Research · Apr 25, 2026
Coronary obstruction in TAVI: a decade of CT screening
The rate has not changed in 13 years
The Spanish TAVI registry (Ojeda et al., JACC Cardiovasc Interv 2023; doi:10.1016/j.jcin.2023.03.024) followed 13,675 patients at 46 centers from 2009 to 2021. Coronary obstruction occurred in 0.80%. The annual rate ranged from 0.3% to 1.3% and did not fall over the period.
Mortality with obstruction was 38.3% vs 4.3% at 30 days and 39.1% vs 9.1% at 1 year. Imaging refinements, risk scores and BASILICA arrived in that period, and the rate stayed where it was in 2009.
Only 31.7% of native-valve obstructions had two or more classical CT risk factors, against 78.3% in valve-in-valve. PCI technical success was 78.0%, and 69.1% with self-expanding valves (88.9% with balloon-expandable).
- For: the largest and longest real-world dataset on the topic: 13,675 patients, 46 centers, 13 years.
- Against: voluntary reporting without external audit; no central CT core lab; no CT data on controls, so risk-factor prevalence in the unobstructed population is unknown.
The flat rate has three possible readings: screening has not improved; it has improved, but TAVI has moved to younger and more anatomically diverse patients; or high-risk patients now go to BASILICA or surgery, leaving an irreducible residual. Probably all three. We miss the diagnosis before the procedure and fail the bailout when obstruction occurs.
Ribeiro 2013: geometry misses 1 in 5
Ribeiro et al. (J Am Coll Cardiol 2013;62:1552; doi:10.1016/j.jacc.2013.07.040) analyzed 44 obstructions among 6688 TAVIs. In the 1:1 analysis matched on age, sex, prior CABG, valve type and size, the per-millimeter OR was 2.17 (1.62 to 2.90) for left coronary height and 1.37 (1.13 to 1.66) for sinus of Valsalva diameter.
86% of obstructions had left coronary height below 12 mm, against 26% of controls. 71% had a sinus below 30 mm, against 33%. Female sex, balloon-expandable valves (0.81% vs 0.34%) and prior bioprosthesis modified risk. But 21% of obstructions occurred with left coronary height above 12 mm.
- For: central core-lab CT; matching on the main confounders; per-millimeter OR instead of an arbitrary cutoff.
- Against: CT missing in 16 of 44 cases (selection bias); no multivariable model, for lack of events; bulky calcium nodules not assessed.
Khan 2023: calcium captures what geometry misses
Khan et al. (JACC Cardiovasc Interv 2023;16:415; doi:10.1016/j.jcin.2022.11.018) compared 60 obstructions with 1381 controls. The rule: cusp height greater than coronary height AND (virtual valve-to-coronary distance ≤ 4 mm OR culprit leaflet calcium > 600 mm³). 24% of obstructed arteries had a distance > 4 mm and were caught only by the calcium term. Apparent AUC 0.93 for the left and 0.94 for the right coronary, in the derivation set.
- For: the largest native-valve CT dataset; variables chosen by mechanism, not by data dredging; calcium catches cases that geometry misses.
- Against: derivation-set AUC, without true external validation; right coronary submodel underpowered (13 events, 3 predictors); the 600 mm³ threshold depends on software and was not tested across platforms.
Jabbour 2018: delayed obstruction
Jabbour et al. (J Am Coll Cardiol 2018;71:1513; doi:10.1016/j.jacc.2018.01.066) analyzed 38 delayed obstructions among 17,092 TAVIs. Timing was bimodal: 47% within 24 h, none between 7 and 59 days, 37% at 60 days or later. Two non-overlapping windows suggest two mechanisms.
Self-expanding valves carried about three times the delayed-obstruction risk of balloon-expandable valves (0.36% vs 0.11%). This is the opposite of Ribeiro’s acute direction and fits continued nitinol expansion. Late obstructions occurred with mean left coronary height 13.8 mm and sinus 29.7 mm, within the range considered safe.
- For: the first registry to describe delayed obstruction as its own entity; the 7-to-59-day gap supports two mechanisms; central CT review.
- Against: no matched control group, so no AUC or score; voluntary reporting from centers with cases (overestimates incidence) and missed sudden out-of-hospital deaths (underestimates it).
ICARO 2025: IVUS of the ostium after implantation
Lunardi et al. (JACC Cardiovasc Interv 2025;18:1147; doi:10.1016/j.jcin.2025.03.016) studied 32 patients (40 vessels), mostly valve-in-valve, with some native and one bicuspid, all at high pre-procedural risk. Post-TAVI IVUS of the ostium was feasible and safe in all 38 vessels without sudden obstruction.
Angiography and IVUS disagreed on the need for chimney stenting in 26% of vessels (κ = 0.41). In 5 vessels angiography looked normal, and IVUS showed a “fish-mouth” residual area of 3.6 mm². In another 5 angiography suggested sequestration, and IVUS showed a 28 mm² paraostial space. With IVUS, chimney stenting was deferred in 63% of high-risk vessels, with 1-year MACCE of 6.2% in both groups.
- For: prospective, multicenter, with an IVUS core lab and blinded expert consensus on stenting.
- Against: 32 patients, only 8 native (1 bicuspid); operator decision as comparator, with no hard outcome to validate it; multivariable analysis underpowered to find IVUS predictors.
It is hypothesis-generating, but it is the first tool aimed at the gap Ojeda measured.
Where screening fails
Pre-procedural CT captures the acute mechanism: the leaflet pinned against the ostium, by geometry or calcium. I think we do this well. CT does not capture delayed obstruction: those patients have normal-range ostial height and sinus. Jabbour’s 14 late cases would have been classified as low risk by Khan’s rule.
During the procedure, IVUS would change the stenting decision in 1 of 4 cases compared with angiography. Protective wires can falsely reassure by displacing the leaflet, and obstruction may sit above the ostium, out of angiography’s reach. ICARO’s 5 fish-mouth cases would have been read as normal on angiography. Bicuspid anatomy, with asymmetric calcium and a raphe, sits outside every derivation cohort discussed here.
What we have and what is missing
| Phase | Tool | Limit |
|---|---|---|
| Pre-procedural | Khan’s rule (cusp–coronary, valve-to-coronary distance, calcium) | Covers the roughly 80% of acute obstructions from ostial leaflet pinning |
| Intra-procedural | Angiography + IVUS at high risk | ICARO: may change management in 1 of 4 cases |
| Post-procedural | Nothing systematic | Pre-discharge CT in high-risk anatomy and review at 60 days, as Jabbour’s group suggested |
Ojeda showed the rate has not moved in a decade. Ribeiro showed geometry misses 1 in 5. Khan showed calcium volume catches most of those, but the threshold depends on software and lacks external validation; we do not know how many of Ribeiro’s missed cases the rule would catch in practice. Jabbour showed the late mechanism escapes CT. ICARO suggests IVUS may close part of the intra-procedural gap.
We screen well for the patient who will obstruct in seconds, and adequately for the one with a bulky leaflet who will obstruct in minutes. With IVUS, we may catch the one who looks normal on angiography. We do not screen for the patient who will obstruct in two months, and we do not yet know how to find them.