Leukemia (2013) 27, 295–304 & 2013 Macmillan Publishers Limited All rights reserved 0887-6924/13 www.nature.com/leu

ORIGINAL ARTICLE Prognostic value of genetic alterations in children with first bone marrow relapse of childhood B-cell precursor acute lymphoblastic leukemia

S Krentz1, J Hof1, A Mendioroz1, R Vaggopoulou1,PDo¨ rge2, C Lottaz3, JC Engelmann3, TWL Groeneveld1,GKo¨ rner1, K Seeger1, C Hagemeier1, G Henze1, C Eckert1, A von Stackelberg1,4 and R Kirschner-Schwabe1,4

Despite risk-adapted treatment, survival of children with relapse of acute lymphoblastic leukemia (ALL) remains poor compared with that of patients with initial diagnosis of ALL. Leukemia-associated genetic alterations may provide novel prognostic factors to refine present relapse treatment strategies. Therefore, we investigated the clinical relevance of 13 recurrent genetic alterations in 204 children treated uniformly for relapsed B-cell precursor ALL according to the ALL-REZ BFM 2002 protocol. The most common alterations were deletions of CDKN2A/2B, IKZF1, PAX5, ETV6, fusion of ETV6-RUNX1 and deletions and/or mutations of TP53. Multivariate analysis identified IKZF1 deletion and TP53 alteration as independent predictors of inferior outcome (P ¼ 0.002 and P ¼ 0.001). Next, we investigated how both alterations can improve the established risk stratification in relapsed ALL. Intermediate-risk relapse patients with low minimal residual disease are currently considered to have a good prognosis. In this group, deletion of IKZF1 and alteration of TP53 identify patients with significantly inferior outcome (Po0.001). In high-risk relapse patients, deletion of IKZF1 is strongly predictive of a second relapse after stem cell transplantation (Po0.001). We conclude that IKZF1 and TP53 represent relevant prognostic factors that should be considered in future risk assessment of children with relapsed ALL to indicate treatment intensification or intervention.

Leukemia (2013) 27, 295–304; doi:10.1038/leu.2012.155 Keywords: IKZF1; TP53; relapse; BCP-ALL; childhood

INTRODUCTION with high levels of minimal residual disease (MRD) at the end of In childhood acute lymphoblastic leukemia (ALL), B20% of the induction therapy (week 5), that is, a burden of 1 or more À 3 8 patients suffer a relapse.1 Children with relapsed ALL generally leukemic cells in 1000 normal cells (X10 ). receive a more intense therapy compared with children with initial Despite this risk-adapted treatment, a substantial number of diagnosis of ALL; still, long-term outcome after relapse remains children with relapsed ALL still suffer a subsequent event, even unsatisfactory with cure rates of merely 30–40%.2–4 in subgroups that present clinically with features of low risk. The German Berlin-Frankfurt-Mu¨nster study group on relapsed Therefore, an early and accurate identification of relapse patients ALL (ALL-REZ BFM) aims at optimising risk-adapted treatment at high risk of therapy failure is essential to further adapt and strategies for children with ALL relapse. In the current ALL-REZ improve treatment intensification and intervention strategies. BFM 2002 protocol, the intensity of chemotherapy and the Leukemia-associated genetic alterations may provide novel prog- allocation of patients to hematopoietic stem cell transplantation nostic tools to refine the risk assessment of children with ALL (HSCT) are based on clinical prognostic factors, that is, the time to relapse. In the treatment of children with initial diagnosis of ALL, relapse relative to primary treatment, the site of relapse and the recurrent chromosomal translocations leading to the fusion genes immunophenotype. An early recurrence of the disease (o6 BCR-ABL1, MLL-AFF1 or ETV6-RUNX1 are established prognostic months after completion of primary treatment), bone marrow factors used for risk stratification of patients.9–14 Furthermore, (BM) involvement, and T-cell immunophenotype are associated genome-wide analyses in B-cell precursor (BCP)-ALL discovered with inferior outcome.5–7 A combination of these prognostic frequent copy number alterations in genes either involved in factors is used to stratify relapse patients into a standard-risk (S1), lymphoid development (for example, EBF1, PAX5, IKZF1) or in cell intermediate-risk (S2) and high-risk group (S3/S4). After a common cycle regulation (for example, CDKN2A, CDKN2B, RB1).15–18 Another intensive induction therapy, these groups are allocated to recurrent deletion affecting the genes CSF2RA, IL3RA and P2RY8 different treatment regimens that consist of multi-agent was identified in the pseudoautosomal region 1 (PAR1) of the sex chemotherapy and radiation therapy if necessary. Further chromosomes, leading to a fusion of the P2RY8 and CRLF2 gene treatment intensification by HSCT is required for high-risk (S3/ (P2RY8-CRLF2).19,20 Of these copy number alterations, particularly S4) patients given that a second complete remission (CR2) can be IKZF1 and PAR1 deletions have been associated with poor induced. HSCT is also indicated for intermediate-risk (S2) patients outcome after primary treatment of children with ALL.21–27

1Department of Pediatric Oncology and Hematology, Charite´ - Universita¨tsmedizin Berlin, Berlin, Germany; 2Department of Pediatrics, University Hospital Schleswig Holstein, Kiel, Germany and 3Institute of Functional Genomics, University of Regensburg Medical School, Regensburg, Germany. Correspondence: Dr R Kirschner-Schwabe, Department of Pediatric Oncology and Hematology, Charite´ – Universita¨tsmedizin Berlin, Campus Virchow, Augustenburger Platz 1, Berlin 13353, Germany. E-mail: [email protected] 4These authors contributed equally to this work. Received 3 April 2012; revised 23 May 2012; accepted 5 June 2012; accepted article preview online 13 June 2012; advance online publication, 3 July 2012 Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 296 In children with relapsed ALL, much less in-depth knowledge is experiment. A gene was considered to contain a deletion if at least one available on the frequency and the prognostic value of genetic MLPA probe targeted to this gene showed a copy number loss. alterations. In previous studies of relapsed ALL, ETV6-RUNX1 has been associated with a favourable and BCR-ABL1 with an adverse Sensitivity and specificity of MLPA data prognosis.28–30 In contrast, deletions of CDKN2A/2B did not affect 31 To test the sensitivity of the MLPA method, we performed serial dilution the outcome after relapse treatment at all. However, these experiments by titrating DNA from the leukemic cell line MHH-cALL2 with analyses were based on heterogeneous patient cohorts enrolled in DNA from healthy donors. The cell line showed a homozygous loss of different ALL-REZ BFM trials. Hence, the aim of our current study CDKN2A/2B and trisomy of all X-chromosome targeted MLPA probes of the was to systematically explore the value of leukemia-associated P335 ‘ALL-IKZF1’ kit. In consistence with Schwab et al.,33 we observed a genetic alterations as prognostic factors in relapsed ALL in a single detection limit of the MLPA method at 20% of abnormal DNA. The majority trial setting. Therefore, we studied the frequency and clinical of analysed leukemic samples (156/204) had a blast percentage of X80% relevance of 13 genetic alterations in 204 children with first BM before enrichment of MNC by Ficoll gradient centrifugation. However, in relapse of BCP-ALL who were treated uniformly according to the five cases the percentage of blasts was between 19 and 30% at sample collection and before enrichment. We were able to detect genetic current ALL-REZ BFM 2002 relapse protocol. alterations by MLPA in samples with blast percentages as low as 19, 24 Specifically, we investigated copy number alterations of EBF1, and 25%. Only two samples below 50% (21 and 28%) were fully negative IKZF1, CDKN2A, CDKN2B, PAX5, ETV6, BTG1, RB1 and the PAR1 with the P335 MLPA kit (Supplementary Figure 1). To test the specificity of region as well as fusion genes ETV6-RUNX1, BCR-ABL1 and the MLPA method, we confirmed the MLPA results by copy number data MLL-AFF1. Furthermore, we integrated in this study the informa- derived from SNP 6.0 microarrays (Affymetrix, Santa Clara, CA, USA) tion on mutations and deletions of the TP53 gene (collectively available from 14 leukemic samples of the studied patients (2 BM samples termed TP53 alterations) from a previous report, in which we from initial diagnosis, 12 BM samples from relapse diagnosis). For EBF1, described a high correlation between TP53 alteration and IKZF1, CDKN2A, CDKN2B, PAX5, ETV6, RB1 and P2RY8, we found a perfect treatment failure after ALL relapse.32 Our present study shows match of MLPA and microarray-derived data (Supplementary Table 1). Only one patient with a PAX5 deletion had a discordant result for PAX5 exon 10, that deletion of IKZF1 and alteration of TP53 are independent whereas PAX5 exons 6 and 8 were shown to be deleted using both the predictors of poor outcome after ALL relapse. Therefore, we methods. Thus, our data confirms the reliability of the P335 ‘ALL-IKZF1’ kit suggest to integrate these genetic factors in future risk described by Schwab et al.33 Validation of the copy number status for assessment of children with relapsed ALL to improve risk- BTG1, CSF2R and IL3R was not possible because these genes were adapted treatment strategies. insufficiently covered by the 6.0 SNP array. For patients with a deletion of CSF2R, IL3R and P2RY8 (PAR1 deletion), we further confirmed the presence of the resulting fusion gene P2RY8-CRLF2 by reverse transcriptase PCR (RT- MATERIALS AND METHODS PCR). Total RNA was available from 12 of the 13 patients with the PAR1 deletion. A total of 1 mg was reverse transcribed using random priming and Patients and samples SuperScript II Reverse Transcriptase in a final volume of 20 ml according to All patients were enrolled in the German multi-center randomised trial the manufacturer’s instructions (Invitrogen, Darmstadt, Germany). A total ALL-REZ BFM 2002 (NCT00114348) approved by the local ethics of 1 ml of the RT reaction was used for subsequent PCR amplification of the committee. Written informed consent was obtained from patients or fusion gene with primers located in P2RY8 exon 1 (forward) and CRLF2 guardians. Patients were selected for the study if they had a first isolated or exon 3 (reverse) described by Cario et al.22 PCR conditions were as follows: combined BM relapse of BCP-ALL, if they were treated according to the 10 min at 95 1C; 35 cycles: 30 s at 95 1C, 45 s at 61 1C, 45 s at 72 1C; 7 min at trial protocol, and if high-quality DNA and RNA from relapse diagnosis 72 1C. The fusion gene was present in all 12 patients tested. Of two were available. The selected cohort included eight patients with Down’s randomly chosen samples, the PCR product was sequenced to confirm the syndrome who were treated according to the trial protocol without fusion transcript. deviations. For details on risk stratification and treatment of patients see the Supplementary Materials and Methods. The median percentage of leukemic blasts in BM or peripheral blood samples from relapse diagnosis Detection of translocation-associated fusion genes was 90% at the time of sample collection (minimum, 19%; maximum, 99%; Fusion genes ETV6-RUNX1, MLL-AFF1 and BCR-ABL1 were detected by RT- Supplementary Figure 1). Before any further analysis, all BM or peripheral PCR. The RT reaction was carried out as described above. A total of 1 mlof blood samples were enriched for mononuclear cells (MNC) by ficoll density cDNA was used for subsequent PCR amplification at 65 1C universal gradient centrifugation. DNA was purified from MNCs using the Gentra annealing temperature. ABL1 was amplified to control for RNA integrity. Puregene Cell Kit (Qiagen, Hilden, Germany). RNA was isolated from MNCs ETV6-RUNX1, and minor and major BCR-ABL1 were detected in duplicate using the Qiagen RNeasy Mini Kit with DNase digestion and including by real-time semi-quantitative PCR using serial dilutions of RNA from homogenisation by QIAshredder (Qiagen). fusion gene-positive cell lines REH, TOM-1 and K562 in RNA derived from MNCs of healthy volunteers as standard. MLL-AFF1 was amplified by conventional nested PCR using a 10 À 3 dilution of RNA from the Detection of copy number alterations cell line SEM as sensitivity control. Primers and probes are given in The copy number status of EBF1, IKZF1, CDKN2A, CDKN2B, PAX5, ETV6, BTG1, Supplementary Table 2. RB1, and of CSF2RA, IL3RA, P2RY8 (PAR1 deletion) was analysed by multiplex ligation-dependent probe amplification (MLPA), using the probe set P335 ‘ALL-IKZF1’ from MRC-Holland (Amsterdam, The Netherlands), Statistical analysis according to the manufacturer’s instructions. The probe targeted to P2RY8 Equality of categorical and continuous variables was analysed by Pearson’s exon 2 was a synthetic MLPA oligo pair that we designed according to the w2 or Fisher’s exact test and by Mann–Whitney U or Kruskal–Wallis test, guidelines of MRC-Holland. We added this probe to the kit to enhance the respectively. Probabilities of event-free (pEFS) and overall survival (pOS) detection of a PAR1 deletion (50-TGTTTGCGTAAAAGGCCACGGTCACCACGT were estimated by Kaplan–Meier analyses and differences were compared TGCAATCTAGATTGGATCTTGCTGGCAC-30;50-GGGTTCCCTAAGGGTTGGAGC with a log-rank test. EFS time was defined as time between relapse TGATACAGGTCATGGTGAGGATGCTGGAATACA-30; length of amplification diagnosis and the date of analysis or a subsequent event (second relapse, product: 112 nucleotides). MLPA amplification products were separated secondary malignancy or death in CR). In case of death during induction or and quantified by capillary electrophoresis using an ABI 3730 DNA non-response to chemotherapy, EFS time was set to zero. OS time was Analyzer with GeneScan 500 LIZ as internal size standard (both Applied defined as time from relapse diagnosis to death or for surviving patients to Biosystems, Life Technologies Corporation, Carlsbad, CA, USA). DNA the date of analysis. The cumulative incidence of relapse (CIR) was fragment analysis and peak sizing was carried out with ABI Peak Scanner analysed according to Kalbfleisch and Prentice, and differences were Software v1.0 (Applied Biosystems, Life Technologies Corporation). Data compared using Gray statistics. Multivariate Cox regression modelling was were normalised using the Coffalyser MLPA-DAT software provided by done for EFS using backward stepwise selection to remove non-significant MRC-Holland (Amsterdam, The Netherlands). Six DNA samples from MNCs factors. Statistical calculations were performed using SPSS PASW 18.0.1 either of healthy blood donors or of BM samples from ALL relapse patients software for Windows (SPSS Inc., Chicago, IL, USA). For CIR statistics, the R with an isolated extramedullary relapse were run as references in each package cmprsk of the R software v2.10.1 was used.34

Leukemia (2013) 295 – 304 & 2013 Macmillan Publishers Limited Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 297 RESULTS AND DISCUSSION and the surviving patients had a mean follow-up time of 4.4 years Characteristics of the studied patients (median, 4.2 years; range 1.5–8.4 years) enabling appropriate We assessed the prognostic value of 13 genetic alterations in 204 survival analysis. Our study covers the most common genetic patients with first isolated or combined BM relapse of B-cell alterations in childhood ALL; however, cytogenetic information on, precursor (BCP) ALL. Bone marrow relapses of BCP origin for example, ploidy could not be included due to a lack of routine characterise the majority (70–75%) of all patients enrolled in the karyotyping in the ALL-REZ BFM 2002 trial. Still, to the best of our current ALL-REZ BFM trial. Relapse of T-cell ALL is observed in only knowledge, the present study represents the most comprehensive 10–15% of patients and was not considered in our study as it analysis so far on the frequency and prognostic value of genetic constitutes a distinct leukemia subtype with a different pattern of alterations in relapsed childhood BCP-ALL. genetic alterations.35,36 In addition, the minor subgroup of isolated extramedullary relapses (10–15% of all patients, ALL-REZ BFM Frequency of genetic alterations in children with first BM relapse 2002 standard-risk group S1 and partially intermediate-risk group of BCP-ALL S2) was excluded from our investigation due to an insufficient The most common genetic alterations were deletions of CDKN2B number of extramedullary specimen. Altogether, our study (37.7%), CDKN2A (37.3%), IKZF1 (33.3%), PAX5 (26.5%), ETV6 (25%), includes more than half of all BCP-ALL patients with a first BM fusion of ETV6-RUNX1 (15.2%) and alterations of TP53 (11.3%; relapse (204/343, 59%) enrolled in the current ALL-REZ BFM 2002 Figure 1a). All other genetic alterations that we investigated, that trial within a 7-year-period (between 2002 and 2009). We is, deletions of BTG1, RB1 and EBF1, as well as amplifications of observed no selection bias regarding clinical, biological and PAX5 and fusion genes P2RY8-CRLF2, MLL-AFF1 and BCR-ABL1, treatment parameters (Supplementary Table 3). All relapse were present in less than 10% of the patients (Figure 1a). The size patients were treated uniformly according to the trial protocol of the deletions in CDKN2A, CDKN2B, IKZF1, PAX5 and ETV6 as

CDKN2B del 37.7% CDKN2A del 37.3% IKZF1 del 33.3% PAX5 del 26.5% ETV6 del 25.0% ETV6-RUNX1 15.2% TP53 alt 11.3% BTG1 del 9.3% RB1 del 6.4% P2RY8-CRLF2 6.4% Down's Syndrome 3.9% PAX5 amp 3.4% MLL-AFF1 2.9% EBF1 del 2.9% BCR-ABL1 2.5% % of all patients (n=204) 0 5 10 15 20 25 30 35 40 45

del del del del del del amp del del CDKN2BCDKN2AIKZF1PAX5ETV6ETV6-RUNX1TP53 BTG1alt RB1 P2RY8-CRLF2Down'sPAX5 SyndromeMLL-AFF1EBF1 BCR-ABL1 CDKN2B del 77 70 33 31 16 98 9 4 8 3 6 222 CDKN2A del 76 33 33 17 78 10 4 8 3 6 122 IKZF1 del 68 22 20 0 5 939 43313 PAX5 del 54 12 67 7 3 5 3 0 0 1 2 ETV6 del 51 161 9 342104 1 ETV6-RUNX1 31 3 7 300005 0 TP53 alt 23 012112 01 BTG1 del 19 1320010 RB1 del 13 000001 P2RY8-CRLF2 13 6 1000 overlap Down's Syndrome p 8 0000 PAX5 amp p 7 000 MLL-AFF1 6 00 exclusion EBF1 del p 6 0 BCR-ABL1 5 Figure 1. Frequency and overlap of genetic alterations in 204 children with first BM relapse of BCP-ALL. (a) Percentage of genetic alterations in the studied cohort. (b) Overlap of genetic alterations. Grey boxes indicate the total number of patients with a respective genetic alteration. Other boxes indicate the number of patients sharing two particular alterations. Blue or red color indicates a significant overlap or exclusion, respectively. Owing to a significant overlap of Down’s syndrome with the P2RY8-CRLF2 fusion gene,20,53 the number of children with Down’s syndrome in our study is given for information purposes. alt, alteration; amp, amplification; del, deletion.

& 2013 Macmillan Publishers Limited Leukemia (2013) 295 – 304 Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 298 detected by MLPA is given in Supplementary Figure 2. The early or early relapse (P ¼ 0.007), whereas nearly all ETV6-RUNX1- frequency of IKZF1 deletion in our relapse patients is approxi- positive patients (29/31) presented with a late relapse after the mately twice as high as the frequency described in children with completion of primary treatment (Po0.001).29–31 Accordingly, initial diagnosis of ALL (33 vs 14–19%).23,25 This makes deletion of patients with a CDKN2A/2B deletion were more often stratified IKZF1 the second most frequent alteration in children with into the high-risk group (P ¼ 0.054), whereas those having the relapsed ALL after deletion of CDKN2A/2B. The increase of IKZF1 ETV6-RUNX1 fusion gene were allocated to the intermediate-risk deletions at ALL relapse presumably results from the high relapse group (Po0.001). In contrast, the site of relapse did not correlate rate of these patients after primary treatment.23–27 However, as with any of the genetic alterations investigated. Regarding the IKZF1 deletions can also be acquired at relapse,23,37,38 we assessed response to treatment, patients with CDKN2A/2B deletion or with the copy number status of IKZF1 in matched samples from initial the ETV6-RUNX1 fusion gene more frequently achieved a second diagnosis in 49 out of the 68 patients with an IKZF1 deletion CR (P ¼ 0.045 and P ¼ 0.023, respectively), whereas alterations of detected at relapse. We observed that the IKZF1 deletion emerged TP53 were associated with a non-response to treatment as at relapse in approximately one fourth of the patients (12/49, described previously (P ¼ 0.020).32 On the level of MRD (in 24%), thus contributing to the high frequency of this alteration in intermediate-risk patients), children having a CDKN2A/2B relapsed ALL (Supplementary Figure 3 and Supplementary deletion also showed a trend towards a better response to Table 4). Together with our previous finding that TP53 alterations treatment (P ¼ 0.057). In contrast, ETV6-RUNX1 and TP53 alteration were acquired in half of the respective relapse patients,32 these were not associated with levels of MRD after induction therapy. data strongly emphasise the need to re-analyse genetic alterations For IKZF1, PAX5 and ETV6 deletion, we observed no correlation at the relapse stage of the disease before their use as prognostic with clinical prognostic factors, risk group assignment or response factors in children with relapsed ALL. to treatment at ALL relapse. In contrast to IKZF1 deletion, the P2RY8-CRLF2 fusion barely increased its frequency at relapse in our study, although it has also been associated with a higher rate of relapse after primary Association of genetic alterations with outcome treatment on some ALL protocols (6.4 vs 4–5% at initial The outcome of relapse patients with different genetic alterations diagnosis).22,39,40 Similar, the adverse prognostic factors BCR- is summarised in Table 2 for alterations present in 410% of ABL1 and MLL-AFF1(refs 10,12,13) show no higher prevalence in our relapse patients (o10% see Supplementary Table 6). Deletions of relapse cohort. In fact, the proportion of BCR-ABL1-positive relapse IKZF1 had an adverse effect on the outcome of children with ALL patients even decreased when compared with the earlier relapse relapse. Both pEFS and pOS were significantly decreased in trial ALL-REZ BFM 90 (2.5 vs 9.8%).28 This likely reflects the patients with an IKZF1 deletion compared with IKZF1 non-deleted amendments in primary treatment for BCR-ABL1-positive patients, patients (pEFS: 0.30±0.06 vs 0.51±0.05, P ¼ 0.002; pOS: for example, high-risk stratification, intensified treatment and 0.36±0.07 vs 0.60±0.05, P ¼ 0.001; Table 2 and Figure 2a). The more frequent transplantation in first continuous remission (CR1) most common event in relapse patients with an IKZF1 deletion from the early nineties onwards.41 The comparatively low was a subsequent relapse shown by an increased cumulative frequency of, for example, P2RY8-CRLF2, BCR-ABL1 and MLL-AFF1 incidence of second relapse (CIR: 0.41±0.07 vs 0.23±0.04; in relapsed ALL (o10%) hampers a reasonable statistical analysis P ¼ 0.006; Table 2 and Figure 2a). Hence, the adverse effect of within our study, particularly if only subgroups of relapse patients IKZF1 deletion on outcome of children with initial diagnosis of are considered. Therefore, we focussed our further analysis on ALL23–27 persists in children with relapsed ALL despite patients those genetic alterations present in more than 10% of relapse receiving an intensified relapse treatment. patients, that is, deletions of CDKN2A, CDKN2B, IKZF1, PAX5, ETV6, Similar to patients with an IKZF1 deletion, children with a fusion gene ETV6-RUNX1 and alterations of TP53. Even so, deletion of CDKN2A/2B had a significantly increased incidence of statistical calculations on the alterations present in o10% of the second relapse compared with CDKN2A/2B non-deleted patients relapse patients are given in the Supplementary Information. (CIR: 0.40±0.06 vs 0.21±0.04; Po.001; Table 2 and Figure 2b). However, as observed in previous ALL-REZ BFM trials,31 there was Overlap of genetic alterations no significantly different pEFS and pOS between patients with or without a CDKN2A/2B deletion (pEFS: 0.45±0.06 vs 0.43±0.05, Several of the genetic alterations occurred simultaneously in the P ¼ 0.990; pOS: 0.48±0.07 vs 0.54±0.05, P ¼ 0.443; Table 2 and patients (Figure 1b). Generally, the pattern of overlap in our Figure 2b). In fact, patients with a CDKN2A/2B deletion more relapse patients resembles that observed in children with initial frequently achieved a second remission (Table 1); however, due to diagnosis of ALL.18,25,42,43 More than 90% of relapse patients with the high incidence of subsequent relapses, the pEFS of patients a CDKN2A deletion also had a deletion of CDKN2B (P 0.001).44 We o with CDKN2A/2B deletion ultimately was similar to the pEFS of therefore considered patients with a CDKN2A and/or CDKN2B patients without the deletion. This ambiguous result adds to the deletion as one group (CDKN2A/2B) in all subsequent clinical existing controversy on the prognostic value of CDKN2A/2B analyses. Deletions of CDKN2A and CDKN2B also coincided deletions (Sulong et al.48 and references therein). It has been significantly with deletions of PAX5 (P ¼ 0.001 and P 0.001) and o postulated earlier that the conflicting prognostic significance of IKZF1 (P ¼ 0.022 and P ¼ 0.032).43,45 Patients with an ETV6-RUNX1 CDKN2A/2B deletions might be caused by an overlap of this fusion gene frequently showed an accompanying deletion of alteration with several distinct genetic subgroups of ALL.48 This is ETV6 (P 0.001),46,47 whereas the ETV6-RUNX1 rearrangement was o in line with the significant overlap of CDKN2A/2B and IKZF1 mutually exclusive with IKZF1 deletions (P 0.001).25,43 Alterations o deletions in our ALL relapse patients (Figure 1b) that might of TP53 were not associated with any of the genetic alterations contribute to the observed increased relapse rate in patients with investigated. CDKN2A/2B deletions. In contrast to CDKN2A/2B deletions, dele- tions of PAX5 were not associated with outcome after first relapse Association of genetic alterations with clinical characteristics at all, although both genes are located on the short arm of Next, we examined the correlation between the genetic altera- chromosome 9. To clarify the different association of CDKN2A/2B tions and clinical prognostic factors, risk group assignment and and PAX5 alterations with outcome, we further separated the response to treatment. For genetic alterations present in 410% of patients into three subgroups based on the MLPA data: (i) patients relapse patients see Table 1 and for o10% of relapse patients see with a deletion of both loci (CDKN2A/2B and PAX5), (ii) patients Supplementary Table 5. As in previous ALL-REZ BFM trials, relapse with a CDKN2A/2B deletion only and (iii) patients with a PAX5 patients with a CDKN2A/2B deletion frequently suffered a very deletion only (Supplementary Table 7). The pEFS and pOS did not

Leukemia (2013) 295 – 304 & 2013 Macmillan Publishers Limited & 03McilnPbihr Limited Publishers Macmillan 2013

Table 1. Association of genetic alterations with prognostic factors, risk stratification and response to treatment of patients with first BM relapse of BCP-ALL

CDKN2A/2B deletion IKZF1 deletion PAX5 deletion ETV6 deletion ETV6-RUNX1 TP53 alteration

All Negative Positive P Negative Postive P Negative Positive P Negative Positive P Negative Positive P Negative Positive P patients n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%)

All patients 204 (100) 121 (59) 83 (41) 136 (67) 68 (33) 150 (74) 54 (26) 153 (75) 51 (25) 173 (85) 31 (15) 181 (89) 23 (11)

Time of relapsea Very early 44 (22) 18 (15) 26 (31) 0.007 26 (19) 18 (27) 0.448 32 (21) 12 (22) 0.926 38 (25) 6 (12) 0.119 44 (25) 0 (–) o0.001 36 (20) 8 (35) 0.192 Early 42 (21) 23 (19) 19 (23) 30 (22) 12 (18) 30 (20) 12 (22) 32 (21) 10 (20) 40 (23) 2 (7) 37 (20) 5 (22) Late 118 (58) 80 (66) 38 (46) 80 (59) 38 (56) 88 (59) 30 (56) 83 (54) 35 (69) 89 (51) 29 (94) 108 (60) 10 (44)

Site of relapseb BM isolated 162 (79) 101 (84) 61 (74) 0.112 106 (78) 56 (82) 0.582 122 (81) 40 (74) 0.326 123 (80) 39 (77) 0.690 140 (81) 22 (71) 0.229 142 (79) 20 (87) 0.423 BM combined 42 (21) 20 (17) 22 (27) 30 (22) 12 (18) 28 (19) 14 (26) 30 (20) 12 (24) 33 (19) 9 (29) 39 (22) 3 (13) eeiso easdcidodBcl rcro ALL precursor Krentz B-cell S childhood relapsed of Genetics Relapse-risk groupc Intermediate-risk (S2) 130 (64) 84 (70) 46 (55) 0.054 88 (65) 42 (62) 0.758 96 (64) 34 (63) 1.000 93 (62) 37 (73) 0.178 101 (58) 29 (94) o0.001 119 (66) 11 (48) 0.109 High-risk (S3/S4) 74 (36) 37 (31) 37 (45) 48 (35) 26 (38) 54 (36) 20 (37) 60 (39) 14 (28) 72 (42) 2 (7) 62 (34) 12 (52)

MRD (S2 only)d al et Low (o10 À 3 ) 64 (55) 34 (48) 30 (67) 0.057 47 (60) 17 (45) 0.163 47 (54) 17 (59) 0.830 48 (57) 16 (50) 0.535 48 (54) 16 (59) 0.665 60 (56) 4 (44) 0.512 High (X10 À 3 ) 52 (45) 37 (52) 15 (33) 31 (40) 21 (55) 40 (46) 12 (41) 36 (43) 16 (50) 41 (46) 11 (41) 47 (44) 5 (56)

CR2 achieved Yes 165 (81) 92 (76) 73 (88) 0.045 113 (83) 52 (77) 0.345 120 (80) 45 (83) 0.689 124 (81) 41 (80) 1.000 135 (78) 30 (97) 0.023 151 (83) 14 (61) 0.020 No 39 (19) 29 (24) 10 (12) 23 (17) 16 (24) 30 (20) 9 (17) 29 (19) 10 (20) 38 (22) 1 (3) 30 (17) 9 (39)

Abbreviations: ALL, acute lymphoblastic leukemia; BCP, B-cell precursor; BM, bone marrow; CR2, second complete remission; MRD, minimal residual disease. aTime of relapse: very early, o18 months after initial diagnosis of ALL; early, X18 months after initial diagnosis of ALL; late, X6 months after completion of primary treatment. bSite of relapse: isolated BM, no evidence of extramedullary disease; combined BM, more than 5% leukemic cells in the BM combined with central nervous system, testis or other extramedullary disease. cRelapse-risk group: high-risk, early isolated or very early isolated or combined BM relapse of BCP-ALL; intermediate risk, late isolated or combined BM relapse or early combined BM relapse of BCP-ALL. dMRD measured at the end of induction therapy (5 weeks): low level, o10 À 3 leukemic cells; high level, X10 À 3 leukemic cells in BM aspirate. Note: Table shows genetic alterations with a prevalence of more than 10% in the study cohort. Bold entries indicate statistically significant results (P-value o0.05). ekma(03 9 304 – 295 (2013) Leukemia 299 Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 300 Table 2. Outcome of children with first BM relapse of BCP-ALL by genetic alteration

Genetic alteration Patients Eventsa pEFS P Relapse CIR P Death pOS P nn±s.e. (Log-rank) n ±s.e. (Gray) n ±s.e. (Log-rank)

CDKN2A/2B deletion Negative 121 63 0.43±0.05 0.990 21 0.21±0.04 o0.001 49 0.54±0.05 0.443 Positive 83 43 0.45±0.06 31 0.40±0.06 39 0.48±0.07

IKZF1 deletion Negative 136 61 0.51±0.05 0.002 27 0.23±0.04 0.006 48 0.60±0.05 0.001 Positive 68 45 0.30±0.06 25 0.41±0.07 40 0.36±0.07

PAX5 deletion Negative 150 78 0.45±0.04 0.865 35 0.27±0.04 0.118 66 0.51±0.05 0.866 Positive 54 28 0.41±0.08 17 0.37±0.08 22 0.56±0.08

ETV6 deletion Negative 153 77 0.45±0.05 0.293 39 0.30±0.04 0.815 63 0.55±0.05 0.318 Positive 51 29 0.42±0.07 13 0.27±0.07 25 0.41±0.10

ETV6-RUNX1 Negative 173 101 0.37±0.04 o0.001 49 0.33±0.04 0.029 84 0.46±0.04 o0.001 Positive 31 5 0.84±0.07 3 0.10±0.06 4 0.84±0.08

TP53 alteration Negative 181 87 0.48±0.04 o0.001 45 0.29±0.04 0.595 74 0.53±0.04 0.030 Positive 23 19 0.10±0.09 7 0.31±0.10 14 0.39±0.10 Abbreviations: ALL, acute lymphoblastic leukemia; BCP, B-cell precursor; CIR, cumulative incidence of relapse; pEFS, probability of event-free survival; pOS, probability of overall survival; s.e., standard error. aEvents: induction death, non-response to chemotherapy, secondary malignancy, second relapse, treatment-related death. Note: Table shows genetic alterations with a prevalence of more than 10% in the study cohort. Bold entries indicate statistically significant results (P-value o0.05). overall survival (probability) overall event-free survival (probability) event-free cumulative incidence of relapse cumulative

p=0.002 p=0.006 p=0.001

years post relapse diagnosis years post relapse diagnosis years post relapse diagnosis

no IKZF1 del: n=136, cens.=75, pEFS=0.51±0.05 no IKZF1 del: n=136, cens.=109, CIR=0.23±0.04 no IKZF1 del: n=136, cens.=88, pOS=0.60±0.05 IKZF1 del: n= 68, cens.=23, pEFS=0.30±0.06 IKZF1 del: n= 68, cens.= 43, CIR=0.41±0.07 IKZF1 del: n= 68, cens.=28, pOS=0.36±0.07 overall survival (probability) overall event-free survival (probability) event-free cumulative incidence of relapse cumulative

p=0.990 p<0.001 p=0.443

years post relapse diagnosis years post relapse diagnosis years post relapse diagnosis

no CDKN2A/2B del: n=121, cens.=58, pEFS=0.43±0.05 no CDKN2A/2B del: n=121, cens.=100, CIR=0.21±0.04 no CDKN2A/2B del: n=121, cens.=72, pOS=0.54±0.05

CDKN2A/2B del: n= 83, cens.=40, pEFS=0.45±0.06 CDKN2A/2B del: n= 83, cens.= 52, CIR=0.40±0.06 CDKN2A/2B del: n= 83, cens.=44, pOS=0.48±0.07 Figure 2. Survival analysis of patients by IKZF1 or CDKN2A/2B deletion. Probability of EFS (left), cumulative incidence of second relapse (CIR, middle) and probability of OS (right) of children with first relapse of BCP-ALL differentiated by IKZF1 deletion (IKZF1 del, a) and by CDKN2A and/or CDKN2B deletion (CDKN2A/2B del, b). cens, censored; del, deletion.

Leukemia (2013) 295 – 304 & 2013 Macmillan Publishers Limited Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 301 Table 3. Multivariate Cox regression model for event-free survival of significantly differ among subgroups (P ¼ 0.629 and 0.811). In patients with first BM relapse of childhood B-cell precursor ALL contrast, the incidence of a second relapse was higher in patients who had a deletion of at least CDKN2A/2B (groups (i) and (ii); Parameter n Hazard ratio 95% Confidence P-value P ¼ 0.002, Supplementary Table 7). Interestingly, these two interval for subgroups also showed an increased frequency of IKZF1 deletion hazard ratio (33% and 56%, respectively, versus 21% in group (iii)), which again indicates that IKZF1 deletion contributes to the higher relapse Lower Upper rates here. Together, these data underscore the notion that Time of relapsea o0.001 deletions of CDKN2A/2B or PAX5 frequently occur as accompany- Late 118 1 ing alterations that might contribute to leukemogenesis but, Early 42 3.13 1.91 5.12 however, are less relevant for resistance to chemotherapy or Very early 41 5.57 3.47 8.95 treatment failure.25 The ETV6-RUNX1 fusion gene was generally associated with a TP53 alteration 0.001 good prognosis in our study, that is, a high rate of second No 181 1 remission (Table 1) and a favourable pEFS and pOS (Table 2), Yes 23 2.29 1.38 3.8 which is comparable to observations from previous ALL-REZ BFM 29,30 IKZF1 deletion 0.002 trials. However, the outcome was worse if an accompanying No 136 1 deletion in the ETV6 gene was present. This deletion occurred in Yes 68 1.89 1.27 2.81 52% of the ETV6-RUNX1-positive relapse patients (vs 20% in ETV6- RUNX1-negative patients, P 0.001) and was associated with a a ¼ Time of relapse: very early, o18 months after initial diagnosis of ALL; decreased pEFS (0.68±0.12) compared with ETV6-RUNX1-positive X X early, 18 months after initial diagnosis of ALL; late, 6 months after patients lacking this extra deletion (1.00±0.00, P ¼ 0.020). completion of primary treatment. Note: multivariate analysis included time to relapse, site of relapse, IKZF1 deletion, ETV6-RUNX1 fusion gene and TP53 Secondary ETV6 deletions have been associated with outcome alteration. Best model is shown. in ETV6-RUNX1-positive ALL previously, but so far the prog- nostic effect of this deletion remains inconsistent (reviewed by

intermediate-risk (S2) patients intermediate-risk (S2), MRD low intermediate-risk (S2), MRD high event-free survival (probability) event-free event-free survival (probability) event-free event-free survival (probability) event-free

p=0.042 p=0.013 p=0.836

years post relapse diagnosis years post relapse diagnosis years post relapse diagnosis

no IKZF1 del: n=88, cens.=61, pEFS=0.63±0.06 no IKZF1 del: n=47, cens.=39, pEFS=0.81±0.06 no IKZF1 del: n=31, cens.=18, pEFS=0.51±0.10 IKZF1 del: n=42, cens.=22, pEFS=0.45±0.09 IKZF1 del: n=17, cens.= 9, pEFS=0.48±0.14 IKZF1 del: n=21, cens.=13, pEFS=0.50±0.14

high-risk (S3/S4) patients high-risk (S3/S4), alloHSCT high-risk (S3/S4), alloHSCT

p=0.012 p=0.001 event-free survival (probability) event-free survival (probability) event-free cumulative incidence of relapse cumulative

p<0.001

years post relapse diagnosis years post relapse diagnosis years post relapse diagnosis no IKZF1 del: n=48, cens.=14, pEFS=0.29±0.07 no IKZF1 del: n=24, cens.=14, pEFS=0.57±0.10 no IKZF1 del: n=24, cens.=20, CIR=0.17±0.08 IKZF1 del: n=26, cens.= 1, pEFS=0.04±0.04 IKZF1 del: n=11, cens.= 1, pEFS=0.09±0.09 IKZF1 del: n=11, cens.= 1, CIR=0.91±0.11 Figure 3. Deletion of IKZF1 identifies patients with poor outcome within different relapse-risk groups. (a) Effect of IKZF1 deletion on the pEFS of intermediate-risk (S2) relapse patients: left, all S2 patients; middle, S2 patients with low MRD level (o10 À 3); right, S2 patients with high MRD level (X10 À 3). (b) Effect of IKZF1 deletion on outcome of high-risk (S3/S4) relapse patients: left, pEFS of all S3/S4 patients; middle and right, pEFS and CIR of S3/S4 patients with allogeneic HSCT in CR2. cens, censored; 2; del, deletion; MRD, MRD at the end of induction therapy (week 5).

& 2013 Macmillan Publishers Limited Leukemia (2013) 295 – 304 Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 302 Peter et al.47). Like our observations, Attarbaschi et al.49 showed time to relapse, deletion of IKZF1 and alteration of TP53 as that the outcome of patients with an accompanying deletion of the strongest predictive factors for an unfavourable pEFS (Table 3). ETV6 was inferior, whereas other studies found no prognostic As in previous ALL-REZ BFM reports, the ETV6-RUNX1 rearrange- correlation.18,46 However, the total number of ETV6-RUNX1- ment did not remain in the model due to its strong association positive patients in our relapse study was limited and validation with the time of relapse, specifically with late relapse, of the prognostic significance of the ETV6 deletion is required. and therefore cannot be regarded as independent prognostic Finally, alterations of the TP53 gene were highly correlated with factor.30 Interestingly, in our study the established prognostic an inferior pEFS and pOS (Po0.001 and P ¼ 0.030; Table 2) factor site of relapse,4,6,7 that is, combined BM relapse versus as already demonstrated in our previous report.32 isolated BM relapse, was also not maintained in our Cox regression model.

Multivariate Cox regression analysis We performed multivariate Cox regression analysis to assess IKZF1 and TP53 as prognostic factors to improve risk assessment at the independence of the genetic alterations that were significantly first BM relapse of BCP-ALL associated with EFS in univariate analysis (IKZF1 deletion, To investigate how IKZF1 deletions and TP53 alterations can aid ETV6-RUNX1 fusion gene and TP53 alteration) as well as the in improving the established risk stratification of children with established clinical prognostic factors of ALL relapse (time to relapsed ALL, we examined their prognostic significance within relapse and site of relapse). The Cox regression model identified the ALL-REZ BFM 2002 risk groups, that is, intermediate-risk (S2)

intermediate-risk (S2), MRD low intermediate-risk (S2), MRD high event-free survival (probability) event-free survival (probability) event-free

p<0.001 p=0.993 0.0 0.0

years post relapse diagnosis years post relapse diagnosis

IKZF1 and TP53 wt: n=43, cens.=39, pEFS=0.89±0.05 IKZF1 and TP53 wt: n=27, cens.=16, pEFS=0.53±0.11 IKZF1 deletion: n=17, cens.= 9, pEFS=0.48±0.14 IKZF1 deletion: n=20, cens.=12, pEFS=0.45±0.15 TP53 alteration: n= 4, cens.= 0, pEFS=0.00±0.00 TP53 alteration: n= 5, cens.= 3, pEFS=0.60±0.22

clinical markers response molecular markers 100

IKZF1 / TP53 wt MRD low 75 late combined BM; IKZF1 / TP53 alt. late isolated BM; S2 early combined BM

50 MRD high % of patients IKZF1 wt 25 early isolated BM; CR2, HSCT very early combined BM; S3/S4 IKZF1 del very early isolated BM non-response, BCP ALL relapse with BM involvement early adverse event 0 Figure 4. IKZF1 deletion and TP53 alteration as prognostic factors to subclassify BCP-ALL relapse patients. (a) Presence of an IKZF1 deletion or TP53 alteration results in a significantly decreased pEFS in intermediate-risk (S2) relapse patients with low MRD level (o10 À 3)attheendofinduction therapy. (b) In contrast, in S2 patients with high MRD level (X10 À 3), IKZF1 deletion and TP53 alteration had no significant prognostic value. (c)We suggest to include IKZF1 and TP53 in a third stratification step following the established risk assessment of relapse patients by clinical and response-based factors. In the intermediate-risk (S2) group with low MRD, the presence of an IKZF1 deletion or TP53 alteration can identify patients with inferior prognosis. For these patients treatment intensification by, for example, HSCT may be considered as it is currently practiced for S2 patients with high MRD. In the high-risk (S3/S4) group, IKZF1 deletion is strongly predictive of a second relapse in children who achieved CR2 and received a HSCT. Therefore, these patients may be eligible for treatment intervention by experimental therapeutic strategies. alt, alteration; BCP, B-cell precursor; BM, bone marrow; cens, censored; CR2, 2nd complete remission; HSCT, hematopoietic stem cell transplantation; MRD, MRD at the end of induction therapy (week 5); pEFS, probability of event-free survival; wt, wild type.

Leukemia (2013) 295 – 304 & 2013 Macmillan Publishers Limited Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 303 and high-risk (S3/S4) patients, as well as in further clinical and assessment of children with ALL relapse in a third step following the prognostic subgroups. In both the intermediate-risk and the high- established stratification by clinical markers and MRD response risk group, deletions of IKZF1 and alterations of TP53 were (Figure 4c). We propose to investigate the IKZF1 and TP53 status in associated with inferior outcome. This is shown for IKZF1 deletion upcoming relapse trials to confirm our results on a prospective basis. in Figures 3a and b, whereas data on the effect of TP53 alteration in these subgroups were published earlier by us.32 According to the ALL-REZ BFM 2002 protocol high-risk relapse CONCLUSION patients who achieve a second CR after chemotherapy further We show that of 13 genetic alterations that were investigated receive a HSCT. In this subgroup of patients, IKZF1 deletion was in our study, IKZF1 deletion and TP53 alteration had an indepen- particularly predictive for a second event identifying children with dent prognostic significance in patients with relapsed pediatric a pEFS as low as 9% (pEFS: 0.09±0.09 vs 0.57±0.10, P ¼ 0.001; BCP-ALL. Provided prospective validation, these genetic factors Figure 3b). Remarkably, these events were exclusively second may be included in the risk assessment of children with ALL relapses that occurred in less than 2 years post HSCT (CIR: relapse to identify subgroups that require intensified and/or 0.91±0.11 vs 0.17±0.08, Po0.001; Figure 3b). Given this alternative treatment. obviously very poor outcome with the current protocol treatment, IKZF1 deletion might represent a valuable novel marker to identify children with high-risk ALL relapse who are eligible for treatment CONFLICT OF INTEREST intervention by experimental therapeutic strategies (Figure 4c). The authors declare no conflict of interest. So far, only MRD before HSCT has been established as reliable marker for outcome after allogeneic HSCT in CR2 (Bader et al.50 and references therein). In a preliminary analysis, the prognostic ACKNOWLEDGEMENTS value of IKZF1 deletion in high-risk relapse patients seemed to be We thank the German Jose´ Carreras Leukemia Foundation (Grants No. DJCLS R 08/17 independent from MRD before HSCT (data not shown); however, to RKS and CE, and DJCLS A 09/01 to AvS), the German National Genome Research the number of children with available MRD data in this subgroup Network (Grants No. 01GS0870 to RKS and CH, and 01GS0882 to CL), the German was small (n 22) and further validation is required. Interestingly, Foundation for Childhood Cancer (Grants No. DKS 2003.08 and 2007.02 to AvS and ¼ GH, and DKS 2003.03 to CE and GH) and KinderLeben Berlin, incorporated society, the association with poor outcome after HSCT was restricted to for financial support. high-risk relapse patients as IKZF1 deletion did not predict a second relapse in intermediate-risk patients after allogeneic HSCT in CR2 (Supplementary Figure 4). This differential effect of IKZF1 REFERENCES in intermediate- and high-risk relapse patients supports the idea 1 Pui CH, Carroll WL, Meshinchi S, Arceci RJ. Biology, risk stratification, and therapy that these groups might constitute two clinically and molecularly of pediatric acute leukemias: an update. J Clin Oncol 2011; 29: 551–565. 4,51,52 distinct entities. Alterations of TP53 did not have any prog- 2 Ko RH, Ji L, Barnette P, Bostrom B, Hutchinson R, Raetz E et al. Outcome of patients nostic value for outcome post HSCT, neither in intermediate-risk treated for relapsed or refractory acute lymphoblastic leukemia: a therapeutic nor in high-risk relapse patients (data not shown); however, advances in childhood leukemia consortium study. J Clin Oncol 2010; 28: 648–654. the number of patients with TP53 alteration per subgroup were 3 van den Berg H, de Groot-Kruseman HA, Damen-Korbijn CM, de Bont ES, limited and larger studies are required. Schouten-van Meeteren AY, Hoogerbrugge PM. Outcome after first relapse in In the intermediate-risk group, we found a particularly predic- children with acute lymphoblastic leukemia: a report based on the Dutch Child- hood Oncology Group (DCOG) relapse all 98 protocol. Pediatr Blood Cancer 2011; tive value of IKZF1 deletion in those patients with low, but not 57: 210–216. with high MRD levels at the end of induction therapy (Figure 3a). 4 Tallen G, Ratei R, Mann G, Kaspers G, Niggli F, Karachunsky A et al. Long-term A similar limitation to MRD good responders has been observed outcome in children with relapsed acute lymphoblastic leukemia after time-point in children with initial diagnosis of ALL, in whom a prognostic and site-of-relapse stratification and intensified short-course multidrug chemo- effect of IKZF1 deletion was described in patients with medium or therapy: results of trial ALL-REZ BFM 90. J Clin Oncol 2010; 28: 2339–2347. low, but not in those with high MRD burden.26 Regarding TP53, 5 Einsiedel HG, von Stackelberg A, Hartmann R, Fengler R, Schrappe M, Janka- we observed the same differential effect in MRD stratified Schaub G et al. Long-term outcome in children with relapsed ALL by risk-stratified intermediate-risk patients. Alterations of TP53 were associated salvage therapy: results of trial acute lymphoblastic leukemia-relapse study of the with poor outcome in intermediate-risk patients with low Berlin-Frankfurt-Munster Group 87. J Clin Oncol 2005; 23: 7942–7950. ± ± 6 Gaynon PS, Qu RP, Chappell RJ, Willoughby ML, Tubergen DG, Steinherz PG et al. MRD (0.00 0.00 vs 0.77 0.06, Po0.001) but not in those with Survival after relapse in childhood acute lymphoblastic leukemia: impact of site high MRD burden (0.60±0.22 vs 0.50±0.09, P ¼ 0.958). Again, and time to first relapse—the Children’s Cancer Group Experience. Cancer 1998; the number of patients with TP53 alteration per subgroup was 82: 1387–1395. limited and validation is required. 7 Nguyen K, Devidas M, Cheng SC, La M, Raetz EA, Carroll WL et al. Factors influ- Currently, intermediate-risk patients with low MRD level are encing survival after relapse from acute lymphoblastic leukemia: a Children’s considered to have a good prognosis with standard intensive Oncology Group study. Leukemia 2008; 22: 2142–2150. consolidation and maintenance chemotherapy8 and are therefore 8 Eckert C, Biondi A, Seeger K, Cazzaniga G, Hartmann R, Beyermann B et al. not allocated to HSCT in the current ALL-REZ BFM Prognostic value of minimal residual disease in relapsed childhood acute lym- 2002 protocol. So far, no additional prognostic factors have phoblastic leukaemia. Lancet 2001; 358: 1239–1241. 9 Kamps WA, Bokkerink JP, Hakvoort-Cammel FG, Veerman AJ, Weening RS, been identified to stratify this group further and to indicate the van Wering ER et al. BFM-oriented treatment for children with acute lympho- requirement of treatment intensification. Therefore, it is blastic leukemia without cranial irradiation and treatment reduction for standard noteworthy that the IKZF1 and TP53 status enables the risk patients: results of DCLSG protocol ALL-8 (1991–1996). Leukemia 2002; 16: distinction of subgroups with significantly different outcome here. 1099–1111. Intermediate-risk relapse patients with low MRD and without 10 Kamps WA, van der Pal-de Bruin KM, Veerman AJ, Fiocco M, Bierings M, Pieters R. deletion of IKZF1 or alteration of TP53 have a particular favourable Long-term results of Dutch Childhood Oncology Group studies for children with pEFS of almost 90% after treatment with chemotherapy only acute lymphoblastic leukemia from 1984 to 2004. Leukemia 2009; 24: 309–319. (0.89±0.05, Po0.001, Figure 4a). In contrast, intermediate-risk 11 Moricke A, Reiter A, Zimmermann M, Gadner H, Stanulla M, Dordelmann M et al. patients with low MRD levels but with an IKZF1 deletion or TP53 Risk-adjusted therapy of acute lymphoblastic leukemia can decrease treatment ± burden and improve survival: treatment results of 2169 unselected pediatric alteration achieved pEFS rates of merely 0.48 0.014 and and adolescent patients enrolled in the trial ALL-BFM 95. Blood 2008; 111: 0.00±0.00, respectively, and might therefore benefit from further 4477–4489. treatment intensification. Hence, we suggest including IKZF1 12 Moricke A, Zimmermann M, Reiter A, Henze G, Schrauder A, Gadner H et al. deletion and TP53 alteration as prognostic factors in the risk Long-term results of five consecutive trials in childhood acute lymphoblastic

& 2013 Macmillan Publishers Limited Leukemia (2013) 295 – 304 Genetics of relapsed childhood B-cell precursor ALL S Krentz et al 304 leukemia performed by the ALL-BFM study group from 1981 to 2000. Leukemia for the detection of copy number abnormalities in B-cell precursor acute 2010; 24: 265–284. lymphoblastic leukemia. Genes Chromosomes Cancer 2010; 49: 1104–1113. 13 Pui CH, Pei D, Sandlund JT, Ribeiro RC, Rubnitz JE, Raimondi SC et al. Long-term 34 Scrucca L, Santucci A, Aversa F. Competing risk analysis using R: an easy guide for results of St Jude Total Therapy Studies 11, 12, 13A, 13B, and 14 for childhood clinicians. Bone Marrow Transplant 2007; 40: 381–387. acute lymphoblastic leukemia. Leukemia 2009; 24: 371–382. 35 Meijerink JP. Genetic rearrangements in relation to immunophenotype and out- 14 Pui CH, Sandlund JT, Pei D, Campana D, Rivera GK, Ribeiro RC et al. Improved come in T-cell acute lymphoblastic leukaemia. Best Pract Res Clin Haematol 2010; outcome for children with acute lymphoblastic leukemia: results of Total 23: 307–318. Therapy Study XIIIB at St Jude Children’s Research Hospital. Blood 2004; 104: 36 Chiaretti S, Foa R. T-cell acute lymphoblastic leukemia. Haematologica 2009; 94: 2690–2696. 160–162. 15 Mullighan CG, Goorha S, Radtke I, Miller CB, Coustan-Smith E, Dalton JD et al. 37 Mullighan CG, Phillips LA, Su X, Ma J, Miller CB, Shurtleff SA et al. Genomic analysis Genome-wide analysis of genetic alterations in acute lymphoblastic leukaemia. of the clonal origins of relapsed acute lymphoblastic leukemia. Science 2008; 322: Nature 2007; 446: 758–764. 1377–1380. 16 Strefford JC, Worley H, Barber K, Wright S, Stewart AR, Robinson HM et al. Genome 38 Yang JJ, Bhojwani D, Yang W, Cai X, Stocco G, Crews K et al. Genome-wide copy complexity in acute lymphoblastic leukemia is revealed by array-based com- number profiling reveals molecular evolution from diagnosis to relapse in parative genomic hybridization. Oncogene 2007; 26: 4306–4318. childhood acute lymphoblastic leukemia. Blood 2008; 112: 4178–4183. 17 Kuiper RP, Schoenmakers EF, van Reijmersdal SV, Hehir-Kwa JY, van Kessel AG, 39 Ensor HM, Schwab C, Russell LJ, Richards SM, Morrison H, Masic D et al. Demo- van Leeuwen FN et al. High-resolution genomic profiling of childhood ALL reveals graphic, clinical, and outcome features of children with acute lymphoblastic novel recurrent genetic lesions affecting pathways involved in lymphocyte dif- leukemia and CRLF2 deregulation: results from the MRC ALL97 clinical trial. Blood ferentiation and cell cycle progression. Leukemia 2007; 21: 1258–1266. 2011; 117: 2129–2136. 18 Kawamata N, Ogawa S, Zimmermann M, Kato M, Sanada M, Hemminki K et al. 40 Harvey RC, Mullighan CG, Chen IM, Wharton W, Mikhail FM, Carroll AJ et al. Molecular allelokaryotyping of pediatric acute lymphoblastic leukemias by high- Rearrangement of CRLF2 is associated with mutation of JAK kinases, alteration of resolution single nucleotide polymorphism oligonucleotide genomic microarray. IKZF1, Hispanic/Latino ethnicity, and a poor outcome in pediatric B-progenitor Blood 2008; 111: 776–784. acute lymphoblastic leukemia. Blood 2010; 115: 5312–5321. 19 Russell LJ, Capasso M, Vater I, Akasaka T, Bernard OA, Calasanz MJ et al. 41 Arico` M, Schrappe M, Hunger SP, Carroll WL, Conter V, Galimberti S et al. Clinical Deregulated expression of cytokine receptor gene, CRLF2, is involved in lymphoid outcome of children with newly diagnosed Philadelphia chromosome-positive transformation in B-cell precursor acute lymphoblastic leukemia. Blood 2009; 114: acute lymphoblastic leukemia treated between 1995 and 2005. J Clin Oncol 2010; 2688–2698. 28: 4755–4761. 20 Mullighan CG, Collins-Underwood JR, Phillips LA, Loudin MG, Liu W, Zhang J et al. 42 Moorman AV, Ensor HM, Richards SM, Chilton L, Schwab C, Kinsey SE et al. Rearrangement of CRLF2 in B-progenitor- and Down syndrome-associated acute Prognostic effect of chromosomal abnormalities in childhood B-cell precursor lymphoblastic leukemia. Nat Genet 2009; 41: 1243–1246. acute lymphoblastic leukaemia: results from the UK Medical Research Council 21 Harvey RC, Mullighan CG, Chen IM, Wharton W, Mikhail FM, Carroll AJ et al. ALL97/99 randomised trial. Lancet Oncol 2010; 11: 429–438. Rearrangement of CRLF2 is associated with mutation of JAK kinases, alteration of 43 Mullighan CG, Miller CB, Radtke I, Phillips LA, Dalton J, Ma J et al. BCR-ABL1 IKZF1, Hispanic/Latino ethnicity, and a poor outcome in pediatric B-progenitor lymphoblastic leukaemia is characterized by the deletion of Ikaros. Nature 2008; acute lymphoblastic leukemia. Blood 2010; 115: 5312–5321. 453: 110–114. 22 Cario G, Zimmermann M, Romey R, Gesk S, Vater I, Harbott J et al. Presence of the 44 Mirebeau D, Acquaviva C, Suciu S, Bertin R, Dastugue N, Robert A et al. The P2RY8-CRLF2 rearrangement is associated with a poor prognosis in non-high-risk prognostic significance of CDKN2A, CDKN2B and MTAP inactivation in B-lineage precursor B-cell acute lymphoblastic leukemia in children treated according to acute lymphoblastic leukemia of childhood. Results of the EORTC studies 58881 the ALL-BFM 2000 protocol. Blood 2010; 115: 5393–5397. and 58951. Haematologica 2006; 91: 881–885. 23 Kuiper RP, Waanders E, van der Velden VH, van Reijmersdal SV, Venkatachalam R, 45 Kim M, Choi JE, She CJ, Hwang SM, Shin HY, Ahn HS et al. PAX5 deletion is Scheijen B et al. IKZF1 deletions predict relapse in uniformly treated pediatric common and concurrently occurs with CDKN2A deletion in B-lineage acute precursor B-ALL. Leukemia 2010; 24: 1258–1264. lymphoblastic leukemia. Blood Cells Mol Dis 2011; 47: 62–66. 24 Mi JQ, Wang X, Yao Y, Lu HJ, Jiang XX, Zhou JF et al. Newly diagnosed acute 46 Alvarez Y, Coll MD, Ortega JJ, Bastida P, Dastugue N, Robert A et al. Genetic lymphoblastic leukemia in China (II): prognosis related to genetic abnormalities in abnormalities associated with the t(12;21) and their impact in the outcome of 56 a series of 1091 cases. Leukemia 2012; 2:23. patients with B-precursor acute lymphoblastic leukemia. Cancer Genet Cytogenet 25 Mullighan CG, Su X, Zhang J, Radtke I, Phillips LA, Miller CB et al. Deletion of IKZF1 2005; 162:21–29. and prognosis in acute lymphoblastic leukemia. N Engl J Med 2009; 7:7. 47 Peter A, Heiden T, Taube T, Korner G, Seeger K. Interphase FISH on TEL/AML1 26 Waanders E, van der Velden VH, van der Schoot CE, van Leeuwen FN, van Reij- positive acute lymphoblastic leukemia relapses - analysis of clinical relevance of mersdal SV, de Haas V et al. Integrated use of minimal residual disease classifi- additional TEL and AML1 copy number changes. Eur J Haematol 2009; 6:6. cation and IKZF1 alteration status accurately predicts 79% of relapses in pediatric 48 Sulong S, Moorman AV, Irving JA, Strefford JC, Konn ZJ, Case MC et al. A com- acute lymphoblastic leukemia. Leukemia 2011; 25: 254–258. prehensive analysis of the CDKN2A gene in childhood acute lymphoblastic leu- 27 Yang YL, Hung CC, Chen JS, Lin KH, Jou ST, Hsiao CC et al. IKZF1 deletions predict kemia reveals genomic deletion, copy number neutral loss of heterozygosity, and a poor prognosis in children with B-cell progenitor acute lymphoblastic leukemia: association with specific cytogenetic subgroups. Blood 2009; 113: 100–107. a multicenter analysis in Taiwan. Cancer Sci 2011; 102: 1874–1881. 49 Attarbaschi A, Mann G, Konig M, Dworzak MN, Trebo MM, Muhlegger N et al. 28 Beyermann B, Adams HP, Henze G. Philadelphia chromosome in relapsed child- Incidence and relevance of secondary chromosome abnormalities in childhood hood acute lymphoblastic leukemia: a matched-pair analysis. Berlin-Frankfurt- TEL/AML1 þ acute lymphoblastic leukemia: an interphase FISH analysis. Leukemia Munster Study Group. J Clin Oncol 1997; 15: 2231–2237. 2004; 18: 1611–1616. 29 Seeger K, Adams HP, Buchwald D, Beyermann B, Kremens B, Niemeyer C et al. TEL- 50 Bader P, Kreyenberg H, Henze GH, Eckert C, Reising M, Willasch A et al. Prognostic AML1 fusion transcript in relapsed childhood acute lymphoblastic leukemia. The value of minimal residual disease quantification before allogeneic stem-cell Berlin-Frankfurt-Munster Study Group. Blood 1998; 91: 1716–1722. transplantation in relapsed childhood acute lymphoblastic leukemia: the ALL-REZ 30 Seeger K, von Stackelberg A, Taube T, Buchwald D, Korner G, Suttorp M et al. BFM Study Group. J Clin Oncol 2009; 27: 377–384. Relapse of TEL-AML1—positive acute lymphoblastic leukemia in childhood: a 51 Kirschner-Schwabe R, Lottaz C, Todling J, Rhein P, Karawajew L, Eckert C et al. matched-pair analysis. J Clin Oncol 2001; 19: 3188–3193. Expression of late cell cycle genes and an increased proliferative capacity char- 31 Graf Einsiedel H, Taube T, Hartmann R, Wellmann S, Seifert G, Henze G et al. acterize very early relapse of childhood acute lymphoblastic leukemia. Clin Cancer Deletion analysis of p16(INKa) and p15(INKb) in relapsed childhood acute lym- Res 2006; 12: 4553–4561. phoblastic leukemia. Blood 2002; 99: 4629–4631. 52 Bhojwani D, Kang H, Moskowitz NP, Min DJ, Lee H, Potter JW et al. Biologic 32 Hof J, Krentz S, van Schewick C, Korner G, Shalapour S, Rhein P et al. Mutations pathways associated with relapse in childhood acute lymphoblastic leukemia: a and deletions of the TP53 gene predict nonresponse to treatment and poor Children’s Oncology Group study. Blood 2006; 108: 711–717. outcome in first relapse of childhood acute lymphoblastic leukemia. J Clin Oncol 53 Hertzberg L, Vendramini E, Ganmore I, Cazzaniga G, Schmitz M, Chalker J et al. 2011; 29: 3185–3193. Down syndrome acute lymphoblastic leukemia, a highly heterogeneous disease 33 Schwab CJ, Jones LR, Morrison H, Ryan SL, Yigittop H, Schouten JP et al. in which aberrant expression of CRLF2 is associated with mutated JAK2: a report Evaluation of multiplex ligation-dependent probe amplification as a method from the International BFM Study Group. Blood 2010; 115: 1006–1017.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia (2013) 295 – 304 & 2013 Macmillan Publishers Limited