Associations between BMI, morphomics and survival in 528 patients with metastatic pancreatic cancer.

Abstract

637 Background: Literature describing associations between patient body mass index (BMI) and survival in pancreatic ductal adenocarcinoma (PDA) remains inconsistent. This heterogeneity may be due to studies using pre-diagnosis BMI and including few metastatic patients despite high prevalence at diagnosis. Furthermore, BMI may not adequately reflect other body compositions such as those reflected via morphomic variables (subcutaneous fat, visceral fat, muscle, and fascia) that may have unique associations with survival. In this study, we sought to explore the relationships between detailed body composition in patients with metastatic PDA on both overall survival (OS) and progression-free survival (PFS). Methods: We evaluated data and scans from a phase 3 clinical trial in which 528 patients with metastatic PDA with no prior chemotherapy were randomized to either FOLFIRINOX or modified FOLFIRINOX plus devimistat. No differences in response rate, median PFS or OS were observed. Of the 528 enrolled patients, 377 had baseline CT scans eligible for morphomic evaluation for which Analytic Morphomics was used to measure body composition. For analysis, morphomic variables were summed across the T10-T12 and L1-L4 vertebrae and then divided into tertiles. BMI at diagnosis was evaluated in tertiles and per WHO standards. Cox proportional hazards models and cubic splines were used to evaluate associations between BMI, morphomic variables and PFS, OS. Univariate and multivariate models were evaluated. Kaplan-Meier plots were created to evaluate median survival times between those in the highest and lowest tertiles (R v4.3.1). Results: We did not observe an association between BMI and PDA patient survival: HR (95% CI) comparing BMI ≥30 (obese) to 18.5-24.9 (healthy weight), PFS 0.96 (0.66, 1.39), p-trend=0.80; OS 0.87 (0.64, 1.18), p-trend=0.30. However, multiple morphomic variables were significantly associated with PFS and/or OS in univariate and multivariate models when comparing the highest to lowest tertile. More subcutaneous fat was associated with longer median OS (12.4 vs 10.5 mos, p=0.006; HR (95% CI) 0.41 (0.25-0.68), p-trend=0.001). More dense visceral fat was associated with shorter median PFS (6.0 vs 9.1 mos, p=0.05, HR (95% CI) 1.52 (1.00-2.33), p-trend=0.047). A higher muscle to fascia ratio was associated with longer PFS (9.3 vs 6.5 mos, p=0.03; HR (95% CI) 0.58 (0.38-0.88), p-trend=0.01) and median OS (13.9 vs 10 mos, p=0.005; HR (95% CI) 0.53 (0.38-0.76), p-trend=0.0004). Conclusions: BMI was not associated with PFS or OS in this large prospective study of patients with metastatic PDA receiving chemotherapy. However, morphomic variables including subcutaneous fat area, visceral fat density, and the muscle to fascia ratio were statistically significantly associated with metastatic PDA patient survival.