Gene expression profiling to predict
tt
treat
t
men response in myeloma
Shaji Kumar,
Kumar M.D.
M.D.
Mayo Clinic
St
Sco
S
t
tt d
s
l
a e,
l
Arizona
i
Rh
Roch
oc
t
es er,
er Mi
M
Minneso
Mi
ta
Jacksonville, Florida
Mayo Clinic College of Medicine
Mayo Clinic Comprehensive Cancer Center
Why predict
predict response?
Allows individualization of therapy
Avoid unnecessary toxicity in patients not likely to
respond
Allows selection of therapy likely to induce maximal
response
Various clinical and tumor cell characteristics have
been used
Gene expression profiling
Gene expression profiling provides a molecular
portrait: assess large numbers of genes
sil
imu t
ltaneously
Diagnosis, stratification and prediction of
of t mor
u
behavior based on signatures
Prognostic signatures as well as response prediction
signatures have been develop
gped for other tumors;
eg. Breast cancer
GEP in
in myeloma
myeloma
GEP has been a valuable tool in understanding
tumor biology
GEP signatures have led to development of
molec l
u ar
lar classifications and prognostic gro
gro ping
u
systems
Also, GEP allows indirect assessment of IgH
translocation status
GEP and drug response
In vitro signatures have been developed for
individual drugs using cell lines
These studies give an insight into potential
mechanisms of
of action
action and resistance
Very few
few studies have examined
examined genetic signatures
that predict response to therapy
Using GEP
GEP for predicting response
response
Three sets of information can be derived from the
GEP of myeloma cells
Individual genes differentially expressed in responders and
non-responders
IgH translocation status among responders and non-
responders
Functional pathways differentially
differentially regulated between
responders and non-responders
Predicting resistance to Thal Dex
GEP was
was performed on myeloma cells
cells from 30
patients with newly diagnosed myeloma treated with
thalidomide and dexamethasone
Laboratory data from diagnosis and during therapy
was revid
iewed
Response: A decrease in
in serum
serum M protein
protein by 50% or
a decrease in 50% of urine M protein in the absence
of a serum M spike
Methods
Bone marrow mononuclear cells were separated from
aspirates by Ficoll or ACK lysis
Tumor cells were isolated by CD138 selection
Purity confirmed by cytospin slide
Total RNA isolated using Qiagen RNeasy, cDNA
generated
Methods
The cDNA was hybridized onto an Affymetrix U133A
chip and scanned using a Genechip 300 scanner and
GeneChip 5 0
. software
software (Affymetrix)
(Affymetrix)
The CEL
CEL files from
from the
the analysis were
were analyzed using
using
Genespring 7.2® software (Agilent Technologies);
preprocessed using GCRMA followed by per gene
normalization to
to median
median.
The output was further analyzed using Ingenuity®
The output was further analyzed using Ingenuity
Pathway Analysis program.
Approach
1.
Identify genes that can reliably predict likelihood of
response to thalidomide and dexamethasone?
2. Correlate IgH translocation status assessed by GEP
to response
3. Can
Can comparison of gene expression profiles from
from
non-responders and responders provide insights
into mechanisms of drug activity
gy and resistance
Predictors of
of response
response
Supervised analysis on 30 patients receiving Thal + Dex
combination
Class Prediction analysis using Support Vector
Machines; Golub
Golub (signal to noise ratio) method
Using the set of 30 chips as training set, we arrived at
set of 25 genes that correctly classified non-responders
from responders
Non-responders
Responders
enesG
Dose response
25
2.5
2
1.5
1
0.5
0
1
3
5
7
9
11131517
1921232527
29
By response category
category
100
212854_x_at
211750_x_at
200022
200
at
_
201154_x_at
10
209251_x_at
211072_x_at
213646_x_at
201029_s_at
201090_x_at
212639_x_at
217526_at
202378_s_at
1
201234_at
NR
PR
CR
201653_at
221581_s_at
218165_at
211058_x_at
0.1
Chr Loc
Gene Name
Protein
4q12
SEC3L1
Exocyst complex component 1
3q26.1
SMC4L1
Structural maintenance of chromosomes (SMC) family member
12q12
TUBA6
Tubulin alpha 6
14q22
CNIH
Cornichon homolog
7q11
WBSCR5
Linker for activation of T cells family, member 2
16p12
NFATC2IP
Nuclear factor of activated T-cells
Xp22
CD99
MIC2, T cell activation
1p31
LEPR
Leptin receptor gene
gene related
-
protein
protein
1p32
USP1
Member of the ubiquitin-specific processing (UBP) family of
proteases
1p35
FLJ11730
FLJ1
Sarcoma antigen
antigen NY SAR
-
91
-
10q11
CSTF2T
Cleavage stimulation factor
10q23
FAM35A
Family with sequence similarity 35, member A
11p15
ILK
Integrin-linked kinase, apoptosis suppressor
Chr Loc
Gene
Function
21q22.3
PTTG1IP
pituitary tumor-transforming 1 interacting protein
15q25.1
IDH3A
isocitrate dehydrogenase 3 (NAD+) alpha
10p11.2
SVIL
supervillin
15q2
ADAM10
ADAM metallopeptidase domain 10
19q13.43
UBE2S
ubiquitin-conjugating enzyme E2S
11q22.3
ACAT1
acetyl-Coenzyme A acetyltransferase 1
2p23
HADHA
hydroxyacyl-Coenzyme A dehydrogenase, alpha subunit
15q21.1
EIF3S1
eukaryotic translation initiation factor 3, subunit 1 alpha, 35kDa
21q22.3
LSS
lanosterol synthase
9p13.3
KIAA1539
KIAA1539
14q24 3
.
NEK9
NIMA (never in mitosis gene a)
a)- related kinase
kinase 9
3q27.1
EIF2B5
eukaryotic translation initiation factor 2B, subunit 5 epsilon, 82kDa
4p16
CTBP1
C-terminal binding protein 1
12q13.1
HNRPA1
heterogeneous nuclear ribonucleop
gprotein A1
7q32.3
TNPO3
transportin 3
15q26.1
NGRN
neugrin, neurite outgrowth associated
7p13
DDX56
DEAD (Asp-Glu-Ala-Asp) box polypeptide 56
3p21.3
QARS
glutaminyl-tRNA synthetase
7q22.3
RINT1
RAD50 interactor 1
15q24.1
EDC3
enhancer of mRNA decapping 3 homolog
Correlation with Primary IgH
translocations
None of the differentially
differentially predictor
predictor genes were
associated with the five common translocations in MM
(11q13 (cyclin D1), 6p21 (cyclin D3), 4p16 (FGFR3-
MMSET), 16q23 (c-maf), or 20q11 (maf-B)).
However all
all 5 non responders
-
had one of the primary
primary
IgH translocations (11q13 -1 pt, 6p21 -1 pt, 4p16 -2 pts,
20q11 -1pt)
In comparison, 7 of the 25 responders had one of the
tl
transl
t
oca i
tions.
Response and Primary IgH
Translocations
7 of 12 patients (56%) with a primary IgH
translocation responded to Thal Dex
versus
18 f
o 18
18
t
pa i
ti
t
en s (100%)
i
w th
ith no
i
pr mary I H
g
translocations identified responded to Thal Dex
Pathway analysis
Ingenuity Pathway Analysis Program
521 genes which were two fold or more differentially
expressed between responders vs non responders
Functionally grouped networks were identified to which
289 specific genes were grouped "focus genes"
Canonical pathways into which the genes could be
groupedi
d were examined
Conclusion
We have identified a set of 15 genes that can potentially
predict for response to thalidomide and dexamethasone
Genes and networks involved in immune response and
cytokine signaling
signaling were
were up
up regulated
regulated in non-
non responders
Although patients with and without primary IgH
translocations respond well to thalidomide
dexamethasone, those without the translocations had
much higher response rates
rates.
Predicting CR
CR with Thal
Thal Dex
32 patients enrolled in Bologna 2002 trial
GEP was performed on 138 sorted myeloma cells
Affymetrix HG133 Plus Microarray platform
Gene Expression Profiling (GEP) of Myeloma (MM) Cells To Predict Attainment (near)
Complete Response to Primary Therapy with Thalidomide-Dexamethasone (Thali-Dex)
for Newly Diagnosed MM. Carolina Terragna et al, ASH 2006
nCR with Thal
Thal Dex
Data analyzed using Genespring 7.3
Using a similar approach, set of 10 genes identified
that were able to predict attainment of nCR with Thal
Dex
Genes predicting nCR
nCR
Chr loc
Pb
Probe ID
ID
Gene
Ft
Func i
tion
1p32.1
225792_at
HOOK1
Microtubule and cytoskeleton organization
1q22
236223_s_at
RIT1
Small GTPase mediated signal transduction
1p35.3
225282_at
SMAP1L Regulation of GTPase activity
2q33
239629_at
CFLAR
Regulation of apoptosis
2p13
227027_at
Tumor rejection antigen
11 14
p
3
.
205848
t
_a
GAS2
At
Apoptosis
d
an
l
cell cycle
t
con l
ro
12p13
200951_s_at
CCND2
Cyclin D2
14q32 31
.
200628 s
_
a
_ t
at
WARS
Protein biosynthesis
biosynthesis, regulation of
of proliferation
proliferation
unknown 242121_at
RNF12
Regulation of transcription
GEP signature
signature predicting
predicting for EFS
Baseline and 48 hr GEP was examined among
patients in TT2
Set of 151 genes were identified that predicted
superiE
ior
FS
EFS
Changes across the paired samples were predictive
for some
some of
of the
the genes
However, difficult to use this data from the context of
thalidomide specific
specific effect
Burington B, Shaughnessy J, Barlogie B, John C. Gene Expression Profiling
(GEP) of Purified Plasma Cells
Cells at Baseline and
and 48hr after
after-Dexamethasone (D) or
or
Thalidomide (T) Improve Outcome Predicition of Baseline GEP Alone in
Patients with Multiple Myeloma (MM) Treated with Total Therapy 2 (TT2). Blood.
2005;106:150a.
Bortezomib
Baseline CD138 selected MM cells were studied
using GEP
Patients from 4 different clinical trials (024, 025, 039,
040) were studied
A large set of gene were identified that were
associated with response to
to bortezomib
The highly expressed genes from responders
included those involved in adhesion, cytokines,
NFkB activity as well as hypoxia
Mulligan, G. et al. Blood 2007;109:3177-3188
Mulligan, G. et al. Blood 2007;109:3177-3188
High dose therapy
251 patients from IFM99 trials were studied using
GEP
Supervised methods were performed to identify
expressii
ion signatures associated with
ith survi
l
va
15 genes were identified and a score developed
It was
ti
prognos c for
l
overa lll
i
surv val independent of
ISS, del13 and t(4;14)
Decaux O, Magrangeas F, Lode L, et al. Molecular Prognosis in Multiple
Myeloma: The IFM Experience. ASH
ASH Annual Meeti
ting Ab
Abstracts. 2006;108:3491.
Future Directions
Currently using a similar approach to identify
responders to Revlimid based regimens
Validations of different genes underway
Cellular expression of some of the downstream
proteins based on differentially expressed genes are
being studied
Future directions
directions
Eventually a focused array spotted with genes
predicting resistance and sustained responses to
different agents need to be developed
And/or
A panel of markers that can be assessed by a
Tissue Microarray based examinati
tion of bone
marrow can be developed
Acknowledgements
Mayo Clinic Dysproteinemia Group
Eastern Cooperative Oncology Group