5 March 2025 in IEEE Journal of Biomedical and Health Informatics
Applying deep learning to predict patient prognostic survival outcomes using histological whole-slide images (WSIs) and genomic data is challenging due to the morphological and transcriptomic heterogeneity present in the tumor microenvironment. Existing deep learning-enabled methods often exhibit learning biases, primarily because the genomic knowledge used to guide directional feature extraction from WSIs may be irrelevant or incomplete. This results in a suboptimal and sometimes myopic understanding of the overall pathological landscape, potentially overlooking crucial histological insights. To tackle these challenges, we propose the CounterFactual Bidirectional Co-Attention Transformer framework. By integrating a bidirectional co-attention layer, our framework fosters effective feature interactions between the genomic and histology modalities and ensures consistent identification of prognostic features from WSIs. Using counterfactual reasoning, our model utilizes causality to model unimodal and multimodal knowledge for cancer risk stratification. This approach directly addresses and reduces bias, enables the exploration of ’what-if' scenarios, and offers a deeper understanding of how different features influence survival outcomes. Our framework, validated across eight diverse cancer benchmark datasets from The Cancer Genome Atlas (TCGA), represents a major improvement over current histology-genomic model learning methods. It shows an average 2.5% improvement in c-index performance over 18 state-of-the-art models in predicting patient prognoses across eight cancer types. Our code is released at https://github.com/BusyJzy599/CFBCT-main.
29 August 2024 in Transplantation Direct
Background.: The number of patients waiting for heart transplant far exceeds the number of hearts available. Donation after circulatory death (DCD) combined with machine perfusion can increase the number of transplantable hearts by as much as 48%. Emerging studies also suggest machine perfusion could enable allograft “reconditioning” to optimize outcomes. However, a detailed understanding of the energetic substrates and metabolic changes during perfusion is lacking. Methods.: Metabolites were analyzed using 1-dimensional 1H and 2-dimensional 13C-1H heteronuclear spectrum quantum correlation nuclear magnetic resonance spectroscopy on serial perfusate samples (N = 98) from 32 DCD hearts that were successfully transplanted. Wilcoxon signed-rank and Kruskal-Wallis tests were used to test for significant differences in metabolite resonances during perfusion and network analysis was used to uncover altered metabolic pathways. Results.: Metabolite differences were observed comparing baseline perfusate to samples from hearts at time points 1–2, 3–4, and 5–6 h of perfusion and all pairwise combinations. Among the most significant changes observed were a steady decrease in fatty acids and succinate and an increase in amino acids, especially alanine, glutamine, and glycine. This core set of metabolites was also altered in a DCD porcine model perfused with a nonblood-based perfusate. Conclusions.: Temporal metabolic changes were identified during ex vivo perfusion of DCD hearts. Fatty acids, which are normally the predominant myocardial energy source, are rapidly depleted, while amino acids such as alanine, glutamine, and glycine increase. We also noted depletion of ketone, β-hydroxybutyric acid, which is known to have cardioprotective properties. Collectively, these results suggest a shift in energy substrates and provide a basis to design optimal preservation techniques during perfusion.
19 June 2024 in JACC: Advances
26 August 2026 in Interdisciplinary CardioVascular and Thoracic Surgery
Objectives: The use of in situ internal thoracic artery (ITA) grafts ipsilateral to upper extremity arterio-venous fistulas (UE-AVFs) during coronary artery bypass grafting (CABG) in dialysis-dependent patients raises concerns of coronary steal and compromised graft perfusion. However, the clinical significance of ITA-AVF laterality remains uncertain. Methods: We retrospectively analysed dialysis-dependent patients who underwent CABG between February 2015 and December 2023 at 2 institutions. Patients with non-UE-AVF access, total vein grafting, or composite vein-artery configurations were excluded. The final cohort (N = 122) was divided into 2 groups: ipsilateral in situ ITA grafts (N = 66) and contralateral in situ or free ITA grafts (N = 56). The primary outcome was the incidence of major adverse cardiovascular and cerebrovascular events (MACCE); secondary outcomes included perioperative complications, haemodialysis-induced ischaemia, and 3-year survival. Results: Baseline and operative characteristics were largely comparable between groups. Perioperative outcomes, including surgical site infection, 30-day mortality, and prolonged ventilation, showed no significant differences. Over a median follow-up of 2.4 years, MACCE occurred in 44% of patients (39% ipsilateral vs 50% contralateral/free ITA; P = .24), with no significant differences in 3-year mortality (29% vs 21%; P = .352). Multivariable analyses showed no association between ITA-AVF laterality and MACCE (adjusted odds ratio 0.68, P = .31) or 3-year mortality (adjusted hazard ratio 1.56, P = .291). Conclusions: In this modestly sized retrospective cohort, ipsilateral in situ ITA grafting was not associated with significantly worse perioperative or long-term outcomes. Our findings should be interpreted as hypothesis-generating, and larger multicentre studies are needed to more definitively evaluate the safety of this grafting strategy in dialysis-dependent patients with UE-AVFs.