Search

Background: Colorectal cancer (CRC) is a leading cause of cancer-related mortality worldwide, and its incidence is increasing in younger persons. Established CRC screening methods include fecal immunochemical testing, which requires handling of stool, and colonoscopy, which is invasive. Many persons who remain unscreened might accept a blood-based screening test, which could have profound public health impact. DNA methylation is a stable, early, and tissue-specific event in cancer development and progression. Measuring the methylation status of tumor-derived cell-free DNA in plasma could enable identification of early-stage CRC. We first developed a plasma-based methylated DNA panel for all stages of CRC (Study 1). We then validated this panel in a subsequent study with independent samples (Study 2). Methods: Differentially methylated regions (DMRs) were initially selected by analyzing tissue data from The Cancer Genome Atlas (TCGA) database. In Study 1, candidate regions were evaluated in plasma samples of 215 patients (93 CRC, 122 controls including 31 with non-advanced adenoma) using methylation-sensitive restriction enzyme qPCR (MSRE-qPCR). This training set was used for marker evaluation and construction of a prediction model. The most promising methylation markers were selected by a support vector machine (SVM)-based machine learning classifier using a random forest algorithm. In Study 2, the selected markers were validated in an independent validation set of 724 samples collected in Spain, Ukraine, the UK, and the US (152 CRC, 622 controls including 148 with non-advanced adenoma and 52 with non-CRC cancer). Results: Tissue data analysis yielded 180 potential DMRs. In Study 1, a panel consisting of the top 12 methylation markers was selected. In Study 2, the prediction model based on this panel correctly classified 117/152 CRC (77%) patients. Sensitivity improved with CRC stage, ranging from 71% (25/35) for stage I, 76% (37/49) for stage II, 77% (37/48) for stage III, to 100% (14/14) for stage IV. Specificity of the model was 88% (544/622). Among non-CRC cancer cases, specificity was 83% overall (73% [19/26] for lung cancer and 92% [24/26] for breast cancer). Conclusions: We developed a prediction model based on a novel plasma 12-marker methylation panel that demonstrated highly promising test performance characteristics for early-stage CRC detection in a validation study using independent samples. This method could serve as the basis for a highly accurate and minimally invasive blood-based CRC screening test. Citation Format: Kristi Kruusmaa, Marko Bitenc, Walter Pulverer, Andreas Weinhaeusel, José Luis Rodrigo Agudo, Andrés Barrientos Delgado, Andrés Sánchez Yagüe, José Manuel Rodríguez Laiz, Marko Chersicola, Primož Knap, Silvia Schönthaler, Uri Ladabaum. Development and clinical performance of an accurate cell-free DNA (cfDNA) methylation assay for early detection of colorectal cancer [abstract]. In: Proceedings of the AACR Special Conference on Advances in Liquid Biopsies; Jan 13-16, 2020; Miami, FL. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(11_Suppl):Abstract nr A05.