• chronic lymphocytic leukemia;
  • soluble CD23;
  • cytometric bead array;
  • flow cytometry;
  • ELISA;
  • prognostic factor


  1. Top of page
  2. Abstract

The soluble form of the transmembrane glycoprotein CD23 corresponding to the low-affinity receptor for the immunoglobulin E (sCD23) is found in the serum of patients with chronic lymphocytic leukemia (CLL). In this disease, an increase in sCD23 level is predictive of poor prognosis at diagnosis as well as during clinical outcome. Quantification of sCD23 is classically performed by enzyme-linked immunosorbent assay (ELISA), a method not routinely used in hematology laboratories. Our aim was to apply cytometric bead array (CBA) technology to measure sCD23 levels. We tested 420 serum samples, 360 from patients and 60 from healthy volunteers. We selected three pairs of monoclonal antibodies recognizing the CD23 molecule that were tested in various conditions of temperature, centrifugation, washing, or chemical supplementation. Satisfactory performances in terms of repeatability (CV: 5%) and reproducibility (CV: 6%) were obtained with the selected pair of antibodies, with a threshold of positivity at 6 ng/mL. CBA and ELISA techniques were correlated with a Spearman coefficient at 0.99. The reproducibility and reliability of the sCD23 CBA assay were confirmed, with a Spearman coefficient at 0.99 in a series of 23 CLL patients and 13 controls tested in two laboratories equipped with different cytometers and using different lots of CBA reagents. Data obtained with serum and plasma samples were correlated with a Spearman coefficient at 0.99. Our study validates a simple method that allows the clinicians to benefit from an indicator of prognosis at the diagnosis as well as a marker of the evolution of CLL disease. © 2013 International Clinical Cytometry Society

Chronic lymphocytic leukemia (CLL) is the most common adult leukemia in Western Europe and North America, characterized by a proliferation of CD5 positive monoclonal B-cells in peripheral blood, bone marrow, and lymphoid tissues. Diagnosis of CLL is based on the presence in peripheral blood of more than 5 × 109/L leukemic cells with a Matutes score at 4 or 5 ([1, 2]). Leukemic B-cells constitutively overexpress CD23 antigen when compared with normal B-cells as a result of the abnormal upregulation of the CD23 gene ([3, 4]).

CD23 is a 45-kDa transmembrane glycoprotein corresponding to the low-affinity receptor for the immunoglobulin E ([5]). There are two isoforms of the CD23 antigen (type A and B) only differing by their intracytoplasmic domain ([6]). This protein can be expressed on monocytes, T-cells, platelets, eosinophils, Langerhans cells, and follicular dendritic cells ([4, 7]). The CD23 antigen is expressed on a variety of hematopoietic cells and especially on B-cells linked to pathological conditions such as CLL ([8]). The CD23 protein can be cleaved and released into the serum as a soluble form of CD23 (sCD23). This sCD23 is a 25 kDa fragment that can be found in serum, plasma, and urine in CLL ([9]).

The increased concentration of sCD23 in the serum of CLL patients was first reported in 1988 ([10]). In addition, it was demonstrated that the sCD23 level correlated with the clinical stage of the disease and with the lymphocyte count ([10]). Subsequently, Sarfati and coworkers demonstrated that: (i) CD23 was directly involved in the proliferation of normal and leukemic B-cells ([11]), and that (ii) sCD23 levels provide significant and additional prognostic information in terms of overall survival, particularly in early-stage patients ([12]). In addition, Saka et al. showed that serum sCD23 values were increased in CLL patients and could be used for clinical follow-up, predicting tumor mass and prognosis ([13]). Finally, Meuleman et al. provided evidence that the doubling time of sCD23 (sCD23DT) in the serum of CLL patients correlated with an increased risk of disease progression ([14]). This group also demonstrated that sCD23DT is a relevant prognostic factor for time-to-treatment and for overall survival.

sCD23 is currently quantified by Enzyme-Linked ImmunoSorbent Assay (ELISA), a method not frequently applied by hematology laboratories. The cytometric bead array (CBA) can be a very convenient alternative technology for sCD23 quantification ([15, 16]) because the diagnosis of CLL is routinely performed by flow cytometry. Given a Matutes score of 4 or 5 is a component of the diagnosis, we propose here to evaluate whether sCD23 CBA technology can be implemented on the same blood sample throughout the full laboratory procedure in the diagnostic or prognostic steps in CLL.


  1. Top of page
  2. Abstract

Patients and Controls

This study was performed on 420 serum samples consisting of 312 samples from 180 typical CLL patients with Matutes score at 4 or 5. Of these, 146 were patients from the Pitie-Salpetriere hospital (Paris, France) and 34 from the Notre Dame hospital (Montreal, Canada) (Table 1). Thirty-two patient samples were collected at the time of diagnosis and 280 samples during the follow-up period. Control samples were obtained from healthy adult volunteers, 50 in Paris and 10 in Montreal. An additional series of 48 samples from patients with B-cell non-Hodgkin lymphoma (B-NHL) were added for overall correlation.

Table 1. Summary of CLL Patients and Controls.
  n (patients)Sex (M/F)Agen (samples)DiagnosisFollow-up
  1. Abbreviation: CLL, chronic lymphocytic leukemia

CLLParis14691/5568 [37–90]26032228
Montreal3415/1972 [53–90]52052
ControlsParis5023/2743 [21–67]50--
Montreal104/659 [31–69]10--

Analyses were carried out on frozen serum or plasma samples obtained on a dry tube, or on a lithium heparin tube. The serum samples of the patient were assayed as directed by the treating physician. Control and CLL subjects provided a written, informed consent before their participation into the study. These studies were approved by the respective local ethic committees.

Enzyme-Linked ImmunoSorbent Assay

Serum and plasma sCD23 levels were measured by ELISA (sCD23 Direct Human ELISA Kit, Biosource, Nivelles, Belgium) according to the manufacturer's recommendations.

Cytometric Bead Array

Quantification of sCD23 was determined in serum and in plasma samples with the newly designed CBA. The coating of the beads with capture monoclonal antibody (moAb) (a-CD23, clone EBVCS5, BD Biosciences, Le Pont de Claix, France) was performed by BD Biosciences. Standards (human recombinant CD23—M. Sarfati, Canada) ([17, 18]) or patient samples were incubated in a first step with capture beads (Fig. 1A). In a second step, a phycoerythrin (PE) conjugated detection moAb (a-CD23, clone M-L233, BD Biosciences) was added (Fig. 1B). After a second incubation, samples were analyzed by flow cytometry using a BD FACSCanto II™ or a BD FACSAria II™. The capture beads were selected through the APC/APC-Cy7 channel upon excitation with the red laser (633 nm). The sCD23 quantification was performed based on PE fluorescence excitation with the blue laser (488 nm). Results were analyzed by BD FACSDiva software (v. 6.1.3).


Figure 1. Schematic representation of CBA principle. A: In a first step, coated beads and patient samples or recombinant protein (standard) were mixed during 1 h. B: In a second step, antibodies coupled to PE (moAb-PE) were added. After 2 h incubation, complexes were formed between beads, sCD23, and moAb-PE. C: Beads were selected in APC/APC-Cy7 dot plot and gated to allow for the quantification of the PE fluorescence intensity. A standard curve was constructed with human recombinant sCD23 to quantify the sCD23 content of each sample. Abbreviations: sCD23, soluble CD23; CBA, cytometric bead array; moAb, monoclonal antibody.

Download figure to PowerPoint

Concentrations were calculated using the recombinant CD23 standard curve (Fig. 1C).

During this step of development, three moAb pairs were tested: pair#1, moAb-135 and moAb-176 ([19]); pair#2, moAb clone M-L233 (BD Biosciences) and moAb clone EBVCS5 (BD Biosciences); pair#3, same antibody as pair#2 but reversed. The best results were observed with pair#3, as documented in result section. To note that moAbs used in the CBA assay differ from those used in ELISA technique.

Statistical Analysis

All results were exported to an internal Microsoft Excel table. Spearman coefficients were used for comparison between the two techniques and the two laboratories. Bland and Altman plots calculated with MedCalc software (version 12.7.0) were used to evaluate the bias between the two techniques. Data have been put down on a logarithmic scale.


  1. Top of page
  2. Abstract

Technical Development

To develop a CBA technique to quantify sCD23, we first identified 4 moAbs that would recognize the extracellular domain of CD23 molecule with high affinity. In a second step, we selected 3 moAbs pairs so that recognized two different epitopes on the CD23 molecule (sandwich assay). For each of these pairs created, one moAb (capture type) was conjugated to a selected cytometric bead type and coupled by BD Biosciences. The second moAb of the chosen pair (detector type) was labeled with PE. The human recombinant sCD23 was used for construction of the standard curve. A linear relationship was obtained up to 18 ng/mL (data not shown).

Several experimental conditions were evaluated for obtaining optimal stability of the antigen-antibodies complex. We carefully investigated the influence of ([1]) temperature (+4°C or room temperature), ([2]) the speed and duration of centrifugation, ([3]) the composition of the assay buffer, ([4]) the washing steps, and ([5]) various reagents (paraformaldehyde, anti-proteases). The assay conditions that were considered as most critical, such as the centrifugation step and the assay temperature, were assessed with all three selected moAb pairs. As depicted in Figure 2A, the results obtained with one of the moAb pairs revealed that optimal conditions (test conditions #4) included room temperature (20°C), no centrifugation, and no washing steps, whatever the moAb pairs tested. These conditions resulted in the highest mean fluorescence and the best stability in time (lines t = 0 and t = 60 min are surimposed). These optimized conditions were slightly different to the standard marketed CBA kits. The addition of other reagent, like paraformaldehyde, did not result in additional benefits (data not shown).


Figure 2. Technical conditions and choice of moAb pairs. A: Comparison of four different conditions: Test #1: cold temperature, without washing; Test #2: cold temperature, washing with centrifugation; Test #3: room temperature, washing with centrifugation; Test #4: room temperature, without washing. Each test was measured at t = 0 (full lines) and t = 60 min (dashed lines). Best conditions were detected in test #4: highest fluorescence signal; 1 h signal stability. B: Comparison of the three pairs of moAbs under optimized conditions (room temperature, without washing) and measurement after 30 min (dashed lines). Best moAb pair was pair #3: highest mean fluorescence value. Abbreviation: moAb, monoclonal antibody.

Download figure to PowerPoint

Three moAb pairs were evaluated with these optimized conditions (Fig. 2B). The moAb pair #3 allowed to obtain the highest fluorescence and the best complex stability. This pair was, therefore, chosen to complete all further experiments.

Technical Validation

The optimized sCD23 CBA test was next evaluated for performance of repeatability and reproducibility.

For intra-assay repeatability, four samples with different levels of sCD23 (1.2 ng/mL, 33 ng/mL, 60 ng/mL, and 167 ng/mL) were assayed 10 times on the same day. The coefficients of variation (CV) were, respectively, 8.5%, 2.1%, 5%, and 4.4%. The mean value results in a CV of 5%. As can be expected, CV was decreasing with increasing value of sCD23.

For inter-assay reproducibility, we analyzed the four same samples with variable reagent conditions: 2 lots of beads, 2 cytometers (two different BD FACSCanto II), and different lots of assay buffer and of antibody detector. The CV were 9.9%, 5.4%, 4.7%, and 4.4%, respectively. The mean value gives a CV of 6%.

Correlation Between CBA and ELISA Technique

The sCD23 CBA method was compared with the ELISA technique (sCD23 Direct Human ELISA Kit—Biosource) ([15]) in a large cohort of 420 samples, 312 from CLL patients, 48 from B-NHL patients, and 60 controls (Table 1). As shown in Figure 3A, the overall correlation between CBA and ELISA (all 420 samples) was very good with a Spearman coefficient (r) of 0.99 (Fig. 3A, left column). On the whole cohort of samples, the Bland and Altman plots show a mean ratio of ELISA on CBA of 1.07 [0.46; 2.5] (Fig. 3A, right column). Similar results were obtained whatever the subgroup of patients. For CLL samples and control samples evaluated in Paris, the Bland and Altman plots show a mean ratio of ELISA and CBA of 1.05 [0.44; 2.5] and of 1.23 [0.49; 3.02], respectively (Figs. 3B and 3C). For CLL and control samples evaluated in Montreal (Fig. 3D), the Bland and Altman plots show a mean ratio of ELISA on CBA of 1.1 [0.55; 2.19]. In all these situations, the mean values of the ratios of the both techniques were close to 0. For very low values (geometric mean < 1 ng/mL), there is a clear bias with larger values measured with ELISA than those obtained with CBA technique, that is potentially explained by the difference of the detection limits of each method. However, these very low values have no clinical relevance. The cut-off value for positivity, calculated from the average of all 60 control samples, was set at 6 ng/mL.


Figure 3. Comparison between CBA and ELISA techniques for sCD23 quantification. Correlations (Spearman coefficient, r, on the left column) and bias (Bland and Altman plots, on the right column). For Bland and Altman plots, data have been put down on a logarithmic scale. The geometric mean of CBA and ELISA values and the ratio of ELISA/CBA values are visualized on horizontal axis and vertical axis, respectively. A: Overall correlation for 420 serum samples (312 CLL + 60 controls + 48 B-NHL). B: Evaluation in Paris: 260 CLL samples. C: Evaluation in Paris: 50 control samples. D: Evaluation in Montreal: 52 CLL and 10 control samples. Abbreviations: sCD23, soluble CD23; B-NHL, B-cell non-Hodgkin lymphoma; CBA, cytometric bead array; CLL, chronic lymphocytic leukemia; ELISA, Enzyme-Linked ImmunoSorbent Assay.

Download figure to PowerPoint

Correlation CBA Paris/Montreal

To further evaluate assay reliability and reproducibility, the technique was performed on two different cytometer models in Paris (BD FACSCanto II) and in Montreal (BD FACSAria II). A total of 24 CLL and 13 control samples were used. The Spearman coefficient (r) of the CBA values for this limited data set was 0.99 and the mean ratio of the results obtained in the two laboratories were close to 0 as visualized on the Bland-Altman plots (mean: 0.91 [0.59; 1.38]) (Fig. 4A left and right column, respectively). The CV were 7%. These data demonstrate that the CBA assay can be implemented with nearly identical results in two different laboratories using different lots of CBA reagents for both capture and detector components.


Figure 4. Comparison between two laboratories for sCD23 quantification by CBA technique and comparison between serum and plasma. Correlation (Spearman coefficient, r, on the left column) and bias (Bland and Altman plots, on the right column). For Bland and Altman plots, data have been put down on a logarithmic scale. A: Paris versus Montreal comparison: results of 37 serum samples (24 patients and 13 controls). The geometric mean of sCD23 CBA Paris and CBA Montreal values is visualized on horizontal axis. The ratio of sCD23 CBA Paris/CBA Montreal values is visualized on vertical axis. B: Serum versus plasma (lithium heparin tube) comparison: results of 17 samples (11 CLL and 6 controls). The geometric mean of serum and plasma sCD23 values and the ratio of serum/plasma sCD23 values are visualized on horizontal axis and vertical axis, respectively. Abbreviations: CBA, cytometric bead array; CLL, chronic lymphocytic leukemia; sCD23, soluble CD23.

Download figure to PowerPoint

Correlation Between Serum and Plasma

We further examined the potential interference of anticoagulant in sCD23 measurement by the CBA method. We compared the results obtained from serum with those from plasma. Samples were collected from 11 CLL patients and six controls (plasma samples were collected on lithium heparin). The Spearman coefficient (r) observed between the two sample types was 0.99 (Fig. 4B left column) and the mean ratio of values obtained in serum and plasma (mean: 1.02 [0.85; 1.23]) were close to 0 in the Bland-Altman plots (Fig. 4B, right column).


  1. Top of page
  2. Abstract

We report here the establishment of a simple and robust quantitative test for sCD23 using the CBA technology.

Our results demonstrate a very good reproducibility and reliability of this newly developed CBA technique for the detection and quantification of sCD23. We observed a high correlation between the proposed CBA and the reference ELISA assay. This correlation is observed for patient samples as well as for controls, confirming the CBA assay as an acceptable method for the quantification of sCD23 in CLL. The implementation of this CBA assay implies the availability of a flow cytometer device in the laboratory. Both the CBA and the ELISA techniques produce the highest benefit when analyzing larger series of samples. These aspects could be considered as potential limitations. However, most hematology laboratories are generally equipped with flow cytometry systems because immunophenotyping of tumoral cells is considered as a required step in the diagnostic process. As recommended by the WHO classification in 2008 ([1]), the diagnosis of CLL is defined by a specific phenotypic profile and the determination of sCD23 quantification may easily been included in the diagnostic strategy. In contrast, ELISA plate readers are not commonly available in the hematology lab. Interestingly, the CBA assay can be performed on plasma samples obtained on a lithium heparin tube. Thus, the newly proposed CBA assay allows the use of a single patient sample tube to determine immunophenotyping and quantification of sCD23 on the same analytical platform. Frozen plasma samples can be stored for bulk analysis saving valuable standard material. Frozen samples can also be easily shipped between laboratories for sCD23 quantification under patient monitoring conditions. Of note, to interpret the results of sCD23 measurement, it remains important to take into account the clinical status of the patient, particularly in case of allergic or auto-immune disease ([20, 21]).

Serum thymidine kinase (sTK) is a serum marker with prognostic value in previously untreated patients diagnosed with Binet stage A CLL ([22]). An elevated sTK level is correlated with a high sCD23 serum level ([23]). Under this condition, the newly proposed CBA technique for the quantitative determination of sCD23 can substitute the sTK radioimmunoassay as a less desirable and more regulations-restrained technique. As an additional limitation, the sTK level may be influenced by various factors unrelated to CLL like viral infections or vitamin B12 deficiency, possibly inducing an increase of serum levels ([24, 25]). Furthermore, sTK results are also generally quantified in U/L (nonstandardized units). Assay results, therefore, are not comparable from one laboratory to another. In contrast, with flow cytometer, standardization being increasingly implemented, CLL patients could be monitored with the same technique independent of the laboratory site. Under the conditions mentioned above, the newly proposed CBA appears to be a most suitable technique for the quantification of sCD23 in serum samples of CLL patients.

Many reports have been generated to demonstrate the prognostic value of sCD23 at diagnosis. The correlation with clinical outcome has only recently been put into light ([12, 25, 26]). The importance in monitoring disease was suggested by Meuleman et al. by studying the doubling time of sCD23 (sCD23DT) ([14]). This study showed that the sCD23DT was correlated with an increased risk of progression. Their finding supports the use of the sCD23DT during disease monitoring through an objective and simple measurement ([14]). Similar to CD20 and CD52, CD23 appears to be a good candidate for immunotherapy in CLL. Also in such application, the proposed CBA for sCD23 could be a tool of preference in the follow up of the patients ([27-29]). Finally, considering the sensitivity and accuracy of the CBA measurement for low sCD23 values, it would be of interest to evaluate the performance of this newly validated technology for sCD23 quantification in minimal residual disease conditions.


  1. Top of page
  2. Abstract

The authors thank Dr. Guy Biron, MD, PhD, and the Department of Hematology of CHUM as well as Manuel Rubio, Patrick Bonnemye, Martine Brissard, Stéphanie Gueguen, Sylvie Baudet and Olivier Munoz for their helpful technical contribution and Dr. Audrey Brignoli for her statistical assistance.


  1. Top of page
  2. Abstract
  • 1
    Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, Vardiman JW, editors. World Health Organisation classification of Tumours of Haematopoietic and Lymphoid Tissues. IARC, Geneva, Switzerland: WHO Press; 2008.
  • 2
    Hallek M, Cheson BD, Catovsky D, Caligaris-Cappio F, Dighiero G, Döhner H, Hillmen P, Keating MJ, Montserrat E, Rai KR, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: A report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood 2008;111:54465456.
  • 3
    Sarfati M. Interleukin-4 and the two isoforms of CD23 antigen: Their possible contribution to the physiopathology of chronic lymphocytic leukemia disease. Nouv Rev Fr Hematol 1993;35:109113.
  • 4
    Fournier S, Delespesse G, Rubio M, Biron G, Sarfati M. CD23 antigen regulation and signaling in chronic lymphocytic leukemia. J Clin Invest 1992;89:13121321.
  • 5
    Lüdin C, Hofstetter H, Sarfati M, Levy CA, Suter U, Alaimo D, Kilchherr E, Frost H, Delespesse GCloning and expression of the cDNA coding for a human lymphocyte IgE receptor. EMBO J 1987;6:109114.
  • 6
    Fournier S, Tran ID, Suter U, Biron G, Delespesse G, Sarfati M. The in vivo expression of type B CD23 mRNA in B-chronic lymphocytic leukemic cells is associated with an abnormally low CD23 upregulation by IL-4: Comparison with their normal cellular counterparts. Leuk Res 1991;15:609618.
  • 7
    Delespesse G, Sarfati M, Hofstetter H. Human IgE-binding factors. Immunol Today 1989;10:159164.
  • 8
    Sarfati M, Fournier S, Wu CY, Delespesse G. Expression, regulation and function of human Fc epsilon RII (CD23) antigen. Immunol Res 1992;11:260272.
  • 9
    Sarfati M. CD23 and chronic lymphocytic leukemia. Blood Cells 1993;19:591599.
  • 10
    Sarfati M, Bron D, Lagneaux L, Fonteyn C, Frost H, Delespesse G. Elevation of IgE-binding factors in serum of patients with B cell-derived chronic lymphocytic leukemia. Blood 1988;71:9498.
  • 11
    Fournier S, Rubio M, Delespesse G, Sarfati M. Role for low-affinity receptor for IgE (CD23) in normal and leukemic B-cell proliferation. Blood 1994;84:18811886.
  • 12
    Sarfati M, Chevret S, Chastang C, Biron G, Stryckmans P, Delespesse G, Binet JL, Merle-Beral H, Bron D. Prognostic importance of serum soluble CD23 level in chronic lymphocytic leukemia. Blood 1996;88:42594264.
  • 13
    Saka B, Aktan M, Sami U, Oner D, Sanem O, Dinçol G. Prognostic importance of soluble CD23 in B-cell chronic lymphocytic leukemia. Clin Lab Haematol 2006;28:3035.
  • 14
    Meuleman N, Stamatopoulos B, Dejeneffe M, El Housni H, Lagneaux L, Bron D. Doubling time of soluble CD23: A powerful prognostic factor for newly diagnosed and untreated stage A chronic lymphocytic leukemia patients. Leukemia 2008;22:18821890.
  • 15
    Castillo L, Maccallum DM. Cytokine measurement using cytometric bead arrays. Methods Mol Biol 2012;845:425434.
  • 16
    Elshal MF, McCoy JP. Multiplex bead array assays: Performance evaluation and comparison of sensitivity to ELISA. Methods 2006;38:317323.
  • 17
    Mossalayi MD, Arock M, Delespesse G, Hofstetter H, Bettler B, Dalloul AH, Kilchherr E, Quaaz F, Debré P, Sarfati M. Cytokine effects of CD23 are mediated by an epitope distinct from the IgE binding site. EMBO J 1992;11:317323.
  • 18
    Sarfati M, Bettler B, Letellier M, Fournier S, Rubio-Trujillo M, Hofstetter H, Delespesse GNative and recombinant soluble CD23 fragments with IgE suppressive activity. Immunology 1992;76:662667.
  • 19
    Rector E, Nakajima T, Rocha C, Duncan D, Lestourgeon D, Mitchell RS, Fischer J, Sehon AH, Delespesse GDetection and characterization of monoclonal antibodies specific to IgE receptors on human lymphocytes by flow cytometry. Immunology 1985;55:481488.
  • 20
    Rogala B, Rymarczyk B. Soluble CD23 in allergic diseases. Arch Immunol Ther Exp (Warsz) 1999;47:251255.
  • 21
    Sayinalp S, Akalin S, Sayinalp N, Erbas T, Bayraktar M, Ozcebe OI, Dündar S, Kirazli S. Serum immunoglobulin E and soluble CD23 in patients with Graves' disease. Horm Metab Res 1996;28:133137.
  • 22
    Hallek M, Langenmayer I, Nerl C, Knauf W, Dietzfelbinger H, Adorf D, Ostwald M, Busch R, Kuhn-rcia-ManeroHallek I, Thiel E, et al. Elevated serum thymidine kinase levels identify a subgroup at high risk of disease progression in early, nonsmoldering chronic lymphocytic leukemia. Blood 1999;93:17321737.
  • 23
    Knauf WU, Langenmayer I, Ehlers B, Mohr B, Adorf D, Nerl CH, Hallek M, Zwingers TH, Emmerich B, Thiel ESerum levels of soluble CD23, but not soluble CD25, predict disease progression in early stage B-cell chronic lymphocytic leukemia. Leuk Lymphoma 1997;27:523532.
  • 24
    Hallek M, Wanders L, Strohmeyer S, Emmerich B. Thymidine kinase: A tumor marker with prognostic value for non-Hodgkin's lymphoma and a broad range of potential clinical applications. Ann Hematol 1992;65:15.
  • 25
    Zenz T, Mertens D, Küppers R, Döhner H, Stilgenbauer S. From pathogenesis to treatment of chronic lymphocytic leukaemia. Nat Rev Cancer 2010;10:3750.
  • 26
    Cramer P, Hallek M. Prognostic factors in chronic lymphocytic leukemia-what do we need to know? Nat Rev Clin Oncol 2011;8:3847.
  • 27
    Giordano Attianese GM, Marin V, Hoyos V, Savoldo B, Pizzitola I, Tettamanti S, Agostoni V, Parma M, Ponzoni M, Bertilaccio MT, et al. In vitro and in vivo model of a novel immunotherapy approach for chronic lymphocytic leukemia by anti-CD23 chimeric antigen receptor. Blood 2011;117:47364745.
  • 28
    Alatrash G, Albitar M, O'Brien S, Wang X, Manshouri T, Faderl S, Ferrajoli A, Burger J, Garcia-Manero G, Kantarjian HM, et al. Circulating CD52 and CD20 levels at end of treatment predict for progression and survival in patients with chronic lymphocytic leukaemia treated with fludarabine, cyclophosphamide and rituximab (FCR). Br J Haematol 2010;148:386393.
  • 29
    Manshouri T, Do KA, Wang X, Giles FJ, O'Brien SM, Saffer H, Thomas D, Jilani I, Kantarjian HM, Keating MJ, et al. Circulating CD20 is detectable in the plasma of patients with chronic lymphocytic leukemia and is of prognostic significance. Blood 2003;101:25072513.