Transcript ASCO_2008_files/Rothenberg Discussant Tailored Tx in PC ASCo

Developing a Patient-Tailored Strategy
to Stratify and Treat Patients with
Pancreatic Cancer
Mace L. Rothenberg, M.D.
Professor of Medicine
Ingram Professor of Cancer Research
Conflict of Interest Disclosure
Consultant or Advisory Role
Antigenics
OSI
Array BioPharma
Pfizer
Bristol Myers-Squibb
Roche
Genentech
sanofi-aventis
Idera
Synta
ImClone
Takeda
Johnson & Johnson
Zymogenetics
Novacea
Stock Ownership
Synta
Targeted Therapeutics
Patient-Tailored Strategy for
Pancreatic Cancer Treatment
How Are These Presentations Related?
•
•
•
•
Development of a more robust preclinical model
Pharmacogenomics
Randomized Phase II clinical trial
Patterns of gene expression and amplification
•
Each looks at a different, and complementary
approach to the development of improved
therapy
Tumor genetics, host genomics, effect of
therapy in the mouse, effect of therapy in the
patient
•
Patient-Tailored Strategy for
Pancreatic Cancer Treatment
A prospective validation of a direct tumor
xenograft model in pancreatic ductal
adenocarcinoma (PDA)
Antonio Jimeno, Ignacio Garrido-Laguna, Maria
Uson-Olaso, Dante Trusty, Ralph Hruban, Richard
Schulick, John Cameron, Anirban Maitra, and
Manuel Hidalgo
Abstract #4500
Direct Tumor Xenograft Model in
Pancreatic Cancer
What Did They Do and Why is it Important?
• Obtained fresh pancreatic ductal carcinoma from the
operating room which were then implanted directly into
nude mice
• This maintains the genetic integrity of the original
tumor over multiple passages
• Evaluated response of tumor explants to drugs to
determine if this model has promise in predicting
• clinical response to the agent in the patient
• potential activity of investigational drugs against pancreatic
cancer
Direct Tumor Xenograft Model in
Pancreatic Cancer
Key Findings
• Explantation and propagation of human pancreatic
ductal adenocarcinomas in nude mice is feasible
• In a limited number of patients, information from this
model was be used to select treatment upon relapse
• In 4 patients, there was concordance between ex vivo
sensitivity and clinical response
• Concordance between ex vivo sensitivity and gene
expression patterns of tumor cell lines known to be
sensitive to gemcitabine and docetaxel
Direct Tumor Xenograft Model in
Pancreatic Cancer
Limitations and Challenges
• Only 20% of all consenting patients yielded successful
xenografts
• Model lacks certain elements that may be critical in
determining treatment response
• Immune system
• Tumor-stromal-vascular interactions
• Very few drugs have demonstrated greater activity than the
standard agent used to treat this disease: gemcitabine
• Clinical “successes” consisted of 2 pts with SD x 4 and 6
months
• Can PD at 100 days be considered a success?
• Poor correlation between ex vivo sensitivity and PFS with
adjuvant gemcitabine: R2 = .08
Patient-Tailored Strategy for
Pancreatic Cancer Treatment
Significant Effect of Single Nucleotide
Polymorphisms of Gemcitabine Metabolic
Genes on Pancreatic Cancer Survival and
Drug Toxicity
Milind Javle, Taro Okazaki, Robert Wolff, Chris Crane,
Linus Ho, Gauri Varadhachary, Douglas B. Evans,
James L. Abbruzzese, Donghui Li.
Abstract #4501
Correlation of Gemcitabine Metabolic
SNPs and Clinical Outcome
What Did They Do and Why is it Important?
• Identified SNPs in gemcitabine-associated
cellular transporter, anabolic and catabolic genes
• Correlated SNP patterns (haplotypes) with
therapeutic outcomes as well as patterns and
severity of toxicity
• This approach has the potential to identify
patients at increased risk of harm as well as
patients more likely to benefit from treatment
Correlation of Gemcitabine Metabolic
SNPs and Clinical Outcome
Key Findings
• Grade 3-4  ANC was increased 2-3 fold for those patients
with CT or TT genotype at position 111 in the cytidine
deaminase gene or the CC/CT genotype in position 1205 in
the deoxycytidine kinase gene.
• Longer survival was observed in those patients with the
same genotype in the cytidine deaminase gene or with an AA
or AG genotype in position 42 of the ribonucleotide
reductase M1 gene.
• Two models, one comprised of selected haplotypes of
metabolic genes and the other with selected haplotypes of
transporter genes, were associated different patterns of
survival
Correlation of Gemcitabine Metabolic
SNPs and Clinical Outcome
Limitations and Challenges
• Data generated from a relatively small dataset (120
patients)
• Combined data from 2 clinical trials with different
chemotherapy regimens
• Gemcitabine alone (54 patients)
• Gemcitabine + cisplatin (66 patients)
• All patients received radiation in addition to
chemotherapy
• Was the predictive value of these SNPs still seen in
multivariate analysis that included known clinical
prognostic variables such as tumor grade, baseline
CA19-9, + or - margin or LN?
Correlation of Gemcitabine Metabolic
SNPs and Clinical Outcome
Conclusions
• These findings should be validated in an independent
cohort involving a larger number of patients
• Agree (this has been the downfall of prior efforts)
• But does the model need further refinement first?
• Prospectively select SNPs based on biological
pathways (for gemcitabine and/or other drugs)
• Agree
Results are encouraging but more work needs to be
done before it is likely to have a major clinical impact
on patient management
Patient-Tailored Strategy for
Pancreatic Cancer Treatment
Final analysis of a randomized phase II trial of
bevacizumab and gemcitabine plus cetuximab
or erlotinib in patients with advanced pancreatic
cancer
Hedy Lee Kindler, Tara Gangadhar, Theodore Karrison,
Howard Hochster, Malcolm Moore, Kenneth Micetich,
Weijing Sun, Daniel Catenacci, Walter M Stadler, and
Everett E Vokes for the University
of Chicago Phase II Consortium
Abstract #4502
Addition of EGFR Inhibitors to Gemcitabine +
Bevacizumab in Advanced Pancreatic Cancer
What Did They Do and Why is it Important?
• Tried to build upon exciting early clinical data for
gemcitabine + VEGF inhibitor (bevacizumab) and
gemcitabine + EGFR inhibitor (erlotinib or cetuximab)
• Promising preclinical data on the effect of
simultaneous inhibition of multiple growth factor
signaling pathways
• Randomized Phase II design could help identify if
either of these regimens should be taken to Phase III
Addition of EGFR Inhibitors to Gemcitabine +
Bevacizumab in Advanced Pancreatic Cancer
Key Findings
• Primary endpoint of RR ≥ 35% for either arm was not met
• Is RR the best indicator of activity in this disease?
• Both regimens were relatively well tolerated
• Patients with better PS and locally advanced disease did
best
• Appearance of rash during treatment correlated with 
PFS for GBE but not for GBC group
• Early HTN may predict for response but not for PFS or OS
• Neither VEGF nor soluble VEGFR2 correlated with clinical
outcome
Addition of EGFR Inhibitors to Gemcitabine +
Bevacizumab in Advanced Pancreatic Cancer
Issues to Consider
• Interference between VEGF inhibitor and EGFR inhibitor
• biologically
• pharmacologically
• Presence of KRAS mutation in >90% of all pancreatic
cancers
• Likely to be a cause of resistance to EGFR inhibitors in CRC
and NSCLC
• But why was NCIC PA.3 study positive? (Death HR = 0.81, p
= .028 for combination of gemcitabine + erlotinib).
• Is there a definable subset of patients with PDAC who
benefit from an EGFR inhibitor?
Addition of EGFR Inhibitors to Gemcitabine +
Bevacizumab in Advanced Pancreatic Cancer
Two Alternative Interpretation of Results
Trial was a Failure
Trial was a Success
1. The trial didn’t meet
protocol-specified
endpoint for success
2. With >90% of PDAC
carrying KRAS
mutations, the lack of
benefit is not surprising
3. In mCRC, the addition of
an EGFR inhibitor to
chemotherapy and
bevacizumab doesn’t
improve outcomes
1. Requirement of a RR
≥35% was unrealistic
2. RR is not a reliable
indicator of efficacy in
PDAC
3. OS = 7-8 months
compares favorably with
gemcitabine (6 mo), gem
+ erlotinib (6.4 mo), gem
+ bevacizumab (5.7 mo),
and gem + cetuximab
(6.4 mo)
Patient-Tailored Strategy for
Pancreatic Cancer Treatment
Whole Genome Expression Analysis of
Pancreatic Adenocarcinoma Predicts
Prognosis after Surgery
Eric A. Collisson, R. Mori, A.C. Hoffmann, G.E.
Kim, R.Hajnal, P.V. Danenberg, J. Cooc, K.D.
Danenberg, M.A. Tempero
Abstract #4503
Whole Genome Expression
Analysis in PDAC
What Did They Do and Why is it Important?
• Tried to identify a genetic signature that
distinguished long from short survivors
following Whipple resection of PDAC
• Performed this on gross and microdissected
tumor
• Also evaluated tumors for differences in gene
copy number
Whole Genome Expression
Analysis in PDAC
Key Findings
• Expression pattern of PDAC differs dramatically
from NL pancreas
• Gene expression profile differences identified
from microdissected specimens were not
maintained when gross tumor was analyzed
• Only 3 of >700 genes overlapped suggesting
dilutional effect of stroma and acini
• Two genetic loci appear to be amplified
disproportionately in patients with PDAC and short
survival: 6p21.1-21.3 and 8p12 - 8p11.23 (LETM2)
Whole Genome Expression
Analysis in PDAC
Issues to Consider
• Was the genetic signature an independent predictive or
prognostic tool after taking known clinical factors into
account?
• Although adjuvant therapy is the only statistically significant
difference, there are notable imbalances in gender, and histologic
grade.
• The short surviving patients survived a median of 3
months; why so short?
• Does this analysis include death from any cause or just diseasespecific death?
• Will either model hold up in a larger, independent cohort?
• What are the biological consequences of the
overexpressed or amplified genes?