Deep Learning Improves Pancreatic Cancer Diagnosis Using RNA-Based Variants.

For optimal pancreatic cancer treatment, early and accurate diagnosis is vital. Blood-derived biomarkers and genetic predispositions can contribute to early diagnosis, but they often have limited accuracy or applicability. Here, we seek to exploit the synergy between them by combining the biomarker CA19-9 with RNA-based variants. We use deep sequencing and deep learning to improve differentiating pancreatic cancer and chronic pancreatitis. We obtained samples of nucleated cells found in peripheral blood from 268 patients suffering from resectable, non-resectable pancreatic cancer, and chronic pancreatitis. We sequenced RNA with high coverage and obtained millions of variants. The high-quality variants served as input together with CA19-9 values to deep learning models. Our model achieved an area under the curve (AUC) of 96% in differentiating resectable cancer from pancreatitis using a test cohort. Moreover, we identified variants to estimate survival in resectable cancer. We show that the blood transcriptome harbours variants, which can substantially improve noninvasive clinical diagnosis.

Sensitive Quantification of Cell-Free Tumor DNA for Early Detection of Recurrence in Colorectal Cancer.

The detection of plasma cell-free tumor DNA (ctDNA) is prognostic in colorectal cancer (CRC) and has potential for early prediction of disease recurrence. In clinical routine, ctDNA-based diagnostics are limited by the low concentration of ctDNA and error rates of standard next-generation sequencing (NGS) approaches. We evaluated the potential to increase the stability and yield of plasma cell-free DNA (cfDNA) for routine diagnostic purposes using different blood collection tubes and various manual or automated cfDNA extraction protocols. Sensitivity for low-level ctDNA was measured in KRAS-mutant cfDNA using an error-reduced NGS procedure. To test the applicability of rapid evaluation of ctDNA persistence in clinical routine, we prospectively analyzed postoperative samples of 67 CRC (stage II) patients. ctDNA detection was linear between 0.0045 and 45%, with high sensitivity (94%) and specificity (100%) for mutations at 0.1% VAF. The stability and yield of cfDNA were superior when using Streck BCT tubes and a protocol by Zymo Research. Sensitivity for ctDNA increased 1.5-fold by the integration of variant reads from triplicate PCRs and with PCR template concentration. In clinical samples, ctDNA persistence was found in ∼9% of samples, drawn 2 weeks after surgery. Moreover, in a retrospective analysis of 14 CRC patients with relapse during adjuvant therapy, we successfully detected ctDNA (median 0.38% VAF; range 0.18-5.04% VAF) in 92.85% of patients significantly prior (median 112 days) to imaging-based surveillance. Using optimized pre-analytical conditions, the detection of postoperative ctDNA is feasible with excellent sensitivity and allows the prediction of CRC recurrence in routine oncology testing.