2025 ESC/EACTS Valvular Guidelines: imaging perspective over a new treatment direction

The 2025 ESC/EACTS Guidelines on valvular heart disease arrive at a moment when ‘treatment direction’ is no longer a single choice between surgery and catheter therapy but a longitudinal strategy—indexed to lifetime risk, feasibility of repeat procedures, and the real (often competing) priorities of patients. In that landscape, imaging is not simply diagnostic: it is the language that allows the Heart Team to translate anatomy, haemodynamics, symptoms, comorbidity, and procedural risk into a coherent plan. The guideline update explicitly reinforces this centrality—linking Heart Team decision-making, Heart Valve Centres, and ‘advanced diagnostic imaging with more clearly defined diagnostic criteria’ as pillars of contemporary care. What is most striking is not any single new class of recommendation, but the way the document treats imaging as the gatekeeper of ‘when’ to intervene, ‘how’ to intervene, and ‘what comes next’ after the index procedure. That perspective is worth making explicit—because for imagers, the guideline is less a list of threshold values and more an invitation to build reproducible, multimodality pathways that can support increasingly complex treatment trajectories.1 Several early headlines changed to signal the guideline’s intent to modernize selection and timing decisions while preserving a patient-centred, multidisciplinary framework. A notable ‘headline’ change is the age cut-off moving from 75 to 70 years in favour of transcatheter aortic valve intervention (TAVI) for patients with tricuspid aortic valve stenosis, as summarized by EACTS. The official guideline educational materials also visualize age (≥70 vs. 1200 AU in women and >2000 AU in men) to adjudicate severity. This is not merely a technical flow chart; it is a formal acknowledgement that ‘discordant AS is common’ and that multimodality adjudication is no longer optional in high-stakes decisions (e.g. an intervention at borderline gradients). The EACVI/ESC position paper on native valvular regurgitation stresses a multimodality approach that integrates quantification, valve anatomy, and chamber consequences—precisely because management decisions increasingly depend on subtle evidence of decompensation rather than symptoms alone.2 For mitral regurgitation, this includes careful differentiation of mechanism (primary vs. secondary, and now atrial secondary), and for TR, it implies that echo must do more than label ‘severe’: it must quantify right-sided remodelling and function in a reproducible way that can support intervention selection and timing.2 As transcatheter options expand and surgery is recommended earlier in select asymptomatic PMR, the quality of anatomical definition [3-dimensional (3D) leaflet analysis, annular geometry, coaptation gap] and functional reserve assessment becomes the difference between a confident repair-first strategy and therapeutic hesitation.2 CT is increasingly the modality that turns ‘a valve lesion’ into ‘a feasible treatment plan’. The guideline’s integrative AS approach explicitly includes CT calcium scoring thresholds to resolve uncertainty. This elevates CT from a planning tool to a diagnostic arbitrator—particularly relevant in low-flow/low-gradient scenarios where echo alone may be ambiguous. The guideline’s visual framework for choosing intervention mode lists anatomic features—hostile annulus/left ventricular outflow tract calcification, bicuspid anatomy, annular dimensions unsuitable for TAVI, and coronary obstruction risk—as factors favouring surgery, while transfemoral feasibility, porcelain aorta, and intact bypass grafts may favour TAVI. The significance for imagers is direct: ‘CT-derived anatomy now carries guideline-level weight in modality selection’, not just in procedural planning. The same figure explicitly calls out lifetime management issues such as coronary obstruction risk and impaired coronary access after valve-in-valve TAVI. This is an imaging mandate: pre-procedural CT interpretation must anticipate downstream scenarios, not only immediate feasibility. Cardiac MR’s (Magnetic Resonance) strength is not replacing echo; it is resolving what echo cannot reliably quantify when decisions are close. For regurgitation, the EACVI/ESC position paper explicitly integrates CMR into the multimodality assessment of native valve regurgitation, recognizing its ability to quantify regurgitant volume and ventricular remodelling when echo windows, eccentric jets, or multiple lesions create uncertainty.2 In multivalvular disease, where interacting lesions can distort Doppler assumptions, the EACVI consensus statement underscores multimodality imaging as integral to Heart Team discussion and complex decision-making.3 A practical way to state this in guideline-era terms: CMR is often most valuable when the treatment decision depends on chamber consequences, not just valve appearance—especially in mixed or multiple valve disease where ‘severity’ cannot be safely reduced to a single echo measurement.2,3 As procedural volume rises (surgical and transcatheter), prosthetic valve assessment becomes a major imaging workload—and a major source of misclassification if done with a single modality mindset. The 2024 ASE guideline on prosthetic valve function explicitly frames echocardiography as foundational but emphasizes the complementary role of CT and CMR, and even Positron Emission Tomography in selected contexts.4 The deeper implication for 2025 valve care is that ‘prosthetic evaluation should be pre-planned’—baseline post-implant studies, reproducible follow-up protocols, and early escalation to advanced imaging when haemodynamics and symptoms diverge. The guideline direction—especially for TR—implies more transcatheter procedures in anatomically complex spaces. That means intraprocedural imaging must be treated as a determinant of outcome, not a procedural accessory. For TR, the guideline recommendation itself is explicitly conditioned on avoiding severe RV dysfunction or pre-capillary PH.2–10 This places imaging in a dual role: ‘selection’ (who is still likely to benefit) and ‘guidance’ (how to execute safely within complex right-heart anatomy). In practice, this is where echo (often 3D transesophageal echocardiogram and/or Intracardiac Cardiac Ecocardiography) and CT-derived anatomical understanding must converge into a procedural map that the operator can execute. The most intriguing hint about the future may be embedded in the guideline ecosystem itself: the ESC guideline page explicitly notes that ESC Guidelines are protected by copyright and that use of guideline content to train or develop generative artificial intelligence (AI) models requires a formal licence agreement (Figure 1). That statement is not merely legal; it is a signal that AI is expected to intersect with guidelines in clinically meaningful ways—and that governance, attribution, and controlled implementation will matter. Central illustration on the present and the future of new valvular guidelines. So the open question is not whether AI will enter valve imaging—it already has, through automation and decision support—but what kind of AI we actually need. Do we want AI that is excellent at segmenting valves and measuring chambers, yet disconnected from guideline logic? Or do we want ‘guideline-native AI’: systems that (i) follow the same integrative pathways (like the AS discordance algorithm), (ii) encode multimodality uncertainty rather than hiding it, and (iii) output structured, auditable statements that a Heart Team can challenge, accept, or override? Early work suggests feasibility: AI-guided echocardiography acquisition and interpretation can improve workflow and reproducibility, potentially lowering operator dependence—an attractive proposition when guidelines push earlier intervention in asymptomatic disease and expand transcatheter indications where anatomy is decisive.5 At the same time, professional societies have emphasized that AI in cardiovascular imaging must be validated, explainable where possible, and integrated into clinical governance rather than treated as a black box.6 Even clinician-facing primers increasingly stress that the main value may be ‘standardization and scalability’, not magical new biomarkers.7 A plausible ‘next fusion’, then, is not AI replacing multimodality imaging—but AI acting as the connective tissue that makes multimodality practical at scale: ‘Standardized acquisition and reporting’ across centres, enabling reliable longitudinal comparisons in lifetime management. ‘Automated identification of discordant patterns’ (e.g. AS low-flow phenotypes) that trigger the correct next test rather than an incorrect conclusion. ‘Procedural foresight’ in CT planning—anticipating coronary access and redo scenarios as part of the index report, consistent with lifetime framing. ‘Risk stratification that respects phenotype’, particularly in TR where benefit hinges on right ventricle/pulmonary hypertension (RV/PH) boundaries. But that fusion only helps if it remains faithful to the premise of the 2025 guideline direction: patient-centred decisions made by accountable Heart Teams, supported by imaging that is transparent, reproducible, and clinically contextual. The challenge for the next guideline cycle is therefore as much organizational as it is technological: can we build imaging-and-AI workflows that are trustworthy enough to justify earlier intervention and more complex lifetime pathways—without amplifying variability, bias, or overconfidence? That is the real frontier: not ‘AI in imaging’, but AI that strengthens guideline-based care without eroding clinical accountability. None. No new data were generated or analysed in support of this research.