Two DNA-methylation approaches aimed at identifying tissue of origin for cancers of unknown primary advanced at AACR 2026, targeting a major diagnostic bottleneck. One program from Guardant Health presented a methylation-based machine learning classifier to guide treatment selection, while researchers from Kindai University reported a CpG-based methylation model. Guardant’s and Kindai’s datasets were designed to reduce reliance on inconclusive histopathology and immunohistochemistry, which can delay site-directed therapy. In parallel disclosures, the Kindai model reported high identification performance in test and independent validation cohorts. The broader push reflects renewed interest in molecular tissue-of-origin tools as a way to expand the proportion of CUP patients who can access precision regimens rather than empiric chemotherapy.