No abstract

Background. We previously reported a microarray-based diagnostic system for heart transplant endomyocardial biopsies (EMBs), using either 3-archetype (3AA) or 4-archetype (4AA) unsupervised algorithms to estimate rejection. The present study aimed to examine the stability of machine-learning algorithms in new biopsies, compare 3AA vs. 4AA algorithms, assess supervised binary classifiers trained on histologic or molecular diagnoses, create a report combining many scores into an ensemble of estimates, and examine possible automated sign-outs. Methods. We studied 889 EMBs from 454 transplant recipients at eight centers: the initial cohort (N=331) and a new cohort (N=558). Published 3AA algorithms derived in cohort 331 were tested in cohort 558; the 3AA and 4AA models were compared; and supervised binary classifiers were created. Results. Algorithms derived in cohort 331 performed similarly in new biopsies despite differences in case mix. In the combined cohort, the 4AA model, including a parenchymal injury score, retained correlations with histologic rejection and DSA similar to the 3AA model. Supervised molecular classifiers predicted molecular rejection (AUCs>0.87) better than histologic rejection (AUCs<0.78), even when trained on histology diagnoses. A report incorporating many AA and binary classifier scores interpreted by one expert showed highly significant agreement with histology (p<0.001), but with many discrepancies as expected from the known noise in histology. An automated random forest score closely predicted expert diagnoses, confirming potential for automated sign-outs. Conclusions. Molecular algorithms are stable in new populations and can be assembled into an ensemble that combines many supervised and unsupervised estimates of the molecular disease states.

BackgroundBecause of improved long-term survival after heart transplantation (HTx), late graft pathologies such as valvular disease or cardiac allograft vasculopathy (CAV) might need surgical intervention to enhance longer survival and ensure quality of life. To this date, there exist no guidelines for indication of cardiac surgery other than retransplantation after HTx.MethodsIn this retrospective, single-center study, we evaluated patients who underwent cardiac surgery after HTx at our institution.ResultsBetween March 1984 and October 2016, 17 (1.16%) of 1466 HTx patients underwent cardiac surgery other than retransplantation after HTx. Indication were valvular disease (n = 7), CAV (n = 6), and other (n = 4). Of these, 29.4% (n = 5) were emergency procedures and 70.6% were elective cases. Median age at time of surgery was 61 years (interquartile range, 52-66 years); 82.4% (n = 14) were male. Median time to surgery after HTx was 9.3 years (2.7-11.1 years). In-hospital, mortality was 11.8% (n = 2); later need of retransplantation was 11.8% (n = 2) due to progressing CAV 3 to 9 months after surgery. One-year survival was 82.35%; overall survival was 47.1% (n = 8) with a median follow-up of 1477 days (416-2135 days). Overall survival after emergency procedures was 209 days (36-1119.5 days) whereas, for elective procedures, it was 1583.5 days (901.5-4319 days).ConclusionsIncidence of cardiac surgery after HTx in our cohort was low (1.16%) compared with that of other studies. In elective cases, long-term survival was good.

The aim was to evaluate the association of molecular‐level human leukocyte antigen (HLA) mismatching with post‐transplant graft survival, rejection, and cardiac allograft vasculopathy (CAV). We retrospectively analyzed all primary cardiac transplant recipients between 01/1984‐06/2016. 1167 patients fulfilled inclusion criteria and had HLA typing information available. In 312 donor‐recipient pairs, typing at serological split antigen level was available. We used the Epitope MisMatch Algorithm to calculate the number of amino acid differences in antibody‐verified HLA eplets (amino acid mismatch load (AAMM)) between donor and recipient. Patients with a higher HLA‐DR AAMM load had inferior 1‐year graft survival (hazard ratio [HR], 1.14; 95% confidence interval [CI], 1.01–1.28). The HLA‐AB AAMM load showed no impact on graft survival. In the subgroup with available split‐level information, we observed an inferior graft survival for a higher HLA‐DR AAMM load 3 months after transplantation (HR, 1.22; 95% CI, 1.04–1.44) and a higher risk for rejection for an increasing HLA‐AB (HR, 1.70; 95% CI, 1.29–2.24) and HLA‐DR (HR, 1.32; 95% CI, 1.09–1.61) AAMM load. No impact on the development of CAV was found. Molecular‐level HLA mismatch analysis could serve as a tool for risk stratification after heart transplantation and might take us one step further into precision medicine.

Background. The INTERHEART study (ClinicalTrials.gov #NCT02670408) used genome-wide microarrays to detect rejection in endomyocardial biopsies; however, many heart transplants with no rejection have late dysfunction and impaired survival. We used the microarray measurements to develop a molecular classification of parenchymal injury. Methods. In 1320 endomyocardial biopsies from 645 patients previously studied for rejection-associated transcripts, we measured the expression of 10 injury-induced transcript sets: 5 induced by recent injury; 2 reflecting macrophage infiltration; 2 normal heart transcript sets; and immunoglobulin transcripts, which correlate with time. We used archetypal clustering to assign injury groups. Results. Injury transcript sets correlated with impaired function. Archetypal clustering based on the expression of injury transcript sets assigned each biopsy to 1 of 5 injury groups: 87 Severe-injury, 221 Late-injury, and 3 with lesser degrees of injury, 376 No-injury, 526 Mild-injury, and 110 Moderate-injury. Severe-injury had extensive loss of normal transcripts (dedifferentiation) and increase in macrophage and injury-induced transcripts. Late-injury was characterized by high immunoglobulin transcript expression. In Severe- and Late-injury, function was depressed, and short-term graft failure was increased, even in hearts with no rejection. T cell–mediated rejection almost always had parenchymal injury, and 85% had Severe- or Late-injury. In contrast, early antibody-mediated rejection (AMR) had little injury, but late AMR often had the Late-injury state. Conclusions. Characterizing heart transplants for their injury state provides new understanding of dysfunction and outcomes and demonstrates the differential impact of T cell–mediated rejection versus AMR on the parenchyma. Slow deterioration from AMR emerges as a major contributor to late dysfunction.