Back

In lupus nephritis, specific in situ inflammatory states are associated with refractory disease and progression to renal failure

Abraham, R.; Durkee, M. S.; Ai, J.; Veselits, M.; Casella, G.; Asano, Y.; Chang, A.; Ko, K.; Oshinsky, C.; Peninger, E.; Giger, M.; Clark, M. R.

2021-09-04 immunology
10.1101/2021.09.03.458909 bioRxiv
Show abstract

In human lupus nephritis (LN), tubulointerstitial inflammation (TII) on biopsy predicts progression to end stage renal disease (ESRD). However, while approximately half of patients with moderate or severe TII develop ESRD, half do not. Therefore, we hypothesized that TII is heterogeneous, with distinct inflammatory states associated with different renal outcomes. We interrogated renal biopsies from LN longitudinal and cross-sectional cohorts using both conventional and highly multiplexed confocal microscopy. To accurately segment cells across whole biopsies, and to understand their spatial relationships, we developed unique computational pipelines by training and implementing several deep learning models and other computer vision techniques. Surprisingly, across biopsies, high B cell densities were strongly associated with protection from ESRD. In contrast, elevated CD4-T cell population densities, which included CD8, {gamma}{delta} and double negative (CD4-CD8-{delta}-, DN) T cells, were associated with both acute refractory renal failure and gradual progression to ESRD. Interestingly, lymphocytes and dendritic cells were organized into discrete clusters or neighborhoods that could be characterized by the enrichment for specific cell populations. B cells were often organized into large neighborhoods with CD4+ T cells including T follicular helper-like cells. In contrast, the CD4-T cell populations formed small cellular neighborhoods whose frequency predicted subsequent progression to ESRD. These data reveal that in LN, specific in situ inflammatory states are associated with refractory disease and progression to ESRD. One sentence summaryUsing deep machine learning to analyze confocal microscopy data, we demonstrate that in lupus nephritis, CD4-T cell populations, including CD8+ and {gamma}{delta} T cells, organize into specific spatial neighborhoods that predict progression to renal failure.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.