How to cite this article: Patkar N, Alex AA, Bargavi B, Ahmed R, Abraham A, George B, Vishwabandya A, Srivastava A, Mathews V. Standardizing minimal residual disease by flow cytometry for precursor B lineage acute lymphoblastic leukemia in a developing country. Cytometry Part B 2012; 82B: 252–258.
In addition to standard risk criteria at diagnosis, minimal residual disease (MRD) following initiation of therapy is a well-recognized risk factor to predict relapse. Literature from developing countries addressing therapeutic or laboratory practices related to MRD, is largely lacking. In a first paper from India, we describe our experience in establishing a flow cytometry-based MRD assay for precursor B lineage ALL (BCP-ALL) with emphasis on the assay standardization and cost.
Normal templates for B cell development were established in 10 control patients using CD45, CD11a, CD38, CD20, CD10, CD19, CD58, CD34, CD123, and CD22. BCP-ALL samples (n = 42) were characterized at diagnosis to identify a suitable marker for follow-up during mid (D+21) and end of induction (D+33). Both, multiparametric immunophenotyping and single marker detection of LAIP were used for data analysis.
In 95.2% of BCP-ALL at least two informative markers could be obtained when a minimum of four cocktail combinations were used. The combination CD20, CD10, CD45, and CD19 was the most useful (71.4%) followed by combinations containing CD38 (66.7%), CD22 (57.1%), CD11a (52.4%), and CD58 (33.3%). Using our approach, 60 and 47% of patients had detectable MRD at mid and end induction time points, respectively.
Standard prognostic variables for acute lymphoblastic leukemia (ALL) include age, WBC count, immunophenotype, and cytogenetics (1–6). Numerous studies in the last two decades have now demonstrated that minimal residual disease (MRD) is an important adjunct in addition to conventional risk factors (4, 6–8). From ongoing or completed trials which are modifying therapy based on MRD risk, it would appear that MRD directed therapy will become the standard of care for ALL in the coming years (9, 10).
However, literature from resource poor developing countries, which addresses therapeutic or laboratory practices related to MRD, is to the best of our knowledge lacking. It is desirable that MRD analysis be performed on these “resource limited” patients as MRD favorable patients who are favorable by other conventional criteria (e.g., cytogenetics, NCI risk) would be prime candidates for therapy de-escalation. The first step in achieving this goal would be developing a cost-effective assay. Although sensitive, PCR-based methods are labor intensive and more expensive as compared to flow cytometry (9, 11). From a developing country point of view, a PCR-based approach also places logistical hurdles as custom synthesized patient-specific primers may not be available locally and may not be imported into the country and validated, especially before early MRD time-points are reached.
Unfortunately, there is a paucity of consensus guidelines which recommend a common panel of antibodies for MRD analysis resulting in differing laboratory approaches (7, 12–16). Some of these are expensive and difficult to standardize. New entrants in this field are left with much uncertainty about the appropriate panel, methodology, and its adaptability to their system.
In this manuscript we describe our experience in establishing a flow cytometry-based MRD assay for precursor B lineage ALL (BCP-ALL) in a single tertiary referral center in southern India. We describe in detail the methodology, expenditure, assay standardization, and validation procedures. We hope that more centers in developing countries are encouraged to develop MRD in their centers for improved prognostication and to further rationalize therapy.
This study was carried out in the Department of Hematology, Christian Medical College, Vellore, Tamil Nadu, India from December 2010 till May 2011 after approval from the institutional ethics committee.
Control and Patient Samples
After obtaining informed consent from the patients or their relatives, control samples and all consecutive patients with BCP-ALL who opted for diagnosis or treatment were enrolled in the study. For controls (n = 10), bone marrow (BM) samples (2 ml of heparinized BM) were obtained from patients undergoing a BM aspiration as a part of routine investigation for nonmalignant hematological disease (for example, idiopathic thrombocytopenia) or lymphoma for staging (proven to be uninvolved on bone marrow biopsy). Additionally, patients of T-lineage ALL whose bone marrows were aspirated for end-induction morphological evaluation of residual disease were used as controls (n = 2) for studying B cell development. Under similar conditions, bone marrow samples were obtained in the patient group for establishing the diagnostic leukemia associated immunophenotype (LAIP). The control age ranged from 3 to 41 years (median 19 years). For flow cytometric evaluation of residual disease BM samples were acquired at mid (D+21) and end induction time points (D+33). The median age of patients with BCP-ALL was 14.5 years (range 1–72 years and M:F = 0.9:1).
Sample Preparation and Data Acquisition
Samples were prepared using gentle processing techniques and addressing specific issues related to MRD (14). Samples were processed as soon as possible, always within 24 h of receipt. Briefly, red cells were lysed using a freshly prepared in-house ammonium chloride-based RBC lysis solution (17), followed by washing cells once in cell-culture grade phosphate buffered saline [(10× Dulbecco's PBS Salt Solution, diluted with sterile cell culture grade water), (both obtained from Pan Biotech, Germany)] which was filtered with a 0.2-μ filter (Minisart NML, Sartorius Stedium, France). To block nonspecific binding, these cells were incubated for 5 min with 50 μl of human AB blood group positive serum. Following this the cells were stained with the monoclonal antibody combinations seen in Table 1. The appropriate concentrations of these antibodies were determined by titration experiments. All antibodies were purchased from BD Biosciences, USA. Following incubation for 15 min in the dark the samples were washed in PBS and acquired in a four color FACSCalibur flow cytometer (BD Biosciences, USA). For establishing the diagnostic LAIP 50,000 events were acquired. For MRD samples (obtained at day+21 and end induction) and normal controls, 500,000 events were acquired per tube.
Table 1. Panel of Antibodies Used for Minimal Residual Disease Studies with Incidence of Aberrant Phenotypes
Tube 1 20/10/45/19
Tube 2 22/34/45/19
Tube 3 11a/10/45/19
Tube 4* 38/10/34/19
Tube 5* 58/10/34/19
Tube 6* 123/10/34/19
Tubes 1–5 were processed in all 42 patients (NA = not applicable).
Only eight patients evaluated, All antibody combinations have fluorochromes in the order of FITC/PE/PerCP/APC (except *-Per-CP5.5 used instead of Per-CP.).
Instrument setup and quality control was performed daily using CaliBRITE Beads (BD Biosciences, USA). Compensation and PMT performance was verified by monitoring the mean fluorescence intensity of CD4 on lymphocytes across all four channels every 15 days.
List mode data were acquired using Cell Quest Pro (BD Biosciences, USA) and analyzed using FCS Express Professional V3.0 (De Novo Software, Los Angeles, USA). Antigen expression was described as dim, intermediate or overexpressed depending upon their relation to normal counterparts based on standard recommendations (18).
Diagnosis of BCP-ALL
Patients were diagnosed with BCP-ALL according to the WHO 2008 criteria (3). Staining for cytoplasmic Igμ chains was not assessed in any of the cases.
Sequential gating was used for all the tubes. The initial gates broadly selected the lymphocyte and monocyte regions based on forward and side scatter characteristics. The next plot gated on CD19 positive low side scatter events (Fig. 2). For Tubes 1–3 multiparametric assessment of B cells was done in relationship to various antigens by looking at biaxial plots and comparing them to normal patterns as defined by the templates. The gating strategy used for Tube 1 was as described by Borowitz et al. (19). Briefly, the plots analyzed included CD45v/sCD10, CD10 v/sCD20, CD45 v/s CD19, and CD45 v/s CD20. Additionally, the relationship of CD10 positive events with other antigens was also determined. For Tube 2 aberrant overexpression of CD22 in CD34 positive B cells with or without under expression of CD45 (20, 21) was determined. Also, the relationship of CD34 positive cells with other antigens was looked at. Additionally, for Tube 3, aberrant expression of CD11a by B cells was determined (15, 22). For Tubes 4, 5, and 6 the gating strategy was as described by Irving et al. and Campana et al. (14, 16). This included gating on CD34 positive and CD34 negative B cells (CD19 positive) and looking at a biaxial plot of CD10 v/s the discriminatory antigen for that tube (e.g., CD38, CD58, and CD123).
Definition of LAIP
As defined by Irving et al “Antibody combinations in which the leukemic blasts fell into empty spaces, distinct from regions housing normal B cell progenitors were identified as useful LAIP” (12, 20). Every useful LAIP required a unique abnormally expressed marker for that tube. For example, decreased CD45 on both the CD45 vs. CD19 and CD45 vs. CD10 plots was considered LAIP = 1. In case two markers were abnormal for that tube for example underexpression of CD45 and overexpressed CD58, then the LAIP was 2. However it was decided that if a case did not have a useful LAIP in any of the combinations, an overlap with <50% of the normal template would be considered as useful. An attempt was made to obtain at least two LAIPs per sample.
MRD Positive Events
Samples were considered to be MRD positive if at least 50 clustered events were recorded, 500,000 events for that tube could be acquired and the quantification of MRD was >0.01% (16).
Leukemic samples (whose immunophenotype was previously characterized) were serially diluted in normal BM aspirates which demonstrated B cell hyperplasia (16, 19). For Tubes 1 and 2, a multiparametric immunophenotype was evaluated whereas for the rest only single marker abnormalities were analyzed. Serial dilutions were done to a maximum of 1:10,000 for CD45 (n = 2), CD38 (n = 2), CD123 (n = 1), CD22 (n = 1), and CD58 (n = 1). An example of the same can be seen in Figure 1.
The amount of expenditure that occurred (US Dollars) was audited. This analysis was limited only to the cost of the reagents and did not include additional expenses such as cost of the equipment, depreciation of the equipment, annual maintenance contract, salaries for medical and technical personnel. Specifically, the costs included cost of antibodies, PBS, sheath fluid and 0.2-μ filters. The antibody cost was broken down per microliter. Depending upon the tittered volume for each antibody and the combination per tube, expenditure involved in a 4 or 5 tube assays was generated. On summing up these numbers the total reagent cost that occurred per MRD time point was accomplished. Combinations of these tubes were evaluated on the original dataset to minimize the panel so as to make it cost effective (Table 2). Because data on CD123 were available only in eight patients this data was not evaluated for calculations in Table 2.
Table 2. Evaluation of Combinations of Tubes in Contributing to >2 LAIPs per Patients
n = 42
Tubes 1, 2, and 4 (%)
Tubes 1, 2, 4, and 5 (%)
Tubes 1, 2 3, 4, and 5 (%)
Generation of Normal Templates
The antigen expression of normal maturing B lymphoid cells was patterned so as to delineate the boundaries of normal. On analyzing a new BM control, expansion to the definition of normal was made. Variations in the number of cells at a particular stage of development (for example Stage 1 hematogones) were seen but the position of these cells with respect to other developing B cells was constant.
In seven out of eight dilution tests the tumor cells could be detected at sensitivity close to 1:10,000. For the CD58 dilution assay, a clear cluster of cells could not be defined at 1:10,000, possibly due to a technical error in serial dilutions of tumor in normal.
The diagnosis in the majority BCP-ALL was common ALL antigen (CALLA) positive ALL (90.5%), the rest were pro-B ALL (n = 4).Using this panel, a median of 3 LAIPs could be obtained per patient. In 95.2% (n = 40) at least two LAIPs could be obtained, in the remaining two only a single LAIP was possible. The types of aberrancies and their relative frequencies can be appreciated in Table 3, whereas the utility of each of the six tubes can be seen in Table 1. The first three tubes use CD45 and CD19 as common markers. We evaluated the utility of other markers and resulting combinations when CD45 was not contributory to the LAIP (n = 17). We found that Tubes 1 and 2 were more useful in contributing to a LAIP as compared to Tube 3 with a utility of 66.7, 50, and 33.3%, respectively.
Table 3. Relative Frequencies of Phenotypic Aberrations Seen on CD19 Positive Blasts at Diagnosis
Type of phenotypic aberration on CD19 positive blasts
Asynchronous CD22 expression in CD34 positive cells
Asynchronous relationship of CD45 and CD34
The expense incurred in a single MRD analysis is described in Table 4.
Table 4. Cost of MRD Analysis for Precursor B Lineage ALL
MRD assay using Tubes 1, 2, 4, and 5
MRD assay using Tubes 1, 2, 4, 5, and 6
Cost of PBS, sheath fluid and 0.2-μ filter
44 (54 with additional tube 6)
Mid and End Induction MRD
A total of 25 mid induction and 17 end induction samples were obtained. The remainder could not be obtained due to numerous reasons (not opted for treatment, institutional attrition, induction deaths, missed by investigator, or patient unwilling for an invasive procedure). Out of 25 D+21 samples, 15 (60%) were positive [0.01–0.1% (n = 6), 0.1–1 (n = 3), 1–5 (n = 2),>5 (n = 4)]. For D+33 samples 8 (47%) were positive [0.01–0.1 (n = 3), 0.1–1 (n = 4) and >5 (n = 1)]. All the end induction samples were MRD positive at mid-induction time points. The most useful tubes to establish a positive MRD were 1 (84%), 4 (56%), 5 (44%), and 2 (16%). An example of diagnostic BCP-ALL, D+21, and D+33 MRD can be seen in Figure 2.
In this manuscript, we have used two different approaches towards MRD analysis. The first three tubes utilize multiparametric immunophenotyping and the rest look for single marker LAIPs in early and late developing B cells. Numerous groups across Europe and North America have used either of these principles and have shown that their panels can detect LAIP in ∼ 90% of BCP-ALL (Table 5). Most of these have been validated to yield clinically useful information (4, 7, 15, 23, 24). One of the goals of this study was to compare these approaches so as to evaluate the feasibility of adapting them to our system.
Table 5. Frequencies of Detection of LAIP From Various Studies Across Europe and North America in Comparison to Our Data
Irving et al. found the under expression of CD45 to be useful in 75% of BCP-ALL whereas, Campana et al. and Krampera et al. (14,16,26,28) found similar results in 30–50% and 26.6% of BCP-ALLs. In our series underexpressed CD45 was seen in 48.8% of BCP-ALLs. Using multiparametric MRD detection, Weir et al. (19) found antibodies similar to Tube 1 to be useful in 93% of patients, whereas Lucio et al. (12) found combinations containing CD45, CD19, and CD34 to be informative in 22.2% of BCP-ALL. Vidriales et al. (26) stated that CD45, CD10, CD19, and anti-TdT was applicable to 70–80% of BCP-ALL. In our series, we found that Tube 1 was applicable to 71.4% of BCP-ALL at diagnosis and contributed maximally to the detection of MRD. Campana et al. found underexpression of CD38 has applicability in 30–50%, whereas Irving et al. found applicability in 63% and Krampera et al. in 57% of patients (14,16,28). In the BIOMED-1 standardization assay, Lucio et al. found CD38 aberrancies in 55.6% of BCP-ALL (12). We detected aberrancies of CD38 in 66.7% at diagnosis and this was a useful tube to detect MRD at both mid and end induction time points.
CD22 is weakly expressed by early CD34 positive B cell precursors (20,21,29) and aberrant strong expression was seen in 16.7% of our patients. As a part of multiparametric analysis, however, this tube was found to be applicable to 57% of our cases. CD22 when seen solely as a marker of aberrant expression seems to be of limited utility as shown by Campana et al. (20–30%), Irving et al. (5%), and Krampera et al. (0%) (14,16,28,29). Lucio et al., however reported a much higher applicability of 46.2% (12). Using CD22 in our series we were able to detect LAIP in 57.1% of patients using multiparametric immunophenotyping.
Tube 3 was applicable to 52% of cases. We then analyzed the utility of Tubes 1–3 when CD45 was not informative. On doing so, the applicability of these tubes dropped, the lowest being Tube 3 indicating that at least in some of the cases under-expression of CD45 influenced the use of that tube. However, it must be mentioned that the fluorochrome and antigens (CD45 instead of CD45-RA, and CD10PE instead of CD10FITC) used in Tube 3 differ from those used originally by Dworzak et al. (15,24,30). The possibility that this alteration could decrease the overall utility to contribute to LAIP cannot be ruled out.
The over expression of CD58 amongst leukemic cells of BCP-ALL has been described as a useful marker for LAIP (40–60% and 55% applicability by Campana and Irving et al., respectively) (16, 29). We found that the utility of this marker in our study was less useful (Table 1). The overexpression of CD123 has been described and utilized by Djokic et al. in 57% of BCP-ALL cases (31). Recently using gene expression profiling for recognizing candidate markers in BCP-ALL followed by flow cytometry, the overexpression of this antigen as compared to nonleukemic samples was seen in 50.7% of ALLs (32). Our preliminary analysis seems to suggest that this seems to be a useful marker for MRD studies.
Using a very innovative approach, Pedreira et al. (33) have recently described a unique probability dependent approach for the detection of MRD in B cell chronic lymphoproliferative disorders and discussed its potential for use in other hematological neoplasms. It would be most interesting to analyze the normal B cell populations and phenotypic aberrations that we observed in our dataset using this technique. Such a process would perhaps allow for easier detection of MRD positive events, especially in the context of >6 color immunophenotyping.
To make the assay affordable, on decreasing the panel of antibodies to just four tubes (Table 2), two LAIPs could be obtained in over 90% of our patients. Adding the CD11a containing tube did not seem to offer an additional benefit. Our current algorithm would be utilizing Tubes 1, 2, 4, and 5. Such an approach would allow MRD analysis at diagnosis and two follow-up time points to be done in $100–150. However, if only a single LAIP is obtained, then an additional Tube 6 would be added. These costs vary from one institute to another and are only meant to be a general guideline for developing countries. To the best of our knowledge, this is the first data from India on standardization of flow-based MRD for BCP-ALL. The clinical and genetic correlation for these cases is pending and data are being accrued for the same. We hope that this data would encourage more centers in India and other developing countries to develop a cost effective technique.