Cobra-head and other shape-memory abnormalities of nitinol atrial septal occluders: incidence, predisposing factors, and outcomes.

BACKGROUND

Shape-memory abnormalities are seen in some nitinol atrial septal occluders. Variably described as cobra-head, tulip, and others, their incidence, mechanisms, clinical impact, and outcome have not been systematically analysed.

METHODS

We retrospectively reviewed all consecutive device closures in the last 6 years for deformations. Type and size of the occluder, deployment technique, size, and angulation/kinking of the delivery sheath were analysed. Procedural success, duration, and other complications were studied.

RESULTS

A total of 112 devices (11.8%) among 950 occluders used in 936 patients showed deformities. Fourteen of 936 received 2 devices. Deformities were transient and self-correcting in 40%. Multivariate analysis showed significant associations with oversized sheaths (p = 0.004), kinked/angulated sheaths (p < 0.001), special deployment techniques (p < 0.001), and twist in the device waist (p = 0.011). Despite more frequent deformities with Figulla (15.6%) and Amplatzer (13.9%) occluders than Cera occluders (6.6%) and larger devices (>24 mm - 14.6%) than smaller devices (less than or equal to 24 mm - 9.7%), they were not significant on multivariate analysis. In vivo manipulations corrected most deformities; nineteen needed in vitro reformations and four needed a change of device. Despite prolongation of the procedure, repeated attempts (mean 2.76 ± 1.7 attempts, with a range from 1 to 9 attempts), and supraventricular tachycardia in two patients, there were no serious adverse effects.

CONCLUSIONS

Deformations were frequent in 11.8% of atrial septal occluders on a targeted search. Oversized and angulated/kinked sheaths, special techniques like pulmonary vein deployment and twist in device waist during procedure predisposed to deformities. While most deformities were corrected with manipulations, removal of the device was infrequently needed and change of device was rarely required. Long procedural time and multiple attempts for deployment did not affect procedural success.