Flow Cytometric Characterization of CD71 Cell Subsets In

Leukemia (2000) 14, 816–825  2000 Macmillan Publishers Ltd All rights reserved 0887-6924/00 $15.00 www.nature.com/leu BIOMED-1 Concerted Action report: Flow cytometric characterization of CD71 cell subsets in normal bone marrow as a basis for the diagnosis and follow-up of T cell acute lymphoblastic leukemia (T-ALL) A Porwit-MacDonald1, E Bjo¨rklund1, P Lucio2, EG van Lochem3, J Mazur1, A Parreira2, MWM van den Beemd3, ER van Wering4, E Baars4, G Gaipa5, A Biondi5, J Ciudad6, JJM van Dongen3, JF San Miguel6 and A Orfao6

1Department of Pathology, Karolinska Hospital, Stockholm, Sweden; 2Instituto Portugues de Oncologia, Lisboa, Portugal; 3Department of Immunology, Erasmus University, Rotterdam, and University Hospital, Rotterdam, The Netherlands; 4Dutch Childhood Leukemia Study Group, The Hague, The Netherlands; 5Centro di Ricerca M Tettamanti, Monza, Italy; and 6Department of Hematology, University Hospital, Salamanca, Spain

The European BIOMED-1 Concerted Action was initiated in multiparameter flow cytometry for the investigation of MRD 1994 to improve and standardize the flow cytometric detection can reach a high sensitivity and specificity for the identifi- of minimal residual disease (MRD) in acute leukemia (AL). 8–11 Three different protocols were defined to identify the normal cation of residual leukemic T lymphoblasts. A cooperative subsets of B, T and myeloid cells in bone marrow (BM), and study conducted by six European centers involved in the were applied to the different types of AL in order to study BIOMED-1 Concerted Action ‘Investigation of Minimal aberrant immunophenotypes. Using sensitive acquisition Residual Disease in Acute Leukemia: International Standardiz- methods (‘live gate’) T cell subsets in normal BM could be ation and Clinical Evaluation’ applied multiparameter flow identified with five triple-stains: CD7/CD5/CD3, CD7/CD4/CD8, cytometry in which triple antibody combinations were used CD7/CD2/CD3, CD7/CD38/CD34 and TdT/CD7/surface or cytoplasmic (cy)CD3 (antibodies conjugated with FITC/PE/PECy5 or together with sensitive data acquisition methods (such as ‘live PerCP, respectively). The identification of T cell subsets in BM gate’ acquisition). Using this approach we obtained detailed allowed definition of ‘empty spaces’ (ie areas of flow cytometric information on cell subsets that are present in normal BM at plots where normally no cells are found). All studied T-ALL very low frequencies and that may remain undetected using cases (n = 65) were located in ‘empty spaces’ and could be conventional methods of analysis. discriminated from normal T cells. The most informative triple The immunophenotyping methods (staining protocols and staining was TdT/CD7/cyCD3, which was aberrant in 91% of T- ALL cases. In most cases, two or more aberrant patterns were flow cytometry) were carefully standardized in order to found. Apparently the immunophenotypes of T-ALL differ sig- develop a highly reproducible and sensitive approach to nificantly from normal BM T cells. This is mostly caused by identify both major and minor subsets of T cells present in BM their thymocytic origin, but also the neoplastic transformation and their relative distribution according to age. Once normal might have affected antigen expression patterns. Application T cell subsets were identified, we focused our attention on of the five proposed marker combinations in T-ALL contributes the ‘empty spaces’ left in multidimensional flow cytometry dot to standardized detection of MRD, since cells persistent or plots of each studied antigen combination, as has previously reappearing in the ‘empty spaces’ can be easily identified in 12–14 follow-up BM samples during and after treatment. Leukemia been reported for precursor B cells. Subsequently, we (2000) 14, 816–825. used the same triple antibody combinations and flow cytome- Keywords: T cells; bone marrow; acute lymphoblastic leukemia; try protocols to analyze 65 consecutive T-ALL cases diag- immunophenotype; flow cytometry; minimal residual disease nosed in the six participating laboratories. Our aim was to assess the incidence of aberrant phenotypes, which can be used for the follow-up of MRD in T-ALL patients. Introduction

It has long been assumed that blasts from T cell acute lym- Materials and methods phoblastic leukemia (T-ALL) reflect the immunophenotypes of early stages of T cell differentiation.1–3 Most T-ALL cases dis- Specimen collection play phenotypes corresponding to various stages of thymocyte differentiation, which normally are not found in bone marrow (BM) or peripheral blood (PB).4 Normal BM samples: Normal BM samples (n = 35) were The aberrant immunophenotypic characteristics of leu- obtained from healthy donors (n = 29) and patients with idio- kemic T lymphoblasts allow discrimination from T cell subsets pathic thrombocytopenia (ITP) (n = 6), recruited in the six Eur- in BM and thus can be used for detection of minimal residual opean centers involved in the BIOMED-1 Concerted Action disease (MRD).5–8 A comprehensive analysis of the immuno- on MRD. The following centers participated: Portuguese Insti- phenotypic characteristics of T cells and the subsequent tute of Oncology, Lisbon, Portugal; Karolinska Hospital, identification of all T cell subsets in normal BM are important Stockholm, Sweden; University of Salamanca, Salamanca, for the reliable distinction between normal and leukemic T Spain; Department of Immunology, Erasmus University, Rot- cells. terdam and University Hospital, Rotterdam, The Netherlands; Previous studies in T-ALL strongly suggest that the use of Dutch Childhood Leukemia Study Group, The Hague, The Netherlands; and M Tettamanti Research Center, Monza, Italy. Informed consent was obtained from all donors or parents Correspondence: A Porwit-MacDonald, Department of Pathology, in cases of children less than 15 years old in accordance with Karolinska Hospital, 171 76 Stockholm, Sweden; Fax: 46 851775843 J Mazur was on leave from Department of Epidemiology and Health local Medical Ethics Committees. The samples from patients Care Planning, Institute Mother and Child, Warsaw, Poland with ITP were obtained for diagnostic purposes. The median Received 14 June 1999; accepted 15 December 1999 age of the donors was 15 years, ranging from 2 to 75 years BIOMED-1 Concerted Action report A Porwit-MacDonald et al 817 old (mean age 24), the male/female ratio was 0.8. Moreover, studies showed that both intracellular staining procedures 16 follow-up BM samples from children receiving chemo- yielded similar results.14,15 therapy for B-precursor ALL were included to investigate if T For standardization purposes, we used triple combinations cell content and distribution changed during cytostatic of the same Mab clones labeled with identical fluorochromes treatment. (FITC/PE/PECy5 or PerCP) in all immunostainings of normal BM in all participating centers: CD7/CD5/CD3, CD7/CD4/CD8, CD7/CD2/CD3, CD7/CD38/CD34 and Leukemic samples: T-ALL samples (n = 65) were obtained TdT/CD7/cyCD3 (Table 1). Isotype-matched immunoglobulins in the participating centers for diagnostic purposes. The mor- and a sample stained for CD3-FITC/CD4-PE/CD8-PE/Cy5 were phological diagnosis of ALL was made on the basis of micro- used as negative and positive controls, respectively. Addition- scopic investigation of May–Gru¨nwald-Giemsa stained BM ally, expression of CD13, CD33 and CD19 within the CD7+ smears. The immunologic diagnosis of T-ALL was made cell population was studied in BM samples from a total of 24 according to well-established criteria.4 The median age of the normal donors in Salamanca. The immunostaining of leu- patients was 13 years, ranging from 1 to 52 years (mean 16), kemic samples was performed using the same triple antibody and the male/female ratio was 1.8. combinations and sample preparation procedures as for normal BM.

Sample preparation Data acquisition and analysis Bone marrow aspirates were collected in either heparin or EDTA anticoagulant and maintained at room temperature (RT) Data acquisition was performed in all centers in a stan- until processed (not longer than 24 h). The nucleated cell dardized manner, using FACScan flow-cytometers (BD), count was adjusted to 1–2 × 107 cells/ml. For sample prep- equipped with either LysisII or CellQuest (BD) software pro- aration, a stain and then lyse/wash technique was used. grams. Instrument set-up and calibration was performed in a Briefly, 100 µl of BM cell suspension (at least 1 × 106 standardized way as previously described.12 For the analysis nucleated cells) were incubated for 10–15 min (RT, in the of normal BM samples, a two-step acquisition procedure was dark) with saturating amounts of five different triple combi- used. First, 15 × 103 non-gated events were acquired from the nations of monoclonal antibodies (Mab, Table 1). After incu- BM samples. Then a total of 105 to 2 × 106 cells were meas- bation, 2 ml of FACS lysing solution (Becton Dickinson (BD), ured with a live gate set on the basis of CD7 expression and San Jose, CA, USA) diluted 1/10 in distilled water was added low/intermediate side scatter (SSC) characteristics. In this and samples were incubated for another 10 min at RT in the second step, data acquisition exclusively concerned CD7+ dark. Subsequently, cells were washed and re-suspended in 1 cells with lymphoid scatter generally representing approxi- ml of PBS. mately 10% of BM cells (Table 2). In practice, this meant that Cytoplasmic (cy)CD3 and nuclear terminal deoxynucleoti- between 5 × 104 and 2 × 105 CD7+ lymphoid BM cells were dyl transferase (TdT) were detected after staining for surface acquired for further evaluation. The expression of the other T markers (as above), and fixation/permeabilization with Perme- cell markers was analyzed within this CD7+ lymphocyte gate afix (OPM; Ortho, Raritan, NY, USA) or Fix & Perm (F/P; An in order to characterize the different CD7+ cell subsets. der Grub, Vienna, Austria).15 The reagents were used follow- Additional live gates were based on CD34+ and TdT+ events ing the manufacturer’s recommendations. After OPM fixation displaying low/intermediate SSC characteristics. and during permeabilization with F/P, the cells were incu- For all leukemic samples, 15 × 103 events were acquired bated with anti-TdT or CD3 Mab (30 or 15 min, respectively) from the total BM and, whenever appropriate, the expression then washed and resuspended in PBS. Previous comparative of various markers analyzed after gating focused on the leukemic blast cells. The fluorescence intensity of T cell markers observed in normal T-lymphocytes present in each analyzed Table 1 Five triple marker combinations used for T cell subset sample was used as the reference value for describing the analysis intensity of antigen expression in leukemic blasts.

Combination FIT-conjugates PE-conjugates PerCP or PECy5 (Tricolor) Standardization of the procedures and quality control conjugates Prior to the study, all procedures used for the immunopheno- I CD7 CD5 CD3 + (Leu9, BD) (Leu1, BD) (Leu4, BD) typic analysis of the normal compartment of the CD7 BM cells were standardized among the six participating centers II CD7 CD4 CD8 (Leu9, BD) (Leu3, BD) (3B5, Caltag) via collaboration in experimental work and regular laboratory meetings. The same antibody clones, fluorochrome conju- III/IV CD7 CD2 CD3 (Leu9, BD) (Leu5b, BD) (Leu4, BD) gates, and triple-staining combinations were used throughout the study in all six participating laboratories. The agreed anti- V TdT CD7 CD3 (HT6, Dako) (Leu9, BD) (Leu4, BD) body combinations were those providing the most powerful discrimination between the T cell subsets under study. Pro- VI CD7 CD38 CD34 (Leu9, BD) (Leu17, BD) (HPCA-2, BD) cedures for instrument set-up and calibration were carefully compared and one common well-standardized method was FITC, fluoresceine isothiocyanate; PE, phycoerythrin; PerCP, perdi- selected. The light scatter characteristics and the auto- din chlorophyll protein; BD, Becton Dickinson (San Jose, CA, USA); fluorescence levels of normal PB lymphocytes were used as Caltag, Caltag Lab (South San Francisco, CA, USA); Dako, reference values for instrument settings. As described above, (Dakopatts, Glostrup, Denmark). the same two-step acquisition procedure was used in all lab-

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 818 Table 2 Frequency of various T cell subsets present in normal bone marrow

Triple staining T cell subsetsa combination (FITC/PE/PercP 123456 alt PE/Cy5)

CD7/CD5/CD3 CD3−/CD5− CD3+/CD5− CD3+/CD5bright CD3dim/neg/CD5bright NA NA n = 35 I 14.4 (8)/ 1.09 (0.99)/ 83.7 (8)/ 0.7 (0.7)/ 1.42 (0.7) 0.14 (0.1) 8.28 (0.7) 0.069 (0.06)

CD7/CD4/CD8 CD4−/CD8− CD4−/CD8dim CD4−/CD8bright CD4dim/CD8bright CD4bright/CD8dim CD4+/CD8− n = 33 II 14.51 (8.2)/ 8.63 (5.2)/ 34.54 (8.6)/ 0.83 (0.8)/ 1.13 (1.6)/ 40.2 (10.3)/ 1.4 (0.8) 0.8 (0.5) 3.3 (0.8) 0.08 (0.07) 0.11 (0.15) 3.9 (0.99)

CD7/CD2/CD3 CD3+/CD2− CD3+/CD2+ CD3−/CD2−/CD7bright CD3−/CD2+/CD7bright CD3−/CD2+/CD7dim NA n = 26 III/IV 0.37 (0.25)/ 85.83 (8.1)/ 3.8 (3.7)/ 8.1 (4.8)/ 1.63 (2.1)/ 0.04 (0.03) 8.5 (3.4) 0.38 (0.46) 0.8 (0.54) 0.16 (0.2) CD3+/CD7+ (III) CD3−/CD7+ (IV) NA 86.12 (8.3)/8.56 (3.3) 14.12 (7.96)/1.42 (0.96) CD3+/CD2− b CD3+/CD2+ b CD3−/CD2−/CD7bright c CD3−/CD2+/CD7bright c CD3−/CD2+/CD7dim c 0.45 (0.4) 99.5 (0.45) 25.9 (11.4) 60.73 (13) 12.53 (12.7)

aSee Figure 1, results are expressed as percentages of gated CD7 positive cells and percentages of total BM cells, respectively (mean (s.d.)). b% of CD3+/CD7+ cells. c% of CD3−/CD7+ cells. NA, not applicable.

oratories and data analysis was performed with Paint-a-Gate significant differences between the six participating centers Pro (BD) software. A special meeting for personnel of all part- were found regarding the frequency of CD7+ BM cells (P . icipating centers was devoted to the training in calibration, 0.05). instrument settings, and data acquisition and analysis. Expression patterns of the various T cell marker combi- Quality control of the procedures was performed both prior nations analyzed within the CD7+ BM cells are exemplified to and during the study. For this purpose, triple staining of a in Figure 1 and their overall frequencies (expressed both as normal PB samples for CD3-FITC (Leu4, BD)/CD4-PE (Leu3, percentages of CD7+ cells and of total BM cells) are shown in BD)/CD8-PE/Cy5 (CD8-TRI, Caltag Laboratories) was perfor- Table 2. med daily. Finally, the data were centrally re-analyzed at No clearly defined populations of CD34+/CD7+ and Karolinska Hospital. Statistical analysis (see below) did not TdT+/CD7+/cyCD3+ cells were found in any of the normal BM show any significant differences in data obtained at different samples analyzed. This was true both when ‘live gate’ acqui- laboratories. sition was performed on the basis of CD7 expression and for TdT+ or CD34+ cells. Thus, we can assume that frequencies of these cells in normal BM were less than 1 in 104 cells in Statistical analysis 20 studied samples. As shown in Figure 1a, CD7/CD5/CD3 triple staining (I) All the results are expressed as mean with one standard devi- showed the existence of four distinct CD7+ subsets in BM, ation (s.d.). Correlation between antigen expression and the the CD3+/CD5+ subset being dominant (84%); other subsets age of the control BM donors was investigated using t-test and included CD3−/CD5− (14%), CD3−/CD5+ (1%) and Pearson correlation test. In order to assess the statistical sig- CD3+/CD5− (1%) cells (Table 2). Further multiple stainings nificance of the differences observed between results from (performed at Salamanca) showed that most of the CD3−CD5+ various groups, the analysis of variance (ANOVA) was used. events corresponded to natural killer (NK) cells, while the P values less than 0.05 were considered to be statistically CD3+/CD5− cells were TCRγδ+ T lymphocytes (data not significant. shown). The relative frequency of the CD3+/CD5− subpopul- ation expressed as a percentage of total CD7+ cells increased with age (correlation coefficient (corr coeff) 0.4361, P = 0.01). Results However, no significant correlation was found when CD5−/CD3+/CD7+ cells were expressed as a percentage of the Expression patterns of T cell markers in normal bone total number of BM cells. No significant differences were marrow found in the distribution of the other subsets according to age or between participating centers (P . 0.05). The mean frequency of CD7+ cells in BM was 10% ± 3.6% Using the CD7/CD4/CD8 triple combination (II), six subsets (range 5.2–20.4%). The proportion of CD7+ cells increased of CD7+ cells were identified (Figure 1b, Table 2). Two minor with age: 7.9% ± 1.8% in children under 14 vs 14.1% ± 3.8% subsets of CD7+ cells (maximum 0.2% of the total BM cells) in healthy donors 14 or more years of age (P , 0.01). No co-expressed both CD4 and CD8 (populations 4 and 5 in Fig-

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 819 a b

c d

Figure 1 Flow cytometric analysis of CD7+ cells in the normal BM. The upper row shows subpopulations revealed after CD7/SSC gating, in triple staining for CD7/CD5/CD3 (I), CD7/CD4/CD8 (II) and lower row for CD3 positive (III) and negative (IV) subsets in CD7/CD2/CD3 staining, respectively. The frequencies of various populations are given in Table 2. (a) In the CD7/CD5/CD3 (I) staining the cyan dots illustrate population (1) CD3−/CD5−, blue dots population (2) CD3+/CD5−, green dots population (3) CD3+/CD5bright and red dots population (4) CD3dim/CD5bright. (b) In the CD7/CD4/CD8 (II) staining the cyan dots illustrate population (1) CD4−/CD8−, blue dots population (2) CD4−/CD8dim, red dots population (3) CD4−/CD8bright, violet dots population (4) CD4dim/CD8bright, black dots population (5) CD4bright/CD8dim and green dots population (6) CD4+/CD8−. (c) In the CD7/CD2/CD3 staining gated on CD3+ cells (III) the blue dots illustrate population (1) CD3+/CD2− and the green dots population (2) CD3+/CD2+. (d) In the CD7/CD2/CD3 staining gated on CD3− cells (IV) the green dots illustrate population (3) CD3−/CD2−/CD7bright, blue dots population (4) CD3−/CD2+/CD7bright and red dots population (4) CD3−/CD2+/CD7dim. ure 1b, Table 2). In these minor subsets, expression of CD4 should be noted that the intensity of expression of both was always either higher or lower than CD8 and both antigens myeloid-associated markers was ‘dim’ in all samples. never showed similar intensity of expression (when CD4 was The total numbers and distribution of various T cell subsets ‘bright’, the CD8 was ‘dim’ and vice versa). The proportion remained stable during chemotherapy. Table 3 shows the of the CD4−/CD8− cells decreased with age (corr coeff results obtained in 16 patients with B precursor ALL tested at −0.4469, P = 0.01), while the percentage of the CD4+/CD8− various phases of treatment. No significant differences were subset (corr coeff 0.5363, P = 0.01) increased. This was valid found between the patients and healthy donors. No popu- both when the data was expressed as a percentage of total lations of C7/CD34 or cyCD3/TdT cells were found in these CD7+ cells and as a percentage of total BM cells. No signifi- patients. cant differences were found in the frequency of the various T cell subsets between the six participating centers, P , 0.05. The subsets identified by the CD7/CD2/CD3 staining, were Immunophenotypic patterns of T-ALL blast cells analyzed separately for CD3+ (III) (two subsets defined, Figure 1c) and CD3− (IV) cells (three subsets defined, Figure 1d), Based on the findings obtained from the analysis of normal (Table 2). No significant differences were found regarding the BM samples, we established the potential patterns of aberrant distribution of these subsets according to age or between part- antigen expression in T-ALL. These aberrant patterns are icipating centers. Further studies have shown that almost all defined by their position in the so-called ‘empty spaces’ found cells in the CD7+/CD3− population expressed the CD56 on the multidimensional dot plots of normal BM samples NK-cell associated marker (98% ± 1%). (Figure 2). Based upon the ‘empty space’ definition, we found No CD7+/CD19+ could be detected at frequencies higher that all 65 investigated T-ALL cases showed aberrant immuno- than 10−4. In contrast, CD7+/CD13+ and/or CD7+/CD33+ cells phenotypes. The frequencies of each of the various aberrant were detected in 10% of the BM samples at frequencies higher phenotypic patterns are given in Table 4, and several aberrant than 10−4 (mean frequency of 0.06 ± 0.03% and 0.05 ± 0.04% patterns are exemplified in Figure 2. In most of the cases more for CD7+/CD13+ and CD7+/CD33+ cells, respectively). It than one aberration was found: 91% of cases had aberrant

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 820 Table 3 Frequency of various T cell subsets present in bone marrow under cytostatics treatment

Triple staining T cell subsetsa combination (FITC/PE/PercP 123456 alt PE/Cy5)

CD7/CD5/CD3 CD3−/CD5− CD3+/CD5− CD3+/CD5bright CD3dim/neg/CD5bright NA NA n = 16 I 14.7 (8.6)/ 2.9 (5.16)/ 81 (10.6)/ 1.1 (1.02)/ 1.17 (1.3) 0.2 (0.4) 9.1 (9.25) 0.1 (0.13)

CD7/CD4/CD8 CD4−/CD8− CD4−/CD8dim CD4−/CD8bright CD4dim/CD8bright CD4bright/CD8dim CD4+/CD8− n = 16 II 18.7 (8.6)/ 5.5 (3.1)/ 38.7 (7.0)/ 1.28 (1.05)/ 0.5 (0.98)/ 33.4 (12.9)/ 1.6 (1.53) 0.5 (0.47) 4.3 (4.6) 0.1 (0.09) 0.04 (0.06) 4.0 (4.4)

CD7/CD2/CD3 CD3+/CD2− CD3+/CD2+ CD3−/CD2−/CD7bright CD3−/CD2+/CD7bright CD3−/CD2+/CD7dim NA n = 16 III/IV 1.1 (2.0)/ 83 (7.7)/ 5.1 (4.0)/ 9.5 (5.4)/ 1.55 (1.1)/ 0.07 (0.09) 8.6 (8) 0.6 (0.85) 0.9 (1.0) 0.12 (0.1) CD3+/CD7+ (III) CD3−/CD7+ (IV) NA 83.6 (8.6)/7.8 (8.1) 16.4 (1.6)/1.6 (1.9) CD3+/CD2− b CD3+/CD2+ b CD3−/CD2−/CD7bright c CD3−/CD2+/CD7bright c CD3−/CD2+/CD7dim c 0.73 (0.64) 99.3 (0.64) 31.14 (12.6) 56.6 (12.5) 11.3 (8.2)

phenotypes in at least two triple stains, 65% in at least three, gestational age of 7 weeks at which the thymic rudiment is 25% in at least four, and 4% of cases were aberrant in all five colonized by T cell progenitors.16 More recent findings have antibody combinations. shown that human multi-potent progenitor cells expressing Table 5 shows a detailed description of the different CD34 have the capacity to differentiate into T cells once they aberrant phenotypes found in the 65 T-ALL cases included in are placed in the thymic environment.17 Differentiating CD34+ the present study. The most common aberrant pattern was the bone marrow cells that co-express CD7 and display poor co-existence of the TdT together with CD7 and/or cyCD3, ability to mature into the myeloid lineage can also efficiently found in 58 of the 64 studied cases (91%). Co-expression of give rise to T cells in culture systems with cytokines CD34 with CD7 and other surface T cell-associated antigens (interleukin 7).18 These lymphoid committed precursors might was found in 40% of all T-ALL cases analyzed (23/58). In 36 be the true thymic repopulating cells. In the thymus, cells with of 65 cases (55%), a bright expression of CD7 (as compared a similar CD34+CD7+ phenotype include the most primitive to normal T-lymphocytes) was noted. In nine of 65 cases thymocyte precursors. CD34+ thymocytes have no ability to (14%) the intensity of CD7 expression was higher than in nor- differentiate into myeloid cells, but may include NK cell pre- mal NK cells, and one of these cases co-expressed CD56. cursors.19,20 In spite of these findings, it is still controversial Based on these findings, the most common leukemia asso- whether commitment to the T cell lineage occurs in man ciated phenotype was an ectopic (thymus associated) antigen before or after T cell precursors migrate into the thymus.21 expression (91%). Additionally, asynchronous antigen CD7 expression is then retained until the last stages of both expression and rare phenotypes were found in 40% and 72% T and NK cell differentiation.22,23 The vast majority of T-ALL of cases, respectively. With regard to cross-lineage antigen cases also express the CD7 antigen.2–4,24 Based on these expression, none of the analyzed T-ALL cases expressed the observations, we have focused our studies on T cell subsets CD19 B cell-associated marker. ‘Dim’ expression of the in the CD7+ lymphoid population of normal BM. We have myeloid-associated markers CD13 and CD33 was detected in eight of 58 (16%) and six of 58 (10.5%) cases, respectively (four of these showed both markers). Totally, based on the Figure 2 Upper row: flow cytometric plots from staining I–VI (a, b, c, g, h and i, respectively) illustrate the concept of ‘empty spaces’ analysis of myeloid and NK cell-associated markers, only + 13.5% of the T-ALL cases showed an aberrant immunopheno- where no normal CD7 cells can be found in normal BM. The blue dots illustrate normal BM T cells and the circles illustrate patterns of type. aberrant expression. The frequency of T-ALL cases showing the proposed patterns of antigen expression are given in Table 4. Lower row: examples of aberrant patterns of antigen expression in T-ALL. Red dots Discussion represent leukemic blasts and gray dots illustrate normal BM T cells. The following patterns are exemplified: (d) pattern IB in the The CD7 antigen has been classically considered the earliest CD7/CD5/CD3 staining; (e) pattern IIA in the CD7/CD4/CD8 staining; (f) pattern IIIB for CD3 positive cases in the CD7/CD2/CD3 staining; T cell-associated marker expressed during maturation of (j) pattern IVB for CD3 negative cases in the CD7/CD2/CD3 staining; T-lymphocytes. This is based on the presence of (k) pattern VA in TdT/CD7/cyCD3 staining; and (l) pattern VIB in CD45+CD7+CD3−CD4−CD8− cells in human fetal liver at the CD34/CD38/CD7 staining.

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 821 a b c

d e f

g h i

j kl

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 822 Table 4 Frequency of the various patterns of antigen expression found among the 65 T-ALL cases using ﬁve studied marker combinations

Marker combinationsa Normal phenotypes Aberrant phenotypes

123456 ABCD

I CD7/CD5/CD3 2 0 5 18 NA NA 0 11b 29b NA n = 65 (3%) (7.6%) (28%) (17%) (44%) II CD7/CD4/CD8 24d 10e 08e 8d 5f 16d–g 9d,f 0NA n = 65 (37%) (15%) (12%) (12%) (7.6%) (24.6%) (13.8%) III CD7/CD2/CD3 1 2 NA NA NA NA 0 6 8 NA CD3 dim/pos. (6%) (12%) (35%) (47%) casesh n = 17 IV CD7/CD2/CD3 NA NA 2 15c 4NA4 2c 55 CD3 neg casesh (6%) (45%) (12%) (12%) (6%) (15%) (15%) n = 33 V TdT/CD7/cyt.CD3 NA NA NA NA NA NA 58 n = 64 (91%) VI CD7/CD38/CD34 NA NA NA NA NA NA 23 n = 58 (40%)

aResults are expressed as numbers of cases and percentages. Normal patterns of expression are shown in Figure 1 and Table 2, aberrant patterns in Figure 2. bIn two cases patterns B and C were observed. cIn one case patterns 4 and B were found. dIn four cases pattern 1 and 2; in one case patterns 1.A and 5; and in three cases pattern 1 and B were observed. eIn four cases pattern 1 and 2; in one case 2 and A; and in one case 2 and 4 were observed. fIn one case paterns B and 6 and in one case A and B coexisted. gIn one case pattern A and C were found. hIn 12 cases CD3dim and CD3 negative populations coexisted.

used a double-step acquisition strategy in which information Table 5 Most frequent aberrant phenotypes in 65 cases of T ALL on CD7+ events was selectively obtained. This strategy was successfully applied for identification of aberrant cells in fol- Type of aberrant antigen expression Number of cases low-up samples of patients included in this study as exem- found plified in Figure 3. There were no CD7 negative cases in our (%) material. An alternative approach for blast cell gating stra- tegies in CD7 negative or CD7 dim cases would be the use 1 Ectopic antigen expression of other T cell markers as cyCD3 or CD2 or CD5 (Figure 3b). TdT/CD7/cyt.CD3 58/64 (91%) Since the results were derived from six different labora- CD34/CD7 23/58 (40%) CD3dim/CD5dim 11/65 (17%) tories, our first aim was to standardize the techniques and to dim − compare the results in order to avoid variations due to incon- CD4 /CD8 9/65 (14%) sistencies in methodology or data analysis. Antibodies, sample 2 Asynchronous antigen expression preparation, staining methods, instrument settings, data acqui- dim dim sition and data analysis were standardized prior to the CD3 /CD2 6/65 (9%) immunophenotyping of the normal BM and T-ALL samples. 3 Rare phenotypes (,1/10−4 in normal Identical results were obtained in each individual center as BM) reflected both by the localization of the different cell popu- CD2dim or neg/CD3−/CD7+++ 10/33 (30%) lations in the dot plot diagrams and the absence of statistically CD3−/CD5dim 29/65 (44.6%) significant differences in the percentages of each subset analyzed. Statistical analysis confirmed the lack of significant 4 Lineage infidelity inter-laboratory variation of results. + CD13 8/58 (16%) We have found that the overall percentage of CD7 cells in CD33 6/58 (10.5%) normal BM was approximately 10% and increased with age. + Among CD7 BM cells our interest was initially focused on the 5 Overexpression of antigen + + + + identification of CD34 /CD7 and TdT /CD7 cells. Previous CD7 9/65 (14%) reports have described the presence of small populations of + + + + TdT /CD7 and CD34 /CD7 T cell precursors in BM.25–27 6 Low expression or lack of antigen Sorting of human BM CD34+ cells showed that the expression + + − CD34 /CD7 /CD2 phenotype correlated with the transcrip- CD4dim/CD8dim 16/65 (24.6%) tion of CD3ζ but not CD3γ or CD3δ (a pattern of CD3 gene CD3 (ectopic) 60/65 (92%) transcription observed in mature blood NK cells but not in T CD5 42/65 (63%) cells).27 In contrast to the above-mentioned studies, our results CD2 34/54 (62%) show that the frequencies of CD34+/CD7+ cells and

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 823 a the CD7/CD2/CD3 triple staining (75% aberrant cases) fol- lowed by the CD7/CD5/CD3 staining (58% aberrant cases). Aberrant phenotypes were also detected in 38% of T-ALL cases analyzed with the CD7/CD4/CD8 triple staining. Thus, the combined use of only a few common triple stains would allow the flow cytometric follow-up of leukemic cells in all T-ALL cases. Our results are in agreement with previously reported data on the incidence of aberrant phenotypes in T-ALL.6,8,9 In most T-ALL cases three or more aberrant immunophenotypic patterns were found, mainly reflecting ectopic thymic phenotypes. The most frequent ectopic phenotype was characterized by either dim or absent surface CD3 expression and double expression of cyCD3/TdT.28 In seven of 65 studied T-ALL cases (10.7%) leukemic blasts were CD3−/CD4−/CD8−/ CD34+. This phenotype was consistent with the earliest stages of thymic differentiation. Twenty-five of 65 T-ALL cases (38%) b showed a presence of CD4+/CD8+ double positive blasts similar to CD4+/CD8+ thymocytes. These ectopic thymic immunophenotypes are found neither in normal BM nor in PB of ALL patients during and after treatment.29 By comparison to normal BM, T and NK cells, aberrant (low or high) antigen expression was detected in 60% of the cases and was usually related to the dim expression of CD3, CD5 or CD2. Over-expression of CD7 was detected in 14% of the cases and abnormally low expression of CD7 or CD5 was detected in six (9%) and 29 (45%) of the cases, respectively. Our results show that only 12% of analyzed cases were CD13 and/or CD33 positive. We have taken into account only cases with strong expression of the myeloid markers on T cell blasts that could be used for MRD investigation. This may explain the relatively low incidence found in the present study using sensitive CD33-PE and CD13-PE conjugates, as compared to some other reports in the literature.11,30–33 Figure 3 Illustration of flow cytometric detection of MRD using It is obvious from our study that the immunophenotypes of ‘live-gate’ approach in follow-up BM samples obtained from two T-ALL differ significantly from normal BM T cells and NK patients with T-ALL in morphological complete remission on day 29 cells. For a major part this is caused by their thymocyte origin (after induction treatment). (a) Triple staining combination (see ectopic antigen expression, Table 5). However, the neo- TdT/CD7/cyCD3. The ‘live’ gate was set on CD7/SSC. The percentage of CD7+ cells in the BM was 7.04%, 30308 CD7+ lymphoid cells plastic transformation might also have affected antigen were acquired, the total of 430756 cells have passed the flow cyto- expression patterns (Table 5), as has been previously observed meter and 278 TdT/CD7/cyCD3 triple positive lymphoid cells were in B precursor ALL34,35 and in acute myeloid leukemia.36,37 found. MRD was calculated as 0.06% of BM cells. (b) Triple staining Furthermore, it should be noted that T-ALL also differ from TdT/CD5/cyCD3. The diagnostic sample showed dim CD5 expression each other displaying unique immunophenotypic patterns. on leukemic blasts. The ‘live gate’ was set on cyCD3/SSC. The + + In summary, our results show that the use of only five triple percentage of cyCD3 cells in the BM was 8.8%, 10482 cyCD3 lymphoid cells were acquired, a total of 118880 cells passed the flow marker combinations allows detection of ectopic thymic cytometer and 70 triple positive TdT/CD5 dim/cyCD3 cells were phenotypes in all T-ALL cases. In addition, standardization of found. MRD was calculated as 0.04% of BM cells. the methods for cell staining, instrument calibration, data acquisition and data analysis provides an objective and reproducible basis for application of flow cytometric MRD TdT+/CD7+ cells in childhood and adult BM generally are detection in T-ALL patients. In this way, flow cytometric lower than 1/104 BM cells. Therefore the observation of MRD detection provides a relatively inexpensive and rapid CD34+/CD7+ phenotypes, especially if co-expressed together alternative to PCR-based MRD detection via rearranged with cyCD3 and/or TdT at frequencies higher than 10−4 can immunoglobulin or T cell receptor genes.6,8,10,38 be considered as leukemia associated. Most T-ALL in our series (91%) co-expressed TdT and Acknowledgements cyCD3 and/or CD7. This makes the TdT/CD7/cyCD3 triple labeling a powerful and broadly applicable tool for MRD stud- This study was supported by European BIOMED-1 grant ies, in accordance with previous early reports.25,28 Twenty- BMH1-CT94-1675, Swedish Cancer Society, Dutch Cancer three of 58 analyzed T-ALL cases (40%) co-expressed CD34 Society/Koningin Wilhelimina Fonds (grant EUR 94-852), and CD7. It should be noted that in most CD34+ T-ALL cases grant of the Associazione Italiana per la Ricerca sul Cancro, the expression of this marker was only seen on a subset of the by Fondazione M Tettamanti, and a grant from ‘Liga Portug- CD7+ leukemic blast cells. In the CD7/CD38/CD34 labeling, uesa contra o Cancro’. We thank Dakopatts (Glostrup, the information provided by CD38 was marginal, thus limiting Denmark) and Caltag Lab (San Francisco, CA, USA) for pro- the use of this triple antigen combination for MRD studies. viding charge-free reagents for the study and Mr Lewis G The most informative of the other marker combinations was Edgel for editorial assistance.

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 824 References MA, Dalloul A, Debre P. CD34-positive early stages of human T- cell differentiation. Leuk Lymphoma 1995; 17: 43–50. 19 Res P, Martinez-Caceres E, Cristina JA, Staal F, Noteboom E, 1 Janossy G, Bollum FJ, Bradstock KF, Ashley J. Cellular phenotypes Weijer K, Spits H. CD34+CD38dim cells in the human thymus can of normal and leukemic hemopoietic cells determined by analysis differentiate into T, natural killer, and dendritic cells but are dis- with selected antibody combinations. Blood 1980; 56: 430–441. tinct from pluripotent stem cells. Blood 1996; 87: 5196–5206. 2 van Dongen JJ, Krissansen GW, Wolvers-Tettero IL, Comans-Bitter 20 Spits H, Blom B, Jaleco AC, Weijer K, Verschuren MC, van WM, Adriaansen HJ, Hooijkaas H, van Wering ER, Terhorst C. Dongen JJ, Heemskerk MH, Res PC. Early stages in the develop- Cytoplasmic expression of the CD3 antigen as a diagnostic marker ment of human T, natural killer and thymic dendritic cells. Immu- for immature T-cell malignancies. Blood 1988; 71: 603–612. nol Rev 1998; 165: 75–86. 3 Ludwig WD, Harbott J, Bartram CR, Komischke B, Sperling C, 21 Blom B, Res P, Noteboom E, Weijer K, Spits H. Prethymic CD34+ Teichmann JV, Seibt-Jung H, Notter M, Odenwald E, Nehmer A. progenitors capable of developing into T cells are not committed Incidence and prognostic significance of immunophenotypic sub- to the T cell lineage. J Immunol 1997; 158: 3571–3577. groups in childhood acute lymphoblastic leukemia: experience of 22 Lazarovits AI, Osman N, Le Feuvre CE, Ley SC, Crumpton MJ. the BFM study 86. Rec Res Cancer Res 1993; 131: 269–282. CD7 is associated with CD3 and CD45 on human T cells. J Immu- 4 van Dongen JJ, Adriaansen HJ. Immunobiology of leukemia. In: nol 1994; 153: 3956–3966. Henderson ES, Lister TA, Greaves MF (eds). Leukemia, 6th edn, 23 Ginaldi L, Farahat N, Matutes E, De Martinis M, Morilla R, Catov- WB Saunders: Philadelphia, 1996, pp 83–130. sky D. Differential expression of T cell antigens in normal periph- 5 Campana D, Coustan-Smith E, Behm FG. The definition of remission in acute leukemia with immunologic techniques. Bone eral blood lymphocytes: a quantitative analysis by flow cytometry. Marrow Transplant 1991; 8: 429–437. J Clin Pathol 1996; 49: 539–544. 6 van Dongen JJ, Breit TM, Adriaansen HJ, Beishuizen A, Hooijkaas 24 van Wering ER, Beishuizen A, Roeffen ET, van der Linden- H. Detection of minimal residual disease in acute leukemia by Schrever BE, Verhoeven MA, Hahlen K, Hooijkas H, van Dongen immunological marker analysis and polymerase chain reaction. JJ. Immunophenotypic changes between diagnosis and relapse in Leukemia 1992; 6 (Suppl. 1): 47–59. childhood acute lymphoblastic leukemia. Leukemia 1995; 9: 7 Drach J, Drach D, Glassl H, Gattringer C, Huber H. Flow cyto- 1523–1533. metric determination of atypical antigen expression in acute leuke- 25 van Dongen JJ, Hooijkaas H, Comans-Bitter M, Hahlen K, de Klein mia for the study of minimal residual disease. Cytometry 1992; A, van Zanen GE, van’t Veer MB, Abels J, Benner R. Human bone 13: 893–901. marrow cells positive for terminal deoxynucleotidyl transferase 8 Campana D, Pui CH. Detection of minimal residual disease in (TdT), HLA-DR, and a T cell marker may represent prothymocytes. acute leukemia: methodologic advances and clinical significance J Immunol 1985; 135: 3144–3150. (see comments). Blood 1995; 85: 1416–1434. 26 Bender JG, Unverzagt KL, Walker DE, Lee W, Van Epps DE, Smith 9 Orfao A, Ciudad J, Lopez-Berges MC, Lopez A, Vidriales B, DH, Stewart CC, To LB. Identification and comparison of CD34- Caballero MD, Valverde B, Gonzalez M, San Miguel JF. Acute positive cells and their subpopulations from normal peripheral lymphoblastic leukemia (ALL): detection of minimal residual dis- blood and bone marrow using multicolor flow cytometry. Blood ease (MRD) at flow cytometry. Leuk Lymphoma 1994; 13 (Suppl 1991; 77: 2591–2596. 1): 87–90. 27 Long H, Gaffney P, Mortari F, Miller JS. CD3 gamma, CD3 delta, 10 van Dongen JJ, Szczepanski T, de Bruijn MA, van den Beemd and CD3 zeta mRNA in adult human marrow hematopoietic pro- MW, de Bruin-Versteeg S, Wijkhuijs JM, Tibbe GJ, van Gastel- genitors correlates with surface CD2 and CD7 expression. Exp Mol EJ, Groeneveld K, Hooijkaas H. Detection of minimal residual Hematol 1996; 24: 1402–1408. disease in acute leukemia patients. Cytok Mol Ther 1996; 2: 28 Campana D, Thompson JS, Amlot P, Brown S, Janossy G. The 121–133. cytoplasmic expression of CD3 antigens in normal and malignant 11 Ciudad J, San Miguel JF, Lopez-Berges MC, Vidriales B, Valverde cells of the T lymphoid lineage. J Immunol 1987; 138: 648–655. B, Ocqueteau M, Mateos G, Caballero MD, Hernandez J, Moro 29 Caver TE, Slobod KS, Flynn PM, Behm FG, Hudson MM, Turner MJ, Mateos MV, Orfao A. Prognostic value of immunophenotypic EV, Webster RG, Boyett JM, Tassie TL, Pui CH, Hurwitz JL. Pro- detection of minimal residual disease in acute lymphoblastic leu- found abnormality of the B/T lymphocyte ratio during chemo- kemia. J Clin Oncol 1998; 16: 3774–3781. therapy for pediatric acute lymphoblastic leukemia. Leukemia 12 Lucio P, Parreira A, van den Beemd MW, van Lochem EG, van 1998; 12: 619–622. Wering ER, Baars E, Porwit-MacDonald A, Bjorklund E, Gaipa G, 30 Pui CH, Rubnitz JE, Hancock ML, Downing JR, Raimondi SC, Riv- Biondi A, Orfao A, Janossy G, van Dongen JJ, San Miguel JF. Flow era GK, Sandlund JT, Ribeiro RC, Head DR, Relling MW, Evans cytometric analysis of normal B cell differentiation: a frame of ref- WE, Behm FG. Reappraisal of the clinical and biological signifi- erence for the detection of minimal residual disease in precursor- cance of myeloid-associated antigen expression in childhood B-ALL. Leukemia 1999; 13: 419–427. acute lymphoblastic leukemia. J Clin Oncol 1998; 16: 3768– 13 Weir EG, Cowan K, LeBeau P, Borowitz MJ. A limited antibody 3773. panel can distinguish B-precursor acute lymphoblastic leukemia 31 Uckun FM, Sather HN, Gaynon PS, Arthur DC, Trigg ME, Tub- from normal B precursors with four color flow cytometry: impli- ergen DG, Nachman J, Steinherz PG, Sensel MG, Reaman GH. cations for residual disease detection. Leukemia 1999; 13: 558– Clinical features and treatment outcome of children with myeloid 567. antigen positive acute lymphoblastic leukemia: a report from the 14 van Lochem EG, Groeneveld K, te Marvelde JG, van den Beemd Children’s Cancer Group. Blood 1997; 90: 28–35. MW, Hooijkaas H, van Dongen JJ. Flow cytometric detection of 32 Babusikova O, Glasova M, Konikova E, Kusenda J, Cap J, Gyarfas intracellular antigens for immunophenotyping of normal and J, Koubek K. Phenotypic heterogeneity and aberrant markers malignant leukocytes: testing of a new fixation-permeabilization solution (letter). Leukemia 1997; 11: 2208–2210. expression in T-cell leukemia. Neoplasma 1998; 45: 128–134. 15 Groeneveld K, te Marvelde JG, van den Beemd MW, Hooijkaas 33 Boldt DH, Kopecky KJ, Head D, Gehly G, Radich JP, Appelbaum H, van Dongen JJ. Flow cytometric detection of intracellular anti- FR. Expression of myeloid antigens by blast cells in acute lym- gens for immunophenotyping of normal and malignant leukocytes. phoblastic leukemia of adults. The Southwest Oncology Group Leukemia 1996; 10: 1383–1389. experience. Leukemia 1994; 8: 2118–2126. 16 Barcena A, Muench MO, Roncarolo MG, Spits H. Tracing the 34 Ryan D, Kossover S, Mitchell S, Frantz C, Hennessy L, Cohen H. expression of CD7 and other antigens during T and myeloid cell Subpopulations of common acute lymphoblastic leukemia anti- differentiation in the human fetal liver and thymus. Leuk Lym- gen-positive lymphoid cells in normal bone marrow identified by phoma 1995; 17: 1–11. hematopoietic differentiation antigens. Blood 1986; 68: 417–425. 17 Barcena A, Galy AH, Punnonen J, Muench MO, Schols D, Roncar- 35 Hurwitz CA, Loken MR, Graham ML, Karp JE, Borowitz MJ, Pullen olo MG, de Vries JE, Spits H. Lymphoid and myeloid differen- DJ, Civin CI. Asynchronous antigen expression in B lineage acute tiation of fetal liver CD34+ linkeage− cells in human thymic organ lymphoblastic leukemia. Blood 1998; 72: 299–307. culture. J Exp Med 1994; 180: 123–132. 36 Reading CL, Estey EH, Huh YO, Claxton DF, Sanchez G, Ter- 18 Schmitt C, Ktorza S, Sarun S, Verpilleux MP, Blanc C, Deugnier stappen LW, O’Brien MC, Baron S, Deisseroth AB. Expression of

Leukemia BIOMED-1 Concerted Action report A Porwit-MacDonald et al 825 unusual immunophenotype combinations in acute myelogenous Pisa P, Gonzalez M, Bartram CR, Panzer-Grumayer ER, Biondi A, leukemia. Blood 1993; 81: 3083–3090. San Miguel JF, van Dongen JJ. Primers and protocols for stan- 37 Terstappen LW, Safford M, Unterhalt M, Konemann S, Zurlutter dardized detection of minimal residual disease in acute lymphobl- K, Piechotka K, Drescher M, Aul C, Buchner T, Hiddemann W. astic leukemia using immunoglobulin and T cell receptor gene Flow cytometric characterization of acute myeloid leukemia: IV. rearrangements and TAL1 deletions as PCR targets: report of the Comparison to the differentiation pathway of normal hematopo- BIOMED-1 CONCERTED ACTION: investigation of minimal ietic progenitor cells. Leukemia 1992; 6: 993–1000. residual disease in acute leukemia. Leukemia 1999; 13: 110–118. 38 Pongers-Willemse MJ, Seriu T, Stolz F, d’Aniello E, Gameiro P,

Leukemia