Letters to the Editor 1639 individuals. This was performed in a number of studies for flow patient) is too low due to false positive healthy and regenerating cytometry (Kern et al.;6 San Miguel et al.;8 Venditti et al.;9 and bone marrows. others) and also for leukemia-specific genes (Lapillone et al.;3 If we use, for instance, a threshold of 10À3 for clinical Steinbach et al.;4 Matsushita et al.;5 and others). decision-making, it does not matter whether the LPT is 10À4 or However, such analyses can still produce misleading results. 10À6. What really matters is the sensitivity and specificity of the In clinical practice, we do not compare samples that contain diagnostic test at the threshold of 10À3. residual leukemic cells with steady-state healthy bone marrow. We should put less effort into determining LPTs and more Instead, we look at bone marrows that are regenerating after effort into determining sensitivity and specificity of our methods heavy chemotherapy. The expression of presumed leukemia- at the clinically relevant thresholds. specific genes or the frequency of presumed leukemia specific Because of the many issues that influence the quality of each phenotypes can be very different in such samples compared to method of monitoring MRD, the best way of really comparing steady-state healthy bone marrow. two methods is to simultaneously apply both of them in one For instance, the gene CSPG4 was identified as a possible clinical trial. Such studies are strongly needed. MRD marker.4 It was highly expressed in AML samples but not in healthy bone marrow. However, we found that it was useless D Steinbach and K-M Debatin as an MRD marker because expression of CSPG4 was also high Department of Pediatric Oncology, University Hospital Ulm, Ulm, Germany in bone marrow that was free of leukemia but was regenerating E-mail: [email protected] after chemotherapy.4 Therefore, to determine the real sensitivity and specificity of an MRD analysis, it is mandatory to analyze a large number of References leukemia-free, regenerating bone marrows. A second problem of analyzing serial dilutions of leukemic 1 Yin JA, Grimwade D. Minimal residual disease evaluation in acute myeloid leukaemia. Lancet 2002; 360: 160–162. cells in samples of healthy bone marrow is that this does not really 2 Campana D. Minimal residual disease studies in acute leukemia. simulate the situation of minimal residual disease. The important Am J Clin Pathol 2004; 122 (Suppl): S47–S57. clinical question is the amount of leukemic stem cells that is still 3 Lapillonne H, Renneville A, Auvrignon A, Flamant C, Blaise A, Perot C present in the patient. If the leukemic phenotype that is used for et al. High WT1 expression after induction therapy predicts high risk flow cytometry is present on the majority of leukemic cells but not of relapse and death in pediatric acute myeloid leukemia. J Clin on leukemic stem cells, the sensitivity could be lower than Oncol 2006; 24: 1507–1515. 4 Steinbach D, Schramm A, Eggert A, Onda M, Dawczynski K, Rump A anticipated. If the expression of a leukemia-associated gene, like et al. Identification of a set of seven genes for the monitoring WT1, is particularly high in leukemic stem cells, the sensitivity of minimal residual disease in pediatric acute myeloid leukemia. might be much better than anticipated and vice versa. Clin Cancer Res 2006; 12: 2434–2441. 5 Matsushita M, Ikeda H, Kizaki M, Okamoto S, Ogasawara M, Ikeda Y et al. Quantitative monitoring of the PRAME gene for the detection of minimal residual disease in leukaemia. Br J Haematol 2001; 112: Conclusions 916–926. 6 Kern W, Danhauser-Riedl S, Ratei R, Schnittger S, Schoch C, Kolb HJ et al. Detection of minimal residual disease in unselected patients Monitoring MRD has become a strong diagnostic tool in acute with acute myeloid leukemia using multiparameter flow cytometry leukemia. It is widely used for clinical decision-making in acute for definition of leukemia-associated immunophenotypes and lymphoblastic leukemia, chronic myeloid leukemia and acute determination of their frequencies in normal bone marrow. Haema- promyelocytic leukemia. Various methods have been developed tologica 2003; 88: 646–653. to monitor MRD in AML and we should proceed to put them 7 van der Velden VH, Cazzaniga G, Schrauder A, Hancock J, Bader P, into clinical practice. Panzer-Grumayer ER, et al., European Study Group on MRD Sensitivity is an important issue in the comparison of these detection in ALL (ESG-MRD-ALL). Analysis of minimal residual disease by Ig/TCR gene rearrangements: guidelines for interpretation methods. When the term is used, it should always be clear of real-time quantitative PCR data. Leukemia 2007; 21: 604–611. whether we are talking about the LPT or the real sensitivity and 8 San Miguel JF, Vidriales MB, Lopez-Berges C, Diaz-Mediavilla J, specificity of our diagnostic test. Gutierrez N, Canizo C et al. Early immunophenotypical evaluation The LPT is easy to determine and for most methods of of minimal residual disease in acute myeloid leukemia identifies monitoring MRD, is in the range of 10À4–10À6. These figures different patient risk groups and may contribute to postinduction have a very limited meaning because for clinical decision- treatment stratification. Blood 2001; 98: 1746–1751. 9 Venditti A, Buccisano F, Del Poeta G, Maurillo L, Tamburini A, making LPT cannot be used as the cutoff for high versus low Cox C et al. Level of minimal residual disease after consolidation MRD. At such a low threshold, the specificity of monitoring therapy predicts outcome in acute myeloid leukemia. Blood 2000; MRD (likelihood of a negative test result in a truly negative 96: 3948–3952.

ETV6 mutations and loss in AML-M0

Leukemia (2008) 22, 1639–1643; doi:10.1038/leu.2008.34; belonging to the E26 transforming specific (ETS) family of DNA- published online 28 February 2008 binding proteins. ETV6 is known as a proto-oncogene involved in translocation with over 40 partners.1 In acute myeloid leukemia (AML) only a few rare translocations result in transforming fusion ETV6 (ETS translocation-variant gene 6, located on chromosome proteins,1 indicating that the oncogenic role of ETV6 does not 12p), also known as TEL, encodes a transcription repressor play a major part in AML. However, abnormalities of the short

Leukemia Letters to the Editor 1640

Figure 1 Single-nucleotide polymorphism analysis of patients with chromosome 12 abnormalities. (a) Loss-of-heterozygosity values based on haplotype, using the paired normal. Each box represents the combined call between tumor sample and respective control (T cells). Black boxes represent loss of heterozygosity, dark gray boxes no loss and light gray boxes non-informative markers. Boxes are displayed proportionally to the position of the SNP that they represent in relation to the cytogenetic band to the left of the panel. (b) Chromosome copy number calculated for the common region showing loss of heterozygosity in panel a (amplified). Copy number for the tumor samples was inferred using the paired normal as reference and a median smoothing. Gray boxes represent two copies and black boxes one copy (deletion) for each chromosome locus. (c) Schematic representation of the genes present in the minimal deleted overlapping region defined in (b) based on the UCSC Genome Browser (http://genome.ucsc.edu/). Solid black boxes represent clusters of related genes. SNP, single-nucleotide polymorphism.

Table 1 Clinical, hematological, cytogenetic features and mutational status of ETV6

Patient Age Diagnoses Karyotype ETV6

Allele 1 Allele 2

1a 65 AML-M0 46,XX,del(16)(q22?),i(17)(q10),del(20)(q?) Deleted WT 2a 67 sAML-M0 47,XX,t(4;12)(q12;p13),À21,+der21del(q?)x2 Translocated WT 6a 37 AML-M0 46,XY S107DfsX21 V345_Y346insR 9a 47 sAML-M0 52,XX,t(1;4)(p13;p12),+6,+8,t(10;12)(q11;p11),+18,+19,+20,+21 Deleted WT 21 59 AML-M0 46,XY R360X WT 43 64 AML-M0 46,XY,t(4;12)(q12;p13) Translocated WT 45 F AML-M0 ND Deleted WT 58 50 AML-M0 46,XY,idic(21)(p11.2) F103LfsX11 WT Abbreviations: AML, acute myeloid leukemia; ETV6, ETS translocation-variant gene 6; ND, not done; s, secondary; WT, wild type. aPatients in the study by Silva et al.7

Leukemia Letters to the Editor 1641 arm of chromosome 12 (12p) are found in about 5% of AML and investigated specifically in this subtype.4,6 To better understand myelodysplastic syndromes. Most abnormalities consist of total or the role of ETV6 in AML-M0, we studied 52 M0 patients using partial loss of 12p usually affecting ETV6 and CDKN1B, cytogenetic techniques complemented with single-nucleotide implicating these genes as tumor-suppressor genes.1–4 Recently, polymorphism (SNP) arrays and sequencing of ETV6,as heterozygous mutations of ETV6, resulting in loss of repressor described in the Supplementary Information. The cohort was activity, were found in AML, adding to the view that ETV6 might not selected for any parameter rather than being AML-M0, had a have tumor-suppressor characteristics.5 median age of 61 years (one patient was a child) and consisted Whereas translocations and deletions involving ETV6 have of de novo, therapy-related (1 case) and secondary leukemia been reported in AML-M0, ETV6 mutations were never (4 cases).

Figure 2 Schematic representation of FISH analysis of the breakpoints in t(10;12) in patient 9 and t(4;12) in patient 2, and fusion transcripts in patients 2 and 43. (a) Determination of the deletion and translocation break points in patient 9. The deleted region in chromosome 12 is localized between SNP rs252027 to SNP rs747726 (SNPs present in a single copy are not underlined) as determined using the GeneChip 10K array (Figure 1b). By FISH analysis, we determined that the distal translocation break point occurred between BACs RP11-685G3 and RP11-959H9, whereas the proximal break point occurred between BACs RP11-165F6 and RP11-1150D7. BACs represented by white boxes are hemizygously deleted, by black boxes are retained in the original chromosome, by dark gray boxes are translocated to the partner chromosome, whereas light gray boxes show an intrachromosomal cross-hybridization signal on der(12) and a split signal with der(10.) Only known genes are represented and white boxes represent clusters of related genes. (b) BAC clones on chromosome 4 and 12 used to determine the t(4;12) break point in patient 2. The gap in chromosome 4 is approximately of 0.75 Mb. Color code for the BAC probes as in panel (a), with the exception of striped boxes, which represent BACs showing a split signal. (c) Genomic structure of CHIC2 (4q12) and ETV6 (12p13) according to UCSC Genome Browser and positions of primers used for PCR and sequencing (genes are not represented in the same scale). Dotted crossed lines represent the probable area where the t(4;12) occurred in patients 2 and 43. (d) ETV6-CHIC2 transcripts detected in both patients 2 and 43. Sequence of ETV6-CHIC2 cDNA showed an out-of-frame fusion between exon 1 of ETV6 and exon 2 of CHIC2. The arrow indicates the boundary between the ETV6 and CHIC2 exons. BACs in all panels were chosen according to the latest version of the UCSC Genome Browser. BAC, bacterial artificial chromosome; ETV, ETS translocation-variant gene; FISH, fluorescent in situ hybridization.

Leukemia Letters to the Editor 1642 Three patients presented deletions ranging from 3.2 to 14.3 Mb in 12p. The minimal overlap region of deletion between these patients included ETV6 but excluded CDKN1B. Two others patients had a t(4;12)(q12;p13). In both patients, we detected an out-of-frame CHIC2-ETV6 fusion transcript. The reciprocal transcript was not detected, supporting the view that t(4;12) does not result in an oncogenic fusion protein. In addition, we detected ETV6-inactivating biallelic mutations in one patient and mutations leading to ETV6 truncated proteins in another two patients (5.7%). DNA isolated from flow-sorted leukemic cells7 was compared with T-cell control DNA using GeneChip 10K arrays (Affymetrix, Santa Clara, CA, USA). Deletions in 12p were detected as loss of heterozygosity and copy-number reduction in patients 1, 9 and 45 (Figure 1; Table 1). The deletion in patient 9, confirmed by fluorescent in situ hybridization (FISH) analysis (Figure 2a), was Figure 3 Schematic representation of the predicted ETV6-mutant accompanied by a t(10;12)(q11;p11) translocation detected by products. Amino-acid positions are shown under the wild-type ETV6 protein. The pointed (PNT) and ETS DNA-binding domain karyotyping (Table 1). Whereas deletions in patients 1 and 45 are represented by a black and a gray box, respectively. New open included both ETV6 and CDKN1B, the proximal deletion break reading frames resulting from frame shifts are drawn as stripped boxes. point in patient 9 excluded CDKN1B (Figure 1). The minimal Patient 6 has two predicted proteins corresponding to each one of the deleted overlapping region extended from rs252028 (position mutated alleles. ETV6, ETS translocation-variant gene 6. 9814 kb) to rs747726 (position 12 561 kb). Together with the studies by Baens et al.2 andLaStarzaet al.,3 with proximal break point at part of the process resulting in AML.9 Adding to this, many d12s358 (a marker close to rs747726), these are the most telomeric translocations in AML involving break points outside the ETV6 break points, including ETV6 ever reported. Both studies set the locus are accompanied by cryptic deletions that include ETV6,3 distal break point at the PRB3 gene, delimiting a region of 1200 kb suggesting targeting of this gene. This is the case of patient 9 for (Figure 1c).2,3 Because inactivation of the second allele of ETV6,or whom FISH analysis showed that the deletion break points another gene, in the deleted region was never found, haplo- coincided with the translocation break point linking these two insufficiency of ETV6 has been suggested as an AML mechanism.4 events (Figure 2a; Supplementary Figure 1). The t(4;12)(q12;p13) translocation, although rare, is a recurrent We amplified and sequenced all 8 exons of ETV6 using event in AML, particularly in subtype M0.6,8 Previous studies genomic DNA of the 52 patients. Mutations in ETV6 have only have mapped the break points of this translocation within ETV6 been reported previously in five AML patients (M1 and M2) and and three regions in chromosome 4: CHIC2, HSG2 and an area in prostate cancer.5 We found insertions in patients 6 and 58, between these two genes.6,8 Cools et al.6 reported four cases and a point mutation in patient 21 (Figure 3; Table 1). Mutations where the t(4;12)(q12;p13) fused the first three exons in CHIC2 in patient 6 were biallelic, including a frameshift mutation in on 4q12 to exons 2–8 of the ETV6 gene in 12p13, resulting in the one allele, resulting in a truncated protein, and an insertion in expression of a hybrid CHIC2-ETV6 transcript in all cases. In our the remaining allele (Figure 3; Table 1). ETV6-mutant proteins cohort, patients 2 and 43 showed t(4;12)(q12;p13) (Table 1). In with truncation and insertions comparable to ours (Table 1; patient 2, FISH showed splitting signals of bacterial artificial Figure 3) were shown to have impaired transcriptional repres- chromosome (BAC) probes RP11-367N1 and of pooled RP11- sion activity.5 Interestingly, all our mutants resulted in loss of the 96B19 and RP11-418C2 probes corresponding to the positions of ETS domain or its binding activity, but not of the pointed (PNT) CHIC2 and ETV6, respectively (Figure 2b). Unfortunately, domain (Figure 3). Similar mutants reported by Barjesteh van metaphases for patient 43 were not available for FISH analysis. Waalwijk van Doorn-Khosrovani et al.5 showed a dominant- We screened both patients for CHIC2-ETV6 transcripts using the negative effect when ETV6 wild-type constructs were co- same approach described by Cools et al.6 In both patients, and in transfected with higher amounts of mutant construct. However, contrast with that study, we could only detect an ETV6-CHIC2 it is possible that the dominant-negative effect is due to the fusion transcript that consisted of the first exon of ETV6 and exons overexpression of mutant ETV6 protein. In fact, expression of 2–6 of CHIC2 (Figure 2c). The fusion transcript, mainly consisting mutated ETV6 protein was not detected in the patients in the of exon 1 of ETV6, encodes a very short out-of-frame protein Barjesteh van Waalwijk van Doorn-Khosrovani et al.5 study, (Figures 2c and d). We repeatedly failed to detect the reciprocal arguing against a dominant-negative effect. CHIC2-ETV6 transcript in both patients, which, if present, would In conclusion, the number and variety of ETV6 translocations consist of the first exon of CHIC2 and exons 2–8 of ETV6, not resulting in a fusion protein, together with loss of ETV6 by producing again an out-of-frame fusion transcript. In addition, we deletion and heterozygous or homozygous mutations, makes a did not detect loss of the remaining allele in any of the patients mutually supportive and compelling case for loss and haplo- (Figure 1). This finding is in line with the absence of the insufficiency of ETV6 as a leukemogenic step in AML-M0. transforming ability of the CHIC2-ETV6 protein described by Cools et al.,8 and further supports the idea that the fusion protein is not the element of pathogenesis. The heterogeneity of the Acknowledgements t(4;12) adds to this view.1,8 Several AML studies reported different translocations involving ETV6, but lacking functional fusion We thank H Wessels for support with karyotyping. Patient proteins. In some cases, it has been shown that ectopic material was kindly provided by PJ Valk, Erasmus University overexpression of proto-oncogenes at the partner chromosomes Medical Center; W-D Ludwig, Medical University of Berlin; WAF might be the malignant event.1 Still in many other cases, neither a Marijt, Leiden University Medical Center and WR Sperr, Medical functional fusion protein nor an alternative malignant mechanism University of Vienna. This work was supported by grants from the was found.1 This suggests that heterozygous disruption of ETV6 Calouste Gulbenkian Foundation and the Foundation for Science by the translocation results in ETV6 haplo-insufficiency and is and Technology (Portugal) to FPG Silva.

Leukemia Letters to the Editor 1643 1 1 2 2 2,3 FPG Silva , B Morolli , CT Storlazzi , A Zagaria , L Impera , 3 La Starza R, Stella M, Testoni N, Di Bona E, Ciolli S, Marynen P 1 1 4,5 B Klein , H Vrieling , HC Kluin-Nelemans et al. Characterization of 12p molecular events outside ETV6 in and M Giphart-Gassler1 complex karyotypes of acute myeloid malignancies. Br J Haematol 1Department of Toxicogenetics, Leiden University Medical 1999; 107: 340–346. Center, Leiden, The Netherlands; 4 Sato Y, Suto Y, Pietenpol J, Golub TR, Gilliland DG, Davis EM et al. 2Department of Genetics and Microbiology, University of TEL and KIP1 define the smallest region of deletions on 12p13 in Bari, Bari, Italy; hematopoietic malignancies. Blood 1995; 86: 1525–1533. 3Department of Hematology, University of Bari, Bari, Italy; 5 Barjesteh van Waalwijk van Doorn-Khosrovani S, Spensberger D, 4Department of Hematology, University Medical Center de Knegt Y, Tang M, Lowenberg B, Delwel R. Somatic heterozygous Groningen, Groningen, The Netherlands and mutations in ETV6 (TEL) and frequent absence of ETV6 protein in 5Department of Hematology, Leiden University Medical acute myeloid leukemia. Oncogene 2005; 24: 4129–4137. Center, Leiden, The Netherlands 6 Cools J, Bilhou-Nabera C, Wlodarska I, Cabrol C, Talmant P, Bernard P et al. Fusion of a novel gene, BTL, to ETV6 in acute myeloid E-mail: [email protected] leukemias with a t(4;12)(q11–q12;p13). Blood 1999; 94: 1820–1824. 7 Silva FP, Morolli B, Storlazzi CT, Anelli L, Wessels H, Bezrookove V et al. Identification of RUNX1/AML1 as a classical tumor suppressor References gene. Oncogene 2003; 22: 538–547. 8 Cools J, Mentens N, Odero MD, Peeters P, Wlodarska I, Delforge M 1 Bohlander SK. ETV6: a versatile player in leukemogenesis. Semin et al. Evidence for position effects as a variant ETV6-mediated Cancer Biol 2005; 15: 162–174. leukemogenic mechanism in myeloid leukemias with a t(4;12) 2 Baens M, Wlodarska I, Corveleyn A, Hoornaert I, Hagemeijer A, (q11–q12;p13) or t(5;12)(q31;p13). Blood 2002; 99: 1776–1784. Marynen P. A physical, transcript, and deletion map of chromosome 9 Yagasaki F, Jinnai I, Yoshida S, Yokoyama Y, Matsuda A, Kusumoto S region 12p12.3 flanked by ETV6 and CDKN1B: hypermethylation of et al. Fusion of TEL/ETV6 to a novel ACS2 in myelodysplastic the LRP6 CpG island in two leukemia patients with hemizygous syndrome and acute myelogenous leukemia with t(5;12)(q31;p13). del(12p). Genomics 1999; 56: 40–50. Genes Chromosomes Cancer 1999; 26: 192–202.

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

Low frequency of the glutathione-S-transferase T1-null genotype in patients with primary myelodysplastic syndrome and 5q deletion

Leukemia (2008) 22, 1643–1646; doi:10.1038/leu.2008.35; March 1999 and December 2004 (212 males, 111 females, published online 28 February 2008 range 19–92 years, median age 68 years). Diagnosis of MDS was established in the major Greek hospitals, according to the French American British classification criteria. Patients with Myelodysplastic syndromes (MDS) comprise a heterogeneous MDS and history of previous cancer diagnosis or with history group of acquired clonal hematopoietic stem-cell disorders, of chemo/radiotherapy were excluded from the study. Bone characterized by ineffective hematopoiesis and a variable risk of marrow samples at diagnosis were analyzed at the Cytogenetics transformation to acute myeloid leukemia. The etiology of MDS Unit, NCSR ‘Demokritos’, for centralized karyotyping. From the is unknown in the majority of cases; however, models for same specimens, genomic DNA was extracted and used for development of sporadic MDS suggest the role of cumulative subsequent genotypic analysis. Peripheral blood samples from genetic and toxic environmental factors in genetically predis- 330 healthy donors were used as control samples. They were posed individuals.1 matched for sex and age (205 males, 125 females) (range: 20–89 Human glutathione-S-transferases (GSTs) are a multigene years, median 65 years). These individuals were not related to family of phase-II cytosolic enzymes that mediate conjugation the patients and had no past history of any malignancy. of electrophilic compounds to glutathione. They play an Informed consent was obtained from patients and controls at important role in detoxification of numerous environmental each participating center, in compliance with the recommenda- carcinogens. The genes coding for the isoforms glutathione-S- tions of the Declaration of Helsinki. The study protocol was transferase-theta-1 (GSTT1) and -mu-1, (GSTM1) exhibit an approved by the ethical committee of each center participating inherited homozygous deletion polymorphism (null genotype), in the trial. resulting in total lack of enzymatic activity.2 The frequency of Cytogenetic analysis of bone marrow samples was performed homozygous deletion genotypes of GSTT1 and GSTM1 varies by using unstimulated short-term cultures. Karyotypes were des- ethnicity, being 15–30% and 35–60% respectively, among cribed in accordance with the recommendations of the Caucasians.3 Several studies investigating the association International System for Human Cytogenetic Nomenclature and potential causative role of GSTs polymorphisms in MDS (ISCN 2005). Patients were stratified according to the Interna- have come to conflicting conclusions, possibly due to the tional Prognostic Scoring System (IPSS) cytogenetic classifica- heterogeneity of these diseases, ethnic variations and/or tion into three risk groups, good: normal, del(5q) only, del(20q) methodological approaches.4 only, ÀY only; intermediate: þ 8, single miscellaneous, double To investigate the association and potential role of GST poly- abnormalities and poor: complex (that is, X3 anomalies) or morphisms in MDS pathogenesis within a group of individuals chromosome 7 abnormalities.5 To verify 5q31 deletion, double- with highly homogeneous ethnic background, we studied the color fluorescent in situ hybridization studies were performed GSTM1 and GSTT1 genotypes on a large cohort of Greek for EGR1 (5q31) and D5S23/D5S721 (5p15.2) (Vysis Inc., patients with primary MDS and healthy individuals. The study Downers Grove, IL, USA) in a total of 47 MDS cases with included 323 patients diagnosed with primary MDS between À5/del(5q).

Leukemia