Letters to the Editor 1113 References 5 Dworzak MN, Froschl G, Printz D, Mann G, Po¨tschger U, Mu¨hlegger N et al. Prognostic significance and modalities of flow cytometric minimal residual disease detection in childhood acute 1 Woessmann W, Seidemann K, Mann G, Zimmermann M, Burkhardt lymphoblastic leukemia. Blood 2002; 99: 1952–1958. B, Oschlies I et al. The impact of the methotrexate administration 6 Hummel M, Bentink S, Berger H, Klapper W, Wessendorf S, Barth schedule and dose in the treatment of children and adolescents with TF et al. A biologic definition of Burkitt’s lymphoma from B-cell neoplasms: a report of the BFM Group Study NHL-BFM95. transcriptional and genomic profiling. N Engl J Med 2006; 354: Blood 2005; 105: 948–958. 2419–2430. 2 Busch K, Borkhardt A, Wossmann W, Reiter A, Harbott J. Combined 7 Dave SS, Fu K, Wright GW, Lam LT, Kluin P, Boerma EJ et al. polymerase chain reaction methods to detect c-myc/IgH rearrange- Molecular diagnosis of Burkitt’s lymphoma. N Engl J Med 2006; ment in childhood Burkitt’s lymphoma for minimal residual disease 354: 2431–2442. analysis. Haematologica 2004; 89: 818–825. 8 Attarbaschi A, Mann G, Dworzak M, Trebo M, Urban C, Fink FM 3 Mussolin L, Basso K, Pillon M, d’Amore ES, Lombardi A, Luzzatto L et al. Malignant non-Hodgkin’s lymphoma of childhood and et al. Prospective analysis of minimal bone marrow infiltration in adolescence in Austria – therapy results between 1986 and 2000. pediatric Burkitt’s lymphomas by long-distance polymerase chain Wien Klin Wochenschr 2002; 114: 978–986. reaction for t(8;14)(q24;q32). Leukemia 2003; 17: 585–589. 9 Canonico B, Zamai L, Burattini S, Granger V, Mannello F, 4 Dworzak MN, Fritsch G, Fleischer C, Printz D, Fro¨schl G, Buchinger Gobbi P et al. Evaluation of leukocyte stabilisation in P et al. Comparative phenotype mapping of normal vs malignant TransFix-treated blood samples by flow cytometry and trans- pediatric B-lymphopoiesis unveils leukemia-associated aberrations. mission electron microscopy. J Immunol Methods 2004; 295: Exp Hematol 1998; 26: 305–313. 67–78.

Transcriptional features of multiple myeloma patients with 1q gain

Leukemia (2007) 21, 1113–1116. doi:10.1038/sj.leu.2404616; TC3 including tumors that do not fall into any of the other published online 22 February 2007 groups, most of which express CCND2; TC4 showing high CCND2 levels and the presence of the t(4;14) translocation and TC5 expressing the highest levels of CCND2 in association with Abnormalities of are among the most either the t(14;16) or t(14;20) translocation.3 frequent chromosomal alterations in multiple myeloma Assessment of 1q/gain by FISH was performed by using three (MM), being found in up to 45% of patients.1,2 It has been BAC clones specific for the BCL9 (1q21.1), CKS1B (1q22) and reported that the short arm of chromosome 1 is preferentially ARF1 (1q42.13) loci, and setting the threshold as 10%. Specific involved in deletions, whereas the long arm is associated alterations were identified in 40/77 (51.9%) patients; three with amplification. The gain of 1q (1q/gain) can occur as (75%) or four (12.5%) signals of all the 1q probes were found in isochromosomes, duplications or jumping translocations. It 35 patients and, in the remaining five samples (12.5%), the has been widely reported that 1q/gain MM patients are probes mapping to 1q21 and 1q22 showed more signals than characterized by complex karyotypes and aggressive disease, that mapping to 1q42. 1q/gain was observed in the majority of and a close association with poor-risk genetic features, such purified plasma cells (490%) in all but three patients (range as chromosome 13q deletion (D13) and the t(4;14) trans- 12–20%). 1q/gain was significantly absent in TC2 group location has also been described.1 It has been recently (Po10À4) and present in TC3 (P ¼ 0.008), whereas the correla- demonstrated that gains/amplification of 1q21 increase as tion was not significant in the TC1 (P ¼ 0.053) or TC4 the condition goes from smoldering to overt MM, thus (P ¼ 0.142) groups; in addition, 1q extra copies significantly suggesting that these regions contain critical for disease associated with D13 (Po10À4) and polisomy progression.2 These findings along with the limited information (although at a limited significance level of P ¼ 0.038), but not concerning specific transcriptional profiles prompted us to ploidy status (P ¼ 0.0971). molecularly characterize 1q/gain MMs by FISH and microarray To identify a specific transcriptional fingerprint characterizing analyses. 1q/gain, we made a supervised analysis of 1q/gain versus 1q/ Our study includes a panel of 77 MM patients at diagnosis, normal MM patients using SAM algorithm. Seventy-two probe whose characteristics have been deposited in National Center sets (specific for 61 genes) distinguished the 40 1q/gain from the for Biotechnology Information’s Expression Omnibus 37 1q/normal cases (Figure 1a and Table 1a, b). Notably, in the (GEO; http://www.ncbi.nlm.mih.gov/geo, accession number 1q/gain patients 41 of the 43 upregulated genes mapped to GSE6401). Bone marrow plasma cells were purified (490% in 1q12-q44, whereas a significant percentage of the down- all cases) using CD138-immunomagnetic bead selection and regulated genes was localized on 13q (7/18; characterized by FISH for the presence of 11 polisomy, the most 39%) and 11 (6/18, 33%). To verify whether the 1q/gain recurrent IGH translocations, ploidy status, D13, and global reflected transcriptional imbalances of specific chromosomal gene expression profiling using the Affymetrix U133A gene regions, the expression data from the 1q/gain and 1q/normal chips, as described previously.3 Patients were then stratified samples were also analyzed in the context of the physical accordingly to the proposed translocation/cyclin D (TC) localization of the genes using a model-free statistical method classification in five groups: TC1 characterized by the t(11;14) (LAP),4 allowing the identification of five modulated chromoso- or t(6;14) translocations, with the consequent overexpression of mal regions (Figure 1b). In the 1q/gain patients, the region either CCND1 or CCND3, and a nonhyperdiploid status; TC2 1q21.1-q44 (absolute positions: 146,567,360-245,353,955) was showing low/moderate levels of the CCND1 gene in the absence upregulated and 13q12-14 (22,800,966-47,961,101) down- of any primary IGH translocations, and a hyperdiploid status; regulated, which is in line with the significant association

Leukemia Letters to the Editor 1114 between 1q abnormalities and D13; regional downregulation 15q24.1-q25 (72,397,963-78,483,746), which may reflect the was also observed in chromosome regions 11p15 (2,380, higher frequency of hyperdiploid patients in the 1q/normal 098-6,372,930), 11q13-q23 (71,317,730-113,946,523) and group.

Figure 1 (a) Supervised analysis of 1q/gain versus 1q/normal patients. The differentially expressed genes discriminating 1q/gain and 1q/normal classes were identified using Significant Analysis of Microarrays software version 2.21 (SAM; Excel front-end publicly available at http://www- stat.stanford.edu/~tibs/SAM/index.html; cutoff for significance: q-value ¼ 0 with median FDR ¼ 0%, 90th %ile FDR ¼ 0%) and visualized by means of dChip software. The color scale bar represents relative gene expression changes normalized by standard deviations. Information about chromosome 11 trisomy ( þ 11), D13, ploidy status (HD ¼ hyperdiploid), TC classes and 1q extra copies (1q þ ) are included (n ¼ data not available). (b) Regional analysis of 1q/gain versus 1q/normal patients. The whole genome plot of the differentially expressed 858 genes identified using LAP algorithm4 (q-value ¼ 0) shows five modulated chromosomal regions. The vertical axis represents the progressive chromosome number, and the horizontal axis (blue lines) the progressive absolute position of the probes represented on HG-U133A gene chips for each chromosome. The white bars indicate the exact chromosomal locations, and the colored perpendicular lines the locations and up- (red) or downregulation (green) of genes in the 1q/gain patients (see geneplotter package from Bioconductor for details).

Table 1a Forty-three upregulated genes in 1q/gain, ordered by chromosomal location and gene name whenever more than one probe recognized the same gene, the one with the best score is shown

Probe ID Gene symbol Chromosome location Probe ID Gene symbol Chromosome location

202337_at PMF1 1q12 208716_s_at TMCO1 1q22-q25 209044_x_at SF3B4 1q12-q21 201403_s_at MGST3 1q23 210386_s_at MTX1 1q21 208114_s_at ISG20L2 1q23.1 210417_s_at PIK4CB 1q21 215158_s_at DEDD 1q23.3 201771_at SCAMP3 1q21 208684_at COPA 1q23-q25 216591_s_at SDHC 1q21 202846_s_at PIGC 1q23-q25 209561_at THBS3 1q21 202427_s_at BRP44 1q24 202596_at ENSA 1q21.2 214838_at SFT2D2 1q24.2 222212_s_at LASS2 1q21.2 217748_at ADIPOR1 1q32 210460_s_at PSMD4 1q21.2 212165_at C1orf37 1q32.1 221189_s_at TARSL1 1q21.2 202187_s_at PPP2R5A 1q32.2-q32.3 216873_s_at ATP8B2 1q21.3 204478_s_at RABIF 1q32-q41 200052_s_at ILF2 1q21.3 208755_x_at H3F3A 1q41 209609_s_at MRPL9 1q21.3 202374_s_at RAB3GAP2 1q41 203515_s_at PMVK 1q21.3 200065_s_at ARF1 1q42 201378_s_at UBAP2L 1q21.3 221497_x_at EGLN1 1q42.1 217978_s_at UBE2Q1 1q21.3 214170_x_at FH 1q42.1 218270_at MRPL24 1q21-q22 202324_s_at ACBD3 1q42.12 200896_x_at HDGF 1q21-q23 212371_at C1orf121 1q44 201275_at FDPS 1q22 214831_at MED28 4p16 218296_x_at MSTO1 1q22 210859_x_at CLN3 16p12.1 218291_at MAPBPIP 1q22

Leukemia Letters to the Editor 1115 Table 1b Eighteen downregulated genes in 1q/gain

Probe ID Gene symbol Chromosome location Probe ID Gene symbol Chromosome location

209430_at BTAF1 10q22-q23 212603_at MRPS31 13q14.11 200023_s_at EIF3S5 11p15.4 221816_s_at PHF11 13q14.2 200909_s_at RPLP2 11p15.5-p15.4 217731_s_at ITM2B 13q14.3 200019_s_at FAU 11q13 214252_s_at CLN5 13q21.1-q32 211967_at PORIMIN 11q22.1 201960_s_at MYCBP2 13q22 209310_s_at CASP4 11q22.2-q22.3 221702_s_at TM2D3 15q26.3 218826_at SLC35F2 11q22.3 217807_s_at GLTSCR2 19q13.3 208742_s_at SAP18 13q12.11 201889_at FAM3C 7q22.1-q31.1 200012_x_at RPL21 13q12.2 202962_at KIF13B 8p12

Table 2 Accuracy measures in the classification of the 248 MM samples from the independent data set by Zhan et al.5

Putative label class No. of patients Predicted class

1q/gain-MM 1q/normal-MM Specificity (%) Sensitivity (%)

1q/gain-MM 114 93 21 86.1 81.6 1q/normal-MM 134 15 119 85.0 88.8 Total 108 140

Notably, the transcriptional fingerprint identified in our thalidomide uptake does not improve event-free survival.2 database was validated by means of a meta-analysis of a Genes involved in the complex network leading to ER stress- publicly available independent set of 248 MM cases (114 1q/ induced responses are also significantly modulated in 1q/gain gain and 134 1q/normal) profiled on Affymetrix HG-U133 2.0 patients. The upregulation of the CLN3 (chaperone gene), Plus gene chips.5 A PAM model was designed with all of UBAP2L and UBE2Q1 ( cycle) and PSMD4 (protea- the probes trained on a random selection of half 1q/gain and some degradation) could be seen in the context of the ER stress 1q/normal patients, and was then used to classify the entire test set secondary defensive mechanism called ER-associated into the 1q/gain and 1q/normal groups (Table 2). The classification degradation. Finally, we found the downregulation of CASP4, accuracy measures showed a global classification rate of 85.2% and which has been described as localizing to the ER and a mean specificity of 85.5%, suggesting that the identified expression specifically initiating apoptosis in response to ER-stress stimuli. signature is a highly conserved characteristic of the 1q/gain In this regard, it has been shown that ER stress-induced abnormality and is not affected by cohort- or lab-specific biases. apoptosis can play an important role in the sensitivity of The functional genomic annotation of our selected list was malignant cells to certain drugs, including bortezomib. In performed using the Database for Annotation, Visualization and addition, a recent pilot study using human MM cell lines found Integrated Discovery (DAVID) Tool 2006 (US National Institutes that submicromolar concentrations of brefeldin A, an inhibitor of Health at http://david.niaid.nih.gov/david/version2/in- of ER-Golgi protein transport currently being developed as a dex.htm) and indicated that most of the discriminating genes novel anticancer agent, effectively induces apoptosis in MM are involved in a network of specialized biological functions. In cells by activating caspase-2, which localizes to the Golgi particular, the patients with 1q/gain showed upregulation of apparatus.6 These findings suggest that improving our under- genes involved in intracellular protein transport, such as COPA standing of ER stress-induced responses may lead to important and ARF1, which play a role in vesicle-mediated transport from contributions to treatment strategies. the endoplasmic reticulum (ER) to the Golgi region, as well as a With regard to currently available literature, the overexpres- significant overexpression of RABIF and RAB3GAP2 both of sion of COPA, RABIF and ARF1 was reported by Walker et al.7 which are related to the ubiquitous key regulators of membrane- in a cohort of 11 1q/gain patients out of 30 MM patients at vesicle transport Rab GTPases. These findings may also partially diagnosis, and overexpression of the COPA gene was also account for the increased expression of genes coding for described by Carrasco et al.8 in the group of hyperdiploid MM involved in energy production pathways, specifically, with a prevalence of 1q/gain and D13. the citrate cycle (FH and SDHC), glycogen metabolism In conclusion, these data extend our knowledge of the (PPP2R5A) and fatty acid oxidation (ADIPOR1) and steroid specific genes/pathways deregulated in MM patients with 1q biosynthesis (ACBD3, PMVK and FDPS). Among the down- gain, and also provide an input for further therapeutic strategies regulated transcripts in the 1q/gain group, we recognized three in MM. genes involved in translational mechanisms (RPLP2, RPL21 and FAU), and the CASP4 gene. In the context of these data, it is worth noting the recent suggestion that B-cell malignancies Acknowledgements (including MM) may be highly dependent on ER-Golgi protein transport for their survival and that targeting this process may This study was supported by grants from the Associazione Italiana represent a new therapeutic strategy. This could be a promising Ricerca sul Cancro (AIRC) (to AN); from the Italian Ministry of approach, especially in these MM cases that poorly respond to University and Research (FIRB RBNE01TZZ8 and RBAU01935A) current treatments, such as the 1q/gain patients in whom and the OncoSuisse Collaborative Cancer Research Project OCS

Leukemia Letters to the Editor 1116 01517–02–2004 (to SB); KT was supported by a fellowship from from MGUS to relapsed myeloma and is related to prognosis and the Fondazione Italiana Ricerca sul Cancro (FIRC). disease progression following tandem stem-cell transplantation. Blood 2006; 108: 1724–1732. S Fabris1,5, D Ronchetti1,2,5, L Agnelli1, L Baldini2, 3 Agnelli L, Bicciato S, Mattioli M, Fabris S, Intini D, Verdelli D F Morabito3, S Bicciato4, D Basso4, K Todoerti1,2, et al. Molecular classification of multiple myeloma: a distinct L Lombardi1, G Lambertenghi-Deliliers2 and A Neri1,2 transcriptional profile characterizes patients expressing CCND1 1Centro di Genetica Molecolare ed Espressione Genica, and negative for 14q32 translocations. J Clin Oncol 2005; 23: Fondazione IRCCS Ospedale Maggiore Policlinico, 7296–7306. Mangiagalli e Regina Elena, Milan, Italy; 4 Callegaro A, Basso D, Bicciato S. A locally adaptive statistical 2Dipartimento di Scienze Mediche, Universita` degli Studi di procedure (LAP) to identify differentially expressed chromosomal regions. Bioinformatics 2006; 22: 2658–2666. Milano, Milan, Italy; 5 Zhan F, Huang Y, Colla S, Stewart JP, Hanamura I, Gupta S et al. 3U.O. Ematologia, A.O. ‘Annunziata’, Cosenza, Italy and 4 The molecular classification of multiple myeloma. Blood 2006; 108: Dipartimento dei Processi Chimici dell’Ingegneria, Universita` 2020–2028. degli Studi, Padua, Italy 6 Carew JS, Nawrocki ST, Krupnik YV, Dunner Jr K, McConkey DJ, E-mail: [email protected] 5 Keating MJ et al. Targeting endoplasmic reticulum protein transport: The first two authors contributed equally to this work a novel strategy to kill malignant B cells and overcome fludarabine resistance in CLL. Blood 2006; 107: 222–231. 7 Walker BA, Leone PE, Jenner MW, Li C, Gonzalez D, Johnson DC References et al. Integration of global SNP-based mapping and expres- sion arrays reveals key regions, mechanisms, and genes important 1 Fonseca R, Barlogie B, Bataille R, Bastard C, Bergsagel PL, Chesi M in the pathogenesis of multiple myeloma. Blood 2006; 108: et al. Genetics and cytogenetics of multiple myeloma: a workshop 1733–1743. report. Cancer Res 2004; 64: 1546–1558. 8 Carrasco DR, Tonon G, Huang Y, Zhang Y, Sinha R, Feng B et al. 2 Hanamura I, Stewart JP, Huang Y, Zhan F, Santra M, Sawyer JR et al. High-resolution genomic profiles define distinct clinico-pathoge- Frequent gain of chromosome band 1q21 in plasma-cell dyscrasias netic subgroups of multiple myeloma patients. Cancer Cell 2006; 9: detected by fluorescence in situ hybridization: incidence increases 313–325.

Expression of Polycomb-group (PcG) protein BMI-1 predicts prognosis in patients with acute myeloid leukemia

Leukemia (2007) 21, 1116–1122. doi:10.1038/sj.leu.2404623; We studied bone marrow (BM) samples (except two from published online 22 March 2007 peripheral blood (PB) and granulocytic sarcoma patient from the lymph node) obtained from 64 patients with newly diagnosed AML (Supplementary Tables 1 and 2). According to the Helsinki BMI-1, a member of the Polycomb group of gene family, is declaration, informed consent was obtained from the patients essential for the self-renewal of hematopoietic, neural and and donors (used as control), and an institutional review board cancerous stem cells.1 BMI-1 is highly expressed in malignan- approved the study. AML M3 patients were excluded from this cies like B cell non-Hodgkin lymphoma, Hodgkin’s lymphoma, study, due to their different disease biology and different colorectal carcinoma, liver carcinoma, non-small cell lung treatment strategy. All cases of de novo AML were diagnosed cancer, breast carcinoma and medulloblastoma. Interestingly, as per the FAB criteria. Granulocytic sarcoma was diagnosed the study of BMi-1 knockout mice revealed that BMi-1 is using lymph node biopsy. Secondary AML (MDS-AML or t-AML) indispensable for the self-renewal of both leukemic and normal was diagnosed according to the revised recommendations of the hematopoietic stem cells.2 In addition to its vital role in stem international working group on AML.4 Complete remission (CR), cell self-renewal, BMI-1 regulates the proliferative output of partial remission , overall survival (OS), relapse-free survival primitive hematopoietic progenitors in a critical manner.1 We (RFS) and remission duration (RD) were defined as per reported previously BMI-1 as a marker of disease progression in recommendations of the same group.4 BM blasts p5%, absolute cases of myelodysplastic syndrome (MDS).3 Therefore, we neutrophil count X1000 ml and platelet count X100 000 ml assumed that BMI-1 expression could be used to predict the were the criteria for CR. Cytogenetic remission or molecular prognosis of patients with acute myeloid leukemia (AML) and remission was not used to define CR. Flow cytometry with thus could serve as a tool to tailor therapy for this disease. This is appropriate antibody was performed to confirm the diagnosis the first study, in which flow cytometry was used to determine and morphologic leukemia-free state. Survival was measured the expression of BMI-1 in AML blasts of human subjects. from the date of diagnosis to the date of death or last follow-up.

Figure 1 (A) BMI-1 expression in AML cell lines by immunoblot analysis. Immunoblot analysis demonstrated BMI-1 expression in five AML cell lines: KG-1 (3), Mono-7 (4), HEL (5), HL-60 (6) and U-937 (7). WI-38 (1) and IL-2-stimulated peripheral blood T cells (2) expressed much less BMI-1 protein. (B) BrdU incorporation by acute promonocytic leukemic cells, U-937: Cell cycle analysis was performed with the staining pattern of 7-AAD and BrdU in (b–d). In (a), M1, M2 and M3 represent BMI-1 negative cells, cells with lower intensity of BMI-1 and higher intensity of BMI- 1, respectively. Dotted line indicates isotype control. BMI-1-negative U-937 cells were collected in (b), cells with lower intensity of BMI-1 were shown in (c) and cells with higher intensity of BMI-1 were depicted in (d). R4, R5, R6 and R7 are indicative of G0/G1 phase, S phase, G2/M phase and apoptotic cells, respectively. (C) BMI-1 expression in AML patients. Patient 16, a case of CD34 þ AML M1 had higher BMI-1 expression (94.57%). Patient 3, a case of CD34 þ AML M0 had the lower expression (21.88%). Patient 21, a case of CD34–AML M2 had the lower expression (6.91%) comparable to control CD34 þ cells as described in Materials and method. (D) BMI-1 expression (%) in FAB subtypes and secondary AML. Control CD34 þ cells and leukemic blasts were examined for BMI-1 expression. ‘FF’ indicates mean value.

Leukemia