Profiling of circulating tumor DNA for treatment selection in patients with advanced and refractory carcinoma: A prospective, two-stage phase II individualized cancer treatment trial - 11th International Symposium on Circulating Nucleic Acids in Plasma and Serum (CNAPS)

Samantha Perakis ¹ Peter Ulz ¹ Jakob M. Riedl ³ Lukas Scheipner ³ Karl Kashofer ² Gerald Höfler ² Jochen B. Geigl ¹ Michael R. Speicher ¹ Ellen Heitzer ¹ Armin Gerger ³

¹Institute of Human Genetics, Diagnostic and Research Center for Molecular BioMedicine, Medical University of Graz, Graz, Austria
²Diagnostic and Research Institute of Pathology, Medical University of Graz, Graz, Austria
³Department of Internal Medicine, Division of Oncology, Medical University of Graz, Graz, Austria

BACKGROUND: Genome-driven cancer treatment has emerged as a promising strategy for realizing individualized cancer treatment. To this end, non-invasive analysis of circulating tumor DNA (ctDNA) from peripheral blood could eliminate ineffective therapy attempts. In this prospective, non-randomized, open two-stage clinical phase II study (EudraCT: 2014-005341-44), the success of a targeted therapy selected by molecular profiling was evaluated. Patients with metastasized carcinoma were profiled via whole-genome sequencing in parallel with a cancer hotspot panel of clinically relevant genes.

METHODS: Somatic copy number alterations (SCNAs) and mutations were matched to targeted treatments using publicly available databases such as “My Cancer Genome” and annotated results were discussed at a molecular tumor board. Additionally, select molecular profiles were analyzed retrospectively via CureMatch PreciGENE™ to test the algorithm for its potential in stratifying combinatorial therapy options.

RESULTS: After interim analysis of 24 patients, this pan-cancer cohort included: colorectal cancer (n=7), pancreas (n=4), cholangiocarcinoma (n=2), cardia (n=2), CUP (n=2), renal (n=1), gallbladder (n=1), breast (n=1), laryngeal (n=1), gastric (n=1), esophageal (n=1) and bladder (n=1). Informative results could be achieved in 18 patients (75%) and median tumor fraction was 22.7% (range 5.2-40.3). Of these, 11 patients had a molecular target associated with a current existing drug. Furthermore, PreciGENE™ analysis was able to identify either a 2-drug or 3-drug combination in each of the tested cases, including a case for which no therapy was previously identified, challenging the premise of the `one biomarker one drug` approach and encouraging us to now undertake additional retrospective analyses.

CONCLUSIONS: Analysis of patient outcomes is currently ongoing. However, early evidence demonstrated that few patients benefited from this histology-agnostic approach. Retrospective analysis of select cases via the CureMatch PreciGENE™ platform was able to suggest combination therapies overlooked by standard database matching of actionable targets, suggesting the power of a machine learning approach to potentially improving patient outcomes.