Leukemia (2011), 1–13
& 2011 Macmillan Publishers Limited All rights reserved 0887-6924/11
www.nature.com/leu
LEADING ARTICLE
Comprehensive array CGH of normal karyotype myelodysplastic syndromes
reveals hidden recurrent and individual genomic copy number alterations with
prognostic relevance
A Thiel1, M Beier1, D Ingenhag1, K Servan1, M Hein1, V Moeller1, B Betz1, B Hildebrandt1, C Evers1,3, U Germing2
and B Royer-Pokora1
1
Institute of Human Genetics and Anthropology, Medical Faculty, Heinrich Heine University, Duesseldorf, Germany and
Department of Hematology, Oncology and Clinical Immunology, Heinrich Heine University, Duesseldorf, Germany
2
About 40% of patients with myelodysplastic syndromes (MDSs)
present with a normal karyotype, and they are facing different
courses of disease. To advance the biological understanding
and to find molecular prognostic markers, we performed a highresolution oligonucleotide array study of 107 MDS patients
(French American British) with a normal karyotype and clinical
follow-up through the Duesseldorf MDS registry. Recurrent
hidden deletions overlapping with known cytogenetic aberrations or sites of known tumor-associated genes were identified in 4q24 (TET2, 2x), 5q31.2 (2x), 7q22.1 (3x) and 21q22.12
(RUNX1, 2x). One patient with a 7q22.1 deletion had an
additional 5q31.2 deletion of the acute myeloid leukemia/MDS
region, the smallest deletion identified so far and including
the putative tumor suppressor (ts) genes, EGR1 and CTNNA1.
One TET2 deletion was homozygous and one heterozygous,
with a missense mutation in the remaining allele, further
supporting its role as a ts gene. Besides these recurrent
alterations, additional individual imbalances were found in 34
cases; in total, 42/107 (39%) cases had genomic imbalances.
These patients had an inferior survival as compared with
the rest of the patients (P ¼ 0.002). This study emphasizes the
heterogeneity of MDS, but points to interesting genes that may
have diagnostic and prognostic impact.
Leukemia advance online publication, 28 January 2011;
doi:10.1038/leu.2010.293
Keywords: myelodysplastic syndromes; aCGH; normal karyotype;
prognosis
Introduction
The myelodysplastic syndromes (MDSs) are clonal disorders of
the hematopoietic system, with morphologic dysplasia, ineffective hematopoiesis and peripheral blood cytopenias. MDS is
currently classified according to the WHO (World Health
Organization), which is based on the morphological, cytogenetic and hematological criteria and defines six subtypes.1
In addition, the French American British classification, which
comprises five subtypes, is also still in use.2 The International
Prognostic Scoring System (IPSS), a risk stratification system that
takes into account clinical parameters as well as cytogenetic
results, assigns four risk categories for death or transformation
to acute myeloid leukemia (AML) (low, Int-1, Int-2 and high).
Specific cytogenetic abnormalities were also included in
the updated prognostic classification system.3 Approximately
Correspondence: Dr B Royer-Pokora, Institute of Human Genetics
and Anthropology, Medical Faculty, Heinrich Heine University,
Moorenstr. 5, 40225 Duesseldorf, Germany.
E-mail: royer@uni-duesseldorf.de
3
Current address: Institute of Human Genetics, Heidelberg University,
Heidelberg, Germany.
Received 26 October 2010; accepted 8 November 2010
40–50% of MDS cases have a normal karyotype. MDS patients
with a normal karyotype and low-risk clinical parameters are
often assigned into the IPSS low and intermediate-1 risk groups.
In the absence of genetic or biological markers, prognostic
stratification of these patients is difficult. To better prognosticate
these patients, new parameters to identify patients at higher risk
are urgently needed. With the more recently introduced modern
technologies of whole-genome-wide surveys of genetic aberrations, it is hoped that more insights into the biology of disease
progression might be obtained.
Array comparative genomic hybridization (aCGH) studies
with bacterial artificial chromosomes (BACs) or oligonucleotides
using exclusively MDS samples are currently rare. The first
BAC-based aCGH study of MDS and AML patients with trisomy
8 as sole cytogenetic aberration demonstrated that additional
duplications and homozygous deletions can be detected, and
that trisomy 8 is not always the primary genetic event.4 Another
BAC aCGH analysis of 38 MDS patients, 19 with and 19 without
cytogenetic abnormalities, compared with a normal agematched control group revealed many cryptic aberrations,
which were confirmed by other methods.5 A BAC aCGH study
using CD34 þ cells from 44 low-risk MDS patients (IPSS p1)
included 25 cytogenetically normal cases.6 This aCGH analysis
uncovered hidden aberrations in the cytogenetically normal
cases, but confirmed only 11 of 16 cytogenetically visible
aberrations. These authors divided the patients into two groups
based on the presence of a total of p3 Mb genomic aberrations
(TGA) or 43 Mb TGA. It is not surprising that the group of
43 Mb TGA had a significant shorter overall survival, as most
cases with large cytogenetically visible alterations fell into this
group. A separate analysis of the 25 cytogenetically normal
cases also showed a significant difference in overall survival in
the group with 43 Mb TGA, although the numbers are still
small.6 Using oligonucleotide aCGH on MDS/AML with a
5q-syndrome or del(5) as the sole cytogenetic abnormality, we
have described the presence of additional hidden aberrations in
2/12 patients.7 With single-nucleotide polymorphism (SNP)
arrays, copy number alterations as well as loss of heterozygosity
through uniparental disomy (UPD) can be detected. Two reports
described the use of SNP arrays to study low-risk MDS for
hidden genomic imbalances. One SNP array study of 119 lowrisk MDS patients, 77 with a normal karyotype, 36 with visible
cytogenetic alterations and 6 with unavailable cytogenetics,
identified 125 UPDs with a median size of 3.78 Mb in 46% of
the cases studied.8 Another study showed that 82% of MDS
patients harbor gains and losses detected by SNP arrays,
whereas conventional cytogenetics detected alterations
in 50%.9 In addition, UPD was found in 33% of the patients.
In both studies, first correlations of genomic imbalances
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
2
including UPD with survival indicated that the presence of these
may correlate with a shorter survival.8,9 This high rate of UPDs
in MDS was recently challenged in a study that showed, when
using buccal DNA samples from the same patients as controls,
that only a small number of the potential loss of heterozygosity
regions represent tumor-specific UPDs and gains and losses.10 In
that report only 4/33 cytogenetically normal cases had
additional alterations.
To further explore the frequency of hidden genomic
imbalances, specifically recurrent abnormalities, we performed
a large retrospective study on 107 karyotypically normal MDS/
AML patients (WHO) using long oligonucleotides aCGH and
non-amplified DNA. Recurrent and individual imbalances
were detected, which correlated with an inferior survival.
Germany) or Marrow Max Medium (Invitrogen, Gibco,
Karlsruhe, Germany) with 10% fetal calf serum (Cytogen, Sinn,
Germany). In total, 20–25 mitoses were analyzed according to
the ISCN (International System for Human Cytogenetic Nomenclature), with a resolution of 100–300 bands. All 107 cases
had a successful cytogenetic analysis. Material left over after
cytogenetics was stored in methanol–acetic acid (3:1) at 20 1C.
DNA was isolated from archived bone marrow samples with the
QIAmp DNA Blood Midi Kit (Qiagen, Hilden, Germany) or with
a phenol chloroform extraction. DNA concentration and purity
were determined with the NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, NC, USA).
Oligonucleotide aCGH
Materials and methods
Patients, cytogenetic analysis and DNA isolation
In total, 107 cytogenetically normal MDS patients at the time of
diagnosis and before therapy were included in this retrospective
study. The patients had been entered into the Duesseldorf
MDS registry and followed for survival. Median follow-up time
was 28 months (1–150). Median survival time of the entire
cohort was 32 months (1–150). The clinical details of the
patients are shown in Table 1. Approval for the study was
obtained from the local ethics committee.
For conventional G-banding, bone marrow cells were
cultivated for 24, 48 or 72 h in RPMI 1640 (Biochrom, Berlin,
Table 1
Data analysis
Clinical characteristics of the patient cohort (n ¼ 107)
No.
%
107
100
64
43
60
40
65
FAB classification
RA
RARS
RAEB
RAEB-T
59
0
30
18
55
28
17
IPSS
0 Low
0.5–1.0 Int-1
1.5–2.0 Int-2
42.0 High
Unknown
34
35
23
14
1
32
33
21
13
1
WHO classification
RCUD
RCMD
RCMD-RS
RAEB-1
RAEB-2
AML
1
44
14
13
17
18
1
41
13
12
16
17
Cytogenetics
Normal karyotype
Sex
Male
Female
Median age
Abbreviations: AML, acute myeloid leukemia; FAB, French American
British; Int, intermediate; IPSS, International Prognostic Scoring
System; RA, refractory anemia; RAEB, refractory anemia with excess
blasts; RAEB-T, refractory anemia with excess blasts in transformation;
RARS, refractory anemia with ringed sideroblasts; RCMD, refractory
cytopenia with multilineage dysplasia; RCMD-RS, refractory cytopenia
with multilineage dysplasia and ringed sideroblasts; RCUD, refractory
cytopenia with unilineage dysplasia; WHO, World Health Organization.
Leukemia
For aCGH, 44k (n ¼ 37), 105k (n ¼ 21) or 244k (n ¼ 49) 60mer
oligonucleotide microarrays were used (Human Genome CGH
Microarray, Agilent Technologies, Palo Alto, CA, USA). CGH
analyses were performed essentially as described in the protocol
of the manufacturer (Protocol Version 5.0, June 2007; Agilent).
Same-gender reference DNAs were purchased as pools (Promega,
Mannheim, Germany). In total, 1.5 mg of cyanine 5-dUTPlabeled test DNA and cyanine 3-dUTP-labeled reference DNA
were hybridized in the presence of Cot-1-DNA (Invitrogen) for
40 h (244, 105k) and 24 h (44k) at 65 1C. Arrays were washed
(Wash Procedure B, Agilent) and slides were scanned with the
Agilent array scanner G2505B.
Microarray images were scanned and Agilent’s Feature Extraction Software Version 10.1 was used for quantification. Aberrant
regions were determined by automatic breakpoint detection
using the GLAD library11 within the R Statistical Environment
(http://www.r-project.org). All aberrations detected by two
algorithms available in GLAD (smooth functions ‘lawsglad’
and ‘haarseg’), covered by at least three oligonucleotides and
with a minimum log2 ratio of ±0.25, were mapped against the
Database of Genomic Variants (http://projects.tcag.ca/variation)
and UCSC (http://genome.ucsc.edu). In addition, for comparison
we used data obtained from the analyses of mentally retarded
patients in our diagnostic laboratory using the same arrays and
the same evaluation parameters (‘controls’). Imbalances were
eliminated when they overlapped by 80% with those found in
MDS patients. Regions overlapping o80% with known variants
were selected for further visual inspection with Agilent’s
Genomic Workbench Software (Version 5.0.14) to define a set
of aberrations for verification with fluorescence in situ hybridization (FISH), quantitative genomic real-time PCR (qPCR)
or custom arrays. The complete data set is available
at http://www.ncbi.nlm.nih.gov/geo/info/linking.html (accession
no. GSE24602).
Validation of imbalances
Larger deletions and gains of several Mb were verified by FISH.
BACs or fosmids were chosen from UCSC and purchased as
described in Supplementary Table S1. BAC DNA was isolated
with the Qiagen Plasmid Midi Kit and indirectly labeled using
either the BioNick DNA Labeling System (Invitrogen) and
detected with streptavidin-cyanine 3 (Dianova, Hamburg,
Germany) or the DIG-Nick Translation Mix and detected with
anti-DIG-fluorescein (Roche, Mannheim, Germany). At least
200 interphases (and metaphases) were analyzed using a Zeiss
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
3
To determine the exact breakpoints of the deletions, we designed
high-resolution custom arrays covering the affected genomic
segments using the eArray internet platform (https://earray.chem.agilent.com/earray/) of Agilent. Several formats were used, with
an average genomic distance of 125 bp to 2 kb at the sites of the
putative breakpoints. The arrays also covered the rest of the
genome with a distance between 44 and 231 kb.
(Figure 1a). qPCR with specific primers in exon 2 of TET2
confirmed the homozygous deletion (Figure 1b). A custom array
was hybridized to identify the breakpoints. Long-range PCR
followed by sequencing showed that the deletion starts at
46.9 kb proximal and extends to 124.8 kb distal of TET2. This
patient had an additional 6.6-Mb deletion in 2p23.3–p24.1
(Supplementary Figure S1A), which was present in 96% of the
cells as determined by FISH. In a follow-up sample 5 months
later, the 2p deletion was reduced to 84%. The deletion in 2p
harbors several interesting genes involved in epigenetic
regulation (DNMT3A) and chromatin modification (ASXL2)
(Supplementary Figure S1B). The paralogous ASXL1 gene has
been described as frequently mutated in MDS.12 Interestingly,
TET2 is probably involved in the regulation of cytosine
methylation, pointing to two genes altered in this patient with
a function in epigenetic regulation. Besides these two aberrations, no other genomic imbalances were noticed in this case.
The patient had refractory anemia with excess blasts in
transformation (RAEB-T) (AML according to the WHO classification), an IPSS score of high and died 10 months after diagnosis
with AML.
A heterozygous deletion of B700 kb removing the PPA2,
EEF1AL and TET2 genes was identified in case P41 and was
verified by qPCR (Figures 1a and b). FISH determined that it was
present in 67% of the cells. To test for mutations of the
remaining alleles, all exons were analyzed and a missense
mutation, p.C1135Y, in exon 3a of TET2 was detected. The
same missense mutation was recently described as a somatic
mutation in a patient with Myeloproliferative Neoplasm.13 In
this case several additional small aberrations were present,
which were not studied further (Table 2). The patient had lowrisk refractory anemia with multilineage dysplasia and ringed
sideroblasts (IPSS 0), but survived only 3 months.
Identification of breakpoints
5q31. Two cases had interstitial chromosome 5q deletions,
To determine the deletion end points, specific primer pairs were
designed according to the results of the Agilent workbench data
from the custom arrays. Long-range PCR conditions were used
(Expand 20 kbPlus PCR System, Roche). DNA fragments were
extracted from the gel (MEGA-spin Agarose Gel Extraction Kit,
Intron Biotechnology, Gyeonggi-do, Korea), purified with
ExoSAP-IT (Affymetrix, USB, Santa Clara, CA, USA) and
sequenced.
one 17 Mb from 5q23.2 to 5q31.3 (P52) and the other 1.35 Mb
in band 5q31.2 (P100) (Figure 2a). Both deletions were
confirmed by FISH and were present in 45 and 72% of the
cells, respectively (Figure 2b). qPCR of HSPA9 exon 2 confirmed
the heterozygous deletion in both cases (Figure 2c). The larger
deletion in P52 starts at the position of the ZNF608 gene (124 Mb)
and extends to the FGF1 gene (141 Mb), including both the 5qsyndrome region and the more proximal AML/MDS region. No
other additional aberrations were seen in P52, whereas patient
P100 had an additional deletion in 7q22 (see below).
The 5q31 deletion in patient P100 is to our knowledge the
smallest 5q AML region deletion described so far. The deletion
end points were identified by hybridization of a custom array.
Sequences spanning the deleted region were amplified by longrange PCR and extensive sequence homology between the two
breakpoints was observed; the break occurred on the proximal
side in an AluJo and on the distal side in an AluSp element
(manuscript in preparation). On the proximal side the MYOT
gene is disrupted and 23 known genes map to this segment
(Figure 2d). This patient (P100) with the small 5q31 and 7q22.1
deletion was diagnosed with refractory anemia with multilineage dysplasia, IPSS Int-1 and died after 12 months. In
contrast, the patient (P52) with the larger 5q deletion and no
other alteration was diagnosed as having refractory anemia with
multilineage dysplasia and ringed sideroblasts, IPSS 1 and
survived 43 years.
Axioplan microscope (Göttingen, Germany) and ISIS software
(MetaSystems, Altlussheim, Germany).
Smaller deletions and gains were studied with qPCR using
FastStart Universal SYBR Green Master (Rox) (Roche) and an ABI
PRISM 7900 (Applied Biosystems, Carlsbad, CA, USA). Specific
primer pairs were designed for each region. PCRs were run in
triplicates and repeated once. For relative quantification, the
delta delta Ct (threshold cycle) method was applied. P-values for
fold changes were computed with two-sample t-tests, and
corrected for multiple testing across qPCR experiments using the
Bonferroni–Holm method. A combined male and female DNA
pool (Promega), besides one or two normal DNA control
samples, served as calibrator. As reference PRNP, a single copy
gene was used.
TET2 exon amplification and mutation analysis
For mutation analysis of the coding sequence of TET2
(NM_001127208.1), all exons were amplified from sample
P41 with specific primer pairs, mixed with normal DNA and
analyzed by denaturing high-performance liquid chromatography (Supplementary Table S2A, B). Fragments with aberrant
migration were subsequently sequenced.
Custom arrays
Results
We investigated 107 MDS samples of various French American
British subgroups at the time of diagnosis. Larger aberrations
detected in X2 cases were studied first. Recurrent aberrations
were identified in 4q (n ¼ 2), 5q (n ¼ 2), 7q (n ¼ 3) and 21q
(n ¼ 2), and individual genomic imbalances occurred in an
additional 34 cases. Interestingly, the recurrent deletions
occurred on chromosomes 5 and 7 known to be frequently
cytogenetically altered in MDS. The other two recurrent
alterations affected two known tumor-associated genes, TET2
and RUNX1. All verified aberrations (n ¼ 40) and those fulfilling
the above-described criteria but without verification (n ¼ 56) are
listed in Table 2.
Recurrent aberrations
4q24. Two cases had small deletions in 4q24 at the site of the
TET2 gene. In P40, aCGH uncovered a homozygous deletion of
240 kb, removing the entire TET2 gene and flanking sequences
7q22. A 7q22.1 deletion with similar end points was found in
three cases, P15, P70 and P100, with sizes of 2.1, 2.2 and
Leukemia
Leukemia
2q22.2
1p33
2p21
6q13
12q24.22
14q23.2
15q21.3
Xp22.31
15q25.1
21q22.3
1q21.3
2q22.3
3p12.3–p12.2
6p22.1
7q22.2
11q14.2
12q24.33
15q21.2
15q21.2
21q11.2
2p23.1–p22.3
8q22.1
18q23
21q22.12
2p23.3–p24.1
4q24
1q41
4q24
6q13
7q22.2
25
29
30
35
41
39
40
37
38
32
33
34
24
10q21.3
8p21.3
7q21.2
7p14.1
7q22.1
14q13.1–q13.2
15q15.2
1q23.1
1q42.3
2q22.3
2q33.1
3q13.12
4p14
4q31.21
4q32.3
10q11.23–q21.2
11p15.4
11q14.2
14q24.2-q31.1
15q22.2
21q11.2
21q22.12
Chromosome
band
151
144
81
27
105
85
131
47
49
14
31
95
72
35
20
106
218
106
73
105
143
49
46
73
116
63
53
8
76
43
585
852
614
944
674
394
336
991
532
771
736
843
863
084
750
268
280
012
978
674
325
860
915
951
130
018
288
553
303
497
399–151 692 068
134–144 991 698
124–81 889 568
581–27 969 064
362–105 716 890
758–85 447 234
969–131 680 161
272–48 094 388
798–49 652 463
985–14 844 050
202–32 851 786
201–95 969 322
881–72 935 992
180–35 413 600
003–27 373 948
743–106 510 206
255–218 361 910
352–106 712 569
071–74 415 198
362–105 716 890
206–143 355 532
361–50 128 060
926–47 027 643
608–74 406 374
414–116 210 234
158–63 373 454
498–53 329 341
116–8 650 148
328–76 357 394
808–44 303 762
69 398 173–70 729 470
19 625 016–19 760 238
91 570 132–91 648 830
39 957 570–40 265 361
99 812 708–101 895 949
33 820 105–34 786 814
41 199 376–41 346 176
156 900 013–156 930 155
233 921 830–234 089 020
144 852 134–144 991 698
201 422 144–201 439 344
109 272 371–109 284 363
35 841 401–35 919 374
143 199 593–144 012 507
169 376 855–169 619 992
53 000 672–63 148 317
5 198 938–5 220 565
85 394 758–85 465 274
72 158 927–92 488 619
60 007 424–60 104 511
14 771 985–14 844 050
35 201 032–35 911 372
Max.
range (bp)
106.7
139.6
275.4
24.5
42.5
52.5
343.2
103.1
119.7
72.1
1115.6
126.0
72.0
329.0
6624.0
241.0
81.7
700.0
437.1
42.5
30.3
268.0
112.0
454.8
79.8
355.3
41.0
97.0
54.1
806.0
1331.3
135.2
79.0
307.8
2083.0
966.7
147.0
30.1
167.2
139.6
17.2
12.0
78.0
812.9
243.1
10 148.0
21.6
70.5
20 330.0
97.1
72.1
710.0
Max.
size (kb)
Summary of all detected aberrations in 42 patients
1
8
11
14
15
19
20
21
Patient
Table 2
0.529
0.629
0.44
0.724
1.31
0.802
0.408
0.668
1.39
1.19
0.344
0.491
0.669
1.03
0.685
2.92
0.439
0.513
0.283
0.74
1.21
1.05
1.04
0.295
0.444
0.315
1.01
0.76
0.601
0.5
0.44
1.2
0.459
0.308
0.526
0.313
0.803
1.17
0.852
0.486
1.39
1.09
1.11
0.278
0.406
0.26
1.9
1.21
0.267
0.64
0.992
0.7
Log2
ratio
ND
ND
ND
ND
40%
40%
ND
ND
67%
Local gain
100%
94%
96%
ND
ND
ND
ND
100% 5mo
84% 5mo
ND
ND
ND
ND
ND
ND
ND
ND
ND
22,2% 1mo
0% 10mo
0% 12mo
ND
73%
89%
ND
ND
Follow-up
FISH (months)
ND
ND
FISH
0,6 TET2
ND
ND
0,61 RUNX1
ND
0,14 TET2
1,7 RRP1B
0,45 RAB27A
ND
0,48 AGBL4
0,61 MCFD2
ND
ND
ND
EPB24 (mPCR)
ND
0,4 INTS10
ND
qPCR
ND
Yes
Yes
Yes
Yes
Yes
ND
Yes
Yes
ND
Yes
Yes
ND
ND
Yes
Yes
ND
Yes
Custom
array
G
G
G
G
G
G
G
G
G
G
L
G
L
L
L
L
L
L
L
G
G
L
L
L
G
L
L
G
L
G
L
L
G
L
L
L
L
G
G
G
G
G
G
G
G
L
G
G
L
G
G
L
Gain/
loss
NAMPT, PBEF1
TET2
RAB27A gene disrupted by loss
KAL1
DNAJ
SIK1, HSF2BP, RRP1B, PDXK, CSTB, RRP1, AGPAT3, TRAPPC10
(disrupted)
S100 gene cluster
ZEB2
GBE, glycon storage disease
Histone gene cluster
NAMPT, PBEF1
PICALM
GALNT6, NOC4
ATPase
DNAJ
SAMSN1
BIRC6
DRY19L4 (disrupted), INTS8 gain, CCNE2 (disrupted)
MBP, short isoform not disrupted, longer isoform yes
RUNX1
450
TET2
NOS1 gene disrupted by gain
KYNU
AGBL4
MCFD2 entire gene deleted, TTC7A (disrupted)
SAMSN1
RUNX1
TMEM62, CCNDBP1, EPB42, TGM5 all loss, TGM7 (disrupted)
SPTA1
LYST gene disrupted
ZEB2
CLK1
CD47
ARAP2
INPP4B
DDX60
Many
HBD
PICALM (AF10 translocation)
Many
46 genes
TET1
INTS10
AKAP9 (disrupted), CYP51A1
Genes
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
4
73 969 589–74 492 261
33 858 060–34 934 616
137 351 108–138 706 114
99 702 890–101 901 100
176 643 265–176 757 841
27 092 400–27 230 818
43 971 072–44 075 643
33 050 558–33 386 018
169 562 135–170 462 974
22 196 826–22 375 740
6q13
14q13.1–q13.2
5q31.2
7q22.1
2q31.1
7p15.2
17q21.32
18q12.2
3q26
5p14.3
19p13.3
chr21
20p11.21
1p34.1
16q23.1
1p31.3
12p11.21
3p26.2
19p13.2–p13.3
124 049 584–141 931 520
150 232 979–150 334 042
1 320 156–1 349 054
163 124 412–164 113 937
164 245 352–166 260 703
68 968 733–69 033 774
7 620 638–7 885 748
99 897 691–102 120 062
85 253 938–113 809 774
38 792 056–39 153 477
1 034 068–1 051 282
17 243 205–17 557 331
4 347 076–4 361 506
44 537 902–44 550 014
63 180–12 511 914
4 187 605–9 071 610–
9 470 811–12 403 648–
13 398 901–18 843 862
1 380 880–1 404 671
1–46 944 322
25 127 250–25 345 765
45 602 357–45 745 433
76 737 994–76 816 871
63 457 391–63 897 984
31 379 504–33 035 815
3 971 534–4 939 661
5q23.2–q31.2
6q25.1
17p13.3
2q24.2–q24.3
6q26–q27
7q11.22
12p13.31
7q22.1
11q14.1–q23.2
12q12
7p22.3
9p22.2
19p13.3
21q22.3
8p23.1–p23.3
018–101 079 805
788–2 776 136
985–14 844 050
068–1 051 282
533–102 282 321
934–129 282 957
413–132 437 274
071–74 406 374
653–133 235 868
337–88 397 205
890
377
771
034
677
070
267
978
888
035
11q22.1
18p11.32
21q11.2
7p22.3
1p21.1–p21.2
4q28.1–q28.2
5q31.1
6q13
7q33
16q24.3
100
2
14
1
101
129
132
73
132
88
Max.
range (bp)
Chromosome
band
522.7
1076.6
1355.0
2198.0
114.6
138.4
104.6
335.5
901.0
178.9
23.8
46 944.0
219.0
143.0
79.0
441.0
1656.3
968.0
13 262.0
17 874.0
101.1
28.9
990.0
2015.0
65.0
265.1
2222.0
28.5
361.0
17.2
314.1
14.4
12.1
12 449.0
189.8
398.3
72.1
17.2
605.0
212.0
169.9
428.3
347.0
362.0
Max.
size (kb)
0.416
0.305
0.414
0.492
0.456
0.574
0.652
0.402
0.589
0.39
0.873
0.27
0.733
1.19
0.92
0.531
0.369
0.899
0.41
0.818
0.563
1.37
0.604
0.89
0.938
0.558
0.362
0.311
0.478
0.976
1.15
1.2
0.708
0.18
0.591
0.427
0.707
1.01
0.687
0.397
0.412
0.352
0.64
0.536
Log2
ratio
ND
ND
49%
ND
ND
ND
ND
100%
71.5%
87%
41% 4mo
Local gain
ND
ND
ND
41% 4mo
17%
Local gain
ND
ND
Local gain
17%
41% 4mo
ND
ND
ND
85%
64.3%
ND
17%
ND
ND
ND
ND
ND
ND
ND
ND
Local gain
ND
Local gain
48%
ND
Follow-up
FISH (months)
100%
FISH
ND
0,57 HSPA9
0,48 CUX1
ND
ND
2,03 PYGB
0,41 MMACHC
0,48 WWOX
2,16 PGM1
ND
ND
0,4 CNTLN
0,77 CUX1
ND
1,56 LRRK2
1,6 KCNH7
0,50 PDE10A
0,59 AUTS2
0,68 HSPA9
0,43 EXOC4
ND
ND
qPCR
ND
Yes
Yes
Yes
ND
Yes
Yes
Yes
Yes
ND
ND
ND
ND
ND
Yes
ND
Yes
Yes
Yes
ND
Yes
ND
Custom
array
L
L
L
L
G
G
G
G
G
G
G
G
G
L
L
G
L
L
G
L
G
G
G
L
L
L
L
L
G
G
L
G
G
L
L
G
G
G
L
L
L
L
L
G
Gain/
loss
Many
50 genes
HOXD gene cluster
HOXA gene cluster
HOXB gene cluster
BRUNOL4
EVI1
SUMF1 (CNV), SETMAR, ISUMF1 short, ITPR1 (not in CNV), EGO
completely deleted
Many
PYGB, ABHD12 gain entire genes
TESK2, MMACHC
WWOX, fra16q
FOXD3, ALG6, ITGB3BP, EFCAB7, DLEU2 L, PGM1 (disrupted)
Many
4170 genes
47 genes
160 genes
LRRK2 gain entire gene
C7ORF50
CNTLN
MYO1C, INPP5K
KCNH7 (disrupted by gain)
PDE10A deleted
AUTS2
EXOC4, larger isoform disrupted
ANKRD11, SPG7, RPL13, CNE7 (CNV), DPEP1, CHMP1A, CDK10,
SPATA2 L, ZNF276, FANCA (disrupted)
Many
OLF3
MFSD8, neuronal, lipofuscinosis
AF4, HSPA4
TRPC6, gene disrupted, focal segmental glomerulosclerosis
NDC80, SMCHD1
SAMSN1
Genes
Abbreviations: CNV, copy number variant; FISH, fluorescence in situ hybridization; G, gain; L, loss; max., maximal; mo, months; mPCR, multiplex PCR; ND, not determined; negative log2 ratio, loss;
positive log2 ratio, gain.
106
104
100
91
94
98
81
90
79
75
72
74
53
58
60
63
70
52
50
47
49
Patient
Table 2 (Continued )
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
5
Leukemia
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
6
Figure 1 A homozygous and a heterozygous deletion of TET2 (4q24) in cases P40 and P41. (a) Deletions as observed with the Agilent workbench
program (setting: ADM-2, threshold 6, gene view). Left P40 with a homozygous deletion and right P41 with a larger heterozygous deletion.
(b) qPCR of the TET2 gene, demonstrating the homozygous deletion in case P40. Residual normal cells in the sample resulted in the background
relative copy number of 0.14. Case P41 has a relative copy number of 0.6, which shows that the deletion is not present in every cell. This result is
in agreement with the deletion seen in 67% of the cells with FISH. Error bars are 95% confidence intervals, with an asterisk (*) indicating
statistically significant fold changes (Pp0.05).
2.2 Mb, respectively (Figure 3a). FISH confirmed the deletion in
73, 85 and 87% of the cells (Figure 3b). In addition, qPCR with
intron 1/exon 2 primers of the CUX1 gene showed a relative
copy number of 0.48 (P100) and 0.77 (P70) vs the pool
calibrator (Figure 3c). The deleted segment contains 50 named
genes (Figure 3a). Segmental duplications and interspersed
repeats flanking this region hampered the molecular identification of the breakpoints.
Patient P70 had an additional interstitial 28-Mb deletion of
11q (Supplementary Figure S2), which was seen by conventional cytogenetics in one metaphase and was therefore not
included in the karyotype. Several interesting putative tumor
suppressor (ts) genes map to this segment, for example ATM
and MRE11, as well as genes involved in apoptosisFCASP1, 4,
12, BIRC2 and 3. This patient had RAEB-2 at diagnosis,
progressed to RAEB-T after 5 months and died 7 months after
diagnosis. The other patient, P15, with no additional imbalance,
had RAEB-1, IPSS Int-1 and died after 15 months from infection.
21q22. Another recurrent deletion occurred at RUNX1/AML1
in 21q22.12 in two cases (Figure 4a). Both were confirmed by
FISH and were present in 94% (P39) and 89% (P24) of the cells.
In a follow-up sample of case P24 after chemotherapy, the
number of aberrant cells was below the cutoff for the probe.
After hybridization of a custom array, the breakpoints could
be identified and PCR products spanning the deletion were
obtained (Figure 4b) and sequenced. The actual sizes of the
deletions were determined as 692 (P39) and 704 kb (P24). The
larger deletion size identified after sequencing in case P39 was
due to the lower resolution of the 44k array. In P39 the entire
RUNX1 gene was deleted and in P24 only exon 1 and the
50 upstream region (Figure 4c). Patient P24 with RAEB-T received
induction chemotherapy and autologous stem cell transplantation
Leukemia
and died 20 months later from septicemia. Patient P39 had no
other genomic imbalances and had RAEB-2, an IPSS score of 3
and died 10 months after diagnosis of unknown cause.
A complex case with loss of 8ptel, gain of interrupted
regions of 19p and gain of chromosome 21
In one case, P79, aCGH uncovered several imbalances, a
complex rearrangement of chromosome 19 with a gain
interrupted by two regions of a seemingly normal copy number,
a gain of the entire chromosome 21 and loss of 8p22 to pter
(Figure 5a). FISH analysis with RP11-348B12 (19p13) revealed
amplification in 17% of the cells with three to seven signals,
suggesting an unstable karyotype. Both probes, 21q22 and
19p13, hybridized to four chromosomes, most likely representing an unbalanced translocation. The metaphases in Figure 5b
show examples with various numbers of translocation chromosomes and three chromosomes with amplified signals. The 8p
telomere deletion was verified by FISH, and simultaneous
hybridization with probes from 19p13.3 and 8p23.1 showed
that 17% of the cells had both aberrations. A schematic drawing
of the identified alterations is shown in Figure 5b below the FISH
results. In conclusion, all three aberrations were present in the
same cell clone, demonstrating a complex cytogenetic result in
a subset of cells. Limited material prevented further analyses.
Although the cytogenetic analysis revealed a normal karyotype
following the ISCN rules, there were several single aberrations.
A follow-up sample after 4 months revealed cytogenetically
visible aberrations: 46,XY[12]/45,XY,del(5)(q14q22),add(13)
(p11),14,15,17,19,21, þ 4mar[10]. In the follow-up
sample, amplification of chromosome 21 and 19p13 was found
in 41% of the interphase cells and 12% of the metaphases,
demonstrating an increase in abnormal cells with cytogenetics
and FISH. In addition, these results also showed that in this
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
7
Figure 2 Submicroscopic deletions in 5q in P52 and P100. (a) The larger 5q23.2–q31.2 deletion in case P52 and the small deletion in case P100
in the chromosome view setting using Agilent genomic workbench (setting: ADM-2, threshold 6). The array format used is indicated below the
chromosome. (b) FISH analyses with RP11-143H23 confirmed the larger deletion in case P52 and with RP11-166J22 in case P100. Both probes
were cohybridized with BAC probes localized in 5p14. (c) qPCR of the HSPA9 gene confirmed the deletion in both cases, with relative copy
numbers of 0.68 and 0.57. Error bars are 95% confidence intervals, with an asterisk (*) indicating statistically significant fold changes (Pp0.05).
(d) Genes mapping to the small deletion in case P100 (UCSC), deduced from the sequence obtained from the PCR product spanning the deletion.
complex case the abnormal cells do not divide efficiently in
culture. This patient had RAEB-1, IPSS 1 and died 5 months after
diagnosis.
Individual additional aberrations
Besides the eight patients with the recurrent alterations, 34
patients had additional individual copy number aberrations. In
22 cases a single aberration was observed; the other patients had
between 2 and 14 aberrations. As no germ-line DNA was
available from any of the patients, we could not determine
whether these were constitutional. Most alterations were
verified by FISH, qPCR or custom arrays (Table 2). Additional
alterations identified by at least two algorithms but not verified
by other methods are also listed in Table 2.
Some of these alterations were present in a high proportion of
cells and therefore could be germ-line imbalances possibly
contributing to MDS development. However, it should be
pointed out that larger tumor-specific deletions were also
present in a high proportion of cells when analyzed with FISH,
for example the deletion 2p23.3–p24.1 in 96%. The same was
observed by us previously for the del(5q), which was found in
35–91% of interphase cells.7
It is remarkable that two patients had alterations affecting
genes that are associated with elliptocytosis. Patient P20 had a
deletion of 147 kb in 15q15 including the EPB42 (Supplementary Figure S3). Mutations in the EPB42 gene, coding for the
erythrocyte protein band 4.2, are associated with recessive
elliptocytosis and recessively transmitted hemolytic anemia.
Patient P21 had a gain in 1q23.1 involving the SPTA1 (alpha
spectrin, erythrocytic 1) gene and possibly disrupting it.
Mutations in this gene result in a variety of hereditary red blood
cell disorders, including elliptocytosis type 2, pyropoikilocytosis
and spherocytic hemolytic anemia. The same patient had a gain
in 1q42.3 disrupting the LYST gene; mutations in this gene are
associated with Chediak–Higashi syndrome, a rare lysosomal
storage disease. Features of this disease are large eosinophilic,
peroxidase-positive inclusion bodies in myeloblasts and promyelocytes of the bone marrow, neutropenia, and abnormal
susceptibility to infection and malignant lymphoma. Patients
Leukemia
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
8
Figure 3 Identification of three similar 7q22 deletions in P15, P70 and P100. (a) Approximate breakpoints of the deletions in the three cases based
on the GLAD algorithm (UCSC). Segmental duplications on either side are shown. (b) FISH analysis using a probe mapping to 7q22 in combination
with a centromere 7 probe confirmed the deletion in case P15. In case P100, two BAC probes mapping within the deletion confirmed the deletion,
as only one chromosome 7 is labeled. (c) qPCR analysis of two cases (P70 and P100) using primers from the CUX1 gene verified the deletion in
both cases with a relative copy number of 0.77 in P70 and 0.48 in P100. The deletion was found in 84 and 87% of the cells, respectively, by FISH.
Error bars are 95% confidence intervals, with an asterisk (*) indicating statistically significant fold changes (Pp0.05).
often die of hemorrhage, some secondary to thrombocytopenia.
Both alterations affect only one copy of these genes and the
associated diseases are recessive. Therefore, the observed
alterations alone do not cause disease, but a combination of
both (compound heterozygotes) may have contributed to MDS
development. In addition, the same patient had two larger
deletions in 14q and 10q in 40% of the cells as determined by
FISH (Supplementary Figure S4A, B), as well as 10 other smaller
alterations (Table 2). This patient had RAEB-2, an IPSS score
of 1, received allogeneic stem cell transplantation and died
3 years after diagnosis of unknown cause.
Another patient (P30) had three alterations verified by custom
arrays (Supplementary Figure S5), two of which could contribute
to MDS development: a deletion disrupting the RAB27 gene
(autosomal recessive mutations cause Griscelli’s syndrome
and hemophagocytic syndrome) and a loss of the entire
MCFD2 gene (autosomal recessive multiple coagulation factor
deficiency, a rare bleeding disorder). In addition, a gain in the
NOS1 gene, disrupting the genomic locus (blood pressure
Leukemia
control), was identified. Two other imbalances in a lower
percentage of cells were not verified yet. The patient had
refractory anemia with multilineage dysplasia, IPSS Int-1 and
died 6 months after diagnosis of gastrointestinal bleeding.
In patient P38 the large isoform of the MBP gene is disrupted
by a deletion, which ends close to the 50 end of the shorter
(hemopoietic, HMBP) isoform, possibly affecting its expression
(Supplementary Figure S6). The HMBP gene is expressed in
CD34 þ bone marrow cells and all hematopoietic lineages.
Patient P98 carried a 968-kb deletion in 3p26.2 in 100% of the
cells as determined with FISH. The region corresponds in part to
a known copy number variant, but two genes, ITPR1 and EGO,
map outside of the variant region. The EGO gene codes for a
non-coding RNA with increased expression following IL5
stimulation of CD34 þ cells.14 Therefore, this gene might have
a regulatory function in hematopoietic cell differentiation. A
deletion of the WWOX gene occurred in P90, which affected
two separate regions as identified by a custom array (Supplementary Figure S7). WWOX maps to a fragile site (FRA16B)
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
9
Figure 4 RUNX1 deletions in P24 and P39 and identification of the deletion end points. (a) The deletions as seen with the Agilent genomic
workbench program (setting: ADM-2, threshold 6). Only a few oligonucleotides show a deletion in case P39 due to the lower resolution of the 44k
array. (b) PCR products obtained using primers flanking the deletions as identified in custom arrays were obtained in both cases. Sequencing of the
PCR products identified the exact breakpoints. (c) Schematic drawing of the site of the deletions in the RUNX1 gene from both cases.
frequently altered in many different types of tumors. It encodes a
substantiated ts gene.15
Gains occurred in 21/42 cases; some patients had more than
one and most were small. Patient P106 had a gain in 3q26.1 at
the position of the MDS1/EVI1 gene, which affected only the
shorter EVI1 gene (Figure 6a). The gain was an amplification,
present in 49% of the cells, as evidenced by the presence of
EVI1 probe signals (green) on several different chromosomes
(Figure 6b). Limited material of poor quality prevented the
identification of the sites of EVI1 insertion. A high expression of
the EVI1 gene was found to predict poor survival in de novo
AML patients.16 The patient had RAEB-2, IPSS 2 and received an
allogeneic
stem cell transplantation and is still alive after 12 months.
Further gains containing several genes were identified:
for example, in 7q21.2 (P11) containing the CYP51A gene,
encoding a monooxygenase, which is a member of the
cytochrome P450 proteins, catalyzing many reactions involved
in drug metabolism and synthesis of cholesterol and steroids; in
21q22.3, a region containing among others two ribosomal
genes, RRP1B and RRP1 (P34) (Table 2); in 8q22.1 (P38) with a
gain of the entire INTS8 gene, which is a subunit of the
integrator complex with a role in processing of small nuclear
RNAs and disruption of CCNE2 (Supplementary Figure S6); in
16q24 with a gain of several entire genes, for example, the
ribosomal protein RPL13, CHMP1A, a chromatin-modifying
protein, CDK10, a cyclin-dependent kinase with a function in
the G2–M phase of the cell cycle and possibly disruption of the
FANCA gene at the border of the gain (P50) (Supplementary
Figure S8; Table 2). Another gain in case P72 included the entire
LRRK2 gene; point mutations but not gains in this gene are
found in patients with Parkinson’s disease. Of note, it has been
described that overexpression of wild-type human LRRK2
protects Caenorhabditis elegans after exposure to mitochondrial
toxins,17 and therefore this gain of the intact gene may have
contributed to the development of MDS.
Correlation with clinical parameters
The median survival was 56 months in the patient group without
alterations as compared with 20 months in patients with
alterations (P ¼ 0.002). The Kaplan–Meier curve for all patients
with and without aberrations is shown in Figure 7a and the
subset of patients with low-risk MDS (IPSS low and intermediate-1) is shown in Figure 7b (57 vs 26 months, P ¼ 0.017).
This further supports the notion that hidden alterations detected
with aCGH may have a prognostic impact.
Discussion
In this aCGH study of the first large series of exclusively
cytogenetically normal MDS samples, we show that MDS
patients with a normal karyotype have a few recurrent copy
number changes and several individual alterations. The four
recurrent aberrations involve genes/chromosomal segments
previously known to be affected in MDS. Two patients had
deletions in 4q removing the entire TET2 gene. Although TET2
was identified as the most frequently mutated gene in MDS,18–20
its role in disease remains controversial; two reports describe no
impact of TET2 mutations on survival in MDS/MPD (myeloproliferative disease),18,21 another describes an association with
decreased overall survival in AML13 and, most recently, TET2
mutations were reported to be an independent favorable
prognostic factor in MDS.22
Two patients had submicroscopic deletions in 5q; one was
only 1.35 Mb in 5q31.2, affecting the more proximal AML/MDS
region. The deletion includes all candidate ts genes described so
far in this region: CTNNA,23 EGR124 and HSPA9, which shows
increased expression during EPO-induced erythroid differentiation.25,26 In addition, other genes involved in cell cycle
regulation, CDC25C and CDC23, and in chromatin modification, JMJD1B, a lysine-specific histone demethylase and
previously described candidate ts gene from 5q27 are localized
in this segment. A recently described study of exon sequencing
Leukemia
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
10
Figure 5 Complex rearrangements in P79 involving chromosomes 19p, 21 and 8p. (a) Chromosome 8 with the telomeric deletion in a low
percentage of cells, gain of various regions in 19p and gain of the entire chromosome 21, shown in chromosome view using the Agilent workbench
program (setting: ADM-2, threshold 6). (b) FISH analyses using several different probe combinations and schematic drawing of the observed
signals. The data showed that there were four translocated chromosomes, hybridizing with both probes from chromosomes 19 and 21; several
carried an amplified signal of chromosome 19p. Only one normal chromosome 21 and one or two normal chromosomes 19 are present. Although
the deletion in 8ptel can be seen in the Agilent workbench program, the log ratio is very low. By chance a probe close to the homozygous deletion
of a copy number variant (CNV) in 8p was used and the deletion was identified in 17% of the cells with FISH. This deletion was present in the same
clones that also carried the other alterations. Therefore, in this case a complex cytogenetic result was present in B17% of the cells.
for all genes from this segment revealed no mutations in any of
these genes,28 further supporting the notion that the haploinsufficiency of one or more genes from this interval is the
disease mechanism. It was described that the dual-specific
phosphatase CDC25C is inhibited by lenalidomide and that
expression of this gene was lower in AML and MDS with
del(5q);29 therefore, haplodeficiency of proteins might also have
a role. This deletion does not include the typical 5q deletion in
MDS mapping more distal in 5q31–32.
A basically identical deletion in 7q22.1 was found in
three patients and overlaps with a commonly deleted segment
Leukemia
in chronic myelogenous leukemia and maps outside the
commonly deleted segment in MDS and AML.30 A deletion
spanning at least a segment between PLANH1/SERPINE and
CUTL1/CUX1 was described in an apparently balanced
translocation t(7;7)(p13;q22)31 and maps to the deleted
segment in the patients described here. Deletions with similar
breakpoints were described in two patients with refractory
anemia6 and in two patients with chronic myelomonocytic
leukemia.32 Taken together, this del(7q22) with similar
deletion end points seems to be a frequent recurrent hidden
aberration.
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
11
Figure 6 Amplification of EVI1 in P106. (a) A gain in 3q26 was identified, including the entire EVI1 gene and disrupting the longer MDS1/EVI1
gene, as can be seen in gene view (Agilent setting: ADM-2, threshold 6). The position of the FISH probes from the EVI t(3;3) inv(3)(3q26) break
apart probe used for verification of the amplification is marked as 3q26 distal (red) and 3q26 proximal (green). The EVI1 gene itself is not localized
on these probes, but flanked by the proximal green and distal red probes. (b) FISH analysis using this probe mix revealed an amplification of the
proximal green signal and, as deduced from the array result the amplification includes the EVI1 gene, whereas the MDS1 gene maps outside of the
amplification. The interphase (top) shows six green signals and the metaphase (bottom) demonstrates that the amplified DNA containing the EVI1
gene is present on various chromosomes.
Several interesting genes map to this genomic segment,
including EPO, TRF2, EPHB4 and CLDN15. The TFR2 gene
encoding a transferrin receptor is highly expressed in CD34 þ
hematopoietic cells and has an important function in mitochondrial iron metabolism. Mutations in this gene cause one type of
hemochromatosis, associated with iron overload.33 The homeobox gene CUX1 functions as a transcriptional activator or
repressor. EPHB4 is expressed in hematopoietic progenitor cells
and is downregulated during differentiation to mature erythroid
cells upon removal of its ligand.34 Active EphB4 enforces
preferentially megakaryocytic and erythroid differentiation.
It was described that the ephrinB2/EphB4 axis is dysregulated
in multiple myeloma.35 In addition, several genes in this region
are involved in regulation of RAS signaling, for example, RABL5,
RASA4, SH2B2 and GNB2.
Cytogenetically visible monosomy 7 or deletions (7q) are
currently poor IPSS risk factors. At present, it is unknown
whether submicroscopic deletions of 7q22 have the same
clinical impact.
Two patients had deletions disrupting the RUNX1 gene. The
critical regulatory role of RUNX1 for development of all
hematopoietic cell types is well established. The functional
consequence of mutations in RUNX1 can be haploinsufficiency
or a dominant-negative effect. This gene is also the most
frequent target for chromosomal translocations in leukemia and
the fusion protein inhibits the function of the native allele.36
So far only a few RUNX1 deletions have been described in MDS
(for example, Heinrichs et al.10 and Gondek et al.32). Inactivation of the normal RUNX1 function is not sufficient to cause
leukemia and other cooperating alterations were postulated.
Most alterations presented in this work were verified by other
methods and several deletion breakpoints were identified by
sequencing (manuscript in preparation). In our study, 39% of the
samples had hidden aberrations, whereas higher numbers of
aberrations were reported using SNP arrays, which can also
detect copy number neutral allelic imbalances. However,
recently it was shown that using normal genomic DNA from
the same patients for comparison is mandatory to avoid false
discoveries in SNP array studies.37 Furthermore, in most SNP
studies the alterations were not verified with other methods as we
have done for many alterations. Similar to our results, Praulich
et al.,38 who studied childhood MDS with BAC and oligonucleotide arrays, found a low frequency of hidden aberrations, which
were verified with other methods. Future SNP array analyses using
paired normal and tumor samples and aCGH studies using
karyotypically normal samples will have to unravel whether UPD
is indeed a more frequent aberration in MDS.
In our large series of karyotypically normal patients, we found
a significantly shorter time to death from diagnosis for patients
with additional aberrations, which is also seen when only
those patients with a low-risk MDS are analyzed. In addition,
the presence of genes involved in epigenetic and chromatin
modification in several of the altered regions further emphasizes
the important role of epigenetic modifications for MDS
Leukemia
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
12
Figure 7 Prognostic significance of detected aberrations. (a) Kaplan–
Meier curve for time from diagnosis to death for all patients with and
without aberrations. Patients with aberrations have a significant shorter
survival (P ¼ 0.002). (b) Time from diagnosis to death for patients with
low-risk MDS (IPSS 0 and 1) with and without aberrations. When only
the low-risk patients were analyzed, a significant shorter survival was
also observed for patients with aberrations (P ¼ 0.017).
development. It is expected that future functional studies
of the genes in the aberrant regions and pathway analyses
may help to uncover common genetic events leading to MDS
development and, in the future, may reveal new targets for
therapy.
Conflict of interest
The authors declare no conflict of interest
Acknowledgements
This work was supported by a grant from the Wilhelm Sander
Stiftung 2008.027.1 (BR-P) and a grant from the Forschungskommission HHU Duesseldorf 11/06 (CE and BB).
References
1 Brunning RS, Orazi A, Germing U, LeBeau MM, Porwit A,
Baumann I et al. Myelodysplastic syndromes/neoplasms, overview.
In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H,
Thiele J, Vardiman JW (eds). WHO Classification of Tumours of
Haematopoietic and Lympoid Tissues, 4th edn. IARC Press: Lyon,
France, 2008, pp 88–93.
Leukemia
2 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA,
Gralnick HR et al. Proposals for the classification of the
myelodysplastic syndromes. Br J Haematol 1982; 51: 189–199.
3 Malcovati L, Germing U, Kuendgen A, Della Porta MG,
Pascutto C, Invernizzi R et al. Time-dependent prognostic
scoring system for predicting survival and leukemic evolution in
myelodysplastic syndromes. J Clin Oncol 2007; 25: 3503–3510.
4 Paulsson K, Heidenblad M, Strömbeck B, Staaf J, Jönsson G, Borg A
et al. High-resolution genome-wide array-based comparative
genome hybridization reveals cryptic chromosome changes in
AML and MDS cases with trisomy 8 as the sole cytogenetic
aberration. Leukemia 2006; 20: 840–846.
5 O’Keefe CL, Tiu R, Gondek LP, Powers J, Theil KS, Kalaycio M
et al. High-resolution genomic arrays facilitate detection of
novel cryptic chromosomal lesions in myelodysplastic syndromes.
Exp Hematol 2007; 35: 240–251.
6 Starczynowski DT, Vercauteren S, Telenius A, Sung S, Tohyama K,
Brooks-Wilson A et al. High-resolution whole genome tiling path
array CGH analysis of CD34+ cells from patients with low-risk
myelodysplastic syndromes reveals cryptic copy number alterations and predicts overall and leukemia-free survival. Blood 2008;
112: 3412–3424.
7 Evers C, Beier M, Poelitz A, Hildebrandt B, Servan K, Drechsler M
et al. Molecular definition of chromosome arm 5q deletion
end points and detection of hidden aberrations in patients
with myelodysplastic syndromes and isolated del(5q) using
oligonucleotide array CGH. Genes Chromosomes Cancer 2007;
46: 1119–1128.
8 Mohamedali A, Gäken J, Twine NA, Ingram W, Westwood N,
Lea NC et al. Prevalence and prognostic significance of allelic
imbalance by single-nucleotide polymorphism analysis in low-risk
myelodysplastic syndromes. Blood 2007; 110: 3365–3373.
9 Gondek LP, Haddad AS, O0 Keefe CL, Tiu R, Wlodarski MW,
Sekeres MA et al. Detection of cryptic chromosomal lesions
including acquired segmental uniparental disomy in advanced
and low-risk myelodysplastic syndromes. Exp Hematol 2007; 35:
1728–1738.
10 Heinrichs S, Kulkarni RV, Bueso-Ramos CE, Levine RL, Loh ML,
Li C et al. Accurate detection of uniparental disomy and
microdeletions by SNP array analysis in myelodysplastic
syndromes with normal cytogenetics. Leukemia 2009; 23:
1605–1613.
11 Trolet J, Hupé P, Huon I, Lebigot I, Decraene C, Delattre O et al.
Genomic profiling and identification of high-risk uveal melanoma
by array CGH analysis of primary tumors and liver metastases.
Invest Ophthalmol Vis Sci 2009; 50: 2572–2580.
12 Gelsi-Boyer V, Trouplin V, Adélaı̈de J, Bonansea J, Cervera N,
Carbuccia N et al. Mutations of polycomb-associated gene ASXL1
in myelodysplastic syndromes and chronic myelomonocytic
leukaemia. Br J Haematol 2009; 145: 788–800.
13 Abdel-Wahab O, Mullally A, Hedvat C, Garcia-Manero G, Patel J,
Wadleigh M et al. Genetic characterization of TET1, TET2,
and TET3 alterations in myeloid malignancies. Blood 2009; 114:
144–147.
14 Wagner LA, Christensen CJ, Dunn DM, Spangrude GJ,
Georgelas A, Kelley L et al. EGO, a novel, noncoding RNA gene,
regulates eosinophil granule protein transcript expression.
Blood 2007; 109: 5191–5198.
15 Del Mare S, Salah Z, Aqeilan RI. WWOX: its genomics, partners,
and functions. J Cell Biochem 2009; 108: 737–745.
16 van Barjesteh Waalwijk Doorn-Khosrovani S, Erpelinck C, van
Putten WLJ, Valk PJM, van der Poel-van de Luytgaarde S, Hack R
et al. High EVI1 expression predicts poor survival in acute myeloid
leukemia: a study of 319 de novo AML patients. Blood 2003; 101:
837–845.
17 Saha S, Guillily MD, Ferree A, Lanceta J, Chan D, Ghosh J et al.
LRRK2 modulates vulnerability to mitochondrial dysfunction
in Caenorhabditis elegans. J Neurosci 2009; 29: 9210–9218.
18 Jankowska AM, Szpurka H, Tiu RV, Makishima H, Afable M,
Huh J et al. Loss of heterozygosity 4q24 and TET2 mutations
associated with myelodysplastic/myeloproliferative neoplasms.
Blood 2009; 113: 6403–6410.
19 Langemeijer SMC, Kuiper RP, Berends M, Knops R, Aslanyan MG,
Massop M et al. Acquired mutations in TET2 are common in
myelodysplastic syndromes. Nat Genet 2009; 41: 838–842.
Prognostic relevance of hidden imbalances in MDS
A Thiel et al
13
20 Delhommeau F, Dupont S, Della Valle V, James C, Trannoy S,
Massé A et al. Mutation in TET2 in myeloid cancers. N Engl J Med
2009; 360: 2289–2301.
21 Smith AE, Mohamedali AM, Kulasekararaj A, Lim Z, Gäken J,
Lea NC et al. Next-generation sequencing of the TET2 gene in 355
MDS and CMML patients reveals low abundance mutant
clones with early origins, but indicates no definite prognostic
value. Blood 2010; 116: 3923–3932.
22 Kosmider O, Gelsi-Boyer V, Cheok M, Grabar S, Della-Valle V,
Picard F et al. TET2 mutation is an independent favorable
prognostic factor in myelodysplastic syndromes (MDSs). Blood
2009; 114: 3285–3291.
23 Liu TX, Becker MW, Jelinek J, Wu W, Deng M, Mikhalkevich N
et al. Chromosome 5q deletion and epigenetic suppression of
the gene encoding alpha-catenin (CTNNA1) in myeloid cell
transformation. Nat Med 2007; 13: 78–83.
24 Joslin JM, Fernald AA, Tennant TR, Davis EM, Kogan SC, Anastasi J
et al. Haploinsufficiency of EGR1, a candidate gene in the del(5q),
leads to the development of myeloid disorders. Blood 2007; 110:
719–726.
25 Xie H, Hu Z, Chyna B, Horrigan SK, Westbrook CA.
Human mortalin (HSPA9): a candidate for the myeloid
leukemia tumor suppressor gene on 5q31. Leukemia 2000; 14:
2128–2134.
26 Ohtsuka R, Abe Y, Fujii T, Yamamoto M, Nishimura J, Takayanagi
R et al. Mortalin is a novel mediator of erythropoietin signaling.
Eur J Haematol 2007; 79: 114–125.
27 Hu Z, Gomes I, Horrigan SK, Kravarusic J, Mar B, Arbieva Z et al.
A novel nuclear protein, 5qNCA (LOC51780) is a candidate for the
myeloid leukemia tumor suppressor gene on chromosome 5 band
q31. Oncogene 2001; 20: 6946–6954.
28 Graubert TA, Payton MA, Shao J, Walgren RA, Monahan RS,
Frater JL et al. Integrated genomic analysis implicates haploinsufficiency of multiple chromosome 5q31.2 genes in de novo
myelodysplastic syndromes pathogenesis. PLoS ONE 2009; 4:
e4583.
29 Wei S, Chen X, Rocha K, Epling-Burnette PK, Djeu JY, Liu Q et al.
A critical role for phosphatase haplodeficiency in the selective
suppression of deletion 5q MDS by lenalidomide. Proc Natl Acad
Sci USA 2009; 106: 12974–12979.
30 Fischer K, Fröhling S, Scherer SW, McAllister Brown J, Scholl C,
Stilgenbauer S et al. Molecular cytogenetic delineation of deletions
and translocations involving chromosome band 7q22 in myeloid
leukemias. Blood 1997; 89: 2036–2041.
31 Tosi S, Scherer SW, Giudici G, Czepulkowski B, Biondi A,
Kearney L. Delineation of multiple deleted regions in 7q in
myeloid disorders. Genes Chromosomes Cancer 1999; 25:
384–392.
32 Gondek LP, Tiu R, O0 Keefe CL, Sekeres MA, Theil KS,
Maciejewski JP. Chromosomal lesions and uniparental disomy
detected by SNP arrays in MDS, MDS/MPD, and MDS-derived
AML. Blood 2008; 111: 1534–1542.
33 Hofmann W, Tong X, Ajioka RS, Kushner JP, Koeffler HP.
Mutation analysis of transferrin-receptor 2 in patients with atypical
hemochromatosis. Blood 2002; 100: 1099–1100.
34 Wang Z, Miura N, Bonelli A, Mole P, Carlesso N, Olson DP et al.
Receptor tyrosine kinase, EphB4 (HTK), accelerates differentiation
of select human hematopoietic cells. Blood 2002; 99: 2740–2747.
35 Pennisi A, Ling W, Li X, Khan S, Shaughnessy JD, Barlogie B et al.
The ephrinB2/EphB4 axis is dysregulated in osteoprogenitors from
myeloma patients and its activation affects myeloma bone disease
and tumor growth. Blood 2009; 114: 1803–1812.
36 Speck NA, Gilliland DG. Core-binding factors in haematopoiesis
and leukaemia. Nat Rev Cancer 2002; 2: 502–513.
37 Heinrichs S, Li C, Look AT. SNP array analysis in hematologic
malignancies: avoiding false discoveries. Blood 2010; 115:
4157–4161.
38 Praulich I, Tauscher M, Göhring G, Glaser S, Hofmann W,
Feurstein S et al. Clonal heterogeneity in childhood myelodysplastic syndromesFchallenge for the detection of chromosomal
imbalances by array-CGH. Genes Chromosomes Cancer 2010; 49:
885–900.
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
Leukemia