Levels of residual disease (RD) are an independent predictor of progression-free survival (PFS) and overall survival (OS) in patients treated for chronic lymphocytic leukemia (CLL). We modified the international standardized approach (ISA) to RD detection using flow cytometry by developing a single tube 10 color antibody assay.
A single tube incorporated the following monoclonal antibodies: CD81FITC, CD22PE, CD3ECD, CD5PercP5.5, CD20PECY7, CD79bAPC, CD38A700, CD43APC Alexa750, CD19eFluor 450, and CD45KO. A modified ISA gating strategy was developed that removed contaminating events. Sensitivity assays were performed using dilution with normal peripheral blood and bone marrow. Clinical samples were compared using the ISA and the single tube assay.
Dilution studies showed that sensitivity of 0.001% was achievable when a minimum of 1.8 × 106 total events were acquired. One hundred twenty-nine samples were analyzed and showed RD levels from 0.003 to 22%. In 80 samples analyzed with both assays, there was an excellent correlation between the two methods (slope = 1.0, intercept = 0.07 and R2 = 0.992) and results from Bland–Altman analysis showed a bias of 0.04 ± 0.38 with 95% confidence interval of −0.71 to 0.79. Removal of contaminating events in the single tube assay led to a significant reduction in RD values (P = 0.0014).
The single tube 10-color assay for the detection of RD in CLL provides equivalent results to the ISA but requires fewer cells, uses fewer reagents, and allows for simpler analysis. By directly removing contaminating events, it improves the accuracy of CLL RD detection and may reclassify the status of some patients following chemotherapy. © 2013 International Clinical Cytometry Society
The eradication of minimal residual disease (MRD) following treatment for chronic lymphocytic leukemia (CLL) predicts for improved outcome irrespective of the therapy used to achieve this goal (1–3). Patients achieving MRD negativity determined by highly sensitive flow cytometry or polymerase chain reaction demonstrate prolonged progression-free survival (PFS) and overall survival (OS). Morphological response criteria alone fail to predict for these improved outcomes (4–6). Recent advances in chemoimmunotherapy make the use of PFS and OS problematic for timely analysis of trials involving new therapies. As a result, although the level of MRD achieved is not included in the standard iwCLL definition of clinical remission (CR) (7), it is increasingly used as a secondary or even primary endpoint in clinical trials (1).
There are two major approaches used for measuring MRD in CLL. Allele-specific oligonucleotide PCR (ASO-PCR) of the immunoglobulin gene of the B-CLL clone is generally accepted to show the highest sensitivity for MRD detection (8, 9) and considered to be the gold standard. However, ASO-PCR is labor intensive and can only be performed on patients that have a pre-treatment sample available. Results are not generally available rapidly enough for real time decision making. An alternative technique is multiparameter flow cytometry which has the advantages of rapid analysis without requirement for patient-specific reagents. Sensitivity varies but may approach that of ASO-PCR (8). Historically, CD19+CD5+ co-expression with demonstration of clonality by immunoglobulin light chain restriction has been the simplest and most commonly used flow cytometry method for evaluation of MRD in CLL. However, this approach lacks sensitivity and is limited by the inclusion of normal hemopoietic cells and the inability to demonstrate light chain restriction at very low B cell numbers (10). Other approaches have looked at the differential expression of cell surface markers such as CD5, CD20, and CD79b (11) and combinations including CD22, CD23, CD19, and CD5 (12) but these are complicated by treatment effects and phenotypic overlap with normal cells (5, 10, 13).
More recently, the European Research Initiative on CLL (ERIC) proposed an international standardized approach (ISA) and identified three monoclonal antibody combinations (CD5/CD19 with CD20/CD38, CD81/CD22, and CD79b/CD43) for the detection of MRD in CLL. The authors reported a close correlation and 95% concordance with ASO-PCR for the detection of CLL above 0.01% (10−4) using a method applicable to all sample types and therapeutic regimens (14). The CLL8 study of the German CLL Study Group identified three levels of MRD (<10−4, 10−4 to 10−2 and >10−2) with prognostic significance following chemotherapy for previously untreated CLL (1). To achieve optimal accuracy in MRD reporting using the ISA method, 500,000 events must be acquired in at least two of the three test samples. However, post-treatment lymphopenia may be a limiting factor impeding the achievement of this goal. Less often, contamination defined as co-expression of CD19 by CD3 expressing cells may prevent attainment of the sensitivity required for MRD reporting at the 10−4 level.
The Australasian Lymphoma and Leukemia Group recently commenced a trial of lenalidomide consolidation therapy in patients with detectable MRD after completion of the chemoimmunotherapy for CLL (ALLG CLL6). In order to maximize accuracy of MRD detection in this trial, all samples are tested in a central flow cytometry facility. We developed a 10-color single tube flow cytometry assay based on the ISA methodology to optimize the accuracy and sensitivity of MRD detection. Placing all of the informative markers in one tube reduces the total number of lymphocytes required for analysis in patients with lymphopenia following chemotherapy and facilitates direct exclusion of contaminating T cells during cell analysis. The assay was established and intended for use solely by a reference laboratory experienced in multiparameter flow cytometry for CLL in the setting of a multi-institutional clinical trial.
MATERIALS AND METHODS
Patients, Treatment Regimens, and Samples
The study population consisted of PB (n = 94) and BM (n = 35) from 58 patients with median age 61 years (range 41–78 years), male:female ratio of 43:15 with a diagnosis of CLL according to the NCI-working group criteria. A total of 129 samples were analyzed at various time points post-treatment for CLL. Treatment protocols consisted of fludarabine alone or in combination with cyclophosphamide and/or Rituximab (n = 52). Six patients were analyzed at various time points after reduced intensity allogeneic stem cell transplantation. As outlined in previous studies of CLL, residual disease levels of <0.01%, 0.01–1%, and >1% were taken to indicated levels of residual disease with significantly different prognostic outcomes (1). Normal blood and BM were obtained from bone marrow transplant donors. In all cases, donors provided informed consent for use of their samples. Assessment was approved by the Human Research Ethics Committee of the Sydney West Area Health Service.
To compare the ISA approach with the single tube 10-color assay, the following antibodies were purchased from Becton Dickinson (BD Biosciences, Australia): CD20 FITC, CD3 PerCp5.5, CD45 FITC, CD19 APC, CD79b PE, CD38 PE, CD81 FITC, CD20 PECY7, CD79b APC, CD5 PerCp5.5, and CD14 PE. Antibodies purchased from Beckman Coulter (Immunotech, Marseilles, France) were CD3 ECD, CD20 FITC, CD43 APC Alexa 750, and CD45 Krome Orange. The following antibodies were purchased from Biolegend (Australian Biosearch) CD22 PE, CD43 FITC, and CD38 A700. CD19 efluor 450 was purchased from eBioscience (Jomar Bioscience, Australia, New Zealand). The panels were constructed according to the published ISA approach (14), i.e., tube 1: CD45 FITC, CD14 PE, CD19 APC, CD3 PerCp 5.5; tube 2: CD81 FITC, CD22PE, CD19 APC, CD5 PerCp5.5; tube3: CD20 FITC, CD38 PE, CD19 APC, CD5 PerCp5.5; tube 4: CD43 FITC, CD79b PE, CD19 APC, CD5 PerCp5.5. The single tube 10-color panel consisted of the following antibodies CD81 FITC, CD22 PE, CD3 ECD, CD5 PercP5.5, CD20 PECY7, CD79b APC, CD38 A700, CD43 APC Alexa 750, CD19 eFluor 450, and CD45 KO.
For both methods ammonium chloride lysis was performed by incubating whole blood or bone marrow with a tenfold excess of ammonium chloride (8.6 g/L in distilled water) prior to the addition of antibodies. Briefly, 10 × 106 cells were incubated with ammonium chloride for 5 min at 37°C centrifuged and washed twice with phosphate buffered saline (PBS). The cells were resuspended in 500 μL of PBS and 100 μL aliquots were then stained with the appropriate volume of pre-titred antibodies (according to the panels described above) for 10 min at RT in the dark, samples were washed twice in PBS and resuspended for acquisition. Sample acquisition was performed on a 10-color Gallios flow cytometer (Beckman Coulter) and analysis was performed using Kaluza version 1.1 software (Beckman Coulter). The compensation matrix for the single 10-color tube was performed according to the manufacturer's instruction. There was no difference identified comparing red cell lysis using ammonium chloride lyse and BD Pharm Lyse™ (data not shown).
A total of 129 CLL samples were analyzed. A subset of 85 samples were analyzed using both the ISA and single tube methods. Gating and analysis were performed according to the ISA approach using the limit of detection (LOD), i.e. results were classified as positive if there were more than the minimum number of CLL cells (50) in at least two of the three MRD test tubes and the average proportion of CLL cells was above 0.01% and above the LOD for that sample (i.e., the contaminating CD19+ CD3+ T cells). A minimum of 500,000 total events were targeted for all tubes. An identical strategy was used for analysis of the single tube assay. There was no difference in values obtained with the single tube assay when 20 and 50 events were taken to define a residual CLL population (data not shown). Therefore, in all subsequent analyses including sensitivity assays a minimum of 20 CLL cells that formed a discrete cluster was used to define the CLL population. The assay was reported as follows: for levels above 0.01%, the value of residual disease was reported; for levels below 0.01% that could be quantified, residual disease was reported as <0.01% but the value was given with a comment indicating that the clinical meaning of this value is uncertain; for levels below 0.01% not satisfying criteria for a cluster (i.e., less than 20 cells), residual disease was reported only as <0.01%; where a total of less than 200,000 events was accumulated, a report was issued saying that the sample was inadequate for analysis.
Determining the Level of Sensitivity for the Single Tube 10-Color Assay
Serial dilutions were prepared by mixing blood from a known CLL patient with over 90% CLL cells in the blood and bone marrow with blood from normal donors such that the CLL cells represented 0.001%–0.1% of the total leukocyte population (this consisted of three dilutions 1 in 10−3, 1 in 10−4, and 1 in 10−5 for both blood and bone marrow). To achieve an MRD level approaching 0.001%, it was necessary to acquire at least 1.8 ×106 total leucocytes per sample. To assess the overlap of normal CD19+ CD5+ and CLL cells, blood from 20 normal donors were also stained using the single tube 10-color assay, 3.0 ×106 cells were labeled with the antibody cocktail and a minimum number of 1.5 × 106 total events were acquired per sample.
Statistical analysis was performed using SPSS (IBM, USA) and GraphPad Prism software (San Diego, CA) for Pearson correlation coefficients, Mann–Whitney t-test, and Bland–Altman, significance was set at <0.05%.
Gating Strategy for the 10-Color Assay
To minimize the level of contamination and reduce intra laboratory variability, a specific gating strategy for the 10-color assay was developed. An initial CD45 versus side scatter plot with a gate placed on all CD45 positive events excludes debris and dead cells. The CD45 positive cells represent the total leucocyte population and the number of events in this gate (F) equals the denominator (Fig. 1A, plot 1). A gate (S) is placed on the lymphoid population (Fig. 1A, plot 1) and events in this gate examined for the expression of CD3 versus CD5 (Fig. 1A, plot 2): the CD3− population (gate AA) contains the CLL cells as well as normal B cells and NK cells; within this population CLL cells can be identified by their expression of CD5 (gate D). The co-expressing CD3+CD5+ population represents residual T cells and contains contaminating CD19+CD3+ events. CD3 negative cells (Fig. 1A, plot 2, gate AA) were analyzed for CD 19 expression versus side scatter (Fig. 1A, plot 3) and CD19 positive cells were gated in a forward versus side scatter gate to exclude doublets (Fig. 1A, plot 4, gate AD). These events (gate AD) were analyzed according to the ISA protocol (14) (Fig. 1B). The CLL cells are identified by their altered staining patterns compared to normal B-cells. Cells falling inside the gated regions are residual CLL cells and those cells falling outside the gates are normal B-cells. The residual CLL cells (Fig. 1C) are enumerated by combining the appropriate gates (Fig. 1B, plots 1–6) using Boolean logic. MRD is calculated by dividing the CLL cells (Fig. 1C, gate E) by the number of total leukocytes (Fig. 1A, gate F). This gating strategy was tested on a series of 20 samples by three operators all experienced in the analysis of list mode data files. There was no significant difference in the results obtained for all data files examined P = 0.95, the proportion of CLL cells ranged from 0.01 to 0.09%.
Utility of the Single Tube 10-Color Assay in Determining MRD
Using the single tube 10-color assay, levels of residual disease in the 129 samples analyzed varied from 0.003 to 22%. Twenty-five samples (19%) showed residual disease >1%, 57 samples (44%) showed residual disease between 1.0 and 0.01%, and 47 samples (37%) showed residual disease below 0.01% (Fig. 2). In seven samples from which a minimum of 1.8 × 106 events were acquired, it was possible to quantify residual disease levels between 0.003 and 0.01% (Fig. 2). Paired peripheral blood and bone marrow samples were obtained from 12 patients following fludarabine, cyclophosphamide, and Rituximab (FCR) therapy. In nine patients, MRD was detected in both peripheral blood and bone marrow, whereas three patients showed residual disease in bone marrow only. Bone marrow samples (n = 4) and peripheral blood (n = 7) from six patients at various time points post allogeneic stem cell transplant (SCT) were also examined for residual disease. Eight samples showed levels above 0.01% and three (all from peripheral blood) samples were <0.01%. Only one of the latter patients had a corresponding bone marrow sample taken and the MRD level was 0.3%. Two patients were monitored for MRD at regular time intervals over a period of 18 months to assess the robustness of the assay for clinical use. Patient 1 (Fig. 3A) was a 57-year-old man who received a second course of six cycles of FCR between April and September 2009. He remained in clinical CR and low but stable levels of MRD were detected from January to September 2011. Subsequently, there was a progressive rise in MRD levels in blood in association with reappearance of splenomegaly. Patient 2 (Fig. 3B) was a 69-year-old man previously treated with oral chlorambucil and in 2009 with three cycles of FCR. From January to October 2011 increasing levels of disease were detected in blood in the presence of a stable lymphocytosis. From November 2011 to March 2012 he received five cycles of FCR because of soft tissue infiltration by CLL cells in the conjunctivae and sinuses.
Comparison of the ISA Approach with the Single Tube 10-Color Assay
To validate the single tube 10color assay, we directly compared results from a cohort of 85 samples analyzed using the ISA method performed according to the standard operating procedure described by Rawstron et al. (14) with results obtained using the single tube 10-color method. Of the 85 samples, five (5.8%) could not be evaluated by the ISA methodology and were excluded from the validation study: four samples had insufficient events to obtain a sensitivity of 0.01% and in one sample the level of contamination due to the co-expression of CD19 and CD3 exceeded the residual CLL population. Of these, only one sample could not be evaluated by the 10-color assay because the sample quality was too poor. The level of residual disease in the 80 samples analyzed varied from <0.01 to 22%. Analysis of all samples showed an excellent correlation between the single tube 10-color assay and the ISA method (regression analysis slope = 1.0, intercept = 0.07 and R2 = 0.992). Bland–Altman analysis showed a bias of 0.04 ± 0.38 with 95% confidence interval from −0.71 to 0.79 (Fig. 4A). Fifty-nine samples showed residual disease of <1.0% (median 0.03 range <0.01–0.63%). There was also excellent correlation for disease levels in this cohort (regression analysis slope = 0.85, intercept = 0 and R2 = 0.98). Bland–Altman analysis showed a bias of 0.0016 ± 0.013 with 95% confidence interval from −0.024 to 0.028 (Fig. 4B). These data indicate that the two methods give comparable results at all levels of residual disease.
Assessing the Level of Contamination Within the Residual CLL Population
Samples from the comparative study, whose MRD levels were below 1%, were assessed for contamination by quantifying the number of events that co-expressed CD19 and CD3 positive events. In 59 samples, the mean contamination rate was 0.01% of total leucocytes (95% confidence interval 0.004–0.017%, mean number of total leukocytes acquired 420,053, range 352,845–582,000). Using the 10-color assay it was possible to determine the contribution of the contaminating events in the CLL MRD population. Of the 59 samples, the contaminating events could be as high as 90% of the CLL population, mean of 15% (range 0.02–90%). There was a significant difference between the CLL MRD values when the contaminating events were removed by gating with the mean CLL MRD by the ISA method 0.076% (range 0.01–0.63%) compared to the mean CLL MRD of the 10-color assay 0.065% (range 0.003–0.52%), P = 0.0014. Three out of 59 (5%) samples were determined to be MRD positive (CLL >0.01%) by the ISA method but when contaminating events were removed by gating by the 10-color assay, MRD levels fell below 0.01%.
Determining the (Threshold) Level of Sensitivity for the 10-Color Assay
Normal blood (n = 3) and bone marrow (n = 3) were mixed with CLL cells to mimic MRD samples at levels that ranged from 0.001 to 0.1%. The total number of events acquired ranged from 1.5 to 1.8 ×106 and the minimum number of events to define a CLL population was set at 20. By removing the contaminating events (events that bind CD19 as well as CD3 antibody), levels of MRD ranging from 0.001 to 0.1% could be obtained for each corresponding sample (Fig. 5). We also assessed 20 normal peripheral blood samples to see whether the assay was able to discriminate between normal CD19+ CD5+ cells and CLL cells. It was necessary to acquire between 1.5 and 1.8 × 106 total leucocytes to achieve sensitivity below 0.01%. Utilizing the same template for analysis as for CLL MRD, there was no overlap between normal CD19+ CD5+ cells and CLL cells. These results suggest that if sufficient events are acquired the 10-color assay has the potential to quantify one CLL cell in 100,000 (10−5) total leucocytes almost approaching the level of sensitivity for ASO-PCR.
Internal Quality Control
Internal data generated from the single 10-color assay allows validation of instrument set up and sample preparation. CD45 expression validates scatter characteristics and estimates total leukocyte population. Gating on CD45 lymphoid gate allows a quick confirmation of the proportion of T, B, and NK cells within the sample, i.e. CD3+ CD5+ T cells versus CD3− CD5− non CLL, B, and NK cells, respectively (Fig. 1A, plot 2). A good separation between the CD3− CD5− cells and the CD3− CD5+ allows easy discrimination between the normal B cells and the residual CLL cells.
Because of the prognostic impact of MRD status, NCI guidelines (7) have recommended that detection of MRD be included in current clinical trials of treatment in CLL. Effective regimens containing CD20 antibodies plus chemotherapy are known to render patients leucopenic, potentially resulting in insufficient cell numbers for the detection of MRD by flow cytometry. As the centralized flow cytometry laboratory providing results for a multicenter clinical trial of lenalidomide treatment for MRD following FCR chemotherapy, we developed a single tube 10-color assay based on the ISA approach. We included in one tube all antibodies contained in the ISA method and developed a sequential gating strategy based on ISA principles that allowed rapid removal of contaminating events during the analysis of MRD. Spiking experiments suggested that the 10-color assay had a sensitivity down to 0.001% and in contradistinction to assays using multiple tubes with fewer antibodies (15), effectively distinguished CLL cells from normal CD5 expressing B cells in blood and bone marrow. The gating strategy also permitted a rapid assessment of common lymphocyte subpopulations, thus acting as a form of internal sample quality control. The gating strategy was established to monitor residual disease in a population of patients with typical CLL and we have not analyzed cases of atypical CLL using this method. Since all markers are included in a single tube, we anticipate that modification of the gating strategy based on the diagnostic phenotype should permit effective analysis of residual disease. Such cases would need to be individually assessed and an appropriate gating strategy defined.
Recent publications of initial therapy of previously untreated patients with CLL have shown that levels of post-treatment residual disease of below 0.01%, 0.01–1%, and over 1% have prognostic significance in terms of both PFS and OS (1, 16). Direct comparison of samples at all levels of residual disease showed excellent correlation between the single tube 10-color assay and the 4 tube 4-color ISA assay. We were unable to evaluate 4.7% (4/85) of samples by the ISA method due to low cell numbers, a finding similar to that of Böttcher et al. (17) who found 5.7% of samples had inadequate cell numbers to achieve a sensitivity of 0.01%. Since the ISA method requires a total of 107 cells, it may be particularly problematic in patients recovering from recent chemoimmunotherapy who are frequently cytopenic. The single tube 10-color assay requires less than 2 × 106 cells to achieve a sensitivity below 0.01%. We were able to acquire sufficient events to achieve a sensitivity of 0.01% in all but one sample.
In contrast to the ISA method in which CD3 and CD19 antibodies are placed in a tube separate to antibodies that quantify MRD, the single tube assay permits contaminating CD3+ CD19+ cells to be directly gated out of the analysis. We observed that a mean of 15% (and up to 90%) of events reported as MRD by the ISA were in fact due to contamination. With median contamination of 0.007% (and values up to 0.028%), the clear implication is that the ISA may overestimate the degree of MRD. In confirmation, our data show that the single tube assay significantly lowered the level of MRD detected. At a level of MRD below 0.01% (arbitrarily defined by the international consensus guidelines for CLL as MRD negative (7)), the single tube assay reclassified MRD disease status in around 5% of cases. Use of the single tube assay thus provides prognostic information that could prevent unnecessary therapy for patients with MRD close to the LOD of the assay.
We studied several patients in a longitudinal fashion and observed results that were reproducible and consistent with clinical progress over up to 18 months of observation. These findings support the notion that the single tube assay is suitable for assessment of MRD and disease stability when used in a repeated fashion in a single flow laboratory. We acknowledge that despite its advantages, the single tube assay cannot at this time be recommended as a realistic widespread alternative to the current ISA method for assessing CLL MRD. Even assuming widespread availability of 10-color flow cytometers and experienced operators, validation of the assay across multiple sites would be essential for more generalized introduction of this technology. However in its current form in a centralized and experienced flow cytometry laboratory, the 10-color single tube assay permits the most accurate flow based CLL MRD detection currently available.