Abstract Cancer therapeutics such as CAR T, Radioligand , ADC, and Bispecifics are all reliant on targeting cancer-selective surface proteins. A major limitation to these tethered killing strategies is the lack of targets that are exclusively expressed on tumor cells and not healthy tissues. Most cancer surface targets exhibit “lineage” expression properties, in which expression is shared by both the tumor and the normal tissue lineage from which the tumor developed. In many cases, expression in normal tissue leads to on-target off-tumor toxicity that can severely limit therapeutic benefit. To address this challenge, we developed Oncolinkage, a computational platform to identify novel tumor targets through their association with genomic changes that drive cancer development. Oncolinkage computationally identifies proteins whose expression is driven by an oncogenic event. While genomic lesions (e.g.amplification, epigenetic dysregulation) activate oncogenes in the developing tumor, many bystander genes are also dysregulated by the non-specific nature of these alterations. Oncolinkage screens for computational signals of cancer genomic lesions mathematically connecting the lesion to target upregulation. Because target expression is “linked” to the pro-tumorigenic lesion, Oncolinked targets are under positive selection, resulting in advantageous properties such as low tumor heterogeneity and resistance to loss. We show that the Cancer/Testis antigens of the MAGE family act as oncolinked targets. Normally restricted to male germ cells, MAGE genes are dysregulated in subsets of multiple solid tumors. Oncolinkage algorithms reveal a distinct set of tumor altered loci that correlate with this tumor-specific expression. We postulate that one or more transcription factors or epigenetic regulators in these loci drives tumor-selective dysregulation of these intracellular targets. Oncolinkage was further used to identify expression-disruptive lesions linked to germ cell restricted surface targets. Among more than a dozen identified oncolinked surface proteins, we found RT1, a protein with testis-restricted expression in normal tissue, but expression in 50% of TNBC and 70% of Ovarian Cancers. Tumor expression was linked to copy number gain on Chr10, the genomic locus of RT1. The known oncogene RET is also present on Chr10 providing a hypothesis that an amplification event targeting the RET oncogene also results in the dysregulation of CAR T target RT1 in subsets of breast and ovarian cancer. The oncolinkage platform is a promising computational approach to identify best-in-class targets for CAR T, RLT, and other tethered killing cancer therapies. Citation Format: McKensie A. Collins, Benjamin T. Hallisey, Andrew Astley, Rick Nicoletti, Murray O. Robinson. Oncolinkage: Discovery of tumor-selective surface targets for cancer therapy abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 496.
Collins et al. (Fri,) studied this question.