TAVI (Transcatheter Aortic Valve Implantation)
Definition
Transcatheter aortic valve implantation (TAVI; TAVR in American usage) is catheter-based implantation of a bioprosthetic valve in the aortic position without sternotomy, most often by the transfemoral route. Randomised evidence has carried it from inoperable and high-risk patients to intermediate- and low-risk populations, and it now accounts for most aortic valve replacement in high-income countries. This page covers indications, device and access selection, the randomised evidence against surgery and between platforms, timing in asymptomatic disease, peri-procedural management, antithrombotic therapy and follow-up. Two adjuncts have their own pages: concepts/PCI-Before-TAVI and concepts/Cerebral-Embolic-Protection-TAVI.
Key Concepts
Indications and Guideline Position
| Indication | ESC 2025 position | Detail |
|---|---|---|
| Symptomatic severe AS | Class I, any surgical risk | Heart Team decision on anatomy, life expectancy, procedural risk and patient preference |
| Asymptomatic severe AS with LVEF ≥50%, normal exercise test, low procedural risk | Class IIa (new 2025) | Early intervention as an alternative to watchful waiting; supported by EARLY TAVR, RECOVERY, AVATAR and EVoLVeD |
| Bicuspid AS at high surgical risk | Class IIb B | If anatomy suitable; historically excluded from major RCTs |
| Severe AR in inoperable patients | Class IIb B | Off-label use of non-dedicated devices carries ~10% risk of a second valve or surgical conversion; dedicated devices carry a ~24% pacemaker rate |
| Moderate AS with HFrEF | Not supported | Evidence insufficient for routine intervention |
(sources/vhd-esc-2025, rating: very high)
- The organising principle is a Heart Team decision, not a fixed age cut-off. ESC 2025 integrates anatomy, life expectancy, procedural risk and an informed patient preference — the last a Class I consideration; ACC/AHA 2020 still used age bands (SAVR <65 years, either 65–80, transfemoral TAVI >80). (sources/vhd-esc-2025; sources/VHD-AHA-2020)
- Anatomy pushes toward surgery when the root is narrow, the coronary ostia are low, commissural alignment is unfavourable, the LVOT is heavily calcified, or the valve is bicuspid — each of these either makes TAVI deployment hazardous or compromises future coronary access. (sources/vhd-esc-2025, rating: very high)
- Risk scores still arbitrate: EuroSCORE II and STS-PROM, with high surgical risk favouring TAVI.
Device and Access Selection
Small annulus (≤430 mm²). SMART is the only randomised head-to-head comparison of the two dominant platforms in this group — a population that is about 87% women and at high risk of prosthesis–patient mismatch (sources/tavr-smart-nejm-2024, rating: very high):
| End point at 12 months | Self-expanding supraannular | Balloon-expandable |
|---|---|---|
| Death, disabling stroke or HF rehospitalisation | 9.4% | 10.6% (non-inferior; HR 0.90) |
| Bioprosthetic-valve dysfunction composite | 9.4% | 41.6% |
| Mean gradient | 7.7 mmHg | 15.7 mmHg |
| Effective orifice area | 1.99 cm² | 1.50 cm² |
| Moderate or severe prosthesis–patient mismatch (30 d) | 11.2% | 35.3% |
| New permanent pacemaker | 12.1% | 7.8% (HR 1.61, 0.98–2.65) |
| Mild or greater aortic regurgitation | 14.1% | 20.3% |
- The dysfunction difference is haemodynamic, not clinical. Haemodynamic structural valve dysfunction was 3.2% vs 32.2% and nonstructural dysfunction 5.9% vs 18.2%, but clinical valve thrombosis (0.3% vs 0.3%), endocarditis (0.6% vs 2.3%) and reintervention (0.9% vs 0.6%) were rare and similar. (sources/tavr-smart-nejm-2024)
- Read it as a haemodynamic gain with a conduction cost, not a mortality benefit: the clinical coprimary end point was equivalent, and the pacemaker interval crosses 1. Whether the lower gradient becomes a durability advantage is what the 5-year follow-up must show — see concepts/Structural-Valve-Deterioration.
- Bicuspid valves: Class IIb B at high surgical risk if anatomy permits. Most device RCTs excluded BAV, including PARTNER 3; SMART allowed it but reported no bicuspid subgroup, so the platform question in BAV remains untested. Observational data suggest more paravalvular leak, second valve implantation and pacemaker requirement. (sources/tavr-smart-nejm-2024; sources/vhd-esc-2025)
- Access route: transfemoral is the default — 97.3% of DEDICATE patients — and iliofemoral anatomy is assessed on the planning CT; non-transfemoral access is reserved for patients in whom the iliofemoral route is unsuitable. (sources/tavr-dedicate-nejm-2024, rating: very high)
- Valve-in-valve is established for a degenerating surgical bioprosthesis, but a valve deployed inside a surgical ring leaves a smaller orifice and higher residual gradients, and the durability of a TAVI valve beyond 8–10 years remains the unresolved question for valve-in-valve chains. (sources/vhd-esc-2025)
Evidence — TAVI versus SAVR
- High surgical risk: TAVI non-inferior or superior to surgery (PARTNER 1, CoreValve High Risk). Intermediate risk: non-inferior (PARTNER 2, SURTAVI). Neither trial is ingested in this wiki; this summary is the one carried by the guidelines. (sources/VHD-AHA-2020, rating: very high; sources/vhd-esc-2025)
- Low risk: two ingested trials, which disagree in emphasis rather than in direction.
| Trial (follow-up) | Population | Primary outcome | Reading |
|---|---|---|---|
| PARTNER 3 (5 years) | n=1,000; STS-PROM 1.9%; mean age 73; transfemoral SAPIEN 3 | Death, stroke or rehospitalisation 22.8% vs 27.2%; HR 0.79 (P=0.07) | No significant difference; the early TAVI advantage is lost by 5 years |
| DEDICATE-DZHK6 (1 year) | n=1,414; median STS-PROM 1.8%; mean age 74; any CE-marked device | Death or stroke 5.4% vs 10.0%; HR 0.53 (0.35–0.79) | Non-inferiority exceeded in an industry-independent trial with operator-chosen devices |
- Temporal attenuation is the key difference. PARTNER 3 favoured TAVR at 1–2 years but not at 5; landmark analysis for years 1–5 gives HR 1.17, and all-cause death was 10.0% vs 8.2% (NS). DEDICATE, by contrast, is a 1-year result. (sources/tavr-partner3-5yr-nejm-2023, rating: very high; sources/tavr-dedicate-nejm-2024)
- The trade-off profile is consistent across trials. TAVI wins on new-onset AF (12.4% vs 30.8% in DEDICATE; 13.7% vs 42.4% in PARTNER 3), major bleeding, acute kidney injury and length of stay; surgery wins on pacemaker implantation, paravalvular leak and vascular access complications. (sources/tavr-dedicate-nejm-2024)
- Cross-trial comparison is unreliable because DEDICATE's 1-year SAVR mortality (6.2%) was roughly three times PARTNER 3's, for reasons discussed under Contradictions.
Timing in Asymptomatic Severe AS
- EARLY TAVR randomised 901 asymptomatic patients with severe AS (mean age 75.8; STS-PROM 1.8%; LVEF ≥50%; negative stress test in 90.6%) to transfemoral SAPIEN 3/3 Ultra or guideline-based surveillance. The primary composite of death, stroke or unplanned cardiovascular hospitalisation was 26.8% vs 45.3% (HR 0.50, 0.40–0.63), but all-cause death (8.4% vs 9.2%; HR 0.93) and stroke (4.2% vs 6.7%; HR 0.62) were not individually significant. (sources/tavi-earlytavr-nejm-2025, rating: very high)
- The composite was driven by hospitalisation (20.9% vs 41.7%; HR 0.43) and by 105 surveillance-arm conversions within 6 months counted as events; 87.0% of the surveillance arm eventually underwent AVR at a median 11.1 months, so the trial compares immediate with delayed TAVI rather than TAVI with indefinite observation.
- RECOVERY points the other way on mortality, with a 10-year all-cause mortality HR of 0.42 for early surgery in very severe asymptomatic AS (Vmax ≥4.5 m/s). The difference is follow-up duration, disease severity and crossover, not a true contradiction — but it is the reason the mortality question in asymptomatic AS is still open. (sources/as-recovery-nejm-2026, rating: very high)
- This evidence base is what supports the ESC 2025 Class IIa for early intervention. Full trial detail is on concepts/Aortic-Stenosis.
Procedural Planning and Concomitant CAD
- CCT is mandatory: annular dimensions, aortic root anatomy, coronary ostia height, LVOT calcification, fluoroscopic projections and iliofemoral access. TOE or CMR substitutes when CCT is compromised by renal failure or arrhythmia. Commissural alignment should be optimised for future coronary access. (sources/vhd-esc-2025, rating: very high)
- CAD screening: CCTA is sufficient to exclude significant CAD when the planning CT is of adequate quality (Class IIa).
- PCI before TAVI: Class IIa B for ≥90% stenosis in vessels ≥2.5 mm, and Class IIb B for ≥70% stenosis in proximal segments of main vessels. Coronary access after TAVI is constrained by frame height, commissural alignment and a narrow root, so the decision is also a future-access decision. (sources/vhd-esc-2025)
- NOTION-3 (n=455; FFR ≤0.80 or ≥90% stenosis; median age 82) tested that Class IIa recommendation: MACE at 2 years was 26% with PCI vs 36% conservative (HR 0.71, 0.51–0.99), driven by MI (HR 0.54) and urgent revascularisation (HR 0.20); all-cause death was neutral (HR 0.85) and bleeding higher (HR 1.51). Full synthesis on concepts/PCI-Before-TAVI. (sources/PCI-TAVI-NOTION3-NEJM-2024, rating: very high)
Cerebral Embolic Protection
- CEP filters (Sentinel) are deployed in the brachiocephalic and left common carotid arteries by right radial access to capture debris during valve crossing and deployment.
- Neither large RCT supports routine use. BHF PROTECT-TAVI (n=7,635, 33 UK centres) was stopped for futility: stroke within 72 hours 2.1% vs 2.2% (P=0.94), with serious adverse events doubled (0.6% vs 0.3%). PROTECTED TAVR (n=3,000) was similarly null for its primary stroke end point. (sources/cep-tavi-bhfprotecttavi-nejm-2025, rating: very high)
- Full device mechanism and the mechanistic disconnect are on concepts/Cerebral-Embolic-Protection-TAVI.
Antithrombotic Therapy
| Situation | ESC 2025 | ACC/AHA 2020 |
|---|---|---|
| No OAC indication | ASA 75–100 mg daily for 12 months (I A); long-term ASA thereafter (IIa C); DAPT not recommended (III B); routine OAC not recommended (III A) | ASA (IIa B-R); DAPT for 3–6 months (IIb B-NR); VKA INR 2.5 for ≥3 months (IIb B-NR) |
| OAC indication (e.g. AF) | OAC alone (I B); OAC plus clopidogrel not recommended | — |
| Low-dose rivaroxaban + ASA | Not recommended — GALILEO halted for harm | III: Harm |
(sources/vhd-esc-2025, rating: very high; sources/VHD-AHA-2020)
- The two guidelines diverge mainly in conviction: ESC 2025 upgraded aspirin to Class I A and made DAPT Class III, whereas ACC/AHA 2020, with less evidence, had aspirin at Class IIa and permitted short-course DAPT.
- DOAC versus VKA after TAVI when anticoagulation is indicated: neither is definitively superior (ATLANTIS, ENVISAGE-TAVI AF). (sources/vhd-esc-2025)
Follow-Up and Complications
- Imaging: TTE within 3 months, at 1 year, then annually; right heart catheterisation is not routine, and CCT or PET is reserved for suspected prosthetic valve thrombosis or endocarditis. (sources/vhd-esc-2025, rating: very high)
- Paravalvular leak: 20.8% vs 3.2% (mild or greater) against surgery in PARTNER 3, though not associated with excess 5-year mortality there; the platform comparison reverses in small annuli, where the self-expanding valve had less regurgitation (14.1% vs 20.3%). (sources/tavr-partner3-5yr-nejm-2023; sources/tavr-smart-nejm-2024)
- Conduction disturbance: new pacemaker 13.5% vs 10.4% against surgery in PARTNER 3, 11.8% vs 6.7% in DEDICATE (HR 1.81), and 12.1% vs 7.8% between platforms in SMART. Right bundle branch block and pre-existing conduction disease are the practical predictors.
- Vascular access: 7.9% vs 0.7% with surgery in DEDICATE (HR 10.64) — the dominant procedural complication of the transfemoral route. (sources/tavr-dedicate-nejm-2024)
- Stroke and valve thrombosis: stroke 5.8% vs 6.4% against surgery at 5 years (PARTNER 3) and 2.9% vs 4.7% at 1 year (DEDICATE); clinically significant valve thrombosis was 2.5% vs 0.2% with TAVI in PARTNER 3, and equal between platforms in SMART.
- Coronary obstruction showed no difference between platforms in SMART; low ostia and a heavily calcified LVOT remain the anatomical risk markers, assessed at planning.
- Structural deterioration, HALT and their definitions are on concepts/Structural-Valve-Deterioration.
Noncardiac Surgery After TAVI
- Noncardiac surgery may proceed even within 30 days of a successful TAVI (COR 2a) when the surgical condition is time-sensitive — the opposite of the post-DES rule, because relieving the fixed obstruction removes the valve-related haemodynamic risk, leaving general surgical and comorbidity risk. (sources/periop-aha-2024, rating: very high)
- Continue aspirin perioperatively (COR 2a) to avoid early leaflet thrombosis.
- The same reasoning applies after mitral transcatheter edge-to-edge repair (COR 2a).
Contradictions / Open Questions
- Durability beyond 8–10 years is the deciding unknown for younger and low-risk patients. PARTNER 3 shows similar bioprosthetic-valve failure at 5 years (3.3% vs 3.8%), but 10-year data are pending, valve-in-valve for a failed TAVI valve carries higher residual gradients, and SMART's haemodynamic advantage has not yet been shown to prevent failure. (sources/tavr-partner3-5yr-nejm-2023, rating: very high; sources/tavr-smart-nejm-2024)
- PARTNER 3 late mortality attenuation: TAVI was superior at 1–2 years and equivalent at 5, with numerically more late deaths (10.0% vs 8.2%, HR 1.24, NS) and a landmark HR of 1.17 for years 1–5. Valve biology, patient selection, differential comorbidity and the pandemic have all been proposed; none has been tested. (sources/tavr-partner3-5yr-nejm-2023)
- Valve thrombosis signal: 2.5% vs 0.2% by VARC-3 criteria at 5 years in PARTNER 3, with no deaths and three strokes. The mechanism is speculative — possibly differential early anticoagulation — and the implications for durability and stroke prevention are unresolved.
- DEDICATE versus PARTNER 3 mortality discordance: 1-year SAVR death was 6.2% in DEDICATE against roughly 2% in PARTNER 3 despite similar STS-PROM scores, attributed to pandemic-era enrolment and a higher proportion of women. This inflates DEDICATE's apparent TAVI benefit and makes cross-trial comparison unreliable. Separately, DEDICATE was a non-inferiority trial (margin HR 1.14); its HR of 0.53 is consistent with superiority but was not designed to claim it. (sources/tavr-dedicate-nejm-2024, rating: very high)
- DEDICATE crossovers: 70 SAVR-allocated patients underwent TAVI, mostly by preference. As-treated analysis agreed with intention-to-treat, but the crossover attenuates the SAVR mortality estimate.
- SMART's platform superiority is haemodynamic, not clinical. The 41.6% balloon-expandable dysfunction rate is built from a gradient threshold, measured prosthesis–patient mismatch and moderate regurgitation; the components representing actual valve failure were rare and equal. The trial's structural dysfunction definition uses an absolute threshold (≥20 mmHg) without requiring a rise from baseline, unlike the 2025 unified definition — so part of the difference is the orifice achieved at implantation, not deterioration. (sources/tavr-smart-nejm-2024, rating: very high)
- SMART leaves the pacemaker penalty unresolved: 12.1% vs 7.8% (HR 1.61, 0.98–2.65) in exactly the population — small annulus, mostly women — where conduction risk is highest and where a pacemaker has real long-term consequences. The trial was not sized to weigh the haemodynamic gain against it. (sources/tavr-smart-nejm-2024)
- EARLY TAVR's primary end point versus its hard outcomes: the composite (HR 0.50) is driven by unplanned hospitalisation (HR 0.43), including 105 early conversions counted as events, while death (HR 0.93) and stroke (HR 0.62) were not individually significant. With 87% crossover at a median 11.1 months, the trial answers "immediate versus delayed TAVI", not "TAVI versus indefinite surveillance". (sources/tavi-earlytavr-nejm-2025, rating: very high)
- EARLY TAVR versus RECOVERY on mortality: RECOVERY (10 years, surgery, very severe AS) found HR 0.42 for all-cause mortality; EARLY TAVR (3.8 years, TAVI) found HR 0.93. Duration, disease severity and crossover plausibly explain the gap, but no trial has yet shown a mortality benefit for early TAVI. (sources/as-recovery-nejm-2026, rating: very high)
- NOTION-3: MACE benefit without a mortality benefit, with excess bleeding. MI and urgent revascularisation drive the result, all-cause death is neutral, and bleeding is nearly doubled. Urgent revascularisation (HR 0.20) may be inflated by open-label knowledge of untreated lesions. ACTIVATION, the earlier and smaller trial with anatomic rather than FFR-guided selection, was neutral — FFR guidance, sample size and follow-up are the likely explanations. (sources/PCI-TAVI-NOTION3-NEJM-2024, rating: very high)
- Cerebral embolic protection captures debris yet prevents no strokes. DWI-MRI shows new lesions in over 70% of TAVI patients and filters reliably capture material, but both large RCTs are null — most plausibly because haemorrhagic and hypoperfusion strokes are not filterable, the vertebrobasilar territory is unprotected, and most MRI lesions are clinically silent. (sources/cep-tavi-bhfprotecttavi-nejm-2025)
- Bicuspid valve evidence gap: excluded from most major RCTs, cleared only as Class IIb at high surgical risk, and not reported separately even in the one trial that allowed it. Whether TAVI's risk profile in BAV (paravalvular leak, second valve, pacemaker) has improved with current devices is unknown.
- Antithrombotic strategy in the first 12 months with a concomitant OAC indication remains unsettled: OAC alone is Class I, but the VKA-versus-DOAC question has no definitive answer from ATLANTIS or ENVISAGE-TAVI AF. (sources/vhd-esc-2025, rating: very high)
Connections
Valve disease
- concepts/Aortic-Stenosis — the parent condition; grading, natural history and the full trial detail
- concepts/Aortic-Regurgitation — off-label TAVI territory
- concepts/Valvular-Heart-Disease — guideline overview
- concepts/Structural-Valve-Deterioration — how dysfunction is defined, and platform differences in gradient and mismatch
- entities/Bicuspid-Aortic-Valve — the population most often excluded from TAVI trials
Adjunctive procedures
- concepts/PCI-Before-TAVI — concomitant CAD; NOTION-3 and ACTIVATION
- concepts/Cerebral-Embolic-Protection-TAVI — routine CEP null result
Comorbidity and pathways
- entities/Atrial-Fibrillation — the commonest OAC indication after TAVI; new-onset AF is markedly less frequent than with surgery
- entities/Chronic-Coronary-Disease — stable CAD in the TAVI population
- concepts/Perioperative-Cardiovascular-Assessment — timing of noncardiac surgery after TAVI
Sources
Guidelines
TAVI versus SAVR
Platform and timing trials
Procedural adjuncts