Left Atrial Appendage Closure by Surgery-2 (LAACS-2)

LAACS-2

Patient / Population Intervention / Exposition Comparaison Critère de jugement

In 1,500 adults undergoing open-heart surgery, concomitant surgical left atrial appendage closure did not reduce stroke or TIA over 2 years compared with leaving the appendage open.

N

1 500patients

Schéma

Multicentre (Denmark, Spain, Sweden), open-label, 1:1 randomised superiority trial with blinded outcome adjudication by two independent neurologists; stratified by site, surgery type and preoperative/planned oral anticoagulation; permuted blocks of 8–16; follow-up at least 2 years; 90% power at α 0.05 to detect a 60% relative risk reduction

Critère

Stroke including transient ischaemic attack, adjudicated by neurologists blinded to allocation, during follow-up (≥2 years)

Pertinence

2Important — l’un des piliers parmi d’autres.

RésultatOver a median 4.0 years, stroke or transient ischaemic attack occurred in 4.28% with closure and 5.04% with the appendage left open (HR 0.85, 95% CI 0.53–1.37; P = 0.52), with no difference in mortality. Among patients whose CHA2DS2-VASc score was above the median, closure was associated with a 56% lower rate of the primary endpoint (P = 0.02); below the median there was no signal.

Dominguez H, Park-Hansen J, et al. Presented at ESC Congress 2026, Hot Line 12, 31 August.

Discussion et critique

Closing the appendage in everyone opened on the table costs a few minutes and, in a cohort where only 4% had atrial fibrillation at baseline, prevented nothing measurable over four years — 4.28% against 5.04%, an interval running from a halving to a 37% increase. That is the honest headline, and it separates this trial from LAAOS III, where every patient had atrial fibrillation and closure did work. The subgroup above the median CHA2DS2-VASc score is where the events were and where the effect appeared, which is arithmetically unsurprising and prespecified only as a subgroup: it should decide the next trial's entry criteria, not today's operating list.