98 - Freie Mitteilung
17. Mai 2019, 14:15 - 15:45, Kursaal Arena, 5. OG
Circulating tumor DNA exposure in peripheral blood using a novel process: Early results of a liquid biopsy feasibility study
J. Douissard1, F. Ris1, M. Hellan2, J. Ouellette2, T. Koessler1, N. C. Buchs1, L. Buehler1, F. Triponez1, C. Toso1, Presenter: J. Douissard1 (1Geneva, 2Dayton/USA)
Standard screening recommendations are available for only a few cancers and most malignancies are discovered through incidental findings or after the presentation of symptoms. Early detection represents a potent solution towards higher rates of surgical resections with curative intent and should increase overall treatment success. Previous research has shown that analyses of circulating tumor DNA in the blood (liquid biopsy) could lower the threshold for cancer discovery before and after initial treatment.
10 ml of peripheral blood was drawn from newly diagnosed patients with pancreatic or colorectal cancer, and a control cohort without known malignancy. DNA was extracted through a custom DNA extraction protocol using Qiagen circulating free nucleid acid extraction kits with 4 ml of plasma. Optimized next generation sequencing (ONGS) was applied to detect elevations of pre-selected mutations that have been described to be associated with cancer. Raw data was analyzed using 2 different machine learning methods: Method 1 used 16 initial samples to train algorithms that were then applied in a blinded fashion to 24 additional samples. Method 2 searched for cancer vs. healthy patterns after unmasking the clinical data in all 40 samples.
The pancreatic cancer cohort included 1 patient with early intra-ductal carcinoma, 1 patient with stage II, 1 patient with stage IV and 2 patients with stage III disease. The colorectal cancer group consisted of 1 patient with stage I, 3 patients with stage II and 2 patients with stage III disease. 29 participants did not have a cancer diagnosis. The predictive machine learning algorithm performed at an overall accuracy of 0.9583, with a specificity of 1 and a sensitivity of 0,74. After unmasking the clinical diagnoses, machine learning showed 100% accuracy in distinguishing cancer patterns from controls.
Despite a very limited learning set, this new approach combining custom DNA extraction, ONGS and machine learning appears to deliver a clinically relevant accuracy for detecting colorectal and pancreatic cancer. As such, it might be a useful tool for early cancer detection for diagnosis, assessing treatment response including a liquid post-operative pathology, and monitoring disease recurrence through blood samples. However, more clinical data is needed to further determine the precise role of this approach.