*Purpose: Molecular assays on ~3mm biopsy (BX) fragments of cannot confirm histologic T-cell mediated rejection (TCMR) in ~50% of routine biopsies, and fail to detect any TCMR in late BXs >10 years post-transplant taken from failing grafts. We tested the hypothesis that this is due to variations in pathology lesions in different regions of interest (ROIs).

*Methods: Spatially resolved RNA-seq was performed on 5 allograft BX with variable inflammation & tubulitis along the core. Two ROIs in the same biopsy (ROI-1 and ROI-2) were examined. Technical reproducibility was assessed by duplicate assays on ROI-2. 15 ROIs were examined to collect data on 11,664,548 – 18,304,170 sequence reads/ROI. Differential gene expression (DE, read count >5 in any sample) between ROIs was performed using by DESeq2 analysis (pAdj<0.05) for groups of samples. CLC Genomics Workbench edgeR was used for PCA analysis of technical replicates.

*Results: On comparing two different ROIs within the same BX, number of DE genes in ROI-1 compared to ROI-2 varied from 77-1198 (median 189, median log2 fold change 0.996, IQR 0.128) (Fig 1). On PCA analysis, rlog2 values of ROI associated gene sets clustered in accordance with degree of inflammation, irrespective of location in the BX core, confirming that regional differences in gene expression (GE) affect molecular classification of BXs. In replicate analyses on the same ROIs, the number of DE genes ranged from 1-9 (median 3, median log2 fold 1.0, IQR 0.0962). Reproducibility of GE signals in ROI-1 vs RO1-2 comparisons varied with the magnitude of the signal. Mean log2 normalized reads for the Q1, Q2, Q3 and Q4 quartiles were 2.64, 4.42, 5.67 and 7.36 respectively and the corresponding coefficients of variation were 27.53%, 8.89%, 6.17% and 12.93%. These data suggests that molecular quantitation of inflammation in BXs is unlikely to be reliable unless it has at least doubled in intensity, and an even greater change might be needed genes with low GE values.

*Conclusions: Different ROIs in the same BX show substantial variation in the number of DE genes, and in quantitative expression of individual mRNAs. These differences affect BX classification, call into question commercial claims that current molecular tests are 99% accurate, explain discrepancies between histologic vs molecular analysis, and emphasize the need to develop more accurate tests tethered to clinical endpoints.

Figure 1. RO1-1 (right end, TCMR area) in this BX had 1198 DE genes compared to ROI-2 (left end, non-inflamed area). Replicates within an RO1-2 were quite concordant, and taken in isolation, created a false sense of reproducibility in the molecular analysis.

« Back to 2022 American Transplant Congress