Abstract
<title>Abstract</title> <p> <bold>Purpose:</bold> Pancreatic cancer (PC) is a highly lethal type of cancer and a highly heterogeneous disease. The current treatment of PC is primarily determined by tumor severity and influences the subsequent course of the disease. Our goal is to address this heterogeneity by identifying distinct PC phenotypes and assessing their relationship to survival and received therapies. <bold>Methods:</bold> We used a registry-based dataset of 1,158 pancreatic cancer patients provided by the Cancer Registry Rhineland-Palatinate and performed an unsupervised hierarchical Ward's linkage analysis using basic patient, tumor, and treatment characteristics. We then explored the relationship between the identified phenotypes and the two-year event-free survival probability, as well as treatment associations within phenotypes using Cox regressions. <bold>Results:</bold> We identified six phenotypes consisting of homogeneous patient, tumor, and treatment characteristics. The young, low-grade, node-negative phenotype~2 defines the favorable prognosis baseline. Male patients in phenotype~5 and elderly female patients in phenotype~1, representing the most typical clinical profile, have twice the risk of death or an event. An even higher risk is observed in advanced profiles characterized by a high proportion of high-grade phenotypes that received surgery (phenotype~4 with partial resection and phenotype~3 with pancreatectomy). Patients with pancreatic cancer (PC) who are metastatic define the high-risk end (phenotype~6). <bold>Conclusion:</bold> The six identified phenotypes capture existing clinical profiles and treatment guidelines. Additionally, they vary significantly in terms of two-year event-free survival, suggesting that they represent clinically meaningful PC profiles that could facilitate the development of novel therapeutic strategies and promote more informed treatment decisions, which are crucial for PC heterogeneity. </p>