Outcomes of simplified left atrial appendage occlusion using the WATCHMAN FLX device: the ROSE-FLX study Authors: Wei Zhou, Qian Tang, Anning Zeng, Jiulin Chen, Fang Wang, Jing Wang, Jingjing Chen, LinYan Shi, Ji Li, Yirong Zhang, Dawei Tian, Dandan Chen, Fuliang Luo, Zheyi Wu, Na Deng, Fei Yan, Qianjiang Pan, Zhongrong Liao, Yongxian Tian, Yifan Mao, Huimin Lu, Xing Zhou, Ya Yang, ZhenFa Zhang, Runze Huang, Wei Li Published: 2026 ABSTRACT AbstractAimsConventional left atrial appendage occlusion (LAAO) procedures typically require transoesophageal echocardiography and general anaesthesia, which may limit feasibility in resource-limited settings. Simplified LAAO (sLAAO) guided exclusively by fluoroscopy offers a practical and efficient alternative. This study evaluated procedural safety, efficacy, and short-term outcomes of sLAAO using the WATCHMAN FLX device under exclusive fluoroscopic guidance in a real-world, multicentre setting.Methods and resultsThe ROSE-FLX study was a prospective, multicentre, single-arm registry including 400 patients with non-valvular atrial fibrillation at high thromboembolic and bleeding risk. All underwent sLAAO under local anaesthesia and exclusive fluoroscopic guidance. Procedural data, peri-procedural complications, and follow-up outcomes were analysed. Predictors of adverse events were determined using Cox regression. Procedural success was 100%, with low major complication rates: pericardial effusion (0.5%), access-site complications (1.3%), and major bleeding (0.8%). Over median follow-up of 194.5 days [interquartile range (IQR) 129.0–265.0], all-cause mortality occurred in three patients (0.8%), transient ischaemic attacks in 13 patients (3.3%), and device-related thrombosis in one patient (0.3%). In multivariable Cox regression, chronic obstructive pulmonary disease (HR 2.86, 95% CI 1.98–4.11, P < 0.001) and higher CHA2DS2-VASc scores (HR 4.69, 95% CI 1.90–11.57, P < 0.001) independently predicted adverse events.ConclusionExclusive fluoroscopy-guided sLAAO with the WATCHMAN FLX device is feasible, safe, and resource efficient, achieving high procedural success and low complication rates, suggesting that this approach may be considered in selected centres lacking advanced echocardiographic or anaesthetic support, pending confirmation from comparative studies.Keywords: Left atrial appendage occlusion, Digital subtraction angiography, Atrial fibrillation, Simplified left atrial appendage occlusion This is the first prospective, multicentre registry to demonstrate the safety and efficacy of a simplified left atrial appendage occlusion (LAAO) procedure using the WATCHMAN FLX device under exclusive fluoroscopic guidance, without intra-procedural echocardiography or general anaesthesia. This is the first prospective, multicentre registry to demonstrate the safety and efficacy of a simplified left atrial appendage occlusion (LAAO) procedure using the WATCHMAN FLX device under exclusive fluoroscopic guidance, without intra-procedural echocardiography or general anaesthesia. The procedure achieved 100% technical success in 400 high-risk patients, with a notably low rate of major peri-procedural complications. The procedure achieved 100% technical success in 400 high-risk patients, with a notably low rate of major peri-procedural complications. While LAA morphology did not impact overall success, it significantly influenced fluoroscopy time, highlighting a unique procedural consideration for the fluoroscopy-only approach. While LAA morphology did not impact overall success, it significantly influenced fluoroscopy time, highlighting a unique procedural consideration for the fluoroscopy-only approach. The incidence of device-related thrombosis (0.3%) and significant peri-device leak was very low, potentially reflecting both device design improvements and a standardized implantation assessment protocol. The incidence of device-related thrombosis (0.3%) and significant peri-device leak was very low, potentially reflecting both device design improvements and a standardized implantation assessment protocol. Introduction Left atrial appendage occlusion (LAAO) is an established percutaneous therapy to reduce stroke risk in patients with non-valvular atrial fibrillation (NVAF) by excluding the left atrial appendage (LAA) as the principal source of thrombus.1 Left atrial appendage occlusion reduces thromboembolic events and may mitigate bleeding risks associated with long-term oral anticoagulation, thereby reducing disability and mortality and lowering healthcare utilization in selected high-risk patients.2–4 Consequently, LAAO is an accepted alternative to long-term anticoagulation for patients with NVAF who are at high risk of stroke or bleeding.5,6 Standard LAAO workflows commonly employ transoesophageal echocardiography (TEE) with general anaesthesia to provide real-time imaging and procedural safety.7,8 Although effective, TEE with general anaesthesia increases procedural complexity and resource requirements and may be poorly tolerated or contraindicated in patients with respiratory or oesophageal comorbidities. Intracardiac echocardiography (ICE) enables procedures under conscious sedation without general anaesthesia and provides real-time device–tissue visualization; however, ICE requires specialized catheters and operator expertise and is associated with increased procedural cost.7,9,10 Comparative data remain limited regarding whether TEE- or ICE-guided LAAO confers superior procedural success or fewer complications, despite the greater procedural complexity and longer procedural times associated with these imaging-dependent workflows.11 Interest has grown in simplified LAAO (sLAAO) workflows guided predominantly by fluoroscopy, as they streamline procedural steps, reduce anaesthesia requirements, shorten procedure duration, and increase patient tolerance—benefits that are particularly relevant in resource-constrained centres. However, most published data on fluoroscopy-only LAAO come from single-centre series, and prospective multicentre evidence is limited.12–15 To address this gap, we conducted the ROSE-FLX multicentre prospective registry enrolling 400 high-risk NVAF patients to evaluate the procedural safety, efficacy, and short-term clinical outcomes of exclusive fluoroscopy-guided sLAAO using the WATCHMAN FLX device. Methods Patient enrolment The ROSE-FLX study was a multicentre, prospective, single-arm registry aimed at assessing the safety and feasibility of a sLAAO approach. The study was led by the Affiliated Hospital of Guizhou Medical University and conducted in collaboration with four nationally certified atrial fibrillation centres in Guizhou Province. The number of implants per centre was as follows: Center A (n = 160), Center B (n = 70), Center C (n = 70), Center D (n = 50), and Center E (n = 50). All participating centres were nationally certified atrial fibrillation centres with established experience in LAAO, and no predefined learning curve phase was required. Prior to study initiation, each centre had independently performed routine LAAO procedures, and operators were experienced interventional cardiologists trained in WATCHMAN FLX implantation. No predefined learning curve phase was required before study inclusion. A total of 400 consecutive adult patients (age ≥ 18 years) with NVAF and elevated thromboembolic and bleeding risk were enrolled. Inclusion criteria were (i) diagnosis of NVAF; (ii) high risk of stroke and bleeding (CHA2DS2-VASc score ≥ 2 for men and ≥3 for women); (iii) clinical indications for LAAO, including prior thromboembolism, contraindication to long-term anticoagulation, or increased bleeding risk; and (iv) provision of written informed consent. Exclusion criteria included (i) presence of active LAA thrombus or other intracardiac thrombi; (ii) severe LAA anatomical abnormalities, such as extremely small (<15 mm) or excessively large (>35 mm) ostial diameters, insufficient appendage depth for adequate device compression, or complex multilobed/tubular morphologies precluding stable device anchoring (the exact number of patients excluded for anatomical unsuitability was not systematically recorded, as screening was performed as part of routine clinical practice prior to study enrolment); (iii) active bleeding or known bleeding diathesis, including recent gastrointestinal haemorrhage; (iv) known allergy to WATCHMAN FLX device materials; (v) severe valvular heart disease requiring surgical intervention; and (vi) life expectancy < 1 year. The study was registered at the Chinese Clinical Trial Registry (ChiCTR2400085158), approved by the ethics committees of all participating centres, and conducted in accordance with the Declaration of Helsinki. All enrolled patients provided written informed consent prior to participation. Preoperative preparation Pre-procedural assessment included blood tests and either TEE (Figure 1) or coronary computed tomography angiography (CCTA) within 24 h to evaluate LAA anatomy and exclude contraindications.7,16 For CCTA, the optimal fluoroscopic projection was selected based on reconstructed images, typically at right anterior oblique (RAO) 30° and caudal (CAU) 20°, with adjustments within ranges of RAO 0°−40° and CAU 0°−40° (Figure 2). Representative transoesophageal echocardiographic images showing measurement of the left atrial appendage (LAA) orifice diameter and depth obtained at angles of (A) 0°, (B) 45°, (C) 90°, and (D) 135°. TEE, transoesophageal echocardiography. Preoperative computed tomography angiography demonstrating measurements of the LAA ostium diameter and depth. (A) Cross-sectional view. (B) Oblique coronal view. (C) Orthogonal view corresponding. (D) Two-chamber right anterior oblique/caudal (RAO/CAU) view generated by planar reconstruction. CAU, caudal; LAA, left atrial appendage; RAO, right anterior oblique. Interventional procedure Procedures were performed in catheterization laboratories equipped with standard digital subtraction angiography systems. Under local anaesthesia (lidocaine infiltration at the femoral puncture site) and sterile conditions, femoral venous access was obtained. No intravenous conscious sedation was used; procedures were performed under local anaesthesia only without midazolam or propofol, ensuring patient tolerance. The overall procedural workflow, including transseptal puncture, anticoagulation strategy, device sizing, deployment, and post-implant assessment, followed contemporary best practice principles for LAA closure, as outlined in the international consensus document on LAAC.17 Transseptal puncture at the fossa ovalis was performed under fluoroscopic guidance; immediately after crossing, intravenous heparin (100 IU/kg) was administered, and activated clotting time (ACT) was maintained at ≥250 s. A pigtail catheter was placed into the distal LAA for angiography. Left atrial appendage angiography was performed using RAO 30°/CAU 20° projections (see Supplementary material online, Video S1). The 4 mm calibration marker on the catheter was used to reference ostial diameter and depth measurements, which were compared with pre-procedural TEE/CCTA. Device size for the WATCHMAN FLX was chosen to be 4–6 mm larger than the measured LAA ostial diameter. After deployment, device position was evaluated in three fluoroscopic projections (RAO 30°/CAU 20°, RAO 30°/CRA 20°, and CAU 20°) to confirm alignment (see Supplementary material online, Videos S2–S4) and to ensure that the exposed shoulder did not exceed one-third of the device diameter (Figure 3A–D). Pre-procedural assessment, device sizing, and post-implantation verification for WATCHMAN FLX left atrial appendage closure. (A) Pre-release LAA angiogram (RAO 30°/CAU 20° view) for ostium diameter and depth measurements. (B–D) Post-release angiograms in multiple views verifying device position, deployment, and seal. (E) Implant sizing chart for WATCHMAN FLX device selection based on LAA measurements. (F) Distribution of LAA morphology types within the study cohort. CAU, caudal; LAA, left atrial appendage; RAO, right anterior oblique. Compression ratio was calculated as the percentage reduction in the final deployed device diameter relative to the measured LAA ostial diameter on fluoroscopy. To minimize inter-operator variability, operators used standardized table height, fixed fluoroscopic source-to-image distance, predefined RAO/CAU and RAO/CRA projections, and consistent patient positioning. The target compression ratio was 10–30% (Figure 3E), which aligned with manufacturer recommendations and prior FLX studies. A standardized tug test was then performed. Device stability was confirmed only if the occluder demonstrated strict synchronous movement with the LAA wall, elastic recoil, and no measurable displacement during traction. Before release, the device was required to fulfil the PASS criteria,18,19 which represent the standard assessment framework for WATCHMAN device implantation and are consistent with the recommendations of the international consensus document on LAAC17: position, appropriate placement within the LAA; anchoring, secure tug test without movement; size—compression ratio 10–30%; and seal, absence of significant peri-device leak (PDL < 5 mm). Peri-device leak was defined on fluoroscopy as contrast extravasation from the left atrium into the distal LAA along the device during angiography in both RAO/CAU and RAO/CRA views. Peri-device leak severity was classified as none, mild–moderate (1 to <5 mm), or severe (≥5 mm). Procedural success required successful deployment meeting all PASS criteria without major complications (death, cardiac tamponade, or device embolization). Postoperative management and follow-up Peri-procedural complications were prespecified and included in-hospital death, cardiac perforation, pericardial effusion or tamponade, stroke, transient ischaemic attack (TIA), device-related thrombus, embolization, access-site complications, and bleeding events. Clinical and imaging follow-up were performed according to routine practice at each centre, with the imaging modality (TEE or CCTA) explicitly selected by the treating physician to ensure consistency with site-specific protocols. Antithrombotic therapy after the procedure was individualized based on ischaemic and bleeding risk, with direct oral anticoagulants (most commonly rivaroxaban) recommended as the first-line therapy. The duration of therapy varied, with some patients transitioning to single or dual antiplatelet therapy after 1–3 months based on bleeding risk and physician assessment. Follow-up typically occurred at 1–3 months and included both clinical evaluation and imaging assessments. Significant PDL was defined as a jet width ≥ 5 mm on either TEE or CCTA, according to established consensus.17,20,21 The composite ‘adverse events’ endpoint included all-cause death, stroke/TIA, major bleeding, device-related thrombosis, and severe PDL (≥5 mm). While rare events such as device-related thrombosis (DRT) and severe PDL had minimal impact on regression estimates, TIA and death contributed most to the associations observed in Cox models, allowing a comprehensive assessment of clinically relevant safety outcomes. Of the 400 consecutively enrolled patients, 348 (87.0%) underwent follow-up imaging (TEE or CCTA) and were included in imaging endpoint analyses. The remaining 52 patients did not undergo imaging follow-up but were included in clinical outcome analyses for stroke, TIA, and death. Study endpoints The primary efficacy endpoint was complete LAA occlusion, defined as the absence of significant PDL (<5 mm) on follow-up imaging (TEE or CCTA). The primary safety endpoint was a composite of major peri-procedural complications, including all-cause death, stroke, cardiac tamponade, and device embolization. Secondary endpoints included device-related thrombus, clinically significant bleeding (as defined by International Society on Thrombosis and Haemostasis criteria), PDL > 5 mm, and long-term all-cause mortality. Statistical methods All statistical analyses were performed using SPSS version 29.0 (IBM Corp., Armonk, NY). Continuous variables are presented as mean ± standard deviation (SD) for normally distributed data or as median with interquartile range (IQR) for non-normal data. Categorical variables are expressed as counts and percentages. Given the single-arm, prospective registry design, descriptive statistics were primarily used to summarize baseline characteristics, procedural parameters, and follow-up outcomes. Associations between baseline variables and adverse events during follow-up were evaluated using univariable Cox proportional hazards regression. Variables with a P value of <0.10 in univariable analysis or deemed clinically relevant were considered for multivariable modelling. To limit overfitting due to the low number of events (n = 20), the final multivariable Cox model was prespecified to include only two clinically prioritized covariates: chronic obstructive pulmonary disease and CHA2DS2-VASc score (entered as a continuous variable). A sensitivity analysis was performed to evaluate whether the choice of follow-up imaging modality (TEE vs. CCTA) affected event detection, using Fisher’s exact test. Statistical significance was defined as a two-sided P < 0.05. Subgroup analysis based on LAA morphology (chicken-wing, cauliflower, cactus, windsock) was performed to evaluate its impact on procedural parameters. Continuous variables were compared using the Kruskal–Wallis test [median (IQR)] due to non-normal distribution; categorical variables were compared using Pearson χ2 test, or Fisher’s exact test if expected cell counts were small. Results Baseline characteristics Among 400 high-risk NVAF patients enrolled across five centres (Table 1), the median age was 70.5 years (IQR 64.0–76.6), 49.8% were male, the median CHA2DS2-VASc score was 4.0 (IQR 3.0–5.0), and the median HAS-BLED score was 2.0 (IQR 2.0–3.0). Prior stroke/TIA was present in 44.8% of patients. Baseline characteristics were broadly similar between patients with and without imaging follow-up (see Supplementary material online, Table S1), suggesting minimal selection bias. Table 1Patient baseline characteristicsVariableValueAge, yrs70.5 (64.0–76.6)Male199 (49.8)BMI, kg/m224.19 ± 3.32Heart failure185 (46.3)Ischaemic heart disease177 (44.3)Hypertension276 (69.0)Diabetes104 (26.0)Hyperlipidaemia60 (15.0)History of stroke/TIA179 (44.8)Chronic obstructive pulmonary disease39 (9.8)Arterial vascular disease193 (48.3)Previous relevant bleeding53 (13.3)CHA2DS2-VASc score4.0 (3.0–5.0)HAS-BLED score2.0 (2.0–3.0)LAD, mm39.96 ± 6.58LVEF, %58.0 (54.0–62.0)INR1.2 (1.0–1.5)NT-proBNP level, pg/mL985.0 (392.9–2284.8)Creatinine, μmol/L79.3 (68.0–99.8)Platelet count, ×109/L177.5 (143.0–227.0)Values are mean ± SD, median (IQR), or n (%).BMI, body mass index; INR, international normalized ratio; LAD, left atrial diameter; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; TIA, transient ischaemic attack. Patient baseline characteristics Values are mean ± SD, median (IQR), or n (%). BMI, body mass index; INR, international normalized ratio; LAD, left atrial diameter; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; TIA, transient ischaemic attack. Pre-procedural imaging and procedural details Pre-procedural imaging used TEE in 58.8% and CCTA in 41.3% of patients. The median LAA landing zone diameter was 26 mm (IQR 22.5–27.9). Device sizes were 31 mm (36.5%), 27 mm (23%), and 35 mm (22%), achieving a median device compression of 14.1% (IQR 12.0–17.1). The median procedure time was 43 min (IQR 32–53), and the median radiation exposure was 272 mGy·cm2 (IQR 186–372) (Table 2). Table 2Pre-procedural imaging and procedural detailsVariableValuePeri-procedural imaging TEE235 (58.8) CCTA165 (41.3) LAA landing zone, mm26.0 (22.5–27.9)Device details Device compression, %14.1 (12.0–17.1) Device size, mm 2018 (4.5) 2456 (14.0) 2792 (23.0) 31146 (36.5) 3588 (22.0)Procedural parameters Procedure time, min43.0 (32.0–53.0) Radiation exposure dose, mGy·cm2272 (186.0–372.0) Fluoroscopy time, min7 (5.0–8.0) Contrast volume, mL65 (58–80)Values are presented as median (IQR) or n (%)CCTA, coronary computed tomography angiography; LAA, left atrial appendage; LAAO, left atrial appendage occlusion; TEE, transoesophageal echocardiography. Pre-procedural imaging and procedural details Values are presented as median (IQR) or n (%) CCTA, coronary computed tomography angiography; LAA, left atrial appendage; LAAO, left atrial appendage occlusion; TEE, transoesophageal echocardiography. Procedural outcomes and in-hospital complications Procedural success was achieved in all 400 cases (100%). Procedural success and complication rates were consistent across participating centres, with no statistically significant between-centre differences in procedural success, complication rates, procedure time, fluoroscopy time, radiation dose, or contrast volume (all P > 0.05; Supplementary material online, Table S2). Correlation between intra-procedural fluoroscopic peri-device leak and follow-up imaging Immediate fluoroscopic assessment after device release revealed no PDL in 374 patients (93.5%), a 1–5 mm leak in 26 patients (6.5%), and no leak ≥ 5 mm (Table 3). Among the 26 patients with mild to moderate PDL (1 to <5 mm) on fluoroscopy, 19 patients had no significant PDL on follow-up TEE or CCTA, six patients had persistent mild to moderate PDL (1 to <5 mm), and one patient did not undergo follow-up imaging; thus, the progression could not be assessed. Left atrial appendage morphology was chicken-wing in 158 (39.5%), cauliflower in 104 (26.0%), cactus in 84 (21.0%), and windsock in 54 (13.5%) (Figure 3F). There were no in-hospital deaths, cerebrovascular accidents, device embolization, or device-related thrombosis. Major in-hospital complications included access-site haematoma in five (1.3%) and pericardial effusion in two (0.5%); none required pericardiocentesis. Major bleeding occurred in three patients (0.8%), all gastrointestinal (melena). Rivaroxaban was the predominant postprocedural antithrombotic regimen (306/400, 76.5%) (see Supplementary material online, Table S3). Table 3Procedural outcomes and complicationsVariableValueProcedural success400 (100)Peri-device leakage (immediate) ≥5 mm0 (0) 1–5 mm26 (6.5) 0 mm374 (93.5)Peri-procedural complications Haemorrhage3 (0.8) Pericardial effusion2 (0.5) Access-site complications5 (1.3)Values are presented as n (%). Procedural outcomes and complications Values are presented as n (%). Follow-up outcomes Only patients with follow-up imaging (n = 348, 87.0%) were included in analyses of imaging endpoints. Patients without imaging follow-up (n = 52) were excluded from imaging analyses to avoid bias; clinical outcome analyses included all 400 patients for TIA/stroke and mortality endpoints. The median follow-up was 194.5 days (IQR 129.0–265.0). Follow-up imaging was available in 348/400 (87.0%): 171/348 (49.1%) by TEE and 177/348 (50.9%) by CCTA; 52 patients were followed by telephone. During follow-up, 3/400 (0.8%) died (one from severe pulmonary infection, one from trauma, one unknown). No ischaemic strokes occurred; 13/400 (3.3%) had TIA. Device-related thrombosis, defined according to contemporary consensus criteria for LAAC imaging surveillance,17 occurred in 1/400 (0.3%), major bleeding in 3/400 (0.8%), and no peri-device flow ≥ 5 mm (Table 4). Table 4Follow-up outcomesFollow-up time, days, median (IQR)194.5 (129.0–265.0)Imaging technique348 (87.0)TEE171 (49.1)CCTA177 (50.9)Death3 (0.8)Stroke/TIA13 (3.3)Device-related thrombosis (DRT)1 (0.3)Major bleeding3 (0.8)Peri-device leakage ≥ 5mm0 (0)Values are presented as n (%). Follow-up outcomes Values are presented as n (%). Predictors of adverse events Candidate predictors of adverse events are shown in Table 5, and Cox proportional hazards regression results are presented in Table 6. Variables significantly associated with adverse events included age, diabetes, chronic obstructive pulmonary disease, prior stroke/TIA, peripheral arterial disease, and higher CHA2DS2-VASc and HAS-BLED scores. In the prespecified multivariable model (limited covariates due to low event count), chronic obstructive pulmonary disease (HR 2.86; 95% CI 1.98–4.11; P < 0.001) and CHA2DS2-VASc score (HR 4.69 per point; 95% CI 1.90–11.57; P < 0.001) were independent predictors. Table 5Distribution of candidate predictors among patients with and without adverse eventsVariableOverall cohort(N = 400)Adverse events (N = 20)Without adverse events (N = 380) P valueAge ≥ 75 years136 (34.0)13 (65.0)123 (32.4)0.003Heart failure185 (46.3)12 (60.0)173 (45.5)0.206Hypertension276 (69.0)13 (65.0)263 (69.2)0.691Diabetes104 (26.0)9 (45.0)95 (25.0)0.047Hyperlipidaemia60 (15.0)6 (30.0)54 (14.2)0.054Chronic obstructive pulmonary disease39 (9.8)8 (40.0)31 (8.2)<0.001History of stroke/TIA179 (44.8)16 (80.0)163 (42.9)0.001Arterial vascular disease193 (48.3)14 (70.0)179 (47.1)0.046CHA2DS2-VASc score ≥ 5175 (43.8)18 (90.0)157 (41.3)<0.001HAS-BLED score ≥ 3158 (39.5)14 (70.0)144 (37.9)0.004Values are presented as n (%).Variables presented in this table were considered as potential predictors in univariable analysis. Distribution of candidate predictors among patients with and without adverse events Values are presented as n (%). Variables presented in this table were considered as potential predictors in univariable analysis. Table 6Cox proportional hazards regression analysis of factors associated with adverse eventsFactorUnivariable HR (95% CI) P valueMultivariable HR (95% CI) P valueAge ≥ 75 years3.98 (1.58–9.99)0.003Heart failure1.88 (0.77–4.61)0.167Hypertension0.88 (0.35–2.20)0.780Diabetes2.63 (1.09–6.36)0.032Hyperlipidaemia2.26 (0.87–5.90)0.095Chronic obstructive pulmonary disease5.00 (2.04–12.45)<0.0012.86 (1.98–4.11)<0.001History of stroke/TIA3.94 (1.31–11.82)0.015Arterial vascular disease3.00 (1.15–7.84)0.025CHA2DS2-VASc score2.81 (2.00–3.94)<0.0014.69 (1.90–11.57)<0.001HAS-BLED score1.23 (0.85–1.76)0.272 Cox proportional hazards regression analysis of factors associated with adverse events Sensitivity analysis Of the 348 patients who completed follow-up imaging (TEE, n = 171; CCTA, n = 177), no cases of significant PDL (≥5 mm) were identified in either group (P = 1.000). Device-related thrombosis was detected in one patient (0.6%) in the CCTA group and in none in the TEE group (P = 1.000). The incidence of TIA/stroke did not differ significantly between groups [TEE: 8/171 (4.7%) vs. CCTA: 5/177 (2.8%); P = 0.408]. Overall, no statistically significant differences were observed between imaging modalities (see Supplementary material online, Table S4). Left atrial appendage morphology and procedural parameters Subgroup analysis based on LAA morphology revealed that fluoroscopy time varied significantly among morphological types (P = 0.036; Supplementary material online, Table S5). Patients with cauliflower-type LAA had the shortest median fluoroscopy time (6.0 min), whereas those with windsock type had the longest (7.0 min). In contrast, no significant differences were observed across the four morphology groups in total procedure time (P = 0.167), radiation dose (P = 0.626), contrast volume (P = 0.678), or incidence of immediate PDL (P = 0.743). Discussion The ROSE-FLX study represents, to our knowledge, the first prospective, multicentre, real-world registry evaluating WATCHMAN FLX implantation performed exclusively under fluoroscopic guidance. Procedural success, defined as fulfilment of PASS criteria and absence of major intra-procedural complications, was achieved in 100% of patients. Peri-procedural adverse events—including access-site haematoma (1.3%), pericardial effusion (0.5%), and major bleeding (0.8%)—were infrequent and generally lower than those reported in prior echocardiography-guided studies.22,23 Imaging endpoint analyses were restricted to the 348 patients with complete follow-up. Given the single-arm design without a contemporary TEE- or ICE-guided control group, causal inference is limited; the findings should be considered hypothesis generating rather than practice changing. Device stability depends on accurate positioning and anchoring, while adequate compression and reliable PDL assessment are critical for success. Earlier EHRA/EAPCI consensus recommended TEE, often under general anaesthesia, for LAAC implantation.17 Recent 2023 guidelines24 allow either TEE or ICE, and note that ICE may be chosen to avoid general anaesthesia, depending on patient preference and institutional resources. In resource-limited settings, standardized fluoroscopic workflows—including predefined imaging projections, repeated angiography, and objective compression targets—can effectively substitute for real-time echocardiography. In our study, repeated fluoroscopic angiography and standard RAO/CAU and RAO/CRA projections allowed precise device placement and assessment of compression and seal, consistent with prior reports.25 This approach provides clear visualization of the LAA ostium and landing zone structures, enabling device positioning and stability assessment comparable to TEE. In addition, the anatomical complexity of the LAA may pose procedural challenges, with the cactus type being the most intricate. Despite no significant impact of LAA morphology on procedural duration, contrast volume, or success—consistent with prior evidence that highlights operator experience and device selection26—fluoroscopy time nonetheless varied significantly (P = 0.036), suggesting that anatomical complexity specifically challenges the fluoroscopic workflow. Consistent with previous comparative device studies27and a recent prospective study on fluoroscopy-only WATCHMAN FLX implantation.28 Our findings further confirm the safety and feasibility of LAAO performed independently by experienced operators under fluoroscopic guidance. Beyond feasibility, the ROSE-FLX registry provides clinical insights into predictors of adverse events after sLAAO. In multivariable analysis (with model covariates prespecified to limit overfitting given the low event count), chronic obstructive pulmonary disease and higher CHA2DS2-VASc scores were independently associated with adverse outcomes.29–31 These associations are biologically plausible—chronic obstructive pulmonary disease may be linked to systemic inflammation, endothelial dysfunction, and delayed endothelialization29 and higher CHA2DS2-VASc reflects greater baseline thromboembolic risk30,31—but our findings should be interpreted cautiously due to the limited number of events and potential residual confounding. In this study, the incidence of DRT was 0.3%, lower than the 1–4% reported in the EWOLUTION and NCDR LAAO registries.32,33 As highlighted in the international LAAC consensus by Potpara et al.,17 the reported incidence of DRT and PDL may be influenced by the imaging modality used for detection. Coronary computed tomography angiography provides higher spatial resolution and may identify small peri-device leaks or subclinical thrombi that could be underestimated by TEE.21 In the present study, follow-up imaging was performed using either TEE or CCTA. Sensitivity analysis showed no significant differences between TEE and CCTA. These modality-related differences, along with individualized antithrombotic therapy, should be considered when interpreting the low incidence of device-related complications. Notably, the rate of PDL was also low, markedly superior to early reports with the WATCHMAN device.34 These lower rates of DRT and PDL are likely related to structural optimizations of the WATCHMAN FLX device. The FLX features a shorter device height, enhanced fabric coverage, and a fully closed distal end and requires less implantation depth. Its flat proximal face includes a recessed delivery cable screw insert to enhance endothelialization and reduce DRT, and the increased number of frame struts improves conformity and device–atrial wall contact, thereby minimizing PDL.35,36 No severe PDL (≥5 mm) was observed by TEE or CCTA. This agreement is consistent with reliable exclusion of clinically significant residual leaks when performed using a standardized workflow. Although one case of DRT was identified on follow-up CCTA, no statistically significant differences were observed between imaging modalities. These findings support the reliability of fluoroscopy-guided implantation while reinforcing the value of structured follow-up imaging surveillance. Importantly, reliance on fluoroscopy alone did not compromise technical performance or safety. Instead, the procedural success observed in ROSE-FLX reflects a combination of operator proficiency, device improvements, and a rigorously standardized fluoroscopic workflow. Gilhofer et al.14 similarly reported that fluoroscopy-only LAAO can achieve long-term outcomes comparable to those obtained under echocardiographic guidance, reinforcing the viability of this method in resource-limited settings. Although procedural volumes varied among centres, outcomes were consistent across sites. All participating centres had prior experience in LAAO implantation, and no formal learning phase was required. These findings suggest that, when performed by trained operators using a standardized fluoroscopic workflow, the simplified LAAO approach can achieve reproducible outcomes across centres with varying procedural volumes. Despite these strengths, the absence of a contemporaneous TEE- or ICE-guided comparison arm represents an important limitation of this prospective single-arm registry and precludes formal assessment of non-inferiority between fluoroscopy-only and imaging-guided LAAO strategies. Because ROSE-FLX did not include a parallel control cohort treated under conventional TEE- or ICE-guided workflows, propensity score–matched analyses were not feasible within the current dataset. Conducting such an analysis using external historical datasets would introduce substantial selection bias due to differences in baseline characteristics, procedural protocols, operator experience, and endpoint adjudication. To contextualize our outcomes, historical reference cohorts from the ITALIAN-FLX registry and the PINNACLE FLX trial demonstrate comparable success, complication, and DRT rates, supporting the plausibility that fluoroscopic-only workflows may achieve similar results. However, these cross-study comparisons should be interpreted cautiously given inherent inter-study heterogeneity. Nevertheless, prospective comparative studies incorporating contemporaneous control groups, ideally using matched or randomized designs, are warranted to strengthen interpretability. Additionally, although all procedures in this registry were performed without ultrasound guidance, emerging use of ICE or paediatric TEE enables non–general anaesthesia workflows while preserving real-time device–tissue visualization. These modalities should be acknowledged as viable alternatives; however, their availability and cost remain prohibitive in many regions, reinforcing the relevance of a purely fluoroscopic approach. Clinical significance and limitations Several notable limitations should be noted. First, despite involving several high-volume centres, all procedures were performed by experienced LAAO specialists, which may reduce the generalizability of the findings to smaller centres. Second, the follow-up period in this study was relatively short (median 194.5 days, ∼6 months), which may be insufficient to capture late complications, such as delayed DRT, PDL, and incomplete endothelialization.21 Subclinical DRT or partial endothelialization cannot be entirely ruled out, and longer-term imaging surveillance would be required to fully assess thromboembolic protection and device-related outcomes. Additionally, this relatively short follow-up did not allow assessment of mortality related to patient profile and comorbidities, as highlighted in the EWOLUTION study, which identified patient characteristics and baseline comorbidities as important predictors of 2-year post-LAAC mortality.37 Third, follow-up imaging was performed using either TEE or CCTA. Although sensitivity analysis showed no significant differences in detection of DRT or PDL ≥ 5 mm, CCTA’s higher spatial resolution may detect small PDLs or subclinical thrombi missed by TEE. These modality-related differences should be considered when interpreting the low observed incidence of device-related complications. Finally, postoperative antithrombotic therapy was tailored to individual risk profiles. Most patients (306/400, 76.5%) received rivaroxaban for 1–3 months post-procedure, followed by single or dual antiplatelet therapy depending on bleeding risk and physician judgment. A minority received initial dual antiplatelet therapy or warfarin per comorbidity or contraindication. This strategy was consistent with contemporary guideline recommendations for post-LAAO antithrombotic management,7,24 in which short-term oral anticoagulation followed by risk-adapted de-escalation is considered reasonable in the absence of device-related thrombosis or significant PDL. During follow-up, therapy adjustments were individualized based on imaging findings and thromboembolic and bleeding risk profiles. Therefore, our results primarily reflect outcomes under a guideline-directed antithrombotic strategy rather than systematic early discontinuation of oral anticoagulation, which may have contributed to the low observed incidence of DRT (0.3%). Additionally, not all patients completed on-site follow-up visits, and some data were obtained via telephone follow-up, possibly introducing recall or reporting bias. While 348 of 400 patients had follow-up imaging, all 400 patients were followed clinically for stroke, TIA, and death. Imaging-derived outcomes (DRT/PDL) are reported for the entire cohort for transparency, though sensitivity analyses were limited to the 348 patients with imaging follow-up. Baseline characteristics between patients with and without imaging follow-up were generally similar, minimizing potential selection bias. However, missing imaging data may still limit detection of subclinical device-related complications. Furthermore, the number of patients excluded due to anatomical unsuitability was not prospectively recorded, which may limit precise quantification of potential selection bias. Conclusions Overall, the findings support sLAAO as a safe, effective, and resource-conscious alternative to standard echocardiography-guided LAAO procedures, particularly in centres lacking advanced imaging or anaesthetic support. Additional large-scale, randomized, and long-term studies are needed to confirm safety and efficacy, enhance patient selection, and establish standardized antithrombotic protocols. Supplementary Material Acknowledgements The authors thank all participating centres and clinical staff for their contribution to patient recruitment, procedure performance, and follow-up data collection. The authors also express their gratitude to the patients who participated in this study. Contributor Information Wei Zhou, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Qian Tang, Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China; The School of Clinical Medicine, Soochow University, Suzhou, Jiangsu 215006, China. Anning Zeng, Department of Cardiology, The People’s Hospital of Qiannan, Duyun, Guizhou 558000, China. Jiulin Chen, Department of Cardiology, People’s Hospital of Qianxinan Buyi, Xingyi, Guizhou 562400, China. Fang Wang, Department of Cardiology, The People’s Hospital of Qiandongnan Miao and Dong Autonomous Prefecture, Kaili, Guizhou 556000, China. Jing Wang, Department of Cardiology, The People’s Hospital of Panzhou City, Panzhou, Guizhou 561600, China. Jingjing Chen, Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. LinYan Shi, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Ji Li, Department of Cardiology, The Second Affiliated Hospital of Guizhou Medical University, Kaili, Guizhou 556000, China. Yirong Zhang, Department of Cardiology, The People’s Hospital of Liupanshui City, Liupanshui, Guizhou 553000, China. Dawei Tian, Department of Cardiology, Qiannan Prefecture Hospital of Traditional Chinese Medicine, Duyun, Guizhou 558000, China. Dandan Chen, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Fuliang Luo, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Zheyi Wu, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Na Deng, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Fei Yan, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, People’s Hospital of Qianxinan Buyi, Xingyi, Guizhou 562400, China. Qianjiang Pan, Department of Cardiology, The Cengong County People’s Hospital, Cengong, Guizhou 557800, China. Zhongrong Liao, Department of Cardiology, The People’s Hospital of Luodian County, Luodian, Guizhou 550100, China. Yongxian Tian, Department of Cardiology, The People’s Hospital of Kaiyang County, Kaiyang, Guizhou 550300, China. Yifan Mao, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Huimin Lu, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Xing Zhou, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Ya Yang, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. ZhenFa Zhang, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Runze Huang, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China. Wei Li, Department of Clinical Medicine, Guizhou Medical University, No. 9 Beijing Road, Yunyan District, Guiyang, Guizhou 550004, China; Department of Cardiology, The Affiliated Hospital of Guizhou Medical University, No. 28 Beijing Road, Guiyang, Guizhou 550004, China; The School of Clinical Medicine, Soochow University, Suzhou, Jiangsu 215006, China. Supplementary material Supplementary material is available at Europace online. Funding This work was supported by the Fund Project of the Guizhou Provincial Health Commission (No. gzwkj2020-1-070). Data availability The data that support the findings of this study are available from the corresponding author, upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement The data that support the findings of this study are available from the corresponding author, upon reasonable request.