Clinical Cancer Genetics

download report

Transcript Clinical Cancer Genetics

Clinical Cancer Genetics
• Our understanding of the genetics of
cancer is beginning to help us design better
• This same understanding, however, has
long (>30 years) been helping us make
decisions about diagnosis and prognosis.
• This is getting better all the time (we’ll talk
about why).
Diagnosis and Prognosis are Important
• Helps tailor therapy (e.g. small cell lung
cancer vs. non-small cell lung cancer).
• Helps tailor therapeutic intensity (e.g.
acute leukemia)
• Helps guide follow-up in patients who
are NED (no evidence of disease, we
never say “cure”).
• Helps patients live their lives.
Cancer Is:
• Inappropriate proliferation and
resistance to differentiation and
apoptosis (Rb, p53, PTEN).
• Genomic instability (p53, MIN vs. CIN,
short telomeres?, others).
• Ability to grow where it ought not (i.e.
metastasis—RAS?, molecular
mechanisms not clear).
p53 status is often only of weak
prognostic value:
Pancreatic cancer survival
Why are the fundamental lesions of
cancer not so good at prognosticating?
1. Technical details (e.g. p53 is hard to measure,
multiple non-equivalent lesions, etc).
2. To be of clinical value, a prognostic variable
has to be really good (easy to determine,
cheap to measure, reproducible, etc etc).
3. Most interesting scientifically: these lesions are
the sine qua non of the cancers themselves.
Comparing apples with apples:
Proliferation /
Aggessive growth
• RAS mutations (20%) not prognostic in
melanoma (1993).
• Almost all (>90%) melanoma has a
RAS or RAF mutation (2002).
• But did we learn something—yes, BRAF inhibitors might make an excellent
melanoma therapy.
If the obvious candidates don’t work, what
1. Things that can be measured:
2. Things that help discern very dissimilar
entities (e.g. the pathologist’s friend).
3. Things that we identified empirically.
The “small round blue cell tumor”
• Classic diagnostic dilemma: poorly
differentiated, rapidly growing tumors of
small children.
• Tumor site, age of child, certain blood
tests helpful (but none are perfect).
• Treatments and prognosis are totally
different (and it could be one of four
What would you do when faced with sick child,
frightened parents, unsure pathologists (not to
mention greedy malpractice attornies, etc.)?
Ewing’s sarcoma: Surgery, chemo, XRT—do OK
Burkitt’s lymphoma: Chemo—do great.
Rhabdomyosarcoma: Surgery, chemo—do so-so.
Neuroblastoma: Surgery, chemo, XRT (or nothing)—do so-so.
First three are uniformly fatal if not (or mis-) treated
How does one decide?
• Cytogenetics:
– t(11;22) = Ewing’s
• Specific Translocations:
– IgH-Myc = Burkitt’s
• Certain amplifications and deletions
– N-myc = neuroblastoma
• Gene expression (by immunostaining)
– Desmin, Myf = rhabdomyosarcoma
What would you do…
•Burkitt’s lymphoma:
•Hi-Dose chemotherapy
•Intrathecal chemotherapy
(by serial lumbar
•Ewing’s sarcoma:
•Surgical womp.
•Different Hi-Dose
•XRT post chemo
•Excellent prognosis
•Good prognosis
• The grand-mother of cancer genetic tests
(Philadelphia chromosome was identified as
“mini-chromosome” in AML in 1960, = t(9;22)
in 1973).
• Done by culturing tumor cells, arresting them
in mitosis, and making metaphase spreads.
• Chromosomes are stained and interpreted by
a cytogeneticist.
• Takes days to > 1 month, often not that
sensitive (many tumors don’t grow in vitro).
Cytogenetics are useful:
• t(9;22) makes bcr-abl
fusion protein.
• Correlates with bad
prognosis in ALL.
• Molecular target of
Gleevec (and predicts
Gleevec response).
• Can be followed as
marker of response (socalled ‘molecular CR’).
Pediatric Acute Lymphoid
Leukemia, 5-year survival
>50 chromosomes
40-50 chromosome ~80%
Ph+ ALL gets an up-front BMT,
other kids get a trial of
Cytogenetics and Prognosis
• Can signify prognosis that is:
– Good: iso12p in mediastinal “carcinoma of
unknown primary” = germ-cell tumor
– Average: 46XX (i.e. normal) in AML
– Bad:
Ph+ in ALL; 7q- in AML
Complex karyotype in solid tumors
• The oncologists’ easy to recall rule to
cytogenetics: if the report goes more than
one page, the prognosis is poor.
• Important Observation: pediatric cancers
tend to have simple cytogenetics, while adult
cancers are more complex
Cytogenetics 2003: Chromosome painting
and Spectral karyotyping (SKY)
Specific Paints
(DAPI counterstain)
Chromosomal Translocations
• Replacing cytogenetics in many areas
(EWS-FLI, BCR-ABL, etc.) when the
target lesion IS KNOWN.
• Usually identified by PCR (DNA), rarely
RT-PCR (RNA…remember, has to be
easy to do).
• Have begun to be used widely for
assesing ‘minimal residual disease.’
Minimal Residual Disease
• 42 year old man with very high white blood
cell count, anemia, high platelets.
• Smear shows lots of well-differentiated
myelocytes (WBCs), some basophils.
• CML (chronic myelogenous leukemia, always
Ph+ positive).
• Treated with hi-dose chemotherapy, total
body irradiation, and BMT.
• Cytogenetic remission in bone marrow at 6
months post-BMT.
PCR on the blood for BCR-ABL
BMT 1 month
3 month 6 months 6.5 months 12 months
Donor lymphocyte
infusions begun
• One problem: Low copy Bcr-Abl can be
found in ‘normal’ people at modest
frequency (carpe diem).
Minimal Residual Disease
# CA cells
BMT 1 month
3 months
6 months 6.5 months 12 months
• Probably 2-4 logs more sensitive than cytogenetics.
• Affords the opportunity to treat small numbers of tumor cells
that are clinically silent, but the cause of relapse
• Consolidation can be good old-fashioned chemotherapy
(HiDAC, stem cell transplants etc), much interest in novel
therapies (immunotherapy, angiogenesis inhibitors,
monoclonal antibodies, etc) in this setting.
Amplifications and Deletions:
• Done by cytogenetics, comparative
cytogenetic hybridization, Southern blot, LOH
assay, rarely quantitative PCR.
• Adult carcinomas characterized by wholesale
gains and losses.
• More of scienitific interest than clinical utility
• A few examples: N-myc copy # important in
neuroblastoma, 14q deletion adverse in
colon cancer
Loss of heterozygosity
Somatic cells
• How tumor
suppressor genes
have classically
been found.
• Gives related but
different result from
CGH (which
measures copy
Comparative Genomic Hybridization
1. Label tumor DNA with green chrome,
label normal DNA red and mix.
2. Hybridize metaphase spreads with
labeled DNA mix. Perform fluorescence
3. Areas of normal copy number are yellow,
tumor amps are more GREEN, dels are
more RED.
4. Computer sums results from several
CGH: Small cell lung cancer
From Charite, Humboldt University Berlin
Array CGH
Done in the same manner as conventional CGH,
except hybridize >1K feature microarrays of
mapped DNA fragments (e.g. BACs) instead of
metaphase spreads.
1 2
3 4 5 6
7 8 9 10 11 12
Assays of gene expression
• Currently, >95% is immunohistochemistry,
ELISA or flow cytometry (that is, antibodies
are used to stain the tumor).
• RNA methods are generally too unreliable for
widespread clinical use.
• RT-PCR is done in a few specific
circumstances (e.g. tyrosinase expression to
rule-in amelanotic melanoma)
IHC / ELISA / Flow
• Conjugated antibodies
bind cognate antigen (e.g.
CD3 on T-cells, neuron
specific enolase as
neuroendocrine marker)
• Ab binding detected by
fluorescence or chemical
reaction (e.g. horseradish
Cyclin E and Breast Cancer
• Cyclin E / cdk2 complex is major activity that
phosphorylates RB and leads to G1-traversal
(i.e. cell divides).
• Despite this important role in regulating
proliferation, the data that cyclin E are an
oncogene or important in cancer are mixed.
• Recent NEJM article (Keyomarsi et al. 2002)
suggests proper measurement of cyclin E
levels is of high prognostic value.
Cyclin E and Breast Cancer
• If correct, obviously we’d try most aggressive therapy in
patients with high cyclin E
• Problems: WB not simple to do in a clinically useful way.
• ?Publication bias (this single gene is better prognosticator
than 10,000 gene microarray).
Tissue Microarrays
• Little pieces of tumor
are embedded in
paraffin in microarray
• Each tumor piece is
hybridized to different
primary antibody
• Can assess hundreds
of IHCs
An interesting observation
about gene expression tests:
• Usually, we measure genes that are
pathologically unimportant (CD3, vimentin,
etc); to help determine tissue of origin.
• We are beginning, however, to have tests for
pathogenic molecules (e.g. Her2-neu in
breast cancer).
• Even better, some of these molecules are
good targets for humanized monoclonal
antibody therapy (e.g. anti-CD20 = Rituxan).
Enzyme assays
• Although not generally thought of as ‘cancer
genetics’; tumor enzyme assays are the
oldest clinically useful tests of gene
• In the old days, all leukemia was typed based
on enzymatic profiles, and myeloperoxidase
is still used to tell AML from ALL (although
now can be done using an antibody to MPO)
-Some intrepid member of the audience should now ask me about
Telomerase activity as tumor marker
• Telomerase is an RNA containing enzyme complex that
maintains the protective cap (‘telomere’) at the end of
• Telomerase activity can be measured by the TRAP assay
(requires intact RNA and PCR).
• Most cancers express telomerase.
• Normal somatic cells don’t express telomerase (so
presence of this activity has been proposed as an early
diagnostic test for cancer).
• Some tumors don’t express telomerase, but still maintain
telomere length (‘ALT’).
Telomerase as marker of pathogenic potential
•Tail vein metastasis assay: Inject tumor cell lines
intravenously into mice.
•Sacrifice animal and count tumors in the lung 6 weeks later.
•All lines are ‘malignant’, assay rather measures ability to
seed and grow in lung (?metastasis?).
+ALT +telomerase
Telomerase and Glioblastoma
• GBM is one of the worst
of all adult cancers
• Generally, 50% survival
measured in weeks.
• ALT+ (telomerase
negative) tumors may do
exceptionally well.
Time (years)
• Even if correct, does not necessarily mean
telomerase inhibitors will be of therapeutic value.
Cancer Genetics: the future
RNA expression profiling on
oligonucleotide microarrays
is capable of measuring the
expression of thousands of
genes in a tumor
Based on expression, one
can “cluster” like tumors and
optimize therapy.
RNA expression profiling
• mRNA from tumor is converted to DNA and
labeled; then hybridized to array.
• array is of oligonucleotides or complementary
DNAs (several versions of arrays at present).
• Arrays represent large numbers of genes
• Tumors are clustered by various statistical
methods (“unsupervised” vs. “supervised”).
• Hypothesis is that tumors in common clusters
will behave in a clinically similar manner.
From van de Vijver et al.
NEJM 2002
Metastasis-free survival:
What is the best microarray technology?
• RNA expression profiling: most mature in terms of clinical
applications. RNA-based assay. Lot of Big Pharma interest.
But…Lots of important molecules regulated at protein level
(e.g. p53), reproducibility issues.
• aCGH: Uses DNA, clinicians are very comfortable using
cytogenetic data for therapeutic decisions. “Normal” is
always known. Quantifies tumor genomic instability.
Sometimes the locus may be more important than the gene
(e.g. del 3p). But…less data than expression profiling
• Tissue microarrays: Hardest to do well, confounded by
tumor heterogeneity (I wouldn’t invest here…).
• Proteomics: Most technically difficult at present, but also
least mature. Can be done on the serum (instead of the
tumor). Immense promise.
A Caution:
• Medicine >< Science.
• Medicine (appropriately) is very
conservative and moves much more slowly
than science.
• Randomized trials are required to change
the standard of care (cost millions, require
years of follow-up).
• Pathologists will be doing IHC and
metaphase spreads 10 years from now.
Example I: Prostate Specific Antigen
• PSA identified as marker of prostate cancer in
• It is, far and away, the best “tumor marker.”
• Still, who, if anyone, should have screening
• What should you do in 70 y/o with high PSA? In
an 80 y/o?
• Clearly PSA has been a boon to radiation
oncologists and urologic surgeons, but still very
unclear if elderly men with indolent cancer
benefit from treatment.
Example II: Autologous stem cell
transplants in breast cancer
• In early 1990’s, several non-randomized trials (Phase
II) demonstrated impressive responses to high-dose
chemotherapy in breast cancer.
• Doses of chemo were so high, to survive patients
required reinfusion of their own hematopoetic stem cells
after chemo (so-called ‘stem-cell transplant’).
• In 1995, Bezwoda et al. reported a 90 patient study with
51% complete remission rate in metastatic breast
cancer with high-dose therapy and stem cell rescue (vs.
5% CR rate in women treated conventionally).
Example II: Autologous stem cell
transplants in breast cancer (cont.)
• Every large CA center in USA (and several small
ones too) began offering ASCT.
• Despite widespread skepticism and vastly increased
cost and toxicity, thousands of women were treated in
this way.
• MA required insurance companies BY LAW to pay for
ASCT in women with high risk breast cancer.
• 2001: Three large randomized trials showed no (or at
best very limited) benefit of ASCT.
• 2001: Bezwoda article was retracted after auditors
concluded the results had been FABRICATED.
Summary: Clinical Cancer Genetics
• In cancer care, good pathology is crucial and
• Clinical cancer genetics, in 2003, is comprised of
cytogenetics, detection of chromosomal
translocations and amps/dels, and limited assays
of gene expression (IHC, ELISA, flow).
• Crucial for diagnosis, therapy, prognostication,
and assessment of minimal residual dz.
• RNA expression profiling, aCGH, and tissue
microarrays hold great promise.
A little less conversation, a little more
If interested in cancer biology, now’s the time to hit
the lab (even if all you want to do is go to med
Ned Sharpless 966-1185
[email protected]