Surface Antigen Phenotype Can Predict TEL-AML1 Rearrangement In

Leukemia (1998) 12, 1764–1770  1998 Stockton Press All rights reserved 0887-6924/98 $12.00 http://www.stockton-press.co.uk/leu Surface antigen phenotype can predict TEL-AML1 rearrangement in childhood B-precursor ALL: a Pediatric Oncology Group study MJ Borowitz1, J Rubnitz2, M Nash2, DJ Pullen3 and B Camitta4

1Johns Hopkins Medical Institutions, Baltimore, MD; 2St Jude Children’s Research Hospital, Memphis, TN; 3University of Mississippi Medical Center, Jackson, MS; 4Milwaukee Children’s Hospital, Milwaukee, WI; and Pediatric Oncology Group, Chicago, IL, USA

The t(12;21)(p13;q22) is the most common translocation in of blast cells.7–14 These distinctive phenotypic properties are childhood B-precursor ALL. It results in a TEL-AML1 rearrange- sometimes morphological, particularly in the case of AML, ment and is associated with a good prognosis. Because many with close associations between morphology and abnormali- chromosomal alterations in leukemia are associated with dis- ␤ tinct cell surface phenotypes, we investigated whether there ties in transcription factors affecting the AML1–CBF- com- was an association seen between surface marker expression plex.10,12 In ALL, T and B-lineage leukemias appear to involve and the TEL-AML1 rearrangement. Of 166 unselected cases of unique transcription factors,15 and among cases of B-precursor B-precursor ALL studied by Southern hybridization, 45 cases ALL there have also been more detailed associations found (27%) showed TEL rearrangement. Blasts of patients with TEL between expression of specific differentiation-associated rearrangement were much more likely to be CD9-negative, 13,14 CD45-positive, CD13 positive, and CD20 negative, but the pre- surface molecules and particular translocations. dictive value of any of these markers for the rearrangement was Because the AML1 gene is an important regulator of differ- very low. However, 93% of patients with the TEL rearrangement entiation, and because abnormalities in this gene have been had blasts that were either negative or only partly positive for shown to be associated with specific phenotypic abnormali- CD9; this phenotype was only seen in 27% of patients without ties in AML,12 we undertook a study to determine if a specific the rearrangement. Only information about CD20 expression pattern of expression of B cell-associated differentiation anti- added to the predictive value of CD9 alone. The predictive value of the phenotype CD9 (negative or partly positive) and CD20 gens could be seen in association with the TEL-AML1. Our (negative or partly positive), for the TEL rearrangement was results indicate that blasts at the differentiation stage charac- prospectively tested on an additional 223 cases, and found to terized by complete or partial lack of both CD9 and CD20 be 88% sensitive and 71% specific for the rearrangement, with were highly likely to have TEL rearrangements; this phenotype a positive predictive value of 47%. Hyperdiploidy, previously can be used to predict cases of abnormal TEL with a high shown to correlate negatively with the rearrangement, was a sensitivity and moderate specificity. slightly more sensitive indicator (94%) but had a much lower predictive value (28%). Three of eight cases found to be rearranged by Southern hybridization but lacking the characteristic phenotype failed to show evidence of the TEL-AML1 Materials and methods rearrangement by polymerase chain reaction, suggesting that at least some of the discordant cases may involve partner Patient population genes other than AML1 in the TEL rearrangement. We conclude that immunophenotyping is highly predictive of the TEL rearrangement. For every 100 patients with B-precursor ALL, The subjects in this study consisted of two groups of patients we estimate that prescreening by phenotyping would eliminate who constituted a subset of those entered on the POG 9400 the need for molecular testing on 57 patients and only two or classification study for newly diagnosed childhood ALL, from three of an expected 24 patients with the TEL rearrangement June 1996 to December 1996. After signing informed consent, would not be detected. all patients whose blasts were determined to be negative for Keywords: childhood; acute lymphoblastic leukemia; TEL; AML1; myeloperoxidase and nonspecific esterase by locally perfor- flow cytometry med standard cytochemistry had aliquots of bone marrow sent to the immunophenotyping reference laboratory at Johns Hop- kins University, and to the reference laboratory of St Jude Chil- Introduction dren’s Research Hospital for ploidy studies116 and for cell banking. A subset of patients with confirmed B-precursor ALL The TEL-AML1 fusion, created by a cryptic t(12;21), is the for whom excess material was available were studied for TEL most common translocation in childhood ALL, occurring in rearrangements. Over the period of this study, 78% of patients 1–3 about 25% of cases. Patients with this translocation have a had sufficient material to complete these studies. favorable prognosis.4–6 Blast cell hyperdiploidy, another important favorable prognostic factor, is negatively correlated 2,5,6 with the presence of this translocation, so that these two Immunophenotyping favorable genetic characteristics identify separate subsets of children. Although the TEL-AML1 translocation is restricted to Samples received at the Johns Hopkins reference laboratory patients with B-precursor ALL, and specifically to common were phenotyped by three color immunofluorescence flow 4,6 ALL, relatively little additional information is available cytometry using a modification of previously published dual 6 about the blast cell characteristics of these cases. color procedures,17 except that a whole marrow lysis pro- Cytogenetic abnormalities in both AML and ALL may be cedure was employed, and CD45-perCP was used as a com- associated with distinct surface antigen phenotypic properties mon reagent to permit specific isolation of the blast population for gating.18 Cases were prescreened either at local Correspondence: MJ Borowitz, (POG 9400) Pediatric Oncology institutions or with the combination of CD45, CD19 and Group, 645 N Michigan Avenue, Suite 910, Chicago, IL 60611, USA CD10 to identify likely cases of B-precursor ALL. All such Received 5 March 1998; accepted 21 July 1998 cases were studied with a reagent panel of phycoerythrin and TEL-AML1 rearrangement in childhood B-precursor ALL MJ Borowitz et al 1765 FITC conjugates as follows: CD19-PE/CD10-FITC; HLADR- gency washes, membranes were analyzed by autoradiogra- PE/CD7-FITC; CD20-PE/CD34-FITC; CD9-PE/CD24-FITC; phy. A subset of the samples that demonstrated TEL rearrange- CD33-PE/CD71-FITC; CD13-PE/CD15-FITC. In addition, a ment by Southern blot were analyzed by reverse-transcriptase separate two-color analysis of CD22-PE/CD45-FITC was per- polymerase chain reaction (RT-PCR) for the expression of the formed. All antibodies were obtained from Becton Dickinson der(21)-encoded TEL/AML1 fusion transcript using described (San Jose, CA, USA) except for CD19-PE (Immunotech, amplification primers and probes.2 Westbrook, ME, USA) and CD9-PE and CD24-FITC (Pharmingen, San Diego, CA, USA). Flow cytometric analysis was performed on a FACScan flow Statistics cytometer equipped with a 488 nm laser. Instrument settings were initially established using standard Calibrite beads Comparisons of frequencies were made using the ␹2 test. Sen- (Becton Dickinson), but then refined to minimize day-to-day sitivity, specificity and predictive value were calculated with variability using QC3 beads (Flow Cytometry Standards, San standard formulas using the presence of a TEL gene rearrange- Juan, PR). Analysis was performed using CellQuest software, ment as the desired outcome to be predicted. first gating on displays of CD45 and right angle scatter to isolate the blast population, and then displaying either dual parameter dot plots or single parameter histograms of expression Results of antigens on the blasts. Cases were defined as B-precursor ALL based on bright uni- One hundred and sixty-six cases of B-precursor ALL in which form expression of CD19 on the CD45-gated blast population; there was sufficient material after completing the required in rare cases CD19 was negative or partly expressed, but the studies for the POG ALL protocols were studied by Southern case was still considered B-precursor ALL because of bright hybridization to detect rearrangements of the TEL gene. Forty- uniform expression of CD22. Antibody positivity was deter- five of these cases (27%) showed TEL rearrangements. The mined by comparison of the shape of either histograms or dot results of these studies were correlated with immunopheno- plots to those of control nonreactive antibodies.19 Initially, type, as shown in Table 1. Immunophenotyping was satisfac- each antibody was categorized simply as positive or negative tory in 159/166 cases including all of the cases with TEL based on whether the histogram showed a significant shift rearrangements. There was no correlation between TEL (usually a minimum of 80–100 channels on a 4 decade log rearrangements and expression of CD19, CD10, HLADR, scale) relative to nonreactive antibodies, or if a distinct subset CD22, CD24 or CD33. However, the other markers tested of the gated blasts, generally amounting to at least 10% of showed statistically significant associations with TEL cells, showed expression that was clearly above the threshold rearrangements. Blasts of patients with TEL rearrangements determined by consideration of the reactivity of the control were much more likely to be CD9-negative and CD45-posi- antibody. In subsequent analyses, these two patterns were dis- tive, and somewhat more likely to be CD13-positive or CD20- tinguished, and cases in which only a subset of cells were negative. However, there was considerable heterogeneity seen to be positive was referred to as having ‘partial’ positivity. seen, and the predictive value of any of these markers for the Figure 1 demonstrates typical reactivity patterns of histograms TEL rearrangement was very low. For example, although 97% considered positive, negative and ‘partial’ positive in the case of TEL+ cases were CD45-positive, 78% of TEL-negative ones of CD9 expression. As illustrated, these histograms were gen- were positive as well; thus, in spite of the statistically signifi- erated based on gating of CD45/RALS displays to isolate leu- cant association the specificity of CD45 positivity for a TEL kemic blasts, and thus represent true distributions of antigen rearrangement was very low. distribution on blast populations. Other antibodies generally To try to improve the predictive value, the immunopheno- showed similar patterns, though in some cases of partial posi- typic information was examined in more detail, as described tivity the positive events constituted a majority of the blast in the materials and methods (Table 2). When looked at in population, rather than the minority, as shown in the Figure. this manner, CD34 expression was also shown to be corre- In an initial exploratory analysis, immunophenotyping data lated with TEL rearrangements. Moreover CD9, which was were reviewed in detail on 166 patients with B-precursor ALL already shown to show a highly significant statistical associ- whose TEL status was known. In this retrospective analysis, ation with TEL rearrangements when only a simple distinction significant associations could be seen with several antigens, between positive and negative was made, now demonstrated but the relationship between combinations of antigens and a pattern that was much more useful as a predictor for the phenotype was investigated in sufficient detail to try to deter- rearrangement. Thus, 93% of patients with TEL rearrange- mine a phenotype which might predict the presence of the ments had blasts that were either CD9-negative, or only par- TEL rearrangement. The accuracy of the phenotype was then tially positive; by contrast only 27% of patients whose blasts validated by having one of us (MJB) send predictions on a lacked TEL rearrangements had this phenotypic pattern. subsequent series of 223 patients without regard to TEL status Although positivity for CD45, or complete or partial absence to the St Jude laboratory for correlation. of CD20 were more sensitive markers for the rearrangement, both were very nonspecific. CD9-positivity and CD45-nega- tivity were highly correlated (data not shown) so that combin- TEL rearrangements ing both of these markers did not enhance the specificity or sensitivity above that of CD9 alone. Similarly, the relatively Genomic DNA was analyzed for TEL gene rearrangements as low sensitivity of CD13 or CD34 expression for the rearrange- previously described.2 Briefly, 10 ␮g of high molecular weight ment made it impossible to use information about these anti- DNA was digested with BamHI, separated electrophoretically gens to improve the predictive value of the CD9 phenotype. in 0.8% agarose gels, and transferred to nylon membranes. However, review of cases classified as CD9-negative or partly Membranes were then hybridized with an ␣-32P-dCTP-labeled positive revealed eight additional cases that were CD20-posi- 466-bp SacI/BamHI TEL cDNA fragment.2 After high strin- tive; as no CD20-positive case was TEL-rearranged, CD20- TEL-AML1 rearrangement in childhood B-precursor ALL MJ Borowitz et al

1766 0 0 0 O>

0.. 0 0.. ...0 0 -a.8 "' •• C> •o :i: c"' C09 50 0.., (J)Ouo Positive 0 (J).., Rt ., 0 8 "' "'

0 0 0 200 400 600 800 1000 0 200 400 600 800 1000 C045P81CP C09PE a b

0 O> ..0 a.AST ~TED ...0 Negative ..0 .. o Control c"' "o 8• C09 0 ., Negative 0 "'

0 0 200 400 600 800 1000 C09PE C

0 O> ..0 Bl.AST ~TED ...0 ..0 Negative .. o Control c "' "o 8• Partial 0 ., Positive 0 "'

0 0 200 400 600 800 1000 d C09PE

Figure 1 Flow cytometric characterization of CD9 antigen expression in ALL. (a) Dual parameter display of CD45 vs right angle light scatter showing gating procedure used to isolate blasts analytically. Single parameter histograms in subsequent ﬁgures represent displays gated on blasts. (b) Typical CD9-positive reaction, showing a homogeneous ‘bell-shaped’ distribution of positivity compared to a negative control. All cases showing this type of distribution were considered positive irrespective of the intensity of antigen expression. (c) Typical negative reaction, in this case superimposable with a negative control. (d) Histogram showing the pattern seen in what is termed ‘partial’ positivity. In this case there is a heterogeneous distribution, with part of the distribution overlapping the negative control, but with a distinct subset of cells of clearly greater intensity of expression.

positive cases could be eliminated and thus improve the PCR analysis of TEL-AML1 fusion specificity of the determination at no cost to the sensitivity. To investigate whether this algorithm could serve as a Although the most common TEL rearrangement in B-precursor means for predicting cases of TEL rearrangements, the immun- ALL involves AML1 on chromosome 21 as the partner gene, ophenotype of an additional series of cases of B-precursor ALL other rearrangements have been seen,20–23 and it is not clear was reviewed in detail without knowledge of TEL status. Cases that they should be expected to display the same phenotype. were considered TEL phenotype-positive if their blasts were To investigate this possibility, eight cases of TEL rearranged either negative or only partly positive for CD9, and also nega- leukemia not predicted by the immunophenotypic profile tive or partly positive for CD20. Other combinations were (including the six cases identified in Table 3 plus two considered TEL phenotype-negative. Satisfactory TEL studies additional false negative cases from the first series of patients) were available on a total of 223 of these cases. As shown were tested by RT-PCR for the presence of the TEL-AML1 in Table 3, the phenotype was highly predictive of the TEL fusion transcript. Five of these cases expressed TEL-AML1, rearrangement in this prospective series of additional cases. whereas three showed no evidence of the fusion transcript. It TEL-AML1 rearrangement in childhood B-precursor ALL MJ Borowitz et al 1767 Table 1 Immunophentoype of cases with and without TEL Correlation of phenotype and ploidy rearrangements Previously it has been recognized that hyperdiploidy showed + − TEL % TEL % P value a significant negative correlation with TEL status.2,5,6 We n = 45 n = 114 therefore investigated the relationship between ploidy status and immunophenotype in our prospectively defined data set. CD10 pos 100 95 NS neg 0 5 Fifty-four of 222 cases were hyperdiploid; only five of these hyperdiploid cases were phenotype-positive. Interestingly, CD34 pos 85 84 NS neg 15 16 three of the five hyperdiploid cases were TEL rearranged. Only 48 of 169 non-hyperdiploid cases showed TEL rearrangement; CD20 pos 40 56 0.019 neg 60 43 thus the predictive value of (lack of) hyperdiploidy for the TEL rearrangement was substantially lower than that of immuno- CD24 pos 100 96 NS neg 0 4 phenotype (28 vs 47%), though the sensitivity and specificity were similar. Because of the high correlation between ploidy CD9 pos 51 96 Ͻ0.0001 neg 49 4 status and phenotype, and the lack of complete specificity of ploidy, adding information about ploidy did not improve the CD33 pos 49 33 0.07 neg 51 67 predictive value of immunophenotype alone. CD45 pos 97 78 0.003 neg 2 22 Discussion CD13 pos 75 42 0.0002 neg 22 58 Many types of acute leukemia are associated with production of abnormal transcription factors that control specific steps in hematopoietic cell differentiation. One manifestation of differentiation is the expression of stage-specific and lineage-spe- Table 2 Correlation of immunophenotype pattern and TEL cific surface membrane proteins. This has led to the obser- rearrangements vation that many types of acute leukemia associated with specific translocations are associated with characteristic TEL+ % TEL− % P value immunophenotypes.8,10,12,14 In this study, we demonstrate that n = 45 n = 114 childhood ALL associated with the TEL-AML1 can be added to the list of such leukemias. pos 7 73 CD9 neg 49 4 Ͻ0.0001 Cases of ALL with rearranged TEL genes had a characteristic part 44 23 immunophenotype in which their blasts failed to express pos 84 46 homogeneous CD20 or CD9 in nearly all cases. CD20 is CD45 neg 2 22 0.0006 known to be an antigen expressed relatively late in B cell part 13 32 development; there are no good studies which speak to the pos 31 23 pattern of acquisition of CD9. In preliminary experiments, CD13 neg 22 58 0.002 however, we have shown that in normal B cell differentiation part 44 19 the earliest CD34+ B cell precursors are CD9-negative or low pos 27 54 density, and that CD9 acquisition roughly follows that shown CD34 neg 15 17 0.015 by CD20 (Borowitz et al, unpublished). Although the AML-1 part 58 30 gene is known to be of critical importance in the regulation pos 0 10 of myeloid differentiation24 less is known about its possible CD20 neg 60 43 0.08 role in lymphoid development. Certainly there are no data to part 40 46 suggest that either CD20 or CD9 are potential target genes for the AML1 gene product. CD9, a member of the tetraspan superfamily of cell surface molecules involved in cell adhesion, has been shown to be upregulated in response to Table 3 Prospective prediction of TEL status from immunopheno- TPA-induced differentiation in HL60 cells.25 typic pattern Although the close association between TEL-AML1 fusion and a common ALL phenotype has been well recognized,2,5,6 Phenotype TEL+ TEL− there has been relatively little work exploring correlation between detailed blast cell phenotypic characteristics and this Yes 45a 50 molecular abnormality. Borkhardt et al6 have shown a statisti- No 6 122 cally significant correlation with myeloid antigen expression Sensitivity Specificity including CD13, a result which was confirmed in a recent 88% 71% study of Baruchel et al26 who also showed a correlation with CD33 expression. Our results confirm a correlation with aOne case deleted TEL. CD13, but not CD33. We also found statistically significant Predictive value; pos, 47%; neg 95%. correlations with expression of CD34 and CD45 which were not previously noted in these other studies. In addition, we further addressed the question of the predictive value of is possible that the three cases demonstrating TEL rearrange- immunophenotype, a much more stringent test of association ments by Southern blot analysis, but lacking TEL-AML1 than a simple statistically significant association. In fact, we expression may represent TEL fused to a novel partner gene. found that many of the markers that were statistically associa- TEL-AML1 rearrangement in childhood B-precursor ALL MJ Borowitz et al 1768 ted with TEL rearrangements had little power to predict the therapy. The specificity of our phenotype for the genetic presence of the rearrangement. abnormality is only about 70%, but the sensitivity is nearly The results of this study extend our previous work showing 90% for predicting TEL rearrangements, and an estimated 92– correlations between immunophenotype and cytogenetic and 93% for predicting TEL-AML1 translocation. This suggests that molecular genetic abnormalities involving the MLL gene13 or phenotyping could be used as a screening test for the genetic the E2A-pbx1 translocation in pre-B ALL.14 These studies share abnormality. Assuming a prevalence rate for this translocation two properties with the current study. First, accurate pheno- of 24%, our results suggest preliminary immunophenotypic typic predictions could not be obtained by simply classifying analysis can eliminate the need for molecular testing in 57% antigen expression as either positive or negative. More com- of patients. Out of 100 patients with B-precursor ALL, only plex descriptions of patterns of expression of combinations of 3/24 patients harboring TEL rearrangements, or only two with antigens was required to demonstrate significant pre- the TEL-AML1 translocation will not be detected. It should be dictability. In the most predictive algorithm in the current cautioned however, that use of this strategy is dependent upon study, blasts that partially expressed either CD9 or CD20 were accurate assessment of the patterns of antigen expression best considered with those that completely lacked the anti- described above, and that the reproducibility of this pattern gens. This same characteristic was seen in the predictive value determination among phenotyping laboratories has not been of CD20 expression for the E2A-pbx1 translocation.14 The established. Nevertheless, when our results are combined with algorithm defined in that study, however, required that CD9 our earlier findings correlating immunophenotype and other be expressed in a bright, homogeneous pattern. Interestingly, clinically important translocations, they suggest a more cost- myeloid leukemias associated with translocations involving effective method for classifying children with B-precursor ALL the AML1 gene are also associated with a distinctive immuno- into important biologically defined risk groups. phenotype, though not surprisingly different antigens are involved.12 The second common feature that our current findings on Acknowledgements the TEL-AML1 share with prior studies is that the algorithm was not a perfect predictor of the genotype. The reasons for This work was supported in part by the following grants from this are not clear, though the group of patients that phenotyp- the National Cancer Institute: CA28476, CA31566, CA15989, ing fails to predict are of potential great interest. The TEL- CA32503, CA30969. AML1 translocation is well known to be associated with an excellent prognosis. It is worth recognizing, however, that dis- ease behavior is fundamentally a phenotypic characteristic. Though many more patients will be required to address this References question, it would be of interest to determine if the few 1 Romana SP, Poirel H, Leconiat M, Flexor M, Mauchauffe M, Jon- patients with TEL rearrangements without the phenotype we veaux P, Macintyre EA, Berger R, Bernard OA. High frequency describe here fare as well as the larger group. of t(12;21) in childhood B-lineage acute lymphoblastic leukemia. It should be noted that although we employed qualitative Blood 1995; 86: 4263–4269. assessments of histogram shape to help classify tumors in this 2 Shurtleff SA, Bujis A, Behm FG, Rubnitz JE, Raimondi SC, Hancock study, we did not employ quantitative analysis of antigen ML, Chan GC, Pui C, Grosveld G, Downing JR. TEL/AML1 fusion expression which we had previously done in other studies.17 resulting from a cryptic (12;21) is the most common genetic lesion in pediatric ALL and defines a subgroup of patients with an excel- In those studies, quantitative analysis was important because, lent prognosis. Leukemia 1995; 9: 1985–1989. in contrast to these studies, there were few differences seen 3 Golub TR, Barker GF, Bohlander SK, Hiebert SW, Ward DC, Bray- between positives and negatives, and important differences Ward P, Morgan E, Raimondi SC, Rowley JD, Gilliland DG. Fusion were seen between high density expression and low. Never- of the TEL gene on 12p13 to the AML1 gene on 21q22 in acute theless, it remains possible that the correlations we describe lymphoblastic leukemia. Proc Natl Acad Sci USA 1995; 92: here could be improved by such quantitative studies. 4917–4921. 2,5 4 McLean TW, Ringold S, Neuberg D. TEL/AML1 dimerizes and is In previous studies, we found that greater than 90% of associated with a favorable outcome in childhood acute lym- cases that demonstrated rearranged TEL also expressed TEL- phoblastic leukemia. Blood 1996; 88: 4252–4258. AML1. In this study, we selectively looked at patients in whom 5 Rubnitz JE, Downing JR, Pui C-H. TEL gene rearrangement in acute we failed to predict the TEL rearrangement by our phenotypic lymphoblastic leukemia: a new genetic marker with prognostic algorithm, and found three of eight cases that were negative significance. J Clin Oncol 1997; 15: 1150–1157. by RT-PCR. It should be noted that TEL may rarely be fused 6 Borkhardt A, Cazzaniga G, Viehmann S, Valsecchi MG, Ludwig 20–23 W-D, Burci L, Mangioni S, Schrappe M, Riehm H, Lampert F et to other genes in B-lineage ALL. It would be of interest to al. Incidence and clinical relevance of TEL/AML1 fusion genes in know if any of these other candidate partner genes were children with acute lymphoblastic leukemia enrolled in the Ger- involved in these three cases. Though the numbers are small, man and Italian multicenter therapy trials. Blood 1997; 90: selection by immunophenotype appeared to identify a sig- 571–577. nificantly larger fraction of cases with discordance between 7 Koeffler HP. Syndromes of acute nonlymphocytic leukemia. Ann the Southern and PCR results than was seen in our previous Intern Med 1987; 107: 748–758. 8 Parkin JL, Arthur DC, Abramson CS, McKenna RW, Kersey JH, study. This suggests that the phenotypic algorithm we propose Heideman RL, Brunning RD. Acute leukemia associated with the may be even more highly correlated with the TEL-AML1 trans- t(4;11) chromosome rearrangement; ultrastructural and immuno- location than with TEL rearrangements. logic characteristics. Blood 1982; 60: 1321–1331. Although more still needs to be done to determine the bio- 9 Bitter MA, LeBeau MM, Rowley JD, Larson RA, Golomb HM, Var- logic significance of the association we describe here, our diman JW. Associations between morphology, karyotype and results have immediate practical consequences. Because the clinical features in myeloid leukemias. Hum Pathol 1987; 18: 211–225. TEL-AML1 translocation is an important independent favor- 10 Adriaansen HJ, te Boekhorst PAW, Hagemeijer AM, van der 2,4,6 able prognostic marker, many new studies of childhood Schoot CE, Delwel HR, van Dongen JJM. Acute myeloid leukemia ALL will require this measurement for assigning patients to M4 with bone marrow eosinophilia (M4E0) and inv(16)(p 13q22) TEL-AML1 rearrangement in childhood B-precursor ALL MJ Borowitz et al 1769 exhibits a specific immunophenoytpe with CD2 expression. Blood typing of acute leukemia by flow cytometric analysis: use of CD45 1993; 81: 3043–3051. and right angle light scatter to gate on leukemic blasts in three- 11 Hunger SP, Tkachuk DC, Amylon MD, Link MP, Carroll AJ, Wel- color analysis. Am J Clin Pathol 1993; 100: 534–540. born JL, Willman CL, Cleary ML. HRX involvement in de novo 19 Borowitz MJ, Bray RA, Gascoyne RD, Melnick S, Parker JW, Picker and secondary leukemias with diverse chromosome 11q23 abnor- LJ, Stetler-Stevenson M. US–Canadian Consensus Recommen- malities. Blood 1993; 81: 3197–3203. dations on the immunophenotypic analysis of hematologic neo- 12 Hurwitz CA, Raimondi SC, Head D, Krance R, Mirro J, Kalwinsky plasia by flow cytometry: data analysis and interpretation. Cyto- DK, Ayers GD, Behm FG. Distinctive immunophenotypic features metry 1997; 30: 236–244. of t(8;21)(q22;q22) acute myeloblastic leukemia in children. 20 Peeters P, Raynaud SD, Cools J. Fusion of TEL, the ETS-variant Blood 1992; 80: 3182–3188. gene 6 (ETV6) to the receptor-associated kinase JAK2 as a result 13 Pui C, Frankel LS, Carroll AJ, Raimondi SC, Shuster JJ, Head DR, of t(9;12) in a lymphoid and t(9;15;12) in a myeloid leukemia. Crist WM, Land VJ, Pullen J, Steuber CP et al. Clinical character- Blood 1997; 90: 2535–2540. istics and treatment outcome of childhood acute lymphoblastic 21 Harada H, Harada Y, Eguchi M, Dohy H, Kamada N. Characteriz- leukemia with the t(4;11)(q21;q23): a collaborative study of 40 ation of acute leukemia with t(4;12). Leuk Lymphoma 1997; 25: cases. Blood 1991; 77: 440–447. 47–53. 14 Borowitz MJ, Hunger SP, Carroll AJ, Shuster JJ, Pullen DJ, Steuber 22 Suto Y, Sato Y, Smith SD, Rowley JD, Bohlander SK. A t(6;12)(q23;p13) results in the fusion of ETV6 to a novel gene, STL, CP, Cleary ML. Predictability of the t(1;19)(q23;p13) from surface in a B cell ALL cell line. Genes Chromosom Cancer 1997; 18: antigen phenotype: implications for screening cases of childhood 254–268. acute lymphoblastic leukemia for molecular analysis: a Pediatric 23 Janssen JW, Ridge SA, Papadopoulos P, Cotter F, Ludwig WD, Oncology Group study. Blood 1993; 82: 1086–1090. Fonatsch C, Rieder H, Ostertag W, Bartram CR, Wiedemann LM. 15 Look AT. Oncogenic transcription factors in the human acute leu- The fusion of TEL and ABL in human acute lymphoblastic leukemias. Science 1997; 278: 1059–1064. kaemia is a rare event. Br J Haematol 1997; 90: 222–224. 16 Look AT, Melvin SL, Williams DL, Brodeur GM, Dahl GV, Kalwin- 24 Olsson I, Bergh G, Ehinger M, Gullberg U. Cell differentiation in sky DK, Murphy SB, Mauer AM. Aneuploidy and percentage of acute myeloid leukemia. Eur J Haematol 1996; 57: 1–16. S-phase cells determined by flow cytometry correlate with cell 25 Xu M, Chen L, Christman JK. Regulation of CD9 expression during phenotype in childhood acute leukemia. Blood 1982; 60: 959– 12-O-tetradecanoyl-phorbol-13- acetate-induced differentiation of 967. human myeloid leukemia (HL-60) cells. Cell Growth Diff 1994; 17 Borowitz MJ, Shuster J, Carroll AJ, Nash M, Look AT, Camitta B, 5: 1225–1234. Mahoney D, Lauer SJ, Pullen DJ. Prognostic significance of fluor- 26 Baruchel A, Cayuela JM, Ballerini P, Landman-Parker J, Cezard V, escence intensity of surface marker expression in childhood B- Firat H, Haddad E, Auclerc MF, Valensi F, Cayre YE, Macintyre precursor acute lymphoblastic leukemia. A Pediatric Oncology EA, Sigaux F. The majority of myeloid-antigen-ositive (My+) child- Group study. Blood 1997; 89: 3960–3966. hood B cell precursor acute lymphoblastic leukaemias express 18 Borowitz MJ, Guenther KL, Shults KE, Stelzer GT. Immunopheno- TEL-AML1 fusion transcripts. Br J Haematol 1997; 99: 101–106.

Appendix Children’s Memorial Hospital (Chicago) Children’s Memorial Hospital (Chicago) CA-07431 Institution Grant Christ Hospital CA-07431 number Rush-Presbyterian CA-07431 Duke University Alberta Pediatric Oncology Consortium Duke University CA-15525 Alberta Children’s Hospital West Virginia University, Charleston CA-15525 Cross Cancer Institute West Virginia University, Morgantown CA-15525 Baylor Eastern Pediatric Oncology Consortium Baylor CA-03161 Eastern Pediatric Oncology Consortium UT/Galveston CA-03161 Mount Sinai Medical School (NY) CA-69428 Bergan-Passiac CCOP University of Maryland CA-69428 Hackensack Medical Center Emory University Boston Floating Hospital Emory University CA-20549 Boston Floating Hospital Florida CCOP Eastern Maine All Children’s Hospital Cardinal Glennon Children’s Hospital Joe DiMaggio Children’s Cardinal Glennon Children’s Hospital Nemours/Orlando Sacred Heart Hospital Carolinas Consortium Tampa Children’s Hospital Carolinas Consortium Carolinas Medical Center CA-69177 Hawaii (NP) CCOP Children’s Hospital Greenville System CA-69177 Cancer Center of Hawaii East Carolina University CA-69177 Johns Hopkins University Presbyterian Hospital CA-69177 Fairfax Hospital CA-28476 Medical University of South Carolina CA-69177 Johns Hopkins University CA-28476 Children’s Hospital Michigan LSU CCOP Children’s Hospital Michigan CA-29691 Children’s Hospital New Orleans/ Hurley Medical Center CA-29691 LSU CCOP St Johns Hospital CA-29691 LSU CCOP TEL-AML1 rearrangement in childhood B-precursor ALL MJ Borowitz et al 1770 Stanford University McGill University Kaiser/Santa Clara CA-33603 Children’s East Ontario CA-33587 Stanford University CA-33603 McGill University CA-33587 University of Arizona CA-33603 Medical College Virginia University of Alabama Medical College Virginia University of Alabama CA-25408 Miami Children’s Hospital University of Arkansas Miami Children’s Hospital University of Arkansas Midwest Children’s Cancer Center University of Florida Midwest Children’s Cancer Center CA-32053 Nemours/Jacksonville New England Consortium University of Florida Dartmouth Hitchcock CA-29293 University of Kansas Massachusetts General Hospital CA-29293 University of Kansas Rhode Island Hospital CA-29293 SUNY Stony Brook CA-29293 University of Miami University of Vermont CA-29293 St Mary’s Hospital University of Miami Oklahoma University Oklahoma University CA-11233 University of Mississippi Medical Center Keesler AFB Hospital CA-15898 Operations Ofﬁce University of Mississippi Medical Center CA-15989 Operations Ofﬁce CA-30969 University of Rochester Roswell Park Cancer Institute University of Rochester Roswell Park Cancer Institute CA-28383 University of Virginia Sainte-Justine University of Virginia University of Laval UC/Davis Schneider Children’s UC/Davis Schneider Children’s UCSD Consortium SPOG Consortium Children’s Hospital (San Diego) CA-28439 SPOG Bern Kaiser Permanente/San Diego CA-28439 SPOG Geneva UC/San Diego CA-28439 SPOG Lausanne US Tex Ped CCOP Suny Syracuse Driscoll Children’s Hospital Suny Syracuse San Antonio MPC & BDC UT/San Antonio SWMSC Cook-Ft Worth Children’s Medical Center CA-33625 USA CCOP SWMSC University of South Alabama Scott and White CA-33625 Unif Serv Onc Cons Southwestern Medical School CA-33625 Madigan Army Medical Center St Christopher’s Hospital Naval Medical Center, Portsmouth St Christopher’s Hospital Tripler Army Medical Center Washington University St Jude Children’s University of Missouri CA-05587 East Tennessee State University CA-31566 St Jude Children’s CA-31566 Wichita CCOP St Francis Regional