No prognostic effect of additional chromosomal abnormalities in children with acute lymphoblastic leukemia and 11q23 abnormalities

Transcription

1 (2005) 19, & 2005 Nature Publishing Group All rights reserved /05 $ No prognostic effect of additional chromosomal abnormalities in children with acute lymphoblastic leukemia and 11q23 abnormalities AV Moorman 1, SC Raimondi 2, C-H Pui 2, A Baruchel 3, A Biondi 4, AJ Carroll 6, E Forestier 5, PS Gaynon 6, J Harbott 7, DO Harms 8, N Heerema 6, R Pieters 9, M Schrappe 7, LB Silverman 10, E Vilmer 11,{, CJ Harrison 1,12 on behalf of the Ponte di Legno Working Group 1 Leukaemia Research Fund Cytogenetics Group, Cancer Sciences Division, University of Southampton, Southampton, UK; 2 St Jude Children s Research Hospital and the University of Tennessee Health Science Center, Memphis, TN, USA; 3 French Acute Lymphoblastic Cooperative Group, France; 4 Associazione Italiana di Ematologia ed Oncologia Pediatrica, Italy; 5 Nordic Society of Pediatric Oncology and Hematology and University of Umeå, Umeå, Sweden; 6 Children s Oncology Group, USA; 7 Berlin-Frankfurt Munster Study Group, Germany; 8 Cooperative ALL Study Group, Germany; 9 Dutch Childhood Oncology Group, Netherlands; 10 Dana Farber Cancer Institute and Harvard Medical School, Boston, MA, USA; 11 European Organization for Research and Treatment of Cancer, France and Belgium; 12 National Cancer Research Institute Childhood Leukaemia Working Party, UK This study characterized the additional chromosomal abnormalities (ACA) associated with 11q23 rearrangements in 450 infants and children with acute lymphoblastic leukemia (ALL) and examined the impact of these ACA on survival. Overall, 213 (47%) cases had ACA but the incidence varied according to patient age and 11q23 subgroup. Infants and patients with t(4;11)(q21;q23) had the lowest incidence of ACA (50/182 (27%) and 57/216 (26%) respectively), whereas patients with del(11)(q23) had the highest incidence (66/93 (71%)). Del(11)(q23) abnormalities were heterogeneous and occasionally secondary to t(9;22)(q34;q11.2). Thus, patients with del(11)(q23) comprised a separate biological entity, which was clearly distinct from those with an 11q23 translocation. The most frequent specific ACA were trisomy X (n ¼ 38), abnormal 12p (n ¼ 32), abnormal 9p (n ¼ 28) and del(6q) (n ¼ 19). The presence of ACA did not change the 5 year event-free survival estimates among children (56% (95% CI 46 65%) vs 62% (54 69%)) or infants (22% (15 29%) vs 18% (9 29%)), nor when the different 11q23 subgroups were analyzed separately. This study has conclusively demonstrated that there is no prognostic effect of secondary chromosomal changes in association with 11q23 abnormalities in childhood ALL. However, characterization of these ACA is important to determine their potential role in initiation of MLL driven leukemogenesis. (2005) 19, doi: /sj.leu Published online 3 March 2005 Keywords: acute lymphoblastic leukemia; childhood; infant; cytogenetics; 11q23; MLL Introduction Acquired chromosomal changes are important predictors of outcome in childhood acute lymphoblastic leukemia (ALL) and are increasingly being used to direct therapy. Patients with ALL and 11q23 abnormalities are generally considered to have a poor prognosis; however, this poor prognosis has been shown to be restricted to patients with rearrangements of the MLL gene. 1,2 Evidence suggests that outcome varies according to the type of MLL rearrangement, the age of the patient or both. For example, in ALL with a t(4;11)(q21;q23), the most frequent 11q23 Correspondence: Dr AV Moorman, LRF Cytogenetics Group, Cancer Sciences Division, University of Southampton, Southampton General Hospital, MP 822, Duthie Building, Tremona Rd, Southampton, SO16 6YD, UK; Fax: þ ; avm@soton.ac.uk { E Vilmer sadly died prior to the completion of this manuscript. Received 6 October 2004; accepted 14 January 2005; Published online 3 March 2005 translocation in ALL, patients aged between 2 and 9 years have an improved overall survival estimate compared with that of other age groups. 3 The prognosis of infants (younger than 1 year) appears to be uniformly poor regardless of the type of MLL rearrangement, 4 whereas among patients with t(11;19)(q23;p13.3), those with T-lineage ALL have survival estimates greater than those of patients with B-lineage ALL patients. 5,6 Most studies investigating the prognostic effects of 11q23 abnormalities in ALL have focused on specific MLL rearrangements with less consideration of the impact of secondary or additional chromosomal abnormalities (ACA). Although no association has been found between survival and ACA, each study analyzed fewer than 50 patients, was restricted to t(4;11)(q21;q23) or both. 3,7 11 The aim of the present study was to characterize the ACA in a large cohort of children with ALL who have been found, by cytogenetic or molecular methods, to have a translocation or deletion involving 11q23 and to assess whether the ACA provide any additional prognostic information. Patients and methods Study population A total of 514 patients with a cytogenetically visible 11q23 abnormality, a molecularly defined MLL rearrangement or both were enrolled on studies of one of 12 cooperative study groups or single institutions in Europe and the United States: AIEOP (n ¼ 13), BFM (n ¼ 57), CCG (n ¼ 93), COALL (n ¼ 7), DCLSG (n ¼ 15), Dana Faber (n ¼ 10), FRALLE (n ¼ 28), POG (n ¼ 110), St Judes (n ¼ 56), UK CLWP (n ¼ 86), EORTC (n ¼ 22) as previously defined, 4,6 and the Nordic Society of Pediatric Oncology and Hematology (NOPHO) (n ¼ 17). 12 Some of the cases included in this data set were also part of the European Concerted Action Workshop on 11q23. 3,5,13,14 All patients had a confirmed diagnosis of ALL and started treatment between 1983 and The median length of follow-up was 7 years 1 month; the range was from 1 year 6 months to 16 years 8 months. Cytogenetic classification All karyotypes were reviewed and described in accordance with the guidelines of the International System for Human Cytogenetic Nomenclature. 15 This cohort of 514 patients comprised

2 (94%) cases with a cytogenetically visible 11q23 and 32 (6%), where the cytogenetics failed, was normal or was not performed, but the MLL gene was shown to be rearranged by fluorescence in situ hybridization (FISH), reverse transcriptase polymerase chain reaction (RT-PCR) or Southern blotting. Among the cases with a cytogenetically visible 11q23 abnormality, the involvement of the MLL gene was confirmed in 45 (9%) cases by molecular methods. The karyotypes were classified as either complete (n ¼ 450) or incomplete (n ¼ 64): the incomplete karyotypes comprised 32 cases with a partial karyotype (11q23 abnormality only reported) and 32 cases in which the 11q23 abnormality was identified solely by FISH, Southern blotting or RT-PCR. All 514 karyotypes were classified into subgroups according to the type of 11q23 abnormality: t(4;11)(q21;q23) (n ¼ 267), t(11;19)(q23;p13.3) (n ¼ 54), other balanced and unbalanced translocations/inversions involving 11q23 (other 11q23) (n ¼ 98) and deletions of 11q23 (del(11)(q23)) (n ¼ 95). The other 11q23 subgroup included the following established 11q23 translocations: t(9;11)(p21;q23) (n ¼ 9), t(10;11)(p12;q23) (n ¼ 11), t(1;11)(p32;q23) (n ¼ 6), t(6;11)(q27;q23) (n ¼ 3). There was breakpoint heterogeneity among the cases described with deletion of 11q23: del(11)(q23) (n ¼ 55), del(11)(q12b14q23) (n ¼ 20), del(11)(q21b22q23) (n ¼ 15) and del(11)(q23q24b25) (n ¼ 5). Only those 450 cases with complete karyotypes were evaluated for the presence of ACA. An ACA was defined as any clonal structural or numerical chromosomal change that accompanied the 11q23 rearrangement, whether present in all cells or restricted to a subpopulation. Three-way translocations involving 11q23 were classified as ACA because at least one locus in addition to those involved in the initial translocation was involved. These additional loci were identified according to the principle that the derived chromosome 11, der(11), was the important translocation product. 16 Thus, der(11) and the usual partner chromosome were classified as the 11q23 translocation, whereas the third partner chromosome was classified as an ACA. Statistical analysis The distribution of cases across different categories was analyzed using the w 2 test. Since age was the most important risk factor in our previous study 4 infants (o1 year) and children (41 year) were considered separately. The principal end point was event-free survival (EFS), which was measured from the start of treatment to the time of the first adverse event, whether relapse or death. This analysis included patients who failed to achieve a complete remission and those who died while their ALL was in remission. Patients who did not suffer an event or die during the follow-up period were censored at the date of last contact. Kaplan Meier life tables and curves were constructed using the log-rank method. 17 The observed:expected (O:E) ratios were from unadjusted log-rank tests that compared two groups. All statistical analyses were performed by using Intercooled Stata v7 (Stata Corporation, College Station, TX, USA). Results Presenting clinical and biological features by type of 11q23 abnormality In this study, 450 (88%) of the 514 patients with ALL and an 11q23 abnormality had a complete karyotype (Table 1). The 64 Table 1 Clinical and biological features of patients with acute lymphoblastic leukemia grouped by 11q23 abnormality Total number of cases (%) (%) of cases with t(4;11) t(11;19) Other 11q23 del(11)(q23) Total 514 (100) 267 (100) 54 (100) 98 (100) 95 (100) Gender Male 264 (51) 133 (50) 23 (43) 55 (56) 53 (56) Female 250 (49) 134 (50) 31 (57) 43 (44) 42 (44) Age (years) a o1 220 (43) 154 (58) 29 (54) 32 (33) 5 (5) (38) 63 (24) 14 (26) 54 (56) 64 (68) X10 96 (19) 49 (18) 11 (20) 11 (11) 25 (27) WBC count ( 10 9 /l) a o (36) 52 (20) 11 (20) 54 (55) 66 (69) (13) 30 (11) 10 (19) 16 (16) 13 (14) X (51) 184 (69) 33 (61) 28 (29) 16 (17) Immunophenotype a B lineage 432 (91) 248 (99) 42 (81) 73 (81) 69 (84) T lineage 42 (9) 2 (1) 10 (19) 17 (19) 13 (16) Karyotype Complete 450 (88) 216 (81) 52 (96) 89 (91) 93 (98) Incomplete 64 (12) 51 (19) 2 (4) 9 (9) 2 (2) ACA b None 237 (53) 159 (74) 26 (50) 25 (28) 27 (29) X1 213 (47) 57 (26) 26 (50) 64 (72) 66 (71) ACA ¼ additional chromosomal abnormality; WBC ¼ white blood cell. a Data for some variables are missing: age (n ¼ 3), WBC count (n ¼ 1), and immunophenotype (n ¼ 40). b Only those cases with a complete karyotype were assessed for the presence of an additional chromosomal abnormality.

3 cases with incomplete karyotypes were significantly more likely to be infants, have a higher white blood cell (WBC) count and have t(4;11) (all three Po0.01). In total, 51 (80%) of these patients had a t(4;11) or MLL-AF4 fusion. In 237 (53%) of the 450 cases with complete karyotypes, the 11q23 rearrangement was the sole chromosomal abnormality observed, whereas in the remaining 213 (47%) cases it was associated with at least one ACA. The incidence of ACA was not uniform across the different 11q23 subgroups (Table 1). Patients with a t(4;11) had a significantly lower incidence of ACA (26%) compared with other subgroups, both individually and collectively (Po0.01). The incidence of ACA varied in relation to other variables, including age, WBC count and immunophenotype, although no gender bias was observed. Infants had a lower incidence of ACA than children (50/182 (27%) vs 163/268 (61%), Po0.01). Although the incidence of t(4;11) was higher among infants than children (123/182 (68%) vs 93/268 (35%), Po0.01), the incidence of ACA in infants with t(4;11) was significantly lower than among children with t(4;11) (23/123 (19%) vs 34/93 (37%), Po0.01). The incidence of ACA was higher in patients with a lower WBC count ( 10 9 /l) (o50: 110/ 172 (64%) vs 450: 103/277 (37%), Po0.01). However, this difference was probably due to the high proportion (65/93 (69%)) of patients with a del(11)(q23), which was associated with a WBC count of o /l. The incidence of ACA among T ALL cases (27/40, 68%) was higher than those with B-lineage ALL (170/372, 46%) (Po0.01). Type and number of ACA Among the 213 patients with ACA, a total of 72 (34%) had both structural and numerical abnormalities, while 93 (44%) and 48 (23%) cases respectively had either structural or numerical abnormalities. The only correlation between the type of ACA and 11q23 subgroup was a higher incidence of structural abnormalities in patients with del(11)(q23) compared with the remaining cases: 61/66 (92%) vs 104/147 (71%), Po0.01. Excluding the handful of cases (n ¼ 13) with high hyperdiploidy, the majority of cases with ACA had just one or two ACA (142/ 200, 71%) (Table 2). The proportion of cases with three or more ACA was significantly greater among those with del(11)(q23) (29/61, 48%) compared with the remaining cases (29/139, 21%) (Po0.01). 559 Table 2 Types of recurrent additional chromosomal abnormalities and their relation to types of 11q23 abnormality Total number of cases of cases with t(4;11) t(11;19) Other 11q23 del(11)(q23) Total cases with ACA Specific Recurrent ACA +X a Abnormal 12p del(12)(p) t(12;v)(p11b13;v) dic(12;v)(p11b13;v) inv(12)(p12b13q12b21) Abnormal 9p del(9)(p)/i(9)(q10) b t(9;v) (p11b13;v) dic(9;v)(p11b13;v) b del(6)(q) del(6)(q13b15q21bqter) c del(6)(q21b23q23bqter) c High hyperdiploidy a a Abnormalities involving 1p i(17)(q10)/del(17)(p) t(9;22)(q34;q11.2) der(4)t(4;11)(q21;q23) 3 3 F F F i(7)(q10) 3 3 d t(2;9)(p11b12;p21b24) of ACAs a Not all cases with ACA had a recurrent abnormality and some had more than one therefore the columns do not add up. ACA ¼ additional chromosomal abnormality; V ¼ various or unknown. a Cases with high hyperdiploidy were excluded. b One case had del(9) and dic(9;v). c No breakpoints were defined in one case. d There were a total of eight cases with t(4;11) and an abnormality of 7p. The other five were: inv(7)(p21q32), inv(7)(p13q32), t(2;7)(p11;p13), add(7)(p?), and t(4;11;7)(q21; q23;p13).

4 560 Numerical chromosomal changes A total of 120 (56%) of the 213 cases with ACA had numerical chromosomal abnormalities, of which 72 (60%) also had a structural chromosomal abnormalities. Chromosome X was the most commonly gained (n ¼ 49), followed by trisomies 21 (n ¼ 22) and 8 (n ¼ 15). These three chromosomes remained the most frequently gained even when the high hyperdiploid karyotypes (n ¼ 13) were removed: þ X(n ¼ 38), þ 8(n ¼ 10), þ 21 (n ¼ 10). A total of 18 (32%) of the 57 cases with t(4;11) and ACA had gained a chromosome X and only two of these were within the context of a high hyperdiploid karyotype. Among 26 cases with t(11;19) and ACA, 12 (46%) had gained a chromosome X while a further four (15%) had trisomy 8; none of these cases had high hyperdiploidy. In contrast, there appeared to be no pattern to the gain of other chromosomes. Monosomy occurred less frequently than trisomy; only 19 (16%) of the 120 cases with a numerical ACA had a modal chromosome number of fewer than 46. The most prevalent monosomy was that of chromosome 21; del(11)(q23) was present in three of the six cases with monosomy 21. Structural chromosomal changes Structural chromosomal abnormalities were observed in 165 (77%) of the 213 patients with ACA, of which 72 (44%) also had numerical abnormalities. Overall, patients with del(11)(q23) accounted for 61 (37%) of the 165 patients with structural ACA. Abnormalities involving 12p were the most common structural changes (n ¼ 32); nearly two-thirds of these (n ¼ 21) occurred in patients with del(11)(q23) (Figure 1). Among cases with balanced and unbalanced 11q23 translocations, 9p (n ¼ 18) was the chromosome arm most frequently involved chromosome arm in ACA followed by 6q (n ¼ 11), 7p (n ¼ 11), 12p (n ¼ 11) and 14q (n ¼ 10). Among 16 cases with three-way or complex translocations, 18 additional chromosomal regions were described. Only two (2p21 and 15q22) were recurrent, seen in two cases each, although the 11q23 partner chromosome was different in each case. The types of abnormality involving 12p, 9p and 6q were often recurrent and frequently associated with del(11)(q23): 21/32 (66%) patients with a 12p ACA had a del(11)(q23) (Table 2). Four patients had t(9;22)(q34;q11.2), three had del(11)(q23) while the fourth was a high hyperdiploid karyotype with an add(11)(q23); the MLL gene status was not known for these cases. Several recurrent ACA were associated with t(4;11), albeit in small numbers of patients (Table 2). In blast cells with ACA, there were no additional copies of the der(11) and extra copies of the derivative partner chromosome were rare: three cases with t(4;11) had an additional copy of der(4)t(4;11). Prognostic associations of ACA We previously reported that age was an independent risk factor among patients with 11q23 abnormalities. 4,6 Therefore, in this study, outcome in infants (o1 year) and children (41 year) was evaluated separately. The median follow-up time was in excess of 7 years. Follow-up was available for all except two children. The 5-year EFS estimate for infants (n ¼ 220) was significantly lower than that of children (n ¼ 292): 21% (95% CI, 16 26%) vs 58% (52 63%), Po There was no difference in EFS estimates between those infants with (n ¼ 182) and without (n ¼ 38) assessable karyotypes (data not shown). However, children with incomplete karyotypes (n ¼ 26) had a worse EFS estimate compared with those with complete karyotypes (n ¼ 266) (Table 3). This difference in survival estimates is probably due to the proportion with t(4;11): among the 26 children without assessable karyotypes 20 (77%) had t(4;11)/ MLL-AF4 compared with 93/266 (35%) among children with a complete karyotype. The presence of ACA did not significantly alter the 5-year EFS estimates of infants with an assessable karyotype (n ¼ 182). The estimates for infants with (n ¼ 50) and without (n ¼ 132) ACA were 18% (9 30%) and 22% (15 29%) respectively (Table 3, Figure 2). Similarly, no significant difference was observed between the 5-year EFS estimates of the 162 children with ACA (62% (54 69%)) and the 104 children with 11q23 as the sole abnormality (57% (46 65%)) (Table 3, Figure 2). The presence of ACA was not associated with outcome when the various 11q23 abnormalities were considered separately within either age category (Table 3). Overall survival was not affected by the presence or absence of ACA (data not shown). The likelihood of detecting ACA is positively correlated with the number of metaphases analyzed. Thus, those cases without Figure 1 Distribution of structural additional chromosomal abnormalities by chromosome arm among 165 cases with acute lymphoblastic leukemia and an 11q23 abnormality.

5 Table 3 Event-free survival estimates for infants and children with acute lymphoblastic leukemia and 11q23 abnormalities, given in relation to the completeness of karyotype and the presence of additional chromosomal abnormalities a 561 Infants (o1 year) Children (41 year) b of cases of events Observed: expected ratio of cases of events Observed: expected ratio Total 220 (100) (100) 125 Karyotype Complete 182 (83) * 266 (91) ** Incomplete 38 (17) (9) ACA c None 132 (73) * 104 (39) * X1 50 (27) (61) q23 abnormality c t(4;11) 123 (100) 93 (100) No ACA 100 (81) * 59 (63) * X1 ACA 23 (19) (37) t(11;19) 27 (100) 25 (100) No ACA 17 (63) * 9 (36) * X1 ACA 10 (37) (64) Other t(11q23) 27 (100) 61 (100) No ACA 14 (52) * 10 (18) * X1 ACA 13 (48) (82) del(11)(q23) 5 (100) 87 (100) No ACA 1 (20) 0 F 26 (30) * X1 ACA 4 (80) 3 F 61 (70) ACA ¼ additional chromosomal abnormality. a Log-rank analysis was used for comparisons. b Two children without follow-up data have been excluded from this table and the relevant analysis: one had t(9;11)(p21;q23) as the sole abnormality and the other had del(11)(q13q23) with ACA. c Only includes cases defined as having a full karyotype. *Not significant (P40.05). **Po Figure 2 Event-free survival estimates for 182 infants (o1 year) and 266 children (41 year) with acute lymphoblastic leukemia and an 11q23 abnormality according to the presence of additional chromosomal abnormalities. ACA (n ¼ 237) were categorized into three groups according to the total number of metaphases analyzed: (a) fewer than 20 (n ¼ 60); (b) 20 or more (n ¼ 65) and (c) unknown (n ¼ 112). The group with 20 or more analyzed metaphases was the least likely to contain misclassified cases (i.e. those with undetected ACA). The 5-year EFS estimate for infants with ACA (n ¼ 50) was then compared with only those infants without ACA whose cytogenetic analysis comprised 20 or more metaphases (n ¼ 40). No difference in 5-year EFS estimates was detected (data not shown). Similarly, no difference in 5-year EFS estimates was observed when the children with ACA (n ¼ 162) were compared with the children without ACA (n ¼ 25) whose cytogenetic analysis comprised 20 or more metaphases (data not shown). Prognosis was not associated with the number of ACA nor the presence of the following specific ACA: gains of chromosomes X, 8 or 21, abnormalities of 9p, deletions of 6q or high hyperdiploidy (data not shown). However, children with a 12p abnormality (n ¼ 31) had a significantly improved outcome when compared with children with other ACA (n ¼ 132) (O/E ¼ 0.37 vs O/E ¼ 1.18, P ¼ 0.008) and all children with assessable karyotypes (n ¼ 237) (O/E ¼ 0.34 vs O/E ¼ 1.10, P ¼ 0.007). Most children with ACA involving an abnormal 12p were in the del(11)(q23) subgroup: 21 out of 31 (68%). Discussion This study is the largest, to date, to characterize the range of ACA associated with rearrangements of 11q23 and to assess the impact of these ACA on the survival of infants and children with ALL. Previously an assessment of this type has been difficult to conduct, because abnormalities involving the chromosomal band 11q23 are heterogeneous at the molecular level, in that they may or may not involve a rearrangement of the MLL gene. 2 Translocations that involve MLL are also heterogeneous, as the MLL gene can be rearranged with a large number of fusion partners. 18 In this study, the MLL status of a number of cases

6 562 remained unconfirmed by molecular techniques; this applied in particular to patients with a del(11)(q23). Therefore, it seemed appropriate to analyze this group separately from the groups with known translocations of MLL. In this large collaborative study, approximately half of the karyotypes that could be assessed showed evidence of additional cytogenetic aberrations. It was noteworthy that the percentage varied according to age and the type of 11q23 abnormality. The largest group of patients had t(4;11) and, although they had a low incidence of ACA, we found several distinctive recurrent secondary abnormalities. These included the gain of chromosome X and abnormalities involving 1p, 7p and 17p, all of which were consistent with previous reports. 3 The second significant group of patients had t(11;19). The recurrent gain of chromosomes X and 8 in association with this translocation also agreed with previous reports. 5 Patients with del(11)(q23), which are the least likely to have MLL gene rearrangements, had the highest incidence of ACA, which included other frequently recognized chromosomal abnormalities in childhood ALL. Abnormalities of 12p, 9p and 6q were present in a considerable proportion of del(11)(q23) cases and three had t(9;22)(q34;q11.2). It is likely that in these Philadelphia chromosome positive cases the del(11)(q23) was a secondary chromosomal change, which has been reported as such in approximately 4% of childhood ALL. 19 Deletions of 11q have been reported in approximately 5% of cases with a t(12;21)(p13;q22). 20 The strong association between abnormalities involving 12p and t(12;21) 20 suggests that this cryptic translocation may occur among the del(11)(q23) patients presented here. Although this proposal has not been tested due to lack of molecular analysis, it is supported by our analysis, which showed that children in this study with an abnormality of 12p had a superior 5-year EFS estimate. The heterogeneity of the del(11)(q23) group and the possibility that several of these deletions were themselves secondary changes highlight the need to analyze this group separately. Neither the overall presence nor the specific type of ACA had an impact on 5-year EFS estimates for the whole population or for the various 11q23 subgroups. This result agrees with those of previous smaller studies, many of which focused on t(4;11). 3,7 11 It was of interest to note that children (41 year) with incomplete karyotypes had a worse EFS estimate than did those with complete karyotypes. This difference was probably due to the identification of MLL-AF4 fusion by molecular screening in a high proportion of children with incomplete karyotypes (80%). This finding further supports our previous observation that children with a t(4;11) have a poorer outcome than those with other 11q23 abnormalities. 4,6 There are two principal difficulties in studying the effects of ACA on the survival of patients with an 11q23 abnormality detected by conventional cytogenetic techniques: firstly, lack of statistical power as these aberrations are infrequent and, secondly, the absence of ACA is difficult to accurately determine. Whereas those cases with ACA were clearly evident, determining whether ACA were not present was more complicated. The accurate classification of cases into the group without ACA depended on the number of metaphases analyzed, which in turn was related to the quality of the preparation and to the proportion of dividing cells. In order to increase the validity and reliability of our findings, we repeated the analysis excluding cases from the group without ACA if the cytogenetic analysis was based on fewer than 20 metaphases. The main shortcoming of the present study was that the patients originated from multiple centers; hence, they were not uniformly treated, nor were the cytogenetic analyses performed to a common standard. Owing to the rarity of infant ALL and childhood ALL with 11q23 abnormalities single institutions rarely accumulate sufficient cases for an in-depth study. Therefore, multicenter collaborative projects are the only way to address such issues. To accurately assess survival in relation to karyotype, we restricted the analysis to cases with complete karyotypes described according to the international standard; thus, variation in the reporting of chromosomal abnormalities was eliminated. Although the patients were not uniformly treated, they were all treated with curative intent, with approximately 90% treated on high-risk protocols. 4 The results of this large study confirm the lack of prognostic effect associated with ACA in infants and children with ALL and an 11q23 rearrangement. We clearly demonstrated the differences in the spectrum of ACA associated with MLL translocations vs deletions of 11q23. These observations emphasized the distinction between cases with MLL rearrangements and those with only a cytogenetically visible 11q23 abnormality, but no proven involvement of the MLL gene. An unresolved issue is the relative prognosis associated with different MLL translocations. Although we have previously reported a significant difference in survival between subgroups among of children with 11q23 abnormalities, 4 the analysis included patients with deletions and a relatively small number of cases with MLL translocations other than t(4;11). This question needs to be addressed in a larger cohort of patients with confirmed 11q23/MLL rearrangements. Although the t(4;11) alone may be sufficient to induce leukemogenesis in infants, 21 further characterization of secondary cytogenetic and genetic abnormalities, for example FLT3 mutations, 22,23 may aid the search for so-called second hits among children with MLL translocations. Acknowledgements AVM and CJH would like to thank the Leukaemia Research Fund in the UK for financial support. The work was also supported in part by grants (CA21765, CA36401, CA51001, CA78224, CA60419 and GM61393) from the US National Institutes of Health, the American Cancer Society FM Kirby Clinical Research Professorship (to CH Pui) and the American Lebanese Syrian Associated Charities (ALSAC). References 1 Behm FG, Raimondi SC, Frestedt JL, Liu Q, Crist WM, Downing JR et al. Rearrangement of MLL gene confers a poor prognosis in childhood acute lymphoblastic leukemia, regardless of presenting age. Blood 1996; 87: Raimondi SC, Frestedt JL, Pui C-H, Downing JR, Head DR, Kersey JH et al. Acute lymphoblastic leukemias with deletion of 11q23 or a novel inversion (11)(p13q23) lack MLL gene rearrangements and have favorable clinical features. Blood 1995; 86: Johansson B, Moorman AV, Haas OA, Watmore AE, Cheung KL, Swanton S et al. Hematologic malignancies with t(4;11)(q21;q23) a cytogenetic, morphologic, immunophenotypic, and clinical study of 183 cases. 1998; 12: Pui C-H, Gaynon PS, Boyett JM, Chessells JM, Baruchel A, Kamps W et al. Outcome of treatment in childhood acute lymphoblastic leukaemia with rearrangements of the 11q23 chromosomal region. Lancet 2002; 359: Moorman AV, Hagemeijer A, Charrin C, Rieder H, Secker-Walker LM. The translocations, t(11;19)(q23;p13.1) and t(11;19)(q23; p13.3): a cytogenetic and clinical profile of 53 patients. 1998; 12:

7 6 Pui C-H, Chessells JM, Camitta B, Baruchel A, Biondi A, Boyett JM et al. Clinical heterogeneity in childhood acute lymphoblastic leukemia with 11q23 rearrangements. 2003; 17: Secker-Walker LM, Stewart EL, Chan L, O Callaghan U, Chessells JM. The (4;11) translocation in acute leukaemia of childhood: the importance of additional chromosomal aberrations. Br J Haematol 1985; 61: Pui C-H, Frankel LS, Carroll AJ, Raimondi SC, Shuster JJ, Head DR et al. Clinical characteristics and treatment outcome of childhood acute lymphoblastic leukemia with the t(4;11)(q21;q23): a collaborative study of 40 cases. Blood 1991; 77: Pui C-H, Hancock ML, Crist WM. Response to Brizard et al Prognosis in children with the t(4;11)(q21;q23) acute leukemia. Blood 1991; 78: Schoch C, Rieder H, Freund M, Hoelzer D, Riehm H, Fonatsch C. Twenty-three cases of acute lymphoblastic leukemia with translocation t(4;11)(q21;q23): the implication of additional chromosomal aberrations. Ann Hematol 1995; 70: Borkhardt A, Wuchter C, Viehmann S, Pils S, Teigler-Schlegel A, Stanulla M et al. Infant acute lymphoblastic leukemia combined cytogenetic, immunophenotypical and molecular analysis of 77 cases. 2002; 16: Forestier E, Johansson B, Gustafsson G, Borgstrom G, Kerndrup G, Johannsson J et al. Prognostic impact of karyotypic findings in childhood acute lymphoblastic leukaemia: a Nordic series comparing two treatment periods. Br J Haematol 2000; 110: Secker-Walker LM. General Report on the European Union Concerted Action Workshop on 11q23, London, UK, May ; 12: Harbott J, Mancini M, Verellen-Dumoulin C, Moorman AV, Secker-Walker LM. Hematological malignancies with a deletion of 11q23: cytogenetic and clinical aspects. 1998; 12: ISCN. An International System for Human Cytogenetic Nomenclature, 1995 edn. Basel, Switzerland: S. Karger, Rowley JD. The der(11) chromosome contains the critical breakpoint junction in the 4;11, 9;11, and 11;19 translocations in acute leukemia. Genes Chromosomes Cancer 1992; 5: Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and Examples. Br J Cancer 1977; 35: Ayton PM, Cleary ML. Molecular mechanisms of leukemogenesis mediated by MLL fusion proteins. Oncogene 2001; 20: Heerema NA, Harbott J, Galimberti S, Camitta BM, Gaynon PS, Janka-Schaub G, et al., for the Acute Lymphoblastic Study Groups. ALL-BFM and CoALL (Germany) AIEOP (Italy) DCLSG (Netherlands) FRALLE (France) CCG DFCI POG and St Jude (USA) and UKALL (UK). Secondary cytogenetic aberrations in childhood Philadelphia chromosome positive acute lymphoblastic leukemia are nonrandom and may be associated with outcome. 2004; 18: Raynaud SD, Dastugue N, Zoccola D, Shurtleff SA, Mathew S, Raimondi SC. Cytogenetic abnormalities associated with the t(12;21): a collaborative study of 169 children with t(12;21)- positive acute lymphoblastic leukemia. 1999; 13: Greaves MF, Maia AT, Wiemels JL, Ford AM. in twins: lessons in natural history. Blood 2003; 102: Taketani T, Taki T, Sugita K, Furuichi Y, Ishii E, Hanada R et al. FLT3 mutations in the activation loop of tyrosine kinase domain are frequently found in infant ALL with MLL rearrangements and pediatric ALL with hyperdiploidy. Blood 2004; 103: Armstrong SA, Mabon ME, Silverman LB, Li A, Gribben JG, Fox EA et al. FLT3 mutations in childhood acute lymphoblastic leukemia. Blood 2004; 103: