12th Congress of the European Hematology ... - Haematologica

12 th Congress of the European Hematology Association
q35.2 (5 cases), 5q23.1-5q31.1 (2 cases), 7q22.1-q32.1 (5 cases), 11q23.3-
11q24.2 (2 cases), and 13q12-q22.1 (7 cases). We analyzed if these
regions contain homozygous mutations in significant genes in AML,
and in 2 cases with 13q12 UPD we identified a homozygous mutation
in FLT3. Although it would be necessary to confirm it in a larger series,
LOH by UPD does not seem to have impact in the outcome of the
patients; however, UPD could lead to alterations in expression levels of
imprinted genes, and could be associated with specific gene expression
patterns. Fisher’s Exact test demonstrated no association (p>0.05)
between the presence of LOH and variables with a prognostic meaning
in AML: FLT3 mutation, age>60, and no complete remission. LOH by
UPD usually occurs in fragile sites in the genome, a common location
for miRNA. Recently, there has been major progress in the identification
of miRNA expression profiles that could be associated with prognostic
factors. Moreover, miRNAs seem to have a role in leukemia pathogenesis.
To study the relationship between LOH regions and miRNAs
expression we profiled expression of 157 miRNAs by using real-time
PCR in 23 patients, and in MO from normal donors. Twenty two miR-
NAs showed differentiation associated expression changes. Consistent
microRNA down-regulation was seen in miR-198, miR-211, miR-139,
miR-302b, miR-127, miR-214, miR-182*, miR-205, miR-105, miR-138
and miR-204; whereas miRNA upregulation was seen in miR-374, miR-
181a, miR-181b, miR-146, miR-210, miR-34a, miR-213, miR-219, miR-
155, miR-17-5p and miR-30e. In conclusion, our results confirm other
studies by showing that the prevalence of LOH by UPD in de novo AML
is high, 60% in our series with normal karyotype. Although it would be
necessary to confirm it in a larger series, this does not seem to have
impact in the outcome of the patients. Interestingly, we identified a subgroup
of 14 patients with no cryptic genomic aberration changes that
was heterogeneous at the molecular level and had different outcome,
confirming that the subclassification of AML patients with normal karyotype
should carry on looking for molecular profiles, including miRNA
differential expression.
0870
SEGMENTAL UNIPARENTAL DISOMY IS THE MOST COMMONLY ACQUIRED GENETIC
ABNORMALITY IN RELAPSED ACUTE MYELOID LEUKEMIA
M. Raghavan
Barts & the London School of Medicine, LONDON, United Kingdom
Background. Relapse is the commonest cause of death in acute myeloid
leukaemia (AML), but the mechanisms leading to relapse are unclear.
Recently, acquisition of segmental uniparental disomy (UPD) by mitotic
recombination (MR) has been reported in 15-20% of AML patients at
diagnosis using whole genome single nucleotide polymorphism (SNP)
arrays. These abnormalities are cytogenetically invisible and are associated
with homozygous mutations in several types of malignancy. Clonal
evolution from heterozygous to homozygous mutations by MR could
provide a mechanism for relapse. Aims. To identify regions of UPD
acquired at relapse of AML and associated gene mutations within the
homozygous region. Methods. DNA from 27 pairs of diagnostic and
relapsed AML samples were analysed using Affymetrix 10K SNP arrays.
Copy number and loss of heterozygosity were analysed using in-house
software. Regions of deletion, gains and segmental UPD were documented
and compared between diagnosis and relapse. Results. Segmental
UPDs were acquired at relapse in eleven AML patients (40%). Six of
these were segmental UPDs of chromosome 13q. FLT3 exon 14-15 lay
in the region of homozygosity. Sequencing of the six cases demonstrated
a change from heterozygosity at diagnosis to homozygosity at relapse
for an internal tandem duplication (ITD) mutation of FLT3, confirmed
by PCR fragment analysis. The mutation was identical between each
diagnosis and relapse pair. Another AML patient acquired segmental
UPD of 19q, which lead to homozygosity of a CEBPA substitution mutation
at position 957, changing from C to T. This is a stop codon that has
previously been described to produce a truncated protein. One AML
patient acquired segmental UPD of chromosome 4q. There is likely to
be an associated homozygous mutation in the region of UPD, but in
view of the large region involved, it was not possible to investigate further.
Another three AML patients had evidence suggesting an acquired
subclone, with UPD of chromosome 13, at relapse. The heterozygous
calls across chromosome 13 at diagnosis became no calls at relapse
because the calling algorithm was unable to interpret a change in the proportion
of alleles. This was confirmed by showing a change in the relative
allele signals between diagnosis and relapse. In one of these cases,
there was also a rise in the FLT3 ITD level at relapse suggesting the subclone
harboured a homozygous FLT3 mutation. Summary/Conclusions.
This study suggests acquisition of segmental UPD by mitotic recombi-
324 | haematologica/the hematology journal | 2007; 92(s1)
nation is the most commonly acquired genetic abnormality at relapse of
AML. It shows there can be possible clonal heterogeneity in the acquisition
of UPD, which suggests acquired UPD may be under detected.
Finally, it suggests AML cells with acquired UPD are resistant to
chemotherapy.
0871
DERIVATION OF A GLOBAL MAP OF DNA HOMOZYGOSITY IN ACUTE MYELOID
LEUKEMIA (AML)
M. Gupta
Barts and The London, LONDON, United Kingdom
Background. Cytogenetic abnormalities are the most important prognostic
factor in AML, but recent small studies have demonstrated frequent
acquired regions of homozygosity invisible by conventional karyotyping.
These regions have a normal copy number, so are called segmental
uniparental disomy (UPD). They result from mitotic recombination
or non-disjunction events and can lead to homozygosity for mutated
genes within the region. Aims. To provide a high-resolution map of
homozygosity and copy number changes (CNCs) in a large and unselected
cohort of patients with acute myeloid leukemia. Materials and Methods.
Diagnostic samples from 463 patients with AML from the MRC
AML10 trial were studied using Affymetrix 10K 2.0 single nucleotide
polymorphism (SNP) arrays. This array identifies 10,204 SNPs across all
chromosomes excepting the Y-chromosome. DNA from ten unrelated
non-leukemic individuals were used as a control. Data was analyzed
using in-house developed software, Genome Oriented Laboratory File
(GOLF). Results. 380 CNCs and regions with homozygosity were identified
in 161 patients (35%). Copy number neutral homozygosity due to
segmental or whole chromosome UPD was observed in 17%, deletions
in 14.5% and gains in 12%. Of all the 380 aberrations, 46.5% were deletions,
31.2% were UPDs and 21.6% were gains. UPDs were most frequent
on chromosomes 11, 13, 2, 1 and 6. The most recurrent regions
were on 13q, 11p and 11q. The most frequent deletions were of 7/7q, 5q,
6p, 6q, 11p and most frequent gains were of 8, 11q, 21q. UPD was seen
across all cytogenetic risk groups. The proportion of patients with deletions,
gains and UPDs amongst the poor-risk cytogenetic group were
71%, 34% and 23%, respectively. The good risk cytogenetic group had
9% deletions, 9% gains and 11% UPDs, and the intermediate risk cytogenetic
group had 9%, 7.5% and 16.5% respectively. UPDs of 13q (5.4%
of patients) were observed exclusively in intermediate risk AML. Within
the intermediate risk group, there were more UPDs amongst normal
karyotype (NK) AML patients (19%) than those with cytogenetic abnormalities
(10.5%). SNP array analysis was able to map in detail copy number
changes in poor risk, complex karyotype AML patients. Monosomy
or deletions of 18/18q (14.3%) and gains of 5p (5.7%) were observed
exclusively in poor-risk AML. Deletions of 12p, 17/17p, 3/3q, 20/20q
(11.4% patients for all four) were more frequent in poor risk patients
than any other group, which is in accordance with other reports. Deletions
of 8/8p, 15/15q and 16/16q were observed in 11.4%, 8.6% and
14.3% of poor-risk patients, respectively. The results of 369 patients by
SNP array analysis largely matched those from cytogenetic karyotyping
available. Of 302 samples that were normal on SNP arrays 12% showed
CNCs by cytogenetic analysis. SNP arrays could also detect many aberrations
missed by karyotyping. The two experimental approaches therefore
complemented each other. Summary/Conclusions. We have identified
recurrent regions of UPD in a large cohort of AMLs. UPD occurs in all
cytogenetic risk groups, but is more frequent in NK AML. This map is an
initial step towards identifying areas containing novel candidate genes
important in AML.