Can transesophageal echocardiography be safely deferred? A low-cost predictive model using D-dimer to exclude left...

Transesophageal echocardiography (TEE) is the reference standard for excluding left atrial thrombus (LAT) in rheumatic severe mitral stenosis (SMS), yet its availability is limited in many low-resource settings. A reliable, low-cost strategy to safely defer TEE would have a significant clinical impact. This study evaluated whether D-dimer-based predictive models can accurately exclude newly detected LAT in treatment-naïve patients with rheumatic SMS.

Sixty-eight consecutive patients with rheumatic SMS who were not receiving anticoagulation underwent D-dimer testing, transthoracic echocardiography, and TEE. Associations with LAT were assessed using conventional and Firth-penalized logistic regression. Diagnostic performance was evaluated using receiver operating characteristic analysis with 1,000-bootstrap internal validation. Two composite models were examined: D-dimer combined with atrial fibrillation (AF) and D-dimer combined with mitral valve area (MVA).

LAT was newly detected in 12 patients (17.6%). D-dimer levels were significantly higher in LAT-positive patients (943.5 vs. 168.5 ng/mL, p < 0.001). The optimal D-dimer cut-off (360.6 ng/mL) achieved 100% sensitivity, 71.4% specificity, and an optimism-corrected AUC of 0.935, outperforming the conventional 500 ng/mL threshold (sensitivity 75%). In adjusted models, D-dimer remained a strong independent predictor of LAT (per SD: AOR 6.91; 95% CI 2.72-17.54). Composite models demonstrated similarly high discrimination (AUC 0.933-0.940); adding AF did not improve performance, whereas incorporating MVA increased specificity to 100%.

D-dimer may help exclude LAT in treatment-naïve patients with rheumatic SMS. Its potential role in supporting more selective TEE use in resource-limited settings warrants further evaluation in larger prospective multicenter studies.