Abstract B013: Host phenotype profiling predicts progression to resection in neoadjuvant pancreatic cancer

Abstract Introduction Neoadjuvant chemotherapy offers a modest survival benefit in pancreatic cancer. Current methods for assessing response to chemotherapy are limited to CA19-9 and post- treatment imaging. However, these methods are not without caveats and focus heavily on tumour biology and staging the tumour, without consideration of host biology or staging the host. We sought to explore the relationship between host biology and outcomes in neoadjuvantly- treated pancreatic cancer, and determine the utility of host phenotype profiling as a tool for predicting chemotherapy response and subsequent progression to resection. Methods Host phenotype profiling was performed by measuring body composition (BC) pre- and post- neoadjuvant chemotherapy in a large international pancreatic cancer cohort (n = 301). Patients with resectable or borderline resectable disease were included if they received chemotherapy with neoadjuvant intent. BC was determined with contrast-enhanced CT scans by measuring single-slice L3 cross-sectional area. Results There were significant differences in BC change from baseline to post-treatment between resected (n = 174) and non-resected patients (n = 127). Patients who did not progress to resection exhibited greater loss in all BC compartments vs those who progressed to resection (muscle area -3.4 vs -0.7, visceral fat -10.1 vs -2.1, subcutaneous fat -12.0 vs -1.4, all p < 0.001, median percentage change per 100 days). We used these values to define a adverse and favourable host phenotypes, which significantly correlated with progression to resection (58.8% adverse phenotype, 74.3% favourable phenotype, p < 0.001). The independence of this effect was confirmed in a logistic regression model with well-established clinico-pathological variables including CA19-19 rise (adverse phenotype OR of non-resection 4.57, 95% CI 1.55-13.48, p = 0.006). Conclusion Host phenotype profiling represents a useful addition to an otherwise limited toolbox of methods for assessing treatment response and selecting for surgery in neoadjuvant pancreatic cancer. Clinicians will be familiar with the subjective “gut feeling” of identifying patients in the clinic who are unlikely to be candidates for surgery, and we demonstrate an objective measurement of this. Early identification of patients on an adverse trajectory may highlight those who would benefit from early intervention such as prehabilitation, salvage change of chemotherapy regimen, or nutritional optimization to switch the phenotype from adverse to favourable and reverse adverse host biology. Citation Format: Adam Bryce, Stephan Dreyer, Ross Dolan, Fieke Froeling, Dario Solinas, Alice Cattelani, Antonio Pea, David Chang. Host phenotype profiling predicts progression to resection in neoadjuvant pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pancreatic Cancer Research; 2024 Sep 15-18; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2024;84(17 Suppl_2):Abstract nr B013.