Title: Gene expression profiling in large cell lymphoma: emerging clinical applications
1Gene expression profiling in large cell lymphoma
emerging clinical applications
- Timothy S. Fenske, M.D.
- April 15, 2005
2Diffuse Large B-cell Lymphoma
- Most common NHL (1/3 of all NHL worldwide)
- Heterogeneous morphology
- Centroblastic type (80 of cases)
- Immunoblastic (10 of cases)
- Anaplastic and Plasmablastic variants also
- Other subtypes recognized
- Intravascular lymphoma
- T cell (or histiocyte) rich B cell lymphoma
- Lymphomatoid granulomatosis-type large BCL
- Primary effusion lymphoma
3DLBCL - pathogenesis
- Rare for DLBCL to have unmutated Ig genes
- Suggests they arise from cells that are GC or
post-GC - Heterogeneous cytogenetic abnormalities
- Translocations involving bcl-6 gene (3q27) are
seen in 35 of cases - block the downregulation of bcl-6 that normally
occurs during differentiation into post-germinal
center cells. - Leads to repression of genes that induce
cell-cycle arrest, apoptosis and differentiation.
- Translocations involving bcl-2 gene, such as
t(1418), seen in lt30 - These cases may evolve from follicular lymphoma
- Can see overexpression of bcl-2 even without
translocations. Correlated with poor prognosis
(indep of IPI) in some studies.
4DLBCL clinical behavior
- Wide variation
- Most common presentation is rapidly enlarging
lymphadenopathy - However, up to 40 can present with extranodal
sites (such as GI, CNS, bone, skin) - Can see unusual presentations such as in
intravascular, primary effusion - 25 stg I, 25 stg II, 50 stg III/IV
- gt10cm masses in ¼ to 1/3 of patients
- BM involved in approx 15
5DLBCL prognosis highly variable
- Overall about 40-50 have prolonged survival
- International Prognostic Index
- Age ( 60 vs gt60)
- LDH (normal vs elevated)
- PS (0-1 vs 2-4)
- Stage (I/II vs III/IV)
- Extranodal involvement (0-1 sites vs gt1 site)
- Low (0-1 factors, 35 of pts) 73 5 yr OS
- Low Intermed (2 factors, 27 of pts) 51
- High Intermed (3 factors, 22 of pts) 43
- High (4-5 factors, 16 of pts) 26
6OS by IPI risk group
NEJM 1993
However, despite this, risk stratified therapies
for high risk patients have not emerged based
on IPI score (or other adverse features).
7Hetergeneity at a molecular level also
- Staudt et al were able to classify DLBCLs into 2
different categories using DNA microarrays - Germinal Center B-cell like (GCB)
- and
- Activated B-cell like (ABC)
Nature, 2000
8CD138
MUM1
bcl-6
CD10
large non-cleaved
small cleaved
MZL
DLBCL (ABC type)
(GC type)
B-CLL
Somatic hypermutation of Ig genes
9Hetergeneity at a molecular level also
- Some of the GCB genes were already known to be
present on GC B-cells by immunohistochem CD10,
CD38, bcl-6 - Bcl-6 levels did not correlate with presence of
translocation, suggesting that it was more
related to GC phenotype and not translocation - ABC genes of interest
- Bcl-2 (again level did not correlate with
translocation) - MUM1 previously known to be translocated to Ig
locus in MM cases - transiently induced during normal lymphocyte
activation and is critical for the proliferation
of B lymphocytes in response to signals from the
antigen receptor
10Molecular subtype correlated with survival
(No validation set, small number of tumors)
Nature, 2000
11Molecular subtype can predict survival
- Rosenwald et al, took 240 DLBCL specimens,
identified GCB, ABC and Type 3 groups
NEJM, 2002
12Again GCB subtype correlated with better survival
NEJM, 2002
13Can a simplified microarray score predict
survival?
- Rosenwald et al took their 160 of their 240 DLBCL
specimens, and identified 17 genes that appeared
particularly predictive of survival - Specifically chose genes from different
biological pathways - GCB, MHC class II, lymph node groups
(favorable) - proliferation, BMP-6, groups (unfavorable)
- expression of each of the 17 genes then
incorporated into a score
14microarrays predict survival
NEJM, 2002
15Microarray predicts survival independent of IPI
score
NEJM, 2002
16Shipp et al could not predict based on cell of
origin
Later, Staudts group was able to resolve this
using their platform and probesets on these
samples
17Moving toward a logistically easier assay
- R. Levys group (Stanford) identified 36 genes
from the literature to test in a qRT-PCR assay to
predict survival in a group of 66 patients
NEJM, 2004
18NEJM, 2004
19NEJM, 2004
20Can DLBCL be classfied into GC vs non-GC using
IHC instead?
- Gene array is expensive and not logistically
feasible for real-time processing of clinical
specimens - qRT-PCR in theory could have a quick turn around
but may not if samples need to be batched, etc. - Technical issues of appropriate controls,
normalization to housekeeping gene - May be possible to classify cell of origin of
DLBCL using limited number of IHC markers - Hans et al took 152 DLBCL cases, 142 of which had
already been subjected to gene expression
profiling, and made tissue microarrays - Subjected the tissue microarrays to staining
using various antibodies - Were able to classify specimens as GC vs non-GC
using 3 antibodies
21CD138
MUM1
bcl-6
CD10
large non-cleaved
small cleaved
MZL
DLBCL (ABC type)
B-CLL
(GC type)
Somatic hypermutation of Ig genes
22CD138
MUM1
bcl-6
CD10
large non-cleaved
small cleaved
MZL
MZL
DLBCL (ABC type)
(GC type)
B-CLL
(GC type)
B-CLL
23IHC using 4 markers mirrors gene array data
Blood, 2004
24IHC may be even better than gene array
Blood, 2004
These behaved more like non-GCB
IHC allows for direct evaluation of neoplastic
cells. Gene array and RT/PCR analyze cDNA from
the entire biopsy.
25No obvious clinical differences for GC vs non-GC
26Other markers may provide additional prognostic
value
Cyclin D2 only in non-GC cases. Esp poor
prognosis.
Bcl-2 not predictive on its own, but identifies
poor risk group of non-GC cases.
Other markers that may predict survival FoxP1
27Confirmation from other groups
GC
IPIgt2, bcl-2, and non-GC
28Rare subtypes of DLBCL
- Primary cutaneous DLBCL of the leg
- Primary breast DLBCL
- Both have relatively poor prognoses
- Both have recently been shown to be non-GC or
ABC subtypes
29Limitations of the data
- Other factors besides cell of origin will
impact prognosis / response to therapy - Liver and kidney function
- Enzymes that may affect chemoRx metabolism or
concentration in tumor cells - Enzymes involved in DNA repair
- Surrounding stromal cells
- Tumor volume / burden
- Presumably why IPI predicts independent of cell
of origin - The patients in these studies were treated with
different regimens - Identification of molecular subgroups may
identify treatments that are particularly
effective for certain patients (e.g. Rituxan for
bcl-2 DLBCL)
30(No Transcript)
31ABC
ABC
GC
32Summary
- Gene array studies have led to a molecular scheme
to classify DLBCL by cell of origin - This scheme has prognostic value, beyond that
provided by strictly clinical parameters such as
IPI score - This classification can be achieved with a
limited number of IHC stains, without the need
for extensive gene array analysis - Similar classification can be achieved using
RT-PCR - Specific cell signalling pathways are being
identified in molecular subtypes of DLBCL
33Future Directions
- Need to prospectively obtain gene array data from
groups of patients treated with the same regimens - CALGB 50303 (pending) R-CHOP vs DA R-EPOCH
- Gene array studies will be done on all patients
- Prospective validation of GC vs non-GC
classification using IHC - ? incorporation of additional markers such as
cyclin D2 - Testing of intensified therapies in a prospective
manner in high risk patients - Identification of additional specific pathways /
targets in molecular subtypes of DLBCL - e.g. NF-kB inhibition strategies in non-GC /
ABC DLBCL
34Blood, 2004
CD10 membrane-associated neutral endopeptidase
expressed in several tissues. In LN, expression
limited to GCs. Bcl-6 zinc finger
transcriptional repressor. Expressed in GCs and
a subset of T cells. MUM1 lymphoid specific
member of the IFN regulatory factor of TFs.
Normally expressed in plasma cells and a small
subset of GC cells.
35(No Transcript)
36(No Transcript)
37(No Transcript)