Marlene Pickl

Predictive markers in early research and companion diagnostic developments in oncology

Predictive biomarkers are discovered and used in oncology research to formulate hypotheses aimed at the identification of patients benefiting from specific therapeutic intervention(s). They pave the way to the development of companion diagnostic tests which are tools readily implemented in the clinic and serve to qualify a patient for treatment with a particular targeted drug or the continued use of a particular drug, thus maximizing the benefit to risk ratio of the medical intervention to the patient. Predictive biomarkers are defined by biological characteristics of the patients or tumor status that can be measured objectively and correlated with clinical outcome: these can be molecular, cellular or biochemical features. Predictive markers need extensive analytical validation - specific for the tool utilized for their assessment - as well as rigorous clinical qualification in the context of the drug treatment for which they define clinical utility. The process of companion diagnostic development is a highly interdisciplinary and complex one, driven by key crucial milestones and accompanying the same and typical process of a whole drug discovery and development continuum, from marker discovery and validation, assay development, clinical qualification until test approval and commercialization.

Abstract 3603: Identification of potential pharmacodynamic markers for HER3 targeted cancer treatment by a multi-technology approach

Proceedings: AACR 103rd Annual Meeting 2012‐‐ Mar 31‐Apr 4, 2012; Chicago, IL The implementation of pharmacodynamic (PD) markers in clinical drug development is becoming a key element for guiding or defining the optimal biological dose during Phase 1 dose escalation. In the past the main focus during Phase 1 was to define the maximum tolerated dose of a novel drug and the nature of its dose-limiting toxicity (DLT). Nowadays, in the context of personalized cancer treatment and in particular with well-tolerated antibody treatments, which often do not show any DLTs, PD markers are used to monitor drug target inhibition and the modulation of associated pathways. The correlation of PD effects with pharmacokinetics (PK) data can therefore guide the dose selection for further clinical development. Here we describe a hypothesis free approach using three different technology platforms to identify potential PD markers for HER3 targeted treatment in pre-clinical models. The technology platforms were selected to enable the evaluation of RNA, miRNA and protein expression in the respective tumor models. By looking at RNA and protein modulation in parallel the likelihood of identifying highly regulated signaling networks might increase significantly. Five different mouse xenograft models were treated once with 10 mg/kg of a glyco-engineered anti-HER3 antibody and tumors were explanted after 1 h, 24 h and 168 h post-treatment. In addition, tumors from vehicle-treated control animals were taken after 1 h and 168 h. Three out of five animal models are known to respond to anti-HER3 therapy whereas the other two do not respond. Tumor samples were analyzed for RNA and miRNA expression using Affymetrix arrays and protein expression by reverse phase array (MD Anderson Cancer Center). Data sets have been analyzed using biostatistics and bioinformatics. The focus was to look for global PD effects across xenograft models which are significantly modulated over time and are not seen in any vehicle control. In addition, we investigated PD markers which are dependent on the response status. Overall we saw that, independent of the technology used, the samples of each individual animal model cluster together. Based on our filter criteria we identified 19 candidate markers on the reverse phase array, 24 candidate genes within gene expression profiling and 19 using miRNA profiling, which are modulated after HER3 targeted treatment. Bioinformatics analysis revealed that some selected molecules are located downstream of HER3 thus suggesting a potential relevance within the HER3 signaling network and specifically RAS activity was altered due to anti-HER3 targeted treatment. Our goal was to compare PD signatures derived from RNA, miRNA and protein array analysis to comprehensively map molecular changes after targeted treatment. Nevertheless, it needs to be demonstrated that the molecular changes, which have been identified in our pre-clinical models, can be translated to clinics. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3603. doi:1538-7445.AM2012-3603