Search

Background: Mucinous colorectal carcinoma (CRC) is found in 10-20% of patients and is associated with worse prognosis and treatment resistance. The early identification of mucinous tumor component at baseline and monitoring resistant clones at follow-up is challenging in clinical practice, which hinders appropriate and timely treatment selection. At CT, being routinely acquired in clinical practice, mucinous tumors can be characterized by semantic features, such as hypoattenuation and more heterogeneous enhancement than the non-mucinous tumors (Wnorowski et al 2019). However, the diagnostic accuracy of such CT findings reaches at most 62% (Young et al 2007). This can be substantially improved by utilizing robust feature quantification using state-of-art machine learning and neural network techniques. Materials and Methods: 7 mucinous and 7 non-mucinous CRC CTs were included in the model development (80% training and 20% validation) and 2 mucinous and 2 non-mucinous independent patients were used to test the model performance. Multiple lesions (primary and metastatic) were semi-automatically segmented in 3D Slicer (N=32 development and N=12 test). Three different classification models were generated using CT images: (1) a logistic regression model based on a newly developed hypodense tissue connectivity (HTC) metric, (2) a logistic regression model using a set of automatically selected radiomics (RAD) features (shape, 1st order and 2nd order) and (3) a convolutional neural network model (CNN) based on the ResNet architecture and automatically selected features. HTC was computed as a ratio between the volume of the connected hypodense tissue (0 Citation Format: Kinga Bernatowicz, Raquel Perez Lopez, Hector Garcia Palmer, Elena Elez Fernandez, Jose Fernandez Navarro, Marta Ligero Hernandez, Alonso Garcia Ruiz, Manuel Escobar Amores. Humans cannot accurately detect mucinous colorectal carcinoma from CT images, can AI help? [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-021.

Purpose: We evaluated whether next-generation sequencing (NGS) of circulating cell-free DNA (cfDNA) could be used for patient selection and as a tumor clone response biomarker in patients with advanced cancers participating in early-phase clinical trials of targeted drugs. Experimental Design: Plasma samples from patients with known tumor mutations who completed at least two courses of investigational targeted therapy were collected monthly, until disease progression. NGS was performed sequentially on the Ion Torrent PGM platform. Results: cfDNA was extracted from 39 patients with various tumor types. Treatments administered targeted mainly the PI3K–AKT–mTOR pathway (n = 28) or MEK (n = 7). Overall, 159 plasma samples were sequenced with a mean sequencing coverage achieved of 1,685X across experiments. At trial initiation (C1D1), 23 of 39 (59%) patients had at least one mutation identified in cfDNA (mean 2, range 1–5). Out of the 44 mutations identified at C1D1, TP53, PIK3CA and KRAS were the top 3 mutated genes identified, with 18 (41%), 9 (20%), 8 (18%) different mutations, respectively. Out of these 23 patients, 13 received a targeted drug matching their tumor profile. For the 23 patients with cfDNA mutation at C1D1, the monitoring of mutation allele frequency (AF) in consecutive plasma samples during treatment with targeted drugs demonstrated potential treatment associated clonal responses. Longitudinal monitoring of cfDNA samples with multiple mutations indicated the presence of separate clones behaving discordantly. Molecular changes at cfDNA mutation level were associated with time to disease progression by RECIST criteria. Conclusions: Targeted NGS of cfDNA has potential clinical utility to monitor the delivery of targeted therapies. Clin Cancer Res; 21(20); 4586–96. ©2015 AACR.