ORIGINAL ARTICLE Flow Cytometric Protein Expression Profiling As a Systematic Approach for Developing Disease-Specific Assays
Total Page：16
File Type：pdf, Size：1020Kb
Leukemia (2006) 20, 2102–2110 & 2006 Nature Publishing Group All rights reserved 0887-6924/06 $30.00 www.nature.com/leu ORIGINAL ARTICLE Flow cytometric protein expression profiling as a systematic approach for developing disease-specific assays: identification of a chronic lymphocytic leukaemia-specific assay for use in rituximab-containing regimens AC Rawstron, R de Tute, AS Jack and P Hillmen Haematological Malignancy Diagnostic Service (HMDS), Leeds Teaching Hospitals, Leeds, UK Depletion of disease below the levels detected by sensitive sustained remissions only occur in patients achieving an MRD- minimal residual disease (MRD) assays is associated with negative complete response.12 Therefore MRD is increasingly prolonged survival in chronic lymphocytic leukaemia (CLL). being used as an end point for therapeutic trials, and several Flow cytometric MRD assays are now sufficiently sensitive and rapid to guide the duration of therapy in CLL, but generally rely studies are now using the assessment of MRD to define the on assessment of CD20 expression, which cannot be accurately duration of therapy. measured during and after therapeutic approaches containing Approaches using allele-specific oligonucleotide polymerase rituximab. The aim of this study was to use analytical software chain reaction (ASO-PCR) to the immunoglobulin gene of the developed for microarray analysis to provide a systematic B-CLL cell are generally accepted to show the highest sensitivity approach for MRD flow assay development. Samples from CLL for MRD detection. However, more recent four-colour ap- patients (n ¼ 49), normal controls (n ¼ 21) and other B-lympho- proaches show sensitivities nearing that of ASO-PCR6,11,13 with proliferative disorders (n ¼ 12) were assessed with a panel of 66 antibodies. The DNA-Chip analysis program was used to lower expense and more rapid provision of results. These identify discriminating antibodies, with hierarchical cluster approaches are dependent on the differential expression of analysis to identify complementary combinations. An iterative proteins by CLL and normal B-cells, particularly CD5, CD79b process was used: increasing numbers of patients were and CD20. CD20 provides the best separation of normal mature assessed with smaller, more targeted antibody panels until a B cells from CLL cells, as the former have very strong highly specific combination (CD81/CD22/CD19/CD5) was identi- homogeneous expression, whereas CLL cells show weak fied. This combination was as sensitive and specific as 11 previously reported assays and potentially applicable to blood expression. However, treatment with rituximab-containing and marrow samples from patients treated with current thera- regimens results in masking of the CD20 epitope and/or peutic approaches including rituximab. This approach to the downregulation of the protein on the cell surface of normal identification of disease-specific antibody combinations for MRD B-cells.14–16 This makes it impossible to differentiate normal analysis is readily applicable to a variety of haematological B cells from CLL cells on the basis of CD20 expression. The disorders. apparent lack of CD20 expression by normal B-cells may Leukemia (2006) 20, 2102–2110. doi:10.1038/sj.leu.2404416; published online 19 October 2006 continue for several months. As such, current MRD-flow Keywords: hierarchical clustering; flow cytometry; minimal residual approaches involving CD20 analysis have a poor sensitivity in disease patients treated with rituximab. Alternative approaches have been identified, including the CD22/CD23/CD19/CD5 combination reported by Sanchez 17 Introduction et al. However, there is significant overlap between CLL cells and normal CD5 þ 23 þ B-cells which are particularly prevalent 11 Conventional therapy in chronic lymphocytic leukaemia (CLL) is after effective therapy. Therefore, in cases which have stronger not curative, partly because there is almost always detectable CD22 expression the CD22/CD23/CD19/CD5 combination is CLL after therapy. Recent therapeutic approaches, such as no longer disease-specific: ideally, CD22 would be combined rituximab combined with purine analogues,1,2 alemtuzumab as with other antibodies that maintain separation of CLL cells from consolidation3,4 or stem cell transplantation,5,6 are associated normal B-cells even when CD22 expression is near normal. with improved response rates with up to 70% of patients Identification of such synergistic combinations requires re- achieving an National Cancer Institute-complete remission after evaluation of previously identified markers in combination with 7,8 novel or rarely reported markers such as CD31, CD39, CD47, Fludarabine/Cyclophosphamide/Rituximab (FCR). In addition, 18–22 these approaches have been associated with eradication of CD63 and CD81. detectable disease in a significant proportion of patients. Several Ideally an MRD assay would also be applied to pretreatment/ studies demonstrate that patients with CLL who achieve an diagnostic samples, in order to ensure that each patient is eradication of detectable minimal residual disease (MRD) using suitable for monitoring with the specific assay. An extension of a highly sensitive technique have prolonged survival irrespec- this goal is that a diagnostic panel would comprise of the MRD tive of the therapy used to achieve this goal.9–11 Patients assays for each B-cell disorder. Therefore the first aim was to achieving an MRD-positive complete response show the same identify antibodies that could separate CLL cells from normal overall survival as those achieving only a partial response; B-cells as well as from other B-cell disorders. To achieve this we incorporated antibodies in our initial panel that are currently used in diagnosis of other disorders (e.g. CD11c and CD103 for Correspondence: Dr AC Rawstron, HMDS, Algernon Firth Building, Hairy Cell Leukaemia). The antibodies identified from the first Leeds General Infirmary, Leeds LS1 3EX, UK. E-mail: [email protected] round of screening could then be used for the development of Received 25 July 2005; revised 10 May 2006; accepted 17 May 2006; disease-specific assays for CLL as well as for other B-cell published online 19 October 2006 disorders. The simultaneous analysis of large numbers of Protein expression profiling in MRD analysis AC Rawstron et al 2103 antigens is complex but recent studies have demonstrated that it Dilutional studies were performed as reported previously.11 is possible to use the software developed for messenger RNA Briefly, leucocytes from a CLL patient were mixed with an microarray analysis to facilitate this approach.23 We have used approximately 10-fold excess of FITC calibrate beadsBD, and this software to provide a systematic approach for MRD flow leucocytes from a normal bone marrow were mixed with an assay development, specifically a combination that is suitable approximately equivalent number of PE calibrate beadsBD, and for detection of CLL cells in patients undergoing therapy with the cell:bead ratio was calculated for both before dilution. The rituximab. sample containing CLL cells was then diluted in serial 1:4 dilutions into normal leucocytes. Samples were stained as above with CD20 FITCIT, CD79b PEIT, CD19 PE-Cy5.5H, and CD5 Methods APCBD or CD81 FITCBD, CD22 PEBD, CD19 PE-Cy5.5H and CD5 APCBD. The excess of FITC beads allows an accurate calculation Patients of actual test:diluent cell ratio, even at high dilutions, and Samples were assessed from patients with CLL (marrow n ¼ 34, independent of pipetting errors. lymph node n ¼ 1 and blood n ¼ 14), Waldenstroms macro- For the assessment of antigen expression during in vitro globulinaemia (marrow n ¼ 3), mantle cell lymphoma (marrow exposure to rituximab, leucocytes from a patient with marginal n ¼ 2, blood n ¼ 1), follicular lymphoma (lymph node n ¼ 2, zone lymphoma (B cells 85% of leucocytes with strong CD20 marrow n ¼ 1), diffuse large cell lymphoma (lymph node n ¼ 1, expression) were prepared by ammonium chloride lysis as marrow n ¼ 2) and normal controls (marrow n ¼ 15, lymph node described above. Cells (106) were incubated with 100 mlof n ¼ 3 and blood n ¼ 3). For neoplastic cases, disorders were therapeutic antibody at serial 1:2 dilutions in FACSFlow from diagnosed according to standard World Health Organization 1:10 to 1:5120 for 20 min. Cells were then washed twice and (WHO) criteria; patients were at presentation for non-CLL incubated with 10 ml of CD20 FITCIT and 10 ml CD19 PE-Cy5H; disorders. CLL patients were either at presentation (n ¼ 13) or complete abrogation of CD20 binding occurred at dilutions up relapse at least 3 months from prior therapy (n ¼ 36). In all cases to 1:500. Leucocytes from the peripheral blood of six patients the samples showed complete replacement of the relevant with active disease (499% of B cells were CLL phenotype and B-lineage compartment with neoplastic cells. All samples were the absolute B-cell count was 410 Â 109/l) were prepared by taken with full-informed patient consent for investigation of a ammonium chloride lysis as described above. For each patient, known or suspected haematological disorder. Treatment with three tubes containing 106 cells were incubated with 100 ml the FCR regimen was performed as reported previously.8 of therapeutic antibody at a 1:100 dilution, and three tubes Samples were anonymized before extended analysis was containing 106 cells were incubated with 100 ml FACSFlow for performed, according to the requirements of the local ethical 20 min. The cells were washed twice,