Logo: to the web site of Uppsala University

Background: We previously reported significant correlations between a direct measure of insulin sensitivity (IS) and blood levels of proteins measured using the Proximity Extension Assay (PEA) in two European cohorts. However, protein correlations with IS within non-European populations, in response to short-term interventions that improve IS, and any causal associations with IS have not yet been established.

Methods: We measured 1470 proteins using the PEA in the plasma of 1015 research participants at Stanford University who underwent one or more direct measures of IS. Association analyses were carried out with multivariable linear regression within and across Stanford subgroups and within each of the two European cohorts. Association statistics were also meta-analyzed after transformation and harmonization of the two direct measures of IS. Lastly, we performed genome-wide association studies of IS and used genetic instruments of plasma proteins from the UK Biobank to identify candidate causal proteins for IS through Mendelian Randomization (MR) analysis.

Results: In age and sex adjusted model, 810 proteins were associated with baseline IS among 652 self-reported European participants in the Stanford cohort at a false discovery rate (FDR) < 0.05. Effect sizes for these proteins were highly correlated with those observed in 122 South Asian, 92 East Asian, 85 Hispanic, and 52 Black/ African American persons (r = 0.68 to 0.83, all P <= 4.3 x 10-113). Meta-analysis of the full Stanford cohort with the two European cohorts (N = 2945) yielded 247 significant protein associations (FDR < 0.05), with 50 remaining significant after further adjustment for body mass index. In a subset of Stanford participants undergoing insulin sensitizing interventions (N = 53 taking thiazolidinediones, N = 66 with weight loss), 79.3 % of protein level changes were directionally consistent with the respective baseline association (observed/expected p = 6.0 x 10(-16)). MR analyses identified ten candidate causal proteins for IS, among which were SELE and ASGR1, proteins with established drug targets currently under investigation.

Conclusion: Plasma proteins measured using the PEA provide a robust signature for IS across diverse populations and after short-term insulin sensitizing interventions highlighting their potential value as universal biomarkers of insulin resistance. A small subset of markers provided insights into potential causal molecular mechanisms and therapeutic targets.

Imbalances in electrolyte concentrations can have severe consequences, but accurate and accessible measurements could improve patient outcomes. The current measurement method based on blood tests is accurate but invasive and time-consuming and is often unavailable for example in remote locations or an ambulance setting. In this paper, we explore the use of deep neural networks (DNNs) for regression tasks to accurately predict continuous electrolyte concentrations from electrocardiograms (ECGs), a quick and widely adopted tool. We analyze our DNN models on a novel dataset of over 290,000 ECGs across four major electrolytes and compare their performance with traditional machine learning models. For improved understanding, we also study the full spectrum from continuous predictions to a binary classification of extreme concentration levels. Finally, we investigate probabilistic regression approaches and explore uncertainty estimates for enhanced clinical usefulness. Our results show that DNNs outperform traditional models but model performance varies significantly across different electrolytes. While discretization leads to good classification performance, it does not address the original problem of continuous concentration level prediction. Probabilistic regression has practical potential, but our uncertainty estimates are not perfectly calibrated. Our study is therefore a first step towards developing an accurate and reliable ECG-based method for electrolyte concentration level prediction—a method with high potential impact within multiple clinical scenarios.

AIMS: In patients with heart failure, treatment patterns in the last years of life have not been assessed at large scale. We aimed to assess whether heart failure treatment patterns up to 5 years prior to death changed over time.

METHODS AND RESULTS: In a cohort study covering the whole Swedish population, we assessed all heart failure patients who died between 1 July 2007 and 31 December 2020 for evidence-based treatments. The proportion on the respective treatment at the time of death was examined by year of death using binomial regression. Looking back in time, treatment discontinuation rates were estimated using Poisson regression on time-split data. Combining these models, the proportion on each medication was estimated up to 5 years prior to death. A total of 364 480 patients died with heart failure during the study period. Half were women, and the median (interquartile range) age at death was 86 (79-90). The use of all heart failure treatments decreased gradually closer to death, but the discontinuation rate of beta blockers decreased over time, resulting in an increasing proportion of patients on treatment at the time of death.

CONCLUSION: In patients with heart failure, a changing pattern of medical treatment during the last years of life was observed, most notably with an increasing use of beta blockers. This may in part be due to a changing pattern of comorbidities over time, with an increase in e.g. hypertension and atrial fibrillation, but a decline in ischaemic heart disease.

Myocardial infarction is a leading cause of death globally but is notoriously difficult to predict. We aimed to identify biomarkers of an imminent first myocardial infarction and design relevant prediction models. Here, we constructed a new case–cohort consortium of 2,018 persons without prior cardiovascular disease from six European cohorts, among whom 420 developed a first myocardial infarction within 6 months after the baseline blood draw. We analyzed 817 proteins and 1,025 metabolites in biobanked blood and 16 clinical variables. Forty-eight proteins, 43 metabolites, age, sex and systolic blood pressure were associated with the risk of an imminent first myocardial infarction. Brain natriuretic peptide was most consistently associated with the risk of imminent myocardial infarction. Using clinically readily available variables, we devised a prediction model for an imminent first myocardial infarction for clinical use in the general population, with good discriminatory performance and potential for motivating primary prevention efforts.

Background: The full spectrum of side effects from COVID-19 vaccinations and infections, including milder symptoms or health effects that do not lead to health care visits, remains unknown. Person-centered self-reporting of symptoms may offer a solution. Monitoring patient-reported outcomes over time will vary in importance for different patients. Individuals have unique needs and preferences, in terms of both communication methods and how the collected information is used to support care.

Objective: This study aims to describe how Symptoms, a system for person-centered self-reporting of symptoms and health-related quality of life, was utilized in investigating COVID-19 vaccine side effects. We illustrate this by presenting data from the Symptoms system in newly vaccinated individuals from the RECOVAC (Register-based large-scale national population study to monitor COVID-19 vaccination effectiveness and safety) study.

Methods: During the COVID-19 pandemic, newly vaccinated individuals were identified as the ideal population to query for milder symptoms related to COVID-19 vaccinations and infections. To this end, we used posters in observation areas at 150 vaccination sites across the Västra Götaland region of Sweden, inviting newly vaccinated individuals to use a novel digital system, Symptoms. In the Symptoms system, users can track their symptoms, functioning, and quality of life as often as they wish, using evidence-based patient-reported outcome measures and short numeric rating scales. These scales cover a prespecified list of symptoms based on common COVID-19 symptoms and previously reported vaccine side effects. Participants could also use numeric rating scales for self-defined symptoms if their symptom was not included on the prespecified list.

Results: A total of 731 people created user accounts and consented to share data for research between July 21, 2021, and September 27, 2022. The majority of users were female (444/731, 60.7%), with a median age of 38 (IQR 30-47) years. Most participants (498/702, 70.9%) did not report any of the comorbidities included in the questionnaire. Of the 731 participants, 563 (77.0%) reported experiencing 1 or more symptoms. The most common symptom was pain at the injection site (486/563, 86.3%), followed by fatigue (181/563, 32.1%) and headache (169/563, 30.0%). In total, 143 unique symptoms were reported. Of these, 29 were from the prespecified list, while the remaining 114 (79.7%) were self-defined entries in the symptom field. This suggests that the flexibility of the self-directed system-allowing individuals to decide which symptoms they consider worth tracking-may be an important feature.

Conclusions: Self-reported symptoms in the Symptoms system appeared to align with previously observed post-COVID-19 vaccination symptoms. The system was relatively easy to use and successfully captured broad, longitudinal data. Its person-centered and self-directed design seemed crucial in capturing the full burden of symptoms experienced by users.

Background Fibroblast growth factor-23 (FGF-23) is associated with a range of cardiovascular and noncardiovascular diseases in conventional epidemiological studies, but substantial residual confounding may exist. Mendelian randomization approaches can help control for such confounding.Methods SCALLOP Consortium data of 19,195 participants were used to generate an FGF-23 genetic score. Data from 337,448 UK Biobank participants were used to estimate associations between higher genetically predicted FGF-23 concentration and the odds of any atherosclerotic cardiovascular disease (n=26,266 events), nonatherosclerotic cardiovascular disease (n=12,652), and noncardiovascular diseases previously linked to FGF-23. Measurements of carotid intima-media thickness and left ventricular mass were available in a subset. Associations with cardiovascular outcomes were also tested in three large case-control consortia: CARDIOGRAMplusC4D (coronary artery disease, n=181,249 cases), MEGASTROKE (stroke, n=34,217), and HERMES (heart failure, n=47,309).Results We identified 34 independent variants for circulating FGF-23, which formed a validated genetic score. There were no associations between genetically predicted FGF-23 and any of the cardiovascular or non cardiovascular outcomes. In UK Biobank, the odds ratio (OR) for any atherosclerotic cardiovascular disease per 1-SD higher genetically predicted logFGF-23 was 1.03 (95% confidence interval [95% CI], 0.98 to 1.08), and for any nonatherosclerotic cardiovascular disease, it was 1.01 (95% CI, 0.94 to 1.09). The ORs in the case-control consortia were 1.00 (95% CI, 0.97 to 1.03) for coronary artery disease, 1.01 (95% CI, 0.95 to 1.07) for stroke, and 1.00 (95% CI, 0.95 to 1.05) for heart failure. In those with imaging, logFGF-23 was not associated with carotid or cardiac abnormalities.Conclusions Genetically predicted FGF-23 levels are not associated with atherosclerotic and nonatherosclerotic cardiovascular diseases, suggesting no important causal link.

Distinct tissue-specific mechanisms mediate insulin action in fasting and postprandial states. Previous genetic studies have largely focused on insulin resistance in the fasting state, where hepatic insulin action dominates. Here we studied genetic variants influencing insulin levels measured 2 h after a glucose challenge in >55,000 participants from three ancestry groups. We identified ten new loci (P < 5 × 10^-8) not previously associated with postchallenge insulin resistance, eight of which were shown to share their genetic architecture with type 2 diabetes in colocalization analyses. We investigated candidate genes at a subset of associated loci in cultured cells and identified nine candidate genes newly implicated in the expression or trafficking of GLUT4, the key glucose transporter in postprandial glucose uptake in muscle and fat. By focusing on postprandial insulin resistance, we highlighted the mechanisms of action at type 2 diabetes loci that are not adequately captured by studies of fasting glycemic traits.

Insulin secretion is critical for glucose homeostasis, and increased levels of the precursor proinsulin relative to insulin indicate pancreatic islet beta-cell stress and insufficient insulin secretory capacity in the setting of insulin resistance. We conducted meta-analyses of genome-wide association results for fasting proinsulin from 16 European-ancestry studies in 45,861 individuals. We found 36 independent signals at 30 loci (p value < 5 × 10^-8), which validated 12 previously reported loci for proinsulin and ten additional loci previously identified for another glycemic trait. Half of the alleles associated with higher proinsulin showed higher rather than lower effects on glucose levels, corresponding to different mechanisms. Proinsulin loci included genes that affect prohormone convertases, beta-cell dysfunction, vesicle trafficking, beta-cell transcriptional regulation, and lysosomes/autophagy processes. We colocalized 11 proinsulin signals with islet expression quantitative trait locus (eQTL) data, suggesting candidate genes, including ARSG, WIPI1, SLC7A14, and SIX3. The NKX6-3/ANK1 proinsulin signal colocalized with a T2D signal and an adipose ANK1 eQTL signal but not the islet NKX6-3 eQTL. Signals were enriched for islet enhancers, and we showed a plausible islet regulatory mechanism for the lead signal in the MADD locus. These results show how detailed genetic studies of an intermediate phenotype can elucidate mechanisms that may predispose one to disease.

BACKGROUND: Proteomic profiling could potentially disclose new pathophysiological pathways for cardiovascular diseases (CVD) and improve prediction at the individual level. We therefore aimed to study the plasma protein profile associated with the incidence of different CVDs.

METHODS: Plasma levels of 245 proteins suspected to be linked to CVD or metabolism were measured in 4 Swedish prospective population-based cohorts (SIMPLER [Swedish Infrastructure for Medical Population-Based Life-Course and Environmental Research], ULSAM (Uppsala Longitudinal Study of Adult Men), EpiHealth, and POEM [Prospective Investigation of Obesity, Energy Production, and Metabolism]) comprising 11 869 individuals, free of CVD diagnoses at baseline. Our primary CVD outcome was defined by a combined end point that included either incident myocardial infarction, stroke, or heart failure.

RESULTS: Using a discovery/validation approach, 42 proteins were associated with our primary composite end point occurring in 1163 subjects. In separate meta-analyses for each of the 3 CVD outcomes, 49 proteins were related to myocardial infarction, 34 to ischemic stroke, and 109 to heart failure. Thirteen proteins were related to all 3 outcomes. Of those, urokinase plasminogen activator surface receptor, adrenomedullin, and KIM-1 (kidney injury molecule 1) were also related to several markers of subclinical CVD in Prospective Investigation of Obesity, Energy production and Metabolism, reflecting myocardial or arterial pathologies. In prediction analysis, a lasso selection of 11 proteins in ULSAM improved the discrimination of CVD by 3.3% (P<0.0001) in SIMPLER when added to traditional risk factors.

CONCLUSIONS: Protein profiling in multiple samples disclosed several new proteins to be associated with subsequent myocardial infarction, stroke, and heart failure, suggesting common pathophysiological pathways for these diseases. KIM-1, urokinase plasminogen activator surface receptor, and adrenomedullin were novel early markers of CVD. A selection of 11 proteins improved the discrimination of CVD.

Aims/hypothesis The euglycaemic-hyperinsulinaemic clamp (EIC) is the reference standard for the measurement of whole-body insulin sensitivity but is laborious and expensive to perform. We aimed to assess the incremental value of high-through-put plasma proteomic profiling in developing signatures correlating with the M value derived from the EIC.

Methods We measured 828 proteins in the fasting plasma of 966 participants from the Relationship between Insulin Sensitivity and Cardiovascular disease (RISC) study and 745 participants from the Uppsala Longitudinal Study of Adult Men (ULSAM) using a high-throughput proximity extension assay. We used the least absolute shrinkage and selection operator (LASSO) approach using clinical variables and protein measures as features. Models were tested within and across cohorts. Our primary model performance metric was the proportion of the M value variance explained (R-2).

Results A standard LASSO model incorporating 53 proteins in addition to routinely available clinical variables increased the M value R-2 from 0.237 (95% CI 0.178, 0.303) to 0.456 (0.372, 0.536) in RISC. A similar pattern was observed in ULSAM, in which the M value R-2 increased from 0.443 (0.360, 0.530) to 0.632 (0.569, 0.698) with the addition of 61 proteins. Models trained in one cohort and tested in the other also demonstrated significant improvements in R-2 despite differences in baseline cohort characteristics and clamp methodology (RISC to ULSAM: 0.491 [0.433, 0.539] for 51 proteins; ULSAM to RISC: 0.369 [0.331, 0.416] for 67 proteins). A randomised LASSO and stability selection algorithm selected only two proteins per cohort (three unique proteins), which improved R-2 but to a lesser degree than in standard LASSO models: 0.352 (0.266, 0.439) in RISC and 0.495 (0.404, 0.585) in ULSAM. Reductions in improvements of R-2 with randomised LASSO and stability selection were less marked in cross-cohort analyses (RISC to ULSAM R-2 0.444 [0.391, 0.497]; ULSAM to RISC R-2 0.348 [0.300, 0.396]). Models of proteins alone were as effective as models that included both clinical variables and proteins using either standard or randomised LASSO. The single most consistently selected protein across all analyses and models was IGF-binding protein 2.

Conclusions/interpretation A plasma proteomic signature identified using a standard LASSO approach improves the cross-sectional estimation of the M value over routine clinical variables. However, a small subset of these proteins identified using a stability selection algorithm affords much of this improvement, especially when considering cross-cohort analyses. Our approach provides opportunities to improve the identification of insulin-resistant individuals at risk of insulin resistance-related adverse health consequences.