Discrepancy analysis comparing two diagnostic platforms offers potential insights into both, without assuming either is always correct. Having optimized the Molecular Microscope® Diagnostic System (MMDx) in renal transplant biopsies, we studied discrepancies within MMDx (reports and sign‐out comments) and between MMDx and histology. Inter‐pathologist discrepancies have previously been documented and were not assessed. Discrepancies cases were classified as ‘clear’ (e.g. ABMR vs. TCMR), ‘boundary’ (e.g. ABMR vs. possible ABMR), or ‘mixed’ (e.g. Mixed vs. ABMR). MMDx report scores showed 99% correlations; sign‐out interpretations showed 7% variation between observers, all located around boundaries. Histology disagreed with MMDx in 37% of biopsies, including 315 clear discrepancies, all with implications for therapy. Discrepancies were widely distributed in all histology diagnoses but increased in some scenarios e.g. histology TCMR contained 14% MMDx ABMR and 20% MMDx no rejection. MMDx usually gave unambiguous diagnoses in cases with ambiguous histology e.g. borderline and transplant glomerulopathy. Histology lesions or features associated with more frequent discrepancies (e.g. tubulitis, arteritis, and polyomavirus nephropathy) were not associated with increased MMDx uncertainty, indicating that MMDx can clarify biopsies with histologic ambiguity. The patterns of histology‐MMDx discrepancies highlight specific histology diagnoses in which MMDx assessment should be considered for guiding therapy.
13 March 2019 in American Journal of Transplantation
We previously reported a system for assessing rejection in kidney transplant biopsies using microarray‐based gene expression data, the Molecular Microscope® Diagnostic System (MMDx). The present study was designed to optimize the accuracy and stability of MMDx diagnoses by replacing single machine learning classifiers with ensembles of diverse classifier methods. We also examined the use of automated report sign‐outs and the agreement between multiple human interpreters of the molecular results. Ensembles generated diagnoses that were both more accurate than the best individual classifiers, and nearly as stable as the best, consistent with expectations from the machine learning literature. Human experts had ~93% agreement (balanced accuracy) signing out the reports, and random forest‐based automated sign‐outs showed similar levels of agreement with the human experts (92% and 94% for predicting the expert MMDx sign‐outs for T cell‐mediated (TCMR) and antibody‐mediated rejection (ABMR) respectively). In most cases disagreements, whether between experts or between experts and automated sign‐outs, were in biopsies near diagnostic thresholds. Considerable disagreement with histology persisted. The balanced accuracies of MMDx sign‐outs for histology diagnoses of TCMR and ABMR were 73% and 78% respectively. Disagreement with histology is largely due to the known noise in histology assessments.
12 November 2018 in American Journal of Transplantation
In kidney transplant biopsies, inflammation in areas of atrophy‐fibrosis (i‐IFTA) is associated with increased risk of failure, presumably because inflammation is evoked by recent parenchymal injury from rejection or other insults, but some cases also have rejection. The present study explored the frequency of rejection in i‐IFTA, by histology Banff 2015 and microarray‐based molecular diagnostic system (MMDx). In unselected indication biopsies (108 i‐IFTA, 73 uninflamed IFTA (i0‐IFTA), and 53 no IFTA), i‐IFTA biopsies were later, more scarred, and had more antibody‐mediated rejection (ABMR) by histology (28%) and MMDx (45%). T cell‐mediated rejection (TCMR) was infrequent in i‐IFTA by histology (8%) and MMDx (16%). Twelve i‐IFTA biopsies (11%) had molecular TCMR not diagnosed by histology, although six were called Borderline and almost all had histologic TCMR lesions. The prominent feature of i‐IFTA biopsies was molecular injury e.g. AKI transcripts. In multivariate analysis of biopsies>1 year post‐transplant, the strongest associations with graft loss were AKI transcripts and histologic atrophy‐scarring; i‐IFTA was not significant when molecular AKI was included. We conclude that i‐IFTA in indication biopsies reflects recent/ongoing parenchymal injury, often with concomitant ABMR but few with TCMR. Thus application of Banff i‐IFTA in the population of late biopsies needs to be reconsidered.
22 February 2017 in American Journal of Transplantation
Histologic assessment of kidney transplant biopsies relies on cortex rather than medulla, but for microarray studies, the proportion cortex in a biopsy is typically unknown and could affect the molecular readings. The present study aimed to develop a molecular estimate of proportion cortex in biopsies and examine its effect on molecular diagnoses. Microarrays from 26 kidney transplant biopsies divided into cortex and medulla components and processed separately showed that many of the most significant differences were in glomerular genes (e.g. NPHS2, NPHS1, CLIC5, PTPRO, PLA2R1, PLCE1, PODXL, and REN). Using NPHS2 (podocin) to estimate proportion cortex, we examined whether proportion cortex influenced molecular assessment in the molecular microscope diagnostic system. In 1190 unselected kidney transplant indication biopsies (Clinicaltrials.govNCT01299168), only 11% had <50% cortex. Molecular scores for antibody‐mediated rejection, T cell–mediated rejection, and injury were independent of proportion cortex. Rejection was diagnosed in many biopsies that were mostly or all medulla. Agreement in molecular diagnoses in paired cortex/medulla samples (23/26) was similar to biological replicates (32/37). We conclude that NPHS2 expression can estimate proportion cortex; that proportion cortex has little influence on molecular diagnosis of rejection; and that, although histology cannot assess medulla, rejection does occur in medulla as well as cortex.
7 August 2024 in American Journal of Transplantation
Most kidney transplant patients who undergo biopsies are classified as having no rejection based on consensus thresholds. However, we hypothesized that because these patients have normal adaptive immune systems, T cell-mediated rejection (TCMR) and antibody-mediated rejection (ABMR) may exist as subthreshold activity in some transplants currently classified as no rejection. To examine this question, we studied genome-wide microarray results from 5086 kidney transplant biopsies (4170 patients). An updated archetypal analysis designated 56% of biopsies as no rejection. Subthreshold molecular TCMR and/or ABMR activity molecular activity was detectable as elevated classifier scores in many biopsies classified as no rejection, with ABMR activity in many TCMR biopsies and TCMR activity in many ABMR biopsies. In biopsies classified as no rejection histologically and molecularly, molecular TCMR classifier scores correlated with increases in histologic TCMR features and molecular injury, lower eGFR, and higher risk of graft loss, and molecular ABMR activity correlated with increased glomerulitis and donor-specific antibody. No rejection biopsies with high subthreshold TCMR or ABMR activity had a higher probability of having TCMR or ABMR respectively diagnosed in a future biopsy. We conclude that many kidney transplant recipients have unrecognized subthreshold TCMR or ABMR activity, with significant implications for future problems.
26 March 2024 in American Journal of Transplantation
The first-generation Molecular Microscope® (MMDx) system for heart transplant endomyocardial biopsies used expression of rejection-associated transcripts (RATs) to diagnose T cell-mediated rejection (TCMR) and antibody-mediated rejection (ABMR) but also detected acute injury. However, the ideal system should detect rejection without being influenced by injury, to permit analysis of the relationship between rejection and parenchymal injury. To achieve this, we developed a new rejection classification in an expanded cohort of 3230 biopsies: 1641 from INTERHEART (ClinicalTrials.gov #NCT02670408), plus 1589 service biopsies added to improve the power of the machine learning algorithms. The new system used six rejection classifiers instead of RATs and generated seven rejection archetypes: No Rejection 48%; Minor 24%; TCMR1 2.3%; TCMR2 2.7%; TCMR/Mixed 2.7%; EABMR 3.9%; and FABMR 16%. Using rejection classifiers eliminated cross-reactions with acute injury, permitting separate assessment of rejection and injury. TCMR was associated with severe recent injury and late atrophy-fibrosis and rarely had normal parenchyma. ABMR was better tolerated, seldom producing severe injury, but in later biopsies was often associated with atrophy-fibrosis, indicating long term risk. Graft survival and LVEF were reduced in hearts with TCMR, but also in hearts with severe-recent injury and atrophy-fibrosis, even without rejection. INTERHEART: ClinicalTrials.gov NCT02670408.