The Molecular Twin artificial-intelligence platform integrates multi-omic data to predict outcomes for pancreatic adenocarcinoma patients.

Contemporary analyses focused on a limited number of clinical and molecular biomarkers have been unable to accurately predict clinical outcomes in pancreatic ductal adenocarcinoma. Here we describe a precision medicine platform known as the Molecular Twin consisting of advanced machine-learning models and use it to analyze a dataset of 6,363 clinical and multi-omic molecular features from patients with resected pancreatic ductal adenocarcinoma to accurately predict disease survival (DS). We show that a full multi-omic model predicts DS with the highest accuracy and that plasma protein is the top single-omic predictor of DS. A parsimonious model learning only 589 multi-omic features demonstrated similar predictive performance as the full multi-omic model. Our platform enables discovery of parsimonious biomarker panels and performance assessment of outcome prediction models learning from resource-intensive panels. This approach has considerable potential to impact clinical care and democratize precision cancer medicine worldwide.

ß-adrenergic receptor inhibition affects cerebral glucose metabolism, motor performance, and inflammatory response after traumatic brain injury.

BACKGROUND: The purpose of this study was to evaluate how ß-adrenergic receptor inhibition after traumatic brain injury (TBI) alters changes in early cerebral glucose metabolism and motor performance, as well as cerebral cytokine and heat shock protein (HSP) expression. METHODS: Mouse cerebral glucose metabolism was measured by microPET fluorodeoxyglucose uptake and converted into standardized uptake values (SUV). Four groups of C57/Bl6 mice (wild type [WT]) were initially evaluated: sham or TBI, followed by tail vein injection of either saline or a nonselective ß-adrenergic receptor inhibitor (propranolol, 4 mg/kg). Then motor performance, cerebral cytokine, and HSP70 expression were studied at 12 hours and 24 hours after sham injury or TBI in WT mice treated with saline or propranolol and in ß1-adrenergic/ß2-adrenergic receptor knockout (BARKO) mice treated with saline. RESULTS: Cerebral glucose metabolism was significantly reduced after TBI (mean SUV TBI, 1.63 vs. sham 1.97, p < 0.01) and propranolol attenuated this reduction (mean SUV propranolol, 1.89 vs. saline 1.63, p < 0.01). Both propranolol and BARKO reduced motor deficits at 24 hours after injury, but only BARKO had an effect at 12 hours after injury. TBI WT mice treated with saline performed worse than propranolol mice at 24 hours after injury on rotarod (23 vs. 44 seconds, p < 0.01) and rearing (130 vs. 338 events, p = 0.01) results. At 24 hours after injury, sham BARKO and TBI BARKO mice were similar on rotarod (21 vs. 19 seconds, p = 0.53), ambulatory testing (2,891 vs. 2,274 events, p = 0.14), and rearing (129 vs. 64 events, p = 0.09) results. Interleukin 1ß expression was affected by BARKO and propranolol after TBI; attenuation of interleukin 6 and increased HSP70 expression were noted only with BARKO. CONCLUSION: ß-adrenergic receptor inhibition affects cerebral glucose metabolism, motor performance, as well as cerebral cytokine and HSP expression after TBI.

A DICOM-based 2nd generation Molecular Imaging Data Grid implementing the IHE XDS-i integration profile.

PURPOSE: A Molecular Imaging Data Grid (MIDG) was developed to address current informatics challenges in archival, sharing, search, and distribution of preclinical imaging studies between animal imaging facilities and investigator sites. This manuscript presents a 2nd generation MIDG replacing the Globus Toolkit with a new system architecture that implements the IHE XDS-i integration profile. Implementation and evaluation were conducted using a 3-site interdisciplinary test-bed at the University of Southern California. METHODS: The 2nd generation MIDG design architecture replaces the initial design's Globus Toolkit with dedicated web services and XML-based messaging for dedicated management and delivery of multi-modality DICOM imaging datasets. The Cross-enterprise Document Sharing for Imaging (XDS-i) integration profile from the field of enterprise radiology informatics was adopted into the MIDG design because streamlined image registration, management, and distribution dataflow are likewise needed in preclinical imaging informatics systems as in enterprise PACS application. Implementation of the MIDG is demonstrated at the University of Southern California Molecular Imaging Center (MIC) and two other sites with specified hardware, software, and network bandwidth. RESULTS: Evaluation of the MIDG involves data upload, download, and fault-tolerance testing scenarios using multi-modality animal imaging datasets collected at the USC Molecular Imaging Center. The upload, download, and fault-tolerance tests of the MIDG were performed multiple times using 12 collected animal study datasets. Upload and download times demonstrated reproducibility and improved real-world performance. Fault-tolerance tests showed that automated failover between Grid Node Servers has minimal impact on normal download times. CONCLUSIONS: Building upon the 1st generation concepts and experiences, the 2nd generation MIDG system improves accessibility of disparate animal-model molecular imaging datasets to users outside a molecular imaging facility's LAN using a new architecture, dataflow, and dedicated DICOM-based management web services. Productivity and efficiency of preclinical research for translational sciences investigators has been further streamlined for multi-center study data registration, management, and distribution.