Deep learning-based pancreas volume assessment in individuals with type 1 diabetes

Pancreas volume is reduced in individuals with diabetes and in autoantibody positive individuals at high risk for developing type 1 diabetes (T1D). Studies investigating pancreas volume are underway to assess pancreas volume in large clinical databases and studies, but manual pancreas annotation is time-consuming and subjective, preventing extension to large studies and databases. This study develops deep learning for automated pancreas volume measurement in individuals with diabetes. A convolutional neural network was trained using manual pancreas annotation on 160 abdominal magnetic resonance imaging (MRI) scans from individuals with T1D, controls, or a combination thereof. Models trained using each cohort were then tested on scans of 25 individuals with T1D. Deep learning and manual segmentations of the pancreas displayed high overlap (Dice coefficient = 0.81) and excellent correlation of pancreas volume measurements (R2 = 0.94). Correlation was highest when training data included individuals both with and without T1D. The pancreas of individuals with T1D can be automatically segmented to measure pancreas volume. This algorithm can be applied to large imaging datasets to quantify the spectrum of human pancreas volume.

Bicarbonate Transport in Cystic Fibrosis and Pancreatitis

CFTR, the cystic fibrosis (CF) gene-encoded epithelial anion channel, has a prominent role in driving chloride, bicarbonate and fluid secretion in the ductal cells of the exocrine pancreas. Whereas severe mutations in CFTR cause fibrosis of the pancreas in utero, CFTR mutants with residual function, or CFTR variants with a normal chloride but defective bicarbonate permeability (CFTRBD), are associated with an enhanced risk of pancreatitis. Recent studies indicate that CFTR function is not only compromised in genetic but also in selected patients with an acquired form of pancreatitis induced by alcohol, bile salts or smoking. In this review, we summarize recent insights into the mechanism and regulation of CFTR-mediated and modulated bicarbonate secretion in the pancreatic duct, including the role of the osmotic stress/chloride sensor WNK1 and the scaffolding protein IRBIT, and current knowledge about the role of CFTR in genetic and acquired forms of pancreatitis. Furthermore, we discuss the perspectives for CFTR modulator therapy in the treatment of exocrine pancreatic insufficiency and pancreatitis and introduce pancreatic organoids as a promising model system to study CFTR function in the human pancreas, its role in the pathology of pancreatitis and its sensitivity to CFTR modulators on a personalized basis.

A combined transcriptional and dynamic roadmap of single human pancreatic endocrine progenitors reveals proliferative capacity and differentiation continuum

Basic helix-loop-helix genes, particularly proneural genes, are well-described triggers of cell differentiation, yet limited information exists on their dynamics, notably in human development. Here, we focus on Neurogenin 3 (NEUROG3), which is crucial for pancreatic endocrine lineage initiation. Using a double reporter to monitor endogenous NEUROG3 transcription and protein expression in single cells in 2D and 3D models of human pancreas development, we show peaks of expression for the RNA and protein at 22 and 11 hours respectively, approximately two-fold slower than in mice, and remarkable heterogeneity in peak expression levels all triggering differentiation. We also reveal that some human endocrine progenitors proliferate once, mainly at the onset of differentiation, rather than forming a subpopulation with sustained proliferation. Using reporter index-sorted single-cell RNA-seq data, we statistically map transcriptome to dynamic behaviors of cells in live imaging and uncover transcriptional states associated with variations in motility as NEUROG3 levels change, a method applicable to other contexts.

FEM: mining biological meaning from cell level in single-cell RNA sequencing data

Background One goal of expression data analysis is to discover the biological significance or function of genes that are differentially expressed. Gene Set Enrichment (GSE) analysis is one of the main tools for function mining that has been widely used. However, every gene expressed in a cell is valuable information for GSE for single-cell RNA sequencing (scRNA-SEQ) data and not should be discarded. Methods We developed the functional expression matrix (FEM) algorithm to utilize the information from all expressed genes. The algorithm converts the gene expression matrix (GEM) into a FEM. The FEM algorithm can provide insight on the biological significance of a single cell. It can also integrate with GEM for downstream analysis. Results We found that FEM performed well with cell clustering and cell-type specific function annotation in three datasets (peripheral blood mononuclear cells, human liver, and human pancreas).

Substance Use Affects Type 1 Diabetes Pancreas Pathology: Implications for Future Studies

Access to human pancreas samples from organ donors has greatly advanced our understanding of type 1 diabetes pathogenesis; however, previous studies have shown that donors have a high rate of substance use, and its impact on pancreatic histopathology in this disease is not well described. One-hundred-thirty-one type 1 diabetes and 111 control organ donor pancreata from persons 12-89 years of age (mean 29.8 ± 15.5 years) within the Network for Pancreatic Organ donors with Diabetes (nPOD) were examined for insulin positivity, insulitis, amyloid staining, acute and chronic pancreatitis, and chronic exocrine changes (acinar atrophy, fibrosis, fatty infiltration, or periductal fibrosis); findings were compared by history of substance use. A secondary analysis compared exocrine pancreatic histopathologic findings in type 1 diabetes versus control organ donors regardless of substance use history. We observed a high but congruent rate of substance use in type 1 diabetes and control organ donors (66.4% and 64% respectively). Among donors with type 1 diabetes (but not controls), islet amyloid (OR 9.96 [1.22, 81.29]) and acute pancreatitis (OR 3.2 [1.06, 9.63]) were more common in alcohol users while chronic exocrine changes (OR 8.86 [1.13, 69.31]) were more common in cocaine users. Substance use impacted the pancreata of donors with type 1 diabetes more than controls. Overall, despite similar rates of substance use, acute pancreatitis (15.3% versus 4.5%, p=0.0061), chronic pancreatitis (29.8% versus 9.9%, p=0.0001), and chronic exocrine changes (73.3% versus 36.9%, p<0.0001) were more common in type 1 diabetes donors than controls. Alcohol and/or cocaine use in type 1 diabetes organ donors increases exocrine pancreas pathology and islet amyloid deposition but does not affect insulitis or insulin positivity. Exocrine pathology in type 1 diabetes donors is common, and further study of the pathophysiology of these changes is needed.

Cholecystokinin-B Receptor-Targeted Nanoparticle for Imaging and Detection of Precancerous Lesions in the Pancreas

Survival from pancreatic cancer remains extremely poor, in part because this malignancy is not diagnosed in the early stages, and precancerous pancreatic intraepithelial neoplasia (PanIN) lesions are not seen on routine radiographic imaging. Since the cholecystokinin-B receptor (CCK-BR) becomes over-expressed in PanIN lesions, it may serve as a target for early detection. We developed a biodegradable fluorescent polyplex nanoparticle (NP) that selectively targets the CCK-BR. The NP was complexed to a fluorescent oligonucleotide with Alexa Fluor 647 for far-red imaging and to an oligonucleotide conjugated to Alexa Fluor 488 for localization by immunohistochemistry. Fluorescence was detected over the pancreas of five- to ten-month-old LSL-KrasG12D/+; P48-Cre (KC) mice only after the injection of the receptor target-specific NP and not after injection of untargeted NP. Ex vivo tissue imaging and selective immunohistochemistry confirmed particle localization only to PanIN lesions in the pancreas and not in other organs, supporting the tissue specificity. A human pancreas tissue microarray demonstrated immunoreactivity for the CCK-BR only in the PanIN lesions and not in normal pancreas tissue. The long-term goal would be to develop this imaging tool for screening human subjects at high risk for pancreatic cancer to enable early cancer detection.

Self-distillation contrastive learning enables clustering-free signature extraction and mapping to multimodal single-cell atlas of multimillion scale

Massively generated single-cell multi-omics datasets are revolutionizing biological studies of heterogenous tissues and organisms, which necessitate powerful computational methods to unleash the full potential of these tremendous data. Here, we present Concerto, stands for self-distillation contrastive learning of cell representations, a self-supervised representation learning framework optimized with asymmetric teacher-student configuration to analyze single-cell multi-omics datasets with scalability up to building 10 million-cell reference within 1.5 hour and querying 10k cells within 8 seconds. Concerto leverages dropout layer as minimal data augmentation to learn meaningful cell representations in a contrastive manner. The teacher module uses attention mechanism to aggregate contextualized gene embeddings within cellular context, while the student module uses simpler dense structure with discreate input. The learned task-agnostic representations can be adapted to a broad range of single-cell computation tasks. 1) Via supervised fine-tuning, Concerto enables automatic cell classification as well as novel cell-type discovery; 2) Attention weights provide model interpretability via automatically extracting specific molecular signatures at single-cell resolution without the needs of clustering; 3) Via source-aware training, Concerto supports efficient data integration by projecting all cells across multiple batches into a joint embedding space. 4) Via batch-aware inference or unsupervised fine-tuning, Concerto enables mapping query cells onto reference and accurately transferring annotations. Concerto can flexibly extend to multi-omics datasets simply through cross-modality summation operation to obtain unified cell embeddings. Using examples from human peripheral blood, human thymus, human pancreas, and mouse tissue atlas, Concerto shows superior performance benchmarking against other top-performing methods. We also demonstrate Concerto recapitulates detailed COVID-19 disease variation through query-to-reference mapping. Concerto can operate on all genes and represents a fully data-driven approach with minimum prior distribution assumptions, eliminating the needs of PCA-like or autoencoder-like dimensionality reduction, which significantly reforms the current best practice. Concerto is a simple, straightforward, robust, and scalable framework, offering a brand new perspective to derive cell representations and can effectively satisfy the emerging paradigm of query-to-reference mapping in the era of atlas-level single-cell multimodal analysis.

Comparative Analysis of the Influence of Extracellular Matrix Biomimetics on the Viability and Insulin-Producing Function of Isolated Pancreatic Islets

The creation of a pancreas tissue-engineered construct based on isolated pancreatic islets is hindered by problems associated with maintaining their viability and insulin-producing function. Both biopolymer and tissue-specific scaffolds can contribute to the maintenance of the structure and function of pancreatic islets in vitro and in vivo. A comparative morphofunctional analysis in vitro of isolated pancreatic islets cultured with a biopolymer collagen-containing scaffold and a tissue-specific scaffold obtained as a result of pancreatic decellularization was performed. The results showed that the use of the scaffolds contributes not only to the maintenance of the cultured islets viability, but also to the prolongation of their insulin-producing functions, compared to the islets monoculture in vitro. A significant increase was found in basal and stimulated (under glucose loading) insulin secreted by the islets cultured with the scaffolds. At the same time, the advantage of using a tissue-specific scaffold in comparison with a biopolymer collagen-containing scaffold was shown. We think that these studies will become a platform for creating a human pancreas tissue-engineered design for the treatment of type 1 diabetes.