Purpose: The B7 family molecules have been shown to regulate immune responses in both positive and negative fashions. Their roles in the progression of human cancers, however, are not well established. The aim of this study was to examine whether leukemic cells of acute myeloid leukemia express functional B7 family molecules and, if so, whether such expression has any clinical significance.

Experimental Design: The expression of four B7 family molecules, B7.1, B7.2, B7-H1, and B7-H2, on leukemic cells from acute myeloid leukemia patients was analyzed by flow cytometry. The function of the expressed molecules was examined by the allogeneic mixed lymphocyte-leukemic cell reaction, and their relationship to the clinical data and survival was analyzed.

Results: Although B7.1 and B7-H1 expressions were rare, the cells from a substantial number of acute myeloid leukemia cases expressed B7.2 and B7-H2 molecules [mean percentages of B7.2- and B7-H2-positive cells were 28.9% (n = 58) and 15.3% (n = 59), respectively]. Patients in whom >25% of leukemic cells expressed B7-H2 had significantly shorter survival, and this B7-H2 positivity had the strongest prognostic value when B7-H2 and other prognostic factors were analyzed together by multivariate analysis (P = 0.0108). Furthermore, B7.2 expression was associated with hyperleukocytosis (P = 0.026). Consistent with this finding, acute myeloid leukemia cells expressing B7.2 and B7-H2 induced allogeneic CD4+ T cells to proliferate and secrete interleukin-4 and interleukin-10 in vitro, effects that were partially blocked by antibodies against B7.2 and B7-H2.

Conclusions: Our results indicate the expression of functional B7.2 and B7-H2 molecules, and these molecules may facilitate progression of acute myeloid leukemia.

Optimal activation of T cells requires second costimulatory signals together with the first signal delivered by engagement of the T-cell receptor with peptide-MHC complex on antigen-presenting cells (APC). The interaction between B7 molecules [i.e., B7.1 (CD80)/B7.2 (CD86) on APCs and CD28/CTLA-4 molecules on T cells] is the most characterized costimulatory signal pathway and is thought to be crucial in eliciting an antitumor immune response (1, 2). On the other hand, it was reported that B7.2-expressing lymphoma cells were more aggressive in transplanted mice and that Th2 cytokines induced by a B7.2-mediated signal might play a role in this growth advantage (3). We have identified several new members of the B7 family: B7-H1, B7-H2, B7-H3, and B7-H4 (47). B7-H1 inhibits the proliferation of T cells through a counterreceptor, programmed death-1 (PD-1) (8). B7-H1 also delivers a stimulatory signal to T cells through a yet unidentified counterreceptor other than PD-1 (9). B7-H1 expression is detected not only on APCs, such as macrophages, B cells, and dendritic cells, but also on activated T cells and some tumor cells, such as lung and colon cancers (4, 5, 10). We reported that B7-H1 on tumor cells suppressed antitumor immunity by inhibiting proliferation of tumor-specific CTL (1012). B7-H2 is expressed on professional APCs and tumor cells, such as glioma and gastric carcinoma (1316). The counterreceptor of B7-H2, inducible costimulator (ICOS), is expressed on activated T cells (17). The B7-H2-ICOS signal induces T cells to proliferate and to secrete both Th1 and Th2 cytokines, such as interleukin (IL)-4 and IFN-γ, but not the potent Th1 cytokine, IL-2. Further, the B7-H2-ICOS signal induces IL-10 production, which plays an important role in modulating and damping immune responses (14, 17). When tested in highly polarized T-cell lines, ICOS blockade reduced the production of Th2 cytokines but not Th1 cytokines (18). Further, in a mouse transplant model, blocking of ICOS prevented the development of Th2-mediated chronic graft-versus-host disease (19). Therefore, the B7-H2-ICOS signal may favor the induction of Th2-type responses and thus may be involved in the negative regulation of cell-mediated immune responses against tumors. However, it has not yet been reported that B7-H2 expression on tumor cells is associated with such a negative regulation, but it has been reported that the B7-H2-ICOS interaction augments CTL expansion in mice transplanted with a potent immunogenic, MHC class I–positive tumor (20).

Because of their various modulatory functions, probably for and against tumor immunity, the pathophysiologic roles of B7 family molecules expressed by human tumor cells are of great interest. In acute myeloid leukemia (AML), it was reported that leukemic cells from a substantial number of patients expressed B7.2 (2123). However, conflicting results exist whether such patients are associated with a poor prognosis, and it remains unknown whether B7.2 on patients' AML cells has an immunologic function. Moreover, neither expression nor function of B7-H1 and B7-H2 molecules on AML cells has been reported. We prepared monoclonal antibodies (mAb) against both B7-H1 and B7-H2 molecules, which were suitable for flow cytometry. In this study, using flow cytometry, we analyzed the expression of various B7 family molecules, B7.1, B7.2, B7-H1, and B7-H2, on patients' AML cells. We showed that AML cells from a substantial number of patients expressed B7.2 and B7-H2 molecules and that these molecules have immunomodulatory functions. We also showed that the expression of these molecules was associated with the proliferative advantage held by leukemic cells as evidenced by the unfavorable outcomes and/or the hyperleukocytosis of the patients.

Subjects and cell preparation. Sixty-one new patients with adult de novo AML, who were diagnosed according to the French-American-British criteria, were the subjects of this study. Their respective French-American-British subtypes were 18 M1, 16 M2, 8 M3, 9 M4, 8 M5, and 2 M6. Cytogenetic data obtained by the standard Giemsa banding method were classified into three prognostic categories, which were defined previously based on other reports (2426). That is, “favorable” is the presence of inv(16), t(15;17), or t(8;21), both with and without any other abnormality; “unfavorable” is −5, −7, 5q−, 7q−, t(9;22), abnormalities of chromosomes 3q and/or 11q, and a complex karyotype (>3 chromosomal abnormalities); and “intermediate” is all other karyotypes. In the present cohort, there were no patients with t(9;22). The prognosis did not differ between patients with and without additional abnormalities in the favorable cytogenetics group. All patients were treated at the Main Hospital of Nippon Medical School (Tokyo, Japan) with intensive chemotherapy according to the Japan Adult Leukemia Study Group protocol, including anthracycline, enocitabine, and 6-mercaptopurine, except for the patients with M3 subtype who were treated with all-trans retinoic acid with or without anthracycline. The median follow-up for patients alive at the time of analysis was 42 months.

Aspirated bone marrow cells or peripheral blood rich in AML cells were obtained from the patients after informed consent was obtained. This study had been approved by the institutional review board. Mononuclear cells, which contained >90% AML cells, were prepared by Ficoll-Hypaque density gradient centrifugation (Sigma, St. Louis, MO). These cells were used immediately or cryopreserved in liquid nitrogen until use. The cryopreserved cells were thawed rapidly and washed with RPMI 1640 containing 10% fetal bovine serum before use. If the viability of the thawed cells, determined by trypan blue dye exclusion, was low, viable cells were collected by density gradient centrifugation using Blastretriever reagent (Japan Immunoresearch Laboratories, Takasaki, Japan) according to the manufacturer's instructions. More than 97% of the cells were viable after this centrifugation.

Cell lines. Twelve cell lines of hematologic malignancies were used. Jurkat, K562, KG-1, BALL-1, SAS413, KML-1, KU812, HL-60, and OIH-1 cells were purchased from RIKEN Cell Bank (Tsukuba, Japan). U937 and Dami cells were kindly supplied by Kirin Brewery Co., Ltd. (Takasaki, Japan). K051 cells were kindly provided by Dr. K. Inokuchi (Nippon Medical School). These cell lines, except for Dami cells, were maintained in RPMI 1640 containing 10% fetal bovine serum. For OIH-1 cells, granulocyte colony-stimulating factor (20 ng/mL; Biosource, Sunnyvale, CA) was added to the medium. Dami cells were maintained in Iscove's modified Dulbecco's medium containing 10% horse serum.

Reverse transcription-PCR. The method was reported previously (7). In brief, cDNA was synthesized from total cellular RNA. Success of cDNA synthesis was monitored by reverse transcription-PCR of β-actin. The primers for B7-H1 and B7-H2 were 5′-GACCTATATGTGGTAGAGTATGGTAGC-3′ and 5′-TTCAGCTGTATGGTTTTCCTCAGGATC-3′ (B7-H1) and 5′-CTGGGATCCAGCAGTGGTCCTTCT-3′ and 5′-CCCTGGGATTCCAGGAGGTTTT-3′ (B7-H2). PCR synthesis was run for 40 cycles, amplifying a 568-bp product for B7-H1 and a 693-bp product for B7-H2.

Flow cytometry. Mouse mAbs against human B7-H1 and B7-H2 were prepared as described previously (27, 28). FITC-conjugated mouse antibodies against human CD80 (L307.4) and CD86 (FUN-1) were purchased from BD PharMingen (San Diego, CA). Flow cytometry was done by the standard method as described previously (27). In brief, the cells were treated with human immunoglobulin to block nonspecific binding and then reacted with each anti-B7 family molecule antibody. To detect B7-H1 or B7-H2 expression, a FITC-conjugated second antibody (anti-mouse IgG; Biosource) was used. Isotype-matched negative controls were used in all assays. Analysis was done on a FACScan (Becton Dickinson, Mountain View, CA), and leukemic blasts were gated on forward versus side scatter plots. The expressions of B7 family molecules were evaluated based on the relative mean fluorescence intensity (MFI) and the percentage of positive cells. The relative MFI of each B7 family molecule staining was calculated as follows: MFI of B7 family molecule staining / MFI of isotype-matched mAb staining.

Allogeneic mixed lymphocyte-leukemic cell reaction. CD4+ T cells from healthy volunteers were positively selected by magnetic sorting using CD4 Microbeads according to the manufacturer's instructions (Miltenyi Biotec, Auburn, CA). The purity of isolated CD4+ T cells was >95% by flow cytometric analysis using a mAb against CD4. U937 cells and leukemic cells from AML patients, which expressed B7-H2 and B7.2, were treated with 1 mg/mL mitomycin C for 1 hour to inhibit their proliferation. After washing with RPMI 1640 containing 10% fetal bovine serum, the mitomycin C–treated cells (2 × 106/mL) were cocultured with allogeneic CD4+ T cells (2 × 106/mL) for 5 to 7 days in 96-well microtiter plates. During the last 15 hours of the culture, [3H]thymidine (1 μCi/well) was added to the wells to determine T-cell proliferation ([3H]thymidine incorporation in T cells). To block the B7-H2-ICOS and B7.2-CD28 pathways, antagonistic mouse anti-human mAbs against ICOS (kindly supplied by Japan Tobacco, Inc., Osaka, Japan) and B7.2 (Becton Dickinson) were added, respectively, to the culture.

In some experiments, culture supernatants of mixed lymphocyte-leukemic cell reaction were collected on day 5 of the culture. The concentrations of IFN-γ, IL-2, IL-4, and IL-10 in the supernatants were measured with sandwich ELISA kits according to the manufacturer's instructions (BD PharMingen).

Statistical analysis. Differences between two groups of data were determined by the χ2 test for categorical variables and Student's t test for continuous variables. Survival curves from the time of diagnosis were obtained by the Kaplan-Meier method and compared by the log-rank test. Patients who underwent stem cell transplantation or who were lost to follow-up were censored at the date of stem cell transplantation or the last follow-up. Multivariate analysis was done by Cox proportional hazards regression analysis. P < 0.05 was considered to be statistically significant.

Expression of B7 family molecules on hematologic cell lines. We analyzed 12 cell lines of hematologic malignances for the expression of various B7 family molecules. First, we analyzed B7-H1 mRNA expression by the PCR method and found that K562 and OIH-1 cells lacked B7-H1 mRNA, whereas KML-1 and HL-60 cells expressed low levels of B7-H1 mRNA. The other eight cell lines expressed high levels of B7-H1 mRNA (Fig. 1A). These results were confirmed by flow cytometric analysis using a mAb to B7-H1 (Fig. 1B; Table 1). B7-H2 mRNA expression was shown for all the cell lines, except Jurkat cells (Fig. 1C), and flow cytometric analysis using a mAb to B7-H2 yielded consistent results (Fig. 1D; Table 1). When these cell lines were treated with IFN-γ, B7-H1 expression was induced in the OIH-1 cells and enhanced in several other cell lines (Table 1). This positive effect of IFN-γ on B7-H1 expression is consistent with the results of our previous report (29). In contrast, the IFN-γ treatment showed little effect on the B7-H2 expression of these cells (Table 1). Tumor necrosis factor-α, which has been recognized as an inducer of B7-H2 in vivo, clearly enhanced B7-H2 expression by OIH-1, K562, and KU812 (Table 1).

Fig. 1.

B7-H1 and B7-H2 expressions analyzed by reverse transcription-PCR and flow cytometry for 12 cell lines of hematologic malignancies. A and C, equal amounts of cDNA from each cell line were amplified using primers specific for B7-H1 (A), B7-H2 (C), and β-actin (bottom of A and C). B and D, flow cytometric analyses of B7-H1 (B) and B7-H2 (D) expressions for four representative cell lines. Bold curves, staining with anti-B7-H1 or anti-B7-H2 antibody; thin curves, staining with isotype-matched control antibody. Percentages of cells positive for B7-H1 and B7-H2 and the relative MFI (in parentheses) are shown.

Fig. 1.

B7-H1 and B7-H2 expressions analyzed by reverse transcription-PCR and flow cytometry for 12 cell lines of hematologic malignancies. A and C, equal amounts of cDNA from each cell line were amplified using primers specific for B7-H1 (A), B7-H2 (C), and β-actin (bottom of A and C). B and D, flow cytometric analyses of B7-H1 (B) and B7-H2 (D) expressions for four representative cell lines. Bold curves, staining with anti-B7-H1 or anti-B7-H2 antibody; thin curves, staining with isotype-matched control antibody. Percentages of cells positive for B7-H1 and B7-H2 and the relative MFI (in parentheses) are shown.

Close modal
Table 1.

Expression of B7 family molecules on cultured cell lines of hematologic malignancies

Cell lineOriginB7-H1
B7-H2
B7.1
B7.2
UnstimulatedIFN-γUnstimulatedIFN-γTumor necrosis factor-αUnstimulatedIFN-γUnstimulatedIFN-γ
Jurkat T-ALL +++* 3.2 +++ 3.43 − 1.07 − 1.08 − 1.00 − 1.00 − 1.00 − 1.00 − 1.00 
BALL-1 B-ALL − 0.96 − 1.29 ++ 2.00 ++ 2.06 ++ 2.10 +++ 8.84 +++ 8.47 +++ 20.62 +++ 22.15 
KG-1 AML ++ 2.84 +++ 4.00 ++ 2.40 ++ 2.09 ++ 2.43 − 1.00 − 1.00 +++ 34.42 +++ 36.64 
K051 AML − 1.28 ++ 2.42 1.77 1.82 ND  − 1.00 − 1.00 1.88 1.87 
HL60 AML (promyelocyte) − 1.27 1.74 1.81 1.60 1.54 +++ 6.67 +++ 6.19 ++ 1.90 ++ 2.44 
U937 AML (monoblast) 1.38 ++ 1.76 1.66 1.33 1.92 − 1.00 − 1.00 +++ 4.22 +++ 4.77 
Dami AML (megakaryoblast) +++ 4.90 +++ 3.27 ++ 2.28 ND  ND  ++ 2.03 1.67 +++ 4.16 +++ 2.97 
K562 CML (erythroleukemia) − 1.00 − 1.00 ++ 2.17 ++ 2.33 +++ 6.70 1.70 1.42 − 1.00 − 1.00 
KU812 CML (basophilic cell) 1.51 1.87 1.85 1.33 ++ 3.02 2.12 1.97 1.68 +++ 3.45 
SAS413 CML − 0.84 ++ 1.46 1.36 1.79 1.52 2.23 2.23 − 1.00 − 1.00 
KML-1 B-cell lymphoma − 1.26 ++ 2.19 1.70 ++ 2.18 2.38 ++ 2.18 ++ 2.05 − 1.20 − 1.18 
OIH-1 Overt leukemia from MDS − 0.93 ++ 2.01 − 1.04 1.65 +++ 8.40 − 1.00 − 1.00 − 1.00 1.95 
Cell lineOriginB7-H1
B7-H2
B7.1
B7.2
UnstimulatedIFN-γUnstimulatedIFN-γTumor necrosis factor-αUnstimulatedIFN-γUnstimulatedIFN-γ
Jurkat T-ALL +++* 3.2 +++ 3.43 − 1.07 − 1.08 − 1.00 − 1.00 − 1.00 − 1.00 − 1.00 
BALL-1 B-ALL − 0.96 − 1.29 ++ 2.00 ++ 2.06 ++ 2.10 +++ 8.84 +++ 8.47 +++ 20.62 +++ 22.15 
KG-1 AML ++ 2.84 +++ 4.00 ++ 2.40 ++ 2.09 ++ 2.43 − 1.00 − 1.00 +++ 34.42 +++ 36.64 
K051 AML − 1.28 ++ 2.42 1.77 1.82 ND  − 1.00 − 1.00 1.88 1.87 
HL60 AML (promyelocyte) − 1.27 1.74 1.81 1.60 1.54 +++ 6.67 +++ 6.19 ++ 1.90 ++ 2.44 
U937 AML (monoblast) 1.38 ++ 1.76 1.66 1.33 1.92 − 1.00 − 1.00 +++ 4.22 +++ 4.77 
Dami AML (megakaryoblast) +++ 4.90 +++ 3.27 ++ 2.28 ND  ND  ++ 2.03 1.67 +++ 4.16 +++ 2.97 
K562 CML (erythroleukemia) − 1.00 − 1.00 ++ 2.17 ++ 2.33 +++ 6.70 1.70 1.42 − 1.00 − 1.00 
KU812 CML (basophilic cell) 1.51 1.87 1.85 1.33 ++ 3.02 2.12 1.97 1.68 +++ 3.45 
SAS413 CML − 0.84 ++ 1.46 1.36 1.79 1.52 2.23 2.23 − 1.00 − 1.00 
KML-1 B-cell lymphoma − 1.26 ++ 2.19 1.70 ++ 2.18 2.38 ++ 2.18 ++ 2.05 − 1.20 − 1.18 
OIH-1 Overt leukemia from MDS − 0.93 ++ 2.01 − 1.04 1.65 +++ 8.40 − 1.00 − 1.00 − 1.00 1.95 

NOTE: Expression of B7 family molecules on unstimulated and cytokine (IFN-γ or tumor necrosis factor-α)–stimulated cell lines was examined by flow cytometry. ALL, acute lymphocytic leukemia; CML, chronic myeloid leukemia; MDS, myelodysplastic syndromes; ND, not done.

*

Percentage of positive cells for B7 family molecules: −, <5%; +/−, 5-10%; +, 10-30%; ++, 30-60%; +++, >60%.

Relative MFI of B7 family molecules.

The expressions of classic B7 molecules, B7.1 and B7.2, on these cell lines were also analyzed by flow cytometry, and the results are shown in Table 1. B7.1 expression was detected in 7 of the 12 cell lines, and IFN-γ did not influence that expression. B7.2 expression was also detected in 7 of the 12 cell lines, and it was extremely high in KG-1, BALL-1, U937, and Dami cells. IFN-γ induced and enhanced B7.2 expression on OIH-1 and KU812 cells, respectively.

Based on these data, we decided to use flow cytometry to analyze the patient samples for the expression of these B7 family molecules.

Expression of B7 molecules on normal bone marrow mononuclear cells and leukemic cells from acute myeloid leukemia patients. It was reported that B7.1 and B7.2 were expressed on normal APCs (i.e., macrophages, dendritic cells, and B cells). In contrast, data regarding B7-H1 and B7-H2 expressions on normal hematopoietic cells are sparse. Thus, we investigated the B7-H1 and B7-H2 expressions on normal bone marrow mononuclear cells by flow cytometry. B7-H1 and B7-H2 expressions were observed on APCs, such as CD14+ and CD19+ cells in bone marrow mononuclear cells but not on CD34+ and CD33+ cells (data not shown).

Next, we analyzed the expression of B7-H1, B7-H2, B7.1, and B7.2 on leukemic cells obtained from the de novo AML patients. We confirmed that two kinds of flow cytometry data for these molecules (i.e., relative MFI and the percentage of positive cells) yielded consistent results (Table 1). Accordingly, in the following sections, we show only flow cytometry data for the percentage of positive cells. Neither B7.1 nor B7-H1 was expressed by a significant percentage of cells in most samples obtained from the first 36 patients. Leukemic cells from only one patient (M6 subtype) expressed B7.1 and B7-H1 at high percentages (35% and 19%, respectively). The B7.1 and B7-H1 expressions (positive percentages) in all other AML samples were <8% (Table 2). We did not analyze for these molecules in the samples from any other patients but analyzed for B7-H2 and B7.2 expressions in the samples from 59 and 58 patients, respectively.

Table 2.

Expression of B7 family molecules on leukemic cells from AML patients

B7 family moleculesTotal no. patientsPercentages of leukemic cells positive for each B7 family molecule
<5%5-10%10-30%30-60%>60%
B7.1 36 34 (94.4)* 1 (2.8) 1 (2.8) 
B7.2 58 17 (29.3) 4 (6.9) 14 (24.1) 14 (24.1) 9 (15.5) 
B7-H1 36 32 (88.9) 3 (8.3) 1 (2.8) 
B7-H2 59 17 (28.8) 8 (13.6) 25 (42.4) 8 (13.6) 1 (1.7) 
B7 family moleculesTotal no. patientsPercentages of leukemic cells positive for each B7 family molecule
<5%5-10%10-30%30-60%>60%
B7.1 36 34 (94.4)* 1 (2.8) 1 (2.8) 
B7.2 58 17 (29.3) 4 (6.9) 14 (24.1) 14 (24.1) 9 (15.5) 
B7-H1 36 32 (88.9) 3 (8.3) 1 (2.8) 
B7-H2 59 17 (28.8) 8 (13.6) 25 (42.4) 8 (13.6) 1 (1.7) 
*

Data are n (%) patients.

The percentage of AML cells expressing B7-H2 ranged from 0% to 68.0% in samples obtained from 59 patients (mean, 15.3%; Fig. 2, left; Table 2). Extremely high B7-H2 expression was detected in some AML patients (M1, M4, M5, and M6 cases). The percentage of cells expressing B7.2 ranged from 0% to 92.0% in samples obtained from 58 patients (mean, 28.9%; Fig. 2, right; Table 2). Normal monocytes express B7.2, and prior reports showed that leukemic cells having a monocytic component (M4 and M5 subtypes) expressed higher levels of B7.2 compared with other AML subtypes. In our cohort, significantly higher expression of B7.2 was seen in M4 and M5 compared with other subtypes (P < 0.001), but there was significant overlap in the B7.2 expression levels among the M1 to M5 subtypes.

Fig. 2.

B7-H2 and B7.2 expressions on leukemic cells from AML patients. Percentages of B7-H2- and B7.2-positive AML cells determined by flow cytometry for each patient. Bars of each AML subtype (M1-M6) are mean values.

Fig. 2.

B7-H2 and B7.2 expressions on leukemic cells from AML patients. Percentages of B7-H2- and B7.2-positive AML cells determined by flow cytometry for each patient. Bars of each AML subtype (M1-M6) are mean values.

Close modal

Function of B7-H2 and B7.2 molecules on acute myeloid leukemia cells. We examined whether B7-H2 and B7.2 molecules on AML cells are functional. First, the optimal doses of mAbs to block ICOS and B7.2 were examined in mixed lymphocyte-leukemic cell reaction using U937 cells and normal CD4+ T cells. In 7-day coculture of these cells, either an anti-ICOS mAb (blockade of the pathway of B7-H2-ICOS) or an anti-B7.2 mAb decreased the proliferation of CD4+ T cells. The optimal dose of each mAb was 5 μg/mL (Fig. 3A). Similarly, when allogenetic CD4+ T cells were cocultured with patients' AML cells expressing both B7-H2 and B7.2 molecules, the anti-ICOS mAb and/or the anti-B7.2 mAb decreased CD4+ T-cell proliferation (Fig. 3B).

Fig. 3.

Allogeneic mixed lymphocyte-leukemic cell reaction. A, normal CD4+ T cells were cultured with mitomycin C–treated U937 cells for 7 days in the presence of control IgG, anti-ICOS mAb, or anti-B7.2 mAb at several doses (1-10 μg/mL), and the CD4+ T-cell proliferation was assessed by [3H]thymidine incorporation. Columns, mean of triplicate cultures; bars, SD. Medium, no antibody was added. B, mitomycin C–treated AML cells from four patients, which expressed B7-H2 and B7.2, were used instead of U937 cells under the same culture conditions as (A). Columns, mean of triplicate cultures of cells from one patient; bars, SD. Similar data were obtained for three other cases. When CD4+ T cells alone or mitomycin C–treated AML cells alone were cultured under the same culture conditions with no antibody, [3H]thymidine incorporation was minimal. C and D, concentrations of cytokines (IFN-γ, IL-2, IL-4, and IL-10) in the culture supernatants of the mixed lymphocyte-leukemic cell reaction, in which mitomycin C–treated U937 cells (C) or AML cells from five patients (D) were cultured similar to (A and B), were measured by ELISA. Columns, mean of triplicate cultures; bars, SD (invisible bars are due to small SD values). The lower limits of detection for IFN-γ, IL-2, and IL-4/IL-10 were 2 pg/mL, 0.1 units/mL, and 1 pg/mL, respectively. ND, not detected.

Fig. 3.

Allogeneic mixed lymphocyte-leukemic cell reaction. A, normal CD4+ T cells were cultured with mitomycin C–treated U937 cells for 7 days in the presence of control IgG, anti-ICOS mAb, or anti-B7.2 mAb at several doses (1-10 μg/mL), and the CD4+ T-cell proliferation was assessed by [3H]thymidine incorporation. Columns, mean of triplicate cultures; bars, SD. Medium, no antibody was added. B, mitomycin C–treated AML cells from four patients, which expressed B7-H2 and B7.2, were used instead of U937 cells under the same culture conditions as (A). Columns, mean of triplicate cultures of cells from one patient; bars, SD. Similar data were obtained for three other cases. When CD4+ T cells alone or mitomycin C–treated AML cells alone were cultured under the same culture conditions with no antibody, [3H]thymidine incorporation was minimal. C and D, concentrations of cytokines (IFN-γ, IL-2, IL-4, and IL-10) in the culture supernatants of the mixed lymphocyte-leukemic cell reaction, in which mitomycin C–treated U937 cells (C) or AML cells from five patients (D) were cultured similar to (A and B), were measured by ELISA. Columns, mean of triplicate cultures; bars, SD (invisible bars are due to small SD values). The lower limits of detection for IFN-γ, IL-2, and IL-4/IL-10 were 2 pg/mL, 0.1 units/mL, and 1 pg/mL, respectively. ND, not detected.

Close modal

In mixed lymphocyte-leukemic cell reaction using U937 cells and CD4+ T cells, both mAbs decreased the production of IFN-γ, IL-4, and IL-10 (Fig. 3C). When leukemic cells from the above AML patients (n = 5) were used instead of U937 cells, the mAb against ICOS decreased the production of IL-4 and IL-10 but had little effect on IFN-γ production. Meanwhile, the mAb against B7.2 decreased the production of each of IFN-γ, IL-4, and IL-10 (Fig. 3D). These results showed that B7-H2 and B7.2 were functional and may modulate the immunologic functions of hosts. IL-2 production was not reduced by the anti-ICOS mAb and reduced only a little by the anti-B7.2 mAb, which are consistent with the reported findings that B7-H2 did not contribute to IL-2 induction, whereas B7.2 did (14).

Prognostic significance of B7-H2 and B7.2 expressions. To examine whether B7-H2 and B7.2 expressions influence the clinical outcome, the survival of AML patients was examined as a function of these molecules' expressions. When AML patients were divided into two groups using various cutoff percentages of B7-H2 positivity, cases in whom >25% of leukemic cells expressed B7-H2 (n = 10, called B7-H2-positive AML in this article) showed significantly shorter overall survival compared with B7-H2-negative AML cases (n = 49; Fig. 4A, top). Essentially, the same results were obtained when M3 cases were excluded from the analysis (P = 0.0212; Fig. 4A, bottom) and when only cases with intermediate karyotypes were analyzed (P = 0.0498). Similarly, when AML patients were divided into two groups using various cutoff percentages of B7.2 positivity, cases in whom >30% of leukemic cells expressed B7.2 (n = 23, called B7.2-positive AML in this article) showed significantly shorter overall survival compared with B7.2-negative AML cases (n = 35; Fig. 4B, top). Even when M3 cases were excluded from this analysis and when only cases with intermediate karyotypes were analyzed, B7.2-positive AML cases showed a shorter overall survival compared with B7.2-negative AML cases without and with statistical significance, respectively [P = 0.0967 (Fig. 4B, bottom) and P = 0.0191]. Only four patients were positive for both B7.2 and B7-H2 molecules, and we thus could not analyze the impact of coexpression of these molecules on the prognosis; however, there was no difference in the overall survival between these four patients and other patients. When patients negative for both B7.2 and B7-H2 molecules (n = 28) were compared with other patients, the former patients had a better overall survival (P = 0.0362). Regarding disease-free survival, the B7-H2- and B7.2-positive patient groups had a shorter disease-free survival compared with the B7-H2- and B7.2-negative patient groups, respectively. However, the differences in these results were not statistically significant.

Fig. 4.

Overall survival of AML patients as a function of B7-H2 (A) and B7.2 (B) expressions. Top, all AML cases; bottom, AML cases excluding M3.

Fig. 4.

Overall survival of AML patients as a function of B7-H2 (A) and B7.2 (B) expressions. Top, all AML cases; bottom, AML cases excluding M3.

Close modal

Univariate analysis of other potential prognostic factors in the present AML cohort showed that age, leukocyte count, and cytogenetic characteristics were prognostic variables having statistical or marginal statistical significance for overall survival (Table 3). Therefore, we analyzed these variables and B7.2 and B7-H2 expressions by multivariate analysis and found that B7-H2 positivity was the strongest independent prognostic factor (P = 0.0108; Table 3). A previous study reported that B7.2 expression was an independent prognostic factor in AML (21), but this was not the case in our cohort.

Table 3.

Univariate and multivariate analyses for variables associated with poor overall survival in AML

VariableUnivariate analysis (P)Multivariate analysis (P)
B7-H2 expression (>25%) 0.0188 0.0108 —* 
B7-2 expression (>30%) 0.0199 —* 0.3895 
Cytogenetics 0.0005 0.0298 0.0807 
Leukocyte count (>1011/L) 0.0912 0.2509 0.4226 
Age (>60 y) 0.0048 0.1016 0.0462 
VariableUnivariate analysis (P)Multivariate analysis (P)
B7-H2 expression (>25%) 0.0188 0.0108 —* 
B7-2 expression (>30%) 0.0199 —* 0.3895 
Cytogenetics 0.0005 0.0298 0.0807 
Leukocyte count (>1011/L) 0.0912 0.2509 0.4226 
Age (>60 y) 0.0048 0.1016 0.0462 
*

Not included in the analysis.

Prognosis was the worst in the unfavorable cytogenetic group and the best in the favorable cytogenetic group.

Comparison of characteristics of acute myeloid leukemia patients as a function of B7-H2 and B7.2 expressions. We compared the differences in clinical and hematologic characteristics (age, sex, leukocyte count, hemoglobin value, platelet count, and cytogenetic risk group at diagnosis) between B7-H2-positive and B7-H2-negative AML patients and between B7.2-positive and B7.2-negative AML patients. There was no difference in these characteristics between the B7-H2-positive and B7-H2-negative AML patients. Meanwhile, the B7.2-positive AML patients had a significantly higher leukocyte count compared with the B7.2-negative AML patients (36.7 × 109/L versus 14.8 × 109/L; P = 0.026; Table 4). These results suggest that B7.2 on AML cells may be related to the increase in leukemic cells in the hosts.

Table 4.

Differences in characteristics of AML patients as a function of B7-H2 or B7.2 expression

B7-H2+B7-H2B7.2+B7.2
No. patients 10 49 23 35 
Male/female 7/3 36/13 14/9 28/7 
Age (y) 57 (43-71) 63 (23-82) 58 (23-82) 57 (33-73) 
Leukocytes (109/L) 29.8 (1.4-181.8) 22.0 (0.9-582.3) 36.7 (2.3-582.3)* 14.8 (0.9-234.6) 
Hemoglobin (g/dL) 7.8 (5.0-11.6) 7.9 (3.7-13.0) 8.0 (5.1-11.6) 8.0 (3.7-13.0) 
Platelets (109/L) 7.1 (2.1-19.0) 5.5 (0.4-23.1) 6.5 (1.4-23.1) 5.3 (0.4-19.0) 
Karyotype     
    Favorable 
    Intermediate 30 17 20 
    Unfavorable 
    Not applicable 
CR rate (%) 60.0 77.6 78.3 71.4 
B7-H2+B7-H2B7.2+B7.2
No. patients 10 49 23 35 
Male/female 7/3 36/13 14/9 28/7 
Age (y) 57 (43-71) 63 (23-82) 58 (23-82) 57 (33-73) 
Leukocytes (109/L) 29.8 (1.4-181.8) 22.0 (0.9-582.3) 36.7 (2.3-582.3)* 14.8 (0.9-234.6) 
Hemoglobin (g/dL) 7.8 (5.0-11.6) 7.9 (3.7-13.0) 8.0 (5.1-11.6) 8.0 (3.7-13.0) 
Platelets (109/L) 7.1 (2.1-19.0) 5.5 (0.4-23.1) 6.5 (1.4-23.1) 5.3 (0.4-19.0) 
Karyotype     
    Favorable 
    Intermediate 30 17 20 
    Unfavorable 
    Not applicable 
CR rate (%) 60.0 77.6 78.3 71.4 

NOTE: Data are n patients or median (range).

*

A significant difference was observed between B7.2-positive and B7.2-negative patients (P = 0.026).

We also investigated whether the expressions of B7-H2 and B7.2 molecules were associated with the individual cytogenetic data, such as a −7, instead of cytogenetic categories (Table 5; Supplementary Table S5). However, there was no significant association between them.

Table 5.

Cytogenetic characteristics and expression of B7-H2 and B7.2 molecules in B7-H2- and B7.2-positive patients

Cytogenetic categoryB7-H2-positive patients (n = 10)
B7.2-positive patients (n = 23)
KaryotypeNo. patients*B7-H2 expression (%)KaryotypeNo. patients*B7.2 expression (%)
Favorable 48, XY, +14, inv(16)(p13;q22), +21 41 46, XY, t(15;17)(q22;q11-12) 34 
Intermediate Normal 46 (range, 26-68) Normal 16 58 (range, 31-84) 
 45, X, −Y 51 46, XX, del(8)(q24), add(22)(q13) 58 
 47, XX, +6 45    
Unfavorable 45, XY, −7 33 46, XX, 11q− 89 
 44, XY, 5q−, −7, 9p+, 9q+, 13q+, −16, −17, 19p+, +mar 35 46, XX, 5q−, 19p− 46 
    46, XX, t(1;11)(p32;q23) 88 
    48, XY, +4, +8, 9q−, 15p+ 92 
    46, XY, t(9;11)(q22;q23) 47 
Cytogenetic categoryB7-H2-positive patients (n = 10)
B7.2-positive patients (n = 23)
KaryotypeNo. patients*B7-H2 expression (%)KaryotypeNo. patients*B7.2 expression (%)
Favorable 48, XY, +14, inv(16)(p13;q22), +21 41 46, XY, t(15;17)(q22;q11-12) 34 
Intermediate Normal 46 (range, 26-68) Normal 16 58 (range, 31-84) 
 45, X, −Y 51 46, XX, del(8)(q24), add(22)(q13) 58 
 47, XX, +6 45    
Unfavorable 45, XY, −7 33 46, XX, 11q− 89 
 44, XY, 5q−, −7, 9p+, 9q+, 13q+, −16, −17, 19p+, +mar 35 46, XX, 5q−, 19p− 46 
    46, XX, t(1;11)(p32;q23) 88 
    48, XY, +4, +8, 9q−, 15p+ 92 
    46, XY, t(9;11)(q22;q23) 47 

NOTE: B7-H2 and B7.2 expressions are percentages of positive cells determined by flow cytometry; for patients with normal karyotype, the percentages indicate median and range.

*

Each abnormal karyotype was observed in 1 patient and normal karyotype was observed in 5 B7-H2-positive and 16 B7.2-positive patients.

Antitumor T-cell–mediated specific immune responses require well-organized, multiple steps of molecular interaction. They include expression of MHC molecules on APCs (professional APCs and/or tumor cells), presentation of tumor-specific peptides in optimal amounts (density) by MHC molecules, optimal costimulation signals between APCs and T cells, intact T-cell receptor–associated signal mechanisms, presence of an appropriate cytokine milieu, and probably many other still unelucidated factors (30). When one of these steps becomes deranged (e.g., the density of peptides presented by APCs is low or excess IL-10 is present), the immune response against tumor cells may become insufficient. In patients who have developed a malignancy, the malignant cells have evaded antitumor immunity using mechanisms that are probably diverse even among patients with the same malignant disease.

In this study, we showed that leukemic cells from a substantial number of AML patients expressed B7.2 and B7-H2 molecules. We also showed that these molecules on AML cells from the examined patients had immunomodulatory functions in vitro. Moreover, the expression of B7-H2 molecules was associated with a poor prognosis even when analyzed by multivariate analysis, and B7.2 was associated with hyperleukocytosis in the patients. These data suggest that expression of these molecules might contribute to the proliferation of AML cells by helping them evade antitumor immune responses. If so, what form might the contributions of B7.2 and B7-H2 molecules take? Yang et al. (31) reported that B7.2 gene-transduced P815 mastocytoma cells regressed, whereas wild-type P815 cells grew lethally in mice. In contrast, Stremmel et al. (3) reported that B7.2-expressing EL4 lymphoma cells grew faster than B7.2-negative EL4 lymphoma cells in mice. They also showed that this proliferative advantage of B7.2-expressing EL4 lymphoma cells was not observed in IL-4 knockout mice. Therefore, a Th2 cytokine, IL-4, probably contributed to the proliferative advantage of B7.2-expressing EL4 lymphoma cells. Further, many studies showed that B7.2 costimulation resulted in a shift to Th2 cytokine production (3235). It was also reported that B7.2 costimulation induced IL-10 production (35), which is a potent inhibitor of antitumor immune responses. Therefore, B7.2-expressing tumors might induce a tumor-specific Th2 response and IL-10 production and thus inhibit antitumor immune responses. B7-H2-ICOS costimulation has been reported to up-regulate secretions of IL-10 and both Th1 and Th2 cytokines (18, 3638). When the ICOS-B7-H2 signal was blocked in experimental animals, which had autoimmune diseases or acute graft-versus-host disease, exacerbation of those diseases occurred (19, 39, 40). Therefore, the B7-H2-ICOS signal is important in controlling excessive immune responses. Our present results also suggest that the expression of B7.2 or B7-H2 on leukemic cells enhanced the production of IL-4 (Th2 cytokine) and IL-10. These cytokines induced by B7.2- or B7-H2-expressing AML cells may inhibit tumor-specific Th1 cell differentiation and CTL activity, resulting in creation of a favorable environment for growth of leukemic cells.

If the expressions of B7.2 and B7-H2 molecules are associated with specific cytogenetic data, said expressions might be able to be used to predict the cytogenetics. However, we could not detect such an association in this study. Recently, it was reported that FLT3-internal tandem duplication was strongly associated with a higher white blood count at diagnosis in some leukemias (41). Therefore, possible association between FLT3-internal tandem duplication and B7.2 expression, which we could not examine in this study, is interesting and should be elucidated in future studies.

In summary, this is the first report to show expression of functional B7 family molecules on leukemic cells from AML patients. Our data suggest that the expression of B7.2 and B7-H2 molecules is associated with the proliferative advantage of AML cells and represents the contributions of these molecules to the pathogenesis of AML. In general, the normal immune response is tightly regulated in space and time to avoid damage to the normal host cells. In leukemia patients, leukemic cells, which have evaded the immune responses and proliferated to a very large number, circulate throughout the body via the blood. AML cells, which constitutively express B7.2 and B7-H2 molecules, thus might derange the normal immune functions of the hosts.

Grant support: Japanese Ministry of Health, Labour, and Welfare grant 14770542 and Maruyama Memorial Fund of Nippon Medical School.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1
Chen L, Linsley PS, Hellstrom KE. Costimulation of T cells for tumor immunity.
Immunol Today
1993
;
14
:
483
–6.
2
Chen L, McGowan P, Ashe S, et al. Tumor immunogenicity determines the effect of B7 costimulation on T cell-mediated tumor immunity.
J Exp Med
1994
;
179
:
523
–32.
3
Stremmel C, Greenfield EA, Howard E, Freeman GJ, Kuchroo VK. B7-2 expressed on EL4 lymphoma suppresses antitumor immunity by an interleukin 4-dependent mechanism.
J Exp Med
1999
;
189
:
919
–30.
4
Dong H, Zhu G, Tamada K, Chen L. B7-H1, a third member of the B7 family, co-stimulates T-cell proliferation and interleukin-10 secretion.
Nat Med
1999
;
5
:
1365
–9.
5
Tamura H, Dong H, Zhu G, et al. B7-H1 costimulation preferentially enhances CD28-independent T-helper cell function.
Blood
2001
;
97
:
1809
–16.
6
Chapoval AI, Ni J, Lau JS, et al. B7-H3: a costimulatory molecule for T cell activation and IFN-γ production.
Nat Immunol
2001
;
2
:
269
–74.
7
Sica GL, Choi IH, Zhu G, et al. B7-H4, a molecule of the B7 family, negatively regulates T cell immunity.
Immunity
2003
;
18
:
849
–61.
8
Freeman GJ, Long AJ, Iwai Y, et al. Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member leads to negative regulation of lymphocyte activation.
J Exp Med
2000
;
192
:
1027
–34.
9
Wang S, Bajorath J, Flies DB, Dong H, Honjo T, Chen L. Molecular modeling and functional mapping of B7-H1 and B7-DC uncouple costimulatory function from PD-1 interaction.
J Exp Med
2003
;
197
:
1083
–91.
10
Dong H, Strome SE, Salomao DR, et al. Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion.
Nat Med
2002
;
8
:
793
–800.
11
Strome SE, Dong H, Tamura H, et al. B7-H1 blockade augments adoptive T-cell immunotherapy for squamous cell carcinoma.
Cancer Res
2003
;
63
:
6501
–5.
12
Hirano F, Kaneko K, Tamura H, et al. Blockade of B7-H1 and PD-1 by monoclonal antibodies potentiates cancer therapeutic immunity.
Cancer Res
2005
;
65
:
1089
–96.
13
Yoshinaga SK, Whoriskey JS, Khare SD, et al. T-cell co-stimulation through B7RP-1 and ICOS.
Nature
1999
;
402
:
827
–32.
14
Wang S, Zhu G, Chapoval AI, et al. Costimulation of T cells by B7-H2, a B7-like molecule that binds ICOS.
Blood
2000
;
96
:
2808
–13.
15
Schreiner B, Wischhusen J, Mitsdoerffer M, et al. Expression of the B7-related molecule ICOSL by human glioma cells in vitro and in vivo.
Glia
2003
;
44
:
296
–301.
16
Chen XL, Cao XD, Kang AJ, Wang KM, Su BS, Wang YL. In situ expression and significance of B7 costimulatory molecules within tissues of human gastric carcinoma.
World J Gastroenterol
2003
;
9
:
1370
–3.
17
Hutloff A, Dittrich AM, Beier KC, et al. ICOS is an inducible T-cell co-stimulator structurally and functionally related to CD28.
Nature
1999
;
397
:
263
–6.
18
Coyle AJ, Lehar S, Lloyd C, et al. The CD28-related molecule ICOS is required for effective T cell-dependent immune responses.
Immunity
2000
;
13
:
95
–105.
19
Ogawa S, Nagamatsu G, Watanabe M, et al. Opposing effects of anti-activation-inducible lymphocyte-immunomodulatory molecule/inducible costimulator antibody on the development of acute versus chronic graft-versus-host disease.
J Immunol
2001
;
167
:
5741
–8.
20
Liu X, Bai XF, Wen J, et al. B7H costimulates clonal expansion of, and cognate destruction of tumor cells by, CD8(+) T lymphocytes in vivo.
J Exp Med
2001
;
194
:
1339
–48.
21
Maeda A, Yamamoto K, Yamashita K, et al. The expression of co-stimulatory molecules and their relationship to the prognosis of human acute myeloid leukaemia: poor prognosis of B7-2-positive leukaemia.
Br J Haematol
1998
;
102
:
1257
–62.
22
Costello RT, Mallet F, Sainty D, Maraninchi D, Gastaut JA, Olive D. Regulation of CD80/B7-1 and CD86/B7-2 molecule expression in human primary acute myeloid leukemia and their role in allogenic immune recognition.
Eur J Immunol
1998
;
28
:
90
–103.
23
Whiteway A, Corbett T, Anderson R, Macdonald I, Prentice HG. Expression of co-stimulatory molecules on acute myeloid leukaemia blasts may effect duration of first remission.
Br J Haematol
2003
;
120
:
442
–51.
24
Dastugue N, Payen C, Lafage-Pochitaloff M, et al. Prognostic significance of karyotype in de novo adult acute myeloid leukemia. The BGMT group.
Leukemia
1995
;
9
:
1491
–8.
25
Grimwade D, Walker H, Oliver F, et al. The importance of diagnostic cytogenetics on outcome in AML: analysis of 1,612 patients entered into the MRC AML 10 trial. The Medical Research Council Adult and Children's Leukaemia Working Parties.
Blood
1998
;
92
:
2322
–33.
26
Slovak ML, Kopecky KJ, Cassileth PA, et al. Karyotypic analysis predicts outcome of preremission and postremission therapy in adult acute myeloid leukemia: a Southwest Oncology Group/Eastern Cooperative Oncology Group Study.
Blood
2000
;
96
:
4075
–83.
27
Dong H, Strome SE, Matteson EL, et al. Costimulating aberrant T cell responses by B7-H1 autoantibodies in rheumatoid arthritis.
J Clin Invest
2003
;
111
:
363
–70.
28
Wang S, Zhu G, Tamada K, Chen L, Bajorath J. Ligand binding sites of inducible costimulator and high avidity mutants with improved function.
J Exp Med
2002
;
195
:
1033
–41.
29
Tamura H, Ogata K, Dong H, Chen L. Immunology of B7-H1 and its roles in human diseases.
Int J Hematol
2003
;
78
:
321
–8.
30
Appleman LJ, Boussiotis VA. T cell anergy and costimulation.
Immunol Rev
2003
;
192
:
161
–80.
31
Yang G, Hellstrom KE, Hellstrom I, Chen L. Antitumor immunity elicited by tumor cells transfected with B7-2, a second ligand for CD28/CTLA-4 costimulatory molecules.
J Immunol
1995
;
154
:
2794
–800.
32
Kuchroo VK, Das MP, Brown JA, et al. B7-1 and B7-2 costimulatory molecules activate differentially the Th1/Th2 developmental pathways: application to autoimmune disease therapy.
Cell
1995
;
80
:
707
–18.
33
Thompson CB. Distinct roles for the costimulatory ligands B7-1 and B7-2 in T helper cell differentiation?
Cell
1995
;
81
:
979
–82.
34
Brown JA, Titus RG, Nabavi N, Glimcher LH. Blockade of CD86 ameliorates Leishmania major infection by down-regulating the Th2 response.
J Infect Dis
1996
;
174
:
1303
–8.
35
Nakajima A, Watanabe N, Yoshino S, Yagita H, Okumura K, Azuma M. Requirement of CD28-CD86 co-stimulation in the interaction between antigen-primed T helper type 2 and B cells.
Int Immunol
1997
;
9
:
637
–44.
36
McAdam AJ, Chang TT, Lumelsky AE, et al. Mouse inducible costimulatory molecule (ICOS) expression is enhanced by CD28 costimulation and regulates differentiation of CD4+ T cells.
J Immunol
2000
;
165
:
5035
–40.
37
Dong C, Juedes AE, Temann UA, et al. ICOS co-stimulatory receptor is essential for T-cell activation and function.
Nature
2001
;
409
:
97
–101.
38
Sperling AI, Bluestone JA. ICOS costimulation: it's not just for TH2 cells anymore.
Nat Immunol
2001
;
2
:
573
–4.
39
Rottman JB, Smith T, Tonra JR, et al. The costimulatory molecule ICOS plays an important role in the immunopathogenesis of EAE.
Nat Immunol
2001
;
2
:
605
–11.
40
Sporici RA, Beswick RL, von Allmen C, et al. ICOS ligand costimulation is required for T-cell encephalitogenicity.
Clin Immunol
2001
;
100
:
277
–88.
41
Kottaridis PD, Gale RE, Frew ME, et al. The presence of a FLT3 internal tandem duplication in patients with acute myeloid leukemia (AML) adds important prognostic information to cytogenetic risk group and response to the first cycle of chemotherapy: analysis of 854 patients from the United Kingdom Medical Research Council AML 10 and 12 trials.
Blood
2001
;
98
:
1752
–9.