Session Information
Session Type: Poster Session C
Session Time: 10:30AM-12:30PM
Background/Purpose: Lupus nephritis (LN) is a major cause of morbidity and mortality in patients with systemic lupus erythematosus (SLE). Despite the advances in the management of LN, its heterogenous nature poses challenges in predicting treatment response. Imaging mass cytometry (IMC) offers a potential solution to dissect the heterogeneity of the treatment response by studying the spatial molecular architecture of LN.
Methods: In total, 47 glomerular and 45 tubulo-interstitial regions of interest (ROI) from 8 Class III, 8 Class IV LN and 3 renal cell carcinoma patients (RCC) were analyzed. All SLE patients had their first kidney biopsy performed for diagnosis. Glomerular ROIs were selected based on viable glomeruli on histopathological evaluation. Normal adjacent tissue in nephrectomy samples from RCC patients was used for comparison. Using a validated panel of metal-conjugated antibodies designed to include markers of major immune cell subsets, resident renal cells and several proteins known to be increased in LN kidneys, we performed simultaneous multiplex imaging with IMC for 35 target proteins. Renal response was assessed at 12 months after induction therapy. Unsupervised Bayesian network analysis (BNA) using the EQ algorithm was performed to determine the associated network with renal response in LN patients using BayesiaLab.
Results: There were increased glomerular, periglomerular and tubulo-interstitial immune infiltrates in LN kidneys compared to RCC controls. Markers such as CD45RO (memory T cells and activated macrophages), HLA-DR (antigen presenting cells) and CD68 (macrophages) were increased in ROIs in LN compared to RCC. Interestingly, there was increased glomerular/periglomerular collagen (collagen I and IV) deposition and fibrosis (fibronectin) in Class V compared to IV LN, despite similar chronicity indexes (Figure 1). In terms of treatment response, glomerular (Figure 2) and tubulo-interstitial immune filtrates and fibrosis were increased in LN patients who did not respond to immunosuppressive therapy. Such markers included CD45RO, HLA-DR, CD14 (macrophages), CD163 (M2 macrophages), collagen I, collagen IV and fibronectin. Lastly, there was evidence of increased epithelial to mesenchymal plasticity (EMP) in the glomerular and tubolo-interstitial ROIs in nonresponders compared to responders, with increased staining for β-catenin and vimentin. An unsupervised BNA assessing all clinical and IMC variables revealed associations between the response node with gender, fibrous crescents and chronicity index while the fibronectin node was strongly associated with CD14+ macrophages for both the glomerular (Figure 3) and tubulointerstitial ROIs.
Conclusion: Through IMC proteomics and machine learning, we identified increased macrophages, fibrosis and EMP as markers associated with poorer treatment outcome for LN. Class V LN, unlike Class IV LN, has hitherto been not associated with glomerulosclerosis on routine histopathological examination. We demonstrate for the first time that collagen deposition and fibrosis was increased in Class V glomeruli compared to Class IV using IMC proteomics.
(A) The targets displayed in the imaging mass cytometry images are shown pseudo-colored, as indicated. Shown data is representative of 53 glomerular regions of interest drawn from 8 Class IV LN patients, 8 Class V LN patients, and 2 RCC controls. (B-G) Mann Whitney U tests were conducted to assess significant changes in glomerular fibrosis markers between Class IV LN (green), Class V LN (red) and RCC controls (blue). Whereas each point represents a single interrogated ROI, the bar-charts show the mean in each group. (A, B) Hypercellularity (DNA) in glomeruli was observed in Class IV and Class V LN biopsies compared to RCC controls. (C-E) In addition, kidneys with Class V LN showed increased periglomerular deposition of fibronectin, collagen I and collagen IV. *, p < 0.05; **, p < 0.01; ***, p < 0.001.
(A) Shown imaging mass cytometry images are representative of 47 glomerular regions of interest drawn from 8 Complete Responder (CR), 4 Partial Responder (PR) and 4 Non-Responder (NR) patients. The targets displayed in this figure are shown pseudo-colored, as indicated. (B-G) Mann Whitney U tests were conducted to assess significant changes in glomerular infiltrating immune cell markers between NR (blue), PR (green) and CR (red). Whereas each point represents a single interrogated ROI, the bar-charts show the mean in each group. (A, B, C, D, E and G) Patients who did not respond to therapy (NR) had significantly increased numbers of immune cell infiltrates into the glomeruli, including CD45RO+ memory T cells, CD4+ T cells, HLA-DR+ antigen presenting cells, CD14+ macrophages and CD163+ M2 macrophages, compared to LN patients who responded to therapy (CR and PR). *, p < 0.05; **, p < 0.01; ***, p < 0.001.
The size of each node was determined using “node force” and is proportional to its impact on other nodes in the network, as determined by conditional probabilities. The connections between nodes are determined by Pearson’s correlation coefficient, showing how variables depend on each other, with the thickness of the connections representing the strength of the correlation. Blue edges represent positive correlations while red edges represent negative correlations.
To cite this abstract in AMA style:
Tay F, Louis Sam Titus A, Maruvada V, Lim S, Cho J, Thamboo T, Mohan C. Elucidating the Molecular Correlates of Treatment Response in Lupus Nephritis via Imaging Mass Cytometry Proteomics and Machine Learning [abstract]. Arthritis Rheumatol. 2024; 76 (suppl 9). https://acrabstracts.org/abstract/elucidating-the-molecular-correlates-of-treatment-response-in-lupus-nephritis-via-imaging-mass-cytometry-proteomics-and-machine-learning/. Accessed .« Back to ACR Convergence 2024
ACR Meeting Abstracts - https://acrabstracts.org/abstract/elucidating-the-molecular-correlates-of-treatment-response-in-lupus-nephritis-via-imaging-mass-cytometry-proteomics-and-machine-learning/