IM³PACT: A New Computational Model to Improve the Design of Glucose-Responsive Insulins

IM³PACT is a unique approach to improve MK-2640, glucose-responsive insulin (GRI). However, it did not function as well on humans as it did on animals. To figure out why, scientists created the IM³PACT computer model. The model helped them find differences in how insulin behaves in animals and humans. The scientists used this understanding to design changes that improved insulin efficiency in humans. These computer models can help in the development of better diabetic insulin therapy.

Glucose Behavior Prediction

Glucose-responsive insulin (GRI) analogs are similar to smart insulin because they alter their action depending on the amount of blood sugar. GRI analogs, such as Merck’s MK-2640, have revolutionized diabetes treatment. Unlike conventional insulin, GRIs can prevent low blood sugar and guarantee that patients receive the correct insulin dose. However, MK-2640 had difficulties in human clinical trials, prompting further investigation. The problem was not potency or distribution but poor removal from the body, even at high glucose levels. Computer modeling aids in the development of superior GRIs that work in tandem with the body’s processes, providing more accessible and safer diabetes treatment.

Designing better drugs with IM³PACT

The IM³PACT model is similar to a computer simulation in that it enables researchers to understand how glucose-responsive insulin interacts with the body. The model is divided into two parts: one focuses on how the GRI responds to glucose levels, and the other mimics how glucose, insulin, glucagon (a blood sugar-raising hormone), and the GRI move and change in the body over time.

The model uses a mechanism called competitive clearance in the first component, in which GRI and glucose compete for contact with certain receptors in the body. They bind to these receptors depending on the amounts of GRI and glucose. The researchers also assess how quickly this connection occurs and whether it may be reversed.

The second component of the model uses equations to depict the body’s systems. It explains how insulin, glucagon, and the GRI move around the body, such as the blood and other organs. It analyzes how rapidly insulin travels from the bloodstream into tissues and how it is utilized in various locations, such as the liver.

Total Physiological Model of the Glucose Regulatory System  

 The physiological model of IM³PACT replicates how substances such as glucose and insulin are moved and worked in the body. It is a virtual representation of your body’s systems, using equations to describe what happens if you inject insulin under the skin or in the abdomen and when you eat. It is believed that the body can be regarded as a network of connected compartments, each representing different parts of our bodies. These chambers are like virtual buckets where substances can enter and exit.

When insulin is injected under the skin, the model accounts for how it gets absorbed and distributed in the body. Similarly, the model considers how the body absorbs glucose from the food when a meal is eaten. It also takes into account how quickly substances move between compartments and how they are used in various organs.

This information allows researchers to see how insulin and glucose engage all through the body. It’s critical for designing higher remedies and predicting how specific insulin doses may have an effect on blood sugar levels. Essentially, IM³PACT provides a digital playground to apprehend the complexity of insulin and glucose in the interior human bodies. With a full-frame physiological version, initially designed for humans, it becomes additionally accelerated to consist of Yucatan minipigs, aligning with preclinical research of MK-2640. Rather than genuinely adjusting parameters primarily based totally on frame mass, real anatomical measurements and the drift fees particular to minipigs had been used. These measurements had been accrued from systematic experiments and present research. While positive vital metabolic parameters and transfer functions, which describe modulation of each compartment’s metabolic rates by glucose and insulin, were not readily available. Through this iterative process, it installed a constant set of version parameters for each minipig and humans, which includes the ones particular to MK-2640.

Comparison between MK-2640 and RHI

Addressing differences between MK-2640 and regular human insulin (RHI), the model considers the unique characteristics of MK-2640. MK-2640 exhibits significantly reduced affinity for insulin receptors (IR) both in vitro and in vivo, making it less effective in glucoregulation compared to RHI. To account for this, an effective concentration was used in the model’s transfer functions, scaled by the IR affinity ratio. Despite the reduced IR affinity, achieving the same efficacy as RHI does not require a proportional increase in dose due to MK-2640’s reduced intrinsic clearance. This was incorporated by introducing a scaling factor (Λ_IRC) representing the ratio of GRI clearance to RHI clearance in various compartments. Additionally, the vascular compartments were adjusted in the model to reflect the observed reduction in the volume of distribution for MK-2640, aligning with experimental data.

Experimental Result Analysis

In the minipig model, the IM³PACT simulations are closely aligned with experimental outcomes, accurately reflecting blood glucose and insulin trajectories following intravenous and subcutaneous doses of both RHI and MK-2640. Notably, the simulations effectively replicated MK-2640’s PK and PD behaviors, including the impact of α-methylmannose (α-MM), a potent MR-binding antagonist. MK-2640 demonstrated a slower clearance and induced hypoglycemia, illustrating the influence of the CCM mechanism. In the euglycemic clamp trials (Trial 1), where infusion rates of GRI maintained steady-state concentrations, IM³PACT matched the clinical observations, validating its predictive capacity. Furthermore, MK-2640’s PD properties were compared to RHI, revealing MK-2640’s lower potency and the dosage needed to achieve comparable effects. In Trial 2, which simulated hyperglycemic conditions, the reduction in clearance rates with increasing glucose levels was less pronounced in humans than in dogs or minipigs, indicating a lesser responsiveness of MK-2640 in humans. The simulations allowed cross-species comparisons, shedding light on clearance variations and their potential implications for clinical translation. Changes were observed in MR binding strengths and IC50 values in vivo compared to in vitro predictions, emphasizing the importance of studying these dynamics in physiological contexts. Additionally, MK-2640’s central compartment volume in humans differed from minipigs, suggesting variances in drug distribution between species.

Probing Poor Translatability, Mapping Competitive Clearance, and Designing for Optimal Performance

The poor translatability of MK-2640 from preclinical to clinical trials can be attributed to critical interspecies distinctions in MK-2640 properties. Through in-silico simulations, several potential failure modes were investigated. For instance, altering human IR-mediated clearance to match that of minipigs did not substantially enhance glucose responsiveness, highlighting that differences in intrinsic clearance are not a significant contributor. Similarly, adopting minipig-like IR affinity or central compartment volume in humans had marginal effects on glucose responsiveness. However, adopting minipig-like MR clearance capacity significantly altered clearance rates, emphasizing the importance of MR-mediated clearance in glucose responsiveness.

Furthermore, changes in MR IC50, which indicate CCM sensitivity, impacted glucose response. The simulation results indicate that changes in MR clearance capability and MR IC50 between species significantly impact MK-2640’s glucose response. Further investigation revealed that, while glucose response is more prominent in the liver, it is still rather small in humans compared to minipigs and dogs, supporting prior findings. The findings highlight the complicated interaction of MK-2640 characteristics and the significance of understanding these dynamics for successful clinical translation.

F_MRCL, or the glucose-GRI landscape for competitive clearance, is a dimensionless proxy for visualizing the extent of MR-mediated MK-2640 clearance based on glucose and GRI concentrations within the liver compartment. This environment, which has been analytically mapped, sheds light on the interaction between glucose and GRI levels. F_MRCL in minipigs and humans varies with glucose and GRI concentrations. The F_MRCL landscapes show that CCM is less glucose-responsive in humans than in minipigs, corroborating prior findings. Furthermore, the F_MRCL proxy is used to investigate the GRI Design Space (GRIDS) for MK-2640, allowing the identification of optimal design parameters for higher-performing alternatives. The simulations show a significant difference in clearance modulation between humans and minipigs, implying that competitive clearance GRI variations of MK-2640 have limited translation potential in humans, even with enhanced potency.

Conclusion

The clinical potential of glucose-responsive insulin analogs (GRIs) is assessed using in-silico modeling based on MK-2640 and regular human insulin (RHI) data. The upgraded modeling platform accurately simulated minipigs’ glucoregulatory system and the mechanistic aspects of the competitive clearance mechanism (CCM). It identified crucial inter-species differences in GRI and mannose-receptor properties that hindered MK-2640’s translation to clinical success. Insufficient clearance capacity and an incompatible competitiveness profile within the relevant glycemic range were highlighted as the main translatability challenges. This research underscores the importance of computational modeling in optimizing GRI design and predicting translatability, presenting a new paradigm for integrated computational and experimental drug development.

Article Source: Reference Paper

Learn More: