Multiscale Mechanistic Modeling in Immunology (in celebration of Denise Kirschner’s 60th birthday)

Marissa Renardy

Applied BioMath (Modeling)

"Capturing CAR-T cell therapy dynamics through semi-mechanistic modeling"

Chimeric antigen receptor T (CAR T) cell therapy has shown remarkable success in treating various leukemias and lymphomas. CARs are engineered to redirect T cells to specific tumor associated antigens. Pharmacokinetic (PK) behavior of CAR T cell therapy is distinct from other therapies due to its 'living' nature; it is characterized by an exponential expansion, fast initial decline (contraction), and slow long-term decline (persistence). Previous models have not mechanistically described all three of these phases. In this work, we develop a semi-mechanistic model of CAR T PK/PD. We use this model to replicate published PK and efficacy data and to explore sources of variability.
Additional authors: Sarah Minucci (Applied BioMath); Scott Gruver (Applied BioMath); Kas Subramanian (Applied BioMath)

---

Pariksheet Nanda

University of Michigan Medical School (Microbiology and Immunology)

"Calibrating multivariate models using CaliPro and approximate Bayesian computing"

Mathematical and computational models are increasingly complex and are typically comprised of one-or-more methods such as ordinary differential equations, partial differential equations, agent-based and rule-based models, etc. Lacking analytical methods, fitting such multivariate biological models to experimental datasets requires iterative parameter sampling-based approaches to establish appropriate ranges of model parameters that capture the corresponding experimental datasets. However, these models typically comprise large numbers of parameters and therefore large degrees of freedom. Thus, fitting these multivariate models to experimental datasets presents significant challenges. We build on our previously published mechanistic, multiscale model of lung granuloma formation from infection by Mycobacteria tuberculosis by calibrating to novel imaging data and metadata from non-human primates to more precisely simulate biological behavior. We apply our model agnostic Calibration Protocol (CaliPro) and explore approximate Bayesian computing (ABC) to highlight strengths and weaknesses among these calibration methods.
Additional authors: Denise Kirschner

---

Maral Budak

University of Michigan Medical School (Department of Microbiology & Immunology)

"Optimizing regimen treatment during the host-pathogen interaction of tuberculosis using a multi-scale computational model"

Tuberculosis (TB) continues to be one of the deadliest infectious diseases in the world, causing ~1.5 million deaths every year. The World Health Organization initiated an End TB Strategy that aims to reduce TB-related deaths in 2035 by 95%. Recent research goals have focused on discovering more effective and more patient-friendly antibiotic drug regimens to increase patient compliance and decrease emergence of resistant TB. To that end, many antibiotics have been identified through in vivo studies and clinical trials that may improve the current standard regimen by shortening treatment time. However, testing every possible combination regimen either in vivo or clinically is not feasible due to experimental and clinical limitations. To identify better regimens more systematically, we simulated pharmacokinetics/pharmacodynamics of various regimens to evaluate efficacies, and then compared our predictions to both clinical trials and nonhuman primate studies. We used GranSim, our well-established hybrid multi-scale agent-based model that simulates granuloma formation and antibiotic treatment, for this task. In addition, we established a multiple-objective surrogate-assisted optimization pipeline using GranSim to discover optimized regimens based on treatment objectives of interest, i.e., minimizing total drug dosage and lowering time needed to sterilize granulomas. Our approach can efficiently test many regimens and successfully identify optimal regimens to inform pre-clinical studies or clinical trials and ultimately accelerate the TB regimen discovery process.
Additional authors: Denise Kirschner (University of Michigan Medical School, Department of Microbiology & Immunology)

---

Christian Michael

University of Michigan - Michigan Medicine (Microbiology & Immunology)

"Towards digital partners of Mycobacterium tuberculosis infection within a virtual city framework"

Mycobacterium tuberculosis (Mtb) is an infectious, airborne microbe that causes tuberculosis (TB), a pandemic infecting roughly a third of the global population. The primary sites of infection are lung granulomas: structures comprising Mtb, immune cells and dead tissue. The high level of granuloma heterogeneity and the slow, complex progression of the disease impacts the relative efficacy of treatments. Considering the challenges of collecting data in TB, it is of critical importance to supplement both experimental and clinical data on TB with computational models that capture the variety of outcomes observed in granulomas both within and between patients . Our group has created HostSim, a host-scale agent-based model of granulomas. Our hybrid framework links lung, lymph node and blood models at multiple spatial scales and is calibrated to experimental data and synthetic data from our fine-grained cell-based granuloma-scale model (GranSim). We next used HostSim to build a Virtual City: a collection of virtual patients, each treated with various medical interventions to quantify impact. We are using Virtual City to build towards Digital Partners - a method for approximating the impact of interventions on specific patients via their most quantitatively similar virtual patients to obtain efficient predictions for effective personalized treatment.
Additional authors: Jennifer Linderman; Denise Kirschner

---

Elsje Pienaar

Purdue University (Biomedical Engineering)

"Biofilm impacts in Non-tuberculous mycobacterial infections in the airway"

Incidence and prevalence of MAC infections are increasing globally, and reinfection is common. Thus, MAC infections present a significant public health challenge. We quantify the impact of MAC biofilms and repeated exposure on infection progression using a computational model of MAC infection in lung airways. MAC biofilms aid epithelial cell invasion, cause premature macrophage apoptosis, and limit antibiotic efficacy. In this computational work we develop an agent-based model that incorporates the interactions between bacteria, biofilm, and immune cells. In this computational model, we perform virtual knockouts to quantify the effects of the biofilm sources (deposited with bacteria vs. formed in the airway), and their impacts on macrophages (inducing apoptosis and slowing phagocytosis). We also quantify the effects of repeated bacterial exposures to assess their impact on infection progression. Our simulations show that chemoattractants released by biofilm-induced apoptosis bias macrophage chemotaxis towards pockets of infected and apoptosed macrophages. This bias results in fewer macrophages finding extracellular bacteria, allowing the extracellular planktonic bacteria to replicate freely. These spatial macrophage trends are further exacerbated with repeated deposition of bacteria. Our model indicates that interventions to abrogate macrophages’ apoptotic responses to bacterial biofilms and/or reduce frequency of patient exposure to bacteria will lower bacterial load, and likely overall risk of infection.
Additional authors: Catherine Weathered; Kelly Pennington; Patricio Escalante

---