Prediction Of Outcome Research Articles

The 5-Item Modified Frailty Index (mFI-5) for risk stratification of patients with infective endocarditis undergoing cardiac surgery

Abstract Background population aging has led to an increased burden of frailty as an emerging global health issue, and variables for its quantification in patients with infective endocarditis (IE) are lacking in current surgical risks scores. The 5-item modified frailty index (mFI-5) score has shown to be effective in the prediction of surgical outcomes in other medical scenarios. Purpose this study aimed to assess the role of the mFI-5 in the prediction of surgical outcomes in patients with left-sided IE undergoing cardiac surgery. Methods a total of 112 cases of confirmed left-sided IE undergoing surgery, were consecutively collected and followed during one year after diagnosis in two tertiary hospitals between 2016 and 2021. The mFI-5 score included 5 baseline patient-specific comorbidities: diabetes, congestive heart failure, non-independent functional status (Barthel index <80), hypertension and chronic obstructive pulmonary disease (range 0-5), and patients were categorised as mFI-low (<2, n=65), and mFI-high (>=2, n=47). Results mean age was 66 [SD 15] years, 67% were men and community-acquired infections were 75%. As shown in Table 1, mFI-high patients were older and, aside from the specific comorbidities included in the score, with more chronic kidney disease. The most isolated pathogen was S. viridans (n=32), predominantly in mFI-low, in contrast to enterococcal and polymicrobial IEs, which were more frequent in mFI-high. Larger vegetations were reported in mFI-high [16 (10) vs 13 (9) mm, p 0.03], with no differences regarding infection site or periannular complications. Acute kidney injury, decompensated heart failure and higher natriuretic peptides were more common in mFI-high. Regarding mortality, in-hospital deaths were more frequent in mFI-high group. When analyzing mFI-score as a discrete variable, in-hospital mortality was associated with higher median scores values [2 (2) vs 1(1) points, p 0.02], being 2 points the optimal cutoff point for prediction [AUC 0.64 (95% CI 0.54-0.73), Fig.1A]. Among survivors who completed follow-up (n=82), rehospitalization and all-cause mortality rates in mFI-high exceeded those from mFI-low. Predictors of in-hospital and 1-year mortality were assessed by logistic regression (Fig. 1B), confirming the independent prognostic association of mFI-5-score in patients with IE undergoing surgery. Conclusions patients with IE undergoing cardiac surgery with higher mFI-5 scores were older, more predisposed to enterococcal and polymicrobial infections, and had a greater comorbidity burden with a more complicated clinical course. The mFI-score was independently associated with in-hospital mortality as well as rehospitalizations and 1-year mortality, positioning itself as an additional tool for risk stratification in this cluster of patients.Table 1Figure 1

Comparison of machine learning and traditional methods for prediction of adverse clinical events after percutaneous coronary intervention

Abstract Background Accurate prediction of clinical outcomes following percutaneous coronary intervention (PCI) is essential for mitigating risk and periprocedural planning. Traditional risk models have demonstrated modest predictive value. Machine learning (ML) models offer an alternative risk stratification that may provide improved predictive accuracy. We performed a systematic review and meta-analysis comparing ML models and traditional risk scores for prediction of clinical outcomes after PCI. Methods This study was reported according to PRISMA guidelines. PubMed, EMBASE, Web of Science and Cochrane databases were searched until 1st November, 2023 for studies comparing ML models with traditional statistical methods for event prediction after PCI. The primary outcome was comparative discrimination measured by C-statistics with 95% confidence intervals between ML models and traditional methods in estimating the risk of all-cause mortality, major bleeding and the composite outcome major adverse cardiovascular events (MACE). Results Thirty-four models were included across 13 observational studies (4105916 patients). The summary C-statistic all ML models across all clinical endpoints was 0.81 (95% CI, 0.78-0.85), compared to traditional methods 0.79 (95% CI, 0.74-0.85). The difference in C-statistic was 0.02 (95% CI, 0.002-0.04, p=0.56). Of included studies, only 1 model was externally validated and 10 models were calibrated. Conclusion ML models did not outperform traditional risk scores in the discrimination of all-cause mortality, major bleeding or MACE following PCI. While integration of ML algorithms into electronic healthcare systems may improve periprocedural risk stratification, immediate implementation in the clinical setting remains uncertain. Further research is required to overcome methodological and validation limitations.

Left atrial strain is associated with cardiovascular outcomes in children with congenitally corrected transposition of the great arteries

Abstract Background The assessment of the systemic right ventricular (sRV) function in congenitally corrected transposition of the great arteries (ccTGA) by echocardiography is restricted by its complex geometry. Purpose To determine the utility of left atrial strain (LAS) and sRV longitudinal strain (sRV-LS) in the prediction of outcomes in children with ccTGA. Methods This nested case-control study included all patients <21 years old with unrepaired ccTGA at our tertiary pediatric institution who had an echocardiogram prior to cardiovascular outcomes (milrinone-dependent heart failure, ventricular assistant device, heart transplant or death). LAS (both R wave and P wave references) and sRV-LS were measured using TomTec AutoStrain. To account for potential confounders, cases and controls were matched (1 cases vs 3 controls) by age ± 1 year and tricuspid regurgitation (TR) severity (none/mild vs. moderate/severe). Univariable conditional logistic regression was employed to assess the association between LAS and sRV-LS with outcomes among matched patients. Results Of 101 patients (64% male) with echocardiogram performed at 1.78 years old (IQR 0.02, 9.80), 14 (14%) developed an outcome. Baseline characteristics, LAS, and sRV-LS are depicted in Table 1. LAS and sRV-LS were not discriminative between age-matched cases and controls. In the matched cases and controls for TR severity, each unit-increase in LAS conduit phase (LAScd R and LAScd P) was associated with 18% and 22% higher odds of outcomes respectively, although these did not achieve statistical significance (LAScd R OR 1.2, 95% CI 1.0-1.4, p=0.09; LAScd P OR 1.2, 95% CI 1.0-1.5, p=0.08, Table 2). Conclusions After controlling for TR severity as a potential confounding factor, the conduit phase of LAS by both R wave and P wave references may have clinical implications as predictors of cardiovascular outcomes in patients with ccTGA. Larger and more well-powered studies are needed to further evaluate this association and studying longitudinal changes in LAS may provide insight in the development of outcomes in this population.

Role of urea to creatinine ratio in the prediction of outcomes in chronic heart failure

Abstract Background Renal function predicts events in chronic heart failure (HF). Urea to creatinine ratio (UCR) is a simple and cheap biomarker which correlates with neuroendocrine pathways. Objectives Evaluate the prognostic role of UCR in patients with chronic HF. Methodology 86 patients with chronic HF from an academic hospital were included (with reduced, mildly reduced or improved left ventricular ejection fraction). UCR was calculated at baseline, and patients were followed for a median of 360 (interquartile range 180-472) days. We plotted ROC curves for NT-proBNP and UCR and evaluated two scenarios using the Kaplan-Meier curve to determine the event-free survival. We grouped patients into four subgroups: a) UCR and NT-proBNP below the cut-off; b) UCR above and NT-proBNP below; c) UCR below and NT-proBNP above; and d) both above the cut-off. A second scenario was analysed, grouped as follows: a) UCR and NT-proBNP below; b) at least one of them above the cut-off; and c) both above the cutoff. The primary endpoint was cardiovascular death or HF hospitalisation. The level of significance was 5%. Results The mean age was 66.1±12.1y, and 60.5% were male. There were 28 events in the follow-up (32.5%). The ROC curve for UCR had an AUC of 0.75 (CI 0.64-0.86; p=0.0002) for a cut-off of 43.8 (sensitivity of 75% and specificity of 70.7%). The AUC for NT-proBNP was 0.64 (0.51-0.77; p=0.043) for a cut-off of 2010 pg/mL (sensitivity of 64.3% and specificity of 65.5%). In the Cox regression model, the UCR was the only independent predictor with HR of 1.06 per 1 unit increase (1.03-1.10; p=0.0002). UCR above cut-off (as a categorical variable) was the only independent predictor, with HR 5.34 (2.26-12.6; p=0.0001). Kaplan-Meier curves in Figures 1 and 2 show the UCR performance associated with NT-proBNP. Conclusions UCR is a simple and cheap biomarker that has been shown to have predictive power, probably due to persistent neuroendocrine activation in patients with low volume. In this population, elevated UCR added risk even in patients with low levels of NT-proBNP.

Elevated levels of plasma inactive stromal cell derived factor-1a predict long-term adverse outcomes in diabetic patients with coronary artery disease

Abstract Background Because of the significantly increased cardiovascular risk by complicating type 2 diabetes (T2D) in patients with coronary artery disease (CAD), the appropriate risk estimation is crucial in patients with T2D following percutaneous coronary intervention (PCI). Nevertheless, there is no established biomarker which can precisely predict outcomes in this particular population. Although stromal cell derived factor-1α (SDF-1α) has been suggested to play a key role in the pathogenesis of T2D, the prognostic impact of SDF-1α in the secondary prevention of CAD is yet to be elucidated. In particular, the ability of SDF-1α isoforms in outcome prediction in this population has not been assessed. Purpose The aim of present study was to investigate the prognostic implication of three isoforms of SDF-1α, such as total, active, and inactive forms of SDF-1α in diabetic patients following PCI. Methods This single-center retrospective analysis involved consecutive patients with T2D who underwent PCI for the first time between 2008 and 2018 (n=849). Primary and secondary outcome measures were all-cause death and the composite of cardiovascular death, non-fatal myocardial infarction, and ischemic stroke (3P-MACE), respectively. For determining plasma levels of SDF-1α, we measured not only total, but also the active type of SDF-1α by ELISA. Inactive isoform of the SDF-1α was calculated by subtracting the active isoform from total SDF-1α. Participants were divided into two groups according to the median of three forms of plasma SDF-1α levels (total SDF-1α: 2270 pg/mL, active SDF-1α: 686 pg/mL, inactive SDF-1α: 1537 pg/mL) at PCI procedure and the incidence and risk of subsequent endpoints following PCI were assessed. Results Kaplan-Meier analyses revealed increased risk of both all-cause death and 3P-MACE in patients with elevated levels of inactive SDF-1α (Figure 1). Constantly, multivariate Cox hazard analyses adjusted by age, sex, body mass index (>25), acute coronary syndrome, and chronic kidney disease (≥ stage 3) repeatedly indicated the 1 higher log-transformed inactive SDF-1α was significantly associated with increased risk of all-cause death (hazard ratio (HR): 2.21, 95% confidence interval (CI): 1.05–4.67, p=0.04) and 3P-MACE (HR: 2.30, 95% CI: 1.01–5.09, p=0.04). However, neither of total and active SDF-1α reached statistical significance (Figure 2). Furthermore, the addition of inactive SDF-1α to baseline models in multivariate Cox proportional hazard analysis significantly enhanced both the net reclassification improvement (NRI) and the integrated discrimination improvement (IDI) for all-cause death (NRI 0.29, p=0.009; IDI 0.007, p=0.03) and 3P-MACE (NRI 0.38, p=0.001; IDI 0.007, p=0.01) while adding total and active SDF-1α did not improve them. Conclusion In diabetic patients following PCI, elevated levels of plasma inactive SDF-1α, not total and active isoforms, might be a useful indicator of adverse outcomes.

Prognostic value of albumin-bilirubin score in patients with chronic heart failure

Abstract Background Liver dysfunction is one of the common comorbidities, and associated with worse clinical outcomes in patients with chronic heart failure (CHF). Recently, the albumin-bilirubin (ALBI) score has been developed as an effective and convenient scoring system for assessing liver function. However, the prognostic significance of the ALBI score in patients with CHF remained to be clarified. Purpose We sought to investigate the clinical significance of ALBI score on outcome prediction in patients with CHF. Methods We prospectively examined 1567 symptomatic CHF patients (median age 74, inter quartile range [IQR] 64-82 years, mean left ventricular ejection fraction [LVEF] 49 ± 15.9 %, median N-terminal pro-brain natriuretic peptide [NT-pro BNP] 801 [IQR 372-1820] pg/ml) in a multicentre registry with 27 Japanese sites between January 2020 and December 2023. The ALBI score was calculated according to the following formula: log10 total bilirubin (mg/dl) × 0.06 + albumin (g/dL) × 10 × (-0.085). Studied patients were divided into three groups according to the ALBI grade; ALBI score ≤ -2.60 (Grade1, n = 691), -2.60 < ALBI score ≤ -1.39 (Grade 2, n = 864), ALBI score > -1.39 (Grade3, n = 12). Primary outcome of interest was a composite of all-cause death and hospitalisation due to worsening heart failure. Results Patient with higher ALBI grade had higher age, New York Heart Association functional class, NT-pro BNP level, prevalence of atrial fibrillation and chronic obstructive pulmonary disease, and lower body mass index, systolic blood pressure, haemoglobin level, serum albumin level and estimated glomerular filtration rate compared to those with lower ALBI grade. LVEF was comparable between the groups. During a median follow-up period of 537 (IQR 285-701) days, the primary outcome occurred in 267 patients (17%). Higher ALBI grade were significantly related to increased incidence of the primary outcome (P for trend < 0.001) (Figure 1). Multivariable Cox regressions showed that a higher ALBI score was independently associated with higher risk of the primary outcome (hazard ratio 2.8, 95% confidential interval (CI) 2.08-3.77, P < 0.001), even after adjustment for the MAGGIC score as an established risk prediction model, NT-pro BNP, aspartate aminotransferase, and platelet counts. Furthermore, addition of the ALBI score to the MAGGIC score significantly increased c-index from 0.70 (95% CI 0.66-0.74) to 0.73 (95% CI 0.69-0.76) (P < 0.001) (Figure 2). The net reclassification improvement afforded by the ALBI score was 26% for the primary outcome (P<0.001). Conclusions Liver dysfunction assessed by the ALBI score was independently associated with worse clinical outcomes in patients with CHF. Moreover, addition of the score to the existing risk prediction model significantly increased the ability of outcome prediction, suggesting beneficial clinical application of the ALBI score in patients with CHF.

The role of fibrosis and inflammation for response and outcome prediction in candidates to cardiac resynchronization therapy: is response the right answer?

Abstract Background Cardiac resynchronization therapy (CRT) is an established treatment in selected patients suffering from heart failure with reduced ejection fraction (HFrEF). It has been proposed that myocardial fibrosis and inflammation could influence CRT "response" and outcome. Purpose Our study investigated the long-term prognostic significance of cardiac biomarkers in HFrEF patients with an indication for CRT. Methods 86 consecutive patients referred for CRT implantation were retrospectively evaluated. The soluble suppression of tumorigenicity 2 (sST2), galectin-3 (Gal-3), N-terminal portion of the B-type natriuretic peptide (NT-proBNP), and estimated glomerular filtration rate (eGFR) were measured at baseline and after 1 year of follow-up. An echocardiographic reduction of LV end-systolic volume ≥15% was used to define a patient as responder to CRT. Multivariate analyses were performed to evaluate their correlation with the primary composite outcome of cardiovascular mortality and heart failure hospitalizations at a mean follow-up of 9 ± 2 years. Results 44% of patients experienced the primary outcome. the mean baseline values of NT-proBNP, Gal-3, and sST2 were significantly higher compared with the patients without cardiovascular events. At the multivariate analyses, baseline Gal-3 [cut-off: 38.5 ng/ml, AUC: 0.63, p=0.03, (OR 7.13 [1.12;45.41], p=0.03), together with TAPSE>17.5 mm (OR 10.86 [3.15;37.44], p<0.001) significantly correlated with the structural response to CRT in several prediction models. Baseline Gal-3 [cut-off: 16.6 ng/ml, AUC: 0.91, p<0.001, HR 8.33 (1.88–33.33), p = 0.005] and sST2 [cut-off: 35.6 ng/ml AUC: 0.91, p<0.001, HR 333 (250–1,000), p = 0.003] significantly correlated with the composite outcome in the prediction models with high likelihood (Table). At Kaplan-Meyer curve patients with both high baseline sST2 and Gal-3 had the lowest survival probability (Figure). Among the parameters evaluated at 1-year follow-up, sST2, eGFR, and the variation from baseline to 1-year of Gal-3 levels showed a strong association with the primary outcome [HR 1.15 (1.08–1.22), p < 0.001; HR: 0.84 (0.74–0.91), p = 0.04; HR: 1.26 (1.10–1.43), p ≤ 0.001, respectively]. Conversely, the echocardiographic definition of CRT response did not correlate with any outcome. Conclusion In HFrEF patients with CRT, sST2, Gal-3, and renal function were associated with the combined endpoint of cardiovascular death and HF hospitalizations at long term follow-up, while the echocardiographic CRT response did not seem to influence the outcome of the patients.

Proteomics-based computational modelling of myocardial metabolism for personalized phenotyping and outcome prediction in advanced heart failure

Abstract Background Perturbations of myocardial metabolism and energy depletion are well-established hallmarks of heart failure (HF), yet methods for their systematic assessment remain limited in humans [1]. Computational modelling of the heart of individual patients represents an emerging precision medicine approach for personalized diagnosis of specific disease mechanisms and tailored clinical decision-making [2,3]. Purpose To determine the ability of computational modelling of patient-specific myocardial metabolism to assess individual bioenergetic phenotypes and their prognostic implications in HF. Methods Based on proteomics-derived enzyme intensity profiles in cardiac biopsies, we recently developed a computational model of myocardial metabolism capturing all pathways along energy-delivering substrates are processed to generate ATP [4]. Using this framework, we studied the individual energetic state in 17 sex-matched nonfailing controls and 48 patients with advanced HF (LVEF 20±10%, 72.9% non-ischemic aetiology, 52.1% females, 29.8% diabetics) undergoing heart transplantation (n=25) or ventricular assist device (VAD) implantation (n=23). The bioenergetic impact of alterations in substrate availability and myocardial workload were simulated, and the model’s ability to predict myocardial reverse remodelling after VAD implantation was assessed. Results Compared to controls, HF patients had a reduced ATP production capacity (p<0.01), although there was considerable interindividual variance, ranging from severely decreased to normal values. While overall oxygen consumption was lower in failing hearts, the amount of ATP generated per mole of oxygen did not differ (p=0.16). Maximum uptake capacity of fatty acids, glucose, and amino acids was comparable. By contrast, maximum uptake capacity of ketone bodies and lactate was reduced in the failing heart (p for both <0.05). There was a shift from fatty acid to glucose utilization in most HF patients, depending on substrate availability and myocardial workload. The ratio of fatty acid to glucose oxidation distinguished patients with LVEF recovery >10% after VAD implantation (C-index 0.93, p<0.01; Picture 1B) and correlated significantly with improvement in post-operative LVEF (r=0.67, p<0.01). ATP production capacity was not associated with reverse remodelling after VAD implantation. Conclusions Computational modelling identified a subset of advanced HF patients with preserved myocardial metabolism despite a similar degree of systolic dysfunction. Myocardial substrate preference but not ATP production capacity was associated with myocardial recovery after VAD implantation, which suggests a potential for substrate manipulation as a therapeutic approach in those with deranged fuel use. Computational assessment of myocardial metabolism in HF may improve understanding of disease heterogeneity, individual risk stratification, and guidance of personalized clinical decision-making in the future.

Combined evaluation of liver and kidney function improves prediction of clinical outcome in patients with chronic heart failure and reduced ejection fraction

Abstract Background/Introduction The liver and the kidneys are important organs for body homeostasis and are both affected by heart failure (HF) either by systemic venous congestion or hypoperfusion. Purpose We aimed to verify whether the Model for End-stage Liver Disease eXcluding International normalized ratio (MELD-XI), as well as a combination of cholestatic liver and kidney function tests carry prognostic value in patients with chronic HF with reduced ejection fraction (HFrEF). The liver parameters included alkaline phosphatase (ALP), gamma glutamyl-transpeptidase (GGT), and total bilirubin, while kidney parameters consisted of eGFR and the tubular damage markers urinary N-acetyl-beta-D-glucosaminidase (NAG) and kidney injury molecule (KIM)-1. Methods We categorized 250 patients with HFrEF on the basis of baseline biomarker levels. We assumed elevated NT-proBNP indicates hemodynamic congestion necessary to induce congestive liver or kidney dysfunction. Thus, in every organ damage category, NT-proBNP level above median was incorporated, next to any 2 out of 3: ALP, GGT or total bilirubin above median for liver dysfunction; any 2 out of 3 NAG or KIM-1 above median or eGFR below median for kidney dysfunction; or coexistence of liver and kidney dysfunction as defined above. Additionally, a subgroup of patients with both NT-proBNP and MELD-XI above the median was analyzed. The primary endpoint was a composite of cardiovascular death, heart transplantation, LVAD implantation, and hospitalization for acute or worsened HF, whichever occurred first. Univariable and multivariable Cox proportional hazard regression models were used. Results Median (Q1-Q3) age of the study population was 68 (58-76) years; 184 (74%) were men, 62 (25%) in NYHA class III or IV, and mean left ventricular ejection fraction (LVEF) was 31±9%. During a median (Q1-Q3) follow-up of 2.2 (1.4-2.5) years, 66 (26%) patients reached the primary endpoint. We observed that liver dysfunction, kidney dysfunction or their coexistence was associated with significantly higher risk of reaching the primary endpoint (HR 3.54, 95% CI 1.75–7.16; HR 1.85, 95% CI 1.00–3.41; HR 2.52, 95% CI 1.03–6.11 respectively) in the clinically adjusted model (Table 1). Higher MELD-XI was associated with higher risk of reaching the primary endpoint in an unadjusted model but not in a clinically adjusted model. Figure 1 presents event-free survival probability of patients falling into organ dysfunction categories. Conclusions Our results show that markers of cholestatic liver function and kidney function predict unfavorable clinical outcome and provide better prognostic value in chronic HFrEF patients than the MELD-XI alone. Thus, screening for liver and kidney damage may help identify high-risk patients in whom more careful observation and individualized therapy is warranted.Fig 1.Event-free survival probability

Pregnancy in women with congenital heart disease : new insights into neonatal risk predictioc

Abstract Background Advances in managing adult congenital heart disease (ACHD) have led to an increased number of women with CHD reaching childbearing age. This demographic shift underscores the need for improved understanding and prediction of complications during pregnancy in this specific ACHD population. Despite progress in maternal cardiac risk assessment, the prediction of neonatal outcomes for ACHD pregnancies remains underdeveloped. Therefore, the aims of this study are (1) to assess neonatal outcomes in a CHD women population, (2) to identify their predictive factors and (3) to propose a new risk score for predicting neonatal complications. Methods This registry study included all women born between 1975 and 1996 diagnosed with ACHD who underwent at least one cardiology consultation for ACHD in Cliniques Universitaires Saint-Luc. A multivariate analysis was performed to identify predictors of neonatal complications and these were incorporated into a new risk index. Its validity was assessed using bootstrap method. This score was then compared with scores adapted from the ZAHARA and CARPREG studies for offspring events prediction. Results Analysis of 491 pregnancies revealed 31.4% of neonatal complications. Four significant predictors of adverse neonatal outcomes were identified: cardiac treatment during pregnancy (OR 14.9, 95%CI [4.6-64.3]), hypertensive disorders of pregnancy (OR 10.4, 95%CI [3.4-39.5]), smoking during pregnancy (OR 10.3, 95% CI [3.0-44.9]), and pre-pregnancy BMI < 18.5 kg/m² (OR 6.2, 95% CI [2.6-16.5]). The risk model demonstrated an AUC of 0.70 (95% CI [0.65-0.75), which remained stable after bootstrap validation. This model significantly outperformed the scores adapted from ZAHARA and CARPREG data. Based on the regression coefficients, a risk score was subsequently developed comprising 5 risk categories. Conclusion One third of ACHD pregnancies are complicated by poor neonatal outcome. These complications are determined by 4 independent factors relating to the cardiac and non-cardiac status of the patients, which have been incorporated into a risk score. Our study is one of the first to propose a predictive risk score of neonatal outcomes in ACHD pregancies, and paves the way for other validation and confirmation studies.Offspring scores: performance comparisonNew risk score for neonatal outcome

Advancements in colorectal cancer detection: The role of immuno‐positron emission tomography, immuno‐single‐photon emission computed tomography, and machine learning applications

AbstractColorectal cancer (CRC) ranks as the world's second most prevalent cancer and third in mortality. Detection and diagnosis are crucial in research and clinical settings. While colonoscopy and computed tomographic colonography are widely used for identifying organic lesions, positron emission tomography (PET) and single‐photon emission computed tomography (SPECT) offer superior visualization of molecular changes. These immuno‐PET and immuno‐SPECT techniques surpass conventional [18F] Fluorodeoxyglucose PET/CT in specificity and sensitivity, improving CRC diagnostics and supporting therapeutic strategies. This review emphasizes the role of immuno‐PET/SPECT in CRC diagnosis and establishing a foundation for therapeutic strategies, facilitating hierarchical management through the identification of treatment‐responsive populations, prediction of therapeutic outcomes, and support for intraoperative imaging. This review introduces the preclinical and clinical utility of immunoconjugates for detecting colorectal adenomas, and primary, metastatic, or recurrent CRC, focusing on specific CRC cell targets like the epidermal growth factor receptor and carcinoembryonic antigen. The review also covers various mAb‐based immunoconjugates and engineered mAb fragments, including diabodies and minibodies. Finally, it looks into the great promise of machine learning in PET or SPECT and it addresses the challenges of translating preclinical successes into clinical practice for colorectal adenoma diagnosis, proposing potential solutions and directions for future research.

Characterisation of plasma proteins and circulating microRNAs in out-of-hospital cardiac arrest patients - approaches for predicting outcome

Abstract Rationale: Early prognostication after out-of-hospital cardiac arrest (OHCA) is an unmet clinical need. There is limited data on serial release kinetics of cardiac, brain and systemic biomarkers such as Neuron Specific Enolase (NSE) and S-100B. miRNAs are sensitive biomarker candidates in acute myocardial infarction (MI). Objective The objective was to characterise circulating biomolecules in the early release kinetics after OHCA, compared with a STEMI control group and assessed with clinical outcome. Methods and Results Serial blood samples of patients with OHCA with STEMI (n=20) and STEMI control (n=20) were collected at baseline and at 6, 12, 24 and 48h after admission. Untargeted plasma proteomics were performed using data-independent acquisition–mass spectrometry (DIA–MS). Candidate proteins and miRNAs were quantified. Proteomics returned n=156 proteins with significantly different kinetics between OHCA and STEMI. Six clusters, depicting distinct biological pathways, with characteristic differences between OHCA and STEMI were identified. The most distinct and significant differences between OHCA and STEMI patients were seen for acute phase reactants LBP, AACT, CRP and the lipid metabolite A2GL as well as endothelial and platelet miR-126 and cardiomyocyte-apoptosis/ischemia-reperfusion related miR-320a. Cardiogenic shock was best characterised by differential kinetics of coagulation and complement related proteins FA9, HRG, FHR1 and inflammation related miR-146. The primary endpoint was poor neurological recovery at 30-days. Individuals with good outcome showed distinct differences in adaptive immune response, complement and coagulation cascades and hemostasis. Circulating plasma levels of hypoxia-associated miR-101, miR-210 and S-100B protein were significantly elevated at baseline in OHCA patients. These molecules further showed higher baseline values in patients with poor outcome. For current prognosis marker NSE, this difference only develops after 48h. Cardiac/muscle miRNAs followed an early release-pattern similar to high-sensitivity cardiac troponin-T (hs-cTnT). Machine learning models were used to calculate the AUC for the prediction of poor neurological outcome at baseline. This was highest for hypoxia-induced miR-210 (AUC: 0.854), which performed significantly better than the protein biomarkers NSE (AUC: 0.646) and S-100B (AUC:0.656). Conclusions Untargeted mass spectrometry reveals distinct biological pathways and candidate biomolecules affected in patients suffering from OHCA compared with STEMI controls, in patients developing cardiogenic shock and in those with adverse versus good neurological outcome. Platelet-, inflammation and hypoxia associated miRNAs are associated with OHCA, cardiogenic shock and prognostication of adverse neurological outcome. Pronounced differences in early release kinetics of RNA and protein biomarkers may improve identification of OHCA patients at risk of poor outcome.

A flexible, calibrated transformer-based risk model for prediction of clinical outcomes in patients with heart failure; an open cohort study using linked UK EHR

Abstract Background Numerous risk scores, both uni- and multi-variable, have been developed to conduct prognostication for a range of clinical outcomes in heart failure (HF) patients [1,2]. However, these models provide only modest discrimination and often depend on complex test data (e.g., ejection fraction) collected outside of routine clinical care [2]. Furthermore, transparency in reporting of key individual-level metrics (e.g., precision, recall) is notably lacking. Purpose We developed and validated a novel deep learning model, the second version of the Transformer-based Risk assessment survival model (TRisk2), for 36-month prediction of all-cause mortality, cardiovascular-related mortality, fatal and non-fatal cardiovascular events, and renal outcomes in patients with HF using routine, linked UK electronic health records (EHR) data. Methods An open cohort of 405 thousand patients with HF between 40 and 90 years of age was identified using linked EHR from 1,063 and 355 English general practices, which were used for TRisk2 model development and external validation, respectively. Comparison was conducted against MAGGIC [2], modified for use on routine linked EHR. Additional analyses compared discriminatory performance in other age groups, by sex, and by baseline disease. All analyses were repeated for prediction of cardiovascular-related mortality, fatal and non-fatal cardiovascular events, and renal outcomes. Results TRisk2 demonstrated superior discrimination with Concordance index (C-index) of 0.828; 95% confidence interval (CI): 0.824 to 0.832 for all-cause mortality prediction outperforming MAGGIC. The proposed model’s performance was found to be stable across stratified analyses by sex, age, and baseline disease status. Both models were overall well-calibrated, and TRisk2 demonstrated greater net benefit than MAGGIC across reasonable decision boundaries in decision curve analyses. At exemplar risk threshold of 50% for 36-month prediction, TRisk2 outperformed MAGGIC by 0.08 for both, precision and recall and identified 10% more events. At the same threshold for 12-month prediction, TRisk2 outperformed MAGGIC by 0.14 and 0.26 precision and recall respectively and captured 70% more events. The well-calibrated TRisk2 similarly outperformed MAGGIC in prediction of cardiovascular-related mortality, fatal and non-fatal cardiovascular events, and renal outcomes. Conclusion Utilising solely routine EHR from primary and secondary care, TRisk2 enabled more accurate prediction than MAGGIC with 10-20% gain in C-index across the four outcome investigations. Integration of TRisk2 into routine clinical care could simultaneously eliminate the reliance of complex tests and improve prognostication of key clinical outcomes, thereby refining management of HF.Models' discrimination on four outcomes

Classification of 90-day mortality risk among patients with cardiogenic shock requiring advanced mechanical circulatory support

Abstract Background The SCAI shock classification aims to provide a simple and practical stratification of the severity of patients with cardiogenic shock (CS) and to immediately offer them crucial treatments such as mechanical circulatory support (MCS). Additionally, the IABP shock II risk score is designed to predict 30-day mortality in patients with CS following acute coronary syndrome (ACS), and the SAVE score can also predict in-hospital mortality for patients after veno-arterial extracorporeal membrane oxygenation (V-A ECMO) insertion due to refractory CS. However, previous studies on SCAI staging in patients with MCS have been inconsistent, and the association with prognosis remains unclear. Purpose The aim of this study is to investigate the correlation between the SCAI shock stage and prognosis in ACS patients requiring advanced MCS, and to compare it with other prognostic scores. Methods From January 2012 to July 2023, 137 ACS patients with advanced MCS (excluding IABP) were evaluated. Upon ICU admission, patients were categorized into SCAI stages B to E. The IABP shock II score and the SAVE score were also calculated to serve as prognostic tools. The type of MCS was categorized as Impella, VA-ECMO with IABP, and ECPELLA (VA-ECMO with Impella). We compared the 90-day survival rates with the SCAI shock stage and other prognostic tools. Result The distribution for SCAI classifications stage B to E was 13 (10%), 38 (28%), 21 (15%), and 65 (47%), respectively. There were no significant differences among the groups regarding age, gender, comorbidities, the ratio of ST-elevation myocardial infarction, door-to-balloon time, and onset-to-reperfusion time. In SCAI Stage E, a significantly higher proportion of patients (86%) underwent extracorporeal cardiopulmonary resuscitation, exhibiting elevated lactate and peak CK-MB levels, along with higher IABP Shock II and lower SAVE scores, compared to other stages. A Kaplan-Meier analysis of 90-day mortality, based on the SCAI, the IABP Shock II, the SAVE categories, and the type of MCS, demonstrated that outcomes deteriorated as the stage advanced across all classifications. In the univariate Cox regression analysis for 90-day mortality, significant differences in outcomes were observed across the four classifications. However, in the multivariate Cox regression analysis that incorporated all four classifications, only the type of MCS (Impella vs. ECMO with IABP; hazard ratio [HR] 0.36; 95% confidence interval [CI], 0.16 - 0.78, p = 0.012) and the SCAI classification (stage E vs. stage C; HR 3.03; 95% CI, 1.51 - 6.08, p = 0.002) demonstrated a significant association with 90-day mortality. Conclusion The SCAI shock stage upon ICU admission suggests the potential for convenient prediction of mid-term outcomes in ACS patients who received advanced MCS.

Low chronotropic response in post-myocardial infarction exercise test predicts worse prognosis in patients with preserved or mildly reduced left ventricular ejection fraction

Abstract Background Chronotropic incompetence is common in post-myocardial infarction (MI) patients and is associated with reduced exercise capacity. However, its prognostic significance and a determination of threshold values for prognosis remain unclear. Methods & Results We assessed 96 post-MI patients 4 weeks after their initial event with left ventricular ejection fraction (LVEF) ≥ 40%. All patients underwent combined stress echocardiography and cardiopulmonary exercise tests. The chronotropic response was measured as a percent of the maximal predicted heart rate at peak exercise (%MPHR). Exclusion criteria included patients with exercise limitations for pulmonary or peripheral reasons and those who declined to complete the study. Follow-up was conducted in 86 patients. The primary outcome measured was a composite of all-cause mortality or hospitalization for ischemic heart disease events (acute MI, unstable angina, or revascularization procedures) or heart failure exacerbation. The median follow-up duration was 4.0 (IQR 2, 5.6) years. The median patient age was 60 (IQR 53, 65) years, 64% of patients were males, and 84% were on beta-blockers. Median LVEF was 57 (IQR 51, 62)%, and median peak VO2 was 19 (IQR 15, 22) mL/kg/min. There were 15 composite end-point events, including 3 deaths, during the follow-up period. The %MPHR threshold of 67% had the best predictive value with an area under the curve (AUC) of 77% for predicting the primary outcome (PPV 31%, NPV 97%, sensitivity 93%, specificity 55%). A significant difference (p = 0.0022) in the occurrence of the composite end-point was noted between patients with a %MPHR below versus above this threshold (Figure. Kaplan-Meier curves for survival free of primary outcome). In the multivariate Cox regression analysis adjusting for demographic, clinical variables, and beta-blocker doses, only chronotropic response and current smoking correlated with the primary end-point (Table. The Cox proportional hazards regression model for prediction of the primary outcome). Conclusions A low chronotropic response to exercise is a negative prognostic marker in post-myocardial infarction patients with either preserved or mildly reduced left ventricular ejection fraction.Figure.Table.

Multi-modal artificial intelligence-based prediction for cardiovascular mortality after transcatheter aortic valve implantation: a time-to-event survival prediction

Abstract Background Predicting long-term outcomes following Transcatheter Aortic Valve Implantation (TAVI) presents a significant challenge. Objective We aimed to predict post-TAVI long-term cardiovascular mortality using multimodal data and employing time-to-event artificial intelligence (AI) algorithms. Methods This study enrolled 3,973 consecutive patients from the prospective TAVI registry (between August 2007 and June 2023). Multimodal information was collected and extracted as numerical data for each patient, encompassing clinical history, medication, laboratory, procedural characteristics, ECG, angiography, echocardiography (TTE and TEE), and CT. Clinical outcomes were adjudicated by an independent clinical event committee. From the multimodal information, 147 features were extracted at baseline before the TAVI procedure. The datasets were divided into a training/validation set (80%) and a hold-out test set (20%) for model development and evaluation, respectively. Cox-LASSO feature selection was employed to identify the most relevant features, followed by the implementation of Random Survival Forest (RSF), Gradient Boosting Survival Analysis (GBSA), and Survival Support Vector Machine (SSVM) to predict the timing and occurrence of cardiac death events in patients. Feature selection, model parameters, and hyperparameters selection were performed using 10-fold cross-validation on the training set. Model performance was evaluated using different metrics, such as the concordance index (c-index) and cumulative Area Under the Curve (C-AUC) on a 20% untouched test set. Results In the patient cohort, 28.4% experienced cardiovascular death. Features were selected from various modalities, including 22 from clinical history, two from medication, eight from laboratory tests, one from ECG, eight from CT scans, and five from echocardiography and angiography. In the univariate Cox-PH model, all features exhibited a C-index ranging from 0.50 to 0.57. In the test set, the conventional Multivariate Cox-PH model achieves a c-index of 0.59 (C-AUC: 0.56). RSF, GBSA, and SSVM achieve a C-index of 0.70 (C-AUC: 0.69), 0.66 (C-AUC: 0.64), and 0.69 (C-AUC: 0.68), respectively, in the holdout test set. Conclusion Integrating multimodal information, encompassing imaging, signal, laboratory data, and clinical history, with AI algorithms has markedly enhanced the prediction of long-term cardiac outcomes following the TAVI procedure. The initially developed model can predict the probability of cardiac death up to 10 years. Furthermore, the RFS model significantly outperformed the conventional Cox model in predicting time-to-event outcomes.

Evaluation of the Effectiveness and Safety of Mesenchymal Stem Cell Treatment in Fistulising Crohn's Disease: An International Real-Life Retrospective Multicentre Cohort Study.

Perianal fistulas of Crohn's disease (CD) create a significant burden on patients' lives. However, the efficacy and safety of adipose-derived mesenchymal stem cell treatment are contradicting, and real-world evidence is lacking. To examine the usability of darvadstrocel therapy in managing perianal CD. We enrolled patients with CD and perianal fistulas in this retrospective multicenter study. The primary outcome was perianal clinical remission (defined as all treated fistulas closed) at weeks 26 and 52. Secondary outcomes were clinical response rates (≥ 1 fistulas closed), perianal activity (PDAI), patient satisfaction, and adverse events. Data were recorded at baseline and weeks 12, 26 and 52. Prediction of primary outcomes was performed by logistic regression. Overall, among 223 patients (male/female ratio: 0.48), perianal clinical remission was achieved in 78.2% and 62.3% until weeks 26 and 52. Baseline PDAI score (OR 0.75), number of fistulas (OR 0.28) and the number of weeks after preparation for surgery (OR 0.98) were associated with treatment failure. The clinical response rates were 84.8% and 79.8% at weeks 26 and 52. Improvement of subjective perianal symptoms was achieved in 77.8% and 78.4% of patients, respectively. Adverse events occurred in 13.5% of patients; perianal abscesses and proctalgia were the most frequently reported. Effectiveness data were higher than in clinical trials. The safety profile was reassuring, and patients' satisfaction was high. Appropriate patient selection, fistula preparation and expertise may help to achieve treatment success.

Prediction of cardiovascular outcomes by subtle changes in the peripheral immune cell landscape - insights from the LURIC single cell RNA-sequencing study (LuRNA)

Abstract Background and Aims Atherosclerosis and its clinical sequelae, myocardial infarction and stroke, represent the main causes of death worldwide. Although preclinical evidence has suggested the existence of an inflammatory response driving disease, the extend of cellular alterations in human atherosclerosis remains enigmatic. Here, we employ single cell RNA-sequencing (scRNA-seq) on peripheral blood mononucleated cells (PBMCs) in a well-defined cardiovascular risk cohort from the "Ludwigshafen Risk and Cardiovascular Health (LURIC) trial" to define changes in the immune cell landscape in atherosclerotic patients. Methods and Results Of 3317 patients enrolled in the LURIC trial between 1997 and 2000, we selected individuals with stable coronary-artery disease (CAD) (n=31) and healthy patients (no CAD) (n=16) by propensity score matching, adjusted for CRP, NTproBNP, Troponin T, LDL-C, and Cystatin-C, in a case-control design. scRNA-seq was performed by droplet sequencing on PBMCs stored in liquid nitrogen. Individual data sets were integrated and changes between healthy individuals and patients with CAD were evaluated on numeric and transcriptional level. Single cell transcriptomes were annotated using an established CITE-seq reference atlas. While bulk RNA-sequencing of PBMC preparations revealed only minor changes between both conditions, we identified several leukocyte populations on a single cell level with specific transcriptomes that were differentially regulated between both conditions. Interestingly, we found that several subsets of Natural Killer (NK) cells, T cells with an enrichment of proliferation-associated pathways, and a population resembling hematopoietic stem cells were more frequent in patients with CAD, while naïve phenotypes of B and T cells were overrepresented in the absence of CAD. Consistently, most T and B cell populations in patients with CAD were enriched in pathways associated with cell activation, immune receptor signalling, and differentiation. In addition, we detected restricted repertoires of T cell receptor sequences in several adaptive immune cell subtypes from patients with CAD, suggestive of increased cellular clonality and antigen-specificity. Finally, in a comparison of cellular frequencies with cardiovascular outcomes over more than 15 years, several immune cell subsets and cell-type specific transcriptional programs predicted cardiovascular death in this case-control scenario in multi-variate adjusted regression analysis. Notably, addition of scRNA-seq derived parameters was superior to traditional risk prediction with clinical characteristics and soluble biomarkers alone. Conclusion Employing scRNAseq, we demonstrate that atherosclerosis is associated with a profound change in the circulating immune cell landscape even in the absence of measurable differences in inflammatory biomarkers. These findings propose the usage of scRNAseq for the discovery of outcome-relevant cellular biomarkers in the future.

Prediction of Adverse Maternal and Fetal Outcome in Preeclampsia Using a Risk Prediction Model Prospective Cohort Study

Prediction of Adverse Maternal and Fetal Outcome in Preeclampsia Using a Risk Prediction Model Prospective Cohort Study

Metabolomic profiling of the entire UK Biobank enables interpretation of familial hypercholesterolemia mutations and prediction of severe cardiovascular outcomes

Abstract Background Familial hypercholesterolemia (FH) is an inherited disorder resulting in high levels of LDL-C from birth. Premeature coronary heart disease occurs in half of affected men by age 50 years(1). While many genetic variants that cause FH are well understood, there are also many variants in FH genes that are of uncertain clinical significance. Furthermore, the variability in residual cardiovascular risk of FH carriers after statin treatment is not fully understood. We here investigate whether metabolomic and genomic data from the 500,000 person UK Biobank cohort can aid in interpretation of candidate pathogenic variants for FH, and characterise future risk of severe cardiac events in the carriers of such variants. Methods We have quantified 249 metabolomic parameters in 490,830 blood samples from UK Biobank cohort. Among 186,483 participants with exome sequence data we identified 896 who carried known pathogenic FH variants and 405 who carried variants of unknown significance in an established FH gene. We fitted a logistic predictive model to identify pathogenic mutations using LDL levels, statin use and 17 principal components of the metabolomic parameters. We further tested whether metabolomic and genomic scores for predicting future risk of myocardial infarction (MI), that were trained in the general population (i.e. individuals without FH), could predict risk in FH carriers. Results A model including all metabolites had a higher accuracy at identifying known pathogenic FH variants (36/56 at FDR < 0.05), compared to LDL and statin use alone (30/56). Our model also identified 8 FH variants marked as uncertain significance in ClinVar with significant (FDR < 0.05) evidence of predicted pathogenicity. We found that our metabolomic risk score predicted 10-year risk of MI among FH carriers (HR=1.96 per SD, p = 0.0011). This was indeed stronger than the effect in the general population (HR = 1.64 per SD, p = 2.5e-168). A combined metabolomic and polygenic risk score had an even larger effect at predicting MI in FH carriers (HR = 3.06 per SD, p = 2.0e-6), and this score was significantly better at predicting MI in FH carriers than in non-carriers (HR = 3.06 vs 1.88 per SD, interaction p = 0.042). Conclusion Familial hypercholesterolemia is an important source of risk for early cardiovascular disease. Population-scale metabolomics can help distinguish pathogenic from benign variants that have not yet been characterised. Metabolomic and genomic scores can also identify individuals at most elevated risk, and provide even more information to FH individuals than in the general population.