Search

Relapsed T-cell acute lymphoblastic leukemia (T-ALL) represents one of the major challenges of pediatric oncology as it is often resistant to treatment and fatal. In the search for genes that define critical steps of relapse and can serve as prognostic markers we compared 67 relapses (REL) and 147 samples collected at initial diagnosis (INI) by targeted sequencing of 313 leukemia-related genes. In addition to the analysis of single nucleotide variants (SNV) and small insertions and deletions (InDels), we made use of the available coverage data for profiling of copy number alterations (CNA). Of the 147 INI patients, 31 were treated according to the ALL-BFM 2000 and 116 according to the AIEOP-BFM ALL 2009 protocol. All REL patients were recruited from the ALL-REZ BFM 2002 trial. We analyzed bone marrow DNA by targeted capture of 313 genes (5964 exons) using the Haloplex Target Enrichment Kit (Agilent). We used Varscan to detect both SNV and InDels. In the absence of available remission samples, we subtracted known SNPs (dbSNP, 1000 gp) and variants present in at least one of 20 non-leukemic samples that we sequenced in parallel to the patients' samples. Only mutations with an AF >10% were considered. Copy number analysis based on read-depth data was validated by MLPA analysis of 14 genes showing a sensitivity of 99% and by low coverage WGS. In total, we have SNV data on all 147 INI and 67 REL patients and CNA data on 144 and 58 patients, respectively. Altogether, we identified relapse specific genetic events in 32 of 67 RELs. Our results confirm that NT5C2 mutations are highly enriched in relapse (REL: 17/67 vs. INI: 1/147; p=0.0001). Although activation of NT5C2 was associated with the occurrence of early first relapse (p=0.02), it did not correlate with induction failure and therefore has no prognostic impact. Similarly, amplifications of chromosome 17 q11.2-24.3, a region that contains genes of the STAT and ABCA families were significantly more frequent in REL than in INI (REL 7/58 vs. INI 3/144, p=0.0068), but also had no prognostic implications. By contrast, TP53 mutations were highly predictive of a second event: all 8 patients who carried a total of 9 TP53 mutations and deletions died within 9 months after first relapse (p=0.002), whereas 17 of the other 58 (29%) survived. Inactivation of TP53 was significantly correlated with higher mutation rates in other genes compared to those with wild-type TP53 (ttest= p In conclusion, targeted sequencing of relapsed T-ALL identified a molecular signature predicting an exquisitely poor outcome in 50% of relapses, who failed salvage treatment. This group of patients does not benefit from current treatment strategies thus identifying a subgroup with a dire clinical need for experimental therapy. Notably, approx. 25% REL patients who failed induction carried RAS and IL7R mutations. These patients may be considered for personalized treatment with either MEK- or JAK/STAT-inhibitors. Another 4 patients carry TP53 missense mutations and may benefit from treatment with p53 refolding compounds. Disclosures No relevant conflicts of interest to declare.

Introduction: Despite the recent identification of the Ph-like subgroup of B-cell precursor Acute Lymphoblastic Leukemia (BCP-ALL), a large number of BCP-ALL patients lack cytogenetic and molecular defined lesions. To get a higher resolution and a broader molecular view of relapsed BCP-ALL, we designed a multi-omics study to reveal age-overriding relapse-driving alterations that may unravel novel molecular targets. Methods: We studied 150 paired samples (initial diagnosis: ID; relapse: REL; complete remission: CR) from 50 patients without known translocations. The cohort consisted of 24 adult and 26 pediatric patients with minimal residual disease < 0.05 % at CR. All patients were treated in population based German study trials (GMALL, BFM). We examined the mutational and copy number status via exome sequencing, obtained expression profiles and fusion-genes via RNA-sequencing and the methylation status via Illumina Methylation Array. Results: With a lenient approach detecting drivers and passengers, we identified significantly more mutations in REL compared to ID samples (adult median: 52 vs 38; pediatric median: 39 vs 27). In addition, we detected 4 hypermutators (more than 100 mutations per sample), 2 were pediatric and 2 were adult samples, 3 of which were REL samples. The most recurrently mutated genes were KRAS (n=15), NRAS (n=15), TP53 (n=13), CDC27 (n=13), KMT2D (n=11), IKZF1 (n=11), CREBBP (n=10) and FLT3 (n=6; Figure 1), with mutations present in both age cohorts. NT5C2, SYK and CHD1 were exclusively mutated in the pediatric cohort with at least 3 mutations. NT5C2 was also specific for early REL. Of all REL mutations, 225 mutations (14%, mean: 4 mutations/patient) were sub-clonal (under < 5% mutation frequency) at ID. Copy number alterations (CNA) varied greatly among pediatric and adult samples: 6% of pediatric and 18% of adult samples had aneuploidies and or copy neutral loss of heterozygosity of whole chromosomes. Chromosomal aberrations at ID persisted at relapse (100 %). Particular targets of CNA affected well-described genes like CDKN2A, CDKN2B, PAX5 on chr9p. Genes preferentially subjected to homozygous deletions were VPREB1 (n=6), SH2B3 (n=4), and ETV6 (n=2). All SH3B2 deletions were found in pediatric samples. On the epi-genomic level, the principal component analysis of the most variable CG-sites revealed a stable methylation profile during the course of the disease. However, we found a clear separation into a smaller pediatric-dominated cluster (n=24; 20 pediatric, 4 adult) and a larger mixed-age cluster (n=76; Fig. 1, Cluster A). Differentially methylated regions, affecting a total of 269 genes, characterized the separation of the smaller cluster, henceforth called Methylation Deregulated (MDR) cluster. The samples of the MDR cluster showed also a distinct gene expression profile by RNA-seq supporting a tight connection between the methylation status and its transcriptional program. A subset of 97 genes was differentially expressed including MAPK and PDGFR genes as most prominently deregulated. Additionally we defined a MDR expression classifier comprising 30 genes (Fig. 1). On the mutational level, the MDR samples had 20 % fewer mutations (mean: 25.3) compared to the remaining samples (mean: 31.3) and fewer CNVs for the most frequently affected genes. Characterising the non-MDR samples, a third of those were categorized as Ph-like ALL using the 15 gene classifier in an unsupervised clustering; this signature also coincided with the presence of well-known fusion-genes (Fig. 1, Cluster B). The remaining samples were defined by chromosomal instability (CI; Fig. 1, Cluster C). In the CI cluster, mutations in epigenetic regulators were twice as frequent when compared to the remaining samples. Conclusions: We describe three distinct clusters in relapsed BCP-ALL, which are characterized by a different genetic alterations: a novel MDR cluster by distinct methylation changes, the Ph-like cluster by gene fusions and the CI cluster by chromosomal instability. The cluster assignment was stable over the course of the disease. All clusters occurred in pediatric and adult patients, with the methylation-driven cluster predominantly in pediatrics. The MDR cluster showed significantly fewer mutations and CNVs compared to the other two clusters. The MDR samples showed activation of the MAPK signaling pathway pointing to actionable therapeutic targets. Figure 1 Figure 1. Disclosures Gökbuget: Pfizer: Honoraria, Research Funding; Amgen: Honoraria, Research Funding.

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=0.002 and P=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 (P

Background Resistance to therapy and subsequent relapse remain major challenges in the clinical management of relapsed childhood acute lymphoblastic leukemia. As the bone marrow environment plays an important role in survival and chemotherapy resistance of leukemia cells by activating different signaling pathways, such as the VLA-4 and PI3K/Akt pathways, we studied the prognostic and biological impact of VLA-4 expression in leukemia cells from children with relapsed B-cell precursor acute lymphoblastic leukemia and its influence on the sensitivity of the leukemia cells to drugs. Design and Methods VLA-4 expression was quantified by real-time polymerase chain reaction in leukemia cells from 56 patients with relapsed acute lymphoblastic leukemia enrolled in the ALL-REZ BFM 2002 trial of the Berlin-Frankfurt-Munster study group. Gene expression changes related to VLA-4 expression were investigated by microarray-based mRNA profiling. The effect of VLA-4 signaling on proliferation and drug resistance was studied in co-cultures of leukemia and stromal cells. Results High expression of VLA-4 at first relapse was associated with adverse prognostic factors, poor molecular response to therapy and significantly worse probabilities of event-free and overall survival. VLA-4 expression was an independent prognostic parameter. Comparing gene expression profiles of leukemia cells with high versus low VLA-4 expression, we identified 27 differentially expressed genes primarily involved in the PI3K/Akt, ephrin and Rho GTPase pathways. Blocking of VLA-4 signaling in combination with cytarabine treatment abolished the growth supportive effect of stromal cells. Conclusions Our results show that high VLA-4 expression is a marker of poor prognosis and a potential therapeutic target in children with relapsed acute lymphoblastic leukemia and confirm that cellular interactions and biological effects related to VLA-4 play a decisive role in the survival of leukemia cells and response to therapy.

A consistently increased mRNA expression of the adhesion receptor CD11b is a hallmark of the reported genomewide gene expression changes in precursor B-cell acute lymphoblastic leukemia (PBC-ALL) after 1 week of induction therapy. To investigate its clinical relevance, CD11b protein expression in leukemic blasts has been prospectively measured at diagnosis (159 patients) and during therapy (53 patients). The initially heterogeneous expression of CD11b inversely correlated with cytoreduction rates measured at clinically significant time points of induction therapy in the ALL–Berlin-Frankfurt-Münster 2000 protocol. CD11b positivity conferred a 5-fold increased risk of minimal residual disease (MRD) after induction therapy (day 33) and of high-risk group assignment after consolidation therapy (day 78). In the multivariate analysis CD11b expression was an independent prognostic factor compared with other clinically relevant parameters at diagnosis. During therapy, CD11b expression increased early in most ALL cases and remained consistently increased during induction/consolidation therapy. In more than 30% of MRD-positive cases, the CD11b expression on blast cells exceeded that of mature memory B cells and improved the discrimination of residual leukemic cells from regenerating bone marrow. Taken together, CD11b expression has considerable implications for prognosis, treatment response monitoring, and MRD detection in childhood PBC-ALL.

PURPOSE. To establish mouse models for RPGR-associated diseases by generating and characterizing an Rpgr mutation (inframe deletion of exon 4) in two different genetic backgrounds (BL/6 and BALB/c).METHODS. Gene targeting in embryonic stem (ES) cells was performed to introduce a in-frame deletion of exon 4 in the Rpgr gene (Rpgr(Delta Ex4)). Subsequently, the mutation was introduced in two different inbred mouse strains by successive breeding. Mutant and wild-type mice of both strains were characterized by electroretinography (ERG) and histology at five time points (1, 3, 6, 9, and 12 months). RPGR transcript amounts were assessed by quantitative RT-PCR. A variety of photoreceptor proteins, including RPGR-ORF15, RPGRIP, PDE6 delta/PrBP delta, rhodopsin, and cone opsin, were localized on retinal sections by immunohistochemistry.RESULTS. Mislocalization of rhodopsin and cone opsin was an early pathologic event in mutant mice of both lines. In contrast, RPGR-ORF15 as well as RPGRIP1 and PDE6 delta/PrBP delta showed similar localizations in mutant and wild-type animals. Functional and histologic studies revealed a mild rod-dominated phenotype in mutant male mice on the BL/6 background, whereas a cone-dominated phenotype was observed for the same mutation in the BALB/c background.CONCLUSIONS. Both Rpgr mutant mouse lines developed retinal disease with a striking effect of the genetic background. Conespecific modifiers might influence the retinal phenotype in the BALB/c strain. The two lines provide models to study RPGR function in rods and cones, respectively. (Invest Ophthalmol Vis Sci. 2010;51:1106-1115) DOI:10.1167/iovs.08-2742

Acute lymphoblastic leukemia (ALL) cells at relapse are frequently more resistant to treatment than primary clones and this may be caused by further genetic changes in the ALL cells at relapse. These acquired genomic abnormalities have not been fully characterized. To examine the additional genomic alterations of ALL at relapse, we performed single nucleotide polymorphism genomic microarry (SNP-chip) analysis on 14 ALL bone marrow samples at initial diagnosis, remission and relapse. Only two cases at initial diagnosis had a normal appearing genome by SNP-chip. All 14 cases had genomic alterations at relapse; and 10 of these had additional genomic abnormalities not present at diagnosis. Deletion of either the INK4A/ARF gene (2 cases) or the NF2 gene (2 cases) at 22q12.2 was an acquired genomic change at relapse. Loss of heterozygosity with normal copy number [uniparental disomy (UPD)] was detected in 3 cases as an additional genomic change at relapse. Interestingly, several genomic alterations, especially deletions, detected at initial diagnosis, disappeared at relapse, suggesting the ALL cells at relapse were minor clones at initial diagnosis and emerged at relapse. For several cases, trisomy at initial diagnosis changed to either UPD (2 cases) or normal appearing genome (2 cases). Further, we found disruption of PTPRD gene occurring at intron 23 as an additional genomic abnormality in one case. In summary, additional genomic changes are very common events in ALL at relapse; whether these abnormalities are associated with resistance to treatment remains to clarified in further studies.

The most common brain tumors in childhood and adolescence are low grade pilocytic astrocytomas (PA). Given that an increasing number of mitochondrial defects have been related to brain tumors and cancer in general, we asked whether PAs harbor mutations of mitochondrial DNA (mtDNA). Sequencing analysis of the complete mitochondrial genome of tumor tissue and corresponding blood samples from 19 patients with PA was performed. Of the 19 PA tissue samples, 16 (84%) showed somatic mtDNA mutations, and a total of 34 different somatic mtDNA mutations were identified. Of the 34 mtDNA mutations, 17 (50%) were found in genomic regions involved in pathways of oxidative phosphorylation. Three of the missense mutations in protein coding regions involved change of one amino acid: M60 V in ATP synthase subunit 6, L236I in cytochrome b, and L112 M in cytochrome c oxidase subunit 1. We were able to demonstrate that mtDNA mutations occur in PA and that they are frequently located in protein coding regions. The PA tumors were found to have the highest percentage of mitochondrial mutations of any of the neuroectodermal tumor entities studied to date. To reveal the prognostic importance of these mutations in the tumor biology of PA, larger series need to be studied prospectively.

Background: Relapse is the leading cause of mortality in children with acute lymphoblastic leukemia (ALL). Studies have shown that most ALL cases are polyclonal at diagnosis and that genetic changes in individual subclones influence sensitity to therapy and subsequent clonal evolution during therapy; but the molecular details remain to be worked out. Among different pathways enriched for mutations at relapse, purine metabolism is particularly interesting for two reasons: first, thiopurines are widely used in the ALL combination chemotherapy regimens, and are prodrugs that are converted by the purine salvage pathway to cytotoxic metabolites. Second, de novo nucleotide biosynthesis is often upregulated in cancer cells, and it is believed that sufficient nucleotide pools are required to maintain genomic stability, could bypass oncogene-induced senescence and promote tumor progression1. Therefore, we focus our current study on de novo purine biosynthesis in drug resistance and tumor relapse of childhood ALL. Methods and Results: Using whole-exome sequencing, we identified relapse-specific mutations in the phosphoribosyl pyrophosphate synthetase 1 gene (PRPS1), which encodes a rate-limiting purine biosynthesis enzyme, in 24/358 (6.7%) relapsed childhood B cell ALL (B-ALL) cases. Targeted sequencing identified mutations in additional genes in de novo purine biosynthesis pathway, providing further genetic evidence for its importance in relapsed ALL. All individuals with PRPS1 mutation relapsed early on-treatment (P Using various functional assays, we demonstrated that rather than causing a simple gain-of-function effect, the mutations in PRPS1 resulted in the disruption of the normal feedback inhibition of purine synthesis, in which the enzyme remained active despite an increased concentration of nucleoside analogs. PRPS1 mutants increased synthesis of the nucleoside inosine monophosphate, its metabolite hypoxanthine (HX) and de novo purine biosynthesis intermediates (e.g. AICAR, SAICAR) in Reh cells. Increased intracellular HX can competively inhibit the conversion of thiopurines into their active metabolites. Furthermore, inhibition of de novo purine biosynthesis in vitro, either by CRISPR-Cas9 genome editing of de novo purine synthesis pathway genes (GART, ATIC etc.) or treatment with a pathway inhibitor lometrexol (GART inhibitor) alleviated the metabolic disturbance and drug resistance induced by PRPS1 mutations. Using ultra-deep sequencing of unique serial remission samples before clinical relapse, we noticed that the PRPS1 mutant allele fraction increased drastically before clinical relapse, suggesting rapid clonal expansion occurs after the acquisition of a PRPS1 mutation. Interestingly, we also noticed that PPRS1 mutation coexist with RAS mutation in many relapse cases and at single cell resolution. Functional analysis revealed that tumor cells which harbored RAS and PRPS1 double mutations are more drug resistant than those with RAS or PRPS1 mutation alone. Previous studies have shown that oncogenic RAS mutation can also induce various stress responses including oncogene-induced senensence and DNA damage response (DDR), which all could impede tumor cell proliferation during relapse. In vitro, we found PRPS1 mutation can release the replication and metabolic stress caused by RAS mutation, in addition to their role in thiopurine resistance. The PRPS1 mutants not only increase the nucleotide pools but also elevate purine biosynthesis intermediate AICAR, which can activate AMPK and reduce the RAS mutant-induced DDR. We are currently working on in vitro and in vivo models (including patient derived xenograft models) to further test the double mutant's effects on tumor-reinitiation and clonal evolution during ALL relapse. Conclusions: We demonstrated that negative feedback-defective PRPS1 mutants can drive de novo purine biosynthesis, which can exert drug resistance and reduce genomic instability during tumor relapse. Our study highlights the importance of de novo purine biosynthesis in the pathogenesis of relapse, and suggests a diagnostic approach to predicting early relapse and a therapeutic strategy to circumventing resistance in ALL. 1 Li et al. Negative feedback-defective PRPS1 mutants drivee thiopurine resistance in relapsed childhood ALL. Nature Medicine,21(6): 563-571 (2015) Disclosures No relevant conflicts of interest to declare.

Precursor T-cell acute lymphoblastic leukemia (T-ALL) remains one of the major challenges of pediatric oncology, because relapses are frequently refractory to treatment and fatal. We aimed at identifying relapse specific genetic alterations by analyzing a cohort of 147 primary T-ALL patients and of 70 relapsed T-ALL patients with targeted gene panel sequencing. In addition to the analysis of single nucleotide variants (SNVs) and small insertions and deletions (InDels), we made use of the available coverage data to characterize aberrant copy number alterations (CNA). DNA from bone marrow of these 217 pediatric T-ALL patients was analyzed by gene panel sequencing after target capture with Agilent HaloPlex. In the target capture design, exons of 324 genes were included that had been found before by whole exome sequencing to carry somatic mutations in a pilot set of relapsed T-ALL or that have been reported to be mutated in T-ALL in the literature. We did not analyze corresponding remission samples and did not discriminate between germline and somatic alterations. Only mutations with an allele frequency (AF) > 10% were considered and absence of the mutation in the 1000 Genomes variant catalogue was required. Copy number analysis based on read-depth data identified deletions (DEL) and amplifications (AMP). Direct comparison of CNAs by multiplex ligation-dependent probe amplification (MPLA) and gene panel sequencing was possible for 13 overlapping regions covering 14 genes in 185 samples. Recognition rate by coverage analysis was 98% (256/260) for biallelic alterations and 81% (92/114) for monoallelic alterations found by MLPA. On average, gene panel sequencing identified 6.7 mutations in initial diagnosis samples (SNVs: 5.2; InDels: 1.5) and 7.9 mutations in relapse samples (SNVs: 5.9; InDels: 2). In the group of primary leukemia and relapse samples, the average AMP/DEL per patient was 8.2 (AMP: 3.2, DEL: 5.0) and 8.8 (AMP: 4.2, DEL: 4.6), respectively. 31 genes were found to be mutated and 46 deleted/amplified in 10 or more patients (see Table 1 and 2). | Gene | Total # of mutations | # pts with mutation in primary T-ALL (n=147) | # pts with mutation in relapsed T-ALL (n=70) | | ------ | -------------------- | --------------------------------------------------- | -------------------------------------------- | | NOTCH1 | 178 | 88 (60%) | 38 (54%) | | PHF6 | 47 | 24 (16%) | 16 (23%) | | FBXW7 | 42 | 20 (14%) | 17 (24%) | | OBSCN | 35 | 20 (14%) | 10 (14%) | | DNM2 | 28 | 17 (12%) | 10 (14%) | | PTEN | 34 | 20 (14%) | 4 (6%) | | XIRP2 | 24 | 19 (13%) | 5 (7%) | | CDH23 | 23 | 11 (7%) | 11 (16%) | | WT1 | 36 | 11 (7%) | 8 (11%) | | NT5C2 | 22 | 1 (1%) | 17 (24%) | Table 1. Most commonly mutated genes (SNVs and InDels) | Amplifications | Deletions | | -------------- | --------------------- | --------------------- | | Gene | primary T-ALL (n=147) | relapsed T-ALL (n=64) | Gene | primary T-ALL (n=147) | relapsed T-ALL (n=64) | | MYB | 9 (6%) | 9 (15%) | CDKN2A | 102 (70%) | 36 (59%) | | MYC | 11 (8%) | 6 (10%) | CDKN2B | 83 (57%) | 31 (51%) | | NRG1 | 11 (8%) | 3 (5%) | MLLT3 | 28 (19%) | 5 (8%) | | UNC5D | 11 (8%) | 3 (5%) | PHIP | 18 (12%) | 6 (10%) | | NCOA2 | 11 (8%) | 3 (5%) | ELOVL4 | 18 (12%) | 5 (8%) | | PTK2B | 11 (8%) | 3 (5%) | MAP3K7 | 17 (12%) | 5 (8%) | | FDFT1 | 11 (8%) | 3 (5%) | CASP8AP2 | 17 (12%) | 5 (8%) | | ABL1 | 8 (5%) | 3 (5%) | APC | 19 (13%) | 3 (5%) | | CNOT3 | 8 (5%) | 3 (5%) | LEF1 | 16 (11%) | 5 (8%) | | SMG8 | 3 (2%) | 7 (10%) | PAX5 | 15 (10%) | 5 (8%) | Table 2. Most common copy number alterations Potential novel mechanisms of oncogene activation are amplifications of PTK2B, a gene that has been found to be deregulated by fusion in Philadelphia-like BCP-ALL and that is potentially targetable by tyrosine kinase inhibitors, and of MYC, which has long been known to be a key player in T-ALL leukemogenesis and that is amplified in neuroblastoma and medulloblastoma. Enriched in relapse, we identified mutations in NT5C2 (p=1.4E-08), TP53 (p=0.0006) and CCDC88A (p=0.01), and amplifications of a region on chr 17q represented by the genes CLTC, ABCA5, C17orf80 and SRSF2. MLLT3 deletions were enriched in primary samples (p=0.04), consistent with the observation that MLLT3 deletions confer a lower risk of relapse in patients treated on BFM protocols. Conclusion Gene panel sequencing emerges as a suitable tool for a comprehensive genetic characterization of pediatric T-ALL. Within the group of selected genes contained in the panel, CNA were as frequent as point mutations. Only few genes were found to be specifically altered in relapse, indicating that progression to relapse may involve diverse, non-recurrent genetic alterations. Disclosures No relevant conflicts of interest to declare.

The ubiquitous cytosolic 5´nucleotidase II (NT5C2) dephosphorylates purine nucleotide monophosphates and has an important role in cellular purine metabolism. Increased levels of nucleotidase activity have been correlated with resistance to nucleoside analog drugs that are commonly used in the treatment of children with acute lymphoblastic leukemia (ALL). Recently, activating mutations of NT5C2 have been identified in relapsed childhood ALL. NT5C2 mutations were present in 20% and 10% of children with relapsed T-cell ALL and with relapsed B-cell precursor ALL, respectively. In vitro studies showed that NT5C2 mutations conferred an increased resistance to purine analog drugs to ALL cell lines. However, the predictive and prognostic value of NT5C2 mutations for response and outcome of patients has remained elusive and systematic studies are warranted. Therefore, we studied the presence of NT5C2 mutations in 259 children with first relapse of B-cell precursor ALL. The patient cohort was a representative subset the German relapse trial ALL-REZ BFM 2002. NT5C2 exons 9, 13, 15 and 16 were sequenced according to Sanger in leukemic samples taken at diagnosis of first relapse. The clinical significance of NT5C2 mutations was assessed by comparing clinical parameters and survival differences between patients with and without NT5C2 mutation, and by multivariate Cox regression modelling. We identified NT5C2 mutations in 5.8% (15/259) of patients with first relapse of B-cell precursor ALL. This confirms the overall lower NT5C2 mutation rate in children with relapsed B-cell precursor ALL compared to those with relapsed T-cell ALL. Sixty percent (9/15) of the patients with NT5C2 mutation showed a G to A transition in codon 367 in exon 13. This mutation substitutes arginine with glutamine (p.R367Q) and represents a mutation hot spot in relapsed T-cell ALL. Our study reveals that p.R367Q is the predominant site of mutation also in relapsed ALL of B-cell precursor lineage. Patients with NT5C2 mutation significantly more often presented with a very early relapse within 18 months after initial diagnosis (P We conclude that mutation of NT5C2 can serve as predictor for the occurrence of a second relapse independent of response to relapse treatment in children with relapsed B-cell precursor ALL. Disclosures No relevant conflicts of interest to declare.

For most children who relapse with acute lymphoblastic leukaemia (ALL), the prognosis is poor and there is a need for better prognostic biomarkers and novel therapies to improve outcome. Relapse samples from children with B lineage ALL entered into the ALL-REZ BFM 2002 clinical trial were screened for somatic mutations which activate the RAS pathway using DHPLC and Sanger sequencing. Mutations were found in 78 patients, giving an incidence of 37.9% and were made up of NRAS (n=30), KRAS (n=30), FLT3 (n=10) and PTPN11 (n=9); one patient had both a KRAS and FLT3 mutation. Clinically, mutated cases were associated with early relapse which was particularly significant for NRAS/KRAS mutated patients (p=0.001), with a 5 year probability of event free survival of 46.9% versus 60.1% for wild type (p=0.21). In addition, there was an over representation of NRAS/KRAS mutated patients with CNS involvement, 23.3% compared to 10.3% for the mutation negative group (p=0.014). Mutation screening of the matched diagnostic samples found many to be wild type (26 from 54) but using more sensitive allelic specific assays, low level mutated subpopulations were found in 57% of cases (8 from 14), suggesting these cells survived up front therapy and subsequently emerged at relapse. Assessment of p-ERK levels by western blotting in a large cohort of diagnostic ALL samples confirmed that RAS pathway mutated cases almost always had constitutive activation of the pathway (27 from 32; 84.3%), while relatively few wild type patients were p-ERK positive (9 from 48; 18.7%) (p8 after 30 drug doses compared to a mean fold increase of >5 for CV. Mice treated with selumetinib and sacrificed at the end of the study showed a significant decrease in spleen size compared to CV treated mice. There was minimal effect of selumetinib in wild type ALLs. Histological analysis of post-mortem brains from mice engrafted with an NRAS mutant primagraft found extensive meningeal leukaemic infiltration in CV-treated, but not selumetinib-treated mice. Pharmacodynamic assessment in spleens of mice engrafted with mutant primagrafts showed absence of p-ERK and increased levels of the apoptotic biomarkers, Bim and cleaved PARP. In summary, we show that RAS pathway mutations are a common genetic abnormality in relapsed ALL and are associated with high risk features and are often relapse-drivers. Targeted MEK inhibition with selumetinib shows excellent activity in RAS mutated ALL both in vitro and in vivo and may offer clinical benefit for a substantial proportion of children with relapsed ALL. Given our findings, clinical trials of selumetinib in RAS pathway positive relapsed patients are warranted. Disclosures: No relevant conflicts of interest to declare.

Acute lymphoblastic leukemia (ALL) is an aggressive hematological tumor resulting from the malignant transformation of lymphoid progenitors. Despite intensive chemotherapy, 20% of pediatric and over 50% of adult ALL patients fail to achieve a complete remission or relapse after intensified chemotherapy, making disease relapse and resistance to therapy the most significant challenge in the treatment of this disease. Using whole exome sequencing, here we identify mutations in NT5C2, a 5’-nucleotidase enzyme responsible for inactivation of nucleoside analog chemotherapy drugs, in 20/103 (19%) relapse T-ALLs and in 1/35 (3%) relapse B-precursor ALLs analyzed. NT5C2 mutant proteins show increased nucleotidase activity in vitro and conferred resistance to chemotherapy with 6-mercaptopurine and 6-thioguanine when expressed in ALL lymphoblasts. These results support a prominent role for activating mutations in NT5C2 and increased nucleoside analog metabolism in disease progression and chemotherapy resistance in ALL. Citation Format: Gannie Tzoneva, Arianne Perez-Garcia, Zachary Carpenter, Hossein Khiabanian, Valeria Tosello, Maddalena Allegretta, Elisabeth Paietta, Janis Racevkis, Jakob M. Rowe, Martin S. Tallman, Maddalena Paganin, Jana Hoff, Renate Kirschner-Schwabe, Teresa Palomero, Raul Rabadan, Adolfo Ferrando. Activating mutations in the NT5C2 nucleotidase gene drive chemotherapy resistance in relapsed ALL. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 4595. doi:10.1158/1538-7445.AM2013-4595

Abstract 752 Risk stratification in the childhood ALL-BFM clinical trial has been based on molecular/cytogenetic markers and the in vivo response to treatment (blast reduction in peripheral blood after prednisone prephase, bone marrow clearance after induction treatment and minimal residual disease (MRD) after induction and induction consolidation). These criteria classify about 50% of ALL cases into standard-risk (SR) and high-risk (HR) groups. The intermediate-risk (IR) ALL group, showing moderate levels of MRD, provides the majority of relapses, but still lacks specific prognostic markers which could distinguish between those patients subsequently relapse and those remaining relapse-free (IR+r and IR-r). We established a bank of NOD/SCID ALL xenografts representing IR+r and IR-r subgroups (12 vs 13 patients) and characterized the samples in a broad series of functional assays. There was no difference in the sensitivity to drugs (dexamethasone, daunorubicin, L-asparaginase and vincristine) between the IR+r and IR-r groups, however, the rate of spontaneous apoptosis in vitro was significantly higher in the IR+r group than in the IR-r group (69 +/− 7% vs 39 +/− 6%, p=0.02). Given that in the presence of stromal cells the cell death level has been similar in both IR groups (9 +/− 5% vs 7 +/− 7%, p=0.8), these observations suggest that the IR+r ALL is more dependent on survival signals from microenvironment than IR-r ALL. In the NOD/SCID engraftment model, leukemic cells from IR+r patients engrafted more rapidly than cells from IR-r patients (93 +/− 12 days vs 170 +/− 24 days, p=0.006), pointing to a better in vivo cooperation between leukemia cells and microenvironment in the IR+r cases. In order to identify the underlying molecular pathways, gene expression changes induced in the presence and absence of stroma have been investigated at genome-wide level (Human Gene 1.0 ST Affymetrix array). Of the apoptosis-regulating genes, a caspase inhibitior from the IAP family, BIRC3, was found to be up-regulated significantly higher in the IR-r than in the IR+r group (3.7 +/− 1.1 vs 1.7 +/− 0.1-fold up-regulation, p=0.016). The adhesion molecule VCAM1 has been generally up-regulated by the co-incubation with stroma, with a tendency to a higher up-regulation in the IR-r (4.4 +/− 0.7-fold) than in the IR+r (2.7 +/− 1.0-fold) group. We further speculated that interaction with microenvironment is a mutual process which also implicates gene expression changes of the stroma. To this end, mouse stromal cells (MS5 cell line) were incubated with a series of ALL samples (n=3), purified by MACS sorting and investigated using Mouse Gene 1.0 ST Affymetrix arrays. The list of genes whose expression increased by at least 2-fold in all 3 cases (n=10), included the macrophage inflammatory peptide-1alpha (MIP1A/CCL3), a chemokine known to regulate chemotaxis of lymphocytes. Our data suggest that the cooperation of ALL with microenvironment may provide a promising approach to molecularly discriminate ALL cases with relapse and to identify potential relapse-associated targets within the IR-ALL group. Disclosures: No relevant conflicts of interest to declare.

Abstract 1718 Childhood acute lymphoblastic leukemia (ALL) is characterized by recurrent structural and/or numeric genetic aberrations and has been consistently reported to occur with a 20% higher incidence in males relative to females. We performed 100k Affymetrix SNP-array analysis in a selected set of 20 ALL samples. In five out of these 20 patients, a previously described deletion of the 5' region of C20orf94 – a gene coding for an uncharacterized DNA repair-associated protein – was observed. The results were validated by high-resolution custom-made CGH array analysis. As the breakpoints within C20orf94 were defined to recurrent positions, we next applied a PCR assay to screen diagnostic leukemic specimens of a representative cohort of 513 patients enroled into the German multicenter trial ALL-BFM 2000 on treatment of childhood ALL. Here, C20orf94 deletions were detected in 164 patients (32.0%). None of 134 available remission samples exhibited the deletion as well as none of 145 healthy blood donors. Sequencing analysis in 40 patients revealed specific breakpoint junctions with typical characteristics of illegitimate V(D)J recombination in all samples. When analyzing the association of C20orf94 deletion with clinical characteristics in the entire screening cohort, the most significant interrelationships were observed for male gender (75.6% of positives; P < 0.001) and TEL/AML1-rearranged ALL (43.3% of positives; P < 0.001). A negative association was observed for a DNA index of ≥1.16 (3.0% of positives; P < 0.001). In contrast, C20orf94 deletion did not show any effect on treatment outcome. Additional screening of an independent cohort of 232 TEL/AML1-rearranged ALLs identified 145 positive samples and not only confirmed the high incidence of C20orf94 deletion in this subgroup, but also allowed validation of its specific association with male gender (55.9% of positives; P = 0.049). In 21 of 22 (17 TEL/AML1-positive and 5 other pre B cell ALLs) paired initial and relapse leukemic samples C20orf94 deletions were maintained at relapse. Breakpoint sequencing of paired initial and relapse leukemic samples revealed a stable monoclonal breakpoint sequence in 9 patients indicating stability of C20orf94 deletions during disease progression. The remaining samples either developed heterogeneity of C20orf94 breakpoints at relapse after monoclonality at initial diagnosis or initially already showed a polyclonal pattern. Backtracking of C20orf94 deletions to birth in seven TEL/AML1-positive patients using material derived from Guthrie cards did not yield any positive results. In conclusion, we describe the frequent and uniform deletion of the 5' part of C20orf94 in childhood ALL - particularly in males and the TEL/AML1-rearranged subtype. These findings suggest a potential role for C20orf94 deletion in the pathogenesis of childhood ALL and point to differences in illegitimate V(D)J recombination as one potential explanation for the observed and currently only poorly understood gender bias in childhood ALL. Disclosures: No relevant conflicts of interest to declare.

Abstract 2612 Poster Board II-588 Minimal residual disease (MRD)-measurements have been integrated into treatment protocols for childhood acute lymphoblastic leukemia (ALL) using T-cell-receptor(TCR)/immunoglobulin(IG) gene rearrangements. Additionally TCR/IG gene rearrangements can be applied to describe clonal relations and identify cell populations which undergo clonal evolution and selection during and after treatment. Patterns of TCR/IG gene rearrangements might differ between initial and relapse diagnosis and are interesting and very informative in order to describe both the quantitative extent at diagnosis and the kinetics during treatment of main- and subpopulations. In order to minimize the chance of a false negative MRD-result knowledge about the stability of TCR/IG-markers is important. Therefore, stability of TCR/IG-markers between initial and relapse diagnsosis was assessed in previous studies. These studies recommended to use two markers besides considering gene-locus and clonality profile. However, it has never been validated prospectively whether the TCR/IG-markers actually chosen for MRD-quantification and risk-group stratification are sufficient for clinically and biologically reliable MRD-monitoring during frontline treatment. We have aimed to compare marker profiles between initial and relapse diagnosis by a systematic approach in order to evaluate markers acutally chosen for initial MRD-quantification. Further, MRD-response to induction of initial and relapse treatment have been compared. One main objective of our study was to gain new insights in clonal heterogeneity, resistance and clonal evolution. We have performed a prospective study including patients treated according to the frontline-trial ALL-BFM 2000 and who suffered a relapse. The interim analysis included finally 45 patients showing a median time to relapse of 1.5 years (range 0.6 – 3.8). Therefore, mainly (73%) very early (during first 18 months after initial diagnosis) and early (between 18 and 30 months) relapses were included in the study. In 33% of patients (15/45) all markers identified at initial diagnosis remained stable at relapse diagnosis and no additional marker was gained at relapse. The remaining 67% of patients showed at least one clonal change. In 38% (17/45) of patients at least one gain of a TCR/IG gene rearrangement was seen; in 53% (24/45) of patients loss at relapse and in 16% (7/45) V-V-replacement or Vd2-Ja ongoing rearrangement product was seen at relapse. Considering markers actually chosen for initial MRD-quantification, in 62% of patients both markers remained stable, in 27% at least one, in 11% no marker. Early on-therapy relapses showed the same proportion of patients with a marker-gain, than later on-/off-therapy relapses. A backtracking-analysis of markers gained revealed the presence of nearly all markers (17/19: 90%) at initial diagnosis at different quantitative levels independently of time point of relapse. Paired initial and relapse molecular response to induction varied remarkably. Only 5 of 24 patients included in this analysis showed similar response (1 poor, 4 intermediate, 3 good responders) which demonstrates the unexpected high heterogeneity. It is interesting to note that complete identical markers for MRD-monitoring during initial and relapse treatment were only chosen in half of the patients (12/24). These observations challenges the classical and still existing opinion that early relapses derive from resistant leukemia because resistant MRD was not eliminated by polychemotherapy. Based on our data and data from other studies comparing genetic markers in paired samples between initial and relapse diagnosis, the extrapolation of an algorithm for MRD-marker choice is not possible and would be too artificial at the moment. From the biological point of view, the practical restriction of using only one or two markers for MRD quantification at several time points during early and later phases of the treatment should not be recommended. Despite the higher proportion of very early/early relapses in this interim analysis we observed a high heterogeneity of marker-profiles between initial and relapsed diagnosis and molecular response to treatment at both disease stages. Disclosures: No relevant conflicts of interest to declare.

The ETV6/RUNX1 fusion, resulting from the cryptic translocation t(12;21)(p13;q22), is the most common genetic rearrangement in B-cell precursor (BCP) acute lymphoblastic leukemia (ALL) with a prevalence of approximately 20–25% at first presentation. Although generally associated with favorable risk features and advantageous long-term survival rates, similar frequencies of ETV6/RUNX1 positivity at relapse question its prognostic significance. At relapse, a second continuous complete remission can again be achieved in the majority of patients with ETV6/RUNX1-positive ALL, however, a substantial proportion (approximately 25%) exhibit a poor response to treatment and eventually experience a subsequent relapse. Currently, critical secondary genetic events are assumed to be pivotal for t(12;21) positive ALL leukemogenesis. At the chromosomal level, additional numeric aberrations may also contribute to differences both in treatment response and course of disease, and may provide additional prognostic information. Aiming at identifying the incidence and the clinical relevance of additional genetic aberrations at the time point of first ALL relapse diagnosis, we performed whole genome high resolution tiling-path bacterial artificial chromosome (BAC) array CGH of leukemic cell DNA from 53 patients with ETV6/RUNX1-positive first relapse enrolled in the relapse trials ALL-REZ BFM of the Berlin-Frankfurt-Münster study group. Additional genetic aberrations were detected in all of the ETV6/RUNX1-positive leukemic bone marrow samples with a mean number of seven aberrations per ALL. Chromosomal losses were observed approximately 2.5 times more frequently than gains. A high proportion of the identified copy number alterations occurred in recurrently affected chromosomal regions. Copy number alterations most commonly detected by high-resolution array CGH were losses of 12p13 (49%, 26/53), 6q21 (34%, 18/53), 15q15.1 (23%, 12/53), 9p21 (21%, 11/53), 3p21 (21%, 11/53), 5q (19%, 10/53), 19q13 (17%, 9/53), 10q (13%, 7/53), 3p14.2 (11%, 6/53) and gains of 21q22 (32%, 17/53) and of 12p (21%, 11/53). Loss of the whole X-chromosome, detected in 17% (9/53) of the analyzed samples, was observed exclusively in females. In contrast, gain of Xq, identified in 13% (7/53) of the samples, was detectable solely in males. Evaluation of the clinical parameters in relation to recurrent copy number alterations revealed that loss of the whole X-chromosome was associated with a molecular good response to treatment (P=0.03), as assessed by sensitive minimal residual disease (MRD) monitoring. In contrast, loss of 5q31.3, detected in 11% (6/53) of the analyzed samples, was associated with a dismal molecular response to therapy (P=0.019) and with a tendency to a worse outcome (P=0.054). Thus, this study demonstrates that relapsed ETV6/RUNX1-positive ALL is characterized by (a.) multiple additional genomic alterations, in particular by a high incidence of chromosomal losses, which occur predominantly in recurrently affected chromosomal regions, some of which (b.) are of significant prognostic relevance. These findings support the notion that several additional chromosomal changes are not only required for the process of malignant transformation in ETV6/RUNX1-positive ALL (leukemogenesis) but also influence therapeutic success. Perspectively, high resolution genomic profiling will provide valuable information enabling a more refined, individualized therapy and an improved risk stratification in relapsed childhood ALL.

Despite considerable improvements regarding treatment outcome about 20% of children with acute lymphoblastic leukemia (ALL) still suffer from recurrent disease. Especially patients with an early relapse occurring before 6 months of treatment cessation have a poor prognosis and, therefore, urgently need therapy optimization. As an initial step, detection of genetic aberrations that are associated with resistant disease and/or early relapse might be of great value as they could serve as prognostic markers to predict disease recurrence and, in addition, may increase our understanding of mechanisms underlying treatment resistance and early relapse. This knowledge may also help to identify new therapeutic strategies including targeted approaches. Surprisingly, almost no information is available on the genetic evolution from initial diagnosis to early relapse. As a first aim in the process of uncovering the pathomechanism of early relapse we wanted to find out if early relapse of childhood ALL reflects primarily resistant disease with little or no additional aberrations at relapse or represents evolution and selection of a new and more resistant clone possibly triggered through chemotherapy. Recently, high throughput technologies (single nucleotide polymorphism [SNP]–Arrays) have been developed to screen the whole genome with high resolution for submicroscopic, genetic alterations by simultaneously analyzing the SNP genotype and determining the copy number state. This prompted us to perform SNP-array analysis encompassing more than 100.000 SNPs in 20 childhood ALL samples collected at initial diagnosis and to compare them to their counterparts obtained at relapse. The analyzed patients belonged to the intermediate or high risk treatment group; 16 exhibited B-precursor and 4 T-precursor ALL. Exclusion criteria were translocations t(12;21), t(9;22) and (4;11) as well as pre-existing conditions (e.g. Down syndrome). At initial diagnosis, the leukemic sample of all patients already displayed a complex karyotype with multiple genetic lesions (average numbers per patient: 5.15 deletions, 1.35 amplifications, and 0.65 copy number neutral LOH (CNN-LOH)). Most alterations remained stable from initial diagnosis to relapse (99 of 103 (96%) initially observed deletions as well as 24 of 27 (89%) amplifications). Four deletions occurring in 3 different patients at initial diagnosis disappeared at relapse. Nevertheless, these patients had additional stable copy number alterations. Newly occurring aberrations at relapse were less frequent (average numbers per patient: 1.8 new deletions, 0.75 new amplifications, and 0.15 new CNN-LOH). In 3 patients genetic alterations at diagnosis and relapse were entirely identical. The most commonly affected chromosome was 9p (75% of samples). Fourteen patients had either heterozygous or homozygous deletions of CDKN2A at initial diagnosis and relapse with 2 patients progressing from heterozygous to homozygous deletions. PAX5 deletions (mainly heterozygous) were detectable in 9 patients at initial diagnosis and 10 patients at relapse. Other sides of recurrent deletions – mainly detected at relapse – included multiple areas of chromosome 7p (p21.3, p15.3, p14.3, p14.3-14.1, p14.1-p13) as well as chromosome 1q (deletion: q42.12-q44, amplification: q25.2-q31.1), chromosome 17p (deletion: p13.3-p11.2) and 17q (amplification: q21.2-q25.1). Potential pathogenetic implications of specific recurrent genetic lesions will be discussed. In conclusion, our results demonstrate a high degree of genomic stability from initial diagnosis to early relapse of childhood ALL. These results suggest that – at least in some patients – early relapse might already be predetermined by a resistant leukemic clone at the time of initial diagnosis.

Activating mutations within the NOTCH1 gene occur in more than 50% of childhood precursor T-cell lymphoblastic leukemia (T-ALL, Weng et al. Science 2004). Our previous work has shown that in the context of the ALL-BFM 2000 treatment strategy, activating NOTCH1 mutations significantly correlate with a good early treatment response and specified a subgroup of patients with an exceptionally favorable long term outcome. By contrast, most of the relapses occurred in those patients without NOTCH1 mutations, which thus represents a clinically important and challenging subgroup (Breit et al., Blood 2006). We now aimed to further differentiate this subgroup without NOTCH1 mutations by analyzing the mRNA expression profile of primary pediatric T-ALL bone marrow samples. We first validated the biological significance of these analyses by comparing samples with and without activating NOTCH1 mutations and confirmed that the presence of activating NOTCH1 mutations correlates with differential expression of multiple downstream signaling pathways that are known to be activated by NOTCH1. Importantly, the comparison of patients without NOTCH1 mutations and good or poor prednisolone response revealed a specific gene signature that differentiates wild-type NOTCH1 T-ALL with a favourable from those with an unfavourable early treatment response. These discriminatory signatures significantly enrich for specific gene ontology categories (e.g. cell death, cellular growth and proliferation, and molecular transport), suggesting functional relevance of these differentially expressed genes. The specific comparison of different T-ALL subgroups thus reveals prognostic gene expression signatures particularly in the clinically difficult patients without NOTCH1 mutations.

Gene expression profiles (GEPs) are becoming increasingly more important for diagnostic procedures, allowing clinical predictions including treatment response and outcome for various types of cancer. However, established gene signatures not always appear helpful in understanding underlying disease mechanisms. Therefore, we explored an alternative approach for analyzing GEPs of a group of t(4;11) positive infant acute lymphoblastic leukemia patients (n=15), a highly aggressive type of MLL rearranged leukemia. We developed a method that uses a relational database program in combination with a normalization approach and specific discriminators. For normalization, every GEP of a given t(4;11) positive ALL sample was compared with 3 GEPs of normal bone marrow aspirates derived from healthy donors. Using the GeneChip Analysis Suite 5.0 program for single comparison analysis, the resulting gene lists were then compared with each other and only consistently identified genes (present on all three gene lists) were used for further analysis. This noise reduction decreased the amount of potentially deregulated target genes to about 30–40%. The remaining gene lists represented highly significant target genes that were then incorporated into a relational database program using specific discriminators. These discriminators were: upregulation of HOXA9 and HOXA10, presence/absence of the AF4-MLL fusion transcripts in addition to the MLL-AF4 fusion, and the localization of the genomic breakpoint within the MLL gene. This pilot study led to promising results, surprisingly classifying individual t(4;11) positive ALL patients into two distinct subgroups. Both subgroups share about 80 target genes, but also display particular sets of subgroup specific target genes. Importantly, these identified target genes can directly be linked to biological properties of t(4;11) positive leukemia cells, and therefore, allow important novel insights into this aggressive type of leukemia in infants.

In the multicentric ALL-BFM (Berlin-Frankfurt-Munster) study, all patients are uniformly treated during the first week of induction therapy which uses glucocorticoids (GC) as the principal therapeutic agent. The GC response assessed at day 8 of therapy provides one of the basic parameters for further risk stratification. In spite of the clinical significance, molecular mechanisms of GC action in vivo are largely unknown. Our recent genome-wide analysis of gene expression in blasts persisting during induction therapy identified a common set of genes differentially expressed in blasts at day 8 (d8) and at diagnosis (d0) (n=457, false discovery rate

Acute lymphoblastic leukemia (ALL) is the predominant malignancy in childhood (30%). The majority of ALL (75%) exhibit characteristic chromosomal aberrations with prognostic significance that are used to classify the heterogenic disease into subgroups. Among these, the translocation t(12;21) resulting in the fusion of the TEL and AML1 genes is the most frequent chromosomal rearrangement in childhood ALL (20 – 25%). Both genes code for transcription factors essential for normal hematopoiesis. Whereas AML1 functions mostly as an activator of expression, TEL/AML1 acts as a repressor in a dominant manner. To date, detailed knowledge on the effect of TEL/AML1 in the process of malignant transformation is widely lacking. To address the consequence of TEL/AML1 expression, we established stable TEL/AML1 expressing cell lines of B lymphoblastoid and non-B cell origin using an episomal vector system and performed microarray-based gene expression profiling. To this end, we obtained gene expression profiles of a TEL/AML1 expressing non-B cell clone and the corresponding empty vector control. We first set out to gain a measure for the comparability of gene expression profiles in leukemic cells of TEL/AML1 positive ALL patients and the TEL/AML1 expressing non-B cell line. We therefore used published gene expression data of initial and relapsed ALL to calculate diagnostic gene expression signatures that faithfully predict genetic and immunological ALL subtypes in patient samples. We then used these signatures to classify a test set of patient samples and the TEL/AML1 expressing non-B cell into these subtypes. All patient samples were predicted correctly, however, the TEL/AML1 non-B cell line was not classified as TEL/AML1 positive, indicating that the diagnostic signature of TEL/AML1 positive leukemic cells is not induced by TEL/AML1 in the non-B cell line. Nevertheless, we identified several genes differentially expressed in the non-B cell line as compared to the empty vector control. The majority of genes are down-regulated, in agreement with the repressor function of the fusion protein. Among these genes are several transcription factors and genes playing a role in hematopoietic development and immune response. Furthermore, we found several genes with a role in neuronal development and disease. Interestingly, TEL knock out mice die during embryonal development due to apoptosis of mesenchymal and neuronal cells, supporting a function of TEL in the nervous system. Thus, a set of genes is regulated by TEL/AML1 in a non-B cell line, compatible with the proposed functions of TEL and AML1. However, the diagnostic signature of leukemic blasts of TEL/AML1 positive ALL patients was not induced in these cells supporting the view that besides common changes in gene expression cell type specific effects of TEL/AML1 exist and contribute to leukemogenesis.

In childhood acute lymphoblastic leukemia (ALL), persistence of leukemic blasts during therapy is of crucial prognostic significance. To approach the mechanisms of therapy resistance, we addressed genome-wide gene expression in blasts persisting after one week of induction therapy (day 8 blasts) and their molecular signatures as compared with blast cells at initial diagnosis (day 0 blasts). In order to approach this issue experimentally, a procedure has been established including flow sorting of leukemic blasts by their leukemia-associated immunophenotype and preparation of cRNA, starting from a small number of cells. Blast cells from 12 patients with precursor B-cell ALL were investigated using Affymetrix HG U133A microarrays, and genes commonly up- or down-regulated in blast cells under therapy were identified in matched pairs of day 8 and day 0 samples. In spite of the heterogeneous clinical features of the patients (mean rate of cytoreduction after 7 days of initial therapy = 82%, range between 33% and 99%), we were able to determine a set of 310 genes whose expression was commonly changed between day 8 and day 0 with an estimated false discovery rate of 0.05. The identified set of genes indicated inhibited cell cycling, reduced metabolism, and expression changes of multiple factors related to B-cell differentiation. These changes collectively suggested that gene expression in day 8 blasts is shifted towards resting mature B cells. To test this hypothesis, we isolated normal B cells from peripheral blood samples of leukemic patients and compared their gene expression to that of leukemic blasts using Principal Component Analysis (PCA). PCA revealed that day 8 samples are positioned between day 0 samples and normal B-cell samples, and statistical significance of this observation could be established using the Jonckheere-Terpstra test. Changes of B-cell differentiation markers on protein level supported this finding. In addition, we analyzed all genes with regard to the correlation of their expression changes with the rates of cytoreduction in peripheral blood. We observed differential impairment of the key components of the translational machinery including ribosome, eukaryotic 43S preinitiation complex and eukaryotic 48S initiation complex. Overall, expression levels of these factors decreased in therapy-sensitive patients but did not change in therapy-resistant patients. Taken together, investigation of leukemia cells persisting during therapy identifies common and individual expression changes which may potentially affect sensitivity towards anti-leukemic agents and offers new insights into the mechanisms of therapy resistance in ALL.