Increased NKX6.1 expression and decreased ARX expression in alpha cells accompany reduced beta-cell volume in human subjects

Insulinoma with Hyperprocalcitoninemia and Hypercalcitoninemia Showing Coexpression of Insulin and Calcitonin in Its Tumor Cells

Neuroendocrine neoplasms can produce multiple hormones that are released into the bloodstream, causing symptoms that vary depending on the type and quantity of hormones involved. We herein report a 63-year-old asymptomatic patient with pancreatic insulinoma who showed marked elevations in circulating calcitonin and procalcitonin levels that returned to normal following surgery. Immunohistochemical analyses confirmed the co-staining of calcitonin and insulin immunoreactivity in the tumor cells, suggesting a calcitonin-producing insulinoma. This insulinoma released calcitonin and a considerable amount of its precursor peptide, procalcitonin, resulting in both hyperprocalcitoninemia and hypercalcitoninemia.

Adversarial training improves model interpretability in single-cell RNA-seq analysis

Motivation

Predictive computational models must be accurate, robust, and interpretable to be considered reliable in important areas such as biology and medicine. A sufficiently robust model should not have its output affected significantly by a slight change in the input. Also, these models should be able to explain how a decision is made to support user trust in the results. Efforts have been made to improve the robustness and interpretability of predictive computational models independently; however, the interaction of robustness and interpretability is poorly understood.

Results

As an example task, we explore the computational prediction of cell type based on single-cell RNA-seq data and show that it can be made more robust by adversarially training a deep learning model. Surprisingly, we find this also leads to improved model interpretability, as measured by identifying genes important for classification using a range of standard interpretability methods. Our results suggest that adversarial training may be generally useful to improve deep learning robustness and interpretability and that it should be evaluated on a range of tasks.

Availability and implementation

Our Python implementation of all analysis in this publication can be found at: https://github.com/MehrshadSD/robustness-interpretability. The analysis was conducted using numPy 0.2.5, pandas 2.0.3, scanpy 1.9.3, tensorflow 2.10.0, matplotlib 3.7.1, seaborn 0.12.2, sklearn 1.1.1, shap 0.42.0, lime 0.2.0.1, matplotlib_venn 0.11.9.

Tracking Ca2+ Dynamics in NOD Mouse Islets During Spontaneous Diabetes Development

The mechanisms accounting for the functional changes of α- and β-cells over the course of type 1 diabetes (T1D) development are largely unknown. Permitted by our established technology of high spatiotemporal resolution imaging of cytosolic Ca2+ ([Ca2+]c) dynamics on fresh pancreas tissue slices, we tracked the [Ca2+]c dynamic changes, as the assessment of function, in islet α- and β-cells of female nonobese diabetic (NOD) mice during the development of spontaneous diabetes. We showed that, during the phases of islet inflammation, 8 mmol/L glucose-induced synchronized short [Ca2+]c events in β-cells were diminished, whereas long [Ca2+]c events were gradually more triggerable at substimulatory 4 and 6 mmol/L glucose. In the islet destruction phase, the synchronized short [Ca2+]c events in a subset of β-cells resumed at high glucose condition, while the long [Ca2+]c events were significantly elevated already at substimulatory glucose concentrations. In the α-cells, the glucose sensitivity of the [Ca2+]c events persisted throughout the course of T1D development. At the late islet destruction phase, the α-cell [Ca2+]c events exhibited patterns of synchronicity. Our work has uncovered windows of functional recovery in β-cells and potential α-cells functional synchronization in NOD mice over the course of T1D development.

ARTICLE HIGHLIGHTS

In NOD mice β-cells, 8 mmol/L glucose-induced synchronized short [Ca2+]c events diminish in the early phases of islet inflammation, and long Ca2+ events became more sensitive to substimulatory 4 and 6 mmol/L glucose. In the late islet destruction phase, the synchronized short [Ca2+]c events in a subset of β-cells resumed at 8 mmol/L glucose, while the long Ca2+ events were significantly elevated at substimulatory glucose concentrations. In the α-cells, the glucose sensitivity of the [Ca2+]c events persisted throughout the course of type 1 diabetes development. α-Cell [Ca2+]c events occasionally synchronize in the islets with severe β-cell destruction.