Non-Invasive Panel of MicroRNAs Predict Long-Term Outcomes Post-Kidney Transplantation
1Surgery and Pathology, UVA, Charlottesville, VA
2VCU, Richmond, VA.
Meeting: 2015 American Transplant Congress
Abstract number: 37
Keywords: Genomic markers, Renal function
Session Information
Session Name: Concurrent Session: Kidney Complications: Late Graft Failure
Session Type: Concurrent Session
Date: Sunday, May 3, 2015
Session Time: 2:15pm-3:45pm
Presentation Time: 2:51pm-3:03pm
Location: Terrace IV
Background. Currently available methods for evaluating kidney graft function post-transplantation (KT) are either ineffective or inaccurate (e.g., serum creatinine) or highly invasive (e.g., biopsies). We aimed to identify a panel of microRNAs (miRs) that can be measured in urine samples for monitoring graft function.
Methods. A total of 104 unique deceased donor kidney transplant recipients (KTPs) (184 urine samples) undergoing protocol biopsies were evaluated. The patients were divided in 2 groups with two-time point sample evaluation each: training (miRNAs tested using GeneChip miRNA 4.0 Array) n= 24 KTPs and validation set (selected miRs tested using RT-qPCR reactions) n= 80 KTPs. Urine samples were collected at biopsy time (3- and 9-months post-KT). Patients were followed by 4.1+/-0.8 years. Urinary and cell free miR were evaluated. miR array data was pre-processed using the oligo Bioconductor package and analyzed using the R programming environment using the limma package with adjustments for multiple comparisons made by estimating the FDR. TaqMan MicroRNA Assays was used and relative expression analyzed as we previously published. RNU44 was used as endogenous control for miR analysis. To derive a multivariable model consisting of multiple parameters capable of predicting outcome, an L1 penalized Cox proportional hazards model was fit using the glmpath library.
Results. KTPs were classified based on graft function and histological findings at 24-mo post-KT as Progressors (P) or non-Progressors (NP) to chronic renal allograft dysfunction in the training and validation sets. The analysis in the training set showed 125 significant miRs (p < 0.001) between P vs. NP at 3-and 87 miRs at 9-mo post-KT, respectively. A penalized logistic regression model was fit to predict P vs. NP using miRs as predictor variables using glmpath. The final model was selected as that having the minimum AIC. The final model included eight miRs (miR-199a, miR-200b, miR-145, miR146b, miR-29c, miR-29, and miR-155). The set performed with high accuracy in the independent set with 85% sensitivity, 70% specificity, 74% PPV, and 81% NPV. The AUC was 0.86
Conclusions. We have identified and independently validated a set of miRNAs that can be non-invasively tested for predicting and monitoring graft function. These miRNAs seem to have an important role in inflammation, stromal biology, and fibrosis.
To cite this abstract in AMA style:
Suh J, Maluf D, Cathro H, Gehrau R, McConnell I, Brayman K, Dumur C, Mas V. Non-Invasive Panel of MicroRNAs Predict Long-Term Outcomes Post-Kidney Transplantation [abstract]. Am J Transplant. 2015; 15 (suppl 3). https://atcmeetingabstracts.com/abstract/non-invasive-panel-of-micrornas-predict-long-term-outcomes-post-kidney-transplantation/. Accessed October 15, 2024.« Back to 2015 American Transplant Congress