Validation of a Blood and Biopsy Gene Expression-Based Molecular Diagnostics for Subclinical Acute Rejection: Comparing DNA Microarrays Vs. Next-Generation RNA Sequencing
1The Scripps Research Institute, La Jolla, CA
2Northwestern University, Chicago, IL.
Meeting: 2015 American Transplant Congress
Abstract number: B256
Keywords: Genomics, Kidney transplantation
Session Information
Session Name: Poster Session B: Translational Genetics and Proteomics in Transplantation
Session Type: Poster Session
Date: Sunday, May 3, 2015
Session Time: 5:30pm-6:30pm
Presentation Time: 5:30pm-6:30pm
Location: Exhibit Hall E
Background: Microarray-based gene expression signatures for clinical diagnostics are well described. However, microarrays lack advantages offered by newer technologies such as Next Generation Sequencing (NGS): accurate quantification, ability to multiplex clinical samples and capability to study microRNAs, long non-coding RNAs and novel transcripts. We validated our microarray-based molecular diagnostics for both peripheral blood and biopsies to differentiate Subclinical Acute Rejection (SCAR), Clinical Acute rejection (CAR) and Transplant eXcellent subjects (TX) and compared the fidelity of gene expression signatures on both global profiling platforms.
Methods: Matched pairs of blood/biopsy samples from 69 subjects were profiled (CAR=21, SCAR=23 and TX=25) on microarrays (Affymetrix HT U133 Plus PM arrays) and NGS (Ion Torrent Proton). Analyses were done in Partek Genomics Suite using ANOVAs and predictions with the Nearest Centroid Algorithm. NGS depth of ∼15 million reads were obtained for all samples.
Results: We used a 2-step prediction model to predict SCAR. For biopsies overall predictive accuracies of 94% and 91% were obtained with microarrays and NGS. For peripheral blood, a similar approach showed predictive accuracies of 91% and 89%. In the biopsies, for 1066 genes differentially expressed (FDR<1%) in both microarrays and NGS data we saw correlations of >98% for fold-change directionality and r2 values of 0.88 – 0.96 for absolute fold changes. Similarly, for 101 genes differentially expressed (FDR<1%) in blood for both we saw correlations of 79-99% and r2 values of 0.76 – 0.92. There was 28% agreement between differentially expressed genes in blood and biopsies.
Conclusions: The data validate our existing microarray signatures on the orthogonal NGS platform. Both platforms perform well with respect to predictions of phenotype and show high correlations with fold-change for significantly differentially expressed genes. Surprisingly, microarrays detect differential expression of about twice the number of genes as NGS. These differences could be improved by evolving RNAseq protocols. The data does not justify claims that NGS is better than microarrays for gene expression but NGS is very appropriate for the next generation of clinical diagnostics.
To cite this abstract in AMA style:
Kurian S, Friedewald J, Harrison F, Gelbart T, Head S, Ordoukhanian P, Abecassis M, Salomon D. Validation of a Blood and Biopsy Gene Expression-Based Molecular Diagnostics for Subclinical Acute Rejection: Comparing DNA Microarrays Vs. Next-Generation RNA Sequencing [abstract]. Am J Transplant. 2015; 15 (suppl 3). https://atcmeetingabstracts.com/abstract/validation-of-a-blood-and-biopsy-gene-expression-based-molecular-diagnostics-for-subclinical-acute-rejection-comparing-dna-microarrays-vs-next-generation-rna-sequencing/. Accessed November 21, 2024.« Back to 2015 American Transplant Congress