and indicated that this gene could be used as H1 is directly produced by lymphoma cells. a novel prognostic factor.29 We are planning to investigate The expression of nm23-H1 or nm23-H2 proteins on the further the relationship between nm23 and these biological cell surface of leukemia cell lines and lymphoma cell lines prognostic factors. has been reported.25 Willems et al26 have recently reported We have reported that serum nm23-H1 levels, which can that they could not demonstrate nm23 on the surface of any be measured from only a small amount of serum, can be used normal hematopoietic cells including lymphocytes. In this as a prognostic factor for aggressive NHL (not only for DLBCL study, the cell surface nm23-H1 of patients with NHL was but also for PTCL). In this study of cell surface nm23-H1 measured by flow cytometry; 25.9% of cases were positive expression, prognosis was poor in cases that showed Ն40% (Ն20% expression) and, specifically, more positive cases were expression and poorer in cases in which the expression of seen in PTCL and NK/T cell lymphomas. The expression of both cell surface nm23-H1 and serum nm23-H1 was high. cell surface nm23-H1 was high in NK/T cell lymphomas com- Therefore, the combination of two markers appears useful for pared with B cell lymphomas and we reported similar results determining the poorer prognosis group in need of front-line for serum nm23-H1.8 A significant correlation was found peripheral blood stem cell transplant treatment, given that between cell surface nm23-H1 and serum nm23-H1, which prognosis is poor in the cases where expression of either may suggest the possibility of cell surface nm23-H1 being lib- serum nm23-H1 or cell surface nm23-H1 is high. Therefore, erated into serum. We also identified a significant correlation the expression of nm23-H1 can be measured in not only intra- between serum nm23-H1, but not cell surface nm23-H1, and cellular RNA and serum protein, but also on the cell surface.

Leukemia Cell surface nm23-H1 and NHL N Niitsu et al 202 As each one of these nm23-H1 parameters is a prognostic fac- 12 Shipp MA. A predictive model for aggressive non-Hodgkin’s lym- tor, nm23-H1 is thus a fundamental prognostic factor for lymphomas. The International Non-Hodgkin’s Lymphoma Prognostic phoma. In addition, cell surface nm23-H1 expression can be Factors Project. N Engl J Med 1993; 329: 987–994. 13 Niitsu N, Umeda M. Biweekly COP-BLAM (cyclophosphamide, easily measured simultaneously with lymphoma surface vincristine, prednisone, bleomycin, doxorubicin and marker by flow cytometry, making it a potentially useful prog- procarbazine) regimen combined with granulocyte colony stimul- nostic factor for deciding treatment strategy for lymphoma. ating factor (G-CSF) for intermediate-grade non-Hodgkin’s lym- The possibility of new treatment methods that target nm23- phoma. Eur J Haematol 1996; 56: 163–167. H1 on the surface of lymphomas remains to be explored. 14 McKelvey EM, Gottlieb JA, Wilson HE, Haut A, Talley RW, Ste- phens R, Lane M, Gamble JF, Jones SE, Grozea PN, Gutterman J, Coltman C, Moon TE. Hydroxyldaunomycin (Adriamycin) combination chemotherapy in malignant lymphoma. Cancer 1976; 38: Acknowledgements 1484–1493. 15 Kaplan EL, Meier P. Nonparametric estimation from incomplete Supported in part by a grant from the Ministry of Health and observations. J Am Stat Assoc 1958; 53: 457–481. Welfare, and Grants-in-Aid for Scientific Research (C) and 16 Wilcoxon F. Individual comparisons by ranking methods. Bio- Cancer Research, from The Ministry of Education, Culture, metrics Bull 1945; 1: 80–83. 17 Mantel N. Evaluation of survival data and two new rank order Sports, Science, and Technology (MEXT) Japan. statistics arising in its consideration. Cancer Chemother Rep 1966; 50: 163–170. 18 Stahl, JA, Leone A, Rosengard AM, Porter L, Kin CR, Steeg PS. References Identification of a second human nm23 gene, nm23-H2. Cancer Res 1991; 51: 445–449. 1 Steeg PS, Bevilacqua G, Kopper L, Thorgeirsson UP, Talmadge JE, 19 Backer JM, Mendra CE, Kovesdi I, Fairhurst JL, O’Hara B, Eddy Liotta LA, Sobel ME. Evidence for a novel gene associated with RL, Shows TB, Methew S, Murty VV, Chaganti RS. Chromosomal low tumor metastatic potential. J Natl Cancer Inst 1988; 80: localization and nucleoside diphosphate kinase activity of human 200–204. metastasis-suppressor genes NM23-1 and NM23-2. Oncogene 2 De La Rose A, Williams RL, Steeg PS. Nm23/nucleoside diphos- 1993; 8: 497–502. phate kinase: toward a structural and biochemical understanding 20 Chandrasekharappa SC, Gross LA, King SE, Collins FS. The human of its biological functions. BioEssays 1995; 17: 53–62. NME2 gene lies within 18kb of NME1 in chromosome 17. Genes 3 Okabe-Kado J, Kasukabe T, Honma Y, Hayashi M, Henzel WJ, Chromosomes Cancer 1993; 6: 245–248. Hozumi M. Identity of a differentiation inhibiting factor for mouse 21 Gilles AM, Presecan E, Vonica A, Lascu I. Nucleoside diphosphate myeloid leukemia cells with nm23/nucleoside diphosphate kinase. kinase from human erythrocytes. J Biol Chem 1991; 266: 8784– Biochem Biophys Res Commun 1992; 182: 987–994. 8789. 4 Okabe-Kado J, Kasukabe T, Honma Y, Hayashi M, Hozumi M. 22 Leone A, Seeger RC, Hong CM, Hu, YY, Arboleda MJ, Brodeur Purification of a factor inhibiting differentiation from conditioned GM, Stram D, Slamon DJ, Steeg PS. Evidence for nm23 RNA over- medium of nondifferentiating mouse myeloid leukemia cells. J Biol expression, DNA amplification and mutation in aggressive child- Chem 1988; 263: 10994–10999. hood neuroblastoma. Oncogene 1993; 8: 855–865. 5 Lacombe ML, Milon L, Munier A, Mehus JG, Lambeth DO. The 23 Chang C, Zhu X-X, Thoraval D, Ungar D, Rawwas J, Hora N, human nm23/nucleoside diphosphate kinases. J Bioenerg Strahler J, Hanash S, Radany E. nm23-H1 mutation in neuroblas- Biomembr 2000; 32: 247–258. toma. Nature 1994; 370: 335–336. 6 Aryee DN, Simonitsch I, Mosberger I, Kos K, Mann G, Schlogl E, 24 Yokoyama A, Okabe-Kado J, Wakimoto N, Kobayashi H, Sakash- Potschger U, Gadner H, Radaszkiewicz T, Kovar H. Variability of ita A, Maseki N, Nakamaki T, Hino K, Tomoyasu S, Tsuruoka N, nm23-H1/NDPK-A expression in human lymphomas and its Motoyoshi K, Nagata N, Honma Y. Evaluation by multivariate relation to tumor aggressiveness. Br J Cancer 1996; 74: 1693– analysis of the differentiation inhibitory factor nm23 as a prognos- 1698. tic factor in acute myelogenous leukemia and application to other 7 Niitsu N, Okabe-Kado J, Kasukabe T, Yamamoto-Yamaguchi Y, hematologic malignancies. Blood 1998; 91: 1845–1851. Umeda M, Honma Y. Prognostic implications of the differentiation 25 Urano T, Furukawa K, Shiku H. Expression of nm23/NDP kinase inhibitory factor nm23-H1 protein in the plasma of aggressive non- proteins on the cell surface. Oncogene 1993; 8: 1371–1376. Hodgkin’s lymphoma. Blood 1999; 94: 3541–3550. 26 Willems R, Van Bockstaele DR, Landon F, Lenjou M, Nijs G, 8 Niitsu N, Okabe-Kado J, Okamoto M, Takagi T, Yoshida T, Aoki Snoeck HW, Berneman ZN, Slegers H. Decrease in nucleoside S, Hirano M, Honma Y. Serum nm23-H1 protein as a prognostic diphosphate kinase expression during hematopoietic maturation. factor in aggressive non-Hodgkin’s lymphoma. Blood 2001; 97: J Biol Chem 1993; 273: 13663–13668. 1202–1210. 27 MacDonald NJ, de la Rosa A, Steeg PS. The potential roles of 9 Niitsu N, Okabe-Kado J, Nakayama M, Wakimoto N, Sakashita A, nm23 in cancer metastasis and cellular differentiation. Eur J Can- Maseki N, Motoyoshi K, Umeda M, Honma Y. Plasma levels of the cer 1995; 31: 1096–1100. differentiation inhibitory factor nm23-H1 protein and their clinical 28 Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald implications in acute myelogenous leukemia. Blood 2000; 96: A, Boldrick JC, Sabet H, Tran T, Yu X, Powell JI, Yang L, Marti 1080–1086. GE, Moore T, Hudson J Jr, Lu L, Lewis DB, Tibshirani R, Sherlock 10 Harris NL, Jaffe ES, Stein H, Banks PM, Chan JK, Cleary ML, Delsol G, Chan WC, Greiner TC, Weisenburger DD, Armitage JO, G, De Wolf-Peeters C, Falini B, Gatter KC. A revised European– Warnke R, Staudt LM. Distinct types of diffuse large B-cell lym- American classification of lymphoid neoplasms: a proposal from phoma identified by gene expression profiling. Nature 2000; 403: the International Lymphoma Study Group. Blood 1994; 84: 503–511. 1361–1392. 29 Aguiar RT, Yakushijin Y, Kharbanda S, Salgia R, Fletcher JA, Shipp 11 Carbone PP, Kaplan HS, Musshoff K, Smithers DW, Tubiana M. MA. BAL is a novel risk-related gene in diffuse large B-cell lym- Report of the Committee on Hodgkin’s Disease Staging Classi- phomas that enhances cellular migration. Blood 2000; 96: fication. Cancer Res 1971; 31: 1860–1861. 4328–4334.

Leukemia