Longitudinal morphomics and metabolomics as metastatic pancreatic cancer prognosticators.

Abstract

781 Background: Morphomics evaluates patients’ body composition, including subcutaneous fat, visceral fat, muscle, and fascia, from radiologic scans. Pancreatic ductal adenocarcinoma (PDA) is an aggressive malignancy with few prognosticators. We previously showed baseline morphomics were associated with PDA survival. Here, we study how morphomic variables change over time, are associate with survival, and investigate morphomics and metabolism as potentially modifiable targets to improve patient survival. Methods: Of 528 metastatic PDA patients enrolled on the phase 3 Avenger500 trial, patients without a baseline and follow-up CT scan were removed. To reduce potential effects of reverse causation, patients with <28 days of follow-up were also removed. We used Analytic Morphomics to measure patients’ body composition and untargeted metabolomics data to study patients’ metabolism (200 metabolites). We used linear regression in combination with time-varying Cox proportional hazards models to understand morphomic changes over time and their potential associations with survival. We implemented logistic regression to study how morphomic change was associated with death in the short-term and long-term. We studied body composition decrease (≥10%) with subsequent increase (≥10%) – i.e. “rebound” as a biomarker via its associations with survival, using Cox proportional hazards models to study immediate and long-term outcomes. Lastly, we describe the metabolic landscape of patients who rebounded. Results: Of them, 310 were eligible for morphomic change analysis and 301 (97.1%) lost ≥10% of at least one morphomic variable of which 156 (51.8%) experienced rebound. In total, 163 (52.6%) and 255 (82.3%) of patients lost ≥10% of their muscle and/or subcutaneous fat area. Patients in the highest tertile of muscle and subcutaneous fat area loss experienced shorter survival (hazard ratios, 95% confidence intervals (CIs), and p-values of 1.62 (1.12-2.35) p=0.01 and 2.04 (1.37-3.04) p=5.9x10 -04 ), respectively. However, rebound was associated with lower risks of death and longer survival. Subcutaneous fat rebound in the first 24 weeks (wks) represented 84.9% and 90.5% lower odds of death in the following 4 and 16 wks (odds ratios, 95% CIs, and p-values of 0.15 (0.01-0.90) p=8.7x10 -02 and 0.09 (0.02-0.31) p=4.7x10 -04 ), respectively, than patients who but did not rebound. Conclusions: We identified multiple morphomics as novel biomarkers for immediate (4 wk) and long-term (overall) survival, particularly in patients who rebound. We show body composition is intimately associated with metabolism and identify specific biomarkers with interventional plausibility to improve pancreatic cancer patient survival.