1559-P: Heterogeneity of Trajectories of Metabolic Parameters after 50% Beta-Cell Mass Loss by Pancreatectomy

Temporal trajectories of metabolic parameters in the onset of dysglycemia are heterogeneous. We aimed to characterize the temporal trajectories of metabolic parameters after β cell mass reduction by pancreatectomy and to study their heterogeneity. Individuals without known diabetes diagnosis (N = 83) underwent mixed-meal/oral glucose tolerance tests (MMTT/OGTT) and/or hyperglycemic/euglycemic clamp (HC/EC) procedures, before and after surgery. We performed stepwise multivariate linear regression analysis on the glucose tolerance (GT) class (treated as ordinal number, 1 to 3) after surgery, using as independent variables the baselines and changes with surgery of anthropometrics and MMTT- and HC-derived functional parameters of insulin secretion, clearance, and sensitivity (IS), imputed via missForest algorithm when missing. We used the variables selected in this analysis (p<0.01) as input for the reversed graphed embedding (RGE) framework, to identify groups of individuals with extreme combinations of the variables (archetypes). Independent associations with after-surgery GT class (cross-validated R2 = 0.57) were observed for changes in IS and β cell glucose sensitivity (GS), and for baseline IS, GS, 1st phase insulin secretion, insulin secretion at 6 mmol/L glucose, and insulin clearance. IS and the β cell function parameters showed different trajectories combinations in each of the 5 archetypes identified via RGE (median adjusted Rand index = 0.88; N = 16, 8, 15, 13, 18). After surgery, all archetypes included individuals in each of the 3 GT classes (all proportions > 0 at 95% CI). The same β cell mass reduction determines a variety of combinations in changes of IS and β cell functional mechanisms. We identified five archetypes underlying these combinations. The same final GT class can be reached by individuals in any of the archetypes, which shed light on the hidden heterogeneity of glycaemic deterioration. Disclosure R. Bizzotto: None. G. Di Giuseppe: None. L. Soldovieri: None. F. Cinti: None. S. Moffa: None. M. Brunetti: None. G. Ciccarelli: None. S. Alfieri: None. G. Quero: None. A. Mari: Consultant; Lilly Diabetes. A. Giaccari: None. T. Mezza: None.