A deep learning model to triage and predict adenocarcinoma on pancreas cytology whole slide imaging.

Pancreatic fine-needle aspirations are the gold-standard diagnostic procedure for the evaluation of pancreatic ductal adenocarcinoma. A suspicion for malignancy can escalate towards chemotherapy followed by a major surgery and therefore is a high-stakes task for the pathologist. In this paper, we propose a deep learning framework, MIPCL, that can serve as a helpful screening tool, predicting the presence or absence of cancer. We also reproduce two deep learning models that have found success in surgical pathology for our cytopathology study. Our MIPCL significantly improves over both models across all evaluated metrics (F1-Score: 87.97% vs 88.70% vs 91.07%; AUROC: 0.9159 vs. 0.9051 vs 0.9435). Additionally, our model is able to recover the most contributing regions on the slide for the final prediction. We also present a dataset curation strategy that increases the number of training examples from an existing dataset, thereby reducing the resource burden tied to collecting and scanning additional cases.

Promoter methylation of ADAMTS1 and BNC1 as potential biomarkers for early detection of pancreatic cancer in blood.

BACKGROUND: Despite improvements in cancer management, most pancreatic cancers are still diagnosed at an advanced stage. We have recently identified promoter DNA methylation of the genes ADAMTS1 and BNC1 as potential blood biomarkers of pancreas cancer. In this study, we validate this biomarker panel in peripheral cell-free tumor DNA of patients with pancreatic cancer. RESULTS: Sensitivity and specificity for each gene are as follows: ADAMTS1 87.2% and 95.8% (AUC = 0.91; 95% CI 0.71-0.86) and BNC1 64.1% and 93.7% (AUC = 0.79; 95% CI 0.63-0.78). When using methylation of either gene as a combination panel, sensitivity increases to 97.3% and specificity to 91.6% (AUC = 0.95; 95% CI 0.77-0.90). Adding pre-operative CA 19-9 values to the combined two-gene methylation panel did not improve sensitivity. Methylation of ADAMTS1 was found to be positive in 87.5% (7/8) of stage I, 77.8% (7/9) of stage IIA, and 90% (18/20) of stage IIB disease. Similarly, BNC1 was positive in 62.5% (5/8) of stage I patients, 55.6% (5/9) of stage IIA, and 65% (13/20) of patients with stage IIB disease. The two-gene panel (ADAMTS1 and/or BNC1) was positive in 100% (8/8) of stage I, 88.9% (8/9) of stage IIA, and 100% (20/20) of stage IIB disease. The sensitivity and specificity of the two-gene panel for localized pancreatic cancer (stages I and II), where the cancer is eligible for surgical resection with curative potential, was 94.8% and 91.6% respectively. Additionally, the two-gene panel exhibited an AUC of 0.95 (95% CI 0.90-0.98) compared to 57.1% for CA 19-9 alone. CONCLUSION: The methylation status of ADAMTS1 and BNC1 in cfDNA shows promise for detecting pancreatic cancer during the early stages when curative resection of the tumor is still possible. This minimally invasive blood-based biomarker panel could be used as a promising tool for diagnosis and screening in a select subset of high-risk populations.