Gerhard Hindricks

Introduction: Timely diagnosis of transthyretin amyloid cardiomyopathy (ATTR-CM) is critical to improving outcomes and remains challenging. Guidelines recommend assessing for red flags (RFs) to facilitate ATTR-CM detection, but their diagnostic validation remains incomplete. This study aimed to multidimensionally validate these RFs in patients with left ventricular thickening (LVT). Methods: Prospective registry study including all patients with LVT and standardized workup at tertiary amyloidosis center. For validated RFs, “high-risk” (specificity for ATTR-CM>90%) and “low-risk” (false-negative rate<5%) findings were predefined. Results: 425 patients (321 ATTR-CM,104 control group) were included. Heart failure (97%) and polyneuropathies (78%) were among most sensitive clinical findings, while bilateral carpal tunnel syndrome (bCTS) was more specific (80%) but less sensitive (45%). Diverse ECG changes were common, yet mostly unspecific. Marked elevations of NT-proBNP and troponin were most discriminative among non-imaging RFs (Youden’s index[YI]:59%). Apical sparing (Apical-to-Basal-strain-ratio≥2.0) was the most effective, stage-independent echocardiographic parameter (YI62%), followed by biatrial enlargement, advanced diastolic dysfunction, and LVGLS impairment (YI48-55%). Normal ECG without conduction disorders history and normal levels of both NT-proBNP and hs-troponin-T constituted “low-risk” findings. Heart failure with concomitant bCTS, peripheral low voltage, right ventricular dysfunction, 5-5-5 sign, restrictive filling pattern and apical sparing with LVGLS impairment represented “high-risk” RFs. Contrary to most non-imaging RFs, most echocardiographic RFs showed association with advancing disease stage (p<0.05). All aforementioned RFs were associated with ATTR-CM independently of age, sex, and LVT-severity (p<0.05). Conclusions: ATTR-CM is frequently associated with validated RFs that differ in their diagnostic characteristics, which should be considered when evaluating for ATTR-CM.

Rationale: Atrial fibrillation (AF) is sustained by electrical triggers and atrial substrate, but whether AF itself generates circulating factors that acutely reinforce arrhythmogenesis is unknown. Objective: To determine whether rhythm-dependent serum bioactivity directly destabilizes atrial electrophysiology and to identify associated cellular mechanisms and circulating mediators. Methods and Results: Cardiac samples were obtained from patients with persistent atrial fibrillation (AF) during AF and after restoration of sinus rhythm (SR) using electrical cardioversion. In isolated Langendorff-perfused mouse hearts, AF serum increased the duration, frequency, and cumulative burden of spontaneous atrial depolarizations compared with paired control perfusate. In human induced pluripotent stem cell derived atrial cardiomyocytes, AF patient serum decreased Ca²⁺-transient amplitudes without significant effect on kinetics or Ca²⁺ spark activity. Moreover, AF serum reduced peak sodium current without materially altering voltage-dependent gating or paced action-potential properties. Data-independent acquisition proteomics identified rhythm-associated changes in extracellular-matrix, proteolytic, inflammatory, and signaling proteins, including MMP9, Ccathepsin G, and 14-3-3 family members. Targeted pathway perturbation differentially modified serum-induced Ca²⁺ phenotypes, indicating context-dependent, multifactorial bioactivity. Serum-induced Ca²⁺ responses were not associated with indices of atrial disease severity, including low-voltage substrate burden and left atrial pressure. Conclusions: Persistent AF is accompanied by rhythm-dependent circulating bioactivity that acutely promotes atrial electrical instability and Ca²⁺-dependent trigger formation. These findings extend the concept that “AF begets AF” beyond tissue-intrinsic remodeling and identify a potentially reversible serum-mediated feed-forward mechanism linking rhythm status and circulating proteomic signals in an atrial substrate severity independent fashion.

Background: Immunosuppressive therapy (IST) is commonly used in cardiac sarcoidosis, but the strategy to guide it is unclear and its effectiveness uncertain. We assessed the association between IST, guided by clinical course and positron emission tomography, and clinical outcomes in cardiac sarcoidosis. Methods: We compared 83 patients with cardiac sarcoidosis who received IST with 20 who declined therapy. IST was escalated for persistent symptoms and inflammation. Ventricular arrhythmia recurrence, ventricular pacing, New York Heart Association class, and left ventricular (LV) ejection fraction were assessed after 12 months. Death, LV assist device, and heart transplantation were assessed during a median follow‐up of 24.9 months. Results: Sustained ventricular tachycardia recurred in 12% of patients with IST versus 20% of those without IST ( P =0.264). Within the IST group, New York Heart Association class and LV ejection fraction improved during 12 months ( P =0.011 and P =0.002), but the between‐group difference in New York Heart Association change was not significant ( P =0.490). Major adverse cardiovascular events occurred less frequently with IST (10.8% versus 40.0%, P =0.004). After adjustment for baseline LV ejection fraction, IST had a lower hazard of major adverse cardiovascular events (hazard ratio, 0.30 [95% CI, 0.11–0.80]; P =0.017), consistent among sensitivity analyses. Given only 17 events and baseline imbalance, residual confounding cannot be excluded and this is hypothesis‐generating. Conclusions: In this observational registry, glucocorticoid‐based IST was associated with lower rates of death, heart transplantation, and LV assist device implantation. No significant between‐group difference was detected in sustained ventricular tachycardia recurrence, although the study was underpowered to exclude smaller differences. Because treatment was not randomized, these associations cannot establish causality.

Background/Objectives: Heart failure with preserved ejection fraction (HFpEF) is a heterogeneous syndrome. Atrial remodeling represents an important dimension of HFpEF beyond the presence of atrial fibrillation (AF). We aimed to identify clinically meaningful atrial phenotypes using unsupervised clustering. Methods: Overall, 153 patients were analyzed from a prospective HFpEF registry who underwent clinical assessment, laboratory testing, electrocardiography, resting and stress echocardiography and cardiopulmonary exercise testing. Consensus k-means clustering across multiply imputed datasets integrated markers of atrial structure and function, biomarkers, and hemodynamic burden. Clinical characteristics, functional capacity, quality of life and 12-month healthcare utilization were compared between clusters. Sensitivity analyses assessed cluster robustness. Results: Two phenotypes were identified. Cluster 1 (n = 66) was characterized by higher biomarker and hemodynamic burden, impaired renal function, larger left atrial volume index, and reduced left atrial reservoir strain. Variables not included in cluster generation also differed between phenotypes: Cluster 1 showed poorer exercise capacity, shorter six-minute walking distance, worse quality of life, and a higher observed all-cause hospitalization burden. The hospitalization finding was considered exploratory. The principal cluster structure was preserved in sensitivity analyses, including complete-case analysis. Conclusions: The clustering approach identified an HFpEF phenotype characterized by biomarker elevation and advanced atrial remodeling. The phenotype was associated with greater functional impairment and adverse patient-reported characteristics. These clinical associations support the potential relevance of multidimensional atrial phenotyping, although the findings remain hypothesis-generating and require validation in independent cohorts.

Aims: Artificial intelligence (AI) is increasingly being applied in healthcare, including cardiology, with growing interest in its potential to support diagnosis, prognosis, and clinical decision-making. The rapid pace of technological advancements is not matched by equivalent progress in the regulation and appraisal of these systems. This review aims to provide a structured evidence map of AI models in cardiology, by grading the robustness of clinical validation and mapping it against the quantity of publications across main subspecialties. Methods and results: To assess the strength of evidence, we defined, developed, and applied a novel classification tailored for AI models that ranks evidence into five Grades of Validation according to the level of clinical validity. We screened literature from 2014 to 2026 according to the PRISMA guidelines, identifying a total of 2807 articles on AI applications within cardiology, 490 of which met the inclusion criteria (validation was performed in at least one external, independent patient population), revealing a highly heterogeneous distribution of evidence across subspecialties. Most studies in AI focused on improved diagnostics (n = 322, 65.7%); however, advanced integration in prognostic (n = 122, 24.9%) and therapeutic applications remains limited (n = 46, 9.4%). Cardiovascular imaging leads AI applications, especially in coronary computed tomography and echocardiography, followed by non-invasive electrophysiology, predominantly in atrial fibrillation screening and ventricular dysfunction detection. Conclusion: The scarcity of prospective validations across different resources and geographical settings (n = 134, 27.3%) and randomized controlled trials (n = 35, 3.8%) limits the adoption by clinical guidelines and widespread implementation of AI algorithms.

Background: Electronic health records are a central component of digital health strategies worldwide. Early implementation phases are particularly critical for shaping long-term adoption patterns but remain insufficiently studied, especially in large-scale, policy-driven rollouts. In January 2025, Germany introduced the electronic patient record ( elektronische Patientenakte , ePA) as a nationwide opt-out system, creating a unique opportunity to examine early implementation in routine outpatient care. Objective: This study aimed to explore facilitators and barriers influencing the early implementation of the ePA in German outpatient practices from the perspective of health care professionals. Methods: A qualitative study based on semistructured interviews with general practitioners, medical assistants, and physician assistants was conducted between August and December 2025. Participants were recruited nationwide through regional physician associations and a professional organization using a criterion-based purposive recruitment strategy with maximum variation elements. Interviews were analyzed using qualitative content analysis guided by the Consolidated Framework for Implementation Research. Coding combined deductive framework-based and inductive thematic approaches. Results: A total of 49 interviews were conducted. Facilitators and barriers were identified across all 5 Consolidated Framework for Implementation Research domains, with the largest number of categories in the inner setting. Organizational factors such as leadership engagement, internal communication structures, and learning climate were central facilitators of implementation. In contrast, barriers were primarily related to technical instability, heterogeneous functionality of practice management systems, limited interoperability across care sectors, and insufficient patient awareness. Although the ePA was generally perceived as intuitive and potentially beneficial, particularly for improving information availability and reducing redundant diagnostics, its routine use remained strongly dependent on organizational capacity, infrastructure stability, and cross-sectoral integration. Conclusions: Early implementation of the nationwide opt-out ePA in Germany is shaped less by professional resistance than by organizational, technical, and systemic conditions. These findings highlight that successful large-scale digital health implementation requires coordinated strategies addressing technical infrastructure, organizational readiness, governance, and patient engagement beyond the mere deployment of digital systems.

Background and Aims: Randomized trials conducted in the early 2000s established the survival benefit of primary prevention implantable cardioverter-defibrillator (ICD) therapy in patients with reduced left ventricular ejection fraction (LVEF) after myocardial infarction. However, management of myocardial infarction and heart failure has substantially evolved since that time. We investigated whether the estimated association between primary prevention ICD implantation in post-myocardial infarction patients with reduced LVEF and mortality reduction has changed over time. Methods: We analysed individual participant data from 32 214 patients with LVEF ≤35% after myocardial infarction included in the PROFID pooled cohort, comprising 7477 patients carrying a primary prevention ICD (ICD patients) and 24 737 patients without an ICD (non-ICD patients). The primary endpoint was all-cause mortality. Propensity scores were estimated using multivariable logistic regression including age, sex, LVEF, renal function, and diabetes, and overlap weighting was applied to balance treatment groups. Time period-specific analyses were performed across three prespecified time periods defined by inclusion year: 1995–2004, 2005–14, and 2015–20. Weighted cumulative mortality curves were generated for each time period. Temporal changes in the estimated association between ICD implantation and mortality reduction were assessed using a weighted Cox proportional hazards model. Results: A total of 12 097 deaths occurred during a mean follow-up of 43.7 months. The estimated association between ICD implantation and mortality changed significantly across time (P for interaction <.001). In weighted time period-specific analyses, the estimated mortality reduction associated with ICD implantation progressively decreased over more recent periods. The hazard ratio for ICD vs non-ICD patients was 0.54 (95% CI 0.47–0.62; P < .001) in 1995–2004, 0.67 (95% CI 0.62–0.72; P < .001) in 2005–14, and 0.89 (95% CI 0.73–1.07; P = .221) in 2015–20, with negligible separation of the weighted cumulative mortality curves in the most recent time period. Conclusions: In this analysis including a large cohort of post-myocardial infarction patients with reduced LVEF, the estimated mortality reduction associated with primary prevention ICD implantation progressively decreased over time.

Aims: Risk scores are essential to evidence-based cardiovascular care, but manual calculation is labour intensive and error prone. Large language models (LLMs) could automate this process, yet LLMs are limited by their propensity for calculation errors and factual hallucinations. Pipelines separating LLM-based data extraction from deterministic score computation may improve reliability and transparency. Methods and results: We conducted a retrospective diagnostic study at a quaternary heart centre in Germany (January 2020 to July 2023). Patients with atrial fibrillation (n = 179) from an ablation registry and patients with severe aortic stenosis (n = 76) evaluated by a heart team were included. Six LLMs (GPT-5.2, Gemini 3.1 Pro, DeepSeek-R1, Qwen3, GPT-OSS 120B, and Kimi K2.5) were tested in standalone, retrieval-augmented generation (RAG), and pipeline configurations to compute HAS-BLED, CHA2DS2-VASc, and EuroSCORE II scores from routine clinical reports. Accuracy was assessed against expert-adjudicated ground truth using root mean squared error (RMSE) and Krippendorff’s α to evaluate numerical deviation and categorical agreement, respectively. Pipeline-generated scores showed substantially higher agreement with expert adjudication than standalone LLMs, LLMs with RAG, and treating physicians (mean Krippendorff’s α: 0.78 vs. 0.32 vs. 0.39 vs. 0.31) and lower deviation from ground truth (mean RMSE: 0.89 vs. 5.81 vs. 1.85 vs. 1.34). Conclusion: Pipelines combining expert-curated knowledge injection, LLM-based clinical data extraction, and deterministic score calculation enable accurate and scalable cardiovascular risk score computation from unstructured real-world clinical data, outperforming physician-documented scores. Such pipelines could form the basis for clinical decision-support systems that automate routine risk assessment, reduce clinician workload, and promote more consistent evidence-based care.

Aims: Risk stratification for appropriate implantable cardioverter-defibrillator (ICD) therapy remains imprecise when based on conventional clinical variables alone. We aimed to develop and geographically validate a machine-learning model that integrates routinely available clinical, electrocardiogram, and device interrogation/programming parameters, and quantify the incremental value of non-sustained ventricular tachycardia (NSVT). Methods and results: We retrospectively analysed 514 ICD recipients implanted between 2020 and 2025 at two hospital sites (development cohort) and an independent cohort of 220 patients from a third site (external validation). The endpoint was appropriate ICD therapy (anti-tachycardia pacing or shock). Models were trained using nested stratified cross-validation in the development cohort, and the final refitted model was applied to the external cohort without recalibration. In the development cohort, 77/514 patients (15%) experienced appropriate ICD therapy (follow-up 404 days). A base model (histogram-based gradient boosting) excluding NSVT achieved modest discrimination [receiver operating characteristic (ROC)-area under the curve (AUC) 0.624; average precision 0.198]. Adding NSVT improved performance (ROC-AUC 0.805; average precision 0.432). Logistic regression with NSVT achieved comparable discrimination (ROC-AUC 0.78; average precision 0.43), indicating a largely NSVT-driven gain. External validation (event rate 51/220, 23%) confirmed good discrimination (ROC-AUC 0.815; average precision 0.583) with acceptable calibration (Brier score 0.137). At the pre-specified threshold of 0.16 in external validation, sensitivity was 0.78 and specificity was 0.74. Conclusion: A machine-learning model integrating routinely available clinical, electrocardiogram, and ICD programming/interrogation data may enable the prediction of appropriate ICD therapy with preserved performance on geographic external validation. NSVT was the dominant contributor to predictive performance across modelling approaches.