MultiGML: Multimodal graph machine learning for prediction of adverse drug events.

Adverse drug events constitute a major challenge for the success of clinical trials. Several computational strategies have been suggested to estimate the risk of adverse drug events in preclinical drug development. While these approaches have demonstrated high utility in practice, they are at the same time limited to specific information sources. Thus, many current computational approaches neglect a wealth of information which results from the integration of different data sources, such as biological protein function, gene expression, chemical compound structure, cell-based imaging and others. In this work we propose an integrative and explainable multi-modal Graph Machine Learning approach (MultiGML), which fuses knowledge graphs with multiple further data modalities to predict drug related adverse events and general drug target-phenotype associations. MultiGML demonstrates excellent prediction performance compared to alternative algorithms, including various traditional knowledge graph embedding techniques. MultiGML distinguishes itself from alternative techniques by providing in-depth explanations of model predictions, which point towards biological mechanisms associated with predictions of an adverse drug event. Hence, MultiGML could be a versatile tool to support decision making in preclinical drug development.

Artificial intelligence-based clustering and characterization of Parkinson's disease trajectories.

Parkinson's disease (PD) is a highly heterogeneous disease both with respect to arising symptoms and its progression over time. This hampers the design of disease modifying trials for PD as treatments which would potentially show efficacy in specific patient subgroups could be considered ineffective in a heterogeneous trial cohort. Establishing clusters of PD patients based on their progression patterns could help to disentangle the exhibited heterogeneity, highlight clinical differences among patient subgroups, and identify the biological pathways and molecular players which underlie the evident differences. Further, stratification of patients into clusters with distinct progression patterns could help to recruit more homogeneous trial cohorts. In the present work, we applied an artificial intelligence-based algorithm to model and cluster longitudinal PD progression trajectories from the Parkinson's Progression Markers Initiative. Using a combination of six clinical outcome scores covering both motor and non-motor symptoms, we were able to identify specific clusters of PD that showed significantly different patterns of PD progression. The inclusion of genetic variants and biomarker data allowed us to associate the established progression clusters with distinct biological mechanisms, such as perturbations in vesicle transport or neuroprotection. Furthermore, we found that patients of identified progression clusters showed significant differences in their responsiveness to symptomatic treatment. Taken together, our work contributes to a better understanding of the heterogeneity encountered when examining and treating patients with PD, and points towards potential biological pathways and genes that could underlie those differences.

A Sequential Targeting Strategy Interrupts AKT-Driven Subclone-Mediated Progression in Glioblastoma.

PURPOSE: Therapy resistance and fatal disease progression in glioblastoma are thought to result from the dynamics of intra-tumor heterogeneity. This study aimed at identifying and molecularly targeting tumor cells that can survive, adapt, and subclonally expand under primary therapy. EXPERIMENTAL DESIGN: To identify candidate markers and to experimentally access dynamics of subclonal progression in glioblastoma, we established a discovery cohort of paired vital cell samples obtained before and after primary therapy. We further used two independent validation cohorts of paired clinical tissues to test our findings. Follow-up preclinical treatment strategies were evaluated in patient-derived xenografts. RESULTS: We describe, in clinical samples, an archetype of rare ALDH1A1+ tumor cells that enrich and acquire AKT-mediated drug resistance in response to standard-of-care temozolomide (TMZ). Importantly, we observe that drug resistance of ALDH1A1+ cells is not intrinsic, but rather an adaptive mechanism emerging exclusively after TMZ treatment. In patient cells and xenograft models of disease, we recapitulate the enrichment of ALDH1A1+ cells under the influence of TMZ. We demonstrate that their subclonal progression is AKT-driven and can be interfered with by well-timed sequential rather than simultaneous antitumor combination strategy. CONCLUSIONS: Drug-resistant ALDH1A1+/pAKT+ subclones accumulate in patient tissues upon adaptation to TMZ therapy. These subclones may therefore represent a dynamic target in glioblastoma. Our study proposes the combination of TMZ and AKT inhibitors in a sequential treatment schedule as a rationale for future clinical investigation.

A Large Hydatid Cyst in the Brain of a 10-year Child.

Hydatid cyst is the larval form of the parasite, echinococcus granulosus. We operated upon a case of a giant hydatid cyst in the left cerebral hemisphere of a 10-year male child. The patient presented to us with a history of headache, vomiting, vertigo and difficulty in walking. On the examination, there was hemiparesis on the right side and left-sided papilledema. The CT scan showed a large extra-axial cystic lesion in the left frontotemporoparietal area. Craniotomy and excision of the cyst by hydro-dissection was performed. The patient recovered uneventfully and was discharged. Albendazole was given postoperatively for a period of one month. The follow-up CT scan, performed after three months, showed complete resolution of the disease. Key Words: Hydatid cyst, Echinococcus granulosus, Brain, Children.

CADA: phenotype-driven gene prioritization based on a case-enriched knowledge graph.

Many rare syndromes can be well described and delineated from other disorders by a combination of characteristic symptoms. These phenotypic features are best documented with terms of the Human Phenotype Ontology (HPO), which are increasingly used in electronic health records (EHRs), too. Many algorithms that perform HPO-based gene prioritization have also been developed; however, the performance of many such tools suffers from an over-representation of atypical cases in the medical literature. This is certainly the case if the algorithm cannot handle features that occur with reduced frequency in a disorder. With Cada, we built a knowledge graph based on both case annotations and disorder annotations. Using network representation learning, we achieve gene prioritization by link prediction. Our results suggest that Cada exhibits superior performance particularly for patients that present with the pathognomonic findings of a disease. Additionally, information about the frequency of occurrence of a feature can readily be incorporated, when available. Crucial in the design of our approach is the use of the growing amount of phenotype-genotype information that diagnostic labs deposit in databases such as ClinVar. By this means, Cada is an ideal reference tool for differential diagnostics in rare disorders that can also be updated regularly.

Comprehensive Profiling of Blood Coagulation and Fibrinolysis Marker Reveals Elevated Plasmin-Antiplasmin Complexes in Parkinson's Disease.

Parkinson's disease (PD) is the second most common age-related neurodegenerative disease. Accumulating evidence demonstrates that alpha-synuclein (α-Syn), an apparently predominant neuronal protein, is a major contributor to PD pathology. As α-Syn is also highly abundant in blood, particularly in red blood cells (RBCs) and platelets, this in turn raises the question on the function of presumably dysfunctional α-Syn in "peripheral" cells and its putative effect on the other enclosed constituents. Herein, we detected the internal variance in erythrocytes of PD patients by Raman spectroscopy, but no measurable amount of erythrocytic behavioural change (eryptosis) or any haemoglobin variation was noticed. An elevated level of plasmin-antiplasmin complexes (PAP) was observed in the plasma of PD patients, indicating activation of the fibrinolytic system, but platelet activation after thrombin stimulation was not altered. Sex-specific patterns were noticed for blood coagulation factor XIII and factor XII activity in PD patients. Additionally, the alterations in homocysteine levels which have often been observed in PD patients were found to be independent from L-DOPA usage and PAP levels. Furthermore, a selective gene expression analysis identified subsets of genes related to different blood-associated compartments (RBCs, platelets, coagulation-fibrinolysis) also involved in PD-related pathways.

Drug repositioning of antiretroviral ritonavir for combinatorial therapy in glioblastoma.

BACKGROUND: The protease inhibitor ritonavir (RTV) is a clinical-stage inhibitor of the human immunodeficiency virus. In a drug repositioning approach, we here exhibit the additional potential of RTV to augment current treatment of glioblastoma, the most aggressive primary brain tumour of adulthood. METHODS: We explored the antitumour activity of RTV and mechanisms of action in a broad spectrum of short-term expanded clinical cell samples from primary and recurrent glioblastoma and in a cohort of conventional cell lines and non-tumour human neural controls in vitro. To validate RTV efficacy in monotherapeutic and in combinatorial settings, we used patient-derived xenograft models in a series of in vivo studies. RESULTS: RTV monotherapy induced a selective antineoplastic response and demonstrated cytostatic and anti-migratory activity at clinical plasma peak levels. Additional exposure to temozolomide or irradiation further enhanced the effects synergistically, fostered by mechanisms of autophagy and increased endoplasmic reticulum stress. In xenograft models, we consequently observed increasing overall survival under the combinatorial effect of RTV and temozolomide. CONCLUSIONS: Our data establish RTV as a valuable repositioning candidate for further exploration as an adjunct therapeutic in the clinical care of glioblastoma.

Optimal multiparametric set-up modelled for best survival outcomes in palliative treatment of liver malignancies: unsupervised machine learning and 3 PM recommendations.

Over the last decade, a rapid rise in deaths due to liver disease has been observed especially amongst young people. Nowadays liver disease accounts for approximately 2 million deaths per year worldwide: 1 million due to complications of cirrhosis and 1 million due to viral hepatitis and hepatocellular carcinoma. Besides primary liver malignancies, almost all solid tumours are capable to spread metastases to the liver, in particular, gastrointestinal cancers, breast and genitourinary cancers, lung cancer, melanomas and sarcomas. A big portion of liver malignancies undergo palliative care. To this end, the paradigm of the palliative care in the liver cancer management is evolving from "just end of the life" care to careful evaluation of all aspects relevant for the survivorship. In the presented study, an evidence-based approach has been taken to target molecular pathways and subcellular components for modelling most optimal conditions with the longest survival rates for patients diagnosed with advanced liver malignancies who underwent palliative treatments. We developed an unsupervised machine learning (UML) approach to robustly identify patient subgroups based on estimated survival curves for each individual patient and each individual potential biomarker. UML using consensus hierarchical clustering of biomarker derived risk profiles resulted into 3 stable patient subgroups. There were no significant differences in age, gender, therapy, diagnosis or comorbidities across clusters. Survival times across clusters differed significantly. Furthermore, several of the biomarkers demonstrated highly significant pairwise differences between clusters after correction for multiple testing, namely, "comet assay" patterns of classes I, III, IV and expression rates of calgranulin A (S100), SOD2 and profilin-all measured ex vivo in circulating leucocytes. Considering worst, intermediate and best survival curves with regard to identified clusters and corresponding patterns of parameters measured, clear differences were found for "comet assay" and S100 expression patterns. In conclusion, multi-faceted cancer control within the palliative care of liver malignancies is crucial for improved disease outcomes including individualised patient profiling, predictive models and implementation of corresponding cost-effective risks mitigating measures detailed in the paper. The "proof-of-principle" model is presented.

Deep learning for clustering of multivariate clinical patient trajectories with missing values.

BACKGROUND: Precision medicine requires a stratification of patients by disease presentation that is sufficiently informative to allow for selecting treatments on a per-patient basis. For many diseases, such as neurological disorders, this stratification problem translates into a complex problem of clustering multivariate and relatively short time series because (i) these diseases are multifactorial and not well described by single clinical outcome variables and (ii) disease progression needs to be monitored over time. Additionally, clinical data often additionally are hindered by the presence of many missing values, further complicating any clustering attempts. FINDINGS: The problem of clustering multivariate short time series with many missing values is generally not well addressed in the literature. In this work, we propose a deep learning-based method to address this issue, variational deep embedding with recurrence (VaDER). VaDER relies on a Gaussian mixture variational autoencoder framework, which is further extended to (i) model multivariate time series and (ii) directly deal with missing values. We validated VaDER by accurately recovering clusters from simulated and benchmark data with known ground truth clustering, while varying the degree of missingness. We then used VaDER to successfully stratify patients with Alzheimer disease and patients with Parkinson disease into subgroups characterized by clinically divergent disease progression profiles. Additional analyses demonstrated that these clinical differences reflected known underlying aspects of Alzheimer disease and Parkinson disease. CONCLUSIONS: We believe our results show that VaDER can be of great value for future efforts in patient stratification, and multivariate time-series clustering in general.

Implications of Isoprostanes and Matrix Metalloproteinase-7 Having Potential Role in the Development of Colorectal Cancer in Males.

BACKGROUND: Colorectal cancer (CRC) is the third most common type of cancer and leading cause of death worldwide. Major risk factors involved in the development of CRC are increased dietary sources, genetics, and increasing age. Purpose of the study was to find the role of different variables in the progression of CRC. METHODOLOGY: 50 blood samples from CRC patients and 20 samples from control were collected. Serum was separated from the blood by centrifugation. This serum was assessed for several antioxidants like superoxide dismutase (SOD), glutathione, glutathione peroxidase, glutathione reductase, catalase, vitamin A, C, and E, and pro-oxidants such as malondialdehyde, advanced oxidation protein products (AOPPs), and AGEs according to their respective protocols. Matrix metalloproteinase-7 (MMP-7) and isoprostanes were assessed by ELISA kits. RESULTS: Lower levels of GSH (4.86 ± 0.78 vs 9.65 ± 1.13 µg/dl), SOD (0.08 ± 0.012 vs 0.46 ± 0.017 µg/dl), CAT (2.45 ± 0.03 vs 4.22 ± 0.19 µmol/mol of protein), and GRx (5.16 ± 0.06 vs 7.23 ± 0.36 µmol/ml) in the diseased group were recorded as compared with control. Higher levels of GPx (6.64 ± 0.19 mmol/dl) were observed in the subjects in comparison with control group (1.58 ± 0.30 mmol/dl). Highly significant decreased levels of vitamin A (0.81 ± 0.07 vs 2.37 ± 0.15 mg/ml), vitamin E (15.42 ± 1.26 vs 25.96 ± 2.19 mg/ml), and vitamin C (47.67 ± 7.69 vs 80.37 ± 10.21 mg/ml) were observed in the patients in contrast to control group. The reversal of antioxidants in later stages of CRC may be due to compensatory mechanisms in cancerous cells. The levels of MDA (nmol/ml) were also assessed, which shows significantly increased level in CRC patients as compared with control groups (3.67 ± 0.19 vs 1.31 ± 0.27). The levels of protein oxidation products [AGEs (2.74 ± 0.16 vs 0.84 ± 0.05 IU) and AOPPs (1.32 ± 0.02 vs 0.82 ± 0.07 ng/ml)] were significantly increased in subjects as compared with control. The levels of MMP-7 (64.75 ± 3.03 vs 50.61 ± 4.09 ng/ml) and isoprostanes (0.71 ± 0.03 vs 0.16 ± 0.02 ng/ml) were also analyzed. This shows that the levels of isoprostanes increased due to high lipid peroxidation mediate higher levels of MMP-7, which promotes development of CRC. CONCLUSION: Following study suggested that elevated oxidative and inflammatory status along with lipid peroxidation and matrix metalloproteinases are the chief contributors in the progression of CRC.

Towards clinically more relevant dissection of patient heterogeneity via survival-based Bayesian clustering.

MOTIVATION: Discovery of clinically relevant disease sub-types is of prime importance in personalized medicine. Disease sub-type identification has in the past often been explored in an unsupervised machine learning paradigm which involves clustering of patients based on available-omics data, such as gene expression. A follow-up analysis involves determining the clinical relevance of the molecular sub-types such as that reflected by comparing their disease progressions. The above methodology, however, fails to guarantee the separability of the sub-types based on their subtype-specific survival curves. RESULTS: We propose a new algorithm, Survival-based Bayesian Clustering (SBC) which simultaneously clusters heterogeneous-omics and clinical end point data (time to event) in order to discover clinically relevant disease subtypes. For this purpose we formulate a novel Hierarchical Bayesian Graphical Model which combines a Dirichlet Process Gaussian Mixture Model with an Accelerated Failure Time model. In this way we make sure that patients are grouped in the same cluster only when they show similar characteristics with respect to molecular features across data types (e.g. gene expression, mi-RNA) as well as survival times. We extensively test our model in simulation studies and apply it to cancer patient data from the Breast Cancer dataset and The Cancer Genome Atlas repository. Notably, our method is not only able to find clinically relevant sub-groups, but is also able to predict cluster membership and survival on test data in a better way than other competing methods. AVAILABILITY AND IMPLEMENTATION: Our R-code can be accessed as https://github.com/ashar799/SBC. CONTACT: ashar@bit.uni-bonn.de. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Skewed X-chromosome inactivation and XIST locus methylation levels do not contribute to the lower prevalence of Parkinson's disease in females.

Parkinson's disease (PD) is a degenerative disorder of the nervous system and the cause of the majority of sporadic cases is unknown. Females are relatively protected from PD as compared with males and linkage studies suggested a PD susceptibility locus on the X chromosome. To determine a putative association of skewed X-chromosome inactivation (XCI) and PD, we examined XCI patterns using a human androgen receptor gene-based assay (HUMARA) and did not identify any association of skewed or random X inactivation with clinical heterogeneity among female PD patients. In addition, we sought to determine methylation-specific changes at the X-inactive specific transcript (XIST) locus, which is known to be responsible for initiating X inactivation. We observed a trend towards hypomethylation in the gene body region of the XIST locus in PD females which did not reach significance. Furthermore, we extended our analysis of DNA methylation across the entire X-chromosome which revealed no methylation-specific differences between PD females and healthy controls. Thus, we propose that skewed XCI and methylation levels on the entire X chromosome did not reveal changes which could account for the decreased PD susceptibility in females or suitable to use as a biomarker.

Expectation propagation for large scale Bayesian inference of non-linear molecular networks from perturbation data.

Inferring the structure of molecular networks from time series protein or gene expression data provides valuable information about the complex biological processes of the cell. Causal network structure inference has been approached using different methods in the past. Most causal network inference techniques, such as Dynamic Bayesian Networks and ordinary differential equations, are limited by their computational complexity and thus make large scale inference infeasible. This is specifically true if a Bayesian framework is applied in order to deal with the unavoidable uncertainty about the correct model. We devise a novel Bayesian network reverse engineering approach using ordinary differential equations with the ability to include non-linearity. Besides modeling arbitrary, possibly combinatorial and time dependent perturbations with unknown targets, one of our main contributions is the use of Expectation Propagation, an algorithm for approximate Bayesian inference over large scale network structures in short computation time. We further explore the possibility of integrating prior knowledge into network inference. We evaluate the proposed model on DREAM4 and DREAM8 data and find it competitive against several state-of-the-art existing network inference methods.

L-dopa increases α-synuclein DNA methylation in Parkinson's disease patients in vivo and in vitro.

BACKGROUND: Increasing gene dosages of α-synuclein induce familial Parkinson's disease (PD); thus, the hypothesis has been put forward that regulation of gene expression, in particular altered α-synuclein gene methylation, might be associated with sporadic PD and could be used as a biological marker. METHODS: We performed a thorough analysis of α-synuclein methylation in bisulfite-treated DNA from peripheral blood of 490 sporadic PD patients and 485 healthy controls and in addition analyzed the effect of levodopa (L-dopa) on α-synuclein methylation and expression in cultured mononuclear cells. RESULTS: α-Synuclein was hypomethylated in sporadic PD patients, correlated with sex, age, and a polymorphism in the analyzed sequence stretch (rs3756063). α-Synuclein methylation separated healthy individuals from sporadic PD with a specificity of 74% (male) and 78% (female), respectively. α-Synuclein methylation was increased in sporadic PD patients with higher l-dopa dosage, and L-dopa specifically induced methylation of α-synuclein intron 1 in cultured mononuclear cells. CONCLUSIONS: α-Synuclein methylation levels depended on disease status, sex, age, and the genotype of rs3756063. The pharmacological action of L-dopa was not limited to the dopamine precursor function but included epigenetic off-target effects. The hypomethylation of α-synuclein in sporadic PD patients' blood already observed in previous studies was probably underestimated because of effect of L-dopa, which was not known previously. The analysis of α-synuclein methylation can help to identify nonparkinsonian individuals with reasonable specificity, which offers a valuable tool for researchers and clinicians.