Date: Saturday, June 1, 2019
Session Time: 5:30pm-7:30pm
Presentation Time: 5:30pm-7:30pm
Location: Hall C & D
*Purpose: Histologically stable allografts have been shown to have molecular inflammation by prior gene expression studies. Nevertheless, there are no methodologies in place to sub-group histologically stable (STA) molecularly active vs quiescent grafts, which could have different clinical outcomes and prognoses. The more precise (sub-)phenotyping of histologically STA allografts would also allow for better selection of true STA control samples for analysis of graft injury during acute rejection.
*Methods: We leveraged publicly available microarray data from 28 NCBI GEO dataset with 2,273 human transplant kidney biopsies. The diagnoses were histologically confirmed for 510 AR (including ABMR, TCMR, Mixed, and Borderline rejections), 1,154 STA, and 609 normal donor samples. We performed gene expression analysis on AR and Normal samples and identified 8,485 significant genes (FDR p-value < 0.05). A specific feature selection pipeline based on a machine learning algorithm revealed 6 genes, KLF4, CENPJ, PPP1R15, KLF2, TNFAIP3, CXCL9, that were highly associated with AR (AUC = 0.99). We also utilized xCell, a cell type enrichment analysis tool, to identify 39 significant cell types (FDR p-value < 0.05) and performed the feature selection procedure. We identified 7 cell types, NK cells, CD4+ Tcm, CD4+ Tem, CD8+ Tem, CD8+ Tcm, Mast cells, and Th1 cells, that separated outcomes (AUC = 0.92). Random Forest classification models based on selected genes (GE model) and cell types (CT model) were applied to histologically STA samples to identify STA allografts into different molecular subtypes, molecularly AR (mAR) and STA (mSTA).
*Results: 40% of hSTA samples from public data were identified as mAR. To validate the hSTA/mAR subtype predictions, we leveraged available clinical data to compute the change in GFR over the subsequent 5 years after the biopsy and found significant decline in graft function in hSTA/mAR biopsies over those with a hSTA/mSTA phenotype as predicted by the xCell model.
*Conclusions: We have developed a tool of cell type enrichment of 7 specific cell types in stable allografts from gene expression data that can accurately discern molecular heterogeneity in histologically STA allografts and identify those STA grafts that have molecular inflammation (hSTA/mAR). These findings can positively impact the design of mechanistic studies with selection of true STA samples, and identify hSTA/mAR grafts that may require immunosuppression adjustment to preserve allograft function.
To cite this abstract in AMA style:Rychkov D, Sigdel T, Liberto J, Sarwal M, Sirota M. Subtyping Of Histologically Stable Kidney Transplants In Molecular Meta-analysis [abstract]. Am J Transplant. 2019; 19 (suppl 3). https://atcmeetingabstracts.com/abstract/subtyping-of-histologically-stable-kidney-transplants-in-molecular-meta-analysis/. Accessed December 13, 2019.
« Back to 2019 American Transplant Congress