Background: Acute Rejection (AR) is the main cause of the renal failure within one year after transplantation (tx). Its earlier diagnosis is crucial for a patient. However, there is variation in histological diagnosis among pathologists even for stable tx.

Methods: In the first study of its kind, we leveraged publicly available microarray data from the 29 NCBI GEO datasets, with 3,004 human kidney biopsies. We performed computational data quality control and filtering procedures to normalize and analyze histologically stable kidney tx (hSTA) and identify those that molecularly confirm the STA phenotype (hSTA/mSTA) as well as separate and better characterize those hSTA samples that might show early signs of rejection (hSTA/mAR). In this dataset we have 585 allograft samples with AR (including ABMR, TCMR, Mixed, and Borderline rejections) and 1,386 with hSTA, and 620 normal donor samples.

Results: Gene expression analysis (FDR < 0.01, FC > 1.3) revealed 1,732 differentially expressed genes in AR. Hierarchical clustering based on this gene expression signature identified a sizable fraction (30%) of hSTA as molecularly false positives, with highly similar profiles to AR (hSTA/mAR). Next, we utilized xCell, new tool for cell type enrichment analysis (D. Aran et.al, Genome Biology, 2017), to identify a unique signature of 27 cell types (B-H adj. p-value < 0.01) that specifically distinguished allografts with AR. Two major cell types were confirmed to drive the major changes in gene expression in AR: CD8+ Tem and CD8+ Tcm cells (log FC = 37) were highly enriched over all other cell types (log FC < 4). We then scored each STA sample based on a combination of AR specific expression and cell type enrichment features, which allowed us to more precisely phenotype hSTA allografts into hSTA/mSTA and hSTA/mAR. These results were independently validated on longitudinal cohorts of kidney tx bx, where hSTA/mAR bx were found to have higher risk of progression to chronic allograft damage over time.

Conclusions: Our computational analysis of public data allows for precision (re)classification of histologically STA allografts and demonstrates possible causes of discrepancies in current allograft histological phenotyping. Correct identification of hSTA/mSTA allografts is an important deliverable for selection of “true” STA samples for mechanistic studies and for accurate prediction of patient clinical outcomes.

CITATION INFORMATION: Rychkov D., Sigdel T., Sirota M., Sarwal M. Precision (re)Phenotyping of Histologically Stable Kidney Transplants Am J Transplant. 2017;17 (suppl 3).

« Back to 2018 American Transplant Congress