Challenges in Incorporating Integral NGS into Early Clinical

Challenges in Incorporating Integral NGS into Early Clinical Trials Dr Shivaani Kummar Head, Early Clinical Trials Development Office of the Director Division of Cancer Treatment and Diagnosis National Cancer Institute May 3, 2012

Clinical Applications of Cancer Genomics § Prevention: prediction of disease risk based on inherited or early somatic changes before neoplastic transformation § Diagnostic: early disease diagnosis § Therapeutic: identify cancer subtypes likely to respond; treatment selection-sensitivity or resistance to an agent § Prognostic: Identify subsets with good or poor prognosis

Molecular Profiling to Assign “Treatments” § Growing interest in early drug development § Hypothesis: Assigning treatment based on current MP technologies would provide superior clinical benefit to this patient population as opposed to treatment based on an experienced clinician’s best judgment § Evaluate this approach in patients with refractory tumors, essentially a population eligible for phase 1 studies

Personalized medicine in a phase I clinical trials Program: M. D. Anderson Cancer Center Initiative § Tumor molecular analysis for samples from 852 (89%) of 955 consecutive patients with advanced cancer. § 354 (41. 5%) had ≥ 1 aberration: 10% of patients had a PIK 3 CA mutation; 19% KRAS; 8% NRAS; 19% BRAF; 3% EGFR; and 2% had a CKIT mutation; 21% had PTEN loss. § Median time to treatment failure (TTF) in 161 pts with 1 aberration treated with matched targeted therapy was 5. 3 months (95%CI: 4. 1, 6. 6) vs 3. 2 months (95%CI: 2. 9 - 4. 0) for prior systemic therapy § No significant difference in TTF or CR+PR in patients with 2 or 3 aberrations between matched targeted therapy vs without matching. Tsimberidou AM, et al. J Clin Oncol 29: 2011 (suppl; abstr CRA 2500)

Molecular Profiling for Potential Targets and Treatment Selection § Nine centers across US, obtained fresh biopsies, IHC/FISH and oligonucleotide microarray § Compared PFS for regimen chosen based on MP results versus PFS for most recent prior regimen ( ‘patients as their own controls’) § PFS on MP-selected therapy/PFS on prior therapy of ≥ 1. 3 was considered positive § Molecular target detected in 84 of 86 pts(98%) § Commercially available agents used for treatment § 18 of 66 (27%) had PFS ratio ≥ 1. 3 Von Hoff DD, et al. J Clin Oncol 2010; 28(33): 4877; Doroshow JH. J Clin Oncol 2010; 28(33): 4869

What have we learnt from our experience so far? § It is feasible to obtain fresh tissue, perform MP § Turn around time 3 -4 weeks (clinically relevant time frame) § To assign therapy need validated assays (CLIA) § Targeted sequencing for actionable mutations versus whole genome sequencing § A subset of patients with refractory tumors may benefit from treatment assignment using MP § Treatment not pre-defined, ad hoc assignment § Assignment bias- availability of appropriate agent, which pts get assigned, other factors

What have we learnt from our experience so far? § TTP on prior systemic therapy § pre-study TTP not well defined using uniform imaging techniques and intervals; § participation in prior trial or design trials with specific TTP lead in period; § randomized design § Includes commercially available and investigational agents (NCI data on phase 1 trials reported a 17% RR) § May include agents such as vemurafenib, doesn’t truly test the hypothesis for assigning therapy with inhibitors of pathways § Importance of target may be disease context dependent (vemurafenib in melanoma vs colorectal cancer) Sosman JA, et al. N Engl J Med 2012; 366(8): 707; Prahallad A, et al Nature 2012; 483(7387): 100

Practical Application of Molecular Profiling in Early Phase Trials § Turn around time of 3 -4 weeks § Interpretation of the data-how do you assign clinical significance? § Finding a target in a patient’s tumor and having access to an agent that inhibits the given target does not imply clinical benefit § Report needs to be in a format to make it easy to interpret with guidance on which class of agents to administer § Multidisciplinary molecular tumor board

Molecular Tumor Board § Multidisciplinary- expertise in clinical oncology (early drug development, disease specific), genomics, signal transduction pathways, genetic counseling, bioethics, pathology § Ad hoc or regularly scheduled meetings/committees § Tumor heterogeneity, biopsy techniques, low tumor purity (microdissection, cell based enrichment, ploidy based sorting) § Which aberrations are clinically significant? § What percent of cells demonstrating mutation are significant?

Molecular Tumor Board Need to address: § Driver versus passenger mutations (single tumor carries an average of 80 somatic mutations that change the amino acid sequences of proteins) § If more than one aberration is present, which one to target first, which sequence? § How to we “validate” the performance of the tumor board? § Will decisions be made the same way on different days?

Informed Consent § Biopsy associated risks § How much information to share? incidental findings? § Genetic Counseling, absence of outcome data to guide counseling § Patient should decide whether to find out about incidental findings § Involve Bioethicists § Verbalize understanding of the consent § Confidentiality issues

M-PACT: Molecular Profiling based Assignment of Cancer Therapeutics Pilot Trial to Assess the Utility of Genetic Sequencing to Determine Therapy and Improve Patient Outcome in Early Phase Trials NCI-Sponsored Clinical Trial

Objective v Assess whether the response rate (CR+PR) and/or 4 - month PFS is improved following treatment with agents chosen based on the presence of specific mutations/amplifications in patient tumors. ¬ Only patients with pre-defined mutations/amplifications of interest will be eligible ¬ Cohort A: Receive treatment based on an agent prospectively identified to work on that mutation/pathway ¬ Cohort B: Receive treatment with one of the targeted agents in the complementary set (identified to not work on one of the detected mutations/pathways) 13

Statistical Design v Patients will be randomized 2: 1 in Cohort A vs Cohort B v Within cohort A, up to 30 patients will be treated within each of the treatment strata; discriminate between tumor response rates of 25% vs. 5% and 2 -month PFS rates of 80% vs. 50% (corresponding to median PFS of 4. 8 vs. 2 months). v The two arms will be compared with respect to both objective response rate and PFS v Within each stratum of Cohort A, objective response rate and PFS will be assessed against the historical standards of 5% response rate and 50% 2 -month PFS 14

Acknowledgements National Cancer Institute Developmental Therapeutics Team All the patients who participate in early-phase clinical trials