CANDiT
Sinha S, Alcantara J, Perry K, Castillo V, Ondersma AK, Banerjee S, McLaren E, Espinoza CR, Taheri S, Vidales E, Tindle C, Adel A, Amirfakhri S, Sawires JR, Yang J, Bouvet M, Ghosh P
2025 · Cell reports. Medicine
Reactivating lineage commitment to differentiate, and hence eliminate, cancer stem cells (CSCs) remains a therapeutic challenge.
Abstract
From the original paper, Cell reports. Medicine · PubMed
Reactivating lineage commitment to differentiate, and hence eliminate, cancer stem cells (CSCs) remains a therapeutic challenge. Here, we present CANDiT (cancer-associated nodes for differentiation targeting), a machine learning framework that identifies transcriptomic vulnerabilities for differentiation therapy in colorectal cancer (CRC). Centering on CDX2-a master intestinal lineage factor lost in high-risk, poorly differentiated CRCs-we identify PRKAB1, a stress polarity sensor, as a top therapeutic target. A clinical-grade PRKAB1 agonist reactivates lineage programs, dismantles Wnt/YAP-driven stemness, and selectively eliminates CDX2-low CSCs across CRC cell lines, xenografts, and patient-derived organoids (PDOs). Multivariate analysis reveals a strong therapeutic index tied to the CDX2-low state. A 50-gene response signature, derived from integrated modeling across all platforms, predicts ∼50% reduction in recurrence and mortality risk. Like immunotherapy, CANDiT resurrects a physiologic program-differentiation-to selectively eliminate CSCs, offering a scalable, precision framework for lineage restoration in solid tumors.
Summary
Level-1 network-informed target discovery for CRC differentiation therapy: BoNE builds a CDX2-seeded Clustered Boolean Implication Network on bulk GPL570 colon compendia (refined on GSE77953), ranks CDX2-proximal nodes by classification accuracy, and maps PRKAB1 to the clinical-grade agonist PF-06409577 via PDB/ClinicalTrials.gov. Validation combines in silico bulk cohort checks with wet-lab testing in cell lines, xenografts, and PDOs; the paper does not benchmark CANDiT head-to-head against other computational reprogramming methods.
Why this class
Level 1 because the method scores candidate interventions through static regulatory-network influence relative to a target GRN, without computing the post-intervention state . Representation family is regulatory-network influence. Cited in §3.2 of the review. Editorial rationale pending review by the maintainer.