Leveraging a Diabetes QSP Model to Drive Decisions in Target Identification and Validation for Proinsulin to Insulin Conversion Therapy
Maria Trujillo PhD Principal Scientist, Merck and Co Inc, Kenilworth, NJ July 18, 2019 12:00-1:00 PM EDT Registration (Free) at https://register.gotowebinar.com/register/4089794427217565195?source=website Abstract: Proinsulin is a precursor to insulin that is co-secreted into the blood by the beta cell as a result of incomplete processing. Circulating proinsulin levels increase with increasing insulin resistance in type 2 diabetes mellitus (T2DM). Unlike insulin, proinsulin has limited activity on the insulin receptor. To assess whether the development of peptides engineered to convert proinsulin to insulin in the blood would provide therapeutic value in T2DM, we leveraged a diabetes quantitative systems pharmacology (QSP) model (a physiologically based computational model of glucose homeostasis in humans); internal clinical datasets, and external data from the literature. In silico hypothesis testing included 1) the addition and qualification of proinsulin biology into our diabetes QSP model, 2) the creation of virtual patients (VP) to determine whether proinsulin conversion therapy may provide value to a subpopulation of patients with T2DM based on phenotypic traits, either as a monotherapy or in addition to standards of care (sulfonylureas and metformin), and 3) the simulation of a phase 3 clinical trial with relevant endpoints (including HbA1c and glucose, insulin, and proinsulin) and additional mechanistic readouts (changes in circulating hormones and metabolites during meals and glucose tolerance tests) to interrogate and interpret results. As monotherapy, proinsulin conversion to insulin led to a ~0.2% reduction in HbA1C in diabetic VPs with lesser effects (~0.1%) when added to a standard of care. Virtual patients with higher proinsulin: insulin ratios at baseline showed the greatest reductions. However, to achieve a clinically meaningful HbA1C reduction of ≥ 0.5%, most VPs needed ratios above the reported physiological range. The minimal influence of proinsulin conversion could be explained by the proinsulin secretion and degradation rates relative to respective rates for insulin; these system dynamics were a key learning from the QSP modeling effort. The lack of projected impact on HbA1C through conversion of proinsulin to insulin was not intuitive prior to the in silico hypothesis testing using QSP approaches. The simulation results were examined and challenged with rigor both quantitatively and qualitatively and led to a recommendation not to pursue proinsulin conversion as a potential T2DM therapy. The QSP modeling approach was chosen to capture not only the dynamic interplay between proinsulin and insulin kinetics but their impact on a complex multi-organ system that maintains glucose homeostasis in the body. By thoroughly evaluating the putative therapeutic in diabetic VPs in a Phase 3 setting, we were able to generate sufficient scientific rationale for the termination decision. This effort demonstrates how in silico hypothesis testing through QSP modeling may aid in target identification and validation efforts in the discovery space, conserving R&D resources for targets with greater probability of clinical success.
