BACKGROUND:Heart failure (HF) and chronic obstructive pulmonary disease (COPD) are leading causes of acute dyspnea in the emergency department (ED) and frequently coexist. However, their combined impact on short- and long-term outcomes in the acute setting remains insufficiently characterized. METHODS:We conducted a monocentric observational study based on the PARADISE cohort, including patients admitted to the ED for acute dyspnea between 2010 and 2019. Patients with a primary diagnosis of HF or COPD were included and stratified according to the presence of the alternate condition. The primary outcomes were in-hospital and post-discharge all-cause mortality, assessed using multivariable regression models. RESULTS:Among 5,131 patients, 3,543 had a primary diagnosis of HF and 1,588 of COPD. Concomitant disease was present in approximately 20% of patients in both groups.In the primary HF cohort, patients with COPD had lower in-hospital mortality compared with those without COPD (8.5% vs. 11.7%, p = 0.014), but similar overall mortality (69.3% vs. 68.9%). After adjustment, COPD remained associated with lower in-hospital mortality (OR 0.74; 95% CI 0.55-0.99; p = 0.050) and with a modest increase in long-term mortality (HR 1.22; 95% CI 1.09-1.36; p < 0.001).In the primary COPD cohort, patients with HF had higher in-hospital mortality (7.4% vs. 3.4%, p = 0.001) and markedly higher long-term mortality (68.7% vs. 48.7%, p < 0.001). After adjustment, HF was not significantly associated with in-hospital mortality (OR 1.54; 95% CI 0.87-2.65; p = 0.13), but remained strongly associated with increased long-term mortality (HR 1.48; 95% CI 1.25-1.76; p < 0.001). CONCLUSIONS:HF and COPD frequently coexist and are both associated with an increased long-term mortality risk, with a greater prognostic impact of HF in patients with COPD. These findings highlight the importance of systematic identification and optimized management of both conditions in this high-risk population.

AIMS:Diagnosing heart failure with preserved ejection fraction (HFpEF) remains challenging, particularly in older individuals. We hypothesized that machine learning (ML) approaches could improve diagnostic accuracy compared with HFpEF scores. METHODS:We evaluated the diagnostic performance of four supervised ML algorithms (random forest [RF], extreme gradient boosting [XGBoost], support vector machines, and decision trees) to identify HFpEF in individuals aged 60 to 80 years. The models were trained on three derivation cohorts (N = 1474; HFpEF: KaRen, MEDIA cohorts; community-based without HF: Malmö Preventive Project) and validated in two independent cohorts (N = 542; HFpEF: HF-Nancy cohort; community-based without HF: STANISLAS cohort). Performance metrics included accuracy, F-measure, area under the receiver operating characteristic curve (AUC), and C-index. ML models were also compared with HFA-PEFF, H2FPEF, and HFpEF-ABA scores. RESULTS:Among 2017 participants, RF and XGBoost demonstrated the highest diagnostic value, outperforming traditional HFpEF scores (AUC: RF, 0.98; XGBoost, 0.96; HFA-PEFF, 0.86; H2FPEF, 0.79). RF and XGBoost also showed the greatest gain in discriminative capacity among ML algorithms when compared with H2FPEF (ΔC-index: RF +0.20, XGBoost +0.18), HFA-PEFF (ΔC-index: RF +0.12, XGBoost +0.10), and HFpEF-ABA score (ΔC-index: RF +0.17, XGBoost +0.15). Elevated natriuretic peptides were by far the most influential feature in both RF and XGBoost models (36% of model explainability). CONCLUSIONS:Machine learning algorithms, particularly RF and XGBoost, demonstrated superior diagnostic accuracy compared to established HFpEF scoring systems. These findings support the potential integration of ML-based tools into clinical workflows to facilitate earlier identification of HFpEF and prompt initiation of guideline-recommended therapies.

Aims Cardiovascular (CV) trials have yielded neutral results in haemodialysis. A better understanding of patient profiles is needed to personalize treatment strategies in order to improve CV outcomes in this setting. This study sought to identify biological phenotypes based on proteomic data using machine learning approaches in patients undergoing haemodialysis.Methods and results A clustering analysis using 253 plasma protein biomarkers was performed in 382 patients (machine learning derivation analysis) from the AURORA trial, which tested the effect of rosuvastatin on CV outcomes in patients on haemodialysis. A decision tree was subsequently constructed to predict cluster membership and assess its association with CV outcomes in another subset of the trial (n = 389 patients, validation analysis). Four phenotypes were identified, namely 'cytokine storm signalling', 'toll-like receptors (TLRs) signalling', 'multiple pathways related to inflammation and fibrosis' phenotypes, as well as a 'reference phenotype' which exhibited the least biological abnormalities. In multivariable analysis of the validation study, after adjusting for key prognostic factors, the TLRs phenotype was significantly associated with CV death, all-cause mortality, and MACE (HR = 1.65 [1.13-2.41], 1.43 [1.03-1.98], and 1.48 [1.04-2.10], respectively).Conclusion Using unsupervised machine learning on proteomic data, we identified four mechanistic biological phenotypes involving cytokine storm and TLRs signalling, inflammation and fibrosis. These biological phenotypes may contribute to CV prognosis and pave the way for personalized therapy in haemodialysis.

BACKGROUND:Acute heart failure (AHF) and respiratory infection (RI) frequently coexist, with the latter commonly regarded as a trigger of AHF decompensation. However, the independent and combined prognostic impact of these conditions on survival is not well studied. We therefore assessed the association of AHF and RI, both separately and in combination, with subsequent mortality. METHODS:Patients discharged with diagnoses of AHF, RI, or both were identified from the PARADISE study, a large cohort of patients hospitalised for acute dyspnoea. Associations with in-hospital and post-discharge mortality were assessed using multivariable binomial logistic regression and Cox proportional hazards models, respectively. RESULTS:Among 11,679 patients, 4,349 (37%) had AHF alone, 5,091 (44%) had RI alone, and 2,239 (19%) had both AHF and RI. In-hospital mortality was highest in patients with concomitant AHF and RI (21.9%), whereas post-discharge mortality was highest among those with AHF (55.2%). After multivariable adjustment, the coexistence of AHF and RI was associated with higher in-hospital mortality compared with AHF alone (adjusted OR [aOR]: 1.62, 1.33-1.98, P<0.001), but not with higher post-discharge mortality (adjusted HR [aHR]: 0.99, 0.88-1.11, P=0.9). Compared with AHF alone, RI alone was not associated with a higher risk of death both during hospitalization (aOR 1.11, 0.89-1.39, P=0.3) and after discharge (aHR 1.07, 0.97-1.17, P=0.2). Results from sensitivity analyses including natriuretic peptides confirm those results. CONCLUSION:Patients with concomitant AHF and RI showed an increased in-hospital risk but no excess post-discharge risk compared with AHF alone, whereas RI alone is not associated with increased mortality both in-hospital and post-discharge.

Heart failure (HF) is a heterogeneous condition for which traditional classification methods do not allow for optimal patient stratification. Machine learning (ML) offers an innovative approach by segmenting patients into homogeneous subgroups based on multidimensional data (clinical, biological, imaging). The ML-assisted segmentation process follows the following steps: data acquisition and preparation: Integration of heterogeneous data (medical records, biomarkers, imaging). Unsupervised clustering: identification of clinical phenotypes using algorithms tailored to medical data (latent class model, K-Prototypes, etc.). Cluster validation: comparison with existing classifications and prognostic assessment using metrics such as the cindex. Cluster prediction: development of supervised models (random forest, neural networks) to assign new patients to these groups. Clinical application: integration into clinical tools to improve treatment personalization and HF prevention. Challenges include external validation, interoperability with electronic medical records, and model explainability to promote clinical adoption. Conclusion. -ML is revolutionizing HF segmentation by enabling more precise patient stratification, paving the way for precision medicine to improve prediction, management, and treatment personalization. (c) 2025 l'Academie nationale de medecine. Published by Elsevier Masson SAS. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

Aims:Data-driven clustering techniques may improve heart failure (HF) categorisation and provide prognostic insights. The present study aimed to elucidate the underlying pathophysiology of acute HF phenotypes based on pulmonary and systemic congestion at both the tissue (PTC, pulmonary tissue congestion; STC, systemic tissue congestion) and intravascular (PIVC, pulmonary intravascular congestion; SIVC, systemic intravascular congestion) level and to assess the association of identified phenotypes with a composite outcome of HF hospitalisation and death. Methods and results:Nineteen clinical, laboratory, and echocardiographic congestion markers were analyzed using clustering techniques to identify phenotypes in patients with worsening HF in the Nancy-HF cohort (n = 741), followed by validation of the clustering model in the BIOSTAT-CHF cohort (n = 4254). Network analysis was conducted using 363 proteins to identify underlying biological pathways. Five congestion phenotypes were identified: (1) PTC-dilated left ventricle (LV), (2) PTC-HFpEF, (3) PTC, STC-atrial fibrillation (AF), (4) PIVC-dilated left atrium (LA) and LV and (5) Global congestion. Compared with the 'PTC-dilated LV' phenotype, the risk of composite outcome was higher in 'PTC, STC-AF' and 'Global' congestion phenotypes [adjusted HR: 1.74 (1.13-2.67) and 2.41 (1.60-3.63), respectively]. In BIOSTAT-CHF, 'Global' congestion phenotype was associated with significantly higher risk [HR: 1.64 (1.04-2.58)]. In network analysis, the immune response pathway was linked to all phenotypes. 'PTC-HFpEF' was related to lipid, protein and angiotensin metabolism, 'PTC, STC-AF' was related to kinase-mediated signalling, extracellular matrix organisation and TNF-regulated cell death, while 'PIVC-dilated LA & LV' was related to kinase-mediated signalling and hemostasis. Conclusion:In worsening HF, clustering techniques identified clinical congestion profiles associated with both long-term clinical risk and differences in biomarkers, suggesting potential different underlying pathophysiologies. These clusters can be applied using the available online model to identify phenotypes as well as associated risks (https://cic-p-nancy.fr/ai-cong-hf/).

L’insuffisance cardiaque (IC) est une pathologie hétérogène dont la classification traditionnelle ne permet pas une stratification optimale des patients. Le machine learning (ML) apporte une approche innovante en segmentant les patients en sous-groupes homogènes à partir de données multidimensionnelles (clinique, biologique, imagerie). Le processus de segmentation assistée par ML passe par les étapes suivantes : acquisition et préparation des données : intégration de données hétérogènes (dossiers médicaux, biomarqueurs, imagerie). Clustering non supervisé : identification de phénotypes cliniques via des algorithmes adaptés aux données médicales (latent class model, K-Prototypes, etc.). Validation des clusters : comparaison aux classifications existantes et évaluation pronostique via des métriques comme le c-index. Prédiction des clusters : développement de modèles supervisés (random forest, réseaux de neurones) pour assigner de nouveaux patients à ces groupes. Application clinique : intégration dans les outils cliniques pour améliorer la personnalisation des traitements et la prévention de l’IC. Les défis incluent la validation externe, l’interopérabilité avec les dossiers médicaux électroniques et l’explicabilité des modèles pour favoriser leur adoption clinique. Conclusion Le ML révolutionne la segmentation en IC en permettant une stratification plus précise des patients, ouvrant la voie à une médecine de précision pour une meilleure prédiction, prise en charge et personnalisation des traitements.

Unsupervised machine learning can improve the characterization and stratification of patients with cardiovascular diseases (CVDs). Clustering algorithms, which group patients based on patterns in clinical data, can reveal distinct subgroups that may differ in prognosis and treatment response. Despite increasing research in this area, the practical use of clustering methods in routine clinical care remains limited by the lack of accessible tools and rigorous external validation. This review presents a systematic framework for applying unsupervised machine learning techniques to CVD research. The framework outlines a stepwise process-from identifying patient clusters and establishing their associations with clinical outcomes to developing predictive models for assigning new patients to these clusters. This approach aims to generate robust, externally validated models that can be integrated into clinical practice to support improved risk stratification and personalized treatment strategies. This framework can enhance the usefulness of clustering in CVD research, by providing valuable resource for medical professionals, stakeholders, and researchers in exploring more effective strategies for managing CVDs.

AimsHigh left ventricular filling pressure increases left atrial volume and causes myocardial fibrosis, which may decrease with spironolactone. We studied clinical and proteomic characteristics associated with left atrial volume indexed by body surface area (LAVi), and whether LAVi influences the response to spironolactone on biomarker expression and clinical variables.Methods and resultsIn the HOMAGE trial, where people at risk of heart failure were randomized to spironolactone or control, we analysed 421 participants with available LAVi and 276 proteomic measurements (Olink) at baseline, month 1 and 9 (mean age 73 ± 6 years; women 26%; LAVi 32 ± 9 ml/m2). Circulating proteins associated with LAVi were also assessed in asymptomatic individuals from a population‐based cohort (STANISLAS; n = 1640; mean age 49 ± 14 years; women 51%; LAVi 23 ± 7 ml/m2). In both studies, greater LAVi was significantly associated with greater left ventricular masses and volumes. In HOMAGE, after adjustment and correction for multiple testing, greater LAVi was associated with higher concentrations of matrix metallopeptidase‐2 (MMP‐2), insulin‐like growth factor binding protein‐2 (IGFBP‐2) and N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP) (false discovery rates [FDR] <0.05). These associations were externally replicated in STANISLAS (all FDR <0.05). Among these biomarkers, spironolactone decreased concentrations of MMP‐2 and NT‐proBNP, regardless of baseline LAVi (pinteraction > 0.10). Spironolactone also significantly reduced LAVi, improved left ventricular ejection fraction, lowered E/e', blood pressure and serum procollagen type I C‐terminal propeptide (PICP) concentration, a collagen synthesis marker, regardless of baseline LAVi (pinteraction > 0.10).ConclusionIn individuals without heart failure, LAVi was associated with MMP‐2, IGFBP‐2 and NT‐proBNP. Spironolactone reduced these biomarker concentrations as well as LAVi and PICP, irrespective of left atrial size.

ABSTRACT Background Identifying the biomarkers associated with new-onset glomerular filtration rate (GFR) decrease in an initially healthy population could offer a better understanding of kidney function decline and help improving patient management. Methods Here we described the proteomic and transcriptomic footprints associated with new-onset kidney function decline in an initially healthy and well-characterized population with a 20-year follow-up. This study was based on 1087 individuals from the familial longitudinal Suivi Temporaire Annuel Non-Invasif de la Santé des Lorrains Assurés Sociaux (STANISLAS) cohort who attended both visit 1 (from 1993 to 1995) and visit 4 (from 2011 to 2016). New-onset kidney function decline was approached both in quantitative (GFR slope for each individual) and qualitative (defined as a decrease in GFR of >15 ml/min/1.7 m2) ways. We analysed associations of 445 proteins measured both at visit 1 and visit 4 using Olink Proseek® panels and 119 765 genes expressions measured at visit 4 with GFR decline. Associations were assessed using multivariable models. The Bonferroni correction was applied. Results We found several proteins (including PLC, placental growth factor (PGF), members of the tumour necrosis factor receptor superfamily), genes (including CCL18, SESN3), and a newly discovered miRNA—mRNA pair (MIR1205–DNAJC6) to be independently associated with new-onset kidney function decline. Complex network analysis highlighted both extracellular matrix and cardiovascular remodelling (since visit 1) as well as inflammation (at visit 4) as key features of early GFR decrease. Conclusions These findings lay the foundation to further assess whether the proteins and genes herein identified may represent potential biomarkers or therapeutic targets to prevent renal function impairment.

AIMS:Patients experiencing ischaemic heart failure with reduced ejection fraction (HFrEF) represent a diverse group. We hypothesize that machine learning clustering can help separate distinctive patient phenotypes, paving the way for personalized management. METHODS AND RESULTS:A total of 8591 ischaemic HFrEF patients pooled from the EPHESUS and CAPRICORN trials (64 ± 12 years; 28% women) were included in this analysis. Clusters were identified using both clinical and biological variables. Association between clusters and the composite of (i) heart failure hospitalization or all-cause death, (ii) cardiovascular (CV) hospitalization or all-cause death, and (iii) major adverse CV events was assessed. The derived algorithm was applied in the COMMANDER-HF trial (n = 5022) for external validation. Five clinical distinctive clusters were identified: Cluster 1 (n = 2161) with the older patients, higher prevalence of atrial fibrillation and previous CV events; Cluster 2 (n = 1376) with the higher prevalence of older hypertensive women and smoking habit; Cluster 3 (n = 1157) with the higher prevalence of diabetes and peripheral artery disease; Cluster 4 (n = 2073) with relatively younger patients, mostly men and with the higher left ventricular ejection fraction; Cluster 5 (n = 1824) with the younger patients and lower CV events burden. Cluster membership was efficiently predicted by a random forest algorithm. Clusters were significantly associated with outcomes in derivation and validation datasets, with Cluster 1 having the highest risk, and Cluster 4 the lowest. Mineralocorticoid receptor antagonist benefit on CV hospitalization or all-cause death was magnified in clusters with the lowest risk of events (Clusters 2 and 4). CONCLUSIONS:Clustering reveals distinct risk subgroups in the heterogeneous array of ischaemic HFrEF patients. This classification, accessible online, could enhance future outcome predictions for ischaemic HFrEF cases.

AimsImpaired left ventricular–arterial coupling (VAC) has been shown to correlate with worse prognosis in cardiac diseases and heart failure (HF). The extent of the relationship between VAC and circulating biomarkers associated with HF has been scarcely documented. We aimed to explore associations of VAC with proteins involved in HF pathophysiology within a large population‐based cohort of middle‐aged individuals.Methods and resultsIn the forth visit of the STANISLAS family cohort, involving 1309 participants (mean age 48 ± 14 years; 48% male) from parent and children generations, we analysed the association of 32 HF‐related proteins with non‐invasively assessed VAC using pulse wave velocity (PWV)/global longitudinal strain (GLS) and arterial elastance (Ea)/ventricular end‐systolic elastance (Ees). Among the 32 tested proteins, fatty acid‐binding protein adipocyte 4, interleukin‐6, growth differentiation factor 15, matrix metalloproteinase (MMP)‐1, and MMP‐9 and adrenomedullin were positively associated with PWV/GLS whereas transforming growth factor beta receptor type 3, MMP‐2 and N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP) were negatively associated. In multivariable models, only MMP‐2 and NT‐proBNP were significantly and inversely associated with PWV/GLS in the whole population and in the parent generation. Higher levels of NT‐proBNP were also negatively associated with Ea/Ees in the whole cohort but this association did not persist in the parent subgroup.ConclusionElevated MMP‐2 and NT‐proBNP levels correlate with better VAC (lower PWV/GLS), possibly indicating a compensatory cardiovascular response to regulate left ventricular pressure amidst cardiac remodelling and overload.

Background: The associations between childhood adiposity and adult increased carotid intima-media thickness (cIMT) have been well established, which might be corroborated by the association between adiposity in children and inflammation in adults. However, longitudinal data regarding biological pathways associated with childhood adiposity are lacking. Methods: The current study included participants from the STANISLAS cohort who had adiposity measurements at age 5-18 years [ N = 519, mean (SD) age, 13.0 (2.9) years; 46.4% male], and who were measured with cIMT, vascular-related and metabolic-related proteins at a median follow-up of 19 +/- 2 years. BMI, waist-to-height ratio and waist circumference were converted to age-specific and sex-specific z -scores. Results: A minority of children were overweight/obese (16.2% overweight-BMI z -score >1; 1.3% obesity- z -score >2). Higher BMI, waist-height ratio and waist circumference in children were significantly associated with greater adult cIMT in univariable analysis, although not after adjusting for C-reactive protein. These associations were more pronounced in those with consistently high adiposity status from childhood to middle adulthood. Participants with higher adiposity during childhood (BMI or waist-height ratio) had higher levels of insulin-like growth factor-binding protein-1, protein-2, matrix metalloproteinase-3, osteopontin, hemoglobin and C-reactive protein in adulthood. Network analysis showed that IL-6, insulin-like growth factor-1 and fibronectin were the key proteins associated with childhood adiposity. Conclusion: In a population-based cohort followed for 20 years, higher BMI or waist-to-height ratio in childhood was significantly associated with greater cIMT and enhanced levels of proteins reflective of inflammation, supporting the importance of inflammation as progressive atherosclerosis in childhood adiposity.

Machine learning is now an essential part of any biomedical study but its integration into real effective Learning Health Systems, including the whole process of Knowledge Discovery from Data (KDD), is not yet realised. We propose an original extension of the KDD process model that involves an inductive database. We designed for the first time a generic model of Inductive Clinical DataBase (ICDB) aimed at hosting both patient data and learned models. We report experiments conducted on patient data in the frame of a project dedicated to fight heart failure. The results show how the ICDB approach allows to identify biomarker combinations, specific and predictive of heart fibrosis phenotype, that put forward hypotheses relative to underlying mechanisms. Two main scenarios were considered, a local-to-global KDD scenario and a trans-cohort alignment scenario. This promising proof of concept enables us to draw the contours of a next-generation Knowledge Discovery Environment (KDE).

Background and aims: Smoking may lead to premature ageing, but the impact on the cardiovascular system and circulating proteins needs further investigation. In the present study, we aim to understand the impact of smoking on heart and vessels and circulating biomarkers of multiple domains including cardiovascular damage, premature ageing and cancer-related pathways. Methods: The STANISLAS Cohort is a longitudinal familial cohort with detailed cardiovascular examination and biomarker assessment. This study included all the participants enrolled in the fourth visit of the STANISLAS Cohort for whom information on smoking habits was available (n = 1696). We assessed pulse wave velocity, intima-media thickness, echocardiographic parameters and a total of 460 proteins to study the association of circulating plasma proteins with smoking status (never vs. past vs. current smoking) while adjusting for potential confounders. Results: Current smokers were approximately 18 years younger but had higher left ventricular mass index (LVMi) and similar pulse wave velocity (PWV), carotid intima media thickness (cIMT), frequency of hypertension, diabetes and carotid plaques compared to the much older never smokers. After multivariate selection, 25 proteins were independently associated with current or past smoking. Current smoking was strongly associated with higher levels of EDIL-3, CCL11, TNFSF13B, KIT, and lower levels of IL-12B and PLTP (p < 0.0001) while past smoking was associated with FGF-21, CHIT1, and lower levels of CXCL10, IL1RL2 and RAGE (p < 0.01). Conclusions: Current smoking is associated with signs of early onset of cardiovascular ageing and protein biomarkers that regulate inflammation, endothelial function, metabolism, oncological processes and apoptosis.

Aims:End-stage renal disease (ESRD) treated by chronic hemodialysis (HD) is associated with poor cardiovascular (CV) outcomes, with no available evidence-based therapeutics. A multiplexed proteomic approach may identify new pathophysiological pathways associated with CV outcomes, potentially actionable for precision medicine.Methods and results:The AURORA trial was an international, multicentre, randomized, double-blind trial involving 2776 patients undergoing maintenance HD. Rosuvastatin vs. placebo had no significant effect on the composite primary endpoint of death from CV causes, nonfatal myocardial infarction or nonfatal stroke. We first compared CV risk-matched cases and controls (n = 410) to identify novel biomarkers using a multiplex proximity extension immunoassay (276 proteomic biomarkers assessed with OlinkTM). We replicated our findings in 200 unmatched cases and 200 controls. External validation was conducted from a multicentre real-life Danish cohort [Aarhus-Aalborg (AA), n = 331 patients] in which 92 OlinkTM biomarkers were assessed. In AURORA, only N-terminal pro-brain natriuretic peptide (NT-proBNP, positive association) and stem cell factor (SCF) (negative association) were found consistently associated with the trial's primary outcome across exploration and replication phases, independently from the baseline characteristics. Stem cell factor displayed a lower added predictive ability compared with NT-ProBNP. In the AA cohort, in multivariable analyses, BNP was found significantly associated with major CV events, while higher SCF was associated with less frequent CV deaths.Conclusions:Our findings suggest that NT-proBNP and SCF may help identify ESRD patients with respectively high and low CV risk, beyond classical clinical predictors and also point at novel pathways for prevention and treatment.

Patients with heart failure (HF) and coronary artery disease (CAD) have a high risk for cardiovascular (CV) events including HF hospitalization, stroke, myocardial infarction (MI) and sudden cardiac death (SCD). The present study evaluated associations of proteomic biomarkers with CV outcome in patients with CAD and HF with reduced ejection fraction (HFrEF), shortly after a worsening HF episode. We performed a case-control study within the COMMANDER HF international, double-blind, randomized placebo-controlled trial investigating the effects of the factor-Xa inhibitor rivaroxaban. Patients with the following first clinical events: HF hospitalization, SCD and the composite of MI or stroke were matched with corresponding controls for age, sex and study drug. Plasma concentrations of 276 proteins with known associations with CV and cardiometabolic mechanisms were analyzed. Results were corrected for multiple testing using false discovery rate (FDR). In 485 cases and 455 controls, 49 proteins were significantly associated with clinical events of which seven had an adjusted FDR < 0.001 (NT-proBNP, BNP, T-cell immunoglobulin and mucin domain containing 4 (TIMD4), fibroblast growth factor 23 (FGF-23), growth differentiation factor-15 (GDF-15), pulmonary surfactant-associated protein D (PSP-D) and Spondin-1 (SPON1)). No significant interactions were identified between the type of clinical event (MI/stroke, SCD or HFH) and specific biomarkers (all interaction FDR > 0.20). When adding the biomarkers significantly associated with the above outcome to a clinical model (including NT-proBNP), the C-index increase was 0.057 (0.033-0.082), p < 0.0001 and the net reclassification index was 54.9 (42.5 to 67.3), p < 0.0001. In patients with HFrEF and CAD following HF hospitalization, we found that NT-proBNP, BNP, TIMD4, FGF-23, GDF-15, PSP-D and SPON1, biomarkers broadly associated with inflammation and remodeling mechanistic pathways, were strong but indiscriminate predictors of a variety of individual CV events.

Cereal crops are frequently affected by toxigenic Fusarium species, among which the most common and worrying in Europe are Fusarium graminearum and Fusarium culmorum. These species are the causal agents of grain contamination with type B trichothecene (TCTB) mycotoxins. To help reduce the use of synthetic fungicides while guaranteeing low mycotoxin levels, there is an urgent need to develop new, efficient and environmentally-friendly plant protection solutions. Previously, F. graminearum proteins that could serve as putative targets to block the fungal spread and toxin production were identified and a virtual screening undertaken. Here, two selected compounds, M1 and M2, predicted, respectively, as the top compounds acting on the trichodiene synthase, a key enzyme of TCTB biosynthesis, and the 24-sterol-C-methyltransferase, a protein involved in ergosterol biosynthesis, were submitted for biological tests. Corroborating in silico predictions, M1 was shown to significantly inhibit TCTB yield by a panel of strains. Results were less obvious with M2 that induced only a slight reduction in fungal biomass. To go further, seven M1 analogs were assessed, which allowed evidencing of the physicochemical properties crucial for the anti-mycotoxin activity. Altogether, our results provide the first evidence of the promising potential of computational approaches to discover new anti-mycotoxin solutions.

The choice of the most appropriate unsupervised machine-learning method for “heterogeneous” or “mixed” data, i.e. with both continuous and categorical variables, can be challenging. Our aim was to examine the performance of various clustering strategies for mixed data using both simulated and real-life data. We conducted a benchmark analysis of “ready-to-use” tools in R comparing 4 model-based (Kamila algorithm, Latent Class Analysis, Latent Class Model [LCM] and Clustering by Mixture Modeling) and 5 distance/dissimilarity-based (Gower distance or Unsupervised Extra Trees dissimilarity followed by hierarchical clustering or Partitioning Around Medoids, K-prototypes) clustering methods. Clustering performances were assessed by Adjusted Rand Index (ARI) on 1000 generated virtual populations consisting of mixed variables using 7 scenarios with varying population sizes, number of clusters, number of continuous and categorical variables, proportions of relevant (non-noisy) variables and degree of variable relevance (low, mild, high). Clustering methods were then applied on the EPHESUS randomized clinical trial data (a heart failure trial evaluating the effect of eplerenone) allowing to illustrate the differences between different clustering techniques. The simulations revealed the dominance of K-prototypes, Kamila and LCM models over all other methods. Overall, methods using dissimilarity matrices in classical algorithms such as Partitioning Around Medoids and Hierarchical Clustering had a lower ARI compared to model-based methods in all scenarios. When applying clustering methods to a real-life clinical dataset, LCM showed promising results with regard to differences in (1) clinical profiles across clusters, (2) prognostic performance (highest C-index) and (3) identification of patient subgroups with substantial treatment benefit. The present findings suggest key differences in clustering performance between the tested algorithms (limited to tools readily available in R). In most of the tested scenarios, model-based methods (in particular the Kamila and LCM packages) and K-prototypes typically performed best in the setting of heterogeneous data.

Insulin-like Growth Factor Binding Protein 2 (IGFBP2) is a member of the IGFBP family which is present in the heart; cardiac IGFBP2 mRNA levels have furthermore been shown to rise in animal models of ischemia-induced heart failure (HF) [ [1] Berry M. Galinier M. Delmas C. et al. Proteomics analysis reveals IGFBP2 as a candidate diagnostic biomarker for heart failure. IJC Metabolic & Endocrine. 2015; 6: 5-12 Crossref Scopus (13) Google Scholar ]. The group of Dr. P. Rouet previously showed that IGFBP2 is a powerful diagnostic biomarker for HF whose accuracy for identifying acute HF is a valuable adjunct to brain natriuretic peptide (BNP) [ [1] Berry M. Galinier M. Delmas C. et al. Proteomics analysis reveals IGFBP2 as a candidate diagnostic biomarker for heart failure. IJC Metabolic & Endocrine. 2015; 6: 5-12 Crossref Scopus (13) Google Scholar ]: The area under the curve of the receiver operating characteristic (ROC) curve for HF was 0.908 (0.830–0.958) for IGFBP2 versus 0.857 (0.770–0.921) for BNP. In addition, IGFBP2 improved the model's accuracy in predicting HF as well as the net reclassification index (NRI) (0.574 (0.340–0.750) p < .001) on top of BNP. Insulin-like Growth Factor Binding Protein 2 predicts mortality risk in heart failureInternational Journal of CardiologyVol. 300PreviewInsulin-like Growth Factor Binding Protein 2 (IGFBP2) showed greater heart failure (HF) diagnostic accuracy than the "grey zone" B-type natriuretic peptides, and may have prognostic utility as well. Full-Text PDF