There is a substantial history of using angiotensin converting enzyme inhibitors (ACEis) to treat congestive heart failure (CHF) by modulating components of the renin-angiotensin-aldosterone system (RAAS) [1]. Optimizing the scheduling of such drugs using lab experiments can be prohibitively difficult. In general, down-regulating the RAAS system (especially angiotensin II) has been associated with improved long-term prognosis in CHF patients [1,2]. This has motivated more recent interest in angiotensin II receptor blockers [1,3]. However, an interesting property of ACEis is their effect on the so-called good RAAS components such as angiotensin 1-7, which is associated with reduced and preserved ejection fraction and reduced risk of heart failure [4].

In this ongoing study in beagles, we have been building a preclinical simulation engine to assist our search for a benazepril dosage which both reduces the RAAS components associated with morbidity and heart failure to a minimum and increases the complementary RAAS components to a maximum.

The first two panels allow the user to specify the parameters of the simulation, and the last panel allows the user to view the effect on four molecules - benazeprilat (the primary active metabolite of benazepril), angiotensin I, angiotensin II, and aldosterone. The prediction distribution (blue) is produced via MCMC simulation methods, from 90% to 10% in steps of 20%. The median experimental response is indicated via a solid black line and the median for simulated placebo group is indicated via the dashed red line. The AUC is calculated between the median experimental response and median placebo reference. It allows for comparison between separate simulations. Preliminary model details presented at PAGE 29.

Bevacizumab-pemetrexed/cisplatin combination therapy (BEV-PEM/CIS) is a first line therapeutic for non-small cell lung cancer (NSCLC) [1]. BEV-PEM/CIS has a narrow therapeutic window – the range of dosages which both reduce the size and spread of the cancer and do not lead to overdose or a damaging accumulation of side-effects. Those side-effects include damage to rapidly dividing healthy cells such as bone marrow and blood cell progenitors. For this reason, BEV-PEM/CIS dosages cannot be scaled to treat aggressive NSCLC [2]. Previous research indicates that administering antiangiogenics (BEV) and and chemotherapeutics (PEM/CIS) sequentially, rather than concomitantly, would greatly improve the efficacy of the combination therapy without leading to additional side-effects. Unfortunately, the optimal gap between BEV and PEM/CIS administration in humans has not been determined [3].

To address this need, we developed a robust preclinical mathematical model of BEV-PEM/CIS. We then scaled that mathematical model to make a first prediction of optimal BEV-PEM/CIS administration in humans. Adjacent is an interactive three-dimensional surface representing the predicted tumor growth over time with respect to the gap between BEV and PEM/CIS administration. It was designed to give an intuitive sense of the behavior of the model. The predicted optimal gap between bevacizumab and pemetrexed/cisplatin administration is 1.2 days. Dosing sequentially with an optimal gap rather than concomitantly improved predicted therapy efficacy (defined as relative tumor volume reduction) by 89.0% over 85 days of treatment. Details of the modeling processed are detailed in a CPT manuscript.

Using a model previously published in the The Journal of Pharmacology and Experimental Therapeutics by Antari Khot et al. as a basis for parameter estimates and model structure, I produced a simulation engine for exogenously administered CAR-T cells in mice using deSolve 1.2.8 within R 4.0.2 [1]. The manuscript by Khot et al. lacked enough documentation and detail for a reader to be able to fully reproduce the model structure documented. So, small modifications to the reported model were necessary to create a stable system. Most modifications were related to kidney and lymph flow dynamics.

I hope others will use my code as a basis to innovate on Khot et al.'s contribution to the modeling community. Adjacent is a panel showing the expected behavior of some of the compartments in the model. All code is annotated and dated. See schnebeni/cart-cell-model on github to download a copy of the project.