Search

Algorithms for automated analysis of intravascular ultrasound (IVUS) images can be disturbed by guidewires, which are often encountered when treating bifurcations in percutaneous coronary interventions. Detecting guidewires in advance can therefore help avoiding potential errors. This task is not trivial, since guidewires appear rather small compared to other relevant objects in IVUS images. We employed CNNs with additional multi-task learning as well as different guidewire-specific regularizations to enable and improve guidewire detection. In this context, we developed a network block which generates heatmaps that highlight guidewires without the need of localization annotations. The guidewire detection results reach values of 0.931 in terms of the F1-score and 0.996 in terms of area under curve (AUC). Comparing thresholded guidewire heatmaps with ground truth segmentation masks leads to a Dice score of 23.1 % and an average Hausdorff distance of 1.45 mm. Guidewire detection has proven to be a task that CNNs can handle quite well. Employing multi-task learning and guidewire-specific regularizations further improve detection results and enable generation of heatmaps that indicate the position of guidewires without actual labels.

Knowing the shape of vascular calcifications is crucial for appropriate planning and conductance of percutaneous coronary interventions. The clinical workflow can therefore benefit from automatic segmentation of calcified plaques in intravascular ultrasound (IVUS) images. To solve segmentation problems with convolutional neural networks (CNNs), large datasets are usually required. However, datasets are often rather small in the medical domain. Hence, developing and investigating methods for increasing CNN performance on small datasets can help on the way towards clinically relevant results. We compared two state-of-the-art CNN architectures for segmentation, U-Net and DeepLabV3, and investigated how incorporating auxiliary image data with vessel wall and lumen annotations improves the calcium segmentation performance by using these either for pretraining or multi-task training. DeepLabV3 outperforms U-Net with up to 6.3 % by means of the Dice coefficient and 36.5 % by means of the average Hausdorff distance. Using auxiliary data improves the segmentation performance in both cases, whereas the multi-task approach outperforms the pre-training approach. The improvements of the multi-task approach in contrast to not using auxiliary data at all is 5.7 % for the Dice coefficient and 42.9 % for the average Hausdorff distance. Automatic segmentation of calcified plaques in IVUS images is a demanding task due to their relatively small size compared to the image dimensions and due to visual ambiguities with other image structures. We showed that this problem can generally be tackled by CNNs. Furthermore, we were able to improve the performance by a multi-task learning approach with auxiliary segmentation data.

Background and Aims Recent investigations have suggested an interdependence of lipoprotein(a) [Lp(a)]-related risk for cardiovascular disease with background inflammatory burden. The aim the present analysis was to investigate whether high-sensitive C-reactive protein (hsCRP) modulates the association between Lp(a) and coronary heart disease (CHD) in the general population. Methods Data from 71 678 participants from 8 European prospective population-based cohort studies were used (65 661 without/6017 with established CHD at baseline; median follow-up 9.8/13.8 years, respectively). Fine and Gray competing risk-adjusted models were calculated according to accompanying hsCRP concentration (<2 and ≥2 mg/L). Results Among CHD-free individuals, increased Lp(a) levels were associated with incident CHD irrespective of hsCRP concentration: fully adjusted sub-distribution hazard ratios [sHRs (95% confidence interval)] for the highest vs. lowest fifth of Lp(a) distribution were 1.45 (1.23–1.72) and 1.48 (1.23–1.78) for a hsCRP group of <2 and ≥2 mg/L, respectively, with no interaction found between these two biomarkers on CHD risk (P interaction = 0.82). In those with established CHD, similar associations were seen only among individuals with hsCRP ≥ 2 mg/L [1.34 (1.03–1.76)], whereas among participants with a hsCRP concentration <2 mg/L, there was no clear association between Lp(a) and future CHD events [1.29 (0.98–1.71)] (highest vs. lowest fifth, fully adjusted models; P interaction = 0.024). Conclusions While among CHD-free individuals Lp(a) was significantly associated with incident CHD regardless of hsCRP, in participants with CHD at baseline, Lp(a) was related to recurrent CHD events only in those with residual inflammatory risk. These findings might guide adequate selection of high-risk patients for forthcoming Lp(a)-targeting compounds. [ABSTRACT FROM AUTHOR]

Background Stroke after a cardiovascular procedure (CVP) is a devastating complication adversely affecting outcome. Mechanical thrombectomy (MT) has not been investigated systematically in this population. Objective To carry out a retrospective study in patients undergoing MT for early stroke after CVP, aiming to further characterize this cohort of patients, and to evaluate the efficacy, safety, procedural characteristics, and outcome of MT. Methods A single-center stroke registry of patients who received MT was analyzed. Baseline and procedural parameters, mortality, functional outcome, recanalization rates, and complications were evaluated. Propensity score matching was carried out, identifying a control cohort with non-periprocedural large vessel occlusion (LVO). Results Overall 913 patients were included (mean age 73.0 (±13.0) years, 52.5% female, median National Institutes of Health Stroke Scale score 15 (10-19)). Eleven patients with a LVO after a recent (<30 days postoperatively) CVP were identified (n=3 transcatheter aortic valve and n=1 surgical aortic valve replacements (SAVR), n=3 coronary bypass grafting (CABG) surgeries, n=2 SAVR+CABG, and n=2 aortic surgeries). After matching, 8 patients in the CVP group were compared with 16 patients in the matched cohort. Comparable rates of reperfusion were achieved. Time from symptom onset to groin puncture (171.5 min (136.3, 178.3) vs 284.0 min (215.0, 490.5); p=0.039), as well as recanalization (195.0 min (146.0, 201.0) vs 419.0 min (274.0, 613.0); p=0.028) was faster in the CVP group. However, this was not reflected by an improved outcome (modified Rankin Scale score after 90 days: 5.5 (3.3, 6.0) vs 5.0 (4.0, 6.0), mortality after 90 days 50.0% vs 37.5%). Complications did not differ between the groups. Conclusions Use of MT for LVO stroke in patients after a recent CVP is a safe and efficient treatment in comparison with patients with a non-periprocedural LVO undergoing MT. [ABSTRACT FROM AUTHOR]

Background: The relevance of blood lipid concentrations to long-term incidence of cardiovascular disease and the relevance of lipid-lowering therapy for cardiovascular disease outcomes is unclear. We investigated the cardiovascular disease risk associated with the full spectrum of bloodstream non-HDL cholesterol concentrations. We also created an easy-to-use tool to estimate the long-term probabilities for a cardiovascular disease event associated with non-HDL cholesterol and modelled its risk reduction by lipid-lowering treatment.Methods: In this risk-evaluation and risk-modelling study, we used Multinational Cardiovascular Risk Consortium data from 19 countries across Europe, Australia, and North America. Individuals without prevalent cardiovascular disease at baseline and with robust available data on cardiovascular disease outcomes were included. The primary composite endpoint of atherosclerotic cardiovascular disease was defined as the occurrence of the coronary heart disease event or ischaemic stroke. Sex-specific multivariable analyses were computed using non-HDL cholesterol categories according to the European guideline thresholds, adjusted for age, sex, cohort, and classical modifiable cardiovascular risk factors. In a derivation and validation design, we created a tool to estimate the probabilities of a cardiovascular disease event by the age of 75 years, dependent on age, sex, and risk factors, and the associated modelled risk reduction, assuming a 50% reduction of non-HDL cholesterol.Findings: Of the 524 444 individuals in the 44 cohorts in the Consortium database, we identified 398 846 individuals belonging to 38 cohorts (184 055 [48·7%] women; median age 51·0 years [IQR 40·7-59·7]). 199 415 individuals were included in the derivation cohort (91 786 [48·4%] women) and 199 431 (92 269 [49·1%] women) in the validation cohort. During a maximum follow-up of 43·6 years (median 13·5 years, IQR 7·0-20·1), 54 542 cardiovascular endpoints occurred. Incidence curve analyses showed progressively higher 30-year cardiovascular disease event-rates for increasing non-HDL cholesterol categories (from 7·7% for non-HDL cholesterol <2·6 mmol/L to 33·7% for ≥5·7 mmol/L in women and from 12·8% to 43·6% in men; p<0·0001). Multivariable adjusted Cox models with non-HDL cholesterol lower than 2·6 mmol/L as reference showed an increase in the association between non-HDL cholesterol concentration and cardiovascular disease for both sexes (from hazard ratio 1·1, 95% CI 1·0-1·3 for non-HDL cholesterol 2·6 to <3·7 mmol/L to 1·9, 1·6-2·2 for ≥5·7 mmol/L in women and from 1·1, 1·0-1·3 to 2·3, 2·0-2·5 in men). The derived tool allowed the estimation of cardiovascular disease event probabilities specific for non-HDL cholesterol with high comparability between the derivation and validation cohorts as reflected by smooth calibration curves analyses and a root mean square error lower than 1% for the estimated probabilities of cardiovascular disease. A 50% reduction of non-HDL cholesterol concentrations was associated with reduced risk of a cardiovascular disease event by the age of 75 years, and this risk reduction was greater the earlier cholesterol concentrations were reduced.Interpretation: Non-HDL cholesterol concentrations in blood are strongly associated with long-term risk of atherosclerotic cardiovascular disease. We provide a simple tool for individual long-term risk assessment and the potential benefit of early lipid-lowering intervention. These data could be useful for physician-patient communication about primary prevention strategies.Funding: EU Framework Programme, UK Medical Research Council, and German Centre for Cardiovascular Research. [ABSTRACT FROM AUTHOR]

Aims: As promising compounds to lower Lipoprotein(a) (Lp(a)) are emerging, the need for a precise characterization and comparability of the Lp(a)-associated cardiovascular risk is increasing. Therefore, we aimed to evaluate the distribution of Lp(a) concentrations across the European population, to characterize the association with cardiovascular outcomes and to provide high comparability of the Lp(a)-associated cardiovascular risk by use of centrally determined Lp(a) concentrations. Methods and results: Based on the Biomarkers for Cardiovascular Risk Assessment in Europe (BiomarCaRE)-project, we analysed data of 56 804 participants from 7 prospective population-based cohorts across Europe with a maximum follow-up of 24 years. All Lp(a) measurements were performed in the central BiomarCaRE laboratory (Biokit Quantia Lp(a)-Test; Abbott Diagnostics). The three endpoints considered were incident major coronary events (MCE), incident cardiovascular disease (CVD) events, and total mortality. We found lower Lp(a) levels in Northern European cohorts (median 4.9mg/dL) compared to central (median 7.9mg/dL) and Southern European cohorts (10.9 mg/dL) (Jonckheere-Terpstra test P< 0.001). Kaplan-Meier curves showed the highest event rate of MCE and CVD events for Lp(a) levels≥90th percentile (log-rank test: P< 0.001 for MCE and CVD). Cox regression models adjusted for age, sex, and cardiovascular risk factors revealed a significant association of Lp(a) levels with MCE and CVD with a hazard ratio (HR) of 1.30 for MCE [95% confidence interval (CI) 1.15-1.46] and of 1.25 for CVD (95% CI 1.12-1.39) for Lp(a) levels in the 67-89th percentile and a HR of 1.49 for MCE (95% CI 1.29-1.73) and of 1.44 for CVD (95% CI 1.25-1.65) for Lp(a) levels>≥90th percentile vs. Lp(a) levels in the lowest third (P< 0.001 for all). There was no significant association between Lp(a) levels and total mortality. Subgroup analysis for a continuous version of cube root transformed Lp(a) identified the highest Lp(a)-associated risk in individuals with diabetes [HR for MCE 1.31 (95% CI 1.15-1.50)] and for CVD 1.22 (95% CI 1.08-1.38) compared to those without diabetes [HR for MCE 1.15 (95% CI 1.08-1.21; HR for CVD 1.13 (1.07-1.19)] while no difference of the Lp(a)-associated risk were seen for other cardiovascular high risk states. The addition of Lp(a) levels to a prognostic model for MCE and CVD revealed only a marginal but significant C-index discrimination measure increase (0.001 for MCE and CVD; P < 0.05) and net reclassification improvement (0.010 for MCE and 0.011 for CVD). Conclusion: In this large dataset on harmonized Lp(a) determination, we observed regional differences within the European population. Elevated Lp(a) was robustly associated with an increased risk for MCE and CVD in particular among individuals with diabetes. These results may lead to better identification of target populations who might benefit from future Lp(a)-lowering therapies. [ABSTRACT FROM AUTHOR]

Conventional low-density lipoprotein cholesterol (LDL-C) quantification includes cholesterol attributable to lipoprotein(a) (Lp(a)-C) due to their overlapping densities. The purposes of this study were to compare the association between LDL-C and LDL-C corrected for Lp(a)-C (LDL Lp(a)corr) with incident coronary heart disease (CHD) in the general population and to investigate whether concomitant Lp(a) values influence the association of LDL-C or apolipoprotein B (apoB) with coronary events. Among 68,748 CHD-free subjects at baseline LDL Lp(a)corr was calculated as "LDL-C—Lp(a)-C," where Lp(a)-C was 30% or 17.3% of total Lp(a) mass. Fine and Gray competing risk-adjusted models were applied for the association between the outcome incident CHD and: 1) LDL-C and LDL Lp(a)corr in the total sample; and 2) LDL-C and apoB after stratification by Lp(a) mass (≥/<90th percentile). Similar risk estimates for incident CHD were found for LDL-C and LDL-C Lp(a) corr30 or LDL-C Lp(a) corr17.3 (subdistribution HR with 95% CI) were 2.73 (95% CI: 2.34-3.20) vs 2.51 (95% CI: 2.15-2.93) vs 2.64 (95% CI: 2.26-3.10), respectively (top vs bottom fifth; fully adjusted models). Categorization by Lp(a) mass resulted in higher subdistribution HRs for uncorrected LDL-C and incident CHD at Lp(a) ≥90th percentile (4.38 [95% CI: 2.08-9.22]) vs 2.60 [95% CI: 2.21-3.07]) at Lp(a) <90th percentile (top vs bottom fifth; P interaction 0.39). In contrast, apoB risk estimates were lower in subjects with higher Lp(a) mass (2.43 [95% CI: 1.34-4.40]) than in Lp(a) <90th percentile (3.34 [95% CI: 2.78-4.01]) (P interaction 0.49). Correction of LDL-C for its Lp(a)-C content provided no meaningful information on CHD-risk estimation at the population level. Simple categorization of Lp(a) mass (≥/<90th percentile) influenced the association between LDL-C or apoB with future CHD mostly at higher Lp(a) levels. [ABSTRACT FROM AUTHOR]

Cardiac allograft vasculopathy (CAV) remains a major long-term complication in heart transplant (HT) recipients related to increased mortality. We aimed to identify non-immune recipient- and donor-related risk factors for the development of CAV in HT patients. 40,647 recipients, prospectively enrolled from April 1995 to January 2019 in the Organ Procurement and Transplantation Network (OPTN), were analyzed after exclusion of pediatric patients, those with missing information on CAV, and re-transplantation. Multivariable-adjusted Cox regression analyses were performed to identify recipient- and donor-related risk factors for CAV. 5-year population attributable risk for classical cardiovascular risk factors was calculated to estimate the recipients' CAV risk. Analyses were based on OPTN data (June 30, 2019). Of 40,647 post-transplant patients, 14,698 (36.2%) developed CAV with a higher incidence in males (37.3%) than in females (32.6%) (p < 0.001). The mean follow-up time was 68.2 months. In recipients, male sex, African American and Asian ethnicity, ischemic cardiomyopathy, body mass index and smoking were associated with CAV occurrence. In donors, older age, male sex, smoking, diabetes and arterial hypertension were related to CAV. Results remained fairly stable after analysis of different time periods. 5-year attributable CAV risk for classical cardiovascular risk factors was 9.1%. In this large registry with known limitations concerning data completeness, CAV incidence was higher in males than in females. Next to male sex and donor age, the classical cardiovascular risk factors were related to incident CAV. Classical cardiovascular risk factors played only a minor role for the 5-year attributable CAV risk. • In this large-scale registry study with more than 40,000 patients, CAV incidence was higher in males than in females. • Next to male sex, older donor age and ischemic cardiomyopathy, the classical cardiovascular risk factors were related to new-onset CAV. • Classical cardiovascular risk factors of the recipient combined, which are potentially modifiable in HT recipient's aftercare, accounted only for about 9% of attributable 5-year CAV risk. [ABSTRACT FROM AUTHOR]

Background: Acute ischaemic stroke (AIS) after percutaneous coronary intervention (PCI) is a rare, but debilitating, complication. However, contemporary data from real-world unselected patients are scarce., Aims: We aimed to explore the temporal trends, outcomes and variables associated with AIS as well as in-hospital all-cause mortality in a nationwide cohort., Methods: A retrospective analysis of healthcare records from 2006-2021 was implemented. Patients were stratified according to the occurrence of AIS in the setting of PCI. The temporal trends of AIS were analysed. A stepwise regression model was used to identify variables associated with AIS and in-hospital all-cause mortality., Results: A total of 4,910,430 PCIs were included for the current analysis. AIS occurred in 4,098 cases (0.08%). An incremental increase in the incidence of AIS after PCI from 0.03% to 0.14% per year was observed from 2006-2021. The strongest associations with AIS after PCI included carotid artery disease, medical history of stroke, atrial fibrillation, presentation with an ST-segment elevation myocardial infarction (STEMI) or non-STEMI and coronary thrombectomy. For patients with AIS, a higher in-hospital all-cause mortality (18.11% vs 3.29%; p<0.001) was documented. With regard to all-cause mortality, the strongest correlations in the stroke cohort were found for cardiogenic shock, dialysis and clinical presentation with a STEMI., Conclusions: In an unselected nationwide cohort of patients hospitalised for PCI, a gradual increase in AIS incidence was noted. We identified several variables associated with AIS as well as with in-hospital mortality. Hereby, clinicians might identify the patient population at risk for a peri-interventional AIS as well as those at risk for an adverse in-hospital outcome after PCI.

Background: Robotic-assisted percutaneous coronary intervention (rPCI) has proven to be feasible and safe. Comparative analyses of rPCI versus manual PCI (mPCI) are scarce., Aims: We aimed to investigate procedural aspects and outcomes of rPCI using the second-generation CorPath GRX Vascular Robotic System compared with mPCI in patients with chronic coronary syndrome and non-ST-segment elevation myocardial infarction acute coronary syndrome., Methods: From January to April 2021, 70 patients underwent rPCI at the University Heart & Vascular Center Hamburg-Eppendorf and were recruited into the INTERCATH study. By propensity score matching, a control cohort of 210 patients who underwent mPCI from 2015-2021 was identified. Co-primary endpoints were one-year all-cause mortality and major adverse cardiovascular events (MACE) as a composite of cardiovascular death, unplanned target lesion revascularisation, myocardial infarction, and stroke., Results: The median age of the patients (n=280) was 70.7 (25th percentile-75th percentile: 62.0-78.0) years, and 24.6% were female. The Gensini score (28.5 [16.2-48.1] vs 28.0 [15.5-47.0]; p=0.78) was comparable between rPCI versus mPCI. During the PCI procedure, total contrast fluid volume did not differ, whilst longer fluoroscopy times (20.4 min [13.8-27.2] vs 14.4 min [10.4-24.3]; p=0.001) were documented in the rPCI versus mPCI cohort. After 12 months of follow-up, neither all-cause mortality (p=0.22) nor MACE (p=0.25) differed between the groups., Conclusions: rPCI was associated with longer fluoroscopy times compared with mPCI, though without increased use of contrast medium. One-year follow-up revealed no differences in all-cause mortality or MACE, supporting the safety of a robotic-assisted approach.

Background: The recently updated European Society of Cardiology (ESC) dyslipidaemia guidelines recommend a lower low-density lipoprotein cholesterol (LDL-C) goal of <55 mg/dL for patients with atherosclerotic cardiovascular disease (ASCVD), with a concomitant Class IA upgrade for proprotein convertase subtilisin/kexin type 9 inhibitors (PCSK9i) for patients not reaching their LDL-C goal under conventional lipid-lowering therapy., Aims: We aim to quantify the need for PCSK9i and the related costs to achieve the revised LDL-C goal in ASCVD patients compared to former ESC recommendations, in particular the risk-based 2017 ESC consensus update., Methods and Results: We included patients with ASCVD from an observational cohort study ongoing since 2015. A Monte Carlo simulation incorporating a treatment algorithm adding sequentially a statin, ezetimibe, and a PCSK9i was applied with consideration of partial and total statin intolerance. The need for PCSK9i was calculated for three different ESC recommendations (2019 guidelines, 2016 guidelines, 2017 consensus update). Preventable events and treatment costs due to PCSK9i were calculated for a range of annual event rates from 2% to 8% and annual treatment costs of ca. 6050 €. We included 1780 patients (mean age 69.5 years). Median LDL-C at baseline was 85.0 mg/dL, with 61% of patients taking lipid-lowering medication. The need for PCSK9i was simulated to be 42.0% (ESC 2019), 31.9% (ESC 2016), and 5.0% (ESC 2017). The LDL-C goals were achieved in 97.9%, 99.1%, and 60.9% of patients, respectively. Annual treatment cost for PCSK9i per 1 000 000 ASCVD patients would be 2.54 billion € (ESC 2019) compared to 0.30 billion € (ESC 2017). Costs per prevented event due to PCSK9i initiation differed widely, e.g. 887 000 € for an event rate of 3% and a treatment goal of <55 mg/dL compared to 205 000 € for an event rate of 7% and risk-based use of PCSK9i., Conclusion: The revised LDL-C treatment goals increase the projected need for PCSK9i with a substantial increase in associated treatment cost. An allocation strategy based on residual LDL-C and clinical or angiographic risk factors leads to a more tailored target population for PCSK9i with a reasonable benefit/cost ratio., (Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2020. For permissions, please email: journals.permissions@oup.com.)

Coronary artery disease (CAD) has a complex etiology involving numerous environmental and genetic factors of disease risk. To date, the genetic 9p21 locus represents the most robust genetic finding for prevalent and incident CAD. However, limited information is available on the genetic background of the severity and distribution of CAD. CAD manifests itself as stable CAD or acute coronary syndrome. The Gensini score quantifies the extent CAD but requires coronary angiography. Here, we aimed to identify novel genetic variants associated with Gensini score severity and distribution of CAD. A two-stage approach including a discovery and a replication stage was used to assess genetic variants. In the discovery phase, a meta-analysis of genome-wide association data of 4,930 CAD-subjects assessed by the Gensini score was performed. Selected single nucleotide polymorphisms (SNPs) were replicated in 2,283 CAD-subjects by de novo genotyping. We identified genetic loci located on chromosome 2 and 9 to be associated with Gensini score severity and distribution of CAD in the discovery stage. Although the loci on chromosome 2 could not be replicated in the second stage, the known CAD-locus on chromosome 9p21, represented by rs133349, was identified and, thus, was confirmed as risk locus for CAD severity.

Background: The determination of cardiac troponin is essential for diagnosing myocardial infarction. A troponin I assay has recently been developed that provides the highest analytical sensitivity to date., Methods: The analysis included 1560 patients with chest pain, of whom 1098 were diagnosed with non-coronary chest pain, 189 with unstable angina pectoris and 273 with non-ST-segment elevation myocardial infarction. The troponin I concentration was determined on admission (0 hours) and 3 hours later. The diagnostic algorithm incorporated troponin I elevation above the gender-specific 99th percentile as well as predefined relative or absolute 3-hour changes in the troponin I concentration (delta)., Results: The diagnostic criterion of troponin I above the 99th percentile resulted in a negative predictive value of 98.0% and 98.2% in men and women, respectively. For rule-in of non-ST-segment elevation myocardial infarction, the use of absolute deltas yielded higher positive predictive values and sensitivities compared to relative deltas. With detection rates of about 85% and 82% in men and women, respectively, non-ST-segment elevation myocardial infarction was diagnosed with a positive predictive value close to 84% in men and 80% in women., Conclusions: The investigational troponin I assay provides an excellent non-ST-segment elevation myocardial infarction rule out. With gender-specific differences, the application of absolute changes in troponin concentration was superior to relative changes to rule in patients with non-ST-segment elevation myocardial infarction.