Bleeding-related conditions, procedural safety, and mortality after left atrial appendage closure Authors: Grégoire Massoullié, Come Fleury, Emeline Tourret, Simon Berger, Catherine Basthier-Denis, Guillaume Clerfond, Pierre-Antoine Catalan, Aurélie Chalard, Clément Riocreux, Aurélien Lebreton, Bruno Pereira, Romain Eschalier Published: 2026 Left atrial appendage closure (LAAC) is increasingly used for stroke prevention in atrial fibrillation patients unsuitable for long-term oral anticoagulation.1,2 In daily practice, referral is often driven by bleeding-prone conditions—thrombocytopenia, advanced renal dysfunction, severe anaemia, or advanced age—that may be intuitively perceived as procedural risk. We hypothesized that these conditions mainly reflect long-term clinical-biological vulnerability rather than acute procedural hazard. We retrospectively analysed 747 consecutive patients undergoing LAAC between 2016 and 2023 in a tertiary centre. All indications were validated by a multidisciplinary Heart Team, with geriatric and specialist assessment when appropriate. This cohort therefore reflects outcomes after structured multidisciplinary selection rather than an unfiltered population of high-risk patients. Procedures were performed under general anaesthesia or conscious sedation with transoesophageal or micro-transoesophageal echocardiography guidance, as previously described.3 Clinical follow-up was scheduled at 3 months. Device surveillance imaging was performed according to local practice using cardiac computed tomography (CT) or transoesophageal echocardiography when clinically appropriate. Device-related thrombus was defined as thrombus on the atrial surface of the device detected by CT or transoesophageal echocardiography. The primary procedural outcome was any peri-procedural complication, including access-site complications. Major peri-procedural adverse events were analysed separately and included peri-procedural death, tamponade, stroke or transient ischaemic attack, intracranial haemorrhage, device embolization, anaphylactic shock, or clinically significant airway bleeding. Access-site complications were reported separately because of their heterogeneous clinical severity. The primary long-term outcome was all-cause mortality. A parsimonious exploratory score was derived from independent baseline mortality predictors, assigning one point for age ≥ 85 years, male sex, estimated glomerular filtration rate < 30 mL/min/1.73 m2, haemoglobin < 10 g/dL, fibrinogen ≥ 4 g/L, and platelet count < 150 × 109/L. The score was grouped as low (0–1), intermediate (2–3), and high risk (≥4). Successful implantation was achieved in 736/747 patients (98.5%; three aborted before device deployment). Any peri-procedural complication occurred in 38 patients (5.1%), including access-site complications in 16 (2.1%). Major peri-procedural adverse events occurred in 13 patients (1.7%). Individual major events included cardiac tamponade in six patients (0.8%), device embolization in four (0.5%), stroke or transient ischaemic attack in one (0.1%), intracranial haemorrhage in two (0.3%), and in-hospital death in three (0.4%), including one procedure-related death. Among attempted implantations, devices were Amulet in 338 (45.4%), Watchman in 116 (15.6%), and Watchman FLX in 290 (39.0%). Device type was not associated with all-cause mortality in unadjusted analysis (log-rank P = 0.925) or after adjustment for score components. Bleeding-prone or clinically vulnerable conditions were not associated with excess overall peri-procedural complications: 4.8% vs. 5.2% in patients aged ≥85 years vs. younger patients (P = 1.000), 8.6% vs. 5.0% in patients with platelet count < 150 × 109/L vs. ≥150 × 109/L (P = 0.187), 4.6% vs. 5.0% in patients with estimated glomerular filtration rate < 30 vs. ≥30 mL/min/1.73 m2 (P = 1.000), and 5.6% vs. 5.3% in patients with haemoglobin < 10 vs. ≥10 g/dL (P = 0.804). No tamponade occurred in patients with thrombocytopenia or advanced renal dysfunction. Post-procedural therapy was predominantly single antiplatelet therapy (94.0%), reflecting contraindication to anticoagulation and multidisciplinary assessment of bleeding risk. Dual antiplatelet therapy or therapeutic-dose anticoagulation was reserved for selected patients after Heart Team assessment. Three-month CT was available in 400 patients (53.5%). Device-related thrombus was identified in eight CT-evaluated patients (2.0%) and did not significantly differ across device types. At 3 years, major bleeding occurred in 9 patients (1.3%), ischaemic stroke in 10 (1.4%), and all-cause death in 227 (31.7%). Long-term mortality was associated with baseline clinical-biological markers rather than procedure-related events. As summarized in Figure 1, the exploratory LAAC score stratified 3-year survival: 85.8%, 68.5%, and 52.6% in low-, intermediate-, and high-risk groups, respectively. Compared with low-risk patients, mortality was higher in intermediate-risk (hazard ratio 2.58, 95% confidence interval 1.59–4.20; P < 0.001) and high-risk patients (hazard ratio 4.39, 95% confidence interval 2.67–7.21; P < 0.001). Procedural safety and mortality stratification after left atrial appendage closure. (A) Study population and objectives in consecutive high-risk patients referred for LAAC after multidisciplinary Heart Team selection. (B) Main referral indications leading to LAAC. (C) Procedural safety profile, showing overall event rates, major peri-procedural adverse events, access-site complications, tamponade, and comparisons of any peri-procedural complication rates between vulnerable subgroups and their respective comparators. (D) Components of the exploratory LAAC mortality score, with one point assigned to each baseline clinical-biological marker. (E) Accumulation of score components according to total score. (F) Three-year survival according to LAAC risk category. CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; Hb, haemoglobin; LAAC, left atrial appendage closure; PLT, platelet count. These results are consistent with previous LAAC cohorts showing that mortality after implantation is largely driven by baseline comorbidity rather than device-related complications.4,5 Bleeding indication categories provided limited incremental prognostic information beyond biological markers, suggesting that transversal markers of vulnerability may capture prognosis more effectively than disease-specific labels. This matters because bleeding causes often drive LAAC referral, whereas biological markers may better reflect residual life expectancy and capacity to benefit. This study addresses a practical distinction in contemporary LAAC: bleeding-related conditions should not automatically be equated with procedural risk. In this selected high-risk cohort, procedural safety was preserved across subgroups traditionally perceived as challenging, whereas long-term mortality reflected baseline clinical-biological vulnerability. The key clinical issue is therefore not only whether LAAC can be performed safely in a high-volume centre, but whether procedural success translates into meaningful benefit within the patient’s broader clinical trajectory. The score should therefore be viewed as a tool for prognostic framing and follow-up planning, not as a gatekeeping instrument to deny LAAC. This distinction is increasingly relevant after OPTION and CHAMPION-AF, which evaluated LAAC in selected anticoagulation-eligible populations, and CLOSURE-AF, which challenged the net benefit of LAAC in older patients at high thromboembolic and bleeding risk.1,2,6 In this context, the neutral CLOSURE-AF results reinforce the need to identify patients in whom technical success may not translate into durable clinical benefit. Real-world registry and subgroup data have also documented LAAC in heterogeneous high-risk populations, including patients with thrombocytopenia, end-stage renal disease, and advanced age.7–10 Accordingly, risk stratification may be most useful in the historical contraindication-to-anticoagulation population, whereas its relevance in younger or anticoagulation-eligible candidates remains uncertain. Several caveats are essential. First, these findings should not be extrapolated to unselected patients with severe thrombocytopenia, advanced renal dysfunction, or very advanced age. Second, cause-specific mortality was unavailable, limiting distinction between cardiovascular and non-cardiovascular death. Therefore, the interpretation that mortality reflects non-procedural vulnerability is supported indirectly by the low rates of stroke, major bleeding, device-related thrombus, and procedural events, but cannot be confirmed by adjudicated cause-specific mortality. Third, the score is exploratory and hypothesis-generating; it should not determine LAAC eligibility before external validation against independent cohorts and established frailty or mortality tools. In conclusion, among high-risk patients undergoing LAAC for contraindication to anticoagulation, bleeding-prone conditions were not associated with excess procedural complications, whereas baseline clinical-biological markers stratified long-term mortality. A simple exploratory score may help multidisciplinary atrial fibrillation teams contextualize procedural success and plan follow-up, without replacing individualized Heart Team judgement. AI statement The authors used a large language model during manuscript preparation for language editing, rephrasing, and organization of selected sections of the text. All scientific content, analyses, interpretations, and final editorial decisions were reviewed and validated by the authors. This work was performed in Clermont-Ferrand University Hospital. Contributor Information Grégoire Massoullié, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France; Image-Guided Therapies (TGI), Université Clermont Auvergne, CNRS, SIGMA Clermont, Institut Pascal, 4 Avenue Blaise Pascal, 63178 Aubière, France. Come Fleury, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France. Emeline Tourret, Hematology Department, Clermont-Ferrand University Hospital, Clermont-Ferrand, France. Simon Berger, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France. Catherine Basthier-Denis, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France. Guillaume Clerfond, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France; Image-Guided Therapies (TGI), Université Clermont Auvergne, CNRS, SIGMA Clermont, Institut Pascal, 4 Avenue Blaise Pascal, 63178 Aubière, France. Pierre-Antoine Catalan, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France. Aurélie Chalard, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France. Clément Riocreux, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France. Aurélien Lebreton, Hematology Department, Clermont-Ferrand University Hospital, Clermont-Ferrand, France. Bruno Pereira, Biostatistics Unit (Clinical Research and Innovation Direction), Clermont-Ferrand University Hospital, Clermont-Ferrand F-63000, France. Romain Eschalier, Cardiology Department, Clermont-Ferrand University Hospital, 58 Rue Montalembert, Clermont-Ferrand F-63000, France; Image-Guided Therapies (TGI), Université Clermont Auvergne, CNRS, SIGMA Clermont, Institut Pascal, 4 Avenue Blaise Pascal, 63178 Aubière, France. Funding None declared. Data availability The data underlying this article will be shared on reasonable request to the corresponding author. References Associated Data Data Availability Statement The data underlying this article will be shared on reasonable request to the corresponding author.