Downregulated Expression of Genes Mapping on Chromosome 9 in Chronic Myeloid Leukemia Cases Bearing Genomic Deletions on Der(9)

Letters to the Editor 813 regimen, with survival of 16 versus 27% respectively, hazard 1Division of Haematology/Oncology, ratio 1.33 (95% CI 1.01–1.77; P ¼ 0.05).8 When interpreted Department of Paediatrics, The Hospital for Sick Children, Toronto, Ontario, Canada; in the context of our meta-analysis results, the trial results 2 questions the role of fludarabine, cytarabine and priming Department of Paediatrics, Program in Child Health Evaluative Sciences, The Hospital for Sick Children, G-CSF in patients with high-risk AML. Toronto, Ontario, Canada; There are several limitations to this meta-analysis. First, the 3Department of Medicine, University Health Network, 95% CI for all survival outcomes other than remission rates does Toronto, Ontario, Canada; not exclude the possibility of clinically meaningful effects. 4Department of Hematology-Oncology, Children’s Mercy However, given the point estimate of a 1% absolute risk Hospitals and Clinics, Kansas City, MO, USA and difference in overall survival, it is unclear whether CSF priming 5Department of Oncology, Children’s Hospital of would warrant prioritization for future study in AML trials, as a Philadelphia, Philadelphia, PA, USA definitive trial would likely require several thousand patients. E-mail: [email protected] Second, the number of studies did not permit meta-regression or further stratified analyses. In addition, there was substantial References heterogeneity in the types of chemotherapy administered concurrent with CSF priming. However, on balance, we did 1 Disis ML. Clinical use of subcutaneous G-CSF or GM-CSF in not find clinically important differences in the effect of CSF malignancy. Oncology (Williston Park) 2005; 19 (4 Suppl 2): 5–9. priming for the variables we examined. 2 Sung L, Nathan PC, Alibhai SM, Tomlinson GA, Beyene J. Meta- In conclusion, the results of this meta-analysis analysis: effect of prophylactic hematopoietic colony-stimulating demonstrate that CSF priming does not improve outcome factors on mortality and outcomes of infection. Ann Intern Med 2007; 147: 400–411. in AML and thus should not be used in routine clinical 3 Becker PS. Growth factor priming in therapy of acute myelogenous care. Although this meta-analysis did not definitively address leukemia. Curr Hematol Rep 2004; 3: 413–418. the question of CSF priming in standard risk patients, the overall 4 Lowenberg B, van Putten W, Theobald M, Gmur J, Verdonck L, results of the analysis do not lend strong support to study this Sonneveld P et al. Effect of priming with granulocyte colony- question further. stimulating factor on the outcome of chemotherapy for acute myeloid leukemia. N Engl J Med 2003; 349: 743–752. 5 Landis JR, Koch GG. The measurement of observer agreement for Acknowledgements categorical data. Biometrics 1977; 33: 159–174. 6 DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–188. LS is supported by a career development award with the Canadian 7 Thomas X, Raffoux E, Botton S, Pautas C, Arnaud P, de Revel T Child Health Clinician Scientist Training Program, a strategic et al. Effect of priming with granulocyte-macrophage colony- program with the Canadian Institutes of Health Research. SMHA is stimulating factor in younger adults with newly diagnosed acute a Research Scientist of the National Cancer Institute of Canada. myeloid leukemia: a trial by the Acute Leukemia French Association JB would like to acknowledge funding from the Canadian Institute (ALFA) Group. Leukemia 2007; 21: 453–461. 8 Milligan DW, Wheatley K, Littlewood T, Craig JI, Burnett AK. of Health Research. Fludarabine and cytosine are less effective than standard ADE chemotherapy in high-risk acute myeloid leukemia, and addition of 1,2 3 2 4 2 L Sung , SMH Alibhai , J Beyene , A Gamis , R Almeida , G-CSF and ATRA are not beneficial: results of the MRC AML-HR S Smith5 and R Aplenc5 randomized trial. Blood 2006; 107: 4614–4622.

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

Downregulated expression of genes mapping on chromosome 9 in chronic myeloid leukemia cases bearing genomic deletions on der(9)

Leukemia (2009) 23, 813–816; doi:10.1038/leu.2008.311; microdeletions could be associated with the loss of tumor published online 6 November 2008 suppressor genes or genes implicated in cell-cycle regulation, resulting in a proliferative advantage of the leukemic clone.4 This phenomenon may act by a gene dosage effect known as Genomic deletions flanking the breakpoint on der(9)t(9;22) haploinsufficiency: mutation or loss of a single allele may be occur in 10–18% of patients with chronic myeloid leukemia sufficient to elicit a cellular phenotype that leads to tumorigen- (CML). These microdeletions have a variable extension and esis without the inactivation of the second allele. involve sequences on chromosomes 9 and/or 22, located The aim of our study was therefore to assess whether the proximally to the ABL and distally to the BCR gene.1 The sequences loss on der(9) in CML patients was associated with a deletions on der(9) are associated with a poor prognosis of gene product decrease. To date, an expression study of genes interferon-a (IFN-a) therapy, whereas controversial data about mapping within the deleted regions on der(9) has never been their influence on the response to imatinib are available.2,3 The performed. We have reported earlier a detailed molecular molecular mechanisms responsible for this unfavorable prog- cytogenetic characterization of 60 (18%) out of 334 CML nosis are still unclear. The most probable hypothesis is that patients, using panels of bacterial artificial chromosome clones

Leukemia Letters to the Editor 814 mapping next to the t(9;22) breakpoint.5 We showed that the controls the cellular lifetime together with localization of many extent of the genomic deletion on der(9) ranged from a few critical proteins. The loss of PPP2R4, ASB6, SH3GLB2 and hundred kilobases to many megabases, causing the loss of PKN3 was detected in 60, 57, 43 and 33% of cases with several genes with a known function. deletions, respectively. Noteworthily, the PPP2R4 gene is one of To provide more accurate information about the significance the regulatory subunits of the phosphatase 2A (PP2A) protein, a of the deletion status of these genes with respect to their serine/threonine phosphatase, which acts as a tumor suppressor functional biological meaning, we performed an expression in CML by antagonizing BCR-ABL. The inactivation of PP2A has analysis of 28 genes by quantitative real-time PCR in 30 CML been reported earlier to be associated with disease progression patients (see Supplementary Information for full description of in CML, as its activity in blast crisis patient cells is markedly Materials and methods). Expression data analysis, performed by inhibited.7 ASB6 contains a C-terminal SOCS box motif, typical the relative expression software tool,6 revealed a lower level of of negative regulators of the cytokine receptors. SH3GLB2 is all the 28 examined target gene transcripts in the group of involved in the activation of the apoptotic cell death mechanism deleted patients as compared with CML cases without deletions and in mediating the generation of signaling complexes. PKN3 is (Supplementary Table). However, the difference was statistically one of the target protein kinases for the small GTPase Rho and significant for six genes: protein kinase PKNbeta (PKN3, can be regulated by various signal transduction pathways that P ¼ 0.003), SH3-domain GRB2-like endophilin B2 (SH3GLB2, mediate cell growth and transformation. The subset of genes P ¼ 0.0018), protein phosphatase 2A regulatory subunit B0 mapping on chromosome 22 did not reveal any statistically (PPP2R4, P ¼ 0.007), ankyrin repeat and SOCS box-containing significant downregulation (Supplementary Table). Earlier papers 6 isoform (ASB6, P ¼ 0.002), ubiquitin-specific protease 20 took into consideration the potential role, in the pathogenesis of (USP20, P ¼ 0.010) and torsin family 1 member B (TOR1B, CML with der(9) deletion, of the tumor suppressor gene SWI/SNF- P ¼ 0.009) (Figure 1a, Supplementary Table). The expression related, matrix-associated actin (SMARCB1) gene, which is levels of the downregulated genes were 0.096, 0.222, 0.223, implicated in leukemogenesis through the remodeling of a local 0.198, 0.222 and 0.293 for PKN3, SH3GLB2, PPP2R4, ASB6, chromatin structure.1 In this study, we could not confirm a USP20 and TOR1B, respectively (Supplementary Table). All the significant downregulation of the SMARCB1 gene, excluding its six downregulated genes are located on chromosome 9 involvement in the poor prognosis of this subset of patients. It can sequences, centromeric to the ABL gene (Figure 1b) and are be noted that all six downregulated genes were located inside a implicated in crucial cellular pathways. It can be noted that the region starting at 1 Mb up to 2 Mb centromeric to ABL (Figure 1b). downregulated genes USP20 and TOR1B were lost in 67% of This observation could clarify earlier results showing a worse the 30 CML patients bearing der(9) deletions identified in our outcome for CML patients with large deletions.8 The down- study. Low levels of TOR1B transcripts could confer to the cell a regulation did not affect all these six genes at the same time, but greater vulnerability to oxidative stress, as this protein is at least one of them was significantly downregulated in 19 out of involved in conformational modeling of proteins and protection 21 patients (90%) bearing chromosome 9 sequences loss of from stress. USP20 possesses a deubiquitinating activity and 41 Mb. All the nine patients with a deletion of p1 Mb did not

Figure 1 Expression analysis and chromosomal localization of the six downregulated genes. (a) A graphic representation of gene expression data obtained by relative expression software tool (REST) analysis. The ﬁgure shows boxplots of gene expression where the top and bottom walls of each box indicate the 75th and 25th percentiles, whereas the dotted line represents the median. Whiskers above and below the box extend to the 90th and 10th percentiles. The reference gene expression is indicated by the red box and its median value corresponds to 1 on the expression ratio axis. (b) Chromosomal localization of the six downregulated genes identiﬁed in the CML cases with der(9) deletions analyzed in this study; the distance (in Mb) of the region where all six genes are mapped from the 50-end of ABL is reported. The diagram is not in scale.

Leukemia Letters to the Editor 815 Table 1 Characteristics of CML patients

Case no. Sex/age Sokal Response Response Response CCyR PKN3 SH3GLB2 PPP2R4 ASB6 USP20 TOR1B risk to IFN-a to imatinib to other TKI

1 F/69 I No No NE No del del del del del del 2 M/43 H No No Yes Yes del del del del del del 3 F/60 I No No NE No del del del del del del 4 M/35 L No NE NE No und und del del del del 5 M/49 L NE NE NE No del del del del del del 6 M/62 I No Yes NE Yes und del und und del del 7 M/40 L No NE NE No del del del del del del 8 F/62 L No Yes NE Yes del del del und und und 9 F/80 NA NA NA NA NA und und del del del del 10 F/29 L NA NA NA NA del del del del del del 11 F/63 I NE Yes NE Yes del del del del del del 12 M/71 I NE NE NE No del del del del del del 13 M/NA NA NA NA NA NA und und del del del del 14 M/40 L NE Yes NE Yes und und del del del del 15 M/NA NA NA NA NA NA und und del del del del 16 M/NA NA NA NA NA NA und del del del del del 17 F/29 L NE Yes NE Yes del del del del del del 18 M/44 I NE Yes NE Yes und und und und del del 19 F/51 H NE No Yes No und und del del del del 20 M/25 H NE Yes NE Yes und del del del del del 21 F/NA NA NA NA NA NA und und und und del del 22 F/46 H No No NE No und und und und und und 23 M/41 I No NE NE No und und und und und und 24 M/55 L NE NE NE No und und und und und und 25 F/65 I No Yes NE Yes und und und und und und 26 F/70 I NE NE NE No und und und und und und 27 M/28 L NE Yes NE Yes und und und und und und 28 F/NA NA NA NA NA NA und und und und und und 29 M/43 L NE Yes NE Yes und und und und und und 30 M/84 I NE Yes NE Yes und und und und und und Abbreviations: CcyR, complete cytogenetic response; CML, chronic myeloid leukemia; del, deleted; F, female; H, high; I, intermediate; IFN-a, interferon-a; L, low; M, male; NA, not available; NE, not evaluable; TKI, tyrosine kinase inhibitor; und, undeleted. Presenting characteristics (sex, age and Sokal risk), response to treatment (IFN-a, imatinib and other tyrosine kinase inhibitor) and deletion status of the 6 downregulated genes in the 30 analyzed CML patients. show any significant downregulation of the genes mapping on as a first-line treatment achieved CyCR and (iii) the majority of chromosome 9 deleted sequences (case nos. 22–30) (Table 1). CML patients with der(9) deletions 41 Mb treated with imatinib Among the 21 patients bearing chromosome 9 sequences in late CP did not achieve cytogenetic response. Therefore, it deletion on der(9) 41 Mb, 7 (33%) did not respond to IFN-a seems that the presence of deletions may exert a prognostic therapy. Among six (29%) patients treated with imatinib as a first- impact with regard to response to imatinib treatment, especially line treatment, five (83%) obtained a complete cytogenetic when it is administered in late CP. In fact, it is possible that response (CCyR). Among five patients started imatinib therapy in the insufficient gene dosage effect observed in these cases may the late chronic phase (CP) after IFN-a treatment failure, three easily be neutralized in the early rather than late CP of CML, (60%) did not achieve any kind of cytogenetic response, whereas probably because a lasting haploinsufficiency may be able to the remaining two (case no. 6 and 8) obtained CCyR. trigger several cellular pathways accountable for the absence of A total of nine (30%) patients showed the loss of chromosome response to imatinib therapy. 9 genomic sequences of p1 Mb. Among them, three (33%) In conclusion, we have shown, for the first time, a down- were treated with IFN-a and none of them obtained response regulated expression of genes located on the der(9) chromosome and three (33%) patients started imatinib as a first-line therapy in CML patients bearing genomic microdeletions. These findings and achieved CCyR. Two patients were treated with imatinib support the haploinsufficiency hypothesis, suggesting that in after failure to respond to IFN-a: the first one was in blast crisis these cases one allele could be insufficient to ensure an and did not achieve response to treatment (case no. 22), the adequate gene expression dosage even though the occurrence second one was in late CP and obtained CCyR (case no. 25) of other mechanisms, such as epigenetic gene silencing, cannot (Table 1). In the overall survival analysis, there was no be excluded. Overall, differences of the gene expression levels statistically significant difference between the patients with were detected in 70% of our series of CML cases with der(9) chromosome 9 sequences deletions of p1 Mb and those with deletions; however, the biological meaning and clinical 41 Mb (data not shown). Our patient series was too hetero- implications of this gene downregulation remain to be geneous in terms of treatment to be able to draw definite investigated. conclusions regarding the important topic of the impact of der(9) deletions on the response to imatinib therapy. However, we must point out that (i) none of the CML patients with deletions of Acknowledgements chromosome 9 sequences 41 Mb, representing half of the total deleted cases, responded to IFN-a therapy; (ii) the majority of We thank Ms MVC Pragnell, BA, for language revision of the CML patients with der(9) deletions 41 Mb treated with imatinib paper. The financial support of Associazione Italiana contro le

Leukemia Letters to the Editor 816 Leucemie (AIL)-BARI is gratefully acknowledged. AZ was 3 Quintas-Cardama A, Kantarjian H, Talpaz M, O0Brien S, Garcia- supported by fellowship of the Societa` Italiana di Ematologia Manero G, Verstovsek S et al. Imatinib mesylate therapy may Sperimentale (SIES). overcome the poor prognostic significance of deletions of derivative chromosome 9 in patients with chronic myelogenous leukemia. Blood 2005; 105: 2281–2286. F Albano1,3, L Anelli1,3, A Zagaria1,3, A Pannunzio1, V Liso1, 4 Huntly BJ, Bench A, Green AR. Double jeopardy from a single M Rocchi2,4 and G Specchia1,4 translocation: deletions of the derivative chromosome 9 in chronic 1Department of Hematology, University of Bari, Bari, Italy and myeloid leukemia. Blood 2003b; 102: 1160–1168. 2Department of Genetics and Microbiology, 5 Albano F, Anelli L, Zagaria A, Archidiacono N, Liso V, University of Bari, Bari, Italy Specchia G et al. ‘Home-brew’ FISH assay shows higher efficiency E-mail: [email protected] than BCR-ABL dual color, dual fusion probe in detecting micro- 3These authors contributed equally to this work. deletions and complex rearrangements associated with t(9;22) in 4These authors share the seniorship of this article. chronic myeloid leukemia. Cancer Genet Cytogenet 2007; 174: 121–126. 6 Pfaffl MW, Horgan GW, Dempfle L. Relative expression software References tool (REST) for group-wise comparison and statistical analysis of relative expression results in real-time PCR. Nucleic Acids Res 1 Huntly BJ, Reid AG, Bench AJ, Campbell LJ, Telford N, Shepherd P 2002; 30: e36. et al. Deletions of the derivative chromosome 9 occur at the time of 7 Neviani P, Santhanam R, Trotta R, , Notari M, , Blaser BW, Liu S the Philadelphia translocation and provide a powerful and et al. The tumor suppressor PP2A is functionally inactivated in blast independent prognostic factor in chronic myeloid leukaemia. Blood crisis CML through the inhibitory activity of the BCR/ABL-regulated 2001; 98: 1732–1738. SET protein. Cancer Cell 2005; 8: 355–368. 2 Huntly BJ, Guilhot F, Reid AG, Vassiliou G, Hennig E, Franke C 8 Fourouclas N, Campbell PJ, Bench AJ, Swanton S, Baxter EJ, Huntly et al. Imatinib improves but may not fully reverse the poor prognosis BJ et al. Size matters: the prognostic implications of large and small of patients with CML with derivative chromosome 9 deletions. deletions of the derivative 9 chromosome in chronic myeloid Blood 2003a; 102: 2205–2212. leukemia. Haematologica 2006; 91: 952–955.

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

Gene expression proﬁle of slowly responding subclones might represent different proﬁles already at diagnosis and might be used for prediction of outcome

Leukemia (2009) 23, 816–819; doi:10.1038/leu.2008.315; relapse in oligoclonal disease.8,9 On the basis of these results, we published online 6 November 2008 now hypothesize that the gene expression profile of the slow- responding subclones still present after the first weeks of chemotherapy might be different from the profiles of all leukemic Microarrays for gene expression profiling offers important cells at diagnosis, and are more predictive for outcome as research tools for novel classification of leukemia’s and subclones are being selected during induction therapy. lymphomas, which may have implications for prognosis and To study these differences in the expression profiles of treatment.1,2 In the field of hematologic malignancies, several leukemic cells, we selected 26 genes (5 genes from Staal papers have demonstrated that the use of microarrays indeed et al.3 and 21 top-ranked genes from Beesley et al.4) that showed may provide novel information at diagnosis. Using Affymetrix differential expression in diagnosis and relapse, originating from oligonucleotide microarrays, distinct expression profiles have the above-mentioned studies that compared gene expression been described for the different cytogenetically defined sub- levels between diagnosis and relapse in B-precursor ALL. groups of acute lymphoblastic leukemia (ALL).2 Subsequently, Gene sequences (mRNA and genomic DNA) were obtained also prognostic and disease progression markers have been from public databases NCBI (http://www.ncbi.nlm.nih.gov) and proposed, ultimately leading to the prospect of patient-tailored BLAT (http://genome.ucsc.edu/cgi-bin/hgBlat). Primer design therapy.2 However, so far no clear prognostic profiles have been was performed using Primer Express 2.0 and Oligo6. Finally, defined. In addition, microarray-based comparisons of paired for 23 genes, adequate primers were designed, showing specific diagnosis-relapse samples3–5 from patients with B-precursor amplification (Table 1). ALL have been performed, but so far, not many studies have Paired diagnosis-day15-relapse samples from three relapsed used this approach. Also, some studies focussed on the identi- B-precursor ALL patients were analyzed. On the basis of the fication of glucocorticoid-response genes in children with ALL, markers expressed on the tumor cells (CD34, CD19, and CD10) comparing diagnosisFday 8 gene expression profiles.6,7 cryopreserved mononuclear cells obtained from Ficoll density Treatment of pediatric ALL is increasingly based on the centrifugation before cryopreservation, were purified using an concept of tailoring the intensity of treatment to the risk of MOFlo High Performance Cell Sorter. Total RNA was extracted treatment failure or success. from these cell subsets following the TRIZOL protocol, Clinical studies have shown that it is possible to stratify patients subsequently followed by cDNA synthesis. RNA integrity was according to the levels of minimal residual disease (MRD) after assessed using an Agilent 2100 BioAnalyser (Santa Clara, CA, induction therapy and early during further treatment, as it has USA). The expression levels of 23 genes were determined in a been demonstrated that the MRD level is the best predictive level SYBR Green based real-time PCR assay. for disease outcome. In previous studies we reported that slow All Ct values were normalized to b-glucuronidase (GUS) responding subclones represent the clones causative for leukemic expression. We compared the expression levels in diagnosis and

Leukemia