Discussion

This study is one of the largest single centre study aimed at identifying risk factors that lead to device embolisation after transcatheter duct closure. Four percent of procedures were complicated by device embolisation, which was similar to the world literature. ^{Reference Kang, Jivanji and Mehta7–Reference El-said, Bratincsak and Foerster10} A retrospective analysis of 408 consecutive procedures done over 11 years from 13 institutions was the previous major report of the complications following transcatheter duct closure, where the embolisation rates were similar. ^{Reference Backes, Rivera and Bridge8} Device embolisation often occurred immediately, but late diagnosis as late as 6 weeks was also rarely reported in literature. ^{Reference Kang, Jivanji and Mehta7,Reference Backes, Rivera and Bridge8}

Most embolisations occurred in infants with small body size in our study. A meta-analysis of 38 studies involving transcatheter duct closure in infants reported 5% incidence of device embolisations, with statistically significant increase when the infants were < 6 kg in weight. ^{Reference Backes, Rivera and Bridge8} A three-fold increase in adverse events and five-fold increase in serious adverse events were observed in patients < 6 kg in another multi-centre registry. ^{Reference Tan, Levi and Moore9,Reference El-said, Bratincsak and Foerster10} Use of Amplatzer duct occluder is off-label in infants < 6 kg, though multiple authors reported its safe and effective use. ^{Reference Sivakumar, Francis and Krishnan11}

Embolisation occurred more often to pulmonary artery rather than aorta, similar to our experience. While reporting the procedural serious adverse events, embolisation with successful transcatheter retrieval was classified as a moderate event and that needing surgical retrieval was classified under major event. ^{Reference Kang, Jivanji and Mehta7} Eighty percent of embolised duct occluders had been percutaneously retrieved in published literature, while the others needed surgery. ^{Reference Tan, Levi and Moore9} Three patients in our group among the 15 embolisations needed surgical retrieval.

Embolisation to aorta increases afterload to left ventricle and increases left-to-right ductal shunt, along with risks of bowel gangrene and kidney injury. Embolisation to pulmonary artery reduces cardiac output. ^{Reference Faella and Hijazi12} Percutaneous retrieval of devices carries risk of vascular trauma, tricuspid valve injury, or compartment syndrome. ^{Reference Jang, Son, Lee, Lee and Kim13} There was no mortality in our group of patients. Two devices could not be retrieved as they lodged deep within the small pulmonary lobar branches. A collective heart team decision was made to conservatively observe due to their significant comorbidities and high pulmonary vascular resistance.

Our study observed statistically higher rate of embolisation with the new generation of softer devices such as vascular plugs and Piccolo device, similar to others experience. ^{Reference Kang, Jivanji and Mehta7–Reference Tan, Levi and Moore9} This could be attributed to difficulty in correctly assessing the dimensions of complex ductal anatomy or a learning curve with the recently introduced softer devices and vascular plugs. Device embolisation was significantly higher in those where the device was undersized, which was a known risk factor. ^{Reference El-said, Bratincsak and Foerster10} While four among the eight vascular plugs were undersized in patients who had embolisation, all the 17 plugs used in the other patients who did not have embolisation were selected more than 50% larger than the duct diameter. Unlike older patients, device was not always oversized 2 mm larger than the ductal dimensions in infants weighing under 6 kg to avoid device protrusion into the aorta or pulmonary artery. Risk of device-related obstruction of left pulmonary artery or descending aorta was higher in small infants, though improvement might occur over time with vascular growth. ^{Reference Kang, Jivanji and Mehta7,Reference Tomasulo, Gillespie and Munson14} Softer vascular plugs and Piccolo devices were often chosen to facilitate intraductal deployment, thereby circumventing vessel obstruction. Left pulmonary artery obstruction might be underestimated on echocardiography due to preferential redistribution to the right lung. ^{Reference Kharouf, Heitschmidt and Hijazi15} One among the 15 patients in this group had a successful larger Amplatzer vascular plug II deployment after retrieval of a 6 mm Amplatzer vascular plug IV plug. Left pulmonary artery occlusion observed on late follow-up could not be rectified and it additionally led to marked hypoplasia of the hilar branches.

The reasons behind device embolisation in young infants include inadequate echocardiographic images, device undersized intentionally to avoid protrusion into adjacent vessels, ductal spasm, tension on delivery system, operator related factors such as forward push of the cable, delay in time to release or inadvertent unscrewing, vigorous patient activity, or a complex duct morphology. ^{Reference Sathanandam, Gutfinger and Morray4,Reference Kang, Jivanji and Mehta7} Four patients with pneumonia, lung disease, and significant lung hyperinflation had suboptimal echocardiographic images that could have contributed for the event. Intentional device undersizing was noted in 8 patients where either a chosen duct occluder was similar in size to the duct dimension or the selected vascular plug was less than 150% of the duct diameter. Operator-related factors could not be deciphered in this retrospective analysis. However, there was one instance, where a catheter advanced into the pulmonary artery to record pressures inadvertently dislodged a device. Patient-related factor such as physical movements could be an additional factor as 13 out of the 15 procedures were done on conscious monitored sedation under ketamine.