RESUMO
OBJECTIVES: The gold standard metric for centre-level performance in orthotopic heart transplantation (OHT) is 1-year post-OHT survival. However, it is unclear whether centre performance at 1 year is predictive of longer-term outcomes. This study evaluated factors impacting longer-term centre-level performance in OHT. METHODS: Patients who underwent OHT in the USA between 2010 and 2021 were identified using the United Network of Organ Sharing data registry. The primary outcome was 5-year survival conditional on 1-year survival following OHT. Multivariable Cox proportional hazard models assessed the impact of centre-level 1-year survival rates on 5-year survival rates. Mixed-effect models were used to evaluate between-centre variability in outcomes. RESULTS: Centre-level risk-adjusted 5-year mortality conditional on 1-year survival was not associated with centre-level 1-year survival rates [hazard ratio: 0.99 (0.97-1.01, P = 0.198)]. Predictors of 5-year mortality conditional on 1-year survival included black recipient race, pre-OHT serum creatinine, diabetes and donor age. In mixed-effect modelling, there was substantial variability between centres in 5-year mortality rates conditional on 1-year survival, a finding that persisted after controlling for recipient, donor and institutional factors (P < 0.001). In a crude analysis using Kaplan-Meier, the 5-year survival conditional on 1-year survival was: low volume: 86.5%, intermediate volume: 87.5%, high volume: 86.7% (log-rank P = 0.52). These measured variables only accounted for 21.4% of the between-centre variability in 5-year mortality conditional on 1-year survival. CONCLUSIONS: Centre-level risk-adjusted 1-year outcomes do not correlate with outcomes in the 1- to 5-year period following OHT. Further research is needed to determine what unmeasured centre-level factors contribute to longer-term outcomes in OHT.
RESUMO
PURPOSE: To evaluate a novel deep learning (DL)-based automated coronary labeling approach for structured reporting of coronary artery disease according to the guidelines of the Society of Cardiovascular Computed Tomography (CT) on coronary CT angiography (CCTA). PATIENTS AND METHODS: A retrospective cohort of 104 patients (60.3 ± 10.7 y, 61% males) who had undergone prospectively electrocardiogram-synchronized CCTA were included. Coronary centerlines were automatically extracted, labeled, and validated by 2 expert readers according to Society of Cardiovascular CT guidelines. The DL algorithm was trained on 706 radiologist-annotated cases for the task of automatically labeling coronary artery centerlines. The architecture leverages tree-structured long short-term memory recurrent neural networks to capture the full topological information of the coronary trees by using a two-step approach: a bottom-up encoding step, followed by a top-down decoding step. The first module encodes each sub-tree into fixed-sized vector representations. The decoding module then selectively attends to the aggregated global context to perform the local assignation of labels. To assess the performance of the software, percentage overlap was calculated between the labels of the algorithm and the expert readers. RESULTS: A total number of 1491 segments were identified. The artificial intelligence-based software approach yielded an average overlap of 94.4% compared with the expert readers' labels ranging from 87.1% for the posterior descending artery of the right coronary artery to 100% for the proximal segment of the right coronary artery. The average computational time was 0.5 seconds per case. The interreader overlap was 96.6%. CONCLUSIONS: The presented fully automated DL-based coronary artery labeling algorithm provides fast and precise labeling of the coronary artery segments bearing the potential to improve automated structured reporting for CCTA.