FISH and SNPs ­ exploiting the genome in myeloma
Gareth J Morgan
Morgan, Professor of
of Haematology
Haematology
Clonality
The MRC Myeloma IX study
Included 1970 patients split
split into 2 broad
broad arms younger fitter and older less fit patients
SURVIVAL BY RESPONSE
100
100
2 = 1024.
6
P < .001
80
80
60
60
GNI
CR N= 65
CR N= 339
VIV
40
RU
MR N= 79
PR N= 127
S 40
127
%
VGPR N= 58
EVI
NC N= 92
T
NC N= 12
ALU
PR N= 212
20
VGPR N= 107
MU
MR N= 7
C 20
PD N= 7
ED N= 48
PD N= 17
12
3
456
123
4
56
TIME (YEARS)
TIME (YEARS)
· Broad conclusions are that the inclusion of thalidomide into the induction
regimen CTD
CTD in both settings
settings decreases the number of progressive disease,
increases CR rates and CRs have better PFS/OS compared to standard
treatment.
· DNA, RNA, somatic DNA in large percentage of samples
h
w ich
hich pro
pro ides
v
an
ideal vehicle to examine biology and outcome factors.
FISH evaluation of bad IgH translocation
t(14;20)
Bad IgH translocations
Kaplan-Meier Method
p<0.001
10
1 0
14;20
t(4;14)
00
t(4;14)
0
1
80
Table of Statistics
t(14;20)
Mean M edian IQ R
1133.99
1381
*
t
522.91
485 638
n
60
t(14;16)
rce
n=1019
Pe
40
n=16
An
A y
n ba
b d
a Ig
I H
g t
20
Ht
Kaplan-Meier Method
0 0
500
1000
1500
2000
p<0.001
Survival
100
bad
IgHt
t(14;16)
Kaplan-Meier Method
0
p=0.025
1
100
14;16
80
0
80
1
Table of Statistics
80
Table of Statistics
Mean Median IQ R
Mean M edian IQ R
1132.21
1379
*
825.16
803
*
t
1177.38
1510
*
tn 60
n
60
846.53
784
*
rce
n=858
Pe
rce
40
Pe
40
20
n=170
0 0
500
1000
1500
2000
Survival
20
t(4;14)
Kaplan-Meier Method
p<0.001
0 0
500
1000
1500
2000
100
4;14
0
1
Survival
80
Table of Statistics
Mean Median IQ R
1152.47
1446
*
t
870.14
786 1207
n
60
rce
n=921
Paradox
Pe
40
Some cases of t(4;14) do well and it is also seen in MGUS
n=121
20
Implication
Collabora
Collabo tin
ra
g genet
gene i
t c lesions
lesions mediating
mediatin the
the progno
progn s
o is
s
0 0
500
1000
1500
2000
2500
3000
Question
Survival
What are the collaborating lesions
How can they be identified
Exploiting single nucleotide polymorhisms
- 99.9% of individual DNA sequences will be
identical to that of another person.
- Of the 0.1% difference, over 80% will be single
nucleotide polymorphisms (SNPs).
- Inherited copy number variation is a further source
of variability.
-
Focussing on differences can
id
prov e a
f
use l
u investi
tigati
ti
l
ona
tool.
-
Mapping the myeloma genome
- Important genes and
and pathways
- Prognostic factors
- Predictive factors
- Side effects of
of treatment
treatment
- Causative factors for myeloma
500K SNP-based Mapping Arrays
·
Consists of 2x 250K arrays (Nsp
(Nsp and Sty)
·
DNA is digested, adaptors ligated, DNA amplified, fragmented
and labelled
·
After hybridi
idising, arrays are stained resulting in homozygous or
heterozygous calls
·
Use control sample to compute acquired LOH for tumour
·
Information on copy number and LOH status
·
Median physical distance between SNPs = 2.5 kb
·
Av
A erage distance = 58
5.8 kb
kb
·
85% of human genome within 10 kb of a SNP
Global gene mapping and clinical outcome
pq
gain
loss
Chromosome
Deletions (%)
Chromosome
Gains (%)
1p
29.8 + 4.4 UPD
1q
36.0
6q
33.3
3
27.2
8p
25.4
5
33.3
12
21.9
7
21
13q
58.7
9
35.9
14q
38.1
11
24.6
16q
35.0 + 8.7 UPD
15
36.8
17p
7.0 + 1.75 UPD
19
33.3
18
15.8
21
12.3
20
12.3
22
18.4
X
28.0 + 21.0 UPD
X
8.7
FISH evaluation in MMIX
LOSS
GAIN
LOSS
1p32-
1q+
More common in
in younger
younger patients
17p-
17p
Do worse with maintenance thalidomide
del13
Del 13 by FISH is not a prognostic factor
Exploiting and evaluating mapping data
Region
Chromsome
Genes
Con-1
2p
SRD5A2
Con-2
10q
CYP2C9
Con-3
20
ZMYND8
H1
Hyp-1
5
TERT
SLC
L 22A4
Hyp-2
9
P16
MUSK
Hyp-3
11
CCND1
BIRC2
Hyp-4
15
SRD5A2
1p
1
CDKN2C
ST7L
1q
1
CKS1B
CKS1
TFB
TF 2M
2
44
NFKIB
6q
6
WTAP
8p
8
PPP2R2A
13q
13
RB1
16q
16
WWOX
CYLD
17p
17
TP53
Design multiplex quantitative PCRs
Tagged high MAF SNPs across identified regions
Region
Chromsome
Genes
Con-1
2p
SRD5A2
Con-2
10q
CYP2C9
Con-3
20
ZMYND8
Hyp-1
5
TERT
SLC22A4
Hyp-2
9
P16
MUSK
Hyp-3
11
CCND1
BIRC2
Hyp
Hy -4
15
SRD5A2
D5
1p
1
CDKN2C
ST7L
1q
1
CKS1B
TFB2M
44
NFKIB
6q
6
WTAP
8p
8
PPP2R2A
13q
13
RB1
16q
16
WWOX
CYLD
17p
17
TP53
Defining Tumour Suppresor genes by the integration of SNP and expression data
Cluster samples
by DNA copy-
Collect SNP array data
number/LOH
Define genomic regions
difference
driving subclusters
Map significant genes to
genomic locations and
identify candidate genes
within genomic alterations
Integrate with
SNP array data
Collect gene expression data
Experimental validation
of candidate cancer
genes
Perform supervised analysis
of subclusters
Garraway et al. 2005 Nature
Deletions of 1p ­ FAF1 is an important new gene
Three MDRs on 1p
1p32.3
p
Poor prognosis and implicates
FAF
FA 1
F ­ negative regulator of apoptosis and NFkB
P18 ­ negative regulator of cell cycle
1p12
1p22.1
No significant impact on prognosis
No significant impact on prognosis
Based on FAM46C and USP33 genes
MTF2 and TMED5 genes
3 regions of deletion on chromosome 16q; CYLD and WWOX
Deletion
UPD
Rl
Relapse dlt
dele i
tion
Homozygous deletion
CYLD
Negative regulator of NFkB
cloned at ICR associated with
li
cy
d
n roma tumours
CDH1
ATBF1
WWOX
Mutated in 20% of LOH cases
S
L
Q
P
V
F
Q
S
D
Ter
SL
S
L
Q
P
V
S
N
Q
I
E
Exons sequenced in 14 cases with LOH at CYLD locus and compared to PB DNA
Mutations found
1655 C/TT frameshift giving stop codon
2412 C/T
D805E
Homozygous Deletions on 11q
Molecular classification of myeloma
Aetiological events
Progression events
t(4;14)
t(11;14)
t(6;14)
Translocations
t(14;16)
t(16;20)
Loss of heterozygosity
1p-FAF1, CDKN2C;
Hyperdiploid
16q-CYLD1, WWOX;
14q-TRAF2;
11q-BIRC2/3;
17p-
17p P53;
del(13), Rb.
Gene rearrangements
MYC, NIK.
Mt
Mutati
tion acti ti
va ng
NRAS, KRAS, NIK, NFkB1/2
NFkB1/2; FGFr3
inactivating
P53, CYLD1, TRAF2.
Gene amplification
1q+ CKS1B.
Homozygous deletions
·
Homozygous deletions are important
important markers of
of deleted
deleted tumour
suppressor genes
·
Genes contained within them which are normally expressed but
expression is lost because of the deletion
deletion are
are obviously inactivated
i. normal
ii. LOH
iii. LOH +TSG
LOH
iv. HD
inactivation
TSG
TSG
TSG
TSG
normal
Normal or
abnormal
abnormal
haploinsufficient
Exploiting genes contained within Homozygous Deletions
Aim ­ to create an
an expression
expression
signature based on defined DNA
abnormalities relevant to MM
Mapping array data
Expression array data
nomalized in dChip
nomalized in dChip
Chromosome 1 Chromosome 8
Chromosome 11
Chromosome 11 Chromosome 17
1.75 Mb
14.921 Mb
1.95 Mb
0.53 Mb
0.718 Mb
CDKN2C
BIRC2, BIRC3
BIRC2, BIRC3
Keratin locus
CN<0.93 in
4 adjacent SNPs
Determining cut-off for homozygous deletions
All HD regions
identified
- Used FISH detection of del13 to identify hemizygous deletion
- Confirmed FISH result by SNP array, discounted interstitial
Present in
>1 sample
deletion samples
Relevant HD
- Looked at distribution of copy number across 13q, and used
regions
lower 1% of values as threshold
Identify genes
- Homozygous deletions determined as copy number <0.93 in
in regions
our data series
HD genes
Compare gene
- However, method over-predicts HDs, so filter with expression
expression in deleted vs.
of interest
non-deleted samples
array data
Filtering Criterion
HD median expression
< non del
-
eted median
- HD regions must be present in 4 adj
t
acen SNP
SNPs
expression
- Relevant regions must be present in >1 sample
Genes affected
- Genes in relevant regions must not be expressed
by HDs
Filtering of Homozygously Deleted Genes
3
5
6
Deletion of Cell Death Genes in Myeloma
· When
homozygously
deleted
genes
are
annotated
according to Gene Ontology (GO) terms, there is an over-
representation of Cell Death pathway genes (EMP1, SGK,
TFDP1, BIRC3, TNFRSF10B, CTSB, BIRC2, FOXO1,
ESR1,
PLAGL1,
CDKN2C,
FGF14,
FAF1
and
TNFRSF10D).
· These can be grouped and samples with deleted Cell
Death genes have a worse Overall Survival (OS) and
Progression Free Survival (PFS) than those without
deletion.
· note: individual genes do not affect outcome, only
the network has an effect.
· This can then be extrapolated to expression array data,
where loss of expression of these genes correlates with a
poor overall survival in our dataset (median 32 vs 48
months).
· This has also been validated on published expression
data sets from three other myeloma groups.
Signature is prognostic and validates
MyIX expression 259
Poor prognosis samples
Shaughnessy set approx 600
Signature validates at relapse with velcade
Mulligan data set
set
Large trial derived data-set
derived from the APEX study
evaluating the use of velcade at
relapse
Signatures identify a common group
a. Arkansas signature
c. Overlap between classifiers
70/17
70/1 gene prepon
prepo d
n e
d r
e an
a ce
n
on Chr 1
Gene lists are non overlapping
259 presenting patient with myeloma analysed
89 HR by Cell death signature
64 HR by IFM signature
90 HR by Arkansas signature
yg
Overlap between cases identified by different classifiers is shown below
Total 89
Avalidated gene expression model of high-risk multiple myeloma is defined by
deregulated expression of genes mapping to chromosome 1
John D. Shaughnessy Jr,1 Fenghuang Zhan,1 Bart E. Burington,2 Yongsheng Huang,1 Simona Colla,1 Ichiro Hanamura,1
James P. Stewart,1 Bob Kordsmeier,1 Christopher Randolph,1 David R. Williams,1 Yan Xiao,1 Hongwei Xu,1 Joshua Epstein,1
Elias Anaissie,1 Somashekar G. Krishna,1 Michele Cottler-Fox,1 Klaus Hollmig,1 Abid Mohiuddin,1 Mauricio Pineda-Roman,1
Guido Tricot,1 Frits van Rhee,1 Jeffrey Sawyer,1 Yazan Alsayed,1 Ronald Walker,1 Maurizio Zangari,1 John Crowley,2
and Bart Barlogie1
Cell Death
b. IFM
IFM signature
IFM 53(16)
sit
ignature
Total 64
15 gene identifies 25%
signature 37
43
46
(6)
(9)
Arkansas
signature
Total 90
Developing a 6-gene signature
Univariate
1st Multivariate
Evaluating the genes
genes
Univariate
Make Cox
Cox
multivariate
Regression
<0.05p
Bk
Break i t
Significants
n o
ER
independent
T
chunks
Logrank
FIL
test
Complete
p
multivariate
Cox
Compare clustering
clustering
Sensitivity
Make Cox
Six Gene
Specificity
py
Compare
Predictive
with other
Values
e
factors
2nd Multivariate
Evaluation
Six gene signature is independent
Factor
Z
P-valu
-
e
valu
Signature
6 gene signature
2.217
0.027
Cytogenetic factors
Deletion 13q
-0.354
0.72
t(4;14)
-02
0. 78
278
07
0. 8
78
t(11;14)
1.291
0.2
Deletion 17p
-1.396
0.16
Deletion 16q
-0.114
0.91
Six gene signature is clinically important in MMIX and validates
Myeloma
My
IX
IX
Median survival with the signature 12
months, without 45 months.
Logrank p= 2.9x10-8
(N
17,
6-gene sig.
N
241)
Others
XBP1U/S ratios are prognostically important in myeloma
IRE1a and control of XBP1u/s ratios
Unfolded protein response to
process nascent Ig
Plasma cell
XBP1s :
IRE1a
uPR
XBP1u
Ig
ratio
IL4
STAT6
XBP1
XBP1s
growth
B cell
IL6
Myeloma patients (n=257)
survival
XBP1u/s ratios in clinical myeloma samples
XBP1 s/u 1.33
C
--- H
u
L
m
S
u
r
v
i
v
a
l
Myeloma cell lines and patient cells express different
OS (months)
levels of XBP1 and XBP1s high ratios associated
with poor outcome
Integrating genomic and tumour genetic changes
Inherited genetic variation aetiology and side effects
Case control study
i. Side effects
VTE, Blood 2007. Predictive signature of who is at risk.
Bone disease, submitted.
Neuropathy
py
Velcade
Thalidomide
ii. Response
iii. Aetiology
Controls
UK 1958 Birth cohort
3000 Affymetrix, Infinium, 100 500k.
USA NHS and PLCO
1000 Infinium
NL Rotterdam study
2 000 Infinium, 100 500k.
6000
Cases
MRC IX
1 500
BOAC custom
Hovon
800
BOAC custom
IFM
400
BOAC custom
IMF study
400
BOAC custom
New England study
600
BOAC custom
Czech
250
BOAC custom
ECOG
3 950
Therapy associated neuropathy
·
Is there a genetic contribution to the aetiology
ggy of
Velcade/Thalidomide neuropathy and is it the same or different
hinting at different mechanisms underlying this side effect ?
·
Pieter Sonneveld, David Johnson, Sophie Corthals
Velcade exposed.........neuropathy vs no neuropathy = top associated SNPs
Thalidomide............... neuropathy vs no neuropathy = top associated
associated SNPs
Compare and contrast
Anal sis
y
in
in progress
progress
Take SNPs and ask if they effect
Different SNPs associated with neuropathy in both settings
Predictive signatures
expression patterns
Ie are they functional ?
Inherited genetic variation and the cause of myeloma
? Is there anything which differentiates the cause of myeloma which
carries a split IgH gene from other myelomas
IgH split: 330 cases, 405 myeloma controls, permutated p-values 0.01
SNP
CHR
SNP type
Case MAF Control MAF
OR (L95 - U95)
P
Permutated P
rs408820*
6
GSTA4 , locus , C>G
0.09
0.05
2.121 (1.396 - 3.22)
0.0003
0.0004
rs3743591
16
TNFRSF17 , untranslated , G>A
0.08
0.04
2.237 (1.409 - 3.551)
0.0005
0.0019
rs1544094
9
PAX5 , Promoter , G>C
0.10
0.06
1.968 (1.325 - 2.922)
0.0007
0.0007
rs13035
12
RECQL , untranslated , G>T
0.37
0.46
0.6949 (0.5607 - 0.8613)
0.0009
0.0014
rs619487
11
PGR , intron , G>T
0.37
0.30
1.412 (1.135 - 1.757)
0.0020
0.0021
rs16754
11
WT1 , coding-synon , C>T
0.13
0.19
0.6299 (0.4687 - 0.8467)
0.0021
0.0025
rs1395
2
SLC5A6 , coding-nonsynon , G>A
0.38
0.30
1.41 (1.122 - 1.772)
0.0031
0.0021
rs10501815
11
MRE
R 11A , int
n ron
tron , A>C
A>
01
0. 2
12
00
0. 8
08
1.675 (1.181 - 2.377)
0.0036
0.0032
rs215073*
16
ABCC1 , intron , T>C
0.04
0.08
0.5022 (0.3135 - 0.8044)
0.0036
0.0032
rs7805658*
7
SHFM1 , intron , A>G
0.49
0.42
1.359 (1.104 - 1.673)
0.0038
0.0031
rs2058996
2
ABCB11 , intron , T>C
0.48
0.41
1.33 (1.08 - 1.637)
0.0071
0.0078
rs1330286
9
ALDH1A1 , intron , C>G
0.37
0.31
1.341 (1.079 - 1.667)
0.0082
0.0069
rs1800591
4
MTP , locus , T>G
0.23
0.29
0.7277 (0.5739 - 0.9227)
()
0.0086
0.0089
rs1611755
1
SPRR1A , locus , T>C
0.43
0.51
0.7524 (0.6085 - 0.9304)
0.0086
0.0064
rs2000813
18
LIPG , coding-nonsynon , T>C
0.34
0.28
1.343 (1.075 - 1.679)
0.0094
0.0099
Seems to be a diff
different aeti l
o ogy
d
un erlying this subgroup of
myeloma !!!!
Copy number variants
· Copy Number Variants
(CNV):
­ DNA segment 1kb or
larger present at a
variable copy
copy number in
in
comparison with a
reference genome
­ Duplications, loss,
lt
complex rearrangements
­ Implicated in wide range
of human diseases
· Genome-
Genome wide
-
map
map of
CNVs
­ EA 500K mapping arrays
and array CGH
· Copy Number Variable
Regions (CNVRs)
·
Created by merging
overlapping CNVs
Redon et al, Nature 2006
Methods: array processing
· DNA extraction and
and array
hybridisation and
scanning
· GCOS and GTYPE
GTYPE
· Use of Hapmap CEU
reference set
· Quality control
­ Genotype call >93% using
BRLMM
­ Standard deviation <0.3
using CNAG
­ Exclusion of
· IgH, IgK and IgL
· X chromosome
Most frequent CNVRs in myeloma cases
Cases
Case
Region
Size
Control
Rank
Band
Start
End
Gains
Losses
with
(%)
PGenes
ID
(kb)
(%)
CNVR
OR4N4 OR11K1P OR4M2
1
115 15q11.2
18427103 20630718 2203.615
12
8
20 31.7%
43.5% 0.088
TUBGCP5 CYFIP1 NIPA2
NIPA1
OR4Q3 OR4M1 OR4N2
2
110 14q11.2
19272965 19581513 308.548
17
0
17 27.0%
21.3% 0.333
OR4K2 OR4K5 OR4K1
OR4K15 OR4K14
OR4K13
3
121 16p11
16p1 2
.
34143861
143
34647935
647
504 0
. 74
07
10
0
10 15 9
. %
9%
10 1
. %
1% 01
0. 91
19
ANUBL1 FAM21C
FAM21D CTGLF1
FRMPD2L1 FRMPD2L2
10q11.21-
FRMPD2 FAM35B SYT15
4
81 q11 22
45392878 47938625 2545.747
10
0
10 15.9%
12.5% 0.423
GPRIN2 PPYR1 FAM25A
q11.22
FAM25B FAM25C
FAM25G CTGLF7
ANXA8L2 FAM21B
ANXA8 ANXA8L1
5
58 7p11.2
57174657 57730637
555.98
8
0
8 12.7%
0.6% <0.001 ZNF479
17q21.31-
6
129 q21.32
41518102 42164185 646.083
5
3
8 12.7%
15.8% 0.711
LRRC37A LRRC37A2
ARL17P1 ARL17 NSF
7
65 8p23.1
11941938 12646342 704.404
6
0
6 9.5%
0.4% <0.001 FAM90A25P DEFB130
LONRF1
8
101 12p11.21
31093189 32477931 1384.742
5
1
6 9.5%
10.7% 1.000
DDX11 DDX12 FAM60A
AMN1 C12orf35 BICD1
Role of CNVs in myeloma pathogenesis
CNV
The end
FISH Wessex
Faith Davies
CTRU Leeds
My laboratory
yy
Fiona Ross
Emma Davenport
Walter Gregory
Brian Walker
Laura Chiechio
Lauren Wiseglass
Sue Bell
Dave Johnson
Emma Smith
Alex Szubert
Paula Leone
Ti
Tina
na Bagratuni
Nuria Navarra
Navarra
Matt Jenner
Kevin Boyd
Luis Brito
Nick Dickens
Athanasia Avramidou
David Gonzalez
Clinical colleagues
g
Tony Child
Graham jackson
Roger Owen
Mark Drayson
Andy Rawstron
Sylvia Feyler
Kay Kendall Leukaemia Fund