A deep profile of gene expression across 18 human cancers.

Clinically and biologically valuable information may reside untapped in large cancer gene expression data sets. Deep unsupervised learning has the potential to extract this information with unprecedented efficacy but has thus far been hampered by a lack of biological interpretability and robustness. Here, we present DeepProfile, a comprehensive framework that addresses current challenges in applying unsupervised deep learning to gene expression profiles. We use DeepProfile to learn low-dimensional latent spaces for 18 human cancers from 50,211 transcriptomes. DeepProfile outperforms existing dimensionality reduction methods with respect to biological interpretability. Using DeepProfile interpretability methods, we show that genes that are universally important in defining the latent spaces across all cancer types control immune cell activation, while cancer type-specific genes and pathways define molecular disease subtypes. By linking DeepProfile latent variables to secondary tumor characteristics, we discover that tumor mutation burden is closely associated with the expression of cell cycle-related genes. DNA mismatch repair and MHC class II antigen presentation pathway expression, on the other hand, are consistently associated with patient survival. We validate these results through Kaplan-Meier analyses and nominate tumor-associated macrophages as an important source of survival-correlated MHC class II transcripts. Our results illustrate the power of unsupervised deep learning for discovery of novel cancer biology from existing gene expression data.

Increased epicardial adipose tissue thickness as a sign of subclinical atherosclerosis in patients with rheumatoid arthritis and its relationship with disease activity indices.

Epicardial adipose tissue is a novel cardiometabolic risk factor and indicator of subclinical atherosclerosis. We aimed to evaluate the epicardial adipose tissue thickness in rheumatoid arthritis (RA) patients and its association with disease activity scores. A total of 81 rheumatoid arthritis patients and 70 age- and sex-matched healthy individuals were recruited for this cross-sectional study. Epicardial adipose tissue thickness (EATT) was measured by transthoracic two-dimensional echocardiography. Tender and swollen joint counts were recorded at the time of inclusion. The laboratory tests included erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), rheumatoid factor, anti-citrullinated protein antibodies, and serum lipid levels. Disease activity was calculated based on Disease Activity Scores for 28 joints (DAS-28) ESR and CRP, the Simple Disease Activity Index (SDAI), and the Clinical Disease Activity Index (CDAI). Epicardial adipose tissue thickness was significantly higher in the RA patients compared to the healthy controls (p < 0.001). We found statistically significant correlations of EATT with all disease activity indices (p < 0.001) and CRP (p = 0.002). According to a cut-off value of 6.4 mm determined for epicardial adipose tissue thickness, the RA patients with thickness ≥ 6.4 mm had higher disease activity scores and CRP levels. In the multivariable regression analysis, only SDAI score was found as an independent risk factor for increased EATT (OR, (95%CI), 13.70 (3.88-48.43), p < 0.001). Epicardial adipose tissue thickness measurement by echocardiography is a reliable method for assessing subclinical atherosclerosis in rheumatoid arthritis patients, and a higher disease activity score is an independent risk factor for coronary artery disease.

The ABCs of antiphospholipid syndrome.

Antiphospholipid syndrome (APS) is a thromboinflammatory syndrome characterized by thrombotic, microvascular, obstetric, or non-thrombotic events in the setting of persistent antiphospholipid antibodies (aPL), namely anticardiolipin antibody (aCL), anti-ß2 glycoprotein-I antibody (aß2GPI), and lupus anticoagulant (LA). The diagnosis of APS requires careful assessment of the aPL profile, the clinical phenotype, and additional risk factors. The standard management of aPL-related thrombosis is anticoagulation, which is not effective for microvascular and non-thrombotic events. In parallel to our improved understanding of aPL-related mechanisms, the role of immunosuppression has been increasingly investigated. In this review, we summarize the basic concepts and future perspectives in APS.

The factors predicting development of serious infections in ANCA-associated vasculitis.

BACKGROUND: Anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) is a rare autoimmune disease usually involving small vessels and progressing with necrotizing inflammation. Treatment requires long-term use of immunosuppressive agents to inhibit disease activity. Serious infections (SIs) are a common complication in AAV. OBJECTIVE: The aim of this study was to identify the risk factors for serious infections which required hospitalization in patients with AAV. METHODS: In this retrospective cohort study., we included 84 patients admitted to the Ankara University Faculty of Medicine in the last 10 years with a diagnosis of AAV. RESULTS: In 42 (50%) of 84 patients followed up with the diagnosis of AAV, an infection requiring hospitalization was identified. The patients' total corticosteroid dose, use of pulse steroids, induction regimen, levels of C-reactive protein (CRP) and the presence of pulmonary and renopulmonary involvement were found to be associated with the frequency of infection (p=0.015, p=0.016, p=0.010, p=0.03, p= 0.026 and p=0.029, respectively). In multivariable analysis, it was found that renopulmonary involvement (p=0.002, HR=4.95, 95% CI= 1.804-13.605), age of over 65 (p=0.049, HR=3.37, 95% CI=1.004-11.369) and high CRP levels (p=0.043, HR=1.006, 95% CI=1.000-1.011) constituted independent predictors of serious infection risk. CONCLUSION: The frequency of infection is known to be increased in ANCA-associated vasculitis. Our study showed that renopulmonary involvement, age and elevated CRP levels on admission are independent risk factors of infection.

Uncovering expression signatures of synergistic drug responses via ensembles of explainable machine-learning models.

Machine learning may aid the choice of optimal combinations of anticancer drugs by explaining the molecular basis of their synergy. By combining accurate models with interpretable insights, explainable machine learning promises to accelerate data-driven cancer pharmacology. However, owing to the highly correlated and high-dimensional nature of transcriptomic data, naively applying current explainable machine-learning strategies to large transcriptomic datasets leads to suboptimal outcomes. Here by using feature attribution methods, we show that the quality of the explanations can be increased by leveraging ensembles of explainable machine-learning models. We applied the approach to a dataset of 133 combinations of 46 anticancer drugs tested in ex vivo tumour samples from 285 patients with acute myeloid leukaemia and uncovered a haematopoietic-differentiation signature underlying drug combinations with therapeutic synergy. Ensembles of machine-learning models trained to predict drug combination synergies on the basis of gene-expression data may improve the feature attribution quality of complex machine-learning models.

Frequency of Helicobacter pylori in Patients With Rheumatoid Arthritis Whose Methotrexate Was Stopped Due to Gastrointestinal Intolerance.

OBJECTIVE: The aims of this study were to compare the frequency of Helicobacter pylori between patients with rheumatoid arthritis (RA) with and without methotrexate (MTX)-related gastrointestinal system (GIS) intolerance, and to demonstrate the associated factors with such intolerance. METHODS: The data of 9756 patients with RA who presented between January 2011 and December 2020 were evaluated. Methotrexate-related GIS intolerance was defined as the discontinuation of MTX owing to the dyspeptic symptoms despite supportive measures and was detected in 1742 (31.3%) patients among 5572 MTX users. A total of 390 patients with and without intolerance who had at least 1 gastroscopic evaluation were included in the final analyses. The demographic, clinical, laboratory, and pathologic characteristics of patients with and without MTX-related GIS intolerance were compared. To determine the associated factors with MTX-related GIS intolerance, logistic regression analysis was performed. RESULTS: Of 390 patients, 160 (41.0%) patients had MTX-related GIS intolerance. According to the pathology results, the presence of H. pylori , inflammation, and activity were significantly higher in patients with MTX-related GIS intolerance ( p < 0.001 for each comparison). In multivariable logistic regression analysis, the use of biologic disease-modifying antirheumatic drugs (DMARDs) or targeted synthetic DMARDs was found to be an independently associated factor for MTX-related GIS intolerance (odds ratio [OR], 3.03 for model 1; OR, 3.02 for model 2) in addition to H. pylori presence (OR, 9.13 for model 1; OR, 5.71 for model 2). CONCLUSIONS: In this study, we found that the presence of H. pylori and the use of biologic or targeted synthetic DMARDs were associated with MTX-related GIS intolerance.

Interpretable machine learning prediction of all-cause mortality.

Background: Unlike linear models which are traditionally used to study all-cause mortality, complex machine learning models can capture non-linear interrelations and provide opportunities to identify unexplored risk factors. Explainable artificial intelligence can improve prediction accuracy over linear models and reveal great insights into outcomes like mortality. This paper comprehensively analyzes all-cause mortality by explaining complex machine learning models. Methods: We propose the IMPACT framework that uses XAI technique to explain a state-of-the-art tree ensemble mortality prediction model. We apply IMPACT to understand all-cause mortality for 1-, 3-, 5-, and 10-year follow-up times within the NHANES dataset, which contains 47,261 samples and 151 features. Results: We show that IMPACT models achieve higher accuracy than linear models and neural networks. Using IMPACT, we identify several overlooked risk factors and interaction effects. Furthermore, we identify relationships between laboratory features and mortality that may suggest adjusting established reference intervals. Finally, we develop highly accurate, efficient and interpretable mortality risk scores that can be used by medical professionals and individuals without medical expertise. We ensure generalizability by performing temporal validation of the mortality risk scores and external validation of important findings with the UK Biobank dataset. Conclusions: IMPACT's unique strength is the explainable prediction, which provides insights into the complex, non-linear relationships between mortality and features, while maintaining high accuracy. Our explainable risk scores could help individuals improve self-awareness of their health status and help clinicians identify patients with high risk. IMPACT takes a consequential step towards bringing contemporary developments in XAI to epidemiology.

This study identifies characteristics that will make a person more likely to die sooner than expected based on life expectancy for the population. We developed a computer program and applied it to information obtained about the characteristics and medical history of people from the USA. We identified previously unidentified characteristics that impact how likely it is someone will die sooner than expected, for example the circumference of the arm. We also identified combinations of characteristics that interact to increase the likelihood of death sooner than expected. By adding a person's characteristics to the program, the likelihood of death over the next 5 years can be calculated and characteristics identified that a person could modify to improve their health and reduce their chance of death during this period.

Reducing Peptide Sequence Bias in Quantitative Mass Spectrometry Data with Machine Learning.

Quantitative mass spectrometry measurements of peptides necessarily incorporate sequence-specific biases that reflect the behavior of the peptide during enzymatic digestion and liquid chromatography and in a mass spectrometer. These sequence-specific effects impair quantification accuracy, yielding peptide quantities that are systematically under- or overestimated. We provide empirical evidence for the existence of such biases, and we use a deep neural network, called Pepper, to automatically identify and reduce these biases. The model generalizes to new proteins and new runs within a related set of tandem mass spectrometry experiments, and the learned coefficients themselves reflect expected physicochemical properties of the corresponding peptide sequences. The resulting adjusted abundance measurements are more correlated with mRNA-based gene expression measurements than the unadjusted measurements. Pepper is suitable for data generated on a variety of mass spectrometry instruments and can be used with labeled or label-free approaches and with data-independent or data-dependent acquisition.

Systemic Immune Inflammation Index as a Reliable Disease Activity Marker in Psoriatic Arthritis.

OBJECTIVE: To assess the utility of systemic immune inflammation index (SII) in predicting disease activity in psoriatic arthritis (PsA) patients. STUDY DESIGN: A descriptive study. PLACE AND DURATION OF STUDY: Diskapi Yildirim Beyazit Research and Training Hospital, Ankara, Turkey, from October 2020 to September 2021. METHODOLOGY: This study included 106 PsA and 103 age and gender-matched healthy individuals. Neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), monocyte to lymphocyte ratio (MLR), and SII were calculated from complete blood count parameters. The PsA disease activity was assessed by using disease activity score-ESR and DAS-CRP based on 28 joints and the Disease Activity in Psoriatic Arthritis (DAPSA) scores. The receiver operating characteristic (ROC) curve was performed to evaluate the utility of SII in determining disease activity in PsA patients. RESULTS: The NLR, PLR, MLR, and SII were significantly higher in PsA patients compared to healthy control (p=0.013, p=0.019, p=0.012, and p=0.002, respectively). There were statistically significant positive correlations between the DAS28-ESR, DAS28-CRP, and DAPSA and SII (p<0.001, p<0.001, and p<0.001 respectively). The SII values were significantly higher in PsA patients with moderate to severe disease activity according to DAPSA scores when compared to patients with remission or low disease activity (p<0.001). The cut-off value of 800x109/L was found for predicting disease activity in PsA. CONCLUSION: SII may be an easy, practical, economical, and readily accessible tool for monitoring disease activity and the efficacy of treatment in PsA patients. KEY WORDS: Blood cell count, Psoriatic arthritis, Systematic immune inflammation index (SII).

An automatic integrative method for learning interpretable communities of biological pathways.

Although knowledge of biological pathways is essential for interpreting results from computational biology studies, the growing number of pathway databases complicates efforts to efficiently perform pathway analysis due to high redundancies among pathways from different databases, and inconsistencies in how pathways are created and named. We introduce the PAthway Communities (PAC) framework, which reconciles pathways from different databases and reduces pathway redundancy by revealing informative groups with distinct biological functions. Uniquely applying the Louvain community detection algorithm to a network of 4847 pathways from KEGG, REACTOME and Gene Ontology databases, we identify 35 distinct and automatically annotated communities of pathways and show that they are consistent with expert-curated pathway categories. Further, we demonstrate that our pathway community network can be queried with new gene sets to provide biological context in terms of related pathways and communities. Our approach, combined with an interpretable web tool we provide, will help computational biologists more efficiently contextualize and interpret their biological findings.

Indications and risk factors for hospitalization in patients with primary Sjögren syndrome: experience from a tertiary center in Turkey.

OBJECTIVE: In this study, it was aimed to reveal the hospitalization reasons for patients diagnosed with primary Sjögren syndrome (pSS) and potentially associated factors in a tertiary health center. METHOD: One hundred and sixty-three pSS patients who regularly attended their follow-ups between January 2010 and May 2021 were included in the study. These patients' reasons for hospitalization, duration of hospitalization, and numbers of presenting to the hospital were recorded. The demographic, clinical and serological characteristics of the hospitalized and non-hospitalized patients were compared. RESULTS: Hospitalization occurred in 22.7% of the patients, and the total number of hospitalizations was 79. The hospitalization incidence density rate was 6.21 per 100 patient-years. The most frequently encountered reason for hospitalizations was pSS-related organ involvement (44.3%). Infections (17.7%), malignancy (16.5%), endocrine, and various other reasons were the other indications for hospitalization. While male sex (p = 0.005), the presence of extra-glandular involvement (p < 0.001), and interstitial lung disease (p = 0.001) were more common in the hospitalized patients, anti-nuclear antibody positivity was less frequent (p = 0.032). The usage rate of hydroxychloroquine (p = 0.022) was lower in the hospitalized patients, whereas the use of glucocorticoids (p < 0.001) and azathioprine (p = 0.005) was more frequent. The multivariable analyses revealed a relationship between extra-glandular involvement (OR: 4.57 [1.05-19.84], p = 0.043), glucocorticoid use (OR: 3.23 [1.13-9.21], p = 0.028) and hospitalization. CONCLUSION: pSS-related system involvement and infection accounted for the majority of hospitalizations of the pSS patients. The presence of extra-glandular involvement and glucocorticoid use were found to be associated with hospitalization. Key Points • pSS-related system involvement and infection accounted for the majority of hospitalizations of pSS patients. • The presence of extra-glandular involvement was found to be associated with hospitalization.

Interstitial Lung Disease in Systemic Sclerosis: A Single-center Retrospective Analysis.

BACKGROUND: Systemic sclerosis (SSc) is a systemic autoimmune disease characterized by microangiopathy, inflammation, fibrosis. Interstitial lung disease (ILD) is common among SSc patients. OBJECTIVE: This study aims to define the clinical, laboratory, and serologic characteristics of SSc patients with ILD and to present the frequency of chest computed tomography features. METHODS: Two hundred twenty-six SSc patients who applied to the Rheumatology Department between January 2007 and August 2019 were retrospectively examined. A total of 100 SSc patients with ILD (44.2%) were determined. Clinical, laboratory, and serological features of SSc patients with and without ILD were compared. RESULT: Both groups had similar characteristics in terms of age and sex. The duration of disease (p=0.001) and follow-up time (p=0.001) were longer in SSc patients with ILD. Multivariable logistic regression analysis indicated that the duration of disease (OR: 1.06 (1.01-1.13), p=0.029), presence of gastrointestinal system involvement (OR: 3.29 (1.28-8.46), p=0.013) and anti-SCL70-positivity (OR: 6.04 (2.35-15.49), p <0.001) were associated with ILD. There was an inverse relationship between Anti-CENP-B positivity and the presence of ILD (p=0.001). The assessment regarding the chest computed tomography characteristics of interstitial pneumonia patterns were as follows: 82.5% non-specific interstitial pneumonia, 14.4% usual interstitial pneumonia, and 2.1% desquamative interstitial pneumonia. The most frequent abnormal findings included ground-glass opacification (88.7%), reticulation (64.9%), traction bronchiectasis (57.7%), septal thickening (52.6%) and honeycombing (28.9%). CONCLUSION: We have shown a relationship between anti-SCL70, disease duration, gastrointestinal system involvement, and ILD in SSc patients.

Causes of severe infections in patients with systemic sclerosis and associated factors.

BACKGROUND: Systemic sclerosis (SSc) is a chronic systemic disease characterized by vascular damage, autoimmunity, and fibrosis in the skin and internal organs. In this study, we tried to determine the causes of severe infection in patients with SSc and to reveal the factors associated with severe infection. METHODS: We retrospectively examined 214 SSc patients between January 2010 and August 2020. Forty-seven patients with at least one severe infection and 167 patients without severe infection were compared. RESULTS: A total of 76 episodes of severe infections were detected in 47 (22%) patients. Common infections included pneumonia, infected digital ulcer, urinary tract infections, and osteomyelitis. Female patients had a higher frequency in the group without severe infection (91.6% vs. 80.9%, p = 0.035). Patients with severe infections had a higher frequency of digital ulcers (p < 0.001), cardiac (p = 0.002), and GIS involvement (p < 0.001). In multivariable analysis, digital ulcer presence (OR: 2.849 [1.356-5.898] (p = 0.006) and cardiac involvement (OR: 2.801 [1.248-6.285]) were associated with severe infection. Of the patients with severe infections, 34% had recurrent severe infections. There was no difference in demographic and clinical characteristics between patients with recurrent and nonrecurrent severe infections. DISCUSSION: The presence of digital ulcer and cardiac involvement seem to be associated with a severe infection in patients with systemic sclerosis. In patients with cardiac involvement and digital ulcers, more careful attention may be required for the development of severe infections.

Disseminated Nocardiosis with Ocular Involvement in a Patient with Anca-associated Vasculitis.

A 62-year male patient, diagnosed with antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV), developed proptosis and decrease in visual acuity while on rituximab treatment. As the ophthalmological examination and imaging studies could not exclude tumour of the orbit, enucleation of the orbit was performed. The histopathology displayed necrosis and inflammation. Because the clinical, laboratory and pathological findings of the patient suggested a vasculitis exacerbation, the immunosuppressive treatment was continued. However, the patient developed confusion and hemiplegia with cerebral mass lesions on imaging. The subsequent report of the pathology revealed a nocardial infection of the eye. The patient was diagnosed with nocardiosis with ocular and cerebral involvement. Despite efficient antimicrobial therapy, the disease progressed rapidly causing death. This case is unique as it describes disseminated nocardiosis with ocular and cerebral involvement in an AVV patient. Key Words: Immunosuppression, Nocardiosis, ANCA-associated vasculitis, Proptosis.

Can Subclinical Atherosclerosis Be Assessed More Precisely in Behçet Syndrome Patients by Using a Particular Cutoff Value for Carotid Intima Media Thickness?

OBJECTIVE: Behçet syndrome (BS) is a multisystemic chronic vasculitic disease. Among previous studies, although there are some that showed increased risk of subclinical atherosclerosis in BS, there are also others that showed the opposite. The objective of this study is to evaluate subclinical atherosclerosis in BS by using the cutoff value for intima-media thickness in the 2013 European Society of Cardiology/European Society of Hypertension guideline. METHODS: We conducted a cross-sectional analysis of 100 BS patients and 30 healthy volunteers at a single center in a 4-month period. All ultrasound scans were performed in a blind manner to the clinical assessment, and they were carried out by the same researcher by a B-mode ultrasonography. RESULT: When we grouped the patients based on the presence of subclinical atherosclerosis, the frequency of subclinical atherosclerosis in the BS patients was found to be higher than that in the healthy controls (32% and 7%, respectively; p = 0.006). When a cutoff is used for carotid intima-media thickness, increased atherosclerosis risk is observed in BS patients with vascular involvement (p = 0.043). CONCLUSIONS: Although higher inflammation and increased atherosclerosis in vascular BS patients were expected, this situation was not supported much in previous studies. We think that this may have been caused by mere comparison of numerical data, and usage of a cutoff value could be more significant in distinguishing what is normal and what is abnormal as in several medical parameters.

Serum Calprotectin is Indicating Clinical and Ultrasonographic Disease Activity in Rheumatoid Arthritis, even with Normal C-Reactive Protein Levels.

OBJECTIVE: Calprotectin is an inflammatory biomarker which assesses disease activity in rheumatoid arthritis (RA). The objective of this study was to test whether serum calprotectin is associated with clinical and ultrasonographic disease activity in patients with RA, and to analyse its predicting value for disease activity evaluation despite normal C-Reactive protein (CRP) levels. METHODS: We included 80 patients with RA and 30 healthy subjects. Patients were examined clinically and by ultrasound, (US7 score) along with laboratory parameters (calprotectin, CRP, erythrocyte sedimentation rate [ESR]). Disease activity scores (DAS28) were calculated to assess disease activity. Firstly, patients were divided into four subgroups according to the DAS28-ESR (high, moderate, low disease activity, and remission), then into two subgroups; group-1 (DAS-28≤3.2) and group-2 (DAS28>3.2). The predicting value of calprotectin for disease activity in patients with normal CRP was analysed with univariate and multivariate analysis and receiver operating characteristic curves. RESULTS: Calprotectin levels were higher in RA patients than controls (96.3±45.9 ng/ml, 54.7±50.0 ng/ml, respectively; p<0.001). Calprotectin levels were 74.8±45.5 ng/ml in group-1 (n=37) and 114.7±37.9 ng/ml in group-2 (n=43) (p<0.001). In univariate analyses, calprotectin was significantly correlated with clinical, laboratory, and ultrasound parameters (p<0.05), and was a better predictor of power doppler synovitis than CRP in multivariate analysis (OR=1.014; 95%CI 1.002-1.027; p=0.024). The discriminatory capacity for calprotectin to distinguish ultrasonographically active disease in patients with normal CRP levels using AUC was 0.75 (95%CI 0.56-0.90, p=0.023). CONCLUSIONS: Calprotectin represents disease activity, even in patients who are clinical and ultrasonographical active but have normal CRP levels.

High mobility group box-1 levels may be associated with disease activity of Behcet's disease

BACKGROUND: High mobility group box- 1 (HMGB- 1) is a nuclear protein acting as a proinflammatory molecule. The serum HMGB- 1 levels were found elevated in chronic inflammatory diseases. In this cross-sectional study, serum HMGB- 1 levels in Behcet's disease (BD) patients and healthy controls (HC) were studied. Also, its association with disease activity scores and clinical findings were evaluated. METHODS: Ninety BD patients and 50 age-sex matched HC were included in the study. Disease activity scores were assessed by Behcet Disease Current Activity Form (BDCAF) and Behcet Syndrome Activity Score (BSAS). Serum HMGB- 1 levels were measured using a commercial ELISA kit. A p value of < 0.05 was considered to be statistically significant. RESULTS: Serum HMGB- 1 levels were significantly higher in BD than in HC (43.26 pg/mL and 16.73 pg/mL; p < 0.001, respectively). Serum HMGB- 1 levels were statistically significantly associated with presence of erythema nodosum (EN) and genital ulcers in the last one month prior to recruitment (p = 0.041 and p < 0.001, respectively). BDCAF and BSAS scores were positively correlated with serum HMGB- 1 level ( p = 0.03 and p = 0.02, respectively). DISCUSSION: HMGB - 1 may play a role in the development of BD. Also, due to its positive correlation with disease activity indices, it can be used as a novel disease activity parameter in BD.

Imaging modalities used in diagnosis and follow-up of patients with Takayasu's arteritis

Background/aim: Takayasu's arteritis (TA) is a rare, large-vessel vasculitis of unknown etiology, affecting aortic arch, and its main branches. Noninvasive imaging methods are frequently used in diagnosis and follow-up in Takayasu's arteritis. Studies investigating optimal timing of follow up imaging are rare. This study is aimed to investigate the radiologic changes in vascular involvements of Takayasu's arteritis patients one year after diagnosis. Materials and methods: Database of our Vasculitis Center was analyzed retrospectively and 97 patients were included into the study. Demographic, clinical, radiological, and therapeutic findings of patients were recorded. Patients with follow-up imaging after approximately one year of diagnosis were recruited into further analysis. Radiological changes and the effect of different immunosuppressive agents on vascular involvements were investigated. Results: Mean age and disease duration of patients were 43.0 and 9.0 years. The most commonly used imaging methods/modalities for the diagnosis of TA were computer tomography-angiography (CT-Ang) (58.8%), magnetic resonance-angiography (MR-Ang) (29.9%), and doppler ultrasonography (11.3%). Subclavian and common carotid arteries were the most frequently involved vessels. Fifty-three patients underwent follow-up imaging after one year of diagnosis and, in 64% of patients, same imaging method had been used. MR- Ang (62.3%) and CT-Ang (35.9%) were the most preferred follow-up imaging studies. Sixty-eight percent of patients had stable vascular involvement, 28% had progression, and 4% had regression. No difference was found in radiological changes regarding patients with usage of different immunosuppressive agents (P = 0.634). There was no association between the change in serum acute phase reactants and radiological disease activity. Conclusion: The most commonly used imaging modality for the diagnosis of TA was CT-Ang, whereas MR-Ang was the most preferred for follow-up. Almost 30% of TA patients in our Vasculitis Center had progression at around one year concordant with previous literature. A follow-up imaging at around one year of treatment seems feasible in management of TA.

Adversarial deconfounding autoencoder for learning robust gene expression embeddings.

MOTIVATION: Increasing number of gene expression profiles has enabled the use of complex models, such as deep unsupervised neural networks, to extract a latent space from these profiles. However, expression profiles, especially when collected in large numbers, inherently contain variations introduced by technical artifacts (e.g. batch effects) and uninteresting biological variables (e.g. age) in addition to the true signals of interest. These sources of variations, called confounders, produce embeddings that fail to transfer to different domains, i.e. an embedding learned from one dataset with a specific confounder distribution does not generalize to different distributions. To remedy this problem, we attempt to disentangle confounders from true signals to generate biologically informative embeddings. RESULTS: In this article, we introduce the Adversarial Deconfounding AutoEncoder (AD-AE) approach to deconfounding gene expression latent spaces. The AD-AE model consists of two neural networks: (i) an autoencoder to generate an embedding that can reconstruct original measurements, and (ii) an adversary trained to predict the confounder from that embedding. We jointly train the networks to generate embeddings that can encode as much information as possible without encoding any confounding signal. By applying AD-AE to two distinct gene expression datasets, we show that our model can (i) generate embeddings that do not encode confounder information, (ii) conserve the biological signals present in the original space and (iii) generalize successfully across different confounder domains. We demonstrate that AD-AE outperforms standard autoencoder and other deconfounding approaches. AVAILABILITY AND IMPLEMENTATION: Our code and data are available at https://gitlab.cs.washington.edu/abdincer/ad-ae. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.