Gradient changes in bioprosthetic valve thrombosis: Duration of anticoagulation and strategies to improve detection

Objective Bioprosthetic valve thrombosis (BPVT) is increasingly recognised as a major cause of prosthetic dysfunction in the first years postimplantation. How early abnormal gradients can be detected prior to diagnosis and how fast they normalise with anticoagulant therapy is unknown. We set forth to (1) evaluate patterns of increase in gradients prior to BPVT diagnosis and (2) characterise time-course of response to anticoagulation. Methods Patients treated with warfarin for BPVT (1999-2019) with clinically significant reduction of mean gradients (≥25%) were identified retrospectively. Recovery was defined as gradient decrease ≥50%, to postimplantation or to normal-range gradients per position, model and size. Time-to-BPVT (implantation-BPVT diagnosis), potential diagnostic delay (first abnormal gradient by position, model and size-BPVT diagnosis) and time-to-recovery (BPVT diagnosis-complete resolution) were recorded. Results 77 patients were identified; 32 (42%) aortic (23 surgical-12 porcine, 11 pericardial; 9 transcatheter); 24 (31%) mitral; 21 (27%) tricuspid. Median time-to-BPVT was 24, 21 and 10 months, respectively. Potential diagnostic delay was median 21 months for aortic, 4 months for mitral, but 0 for tricuspid. Recovery was significantly faster in mitral than aortic (median 2.5 vs 4.8 months, p=0.038) and tricuspid (median 5.9 months, p=0.025) positions. Porcine aortic valves responded faster than pericardial aortic valves (median 2.9 vs 20.3 months, p=0.004). Conclusion Gradients start to increase months before the clinical BPVT diagnosis. Recovery is faster in mitral and surgical aortic porcine valves; a longer warfarin trial seems indicated in tricuspid and surgical aortic pericardial valves.