Kidney Week 2025 Early Program - Advances in Research Conference 2025: Generative Artificial Intelligence in Kidney Disease Research and Management-Prognostic Enrichment and Clinical Trial Design Considerations

Video Transcription

Video Summary

The speaker discusses how predictive AI and enrichment strategies can improve clinical trial protocol design, especially in cardiovascular–kidney–metabolic (CKM) diseases. Because phase 3 trials are long, expensive (often >$100M), and still only achieve ~60% success, the goal is not to “make drugs work,” but to modestly increase the odds of success by improving patient selection while preserving biological rationale and regulatory acceptability.<br /><br />They review FDA support for enrichment (2019 guidance) and distinguish three approaches: reducing variability (e.g., run-in periods to confirm adherence, tolerability, and stable baseline labs), prognostic enrichment (selecting higher-risk patients to increase event rates and reduce sample size), and predictive enrichment (selecting likely responders or those less likely to have adverse effects). Predictive enrichment can narrow labeling and is less proven in CKM than prognostic enrichment.<br /><br />AI/ML applications are emerging, mostly as simulations (e.g., phenomapping with adaptive enrichment, Alzheimer’s progression prediction, imaging-based total kidney volume in ADPKD). In CKD trials, reliance on simple eGFR/albuminuria “boxes” drives screen failures, especially due to low or noisy UACR. The speaker presents modeling showing multi-variable risk scores can identify high-risk patients even with lower albuminuria, potentially speeding recruitment and shortening trials by weeks to months.<br /><br />Q&A highlights concerns about albuminuria noise, preference for pre-specified baseline AI vs multiple interims, the need for broader kidney/volume-related surrogates, and clarification that prognostic markers increase absolute benefit without necessarily indicating treatment-specific responsiveness.

Asset Subtitle

Navdeep Tangri

Meta Tag

Module ARC

Speaker Navdeep Tangri

Keywords

predictive AI

clinical trial protocol design

enrichment strategies

cardiovascular–kidney–metabolic (CKM) diseases

prognostic vs predictive enrichment

chronic kidney disease (CKD) trials

eGFR and albuminuria (UACR) risk scores