Early prediction of decompensation (EPOD) score: Non-invasive determination of cirrhosis decompensation risk.

BACKGROUND & AIMS: Decompensation is a hallmark of disease progression in cirrhotic patients. Early detection of a phase transition from compensated cirrhosis to decompensation would enable targeted therapeutic interventions potentially extending life expectancy. This study aims to (a) identify the predictors of decompensation in a large, multicentric cohort of patients with compensated cirrhosis, (b) to build a reliable prognostic score for decompensation and (c) to evaluate the score in independent cohorts. METHODS: Decompensation was identified in electronic health records data from 6049 cirrhosis patients in the IBM Explorys database training cohort by diagnostic codes for variceal bleeding, encephalopathy, ascites, hepato-renal syndrome and/or jaundice. We identified predictors of clinical decompensation and developed a prognostic score using Cox regression analysis. The score was evaluated using the IBM Explorys database validation cohort (N = 17662), the Penn Medicine BioBank (N = 1326) and the UK Biobank (N = 317). RESULTS: The new Early Prediction of Decompensation (EPOD) score uses platelet count, albumin, and bilirubin concentration. It predicts decompensation during a 3-year follow-up in three validation cohorts with AUROCs of 0.69, 0.69 and 0.77, respectively, and outperforms the well-known MELD and Child-Pugh score in predicting decompensation. Furthermore, the EPOD score predicted the 3-year probability of decompensation. CONCLUSIONS: The EPOD score provides a prediction tool for the risk of decompensation in patients with cirrhosis that outperforms well-known cirrhosis scores. Since EPOD is based on three blood parameters, only, it provides maximal clinical feasibility at minimal costs.

Deletion of Neuronal GLT-1 in Mice Reveals Its Role in Synaptic Glutamate Homeostasis and Mitochondrial Function.

The glutamate transporter GLT-1 is highly expressed in astrocytes but also in neurons, primarily in axon terminals. We generated a conditional neuronal GLT-1 KO using synapsin 1-Cre (synGLT-1 KO) to elucidate the metabolic functions of GLT-1 expressed in neurons, here focusing on the cerebral cortex. Both synaptosomal uptake studies and electron microscopic immunocytochemistry demonstrated knockdown of GLT-1 in the cerebral cortex in the synGLT-1 KO mice. Aspartate content was significantly reduced in cerebral cortical extracts as well as synaptosomes from cerebral cortex of synGLT-1 KO compared with control littermates. 13C-Labeling of tricarboxylic acid cycle intermediates originating from metabolism of [U-13C]-glutamate was significantly reduced in synGLT-1 KO synaptosomes. The decreased aspartate content was due to diminished entry of glutamate into the tricarboxylic acid cycle. Pyruvate recycling, a pathway necessary for full glutamate oxidation, was also decreased. ATP production was significantly increased, despite unaltered oxygen consumption, in isolated mitochondria from the synGLT-1 KO. The density of mitochondria in axon terminals and perisynaptic astrocytes was increased in the synGLT-1 KO. Intramitochondrial cristae density of synGLT-1 KO mice was increased, suggesting increased mitochondrial efficiency, perhaps in compensation for reduced access to glutamate. SynGLT-1 KO synaptosomes exhibited an elevated oxygen consumption rate when stimulated with veratridine, despite a lower baseline oxygen consumption rate in the presence of glucose. GLT-1 expressed in neurons appears to be required to provide glutamate to synaptic mitochondria and is linked to neuronal energy metabolism and mitochondrial function.SIGNIFICANCE STATEMENT All synaptic transmitters need to be cleared from the extracellular space after release, and transporters are used to clear glutamate released from excitatory synapses. GLT-1 is the major glutamate transporter, and most GLT-1 is expressed in astrocytes. Only 5%-10% is expressed in neurons, primarily in axon terminals. The function of GLT-1 in axon terminals remains unknown. Here, we used a conditional KO approach to investigate the significance of the expression of GLT-1 in neurons. We found multiple abnormalities of mitochondrial function, suggesting impairment of glutamate utilization by synaptic mitochondria in the neuronal GLT-1 KO. These data suggest that GLT-1 expressed in axon terminals may be important in maintaining energy metabolism and biosynthetic activities mediated by presynaptic mitochondria.

Mechanisms of sleep and circadian ontogeny through the lens of neurodevelopmental disorders.

Sleep is a mysterious, developmentally regulated behavior fundamental for cognition in both adults and developing animals. A large number of studies offer a substantive body of evidence that demonstrates that the ontogeny of sleep architecture parallels brain development. Sleep deprivation impairs the consolidation of learned tasks into long-term memories and likely links sleep to the neural mechanisms underlying memory and its physiological roots in brain plasticity. Consistent with this notion is the alarming frequency of sleep and circadian rhythm dysfunction in children with neurodevelopmental disorders (NDDs). While the mechanisms underlying sleep dysfunction in most NDDs still remains poorly understood, here we will review several sentinel examples of monogenetic NDDs with both well-established connections to synaptic dysfunction and evidence of sleep or circadian dysfunction: Tuberous Sclerosis Complex, Fragile X Syndrome, and Angelman Syndrome. We suggest that the coincident maturation of sleep with synaptic physiology is one of the core reasons for the commonplace disruption of sleep in NDDs and argue that disorders with well-defined molecular genetics can provide a unique lens for understanding and unraveling the molecular correlates that link the development of sleep and circadian rhythms to health and disease.

Stochastic modelling of T-cell activation.

We investigate a specific part of the human immune system, namely the activation of T-cells, using stochastic tools, especially sharp large deviation results. T-cells have to distinguish reliably between foreign and self peptides which are both presented to them by antigen presenting cells. Our work is based on a model studied by Zint et al. (J Math Bio 57(6):841-861, 2008). We are able to dispense with some restrictive distribution assumptions that were used previously, i.e., we establish a higher robustness of the model. A central issue is the analysis of two new perspectives to the scenario (two different quenched systems) in detail. This means that we do not only analyse the total probability of a T-cell activation (the annealed case) but also consider the probability of an activation of one certain clonotype and the probability of a T-cell activation by a certain antigen presentation profile (the quenched cases). Finally, we see analytically that the probability of T-cell activation increases with the number of presented foreign peptides in all three cases.

Computational Simulation of Virtual Patients Reduces Dataset Bias and Improves Machine Learning-Based Detection of ARDS from Noisy Heterogeneous ICU Datasets.

Goal: Machine learning (ML) technologies that leverage large-scale patient data are promising tools predicting disease evolution in individual patients. However, the limited generalizability of ML models developed on single-center datasets, and their unproven performance in real-world settings, remain significant constraints to their widespread adoption in clinical practice. One approach to tackle this issue is to base learning on large multi-center datasets. However, such heterogeneous datasets can introduce further biases driven by data origin, as data structures and patient cohorts may differ between hospitals. Methods: In this paper, we demonstrate how mechanistic virtual patient (VP) modeling can be used to capture specific features of patients' states and dynamics, while reducing biases introduced by heterogeneous datasets. We show how VP modeling can be used for data augmentation through identification of individualized model parameters approximating disease states of patients with suspected acute respiratory distress syndrome (ARDS) from observational data of mixed origin. We compare the results of an unsupervised learning method (clustering) in two cases: where the learning is based on original patient data and on data derived in the matching procedure of the VP model to real patient data. Results: More robust cluster configurations were observed in clustering using the model-derived data. VP model-based clustering also reduced biases introduced by the inclusion of data from different hospitals and was able to discover an additional cluster with significant ARDS enrichment. Conclusions: Our results indicate that mechanistic VP modeling can be used to significantly reduce biases introduced by learning from heterogeneous datasets and to allow improved discovery of patient cohorts driven exclusively by medical conditions.

Efficient Microfluidic Downstream Processes for Rapid Antibody Hit Confirmation.

Microfluidics has been recently applied to better understand the spatial and temporal progression of the immune response in several species, for tool and biotherapeutic production cell line development and rapid antibody hit discovery. Several technologies have emerged that allow interrogation of large diversities of antibody-secreting cells in defined compartments such as picoliter droplets or nanopens. Mostly primary cells of immunized rodents but also recombinant mammalian libraries are screened for specific binding or directly for the desired function. While post-microfluidic downstream processes appear as standard steps, they represent considerable and interdependent challenges that can lead to high attrition rates even if original selections had been successful. In addition to next-generation sequencing recently described in depth elsewhere, this report aims at in detail explanations of exemplary droplet-based sorting followed by single-cell antibody gene PCR recovery and reproduction or single-cell sub-cultivation for crude supernatant confirmatory studies.

An integrated mammalian library approach for optimization and enhanced microfluidics-assisted antibody hit discovery.

Recent years have seen the development of a variety of mammalian library approaches for display and secretion mode. Advantages include library approaches for engineering, preservation of precious immune repertoires and their repeated interrogation, as well as screening in final therapeutic format and host. Mammalian display approaches for antibody optimization exploit these advantages, necessitating the generation of large libraries but in turn enabling early screening for both manufacturability and target specificity. For suitable libraries, high antibody integration rates and resulting monoclonality need to be balanced - we present a solution for sufficient transmutability and acceptable monoclonality by applying an optimized ratio of coding to non-coding lentivirus. The recent advent of microfluidic-assisted hit discovery represents a perfect match to mammalian libraries in secretion mode, as the lower throughput fits well with the facile generation of libraries comprising a few million functional clones. In the presented work, Chinese Hamster Ovary cells were engineered to both express the target of interest and secrete antibodies in relevant formats, and specific clones were strongly enriched by high throughput screening for autocrine cellular binding. The powerful combination of mammalian secretion libraries and microfluidics-assisted hit discovery could reduce attrition rates and increase the probability to identify the best possible therapeutic antibody hits faster.

Mammalian display to secretion switchable libraries for antibody preselection and high throughput functional screening.

Recently, there has been a co-evolution of mammalian libraries and diverse microfluidic approaches for therapeutic antibody hit discovery. Mammalian libraries enable the preservation of full immune repertoires, produce hit candidates in final format and facilitate broad combinatorial bispecific antibody screening, while several available microfluidic methodologies offer opportunities for rapid high-content screens. Here, we report proof-of-concept studies exploring the potential of combining microfluidic technologies with mammalian libraries for antibody discovery. First, antibody secretion, target co-expression and integration of appropriate reporter cell lines enabled the selection of in-trans acting agonistic bispecific antibodies. Second, a functional screen for internalization was established and comparison of autocrine versus co-encapsulation setups highlighted the advantages of an autocrine one cell approach. Third, synchronization of antibody-secreting cells prior to microfluidic screens reduced assay variability. Furthermore, a display to secretion switchable system was developed and applied for pre-enrichment of antibody clones with high manufacturability in conjunction with subsequent screening for functional properties. These case studies demonstrate the system's feasibility and may serve as basis for further development of integrated workflows combining manufacturability sorting and functional screens for the identification of optimal therapeutic antibody candidates.

Synaptic BMAL1 phosphorylation controls circadian hippocampal plasticity.

The time of day strongly influences adaptive behaviors like long-term memory, but the correlating synaptic and molecular mechanisms remain unclear. The circadian clock comprises a canonical transcription-translation feedback loop (TTFL) strictly dependent on the BMAL1 transcription factor. We report that BMAL1 rhythmically localizes to hippocampal synapses in a manner dependent on its phosphorylation at Ser42 [pBMAL1(S42)]. pBMAL1(S42) regulates the autophosphorylation of synaptic CaMKIIα and circadian rhythms of CaMKIIα-dependent molecular interactions and LTP but not global rest/activity behavior. Therefore, our results suggest a model in which repurposing of the clock protein BMAL1 to synapses locally gates the circadian timing of plasticity.

Defining diurnal fluctuations in mouse choroid plexus and CSF at high molecular, spatial, and temporal resolution.

Transmission and secretion of signals via the choroid plexus (ChP) brain barrier can modulate brain states via regulation of cerebrospinal fluid (CSF) composition. Here, we developed a platform to analyze diurnal variations in male mouse ChP and CSF. Ribosome profiling of ChP epithelial cells revealed diurnal translatome differences in metabolic machinery, secreted proteins, and barrier components. Using ChP and CSF metabolomics and blood-CSF barrier analyses, we observed diurnal changes in metabolites and cellular junctions. We then focused on transthyretin (TTR), a diurnally regulated thyroid hormone chaperone secreted by the ChP. Diurnal variation in ChP TTR depended on Bmal1 clock gene expression. We achieved real-time tracking of CSF-TTR in awake TtrmNeonGreen mice via multi-day intracerebroventricular fiber photometry. Diurnal changes in ChP and CSF TTR levels correlated with CSF thyroid hormone levels. These datasets highlight an integrated platform for investigating diurnal control of brain states by the ChP and CSF.

Application of convex hull analysis for the evaluation of data heterogeneity between patient populations of different origin and implications of hospital bias in downstream machine-learning-based data processing: A comparison of 4 critical-care patient datasets.

Machine learning (ML) models are developed on a learning dataset covering only a small part of the data of interest. If model predictions are accurate for the learning dataset but fail for unseen data then generalization error is considered high. This problem manifests itself within all major sub-fields of ML but is especially relevant in medical applications. Clinical data structures, patient cohorts, and clinical protocols may be highly biased among hospitals such that sampling of representative learning datasets to learn ML models remains a challenge. As ML models exhibit poor predictive performance over data ranges sparsely or not covered by the learning dataset, in this study, we propose a novel method to assess their generalization capability among different hospitals based on the convex hull (CH) overlap between multivariate datasets. To reduce dimensionality effects, we used a two-step approach. First, CH analysis was applied to find mean CH coverage between each of the two datasets, resulting in an upper bound of the prediction range. Second, 4 types of ML models were trained to classify the origin of a dataset (i.e., from which hospital) and to estimate differences in datasets with respect to underlying distributions. To demonstrate the applicability of our method, we used 4 critical-care patient datasets from different hospitals in Germany and USA. We estimated the similarity of these populations and investigated whether ML models developed on one dataset can be reliably applied to another one. We show that the strongest drop in performance was associated with the poor intersection of convex hulls in the corresponding hospitals' datasets and with a high performance of ML methods for dataset discrimination. Hence, we suggest the application of our pipeline as a first tool to assess the transferability of trained models. We emphasize that datasets from different hospitals represent heterogeneous data sources, and the transfer from one database to another should be performed with utmost care to avoid implications during real-world applications of the developed models. Further research is needed to develop methods for the adaptation of ML models to new hospitals. In addition, more work should be aimed at the creation of gold-standard datasets that are large and diverse with data from varied application sites.

Population pharmacokinetic analysis of rivaroxaban in children and comparison to prospective physiologically-based pharmacokinetic predictions.

Rivaroxaban has been investigated in the EINSTEIN-Jr program for the treatment of acute venous thromboembolism (VTE) in children aged 0 to 18 years and in the UNIVERSE program for thromboprophylaxis in children aged 2 to 8 years with congenital heart disease after Fontan-procedure. Physiologically-based pharmacokinetic (PBPK) and population pharmacokinetic (PopPK) modeling were used throughout the pediatric development of rivaroxaban according to the learn-and-confirm paradigm. The development strategy was to match pediatric drug exposures to adult exposure proven to be safe and efficacious. In this analysis, a refined pediatric PopPK model for rivaroxaban based on integrated EINSTEIN-Jr data and interim PK data from part A of the UNIVERSE phase III study was developed and the influence of potential covariates and intrinsic factors on rivaroxaban exposure was assessed. The model adequately described the observed pediatric PK data. PK parameters and exposure metrics estimated by the PopPK model were compared to the predictions from a previously published pediatric PBPK model for rivaroxaban. Ninety-one percent of the individual post hoc clearance estimates were found within the 5th to 95th percentile of the PBPK model predictions. In patients below 2 years of age, however, clearance was underpredicted by the PBPK model. The iterative and integrative use of PBPK and PopPK modeling and simulation played a major role in the establishment of the bodyweight-adjusted rivaroxaban dosing regimen that was ultimately confirmed to be a safe and efficacious dosing regimen for children aged 0 to 18 years with acute VTE in the EINSTEIN-Jr phase III study.

Algorithmic surveillance of ICU patients with acute respiratory distress syndrome (ASIC): protocol for a multicentre stepped-wedge cluster randomised quality improvement strategy.

INTRODUCTION: The acute respiratory distress syndrome (ARDS) is a highly relevant entity in critical care with mortality rates of 40%. Despite extensive scientific efforts, outcome-relevant therapeutic measures are still insufficiently practised at the bedside. Thus, there is a clear need to adhere to early diagnosis and sufficient therapy in ARDS, assuring lower mortality and multiple organ failure. METHODS AND ANALYSIS: In this quality improvement strategy (QIS), a decision support system as a mobile application (ASIC app), which uses available clinical real-time data, is implemented to support physicians in timely diagnosis and improvement of adherence to established guidelines in the treatment of ARDS. ASIC is conducted on 31 intensive care units (ICUs) at 8 German university hospitals. It is designed as a multicentre stepped-wedge cluster randomised QIS. ICUs are combined into 12 clusters which are randomised in 12 steps. After preparation (18 months) and a control phase of 8 months for all clusters, the first cluster enters a roll-in phase (3 months) that is followed by the actual QIS phase. The remaining clusters follow in month wise steps. The coprimary key performance indicators (KPIs) consist of the ARDS diagnostic rate and guideline adherence regarding lung-protective ventilation. Secondary KPIs include the prevalence of organ dysfunction within 28 days after diagnosis or ICU discharge, the treatment duration on ICU and the hospital mortality. Furthermore, the user acceptance and usability of new technologies in medicine are examined. To show improvements in healthcare of patients with ARDS, differences in primary and secondary KPIs between control phase and QIS will be tested. ETHICS AND DISSEMINATION: Ethical approval was obtained from the independent Ethics Committee (EC) at the RWTH Aachen Faculty of Medicine (local EC reference number: EK 102/19) and the respective data protection officer in March 2019. The results of the ASIC QIS will be presented at conferences and published in peer-reviewed journals. TRIAL REGISTRATION NUMBER: DRKS00014330.

Multiscale classification of heart failure phenotypes by unsupervised clustering of unstructured electronic medical record data.

As a leading cause of death and morbidity, heart failure (HF) is responsible for a large portion of healthcare and disability costs worldwide. Current approaches to define specific HF subpopulations may fail to account for the diversity of etiologies, comorbidities, and factors driving disease progression, and therefore have limited value for clinical decision making and development of novel therapies. Here we present a novel and data-driven approach to understand and characterize the real-world manifestation of HF by clustering disease and symptom-related clinical concepts (complaints) captured from unstructured electronic health record clinical notes. We used natural language processing to construct vectorized representations of patient complaints followed by clustering to group HF patients by similarity of complaint vectors. We then identified complaints that were significantly enriched within each cluster using statistical testing. Breaking the HF population into groups of similar patients revealed a clinically interpretable hierarchy of subgroups characterized by similar HF manifestation. Importantly, our methodology revealed well-known etiologies, risk factors, and comorbid conditions of HF (including ischemic heart disease, aortic valve disease, atrial fibrillation, congenital heart disease, various cardiomyopathies, obesity, hypertension, diabetes, and chronic kidney disease) and yielded additional insights into the details of each HF subgroup's clinical manifestation of HF. Our approach is entirely hypothesis free and can therefore be readily applied for discovery of novel insights in alternative diseases or patient populations.

Engineering advanced neural tissue constructs to mitigate acute cerebral inflammation after brain transplantation in rats.

Stroke, traumatic brain injuries, and other similar conditions often lead to significant loss of functional brain tissue and associated disruption of neuronal signaling. A common strategy for replacing lost neurons is the injection of dissociated neural stem cells or differentiated neurons. However, this method is unlikely to be suitable for replacing large brain cavities, and the resulting distribution of neurons may lack the necessary architecture to support appropriate brain function. Engineered neural tissues may be a viable alternative. Cell death is a prominent concern in neuronal grafting studies, a problem that could be magnified with the transplantation of engineered neural tissues. Here, we examined the effect of one contributor to cell death, acute cerebral inflammation, on neuronal survival after the transplantation of bioengineered constructs based on silk scaffolds. We found evidence of a high degree of inflammation and poor neuronal survival after introducing engineered constructs into the motor cortex of rats. Integrating a corticosteroid (methylprednisolone) into the constructs resulted in significantly improved neuron survival during the acute phase of inflammation. The improved construct survival was associated with decreased markers of inflammation and an anti-inflammatory state of the immune system due to the steroid treatment.

A stochastic model for immunotherapy of cancer.

We propose an extension of a standard stochastic individual-based model in population dynamics which broadens the range of biological applications. Our primary motivation is modelling of immunotherapy of malignant tumours. In this context the different actors, T-cells, cytokines or cancer cells, are modelled as single particles (individuals) in the stochastic system. The main expansions of the model are distinguishing cancer cells by phenotype and genotype, including environment-dependent phenotypic plasticity that does not affect the genotype, taking into account the effects of therapy and introducing a competition term which lowers the reproduction rate of an individual in addition to the usual term that increases its death rate. We illustrate the new setup by using it to model various phenomena arising in immunotherapy. Our aim is twofold: on the one hand, we show that the interplay of genetic mutations and phenotypic switches on different timescales as well as the occurrence of metastability phenomena raise new mathematical challenges. On the other hand, we argue why understanding purely stochastic events (which cannot be obtained with deterministic models) may help to understand the resistance of tumours to therapeutic approaches and may have non-trivial consequences on tumour treatment protocols. This is supported through numerical simulations.

The absence of mrp4 has no effect on the recruitment of neutrophils and eosinophils into the lung after LPS, cigarette smoke or allergen challenge.

The multidrug resistance protein 4 (Mrp4) is an ATP-binding cassette transporter that is capable of exporting the second messenger cAMP from cells, a process that might regulate cAMP-mediated anti-inflammatory processes. However, using LPS- or cigarette smoke (CS)-inflammation models, we found that neutrophil numbers in the bronchoalveolar lavage fluid (BALF) were similar in Mrp4(-/-) and Mrp4(+/+) mice treated with LPS or CS. Similarly, neutrophil numbers were not reduced in the BALF of LPS-challenged wt mice after treatment with 10 or 30 mg/kg of the Mrp1/4 inhibitor MK571. The absence of Mrp4 also had no impact on the influx of eosinophils or IL-4 and IL-5 levels in the BALF after OVA airway challenge in mice sensitized with OVA/alum. LPS-induced cytokine release in whole blood ex vivo was also not affected by the absence of Mrp4. These data clearly suggest that Mrp4 deficiency alone is not sufficient to reduce inflammatory processes in vivo. We hypothesized that in combination with PDE4 inhibitors, used at suboptimal concentrations, the anti-inflammatory effect would be more pronounced. However, LPS-induced neutrophil recruitment into the lung was no different between Mrp4(-/-) and Mrp4(+/+) mice treated with 3 mg/kg Roflumilast. Finally, the single and combined administration of 10 and 30 mg/kg MK571 and the specific breast cancer resistance protein (BCRP) inhibitor KO143 showed no reduction of LPS-induced TNFα release into the BALF compared to vehicle treated control animals. Similarly, LPS-induced TNFα release in murine whole blood of Mrp4(+/+) or Mrp4(-/-) mice was not reduced by KO143 (1, 10 µM). Thus, BCRP seems not to be able to compensate for the absence or inhibition of Mrp4 in the used models. Taken together, our data suggest that Mrp4 is not essential for the recruitment of neutrophils into the lung after LPS or CS exposure or of eosinophils after allergen exposure.