1206 Profiling the dark genome of tumours provides new potential therapies for cancer patients

<h3>Background</h3> T cell engagers (TCEs) are a highly interesting class of targeted therapies which have enjoyed recent clinical success, showing the potential to drive robust response rates and increase T cell infiltration in solid tumours. A major challenge for the development of TCEs is that the potency of this drug class warrants cleaner targets with wider therapeutic windows than, for example, ADCs. Currently, the field is severely limited by the availability of clean cancer targets with very few exceptions. Intracellular targets presented to T cells by HLA molecules provide an opportunity to expand the target space addressable by TCEs. A unique advantage of HLA presented targets is that they may be derived from any part of the human genome, including ‘non-coding’ regions – the so-called ‘dark genome’. <h3>Methods</h3> We have developed a large-scale, pan-cancer data architecture based on expression data and immunopeptidomics data from 1,000s of human tissue samples. Based on these data, we have developed algorithms to predict shared cancer targets with large therapeutic windows, with a focus on the novel ‘dark genome’ space. Targets predicted in this manner were screened against T cells from 100s of donors, assessing immunogenicity of each target and identifying strong TCR binders. In parallel, we measure the copy number of these novel HLA presented epitopes on human cancer samples to evaluate targetability in the context of TCEs. <h3>Results</h3> We have identified more than 20 novel targets, primarily originating from the dark genome space. We detail one example of a highly-compelling, fully-validated target originating from the dark genome. We investigated peptide copy number by quantitative mass-spectrometry on patient samples, and demonstrated a surface copy number comparable to PRAME. The target is expressed in multiple cancer indications, including CRC, in &gt;50% of patients. We identified T cells expressing highly-functional TCRs binding the target, took these through affinity maturation using machine learning and functional enrichment, and ultimately generated a TCE with picomolar (pM) target affinity. <h3>Conclusions</h3> We have shown that the dark genome provides a set of valuable targets with large therapeutic windows. We identified rare potent T cell clones and developed TCEs with pM affinity against a multi-tumour dark genome target. Currently, we are finalising the preclinical evaluation of a TCE against this target with focus on CRC patients and expect to be able to enter into IND enabling studies in the first half of 2025. <h3>Ethics Approval</h3> This study was approved by the South Central - Berkshire Research Ethics Committee; approval number 22/SC/0025 and the NHSBT; approval number NCI 2367.