Clinically Applicable AI System for Accurate Diagnosis, Quantitative Measurements, and Prognosis of COVID-19 Pneumonia Using Computed Tomography.

Many COVID-19 patients infected by SARS-CoV-2 virus develop pneumonia (called novel coronavirus pneumonia, NCP) and rapidly progress to respiratory failure. However, rapid diagnosis and identification of high-risk patients for early intervention are challenging. Using a large computed tomography (CT) database from 3,777 patients, we developed an AI system that can diagnose NCP and differentiate it from other common pneumonia and normal controls. The AI system can assist radiologists and physicians in performing a quick diagnosis especially when the health system is overloaded. Significantly, our AI system identified important clinical markers that correlated with the NCP lesion properties. Together with the clinical data, our AI system was able to provide accurate clinical prognosis that can aid clinicians to consider appropriate early clinical management and allocate resources appropriately. We have made this AI system available globally to assist the clinicians to combat COVID-19.

Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning.

The implementation of clinical-decision support algorithms for medical imaging faces challenges with reliability and interpretability. Here, we establish a diagnostic tool based on a deep-learning framework for the screening of patients with common treatable blinding retinal diseases. Our framework utilizes transfer learning, which trains a neural network with a fraction of the data of conventional approaches. Applying this approach to a dataset of optical coherence tomography images, we demonstrate performance comparable to that of human experts in classifying age-related macular degeneration and diabetic macular edema. We also provide a more transparent and interpretable diagnosis by highlighting the regions recognized by the neural network. We further demonstrate the general applicability of our AI system for diagnosis of pediatric pneumonia using chest X-ray images. This tool may ultimately aid in expediting the diagnosis and referral of these treatable conditions, thereby facilitating earlier treatment, resulting in improved clinical outcomes. VIDEO ABSTRACT.

P16INK4a Upregulation Mediated by SIX6 Defines Retinal Ganglion Cell Pathogenesis in Glaucoma.

Glaucoma, a blinding neurodegenerative disease, whose risk factors include elevated intraocular pressure (IOP), age, and genetics, is characterized by accelerated and progressive retinal ganglion cell (RGC) death. Despite decades of research, the mechanism of RGC death in glaucoma is still unknown. Here, we demonstrate that the genetic effect of the SIX6 risk variant (rs33912345, His141Asn) is enhanced by another major POAG risk gene, p16INK4a (cyclin-dependent kinase inhibitor 2A, isoform INK4a). We further show that the upregulation of homozygous SIX6 risk alleles (CC) leads to an increase in p16INK4a expression, with subsequent cellular senescence, as evidenced in a mouse model of elevated IOP and in human POAG eyes. Our data indicate that SIX6 and/or IOP promotes POAG by directly increasing p16INK4a expression, leading to RGC senescence in adult human retinas. Our study provides important insights linking genetic susceptibility to the underlying mechanism of RGC death and provides a unified theory of glaucoma pathogenesis.

Dynamin 1 Restrains Vesicular Release to a Subquantal Mode In Mammalian Adrenal Chromaffin Cells.

Dynamin 1 (dyn1) is required for clathrin-mediated endocytosis in most secretory (neuronal and neuroendocrine) cells. There are two modes of Ca2+-dependent catecholamine release from single dense-core vesicles: full-quantal (quantal) and subquantal in adrenal chromaffin cells, but their relative occurrences and impacts on total secretion remain unclear. To address this fundamental question in neurotransmission area using both sexes of animals, here we report the following: (1) dyn1-KO increased quantal size (QS, but not vesicle size/content) by ≥250% in dyn1-KO mice; (2) the KO-increased QS was rescued by dyn1 (but not its deficient mutant or dyn2); (3) the ratio of quantal versus subquantal events was increased by KO; (4) following a release event, more protein contents were retained in WT versus KO vesicles; and (5) the fusion pore size (dp) was increased from ≤9 to ≥9 nm by KO. Therefore, Ca2+-induced exocytosis is generally a subquantal release in sympathetic adrenal chromaffin cells, implying that neurotransmitter release is generally regulated by dynamin in neuronal cells.SIGNIFICANCE STATEMENT Ca2+-dependent neurotransmitter release from a single vesicle is the primary event in all neurotransmission, including synaptic/neuroendocrine forms. To determine whether Ca2+-dependent vesicular neurotransmitter release is "all-or-none" (quantal), we provide compelling evidence that most Ca2+-induced secretory events occur via the subquantal mode in native adrenal chromaffin cells. This subquantal release mode is promoted by dynamin 1, which is universally required for most secretory cells, including neurons and neuroendocrine cells. The present work with dyn1-KO mice further confirms that Ca2+-dependent transmitter release is mainly via subquantal mode, suggesting that subquantal release could be also important in other types of cells.

Impaired D2 receptor-dependent dopaminergic transmission in prefrontal cortex of awake mouse model of Parkinson's disease.

The loss-of-function mutation in PARK7/DJ-1 is one of the most common causes of autosomal recessive Parkinson's disease, and patients carrying PARK7 mutations often exhibit both a progressive movement disorder and emotional impairment, such as anxiety. However, the causes of the emotional symptom accompanying PARK7-associated and other forms of Parkinson's disease remain largely unexplored. Using two-photon microscopic Ca2+ imaging in awake PARK7-/- and PARK7+/+ mice, we found that (i) PARK7-/- neurons in the frontal association cortex showed substantially higher circuit activity recorded as spontaneous somatic Ca2+ signals; (ii) both basal and evoked dopamine release remained intact, as determined by both electrochemical dopamine recordings and high performance liquid chromatography in vivo; (iii) D2 receptor expression was significantly decreased in postsynaptic frontal association cortical neurons, and the hyper-neuronal activity were rescued by D2 receptor intervention using either local pharmacology or viral D2 receptor over-expression; and (iv) PARK7-/- mice showed anxiety-like behaviours that were rescued by either local D2 receptor pharmacology or overexpression. Thus, for first time, we demonstrated a robust D2 receptor-dependent phenotype of individual neurons within the prefrontal cortex circuit in awake parkinsonian mice that linked with anxiety. Our work sheds light on early-onset phenotypes and the mechanisms underlying Parkinson's disease by imaging brain circuits in an awake mouse model.

DNA methylation markers for diagnosis and prognosis of common cancers.

The ability to identify a specific cancer using minimally invasive biopsy holds great promise for improving the diagnosis, treatment selection, and prediction of prognosis in cancer. Using whole-genome methylation data from The Cancer Genome Atlas (TCGA) and machine learning methods, we evaluated the utility of DNA methylation for differentiating tumor tissue and normal tissue for four common cancers (breast, colon, liver, and lung). We identified cancer markers in a training cohort of 1,619 tumor samples and 173 matched adjacent normal tissue samples. We replicated our findings in a separate TCGA cohort of 791 tumor samples and 93 matched adjacent normal tissue samples, as well as an independent Chinese cohort of 394 tumor samples and 324 matched adjacent normal tissue samples. The DNA methylation analysis could predict cancer versus normal tissue with more than 95% accuracy in these three cohorts, demonstrating accuracy comparable to typical diagnostic methods. This analysis also correctly identified 29 of 30 colorectal cancer metastases to the liver and 32 of 34 colorectal cancer metastases to the lung. We also found that methylation patterns can predict prognosis and survival. We correlated differential methylation of CpG sites predictive of cancer with expression of associated genes known to be important in cancer biology, showing decreased expression with increased methylation, as expected. We verified gene expression profiles in a mouse model of hepatocellular carcinoma. Taken together, these findings demonstrate the utility of methylation biomarkers for the molecular characterization of cancer, with implications for diagnosis and prognosis.

Stretch-induced Ca2+ independent ATP release in hippocampal astrocytes.

KEY POINTS: Similar to neurons, astrocytes actively participate in synaptic transmission via releasing gliotransmitters. The Ca2+ -dependent release of gliotransmitters includes glutamate and ATP. Following an 'on-cell-like' mechanical stimulus to a single astrocyte, Ca2+ independent single, large, non-quantal, ATP release occurs. Astrocytic ATP release is inhibited by either selective antagonist treatment or genetic knockdown of P2X7 receptor channels. Our work suggests that ATP can be released from astrocytes via two independent pathways in hippocampal astrocytes; in addition to the known Ca2+ -dependent vesicular release, larger non-quantal ATP release depends on P2X7 channels following mechanical stretch. ABSTRACT: Astrocytic ATP release is essential for brain functions such as synaptic long-term potentiation for learning and memory. However, whether and how ATP is released via exocytosis remains hotly debated. All previous studies of non-vesicular ATP release have used indirect assays. By contrast, two recent studies report vesicular ATP release using more direct assays. In the present study, using patch clamped 'ATP-sniffer cells', we re-investigated astrocytic ATP release at single-vesicle resolution in hippocampal astrocytes. Following an 'on-cell-like' mechanical stimulus of a single astrocyte, a Ca2+ independent single large non-quantal ATP release occurred, in contrast to the Ca2+ -dependent multiple small quantal ATP release in a chromaffin cell. The mechanical stimulation-induced ATP release from an astrocyte was inhibited by either exposure to a selective antagonist or genetic knockdown of P2X7 receptor channels. Functional P2X7 channels were expressed in astrocytes in hippocampal brain slices. Thus, in addition to small quantal ATP release, larger non-quantal ATP release depends on P2X7 channels in astrocytes.

Circulating tumour DNA methylation markers for diagnosis and prognosis of hepatocellular carcinoma.

An effective blood-based method for the diagnosis and prognosis of hepatocellular carcinoma (HCC) has not yet been developed. Circulating tumour DNA (ctDNA) carrying cancer-specific genetic and epigenetic aberrations may enable a noninvasive 'liquid biopsy' for diagnosis and monitoring of cancer. Here, we identified an HCC-specific methylation marker panel by comparing HCC tissue and normal blood leukocytes and showed that methylation profiles of HCC tumour DNA and matched plasma ctDNA are highly correlated. Using cfDNA samples from a large cohort of 1,098 HCC patients and 835 normal controls, we constructed a diagnostic prediction model that showed high diagnostic specificity and sensitivity (P < 0.001) and was highly correlated with tumour burden, treatment response, and stage. Additionally, we constructed a prognostic prediction model that effectively predicted prognosis and survival (P < 0.001). Together, these findings demonstrate in a large clinical cohort the utility of ctDNA methylation markers in the diagnosis, surveillance, and prognosis of HCC.

Noninvasive detection of fetal subchromosomal abnormalities by semiconductor sequencing of maternal plasma DNA.

Noninvasive prenatal testing (NIPT) using sequencing of fetal cell-free DNA from maternal plasma has enabled accurate prenatal diagnosis of aneuploidy and become increasingly accepted in clinical practice. We investigated whether NIPT using semiconductor sequencing platform (SSP) could reliably detect subchromosomal deletions/duplications in women carrying high-risk fetuses. We first showed that increasing concentration of abnormal DNA and sequencing depth improved detection. Subsequently, we analyzed plasma from 1,456 pregnant women to develop a method for estimating fetal DNA concentration based on the size distribution of DNA fragments. Finally, we collected plasma from 1,476 pregnant women with fetal structural abnormalities detected on ultrasound who also underwent an invasive diagnostic procedure. We used SSP of maternal plasma DNA to detect subchromosomal abnormalities and validated our results with array comparative genomic hybridization (aCGH). With 3.5 million reads, SSP detected 56 of 78 (71.8%) subchromosomal abnormalities detected by aCGH. With increased sequencing depth up to 10 million reads and restriction of the size of abnormalities to more than 1 Mb, sensitivity improved to 69 of 73 (94.5%). Of 55 false-positive samples, 35 were caused by deletions/duplications present in maternal DNA, indicating the necessity of a validation test to exclude maternal karyotype abnormalities. This study shows that detection of fetal subchromosomal abnormalities is a viable extension of NIPT based on SSP. Although we focused on the application of cell-free DNA sequencing for NIPT, we believe that this method has broader applications for genetic diagnosis, such as analysis of circulating tumor DNA for detection of cancer.

Sensory Response of Transplanted Astrocytes in Adult Mammalian Cortex In Vivo.

Glial precursor transplantation provides a potential therapy for brain disorders. Before its clinical application, experimental evidence needs to indicate that engrafted glial cells are functionally incorporated into the existing circuits and become essential partners of neurons for executing fundamental brain functions. While previous experiments supporting for their functional integration have been obtained under in vitro conditions using slice preparations, in vivo evidence for such integration is still lacking. Here, we utilized in vivo two-photon Ca(2+) imaging along with immunohistochemistry, fluorescent indicator labeling-based axon tracing and correlated light/electron microscopy to analyze the profiles and the functional status of glial precursor cell-derived astrocytes in adult mouse neocortex. We show that after being transplanted into somatosensory cortex, precursor-derived astrocytes are able to survive for more than a year and respond with Ca(2+) signals to sensory stimulation. These sensory-evoked responses are mediated by functionally-expressed nicotinic receptors and newly-established synaptic contacts with the host cholinergic afferents. Our results provide in vivo evidence for a functional integration of transplanted astrocytes into adult mammalian neocortex, representing a proof-of-principle for sensory cortex remodeling through addition of essential neural elements. Moreover, we provide strong support for the use of glial precursor transplantation to understand glia-related neural development in vivo.

Calcium influx activates adenylyl cyclase 8 for sustained insulin secretion in rat pancreatic beta cells.

AIMS/HYPOTHESIS: Insulin is a key metabolic regulator in health and diabetes. In pancreatic beta cells, insulin release is regulated by the major second messengers Ca(2+) and cAMP: exocytosis is triggered by Ca(2+) and mediated by the cAMP/protein kinase A (PKA) signalling pathway. However, the causal link between these two processes in primary beta cells remains undefined. METHODS: Time-resolved confocal imaging of fluorescence resonance energy transfer signals was performed to visualise PKA activity, and combined membrane capacitance recordings were used to monitor insulin secretion from patch-clamped rat beta cells. RESULTS: Membrane depolarisation-induced Ca(2+) influx caused an increase in cytosolic PKA activity via activating a Ca(2+)-sensitive adenylyl cyclase 8 (ADCY8) subpool. Glucose stimulation triggered coupled Ca(2+) oscillations and PKA activation. ADCY8 knockdown significantly reduced the level of depolarisation-evoked PKA activation and impaired replenishment of the readily releasable vesicle pool. Pharmacological inhibition of PKA by two inhibitors reduced depolarisation-induced PKA activation to a similar extent and reduced the capacity for sustained vesicle exocytosis and insulin release. CONCLUSIONS/INTERPRETATION: Our findings suggest that depolarisation-induced Ca(2+) influx plays dual roles in regulating exocytosis in rat pancreatic beta cells by triggering vesicle fusion and replenishing the vesicle pool to support sustained insulin release. Therefore, Ca(2+) influx may be important for glucose-stimulated insulin secretion.

Real-time endocytosis imaging as a rapid assay of ligand-GPCR binding in single cells.

Most G protein-coupled receptors (GPCRs) do not generate membrane currents in response to ligand-receptor binding (LRB). Here, we describe a novel technique using endocytosis as a bioassay that can detect activation of a GPCR in a way analogous to patch-clamp recording of an ion channel in a living cell. The confocal imaging technique, termed FM endocytosis imaging (FEI), can record ligand-GPCR binding with high temporal (second) and spatial (micrometer) resolution. LRB leads to internalization of an endocytic vesicle, which can be labeled by a styryl FM dye and visualized as a fluorescent spot. Distinct from the green fluorescence protein-labeling method, FEI can detect LRB endocytosis mediated by essentially any receptors (GPCRs or receptors of tyrosine kinase) in a native cell/cell line. Three modified versions of FEI permit promising applications in functional GPCR studies and drug screening in living cells: 1) LRB can be recorded in "real time" (time scale of seconds); 2) internalized vesicles mediated by different GPCRs can be discriminated by different colors; and 3) a high throughput method can screen ligands of a specific GPCR.

Calcium triggers exocytosis from two types of organelles in a single astrocyte.

Astrocytes release a variety of signaling molecules including glutamate, D-serine, and ATP in a regulated manner. Although the functions of these molecules, from regulating synaptic transmission to controlling specific behavior, are well documented, the identity of their cellular compartment(s) is still unclear. Here we set out to study vesicular exocytosis and glutamate release in mouse hippocampal astrocytes. We found that small vesicles and lysosomes coexisted in the same freshly isolated or cultured astrocytes. Both small vesicles and lysosome fused with the plasma membrane in the same astrocytes in a Ca(2+)-regulated manner, although small vesicles were exocytosed more efficiently than lysosomes. Blockade of the vesicle glutamate transporter or cleavage of synaptobrevin 2 and cellubrevin (both are vesicle-associated membrane proteins) with a clostridial toxin greatly inhibited glutamate release from astrocytes, while lysosome exocytosis remained intact. Thus, both small vesicles and lysosomes contribute to Ca(2+)-dependent vesicular exocytosis, and small vesicles support glutamate release from astrocytes.

Thiopental-induced insulin secretion via activation of IP3-sensitive calcium stores in rat pancreatic ß-cells.

While glucose-stimulated insulin secretion depends on Ca(2+) influx through voltage-gated Ca(2+) channels in the cell membrane of the pancreatic ß-cell, there is also ample evidence for an important role of intracellular Ca(2+) stores in insulin secretion, particularly in relation to drug stimuli. We report here that thiopental, a common anesthetic agent, triggers insulin secretion from the intact pancreas and primary cultured rat pancreatic ß-cells. We investigated the underlying mechanisms by measurements of whole cell K(+) and Ca(2+) currents, membrane potential, cytoplasmic Ca(2+) concentration ([Ca(2+)](i)), and membrane capacitance. Thiopental-induced insulin secretion was first detected by enzyme-linked immunoassay, then further assessed by membrane capacitance measurement, which revealed kinetics distinct from glucose-induced insulin secretion. The thiopental-induced secretion was independent of cell membrane depolarization and closure of ATP-sensitive potassium (K(ATP)) channels. However, accompanied by the insulin secretion stimulated by thiopental, we recorded a significant intracellular [Ca(2+)] increase that was not from Ca(2+) influx across the cell membrane, but from intracellular Ca(2+) stores. The thiopental-induced [Ca(2+)](i) rise in ß-cells was sensitive to thapsigargin, a blocker of the endoplasmic reticulum Ca(2+) pump, as well as to heparin (0.1 mg/ml) and 2-aminoethoxydiphenyl borate (2-APB; 100 µM), drugs that inhibit inositol 1,4,5-trisphosphate (IP(3)) binding to the IP(3) receptor, and to U-73122, a phospholipase C inhibitor, but insensitive to ryanodine. Thapsigargin also diminished thiopental-induced insulin secretion. Thus, we conclude that thiopental-induced insulin secretion is mediated by activation of the intracellular IP(3)-sensitive Ca(2+) store.

Two distinct vesicle pools for depolarization-induced exocytosis in somata of dorsal root ganglion neurons.

The somata of dorsal root ganglion (DRG) neurons release neurotransmitters and neuropeptides. In addition to the conventional Ca2+-dependent secretion (CDS), Ca2+-independent but voltage-dependent secretion (CIVDS) also occurs in the somata of DRG neurons. Electrical stimulation induces both CDS and CIVDS, which differ in size and are coupled with different types of endocytosis contributed by CIVDS and CDS, respectively. However, it is unclear whether they use a common vesicle pool, so we investigated the relationship between the vesicle pools of CDS and CIVDS. Membrane capacitance recording and photolysis of a caged-Ca2+ compound showed that, in low external Ca2+ solutions, the depolarization-induced exocytosis contained two (fast and slow) phases, which were contributed by CIVDS and CDS, respectively. Depletion of the CDS readily releasable pool using photolysis did not affect the CIVDS. When the CIVDS and CDS vesicle pools were depleted by electrical stimulation, the pools had different sizes. Their kinetics of exocytosis-coupled endocytosis were also different. Thus, CIVDS and CDS used different vesicle pools in DRG neurons.

Circulating tumour DNA methylation in hepatocellular carcinoma diagnosis using digital droplet PCR.

OBJECTIVE: To evaluate the performance of a DNA methylation-based digital droplet polymerase chain reaction (ddPCR) assay to detect aberrant DNA methylation in cell-free DNA (cfDNA) and to determine its application in the detection of hepatocellular carcinoma (HCC). METHODS: The present study recruited patients with liver-related diseases and healthy control subjects. Blood samples were used for the extraction of cfDNA, which was then bisulfite converted and the extent of DNA methylation quantified using a ddPCR platform. RESULTS: A total of 97 patients with HCC, 80 healthy control subjects and 46 patients with chronic hepatitis B/C virus infection were enrolled in the study. The level of cfDNA in the HCC group was significantly higher than that in the healthy control group. For the detection of HCC, based on a cut-off value of 15.7% for the cfDNA methylation ratio, the sensitivity and specificity were 78.57% and 89.38%, respectively. The diagnostic accuracy was 85.27%, the positive predictive value was 81.91% and the negative predictive value was 87.20%. The positive likelihood ratio of 15.7% in HCC diagnosis was 7.40, while the negative likelihood ratio was 0.24. CONCLUSIONS: A sensitive methylation-based assay might serve as a liquid biopsy test for diagnosing HCC.

A deep-learning pipeline for the diagnosis and discrimination of viral, non-viral and COVID-19 pneumonia from chest X-ray images.

Common lung diseases are first diagnosed using chest X-rays. Here, we show that a fully automated deep-learning pipeline for the standardization of chest X-ray images, for the visualization of lesions and for disease diagnosis can identify viral pneumonia caused by coronavirus disease 2019 (COVID-19) and assess its severity, and can also discriminate between viral pneumonia caused by COVID-19 and other types of pneumonia. The deep-learning system was developed using a heterogeneous multicentre dataset of 145,202 images, and tested retrospectively and prospectively with thousands of additional images across four patient cohorts and multiple countries. The system generalized across settings, discriminating between viral pneumonia, other types of pneumonia and the absence of disease with areas under the receiver operating characteristic curve (AUCs) of 0.94-0.98; between severe and non-severe COVID-19 with an AUC of 0.87; and between COVID-19 pneumonia and other viral or non-viral pneumonia with AUCs of 0.87-0.97. In an independent set of 440 chest X-rays, the system performed comparably to senior radiologists and improved the performance of junior radiologists. Automated deep-learning systems for the assessment of pneumonia could facilitate early intervention and provide support for clinical decision-making.

Circulating tumor DNA methylation profiles enable early diagnosis, prognosis prediction, and screening for colorectal cancer.

Circulating tumor DNA (ctDNA) has emerged as a useful diagnostic and prognostic biomarker in many cancers. Here, we conducted a study to investigate the potential use of ctDNA methylation markers for the diagnosis and prognostication of colorectal cancer (CRC) and used a prospective cohort to validate their effectiveness in screening patients at high risk of CRC. We first identified CRC-specific methylation signatures by comparing CRC tissues to normal blood leukocytes. Then, we applied a machine learning algorithm to develop a predictive diagnostic and a prognostic model using cell-free DNA (cfDNA) samples from a cohort of 801 patients with CRC and 1021 normal controls. The obtained diagnostic prediction model discriminated patients with CRC from normal controls with high accuracy (area under curve = 0.96). The prognostic prediction model also effectively predicted the prognosis and survival of patients with CRC (P < 0.001). In addition, we generated a ctDNA-based molecular classification of CRC using an unsupervised clustering method and obtained two subgroups of patients with CRC with significantly different overall survival (P = 0.011 in validation cohort). Last, we found that a single ctDNA methylation marker, cg10673833, could yield high sensitivity (89.7%) and specificity (86.8%) for detection of CRC and precancerous lesions in a high-risk population of 1493 participants in a prospective cohort study. Together, our findings showed the value of ctDNA methylation markers in the diagnosis, surveillance, and prognosis of CRC.

DNA methylation markers in the diagnosis and prognosis of common leukemias.

The ability to identify a specific type of leukemia using minimally invasive biopsies holds great promise to improve the diagnosis, treatment selection, and prognosis prediction of patients. Using genome-wide methylation profiling and machine learning methods, we investigated the utility of CpG methylation status to differentiate blood from patients with acute lymphocytic leukemia (ALL) or acute myelogenous leukemia (AML) from normal blood. We established a CpG methylation panel that can distinguish ALL and AML blood from normal blood as well as ALL blood from AML blood with high sensitivity and specificity. We then developed a methylation-based survival classifier with 23 CpGs for ALL and 20 CpGs for AML that could successfully divide patients into high-risk and low-risk groups, with significant differences in clinical outcome in each leukemia type. Together, these findings demonstrate that methylation profiles can be highly sensitive and specific in the accurate diagnosis of ALL and AML, with implications for the prediction of prognosis and treatment selection.