Prioritizing New Agents for Pediatric Oncology Evaluation Malcolm A. Smith, MD, PhD Cancer Therapy Evaluation Program National Cancer Institute May 2007

How Are New Agents Prioritized for Evaluation in Children with Cancer? n Novel mechanism of action n Pre-clinical data model systems High levels of activity in adult pre-clinical modelsHigh levels of activity in adult pre-clinical models Activity against pediatric models (e.g., rhabdomyosarcoma xenograft)Activity against pediatric models (e.g., rhabdomyosarcoma xenograft) n High degree of activity observed in trials in adult patients

Why Is Prioritizing Targeted Therapies so Important?? Old approaches to improving outcome are not sufficient –Maximum treatment intensities have been reached for patients with high risk diagnoses –Outcome improvements have slowed for some diagnoses The opportunity costs of picking the wrong new agents for definitive evaluations –“Costs” to individual patients –The loss of 5 to 10 years of opportunity for improving outcome, which may occur if agents that are ineffective are selected for definitive evaluations.

Number of Phase 3 Trials Initiated in past 20 Years for Selected Childhood Cancers Ewing sarcoma –INT-0091 (1988) –INT-0154 (1995) –AEWS0031 (2001) –AEWS0531 (2008) High Risk Neuroblastoma –POG-8741 (1987) –CCG-3891 (1991) –POG-9340/42 (1993) –A3973 and ANBL0032 (2001) –ANBL0532 (2007) Rhabdomyosarcoma –IRS-III (1984) –IRS-IV (1991) –IRS-V (1999) –ARST0531 (2006)

Why Has the Task Been So Challenging?? Relatively small patient populations make it difficult to obtain reliable answers about the effectiveness of new treatments The menu of molecularly targeted agents is large and is driven by adult cancer drug development Most agents have some rationale for pediatric evaluation, but available preclinical data have often been fragmentary for the role of the agents in modulating growth/survival of specific childhood cancers No uniformly accepted standards for prioritizing agents for evaluation

Examples of New Agents under Investigation (or recently studied) for Children with Cancer AGENTMOLECULAR TARGET BortezomibProteasome 17-AAGHeat shock protein 90 Erlotinib & gefitinibEGF receptor Vorinostat & depsipeptideHistone deacetylase LapatinibEGF receptor & ErbB-2 Bevacizumab & AZD2171 & VEGF TrapVEGF pathway Sunitinib & sorafenibMulti-targeted kinase DecitabineDemethylating agent IspinesibKSP inhibitor Campath-1HCD52 EpratuzumabCD22 DasatinibSrc-Abl inhibitor ImatinibBcr-Abl & other tyrosine kinases IMC-A12, R1507, SCH , etc.IGF-1R Everolimus, temsirolimusmTOR Lenalidomide (Revlimid)Angiogenesis & others

How to Keep Focus of Pediatric Drug Development “child focused”?? Agents developed by pharmaceutical sponsors for adult cancers may (or may not) have limited relevance in the pediatric setting. A more systematic preclinical data driven prioritization process for introducing novel agents into the pediatric oncology setting may help to ensure that selection of agents for pediatric evaluation remains a “child focused” process

Relevant Data for Prioritizing an Agent for Pediatric Evaluation RNA and protein expression pattern consistent with the oncogenic role of the agent’s putative target Genetic evidence in childhood cancers for oncogenic role for putative target Genetic evidence in preclinical models for oncogenic role for the putative target Agents directed against the putative target show in vitro activity within relevant concentration ranges in appropriate cell line models Agents directed against the putative target show in vivo activity against relevant preclinical models Adult activity of agent and number of agents in class already being studied in the pediatric setting

IGF-1 Receptor in Ewing Sarcoma

Ewing sarcoma xenograft EW5

Building upon Molecular Characterization Discoveries in the Childhood Cancer Setting A novel oncogenic fusion gene is found in the majority of pilocytic astrocytomas (AACR 2008) – David T W Jones, Sylvia Kocialkowski, Lu Liu, Danita Pearson, L. Magnus Backlund, Koichi Ichimura, V. Peter Collins. University of Cambridge, Cambridge, United Kingdom, Karolinska Institute, Stockholm, Sweden –The novel rearrangement produces an in-frame fusion gene encoding the kinase domain of B-Raf coupled to the N-terminus of a previously uncharacterized gene –The fusion has constitutive kinase activity –The alteration was not present in > 200 WHO Grade III/IV astrocytomas

Building upon Molecular Characterization Discoveries in the Childhood Cancer Setting Identification of ALK as the major neuroblastoma predisposition gene (Mosse, et al. AACR 2008) Heritable mutations in the tyrosine kinase domain of the ALK proto-oncogene are the major genetic determinants of familial neuroblastoma, Somatically acquired alterations in this gene also occur frequently (~10% of high-risk cases)and may contribute to a more aggressive clinical phenotype. ALK is strong candidate target for evaluating in a subset of children with high-risk neuroblastoma.

Relevant Data for Prioritizing an Agent for Pediatric Evaluation RNA and protein expression pattern consistent with the oncogenic role of the agent’s putative target Genetic evidence in childhood cancers for oncogenic role for putative target Genetic evidence in preclinical models for oncogenic role for the putative target Agents directed against the putative target show in vitro activity within relevant concentration ranges in appropriate cell line models Agents directed against the putative target show in vivo activity against relevant preclinical models Adult activity of agent and number of agents in class already being studied in the pediatric setting

Setting the Bar for Declaring Success Does the in vivo experiment below indicate activity against an intracranial xenograft model for glioblastoma? 24 days to event for controls and 29 for treated animals (Mol Cancer Ther 2007;6(3). 2007) Setting an appropriate bar for declaring a “hit” for in vivo testing is critical: –Lowest bar is “significant difference” between control and treated groups, and progressive disease often meets this bar – More relevant bar in pediatric oncology is tumor regression and “cures”

Tumor Regression Is the Gold Standard in Pediatrics Rh28 (alveolar rhabdomyosarcoma) response to vincristine ALL-17 (B-precursor ALL) response to vincristine

Tumor Regression Is the Gold Standard in Pediatrics Cyclophosphamide single agent activity against PPTP in vivo panel. Comparable activity being observed for novel agents against selected tumor panels. Testing results available at

Preclinical Data Driven Prioritization of Combinations for Evaluation Single agent activity generally prompts incorporation of agent into regimen with standard therapy. –Rituximab for adult and pediatric NHL What if agent has limited single agent activity? –Preclinical demonstration of potentiation of activity of standard agents –Preclinical demonstration of synergistic effects in vitro with in vivo confirmation of supra-additive effects at tolerable doses

Rh18 Rh18 Combination Treatment with Rapamycin and Cytoxan Rh18 Rh30

Relevant Data for Prioritizing an Agent for Pediatric Evaluation RNA and protein expression pattern consistent with the oncogenic role of the agent’s putative target Genetic evidence in childhood cancers for oncogenic role for putative target Genetic evidence in preclinical models for oncogenic role for the putative target Agents directed against the putative target show in vitro activity within relevant concentration ranges in appropriate cell line models Agents directed against the putative target show in vivo activity against relevant preclinical models Adult activity of agent and number of agents in class already being studied in the pediatric setting.

Drug Development Pyramid Phase I Studies Phase II Studies And Pilot Studies Phase III Studies

Clinical Trial “Niches” in Pediatric Sarcoma Setting Relapsed, multiple Single agent or multi-agent phase 2 Phase 1 Recurrence 1 st (or 2 nd ) Multi-agent single arm phase 2 / pilot (multi-agent) Multi-agent randomized phase 2 (selection design or “screening” design) Potential for single agent “window” evaluation prior to multi-agent therapy Newly diagnosed, metastatic Single arm pilot (multi-agent) patients per year Newly diagnosed, localized Phase 3, most often with concurrent randomization patients per year (depending upon how patient population is “risk categorized”)

Conclusions A more systematic preclinical data driven prioritization process may help ensure that selection of agents for pediatric evaluation remains a “child focused” process. Relevant data to use for prioritization include: –RNA and protein expression pattern consistent with the oncogenic role of the agent’s putative target –Genetic evidence in childhood cancers for oncogenic role for putative target –Genetic evidence in preclinical models for oncogenic role for the putative target –Agents directed against the putative target show in vitro activity within relevant concentration ranges in appropriate cell line models –Agents directed against the putative target show in vivo activity against relevant preclinical models –Adult activity of agent and number of agents in class already being studied in the pediatric setting By assembling relevant preclinical datasets for agents entering pediatric evaluation today, the prioritization process may be able to be refined and improved in the coming years.