The Signature Program, a Distinctive Tissue Agnostic Trial Model for Molecularly Pre-Selected Hematological and Solid Tumor Patients

Introduction: Here we describe a novel signal-finding clinical trial protocol series, termed the Novartis “Signature” program. These are tissue-agnostic, genetic alteration (mutation, amplification, translocation, etc.) specific protocols that do not include pre-identified clinical trial sites. In part these are responsive to the increased frequency of molecular profiling in the oncology community, and the incidence of patients whose tumors are identified with actionable genetic alteration yet without automatic access to drugs targeting those alterations. As these patients are identified via standard of care physician-directed profiling, we bring the ‘Protocol to the Patient’, utilizing a rapid study start-up process, such that a de novo site can have one of these trials opened within weeks of the originating patient being identified. Currently the 8 open single agent protocols involve buparlisib (BKM120, PI3Ki), dovitinib (TKI258, multi-kinase inhibitor), binimetinib (MEK162, MEKi), encorafenib (LGX818, RAFi), LEE011 (CDK4/6i), BGJ398 (FGFRi), ceritinib (LDK378, ALKi) and sonidegib (LDE225; SMOi).

Methods:Patients with advanced solid and hematologic cancers and no established standard therapy options are eligible. Patients are locally pre-identified with an actionable genetic alteration relevant to the particular compound being studied; a local test done in a CLIA laboratory is sufficient for eligibility. Central broad molecular profiling of patient’s tumors (including confirmatory analysis of inclusion genetic alteration) is performed post-hoc using a fresh or archived sample. The primary objective of these trials is to assess clinical benefit (SD or better for ≥ 16 weeks for solid tumors or appropriate hematologic criteria) associated with the study related compound. A novel statistical design is incorporated to adaptively cluster patients of like indications into cohorts for independent analysis for futility (minimum 10 patients) or efficacy (minimum 15 patients). Rare tumors of at least 4 patients can be clustered together into unique cohorts for determination of signals. To investigate mechanisms of resistance, an optional biopsy is taken upon progression for patients who have had a prior response to the study compound – these tissues are subjected to a broad molecular profiling analysis.

Results: Between March 2013 - July 2014, 13 academic centers and another 74 unique community/network sites had screened 286 patients. On average the startup timelines were 5.2 weeks across the program. 147 patients with a genetic alteration received matched therapy and have been dosed: primarily colorectal (28; 19%), NSCLC (25; 17%), ovarian (13; 9%) and sarcoma (n=12; 8%). Rare tumors accrued include small intestine/duodenal (n=6), anal (n=3), thymus (n=2), cholangiocarcinoma (n=2) and appendix (n=2). The Signature program also accrued heme malignancies.

Across the whole program, the identifying actionable alterations were mainly PIK3CA (35%), PTEN loss (23%), RAS (18%), NF1 (4.7%), and FGFR2 (3.5%), and also included BRAF, SMO, PTCH1, VEGFR2, RET, PDGFRa, PDGFRb, TRKa, FGFR3, MEK1 and FLT3. The median number of lines of prior treatment was 3 (mean = 3.4; range of 1 to 13), median patient age of 61 years (range of 22 to 82), 40.3 % male, ECOG PS of 0 in 38% of patients and ECOG PS of 1 in 61.5% of patients.

Three patients with AML harboring RAS mutations and one patient with multiple myeloma with a BRAFV600E mutation were included in the binimetinib module. (3 female / 1 male; 1-6 prior lines of therapy; 31-80 years old. One AML patient received prior transplant). Additional data will be presented. Futility/Efficacy analyses for 5 cohorts (4 from buparlisib and 1 from binimetinib) are currently ongoing.

Conclusions: This program effectively allows the enrollment of molecularly profiled patients with genetic alterations linked to cognate targeted agents. The cohort analysis will evaluate clinical activity in multiple tumor types in parallel. The Signature program could enable the identification of clinical signals using patient sparing designs, potentially leading to subsequent confirmatory trials. This trial model has successfully allowed enrollment of molecularly pre-selected rare tumors in a short timeline without pre-determined trial site selection. The next step of this model will include novel drug combination modules.