Abstract
Aim: Growing evidence points to non-genetic mechanisms underlying long-term resistance to cancer therapies. These mechanisms involve pre-existing or therapy-induced transcriptional cell states that confer resistance. However, the relationship between early transcriptional responses to treatment and the eventual emergence of resistant states remains poorly understood. Furthermore, it is unclear whether such early resistance-associated transcriptional responses are evolutionarily conserved. In this study, we examine the similarity between early transcriptional responses and long-term resistant states, assess their clinical relevance, and explore their evolutionary conservation across species. Methods: We integrated datasets on early drug responses and long-term resistance from multiple cancer cell lines, bacteria, and yeast to identify early transcriptional changes predictive of long-term resistance and assess their evolutionary conservation. Using genome-wide CRISPR-Cas9 knockout screens, we evaluated the impact of genes associated with resistant transcriptional states on drug sensitivity. Clinical datasets were analyzed to explore the prognostic value of the identified resistance-associated gene signatures. Results: We found that transcriptional states observed in drug-naive cells and shortly after treatment overlapped with those seen in fully resistant populations. Some of these shared features appear to be evolutionarily conserved. Knockout of genes marking resistant states sensitized ovarian cancer cells to Prexasertib. Moreover, early resistance gene signatures effectively distinguished therapy responders from non-responders in multiple clinical cancer trials and differentiated premalignant breast lesions that progressed to malignancy from those that remained benign. Conclusion: Early cellular transcriptional responses to therapy exhibit key similarities to fully resistant states across different drugs, cancer types, and species. Gene signatures defining these early resistance states have prognostic value in clinical settings.