Authors’ Reply: Computational Lymphocyte Topology: A Roadmap to Mechanism and Clinical Translation?

We thank Dr. Zhou et al.1 for their interest in our work2 in JASN and their thoughtful letter outlining important future directions for this research. Our study represents a paradigm shift in the quantitative assessment of inflammation—the first demonstration that graph-based topologic features of lymphocyte distribution provide prognostic information independent of traditional metrics in glomerular diseases. This initial exploratory investigation serves as a discovery tool, proof-of-concept, and a foundation on which anchoring molecular analysis can provide biologic plausibility to these observations. We fully agree that integrating sophisticated bulk and spatial omics with our topologic framework would significantly enrich our understanding of which cell type drives clinical outcome and response to therapy and provide the mechanistic insights needed to understand whether these patterns actively drive injury versus attempt to repair or represent secondary phenomena. Complementary spatial multiomics studies are currently being conducted across nephrology consortia. Although these rapidly evolving technologies have been proven to advance our understanding of kidney diseases, broad access to biopsy imaging and tissue blocks remains challenging, at least in North America, due to strict institutional and national regulatory bodies' policies. Thus, Dr. Zhou and colleagues' observation is very timely in offering the opportunity to raise awareness of the importance of the application of these methodologies to improve current knowledge and patient care and to call on the medical community, regulatory agencies, and patient advocacy groups to revise current biopsy practices and tissue access regulations. Translation of image analysis–derived discoveries should represent the logical next step toward clinical implementation. Our work intends to broaden the methodologic space by introducing graph topology among various methodologic configurations and by demonstrating the applicability of different topology metrics. The optimal metric may depend on the specific clinical question and context, as different applications may require distinct topologic metrics. Moreover, clinical implementation will require multiple validation steps beyond simplification, including prospective validation in independent cohorts; harmonization across different scanning platforms and institutions; sustained collaborative efforts from the nephrology, pathology, computational, and clinical research communities; and, ultimately, demonstration that such metrics can effectively guide treatment decisions and improve patient outcomes. In conclusion, we are grateful and feel encouraged by Dr. Zhou and colleagues' engagement and look forward to future studies aiming at advancing mechanistic understanding and clinical implementation.