Targeting Tumor Angiogenesis: What Have We Learned So Far? Lee M. Ellis, MD Departments of Cancer Biology and Surgical Oncology UT MD Anderson Cancer Center Houston, Texas, USA Tuesday June 8 Education Session Tumor Biology

Overview: What Have We Learned So Far? Current State in the Clinic –Where have we done well and where have we failed Potential Mechanisms of Action of Anti-VEGF Therapy Biomarkers

VEGFR-1 (Flt-1) NRP-1/ NRP-2 VEGF-DVEGF-C VEGF-B VEGF-A PlGF VEGFR-2 (Flk-1/KDR) VEGFR-3 (Flt-4) Vasculogenesis Angiogenesis Lymphangiogenesis BEVACIZUMAB* VEGF-TRAP 18F1 1121B TG-403 Tyrosine Kinase Inhibitors Sunitinib* Sorafenib* Pazopanib* Axitinib Motesanib Cedirinib Brivinib Many, many others VEGF Targeted Agents in the Clinic or In Clinical Trials Ellis, Hicklin Nat Rev Ca * FDA approved agents

The Scorecard: Phase III Trials Chemo +/- Anti-VEGF Rx in Solid Malignancies TrialDisease SiteLine of Therapy 1 o Endpoint Met  RR Anti- VEGF Rx  PFS mos 5FU/LCV +/- SU5416mCRCFirst LineNo?? IFL +/- BVmCRCFirst LineYes10%4.4 FOLFOX +/- PTK/ZKmCRCFirst LineNo-4%0.2 XELOX/FOLFOX +/- BVmCRCFirst LineYes01.4 FOLFOX +/- BVmCRCRefractoryYes14%2.6 FOLFOX +/- PTK/ZKmCRCRefractoryNo?1.5 FOLFIRI+/- SunitinibmCRCFirst LineNo?? Chemo +/ Cedirinib vs Chemo + BV** mCRCFirst LineNo?? Chemo +/- BV/BVAdjuvant CRC-NoNA 5FU/Cisplatin +/- BVGastricFirst LineNo91.4 ** head to head comparison BV=Bevacizumab, RR= response rate, PFS=Progression free survival 06/10

The Scorecard: Phase III Trials Chemo +/- Anti-VEGF Rx in Solid Malignancies TrialDisease SiteLine of Therapy 1 o Endpoint Met  RR Anti- VEGF Rx  PFS mos Paclitaxel +/- BVmBreast CaFront LineYes22%5.9 Docetaxel +/- BVmBreast CaFront LineYes11% Capecitabine +/-BV*mBreast CaRefractoryNo10%0.7 Chemo +/- BVmBreast CaFront LineYes12-13% Chemo +/- SunitinibmBreast CaFront LineNo?? Chemo +/- SunitinibmBreast CaRefractoryNo?? Carbo/Paclitaxel +/- BV +/- BV OvarianFirst LineNo/Yes?4 Carbo/Paclitaxel +/- BVMelanomaFirst Line (Phase 2) No1.4 06/10

The Scorecard: Phase III Trials Chemo +/- Anti-VEGF Rx in Solid Malignancies TrialDisease SiteLine of Therapy 1 o Endpoint Met  RR Anti- VEGF Rx  PFS mos Gemcitabine +/- BV*Pancreatic CaFirst LineNo1%0 Gemcitabine +/- AxitinibPancreatic CaFirst LineNo?? Gem +/- VEGF-TrapPancreatic CaFirst LineNo?? Gemcitabine/Erlotinib +/- BV* Pancreatic CaFirst LineNo?? Chemo +/- BVProstate CaHormone- refractory No?? Carbo/Paclitaxel +/- BVNSCLCFirst LineYes15%1.9 Gem/Cis +/- BVNSCLCFirst LineYes10-14% Carbo/Paclitaxel +/- Sorafenib NSCLCFirst LineNo3-0.8 BV=Bevacizumab, RR= response rate, PFS=Progression free survival 06/10

The Scorecard: Phase III Trials “Standard of Care” vs Single Agent Anti-VEGF Rx in Solid Malignancies TrialDisease SiteLine of Rx1 o Endpt Met  RR Anti- VEGF Rx  PFS mos Sorafenib vs PlaceboRCCRefractoryYes8%2.7 Sunitinib vs IFNRCCFirst LineYes31%5.9 IFN +/- BVRCCFirst LineYes18%4.8 Pazopanib vs PlaceboRCCBothYes27%5.0 Sorafenib vs PlaceboHCCFirst LineYes2%3.0 Sorafenib vs PlaceboHCCFirst LineYes2%1.4 Sunitinib vs Sorafenib**HCCFirst LineNo?? Sunitinib vs PlaceboPancreatic NET First LineYes9%5.9 ** head to head comparison

What Have We Learned So Far? The efficacy of VEGF-targeted therapy is –Tumor specific –Context specific –Agent specific This is not a simple field to understand –You cannot make broad generalizations across drugs, tumor types, or stage of tumors

What Have We Learned So Far? Despite similar primary targets (VEGFRs), all drugs in this class are NOT created equal –If we “hit” too many targets, is this a bad thing? CAIRO-2 and PACCE Trials Activation of compensatory pathways?

Kinase Inhibition Profiles of VEGFR Targeted Agents Bevacizumab VEGF-Trap Karaman et al. Nat Biotech 2008

Phase III Trial Summary: TKIs vs MoAB (Bev) Excluding head to head comparisons and continuation studies DiseaseMoAB (Bev)TKI Pancreatic0/2 (VEGF Trap 0/1)0/1 Melanoma0/1NA Gastric0/1NA Prostate0/1NA CRC3/30/4 Breast3/40/2 NSCLC2/20/1 HepatocellularNA2/2 Renal Cell1/13/3 Pancreatic NETNA1/1 Total Percent9/16 (56%)6/14 (43%) Excluding HCC/RCC/pNET 8/15 (53%)0/8 (0%) Highly Angiogenic Tumors

MoABs vs TKIs Differences in Specificity Half-life/exposure (metabolism) Tissue Penetration Despite many hypothetical discussions about the advantages of TKIs regarding their ability to target all 3 VEGFR TKRs, and PDGFR, etc., in the tumor types that are NOT highly angiogenic, we cannot ignore the clinical data. My personal view, backed by some preclinical data, is that targeting PDGFR does not add to efficacy, and in some ways, may decrease efficacy

ASCO 2009/2010: For Solid Tumors “Requiring” Multiagent Therapy, Single Agent Bevacizumab May Demonstrate Some Activity as “Maintenance Therapy” But the question remains, how much benefit will patients receive CRC Ovarian Cancer TumorTrialBenefit Observed Remaining Questions CRCNSABP C-08/adjuvantSeparation of curves while Bev administered as single agent Is it feasible to give Bev for an extended period (yrs) in a population where 97% may not benefit from Rx? mCRCMACRO/maintenance Bev following XeLOX/Bev vs XeLOXBev continuation Non-inferior to XeLOX/Bev after first 6 cycles Too few patients to draw firm conclusions of non-inferiority OvarianECOG/maintenance Bev following C/T/Bev vs C/T alone Improved PFSOS not improved - is this due to access to Bev after progression or does this fit with other Bev trials where PFS does not always translate into OS? Eisenhower discussion

Validation, Validation, Validation! Single agent maintenance therapy is not FDA-approved. We must study this concept in more clinical trials before changing practice.

Overview: What Have We Learned So Far? Current State in the Clinic Potential Mechanisms of Action of Anti-VEGF Therapy Biomarkers

Summary of PFS and RRs with VEGF-Targeted Therapies Cancer TypeHow Utilized or Studied Increase PFS Over Standard Care  Increase RR  FDA Approval CRC+ Chemo0-4 months0-10%Y NSCLC+ Chemo0-2 months3-15%Y Breast Cancer+ Chemo1-6 months10-22%Y Renal Cell CaSingle agent3-6 months8-30%Y GBM (Phase 2)Single agent1-2 months15-20%Y HCCSingle agent1.4-3 months2%Y OvarianAfter chemo/Bev 4 months?pending Prostate+ Chemo??N Pancreas+ Chemo00-1%N Melanoma+ Chemo %N

With Such Varied Results With VEGF-Targeted Therapies, There Must be Multiple Mechanisms of Action of This Class of Drugs

Potential Mechanisms of Action of VEGF Targeted Therapies (When it works) Anti-angiogenic “Normalization” of the vasculature Direct effect on tumor cells Vascular “constriction” Offset effects of stress Reverse immuno-suppression due to VEGF Disruption of the cancer stem cell niche

Anti-VEGF Therapy: My Theory on Different Mechanisms of Action in Different Tumor Systems Renal Cell Carcinoma Colon Carcinoma

Overview: What Have We Learned So Far? Current State in the Clinic Potential Mechanisms of Action of Anti-VEGF Therapy Biomarkers

Prognostic marker –associated with clinical outcome, independent of specific treatment Predictive marker –identifies groups receiving different degrees of benefit from a therapy –specific for a given treatment (i.e., Ras for EGFR MoABs in CRC, HER2 expression for trastuzumab in breast cancer) Surrogate / activity marker –modulated by treatment –may be on target tissue (tumor) or surrogate tissue (i.e., lymphocytes, skin) –might or might not correlate with clinical activity A Few Definitions A “marker” is not a “marker” is not a “marker”

We Do Not Have Any Predictive Biomarkers!!!! Comment: We should never subject patients to biopsies for any biomarker study until we have shown proof of principle in preclinical models.

A Snapshot of Predictive Marker Studies for VEGF Targeted Therapies Negative (mostly unpublished) –VEGF Tissue, plasma, anywhere –TSP –VEGFRs –Ras, Raf, P53 Possible/requires further study –SNPs (VEGF, VEGFR-1) –Imaging of vascular flow and function in metastasis Yao et al. NETs, Saturday A predictive marker must be strong enough to allow a “go/no go” for a particular drug for a particular patient Would you deny a patient therapy based on a “negative” biomarker? - Mut Kras for EGFR MoABs in mCRC

ASCO 2010 “Conclusions: Measurement of baseline circulating VEGF levels may be useful as a prognostic biomarker, but not as a predictive biomarker for bevacizumab-based treatment benefit in metastatic colorectal, lung, and renal cell cancers. Circulating vascular endothelial growth factor (VEGF) as a biomarker for bevacizumab-based therapy in metastatic colorectal, non-small cell lung, and renal cell cancers: Analysis of phase III studies. C. Bernaards, et al. Despite innumerable studies of circulating VEGF and it’s lack of validity as a predictive biomarker, invesigators will continue to study this until we get a false positive.

Schneider, et al. J Clin Oncol; 2008 Kaplan-Meier curve for overall survival (OS) in experimental arm by genotype; (A) vascular endothelial growth factor (VEGF) C/A; (B) VEGF-1154 G/A SNPs: VEGF Polymorphisms and Predictive Value in ECOG-2100 (Pac +/- Bev Metastatic Breast Cancer) Caveats: Predictive for OS, but not PFS Small numbers Did not include Pac Rx alone group

VEGFR-1 SNPs in Pancreatic Cancer Bev Group Lambrechts et al. ESMO 2009 SNPs are exploratory and require unbiased validation in larger studies! We should do our best to associate SNPS to biology.

Prognostic marker Predictive marker Surrogate / activity marker –modulated by treatment –may be on target tissue (tumor) or surrogate tissue (i.e., lymphocytes, skin) –might or might not correlate with clinical activity A Few Definitions A “marker” is not a “marker” is not a “marker” From a practical perspective, a surrogate marker is only of use to an oncologist if it is strong enough to force an oncologist to alter Rx.

After a Decade of Studies and MILLIONS of $$ Spent in Search of Biomarkers for Anti- angiogenic Therapy George Sledge ~$40-50

Grade 3/4 Hypertension Is Associated With Improved Median OS in E2100 Median OS: 25.3 mo vs mo p=0.002 Schneider et al; J Clin Oncol, 2008

Hypertension is a Biomarker of Efficacy in Patients with Metastatic Renal Cell Carcinoma Treated with Sunitinib Brian Rini et al (2010)

Dahlberg et al. J Clin Oncol; 28: 2010 Overall Survival and HTN After One Cycle of Therapy: ECOG 4599 High blood pressure (HBP) by the end of cycle 1 was defined as > 150/100 at any previous time or at least a 20-mmHg increase in diastolic blood pressure from baseline.

ASCO 2010 Analysis of early hypertension (HTN) and clinical outcome with bevacizumab (BV). H. Hurwitz, et al. “…The primary HTN endpoint was an SBP increase 20 mmHg or DBP increase 10 mmHg within the first 60 days of Tx. …. Conclusions: HTN during Tx does not predict clinical benefit from BV based on PFS or OS.….”

What We Know (and don’t know) in 2010 About VEGF Targeted Therapy What We (Think) We Know –Single agent anti-VEGF therapy is effective in the most angiogenic tumors RCC, HCC, GBM (NETs?) And….it may have utility as “maintenance therapy” (Ovarian, colorectal cancers- ASCO 2010) –Not all anti-VEGF therapies are created equal This is both good and bad –The benefits of therapy at times may be statistically significant, but clinically marginal What We Do Not Know (but still think we do know) –How it works –The best agents to partner with VEGF- targeted therapies –Biomarkers to select patients (SNPs?) –If it will work in the adjuvant setting (presumed micromets) in diseases other than CRC (C-08 -) –Long term effects of VEGF inhibition on the CV system and CNS (relevant to adjuvant studies) –If it will work through multiple lines of therapy (CRC trials)

Despite Incremental Gains in Many Tumor Types, We Have Not Done As Well as We Had Hoped. We Must Be More Creative in Drug Combinations and Biomarker Discovery!! It is time to move new approaches forward!! “Me too” drugs and trials are unlikely to significantly advance the field