An integrated molecular diagnostic report for heart transplant biopsies using an ensemble of diagnostic algorithms

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.

Molecular T-cell‒mediated rejection in transbronchial and mucosal lung transplant biopsies is associated with future risk of graft loss

“…R4 late correlated with late time of biopsy after transplantation and had features such as immunoglobulin transcripts that are associated with atrophy/ fibrosis in other organs. 9 We concluded that the best estimate of TCMR in mucosal biopsies was the R2 TCMR score and not PC1 because (1) unlike TBBs, PC1 was influenced by R4 late as well as R2 TCMR and (2) the concordance between single-piece mucosal biopsies makes the R2 TCMR score reliable.…”

The XVth Banff Conference on Allograft Pathology the Banff Workshop Heart Report: Improving the diagnostic yield from endomyocardial biopsies and Quilty effect revisited

“…1,2 The molecular similarities of kidney and heart rejection provide the basis for the application of the "molecular microscope" to heart transplantation. 3 Targeted gene expression analysis has emerged as a rapid and cost-effective alternative to pangenomic microarray approaches. 4,5 Techniques using formalin-fixed paraffin-embedded ( EMBs.…”

“…Gene expression analysis of fresh or frozen allograft tissue using pangenomic microarray approaches is a powerful tool for understanding the pathophysiological pathways of rejection 1,2 . The molecular similarities of kidney and heart rejection provide the basis for the application of the “molecular microscope” to heart transplantation 3 . Targeted gene expression analysis has emerged as a rapid and cost‐effective alternative to pangenomic microarray approaches 4,5 .…”

Investigation of hub genes and immune status in heart transplant rejection using endomyocardial biopsies

“…Recently, molecular examination of EMBs has been proposed to improve the precision and accuracy of HTx rejection diagnosis. A new diagnostic system, the Molecular Microscope™ Diagnostic System (MMDx), assesses EMBs in terms of their molecular phenotype, including stable (normal), TCMR, ABMR and injury; the microarray data are available in GSE124897 10,11 . In the present study, we use bioinformatics analysis to screen the differences in gene expression between TCMR/ABMR and stable HTx samples and identify hub genes.…”

“…diagnostic system, the Molecular Microscope™ Diagnostic System (MMDx), assesses EMBs in terms of their molecular phenotype, including stable (normal), TCMR, ABMR and injury; the microarray data are available in GSE124897. 10,11 In the present study, we use bioinformatics analysis to screen the differences in gene expression between TCMR/ ABMR and stable HTx samples and identify hub genes. Additionally, abnormal immune cell infiltration in TCMR/ABMR samples and the significant relationship between hub gene expression and immune cell populations, further implicates these hub genes in HTx immunity.…”