Circulating levels and clinical significance of soluble CD40 in patients with hematologic malignancies

Authors


Abstract

BACKGROUND

CD40 plays a critical role in immunoregulation, and CD40 ligation is being investigated as a therapy for hematologic malignancies. Although soluble CD40 (sCD40) is a potential modulator of both antitumor responses and CD40-based therapies, the levels and significance of sCD40 in patients with hematologic malignancies are unknown.

METHODS

The authors evaluated serum/plasma sCD40 levels using an enzyme-linked immunoassay in patients with acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), chronic lymphocytic leukemia (CLL), mantle cell lymphoma (MCL), and multiple myeloma (MM).

RESULTS

Levels of sCD40 were elevated in serum (>1.697 ng/mL) or plasma (>0.649 ng/mL) from 73% of patients with CLL, 80% of patients with MCL, 40% of patients with AML, 43% of patients with MDS, and 33% of patients with MM. Multivariate analysis of patients with MM demonstrated that elevated sCD40 was a significant, independent predictor of poor survival. In multivariate analysis of patients with AML, sCD40 was a significant prognostic factor when the interaction of age and sCD40 was included as a variable. Further analysis demonstrated that elevated sCD86 levels were associated with significantly shorter survival only in AML patients younger than age 64 years. Release of sCD40 by CLL cells was induced by cross-linking with CD40 monoclonal antibody.

CONCLUSIONS

Many patients with hematologic malignancies have elevated circulating levels of sCD40, and these elevated levels are associated with a poor prognosis at least in patients with MM and AML, suggesting that sCD40 may have a role in modulating antitumor responses and also may be a useful prognostic marker. In addition, the findings suggested that further studies will be required to determine the effect of circulating sCD40 on the clinical effectiveness of CD40-ligating reagents used in the treatment of hematologic malignancies. Cancer 2006. © 2006 American Cancer Society.

The signals delivered after membrane CD40 (mCD40) ligation play a central role in regulating immune responses, both because of the wide expression of mCD40 and because of the diverse effects its signals have on cell growth, activation, and survival.1, 2 With respect to hemopoietic cells, ligation of mCD40 modulates not only B-cell development and function but also increases monocyte tumoricidal activity, dendritic cell (DC) maturation, leukocyte survival, and cytokine release by antigen-presenting cells (APCs) and platelets.1–3 Many hemopoietic malignancies also express mCD404–7 and undergo a range of cellular changes in response to CD40 ligation. Depending on the cell type, these changes can either increase or decrease malignant cell growth, survival, and/or immunogenicity.2, 4, 8

The importance of mCD40 in tumor immunity is well established in both in vitro studies and animal tumor models.4, 8–10 Blockade of mCD40 signaling inhibits in vivo antitumor responses, whereas the provision of increased mCD40 ligation augments responses. It is believed that these effects result predominantly from the stimulatory effect of CD40 ligation on APC maturation, although it also is believed that direct effects on effector cells and tumor cell immunogenecity and/or survival play a role.4, 8, 10, 11 Consequently, there is now considerable interest in the clinical use of CD40-ligating agents, such as CD40 ligand (CD40L) trimer (CD40LT), CD40 monoclonal antibody (mAb), and CD40L-positive cells, as immunotherapeutic agents, particularly in the treatment of hematologic malignancies.4, 8 Initial clinical studies have provided promising results,4, 12, 13 but it is clear that the regulation and role of mCD40 signaling in antitumor responses still are not understood fully and that further optimization of these approaches is needed.

The release of soluble forms of membrane molecules provides an important mechanism by which cells can either enhance or inhibit the signals delivered by their respective membrane-bound counterparts. Studies that used recombinant forms of soluble CD40 (sCD40) demonstrated that the presence of these molecules can inhibit immune responses both in vivo and in vitro, suggesting that in vivo release of functional sCD40 would be immunomodulatory.14, 15 However, studies on in vivo-generated forms of sCD40 have been limited to date. Low levels of circulating sCD40 have been detected in the blood of normal donors, and elevated levels of sCD40 have been reported in patients with renal failure, arthritis, and lung cancer.16–18 Although it has been demonstrated that activated B cells release sCD40 in vitro, the ability of other cell types to release sCD40 and other sources of in vivo sCD40 has not been established.17, 19, 20 In vitro studies that used the sCD40 released from B cells have demonstrated that, like recombinant sCD40, these forms of sCD40 also are inhibitory, at least with respect to B-cell function.17

The release of sCD40 by the immune system and/or by malignant cells provides a potentially powerful mechanism for regulating antitumor responses by modulating the interaction of mCD40 with its ligands. In addition, the presence of sCD40 may modulate the effectiveness of mCD40-directed immunotherapy agents. Monitoring alterations in sCD40 levels also may provide useful prognostic information. However, currently, little is known about the relative levels of sCD40 in patients with hematologic malignancies. In the current study, we analyzed the circulating levels and clinical significance of sCD40 in patients with hematologic malignancies to provide information relevant to understanding the biologic role of sCD40.

MATERIALS AND METHODS

Patients and Specimens

Normal donor, chronic lymphocytic leukemia, and mantle cell lymphoma samples

Normal blood samples were obtained from healthy individuals who voluntarily were donating blood to the New Zealand Blood Transfusion Service (NZBTS) and consequently had undergone comprehensive screening for the presence of infectious or chronic disease. The individual details of these donors were not available, but donors to the NZBTS range in age between 16 years and 60 years and have a male:female ratio of 52:48. Blood was obtained from patients with chronic lymphocytic leukemia (CLL) and mantel cell lymphoma (MCL) who attended the Haematology Department, Christchurch Hospital, New Zealand. Patients with CLL and MCL were discriminated on the basis of CD23 expression and fluorescent in situ hybridization analysis for the t(11;14) translocation. Blood samples from normal donors and from patients with CLL and MCL were collected with appropriate informed consent, according to Canterbury (New Zealand) Ethical Committee guidelines, into either ethylenediamine tetraacetic acid (EDTA) or plain tubes. Blood samples that were collected into plain tubes were allowed to clot at room temperature and were stored at 4°C prior to recovery of the serum fraction. The EDTA-treated blood samples were centrifuged at low speed (× 300 g for 5 minutes), and the plasma fraction was recovered. The plasma and serum fractions were then centrifuged again (× 10,000 g for 30 minutes), and the nonsedimented material was stored frozen (− 70°C) prior to analysis.

Acute myeloid leukemia and myelodysplastic syndrome samples

Plasma samples were obtained from patients with newly diagnosed acute myeloid leukemia (AML) (n = 62 patients) and newly diagnosed myelodysplastic syndrome (MDS) (n = 21 patients) at The University of Texas M. D. Anderson Cancer Center (Houston, TX) after informed consent was obtained according to institutional guidelines. Cytogenetic characteristics were classified, as described previously,21 into good (t[8;21] or inv16), intermediate (diploid), or bad (−5, −7 11q23, or +8). All patients were treated on frontline AML-type chemotherapy clinical research protocols (cytarabine/idarubicin or cytarabine/topotecan-based chemotherapy combinations) and were followed regularly in outpatient clinics. Blood samples were collected 1 or 2 days prior to commencement of chemotherapy; and, after separation, the plasma was stored at − 70°C.

Myeloma samples

Serum samples were collected from patients with myeloma who were participating in the United Kingdom Medical Research Council sixth myelomatosis trial. Patients in the trial were randomized to 1 of 2 treatment regimens: Treatment Group A received induction chemotherapy with doxorubicin, carmustine, melphalan, and cyclophosphamide (ABCM) repeated every 6 weeks; and Treatment Group B received induction chemotherapy with ABCM repeated every 6 weeks plus intermediate-dose prednisilone with the first cycle. Samples from 38 patients in Treatment Group A and from 38 patients in Treatment Group B were shipped on dry ice to Christchurch, New Zealand for analysis.

Reagents

The mAb 89 (CD40; immunoglobulin G1 [IgG1]) was obtained from Coulter Immunotech (Marseilles, France), and polyclonal goat anti-CD40 (GA40) was obtained from R&D systems (Minneapolis, MN). The mAb G28-5 (CD40; IgG1) was prepared by protein A purification of ascites that were obtained using the hybridoma from the American Type Culture Collection (Manassas, VA). The antibody MOPC21 (mouse IgG1) and purified goat Ig were obtained from Sigma (St. Louis, MO). Biotinylated rabbit antigoat Ig (bio-RAG) was obtained from Dako (Glostrup, Denmark), and cross-reactivity with mouse Ig was removed by passage over mouse IgG agarose (Sigma). Peroxidase-conjugated goat antihuman Ig (pGAH), streptavidin-horseradish peroxidase (HRP), and a tetramethylbenzidine (TMB) 1-step substrate system were obtained from Dako.

Expression and Purification of Recombinant CD40-Ig Fusion Protein

CD40-Ig was constructed by fusing the extracellular domain of CD40 with the Fc part of human IgG1. Combinational DNA (cDNA) was prepared from the B-cell line Raji, as described previously.22 The extracellular domain of CD40 was amplified from Raji cDNA by using Taq polymerase with a pair of CD40-specific primers: 5′-CTATAAGCTTATGGTTCGTCTGCCTCTG-3′ (at nucleotides 48-66) and 5′-CTATGGATCCACTTACCTGTTCTCAGCCGATCCTGGGG-3′ (at nucleotides 609-626) containing a HindIII and BamHI site (underlined), respectively, and a splicing donor sequence (in italics). After digestion with HindIII and BamHI the polymerase chain reaction product was purified by gel extraction (QIAgen, Clifton Hill, Victoria, Australia), cloned into the pIg1 vector23 (a gift from L. Williams, Oxford, United Kingdom), and transformed into MC1061/p3-competent bacteria. The resulting plasmid was purified (Wizard Midiprep DNA purification system; Promega, Madison, WI), then transfected into COS-7 cells (Lipofectamine 2000; Life Technologies), and the cells were cultured for 3 days in Dulbecco Modified Eagle Medium (Life Technologies) containing 10% fetal calf serum. The secreted CD40-Ig was purified by protein A chromatography, and reactivity with both CD40 mAb and pGAH was confirmed by enzyme-linked immunosorbent assay (ELISA). The concentration of purified CD40-Ig was determined by ELISA using human IgG1 as the standard, as described previously.22

Isolation and Activation of CLL Peripheral Blood Mononuclear Cells

Human peripheral blood mononuclear cells (PBMCs) were prepared from the blood (EDTA) of patients with CLL who had white blood cell (WBC) counts >80 × 109/L. PBMCs were isolated by centrifugation over Ficoll/Paque (Amersham Pharmacia Biotech, Uppsala, Sweden), then cryopreserved in 10% dimethyl sulfoxide. Immediately prior to use, CLL PBMCs were thawed; and, after washing 3 times, they were resuspended (4 × 106/mL) in media (X-VIVO 15; Bio Whittaker, MD).

Activation studies were performed in 96-well, U-bottomed tissue culture plates (Nunc). Wells were precoated (for 2 hours at 37°C) with 100 μL of either phosphate-buffered saline (PBS) alone, CD40 mAb (G28-5) at 5 μg/mL in PBS, or control mAb (MOPC21) at 5 μg/mL in PBS. After PBS washes, 100 μL of either media alone (X-VIVO 15) or media supplemented with 800 U/mL interleukin 4 (IL-4) (R&D Systems) were added to each well. Then, CLL PBMCs (100 μL at 4 × 106/mL) were added, and the plates were cultured (for 72 hours at 37°C in a 5% CO2 atmosphere) prior to centrifugation and recovery of the supernatants. Supernatants were centrifuged again (×10,000 g, 30 minutes) prior to storage at − 20°C.

sCD40 ELISA

The sCD40 ELISA was performed in 96-well microtiter plates (Maxisorp, Nunc). Unless stated otherwise, all incubations were carried out at 37°C in a 5% CO2 incubator, because we found that this increased ELISA sensitivity.

ELISA plates were precoated (2 hours) with CD40 mAb (mAb89) or isotype control antibody (MOPC21) at 2.5 μg/mL. The CD40 mAb was prepared in PBS, and the control mAb was prepared in bovine serum albumin (BSA)/PBS (12.5 μg/mL), so that the final concentration of BSA present in both mAb coating solutions was 12.5 μg/mL. Nonspecific binding was then blocked by overnight incubation (at 4°C) with 2% BSA/PBS. Samples (100 μL) that were made up in sample diluent (see below) were then applied. After incubation for 2 hours, wells were incubated for 1 hour with GA40 at 1 μg/mL in diluent (0.4% BSA/PBS) prior to incubation for 1 hour with bio-RAG (1:5000 in diluent). Streptavidin-HRP (Dako) was then added in a 1 in 10,000 dilution (in 1% BSA/PBS) and incubated for 1 hour. Between each step, plates were washed thoroughly with 0.1% Tween 20/PBS. Plates were developed by using TMB substrate, the reaction was stopped by acidification (3 M H2SO4), and the optical density (OD) was read at 450 nm.

Standard curves for the estimation of sCD40 concentration were generated by using serial dilutions of CD40-Ig that were prepared in 0.5% BSA/PBS diluted 2:1 (volume/volume) with sample diluent (2% Casein, 0.2% Tween 20, 0.02% NaN3). Stored serum samples, plasma samples, and supernatants were thawed, vortexed, then centrifuged (×10,000 g, 30 minutes) immediately prior to the analysis of nonsedimented material. Samples were diluted 2:1 (volume/volume) with sample diluent prior to ELISA analysis. Each sample was measured in duplicate, and wells coated with the isotype control antibody MOPC21 were used as a measure of the nonspecific background for each individual sample. In a number of experiments, samples were immunodepleted with CD40 (G28-5) or control (MOPC21) mAb prior to analysis, as described previously.24

Statistical Analysis

Associations between sCD40 and other continuous covariates were calculated by using Spearman rank correlations. Differences in the cellular and clinical characteristics of the sCD40High and sCD40Low populations were evaluated by using chi-square statistics. Survival distributions were plotted with the Kaplan–Meier method and were compared by using the log-rank test. Differences between groups were evaluated by the Mann–Whitney U test. Multivariate analysis of prognostic factors was performed using the Cox proportional hazards regression model. In the multivariate analysis of patients with AML, cytogenetics were coded as poor prognosis versus good versus intermediate prognosis,21 sCD40 was coded as high versus low, achievement of complete remission (CR) was coded as yes or no, and other variables were coded according to their division into quartiles. In the multivariate analysis of patients with myeloma, sCD40 was coded as high versus low, World Health Organization performance status (PS) was coded into 2 groups (PS 0-2 and PS 3-4), and all other variables were coded according to their division into quartiles.

RESULTS

Establishment of an sCD40 ELISA

A sandwich ELISA for the detection of sCD40 was developed by using a CD40 mAb as the capture reagent and polyclonal goat anti-CD40 (GA40) for detection. Although to was reactive strongly with CD40 Ig, the ELISA showed no reactivity with control fusion proteins, including CD86-Ig (Fig. 1) and CTLA4-Ig (data not shown). Specificity was confirmed further by the absence of signal when control capture and detection reagents were used in the ELISA (Fig. 1). The sensitivity limit of the ELISA was <50 pg/mL with CD40-Ig used as the internal standard (data not shown). Analysis of human plasma detected a signal significantly above background levels, and this signal was abolished when control capture or detection reagents were used (Fig. 1). Similar results were obtained with human serum (data not shown).

Figure 1.

This graph illustrates the specificity of the soluble CD40 (sCD40) enzyme-linked immunosorbent assay (ELISA). The reactivity of the ELISA with CD40 immunoglobulin (CD40-Ig), CD86-Ig, and plasma from patients with chronic lymphocytic leukemia (CLL Plasma) was analyzed by using different combinations of capture (CD40 monoclonal antibody [mAb]) or isotype-matched control mAb (Con) and detection reagents (GA40 or control Ig [Con Ig]). Data are shown as the mean optical density (OD) ± standard error of the mean and are from 1 representative experiment of 3 performed.

To confirm further the specificity of the sCD40 signal detected in plasma and serum, several samples were immunodepleted with a third-party CD40 antibody prior to analysis by ELISA. A single round of immunodepletion specifically reduced the sCD40 ELISA signal observed in both serum and plasma samples by >80% (n = 3 samples; data not shown).

Levels of sCD40 in Normal and Patient Samples

Plasma and serum levels of sCD40 were determined by ELISA; and, for each individual sample, the background was defined as the signal that was obtained by using a control capture mAb. All samples that were analyzed contained detectable levels of sCD40 (Fig. 2). Normal plasma levels (n = 96 samples) ranged from 0.163 ng/mL to 0.649 ng/mL (median, 0.415 ng/mL). A wider range of levels was observed in patient plasma samples (n = 106 samples; range, 0.138-3.277 ng/mL; median, 0.65 ng/mL). The sCD40 levels in CLL samples ranged from 0.477 ng/mL to 3.277 ng/mL (median, 0.807 ng/mL; n = 37 samples), and they ranged from 0.497 ng/mL to 3.27 ng/mL in MCL samples (median, 0.95 ng/mL; n = 5 samples), from 0.219 ng/mL to 3.606 ng/mL in AML samples (median, 0.573 ng/mL; n = 62 samples), and from 0.138 ng/mL to 1.646 ng/mL in MDS samples (median, 0.517 ng/mL; n = 21 samples). The plasma levels in each patient group were significantly higher than those in normal plasma (P <.02).

Figure 2.

Circulating soluble CD40 (sCD40) levels are illustrated in these scatter plots of either plasma or serum samples that were obtained from normal donors and from patients with hematologic malignancies. AML: acute myeloid leukemia; MDS: myelodysplastic syndrome; CLL: chronic lymphocytic leukemia; MCL: mantel cell lymphoma; MM: multiple myeloma.

Normal serum levels (n = 52 samples) ranged from 0.522 ng/mL to 1.697 ng/mL (median, 1.047 ng/mL) and were significantly higher than the levels observed in normal plasma (P <.0001). The sCD40 levels in myeloma serum samples (n = 76 samples; range, 0.459-5.951 ng/mL; median, 1.305 ng/mL) were significantly higher than the levels observed in samples from normal donors (P = .0002).

Overall 73% of patients with CLL, 80% of patients with MCL, 40% of patients with AML, 43% of patients with MDS, and 33% of patients with myeloma patients had sCD40 levels above the normal range. In a number of experiments, CD40-Ig was added to normal plasma samples, normal serum samples, and patient samples that had low sCD40 levels. This addition resulted in the expected increase in ELISA signal, demonstrating that the detection of low/normal sCD40 levels was because of a lack of sCD40 and because of the presence of an inhibitor of sCD40 detection, such as soluble CD154 (data not shown).

Association of sCD40 Levels with Myeloma Clinical Characteristics

In patients with myeloma, there was no significant correlation observed between sCD40 levels and either age or serum calcium levels (Table 1). However, sCD40 levels were correlated weakly with platelet levels and strongly with both β-2-microtubulin (β2M) and blood urea nitrogen (BUN) levels. The median BUN level was 6.5 mg/dL, and only 7 of 72 patients had levels >20 mg/dL, suggesting that most patients had normal renal function and that sCD40 levels were not affected significantly by renal function.

Table 1. Correlation of Clinical and Laboratory Features with Soluble CD40 Levels
CharacteristicMMAML
R*PRP
  • MM: multiple myeloma; AML: acute myeloid leukemia; WBC: white blood cells; NA: not available; β2M: β-2-microglobulin; BUN: blood urea nitrogen.

  • *

    Spearman rank correlation coefficient.

  • Significant correlation at the .05 level.

Age0.14.240.33<.01
WBCNANA0.44<.001
Platelet0.22.050.03.8
HemoglobinNANA−0.03.79
β2M0.53<.0010.71<.001
CreatinineNANA0.18.15
BUN0.58<.0010.34<.001
Serum Ca0.06.41NANA

The patients with myeloma were divided prospectively into 2 groups based on their serum levels of sCD40. The cut-off level was set at 1.697 ng/mL, a value equal to the upper level observed in normal donors. Patients with levels ≤1.697 ng/mL (n = 51 patients) were defined as the sCD40Low group, and patients with levels >1.697 ng/mL (n = 25 patients) were defined as the sCD40High group.

The prognostic value of sCD40 was evaluated by comparing the overall survival of the sCD40Low group with that of the sCD40High group. Patients in the sCD40High group had a significantly shorter survival (P = .015) than patients in the sCD40Low group (Fig. 3A ). In univariate analysis, other variables that were associated significantly with survival (P<.01) were PS, BUN, β2M, and platelets; whereas serum calcium, treatment protocol, and patient age were not associated significantly with survival. The independence of the prognostic value of sCD40 levels was evaluated by using the Cox proportional hazards regression model. The level of sCD40 (sCD40High vs. sCD40Low) was a significant independent prognostic factor in this model together with PS and platelet levels; whereas BUN, β2M, and patient age were not significant prognostic factors in this model (Table 2).

Figure 3.

These charts illustrate the survival of patients with myeloma based on pretreatment serum levels of soluble CD40 (sCD40). Patients were separated into sCD40Low and sCD40High groups by using an sCD40 level of 1.69 ng/mL as the cut-off value. Survival data are presented as Kaplan–Meier survival curves. (A) All patients with multiple myeloma (MM). (B) Patients with MM who had β-2-microtubulin levels >10 mg/L.

Table 2. Multivariate Analysis of the Correlation between Clinical Characteristics and Survival in Patients with Myeloma
Variable*Risk Ratio95% CIP
  • 95% CI: 95% confidence interval; sCD40: soluble CD40; PS: performance status; BUN: blood urea nitrogen; B2M: β-2-microglobulin.

  • *

    All variables were categorical. sCD40 and PS were divided into 2 groups, and the remaining variables were divided into quartiles.

  • Risk ratios with P<.05.

sCD402.311.22–4.39.011
PS2.131.22–3.75.013
BUN1.160.89–1.50.254
β2M1.050.79–1.40.704
Platelets0.740.58–0.96.024
Age1.130.88–1.45.338

β2M is a widely used and powerful predictor of survival in patients with myeloma; and, in a recent, large, international study, it was reported that levels >10 mg/L identified a group of very-high-risk patients.25 Because of the strong correlation between β2M and sCD40 levels, the correlation between these 2 variables was investigated further by comparing the survival of sCD40High and sCD40Low groups in patients with β2M levels >10 mg/L (Fig. 3B). Although this patient group already had poor overall survival, division on the basis of sCD40 levels separated them further into groups with relatively poor survival (sCD40High) or improved survival (sCD40Low). Although there was a large apparent difference in the survival of the sCD40High and sCD40Low groups, which was significant according to the log-rank test (P = .025), the small numbers of patients in each group mean that this difference must be evaluated with considerable caution. However, these data further support the results of the multivariate analysis, which indicated that sCD40 levels provide additional prognostic information to that supplied by the measurement of β2M levels.

Association of sCD40 Levels with AML Clinical Characteristics

In patients with AML, there was no significant correlation observed between sCD40 levels and either hemoglobin levels or platelet counts (Table 1). There was a similar lack of correlation with creatinine levels and only a weak correlation with BUN, suggesting that sCD40 levels were not influenced significantly by renal function. However, sCD40 levels were correlated strongly with β2M and moderately with leukocyte counts and age. Comparison of sCD40 levels within the different French–American–British AML subtypes suggested that the presence of elevated sCD40 levels was not associated with any particular subtype, although the low numbers of patients within many of the subtypes precluded statistical comparison (data not shown).

The patients with AML were divided prospectively into 2 groups based on their plasma levels of sCD40. The cut-off level was set at 0.649 ng/mL, a value equal to the upper level observed in the normal donor group. Patients with levels ≤0.649 ng/mL were defined as the sCD40Low group, and patients with levels >0.649 were defined as the sCD40High group.

Among the patients with AML, a comparison of the sCD40Low group with the sCD40High group demonstrated no significant differences between the 2 groups with respect to the CR rate (63% vs. 41%, respectively), the percentage of patients with poor cytogenetics (24.3% vs. 24%, respectively), the percentage of patients older than age 64 years (41% vs. 56%, respectively), or the percentage of patients who had antecedent hematologic disorders (51% vs. 62%, respectively). However, the sCD40High group had a significantly greater percentage of patients with elevated WBC counts (>10 × 109/L; 56% vs. 21%; P = .006).

The prognostic value of sCD40 was evaluated by a comparison of the overall survival of the sCD40Low and the sCD40High groups (Fig. 4A ). Patients in the sCD40High group had a significantly shorter survival (P = .042) compared with patients in the sCD40Low group. In univariate analysis, other variables that were associated significantly with survival (P<.05) were achievement of CR, cytogenetics, and β2M levels; whereas patient age, WBC count, creatinine, and BUN levels were not associated significantly with survival. The independence of the prognostic value of sCD40 levels was evaluated by using the Cox proportional hazards regression model (Table 3). sCD40 was a significant independent prognostic factor in this model, together with age, cytogenetics, and achievement of CR when the model included the interaction of age with sCD40. To investigate further the interaction of age and sCD40, the survival of patients in the sCD40Low and sCD40High groups was compared among the patients with AML, who were divided into different age groups. The median age of the patients with AML (64 years) was used as a cut-off age. In patients with AML who were age 64 years or younger, the presence of elevated sCD40 levels was associated with a significantly shorter survival. (Fig. 4B). However, in patients who were older than age 64 years, there was no significant difference in the survival between the sCD40Low group and the sCD40High group (Fig. 4C).

Figure 4.

These charts illustrate the survival of patients with acute myeloid leukemia (AML) based on pretreatment plasma levels of soluble CD40 (sCD40). Patients were separated into sCD40Low and sCD40High groups by using an sCD40 level of 0.649 ng/mL as the cut-off value. Kaplan–Meier survival curves for the sCD40Low and sCD40High groups are presented for all patients with AML (A) patients with AML younger than age 64 years (B), and (C) patients with AML older than age 64 years.

Table 3. Multivariate Analysis of the Correlation between Clinical Characteristics and Survival in Patients with Acute Myeloid Leukemia
Variable*Risk Ratio95% CIP
  • 95% CI: 95% confidence interval; sCD40: soluble CD40; CR: complete response.

  • *

    All variables were categorical. sCD40 and CR were divided into 2 groups, cytogenetics was into 3 groups, and age was divided into quartiles prior to analysis.

  • Risk ratios with P<.05.

sCD407.461.38–40.39.004
Age2.011.08–3.74.01
Cytogenetics4.321.54–12.06.003
Achieve CR0.200.07–0.54.005
Age and sCD400.440.19–1.01.012

Association of sCD40 Levels with MDS Clinical Characteristics

In patients with MDS, there was no significant correlation observed between sCD40 levels and either hemoglobin levels, platelet counts, or age. There was a similar lack of correlation with creatinine and BUN levels, suggesting that sCD40 levels were not influenced significantly by renal function. However, sCD40 levels were correlated strongly with β2M levels (R = .68; P = .001) and WBC counts (R = .64; P = .002).

Association of sCD40 Levels with CLL Clinical Characteristics

Analysis of the patients with CLL demonstrated a weak correlation with age (R = .39; P = .03) but no significant correlation with WBC counts, hemoglobin, or platelet levels. In addition, there was no significant difference (P = .19) between the sCD40 levels among patients with CD38-negative status and CD38-positive status, which we determined by using 7% CD38 positivity as a cut-off level (data not shown).

Release of sCD40 by CLL Cells

It was reported previously that human B cells release sCD40 in response to IL-420 or plate-bound CD40 mAb.19 Therefore, the levels of sCD40 release by CLL cells were determined after culture in the presence of combinations of these stimuli.

CLL cells that were cultured in the presence of either media alone or media plus IL-4 released only low levels of sCD40 (range, 0.028-0.052 ng/mL; data not shown). Similarly, low levels of sCD40 release were observed after culture in wells that were precoated with a control mAb (Fig. 5). However, significantly higher levels of sCD40 release were observed in all cell preparations when the cultures were carried out in wells that were precoated with CD40 mAb. In 2 of the 4 patients analyzed, the presence of IL-4 significantly increased the sCD40 release induced by solid-phase CD40 mAb.

Figure 5.

This chart illustrates the release of soluble CD40 (sCD40) by chronic lymphocytic leukemia (CLL) cells during culture. CLL cells from 4 separate patients were suspended in either media alone or in media with interleukin 4 (IL-4), and then were cultured for 72 hours in wells that were precoated with either control monoclonal antibody (mAb) or CD40 mAb. The levels of sCD40 in the culture supernatants were determined by enzyme-linked immunosorbent assay, and data from triplicate samples are shown as the mean ± standard error of the mean.

DISCUSSION

In the current study we demonstrated that a large proportion of patients with hematologic malignancies had significantly elevated levels of the circulating soluble form of CD40 and that these elevated levels are associated with a poor prognosis, at least in patients with AML and MM. The mechanism that underlies the association between sCD40 and survival is unclear. The release of sCD40, if functional, provides a potentially powerful means by which the CD40-CD40 ligand interaction and, thus, the nature and magnitude of immune responses, may be modulated. Current functional data that were obtained by using recombinant or in vitro-derived forms of sCD40 suggest that the presence of sCD40 would inhibit immune responses.14, 15, 17 The association between high sCD40 levels and poor survival in both AML and MM provides further evidence to support this hypothesis. However, currently, there is no direct evidence that sCD40 can impair immune responses to infection and/or malignant cells. The possibility cannot be discarded that sCD40 may not be relevant functionally but may only be a marker of more aggressive or chemotherapy-resistant malignancies. The prognostic value of sCD40 levels in MM and AML appears to be independent of other established prognostic factors. This suggests that, irrespective of its actual functional roles, sCD40 merits further investigation as a clinically useful prognostic marker not only in AML and MM but also in CLL and MCL.

A number of factors in addition to absolute concentration will influence the function of in vivo-generated sCD40. These include the kinetics of its release relative to the expression of mCD40 and its ligands and its levels of oligomerization, which can alter sCD40 function dramatically.14 The source(s) of circulating sCD40 currently is unknown, although the demonstration in the current report that CLL cells can release sCD40 after CD40 ligation suggests that both normal and malignant cells are potential sources. These results also raise the possibility that the generation of activated, CD40L-positive T cells during an immune response may trigger sCD40 release. The high levels of sCD40 in serum compared with plasma suggest that the clotting process also may induce sCD40 release, possibly through the activation of CD40-positive platelets.3

Although mCD40 is expressed widely by hematologic malignancies, and it is clear that mCD40 ligation induces considerable cellular changes in these cells, the effect of these changes on malignant cell immunogenicity and survival remains controversial.2 Membrane expression of CD40 by AML cells reportedly has been associated with a poor prognosis.7 Therefore, establishing the correlation and functional relation of the membrane and soluble forms of CD40 will require further investigation.

Currently, there is considerable interest in the use of CD40 ligating reagents, such as CD40 mAb, CD40LT, and CD154-positive cells, as therapeutic agents in the treatment of hematologic malignancies, particularly CLL and myeloma.4, 8 The effectiveness of these reagents results predominantly from their ligation of immune system-associated mCD40. The presence of circulating sCD40 potentially may reduce or eliminate these therapeutic effects by providing an alternative pool of material that can bind CD40 ligating reagents before their interaction with mCD40. It is believed that this type of interaction between soluble forms of a target antigen and therapeutic antibodies in patients with leukemia reduces the efficacy of therapeutic antibodies, such as Rituximab and Campath-1H.26 The observation in the current study that, at least in CLL, ligation of mCD40 on leukemia cells can trigger sCD40 release raises the possibility that CD40 ligation-based therapy also may increase sCD40 levels, which may inhibit antitumor responses. Prospective studies on the clinical significance of circulating sCD40 in patients receiving CD40 ligating reagent therapy, thus, will be important in the optimization of these therapies.

The importance of mCD40 in the development of antitumor responses has stimulated considerable interest in this molecule as a target of immunotherapy for patients with hematologic malignancies. The data presented in the current study demonstrate that elevated levels of a circulating, soluble form of CD40 are present in many patients with hematologic malignancies and are associated with poor survival. These findings have considerable implications for CD40-based immunotherapies and provide a strong rationale for further studies on the biologic function, prognostic significance, and therapeutic significance of sCD40 in hematologic malignancies.

Ancillary