Abstract: TH-OR51
Diagnostic Application of NanoString Gene Scores in Transplant Biopsies with Suspicious Features of Antibody-Mediated Rejection
Session Information
- Kidney Transplantation: Breakthroughs from Basic to Translational to Clinical Research
November 04, 2021 | Location: Simulive, Virtual Only
Abstract Time: 04:30 PM - 06:00 PM
Category: Transplantation
- 1901 Transplantation: Basic
Authors
- Beadle, Jack, Imperial College London, London, London, United Kingdom
- Toulza, Frederic Jb, Imperial College London, London, London, United Kingdom
- Willicombe, Michelle, Imperial College London, London, London, United Kingdom
- Roufosse, Candice A., Imperial College London, London, London, United Kingdom
Background
Antibody-mediated rejection (AMR) is the leading cause of renal allograft loss, a diagnosis which is reached using the Banff Classification for Allograft Pathology. Biopsies that only partially fulfil the histological criteria for AMR, those with ‘incomplete phenotypes’, provide a challenge for diagnosis, prognosis and management. The Banff Molecular Diagnostics Working Group has designed a 758-gene panel for use with NanoString with the aim of developing a gene transcript signature to improve the diagnosis of AMR, but this has not been validated in biopsies with features suspicious for AMR, where its use may distinguish between cases that represent AMR from those that do not.
Methods
RNA was extracted from a retrospective cohort of 147 FFPE biopsies that were divided into three groups: biopsies that fulfilled the 2019 Banff Criteria for AMR ('AMR', n=34), those that would meet the criteria for AMR only in the presence of a validated gene signature ('AMRsusp', n=41), and biopsies that showed no features of AMR ('No AMR', n=72) . Gene expression analysis of 758 genes in the Banff Human Organ Transplant panel was carried out using Nanostring® analysis. Gene expression was normalised using twelve housekeeper genes and internal positive-control normalisation
Results
Of the 758 genes, 134 were significantly different (with FDR set at 0.01) in biopsies which fulfilled the full Banff Criteria for AMR, compared to biopsies without features of AMR. Lasso regression was used to define a set of 37 up-regulated genes that were strongly predictive of AMR, and used to create an AMR Gene Score. A Reciever-Operating Characteristic demonstrated that the Gene Score was able to differentiate between AMR and No AMR Cases (AUC 0.8742 ,95% CI 0.8095-0.9388, p<0.0001) and was used to develop a gene score cut-off, maximising sensitivity and specificity. In biopsies suspicious for AMR, but which did not complete the full diagnostic criteria, a high Gene Score was predictive of allograft loss, compared to biopsies with a low gene score (p=0.0065)
Conclusion
Nanostring analysis of gene expression in FFPE biopsy samples can be used to identify biopsies suspicious for AMR that are at higher risk of allograft loss, and may have a role in characterising cases that represent AMR, even in the abscence of full diagnostic criteria.