MF-PCBA: Multifidelity High-Throughput Screening Benchmarks for Drug Discovery and Machine Learning.

High-throughput screening (HTS), as one of the key techniques in drug discovery, is frequently used to identify promising drug candidates in a largely automated and cost-effective way. One of the necessary conditions for successful HTS campaigns is a large and diverse compound library, enabling hundreds of thousands of activity measurements per project. Such collections of data hold great promise for computational and experimental drug discovery efforts, especially when leveraged in combination with modern deep learning techniques, and can potentially lead to improved drug activity predictions and cheaper and more effective experimental design. However, existing collections of machine-learning-ready public datasets do not exploit the multiple data modalities present in real-world HTS projects. Thus, the largest fraction of experimental measurements, corresponding to hundreds of thousands of "noisy" activity values from primary screening, are effectively ignored in the majority of machine learning models of HTS data. To address these limitations, we introduce Multifidelity PubChem BioAssay (MF-PCBA), a curated collection of 60 datasets that includes two data modalities for each dataset, corresponding to primary and confirmatory screening, an aspect that we call multifidelity. Multifidelity data accurately reflect real-world HTS conventions and present a new, challenging task for machine learning: the integration of low- and high-fidelity measurements through molecular representation learning, taking into account the orders-of-magnitude difference in size between the primary and confirmatory screens. Here we detail the steps taken to assemble MF-PCBA in terms of data acquisition from PubChem and the filtering steps required to curate the raw data. We also provide an evaluation of a recent deep-learning-based method for multifidelity integration across the introduced datasets, demonstrating the benefit of leveraging all HTS modalities, and a discussion in terms of the roughness of the molecular activity landscape. In total, MF-PCBA contains over 16.6 million unique molecule-protein interactions. The datasets can be easily assembled by using the source code available at https://github.com/davidbuterez/mf-pcba.

Accelerated FEV1 decline and risk of cardiovascular disease and mortality in a primary care population of COPD patients.

Accelerated lung function decline has been associated with increased risk of cardiovascular disease (CVD) in a general population, but little is known about this association in chronic obstructive pulmonary disease (COPD). We investigated the association between accelerated lung function decline and CVD outcomes and mortality in a primary care COPD population.COPD patients without a history of CVD were identified in the Clinical Practice Research Datalink (CPRD)-GOLD primary care dataset (n=36 382). Accelerated decline in forced expiratory volume in 1 s (FEV1) was defined using the fastest quartile of the COPD population's decline. A Cox regression was used to assess the association between baseline accelerated FEV1 decline and a composite CVD outcome over follow-up (myocardial infarction, ischaemic stroke, heart failure, atrial fibrillation, coronary artery disease and CVD mortality). The model was adjusted for age, sex, smoking status, body mass index, history of asthma, hypertension, diabetes, statin use, Modified Medical Research Council (mMRC) dyspnoea score, exacerbation frequency and baseline FEV1 % predicted.6110 COPD patients (16.8%) had a CVD event during follow-up; median length of follow-up was 3.6 years (interquartile range (IQR) 1.7-6.1 years). Median rate of FEV1 decline was -19.4 mL·year-1 (IQR -40.5-1.9); 9095 patients (25%) had accelerated FEV1 decline (> -40.5 mL·year-1), 27 287 (75%) did not (≤ -40.5 mL·year-1). Risk of CVD and mortality was similar between patients with and without accelerated FEV1 decline (HRadj 0.98, 95% CI 0.90-1.06). Corresponding risk estimates were 0.99 (95% CI 0.83-1.20) for heart failure, 0.89 (95% CI 0.70-1.12) for myocardial infarction, 1.01 (95% CI 0.82-1.23) for stroke, 0.97 (95% CI 0.81-1.15) for atrial fibrillation, 1.02 (95% CI 0.87-1.19) for coronary artery disease and 0.94 (95% CI 0.71-1.25) for CVD mortality. Rather, risk of CVD was associated with a mMRC score ≤2 and two or more exacerbations in the year prior.CVD outcomes and mortality were associated with exacerbation frequency and severity and increased mMRC dyspnoea score but not with accelerated FEV1 decline.

Characteristics, service use and mortality of clusters of multimorbid patients in England: a population-based study.

BACKGROUND: Multimorbidity is associated with mortality and service use, with specific types of multimorbidity having differential effects. Additionally, multimorbidity is often negatively associated with participation in research cohorts. Therefore, we set out to identify clusters of multimorbidity patients and how they are differentially associated with mortality and service use across age groups in a population-representative sample. METHODS: Linked primary and secondary care electronic health records contributed by 382 general practices in England to the Clinical Practice Research Datalink (CPRD) were used. The study included a representative set of multimorbid adults (18 years old or more, N = 113,211) with two or more long-term conditions (a total of 38 conditions were included). A random set of 80% of the multimorbid patients (N = 90,571) were stratified by age groups and clustered using latent class analysis. Consistency between obtained multimorbidity phenotypes, classification quality and associations with demographic characteristics and primary outcomes (GP consultations, hospitalisations, regular medications and mortality) was validated in the remaining 20% of multimorbid patients (N = 22,640). RESULTS: We identified 20 patient clusters across four age strata. The clusters with the highest mortality comprised psychoactive substance and alcohol misuse (aged 18-64); coronary heart disease, depression and pain (aged 65-84); and coronary heart disease, heart failure and atrial fibrillation (aged 85+). The clusters with the highest service use coincided with those with the highest mortality for people aged over 65. For people aged 18-64, the cluster with the highest service use comprised depression, anxiety and pain. The majority of 85+-year-old multimorbid patients belonged to the cluster with the lowest service use and mortality for that age range. Pain featured in 13 clusters. CONCLUSIONS: This work has highlighted patterns of multimorbidity that have implications for health services. These include the importance of psychoactive substance and alcohol misuse in people under the age of 65, of co-morbid depression and coronary heart disease in people aged 65-84 and of cardiovascular disease in people aged 85+.

Inhaled corticosteroids and FEV1 decline in chronic obstructive pulmonary disease: a systematic review.

Rate of FEV1 decline in COPD is heterogeneous and the extent to which inhaled corticosteroids (ICS) influence the rate of decline is unclear. The majority of previous reviews have investigated specific ICS and non-ICS inhalers and have consisted of randomised control trials (RCTs), which have specific inclusion and exclusion criteria and short follow up times. We aimed to investigate the association between change in FEV1 and ICS-containing medications in COPD patients over longer follow up times.MEDLINE and EMBASE were searched and literature comparing change in FEV1 in COPD patients taking ICS-containing medications with patients taking non-ICS-containing medications were identified. Titles, abstract, and full texts were screened and information extracted using the PICO checklist. Risk of bias was assessed using the Cochrane Risk of Bias tool and a descriptive synthesis of the literature was carried out due to high heterogeneity of included studies.Seventeen studies met our inclusion criteria. We found that the difference in change in FEV1 in people using ICS and non-ICS containing medications depended on the study follow-up time. Shorter follow-up studies (1 year or less) were more likely to report an increase in FEV1 from baseline in both patients on ICS and in patients on non-ICS-containing medications, with the majority of these studies showing a greater increase in FEV1 in patients on ICS-containing medications. Longer follow-up studies (greater than 1 year) were more likely to report a decline in FEV1 from baseline in patients on ICS and in patients on non-ICS containing medications but rates of FEV1 decline were similar.Further studies are needed to better understand changes in FEV1 when ICS-containing medications are prescribed and to determine whether ICS-containing medications influence rate of decline in FEV1 in the long term. Results from inclusive trials and observational patient cohorts may provide information more generalisable to a population of COPD patients.

Wigwams: identifying gene modules co-regulated across multiple biological conditions.

MOTIVATION: Identification of modules of co-regulated genes is a crucial first step towards dissecting the regulatory circuitry underlying biological processes. Co-regulated genes are likely to reveal themselves by showing tight co-expression, e.g. high correlation of expression profiles across multiple time series datasets. However, numbers of up- or downregulated genes are often large, making it difficult to discriminate between dependent co-expression resulting from co-regulation and independent co-expression. Furthermore, modules of co-regulated genes may only show tight co-expression across a subset of the time series, i.e. show condition-dependent regulation. RESULTS: Wigwams is a simple and efficient method to identify gene modules showing evidence for co-regulation in multiple time series of gene expression data. Wigwams analyzes similarities of gene expression patterns within each time series (condition) and directly tests the dependence or independence of these across different conditions. The expression pattern of each gene in each subset of conditions is tested statistically as a potential signature of a condition-dependent regulatory mechanism regulating multiple genes. Wigwams does not require particular time points and can process datasets that are on different time scales. Differential expression relative to control conditions can be taken into account. The output is succinct and non-redundant, enabling gene network reconstruction to be focused on those gene modules and combinations of conditions that show evidence for shared regulatory mechanisms. Wigwams was run using six Arabidopsis time series expression datasets, producing a set of biologically significant modules spanning different combinations of conditions. AVAILABILITY AND IMPLEMENTATION: A Matlab implementation of Wigwams, complete with graphical user interfaces and documentation, is available at: warwick.ac.uk/wigwams. .

Arabidopsis defense against Botrytis cinerea: chronology and regulation deciphered by high-resolution temporal transcriptomic analysis.

Transcriptional reprogramming forms a major part of a plant's response to pathogen infection. Many individual components and pathways operating during plant defense have been identified, but our knowledge of how these different components interact is still rudimentary. We generated a high-resolution time series of gene expression profiles from a single Arabidopsis thaliana leaf during infection by the necrotrophic fungal pathogen Botrytis cinerea. Approximately one-third of the Arabidopsis genome is differentially expressed during the first 48 h after infection, with the majority of changes in gene expression occurring before significant lesion development. We used computational tools to obtain a detailed chronology of the defense response against B. cinerea, highlighting the times at which signaling and metabolic processes change, and identify transcription factor families operating at different times after infection. Motif enrichment and network inference predicted regulatory interactions, and testing of one such prediction identified a role for TGA3 in defense against necrotrophic pathogens. These data provide an unprecedented level of detail about transcriptional changes during a defense response and are suited to systems biology analyses to generate predictive models of the gene regulatory networks mediating the Arabidopsis response to B. cinerea.

High-resolution temporal profiling of transcripts during Arabidopsis leaf senescence reveals a distinct chronology of processes and regulation.

Leaf senescence is an essential developmental process that impacts dramatically on crop yields and involves altered regulation of thousands of genes and many metabolic and signaling pathways, resulting in major changes in the leaf. The regulation of senescence is complex, and although senescence regulatory genes have been characterized, there is little information on how these function in the global control of the process. We used microarray analysis to obtain a high-resolution time-course profile of gene expression during development of a single leaf over a 3-week period to senescence. A complex experimental design approach and a combination of methods were used to extract high-quality replicated data and to identify differentially expressed genes. The multiple time points enable the use of highly informative clustering to reveal distinct time points at which signaling and metabolic pathways change. Analysis of motif enrichment, as well as comparison of transcription factor (TF) families showing altered expression over the time course, identify clear groups of TFs active at different stages of leaf development and senescence. These data enable connection of metabolic processes, signaling pathways, and specific TF activity, which will underpin the development of network models to elucidate the process of senescence.

Alzheimer's disease biomarker discovery using SOMAscan multiplexed protein technology.

Blood proteins and their complexes have become the focus of a great deal of interest in the context of their potential as biomarkers of Alzheimer's disease (AD). We used a SOMAscan assay for quantifying 1001 proteins in blood samples from 331 AD, 211 controls, and 149 mild cognitive impaired (MCI) subjects. The strongest associations of protein levels with AD outcomes were prostate-specific antigen complexed to α1-antichymotrypsin (AD diagnosis), pancreatic prohormone (AD diagnosis, left entorhinal cortex atrophy, and left hippocampus atrophy), clusterin (rate of cognitive decline), and fetuin B (left entorhinal atrophy). Multivariate analysis found that a subset of 13 proteins predicted AD with an accuracy of area under the curve of 0.70. Our replication of previous findings provides further evidence that levels of these proteins in plasma are truly associated with AD. The newly identified proteins could be potential biomarkers and are worthy of further investigation.

Transfer learning with graph neural networks for improved molecular property prediction in the multi-fidelity setting.

We investigate the potential of graph neural networks for transfer learning and improving molecular property prediction on sparse and expensive to acquire high-fidelity data by leveraging low-fidelity measurements as an inexpensive proxy for a targeted property of interest. This problem arises in discovery processes that rely on screening funnels for trading off the overall costs against throughput and accuracy. Typically, individual stages in these processes are loosely connected and each one generates data at different scale and fidelity. We consider this setup holistically and demonstrate empirically that existing transfer learning techniques for graph neural networks are generally unable to harness the information from multi-fidelity cascades. Here, we propose several effective transfer learning strategies and study them in transductive and inductive settings. Our analysis involves a collection of more than 28 million unique experimental protein-ligand interactions across 37 targets from drug discovery by high-throughput screening and 12 quantum properties from the dataset QMugs. The results indicate that transfer learning can improve the performance on sparse tasks by up to eight times while using an order of magnitude less high-fidelity training data. Moreover, the proposed methods consistently outperform existing transfer learning strategies for graph-structured data on drug discovery and quantum mechanics datasets.

Urate-lowering therapy in patients with hyperuricemia and heart failure: A retrospective cohort study using the UK Clinical Practice Research Datalink.

BACKGROUND: Elevated serum uric acid (sUA) is associated with heart failure (HF). HYPOTHESIS: Urate-lowering therapy (ULT) in HF is associated with lower risk of HF hospitalization (hHF) and mortality. METHODS: Data on patients with HF and gout or hyperuricemia in the Clinical Practice Research Datalink database linked to the Hospital Episode Statistics and the Office for National Statistics in the United Kingdom were analyzed. Risks of hHF and all-cause mortality or cardiovascular-related mortality by ULT exposure (ULT initiated within ≤6 months of gout or hyperuricemia diagnosis) were analyzed in a propensity score-matched cohort using adjusted Cox proportional hazards regression models. RESULTS: Of 2174 propensity score-matched pairs, patients were predominantly male, aged >70 years, with mean ± standard deviation sUA 9.3 ± 1.8 (ULT-exposed) and 9.4 ± 1.9 mg/dL (ULT-unexposed). At 5 years, ULT-exposed patients had a 43% lower risk of hHF or all-cause mortality (adjusted hazard ratio [HR]: 0.57; 95% confidence interval [CI]: 0.51-0.65) and a 19% lower risk of hHF or cardiovascular-related mortality (adjusted HR: 0.81; 95% CI: 0.71-0.92) versus no ULT exposure. CONCLUSION: ULT was associated with reduced risk of adverse clinical outcomes in patients with HF and gout or hyperuricemia over 5 years.

Effect of COVID-19 on Bronchiectasis Exacerbation Rates: A Retrospective U.S. Insurance Claims Study.

Rationale: Bronchiectasis is a chronic, progressive disease of bronchial dilation, inflammation, and scarring leading to impaired mucociliary clearance and increased susceptibility to infection. Identified causes include previous severe respiratory infections. A small, single-center UK study demonstrated a reduction in bronchiectasis exacerbations during the first year of the coronavirus disease (COVID-19) pandemic. No studies have been conducted in a U.S. (commercially insured) cohort to date. Objectives: To explore the impact of the COVID-19 pandemic on the frequency of exacerbations in a large cohort of commercially insured U.S. patients with bronchiectasis by testing the hypothesis that U.S. patients with bronchiectasis had fewer exacerbations during the pandemic. Methods: This retrospective observational cohort study used health insurance claims data from Optum's deidentified Clinformatics Data Mart database, which included U.S. patients and their covered dependents. Eligible patients were ⩾18 years of age with bronchiectasis; patients with other respiratory conditions were excluded. The main study cohort excluded patients with frequent asthma and/or chronic obstructive pulmonary disease diagnoses. The primary objective was to compare the bronchiectasis exacerbation rates before and during the COVID-19 pandemic. Results: The median number of exacerbations per patient per year decreased significantly from the year before the COVID-19 pandemic to the first year of the pandemic (1 vs. 0; P < 0.01). More patients had zero exacerbations during the first year of the pandemic than the year prior (57% vs. 24%; McNemar's chi-square = 122.56; P < 0.01). Conclusions: In a U.S. population-based study of patients with International Classification of Diseases codes for bronchiectasis, the rate of exacerbations during Year 1 of the COVID-19 pandemic was reduced compared with the 2-year time period preceding the pandemic.

Modelling local and general quantum mechanical properties with attention-based pooling.

Atom-centred neural networks represent the state-of-the-art for approximating the quantum chemical properties of molecules, such as internal energies. While the design of machine learning architectures that respect chemical principles has continued to advance, the final atom pooling operation that is necessary to convert from atomic to molecular representations in most models remains relatively undeveloped. The most common choices, sum and average pooling, compute molecular representations that are naturally a good fit for many physical properties, while satisfying properties such as permutation invariance which are desirable from a geometric deep learning perspective. However, there are growing concerns that such simplistic functions might have limited representational power, while also being suboptimal for physical properties that are highly localised or intensive. Based on recent advances in graph representation learning, we investigate the use of a learnable pooling function that leverages an attention mechanism to model interactions between atom representations. The proposed pooling operation is a drop-in replacement requiring no changes to any of the other architectural components. Using SchNet and DimeNet++ as starting models, we demonstrate consistent uplifts in performance compared to sum and mean pooling and a recent physics-aware pooling operation designed specifically for orbital energies, on several datasets, properties, and levels of theory, with up to 85% improvements depending on the specific task.

Evaluating clinical outcomes and prognosis in patients with cirrhosis and portal hypertension: a retrospective observational cohort study.

OBJECTIVE: Cirrhosis describes the end-stage of chronic liver disease. Irreversible changes in the liver cause portal hypertension, which can progress to serious complications and death. Only a few studies with small sample sizes have investigated the prognosis of cirrhosis with portal hypertension. We used electronic healthcare records to examine liver-related outcomes in patients with diagnosed/suspected portal hypertension. DESIGN: This retrospective observational cohort study used secondary health data between 1 January 2017 and 3 December 2020 from the TriNetX Network, a federated electronic healthcare records platform. Three patient groups with cirrhosis and diagnosed/suspected portal hypertension were identified ('most severe', 'moderate severity' and 'least severe'). Outcomes studied individually and as a composite were variceal haemorrhage, hepatic encephalopathy, complications of ascites and recorded mortality up to 24 months. RESULTS: There were 13 444, 23 299, and 23 836 patients in the most severe, moderate severity and least severe groups, respectively. Mean age was similar across groups; most participants were white. The most common individual outcomes at 24 months were variceal haemorrhage in the most severe group, recorded mortality and hepatic encephalopathy in the moderate severity group, and recorded mortality in the least severe group. Recorded mortality rate was similar across groups. For the composite outcome, cumulative incidence was 59% in the most severe group at 6 months. Alcohol-associated liver disease and metabolic-associated steatohepatitis were significantly associated with the composite outcome across groups. CONCLUSION: Our analysis of a large dataset from electronic healthcare records illustrates the poor prognosis of patients with diagnosed/suspected portal hypertension.

Inhaled Corticosteroid Withdrawal and Change in Lung Function in Primary Care Patients with Chronic Obstructive Pulmonary Disease in England.

Rationale: In chronic obstructive pulmonary disease (COPD), inhaled corticosteroids (ICS) are associated with pneumonia, highlighting the importance of investigating subgroups of patients who may benefit from prolonged ICS use. Despite this, the WISDOM (Withdrawal of Inhaled Steroids during Optimized Bronchodilator Management) trial found a greater decline in forced expiratory volume in 1 second (FEV1) in patients with COPD who withdrew from ICS compared with patients who remained on triple therapy. Objectives: We investigated the association between ICS withdrawal and the rate of FEV1 decline in patients with COPD using routinely collected electronic healthcare records. Methods: Using CPRD (Clinical Practice Research Datalink) Aurum and Hospital episode statistics, we included patients with COPD who had been on triple therapy for at least 4 months. Patients were categorized into those who withdrew from ICS and those who remained on triple therapy during follow-up. Three cohorts were created: 1) patients meeting the WISDOM trial eligibility criteria; 2) patients with COPD not restricted by the WISDOM trial eligibility criteria; and 3) patients who would have been excluded from the WISDOM trial on the basis of their comorbidities. Mixed linear regression was used to model the association between ICS withdrawal and the rate of FEV1 decline (ml/year) adjusted for baseline characteristics. Results: A total of 6,008 patients with COPD met the WISDOM eligibility criteria, of which 9.0% withdrew from ICS. Mean rates of FEV1 declined -7.8 ml/year (95% confidence interval [CI], -19.7 to 4.1) for withdrawers and -15.2 ml/year (95% CI, -18.7 to -11.8) for those who remained on triple therapy (difference, P = 0.264). A total of 60,645 patients with COPD were not restricted by the WISDOM eligibility criteria. The mean rate of FEV1 decline was -32.6 ml/year (95% CI, -33.6 to -31.5) for withdrawers and -36.4 ml/year (95% CI, -39.4 to -33.4) for those who remained on triple therapy. A total of 32,882 patients with COPD were included in the last population representing those who would have been excluded from the WISDOM trial because of their comorbidities. The mean rate of FEV1 decline was -29.4 ml/year (95% CI, -30 to -28.1) in withdrawers and -31.3 ml/year (95% CI, -35 to -27.5) in those who remained on triple therapy. Conclusions: The rate of FEV1 decline was similar between patients on triple therapy and patients who withdrew from ICS regardless of the specific COPD population studied. In routine clinical practice, few patients with COPD meet WISDOM eligibility criteria, and few patients are withdrawn from ICS.

In simulated data and health records, latent class analysis was the optimum multimorbidity clustering algorithm.

BACKGROUND AND OBJECTIVES: To investigate the reproducibility and validity of latent class analysis (LCA) and hierarchical cluster analysis (HCA), multiple correspondence analysis followed by k-means (MCA-kmeans) and k-means (kmeans) for multimorbidity clustering. METHODS: We first investigated clustering algorithms in simulated datasets with 26 diseases of varying prevalence in predetermined clusters, comparing the derived clusters to known clusters using the adjusted Rand Index (aRI). We then them investigated the medical records of male patients, aged 65 to 84 years from 50 UK general practices, with 49 long-term health conditions. We compared within cluster morbidity profiles using the Pearson correlation coefficient and assessed cluster stability using in 400 bootstrap samples. RESULTS: In the simulated datasets, the closest agreement (largest aRI) to known clusters was with LCA and then MCA-kmeans algorithms. In the medical records dataset, all four algorithms identified one cluster of 20-25% of the dataset with about 82% of the same patients across all four algorithms. LCA and MCA-kmeans both found a second cluster of 7% of the dataset. Other clusters were found by only one algorithm. LCA and MCA-kmeans clustering gave the most similar partitioning (aRI 0.54). CONCLUSION: LCA achieved higher aRI than other clustering algorithms.

Temporal clustering by affinity propagation reveals transcriptional modules in Arabidopsis thaliana.

MOTIVATION: Identifying regulatory modules is an important task in the exploratory analysis of gene expression time series data. Clustering algorithms are often used for this purpose. However, gene regulatory events may induce complex temporal features in a gene expression profile, including time delays, inversions and transient correlations, which are not well accounted for by current clustering methods. As the cost of microarray experiments continues to fall, the temporal resolution of time course studies is increasing. This has led to a need to take account of detailed temporal features of this kind. Thus, while standard clustering methods are both widely used and much studied, their shared shortcomings with respect to such temporal features motivates the work presented here. RESULTS: Here, we introduce a temporal clustering approach for high-dimensional gene expression data which takes account of time delays, inversions and transient correlations. We do so by exploiting a recently introduced, message-passing-based algorithm called Affinity Propagation (AP). We take account of temporal features of interest following an approximate but efficient dynamic programming approach due to Qian et al. The resulting approach is demonstrably effective in its ability to discern non-obvious temporal features, yet efficient and robust enough for routine use as an exploratory tool. We show results on validated transcription factor-target pairs in yeast and on gene expression data from a study of Arabidopsis thaliana under pathogen infection. The latter reveals a number of biologically striking findings. AVAILABILITY: Matlab code for our method is available at http://www.wsbc.warwick.ac.uk/stevenkiddle/tcap.html.

Challenges and Pitfalls of Using Repeat Spirometry Recordings in Routine Primary Care Data to Measure FEV1 Decline in a COPD Population.

BACKGROUND: Electronic healthcare records (EHR) are increasingly used in epidemiological studies but are often viewed as lacking quality compared to randomised control trials and prospective cohorts. Studies of patients with chronic obstructive pulmonary disease (COPD) often use the rate of forced expiratory volume in 1 second (FEV1) decline as an outcome; however, its definition and robustness in EHR have not been investigated. We aimed to investigate how the rate of FEV1 decline differs by the criteria used in an EHR database. METHODS: Clinical Practice Research Datalink and Hospital Episode Statistics were used. Patient populations were defined using 8 sets of criteria around repeated FEV1 measurements. At a minimum, patients had a diagnosis of COPD, were ≥35 years old, were current or ex-smokers, and had data recorded from 2004. FEV1 measurements recorded during follow-up were identified. Thereafter, eight populations were defined based on criteria around: i) the exclusion of patients or individual measurements with potential measurement error; ii) minimum number of FEV1 measurements; iii) minimum time interval between measurements; iv) specific timing of measurements; v) minimum follow-up time; and vi) the use of linked data. For each population, the rate of FEV1 decline was estimated using mixed linear regression. RESULTS: For 7/8 patient populations, rates of FEV1 decline (age and sex adjusted) were similar and ranged from -18.7mL/year (95% CI -19.2 to -18.2) to -16.5mL/year (95% CI -17.3 to -15.7). Rates of FEV1 decline in populations that excluded patients with potential measurement error ranged from -79.4mL/year (95% CI -80.7 to -78.2) to -46.8mL/year (95% CI -47.6 to -46.0). CONCLUSION: FEV1 decline remained similar in a COPD population regardless of number of FEV1 measurements, time intervals between measurements, follow-up period, exclusion of specific FEV1 measurements, and linkage to HES. However, exclusion of individuals with questionable data led to selection bias and faster rates of decline.

Characteristics Associated with Accelerated Lung Function Decline in a Primary Care Population with Chronic Obstructive Pulmonary Disease.

Background: Estimates for lung function decline in chronic obstructive pulmonary disease (COPD) have differed by study setting and have not been described in a UK primary care population. Purpose: To describe rates of FEV1 and FVC decline in COPD and investigate characteristics associated with accelerated decline. Patients and Methods: Current/ex-smoking COPD patients (35 years+) who had at least 2 FEV1 or FVC measurements ≥6 months apart were included using Clinical Practice Research Datalink. Patients were followed up for a maximum of 13 years. Accelerated rate of lung function decline was defined as the fastest quartile of decline using mixed linear regression, and association with baseline characteristics was investigated using logistic regression. Results: A total of 72,683 and 50,649 COPD patients had at least 2 FEV1 or FVC measurements, respectively. Median rates of FEV1 and FVC changes or decline were -18.1mL/year (IQR: -31.6 to -6.0) and -22.7mL/year (IQR: -39.9 to -6.7), respectively. Older age, high socioeconomic status, being underweight, high mMRC dyspnoea and frequent AECOPD or severe AECOPD were associated with an accelerated rate of FEV1 and FVC decline. Current smoking, mild airflow obstruction and inhaled corticosteroid treatment were additionally associated with accelerated FEV1 decline whilst women, sputum production and severe airflow obstruction were associated with accelerated FVC decline. Conclusion: Rate of FEV1 and FVC decline was similar and showed similar heterogeneity. Whilst FEV1 and FVC shared associations with baseline characteristics, a few differences highlighted the importance of both lung function measures in COPD progression. We identified important characteristics that should be monitored for disease progression.

Prediction of five-year mortality after COPD diagnosis using primary care records.

Accurate prognosis information after a diagnosis of chronic obstructive pulmonary disease (COPD) would facilitate earlier and better informed decisions about the use of prevention strategies and advanced care plans. We therefore aimed to develop and validate an accurate prognosis model for incident COPD cases using only information present in general practitioner (GP) records at the point of diagnosis. Incident COPD patients between 2004-2012 over the age of 35 were studied using records from 396 general practices in England. We developed a model to predict all-cause five-year mortality at the point of COPD diagnosis, using 47,964 English patients. Our model uses age, gender, smoking status, body mass index, forced expiratory volume in 1-second (FEV1) % predicted and 16 co-morbidities (the same number as the Charlson Co-morbidity Index). The performance of our chosen model was validated in all countries of the UK (N = 48,304). Our model performed well, and performed consistently in validation data. The validation area under the curves in each country varied between 0.783-0.809 and the calibration slopes between 0.911-1.04. Our model performed better in this context than models based on the Charlson Co-morbidity Index or Cambridge Multimorbidity Score. We have developed and validated a model that outperforms general multimorbidity scores at predicting five-year mortality after COPD diagnosis. Our model includes only data routinely collected before COPD diagnosis, allowing it to be readily translated into clinical practice, and has been made available through an online risk calculator (https://skiddle.shinyapps.io/incidentcopdsurvival/).

Dysregulated Antibody, Natural Killer Cell and Immune Mediator Profiles in Autoimmune Thyroid Diseases.

The pathogenesis of autoimmune thyroid diseases (AITD) is poorly understood and the association between different immune features and the germline variants involved in AITD are yet unclear. We previously observed systemic depletion of IgG core fucosylation and antennary α1,2 fucosylation in peripheral blood mononuclear cells in AITD, correlated with anti-thyroid peroxidase antibody (TPOAb) levels. Fucose depletion is known to potentiate strong antibody-mediated NK cell activation and enhanced target antigen-expressing cell killing. In autoimmunity, this may translate to autoantibody-mediated immune cell recruitment and attack of self-antigen expressing normal tissues. Hence, we investigated the crosstalk between immune cell traits, secreted proteins, genetic variants and the glycosylation patterns of serum IgG, in a multi-omic and cross-sectional study of 622 individuals from the TwinsUK cohort, 172 of whom were diagnosed with AITD. We observed associations between two genetic variants (rs505922 and rs687621), AITD status, the secretion of Desmoglein-2 protein, and the profile of two IgG N-glycan traits in AITD, but further studies need to be performed to better understand their crosstalk in AITD. On the other side, enhanced afucosylated IgG was positively associated with activatory CD335- CD314+ CD158b+ NK cell subsets. Increased levels of the apoptosis and inflammation markers Caspase-2 and Interleukin-1α positively associated with AITD. Two genetic variants associated with AITD, rs1521 and rs3094228, were also associated with altered expression of the thyrocyte-expressed ligands known to recognize the NK cell immunoreceptors CD314 and CD158b. Our analyses reveal a combination of heightened Fc-active IgG antibodies, effector cells, cytokines and apoptotic signals in AITD, and AITD genetic variants associated with altered expression of thyrocyte-expressed ligands to NK cell immunoreceptors. Together, TPOAb responses, dysregulated immune features, germline variants associated with immunoactivity profiles, are consistent with a positive autoreactive antibody-dependent NK cell-mediated immune response likely drawn to the thyroid gland in AITD.