Mathematical modeling in autoimmune diseases: from theory to clinical application

The research & development (R&D) of novel therapeutic agents for the treatment of autoimmune diseases is challenged by highly complex pathogenesis and multiple etiologies of these conditions. The number of targeted therapies available on the market is limited, whereas the prevalence of autoimmune conditions in the global population continues to rise. Mathematical modeling of biological systems is an essential tool which may be applied in support of decision-making across R&D drug programs to improve the probability of success in the development of novel medicines. Over the past decades, multiple models of autoimmune diseases have been developed. Models differ in the spectra of quantitative data used in their development and mathematical methods, as well as in the level of "mechanistic granularity" chosen to describe the underlying biology. Yet, all models strive towards the same goal: to quantitatively describe various aspects of the immune response. The aim of this review was to conduct a systematic review and analysis of mathematical models of autoimmune diseases focused on the mechanistic description of the immune system, to consolidate existing quantitative knowledge on autoimmune processes, and to outline potential directions of interest for future model-based analyses. Following a systematic literature review, 38 models describing the onset, progression, and/or the effect of treatment in 13 systemic and organ-specific autoimmune conditions were identified, most models developed for inflammatory bowel disease, multiple sclerosis, and lupus (5 models each). ≥70% of the models were developed as nonlinear systems of ordinary differential equations, others - as partial differential equations, integro-differential equations, Boolean networks, or probabilistic models. Despite covering a relatively wide range of diseases, most models described the same components of the immune system, such as T-cell response, cytokine influence, or the involvement of macrophages in autoimmune processes. All models were thoroughly analyzed with an emphasis on assumptions, limitations, and their potential applications in the development of novel medicines.

An integrative mechanistic model of thymocyte dynamics

Background

The thymus plays a central role in shaping human immune function. A mechanistic, quantitative description of immune cell dynamics and thymic output under homeostatic conditions and various patho-physiological scenarios are of particular interest in drug development applications, e.g., in the identification of potential therapeutic targets and selection of lead drug candidates against infectious diseases.

Methods

We here developed an integrative mathematical model of thymocyte dynamics in human. It incorporates mechanistic features of thymocyte homeostasis as well as spatial constraints of the thymus and considerations of age-dependent involution. All model parameter estimates were obtained based on published physiological data of thymocyte dynamics and thymus properties in mouse and human. We performed model sensitivity analyses to reveal potential therapeutic targets through an identification of processes critically affecting thymic function; we further explored differences in thymic function across healthy subjects, multiple sclerosis patients, and patients on fingolimod treatment.

Results

We found thymic function to be most impacted by the egress, proliferation, differentiation and death rates of those thymocytes which are most differentiated. Model predictions also showed that the clinically observed decrease in relapse risk with age, in multiple sclerosis patients who would have discontinued fingolimod therapy, can be explained mechanistically by decreased thymic output with age. Moreover, we quantified the effects of fingolimod treatment duration on thymic output.

Conclusions

In summary, the proposed model accurately describes, in mechanistic terms, thymic output as a function of age. It may be further used to perform predictive simulations of clinically relevant scenarios which combine specific patho-physiological conditions and pharmacological interventions of interest.

Optimization of the MACE endpoint composition to increase power in studies of lipid-lowering therapies-a model-based meta-analysis

Aims

To develop a model-informed methodology for the optimization of the Major Adverse Cardiac Events (MACE) composite endpoint, based on a model-based meta-analysis across anti-hypercholesterolemia trials of statin and anti-PCSK9 drugs.

Methods and results

Mixed-effects meta-regression modeling of stand-alone MACE outcomes was performed, with therapy type, population demographics, baseline and change over time in lipid biomarkers as predictors. Randomized clinical trials up to June 28, 2022, of either statins or anti-PCSK9 therapies were identified through a systematic review process in PubMed and ClinicalTrials.gov databases. In total, 54 studies (270,471 patients) were collected, reporting 15 different single cardiovascular events. Treatment-mediated decrease in low density lipoprotein cholesterol, baseline levels of remnant and high-density lipoprotein cholesterol as well as non-lipid population characteristics and type of therapy were identified as significant covariates for 10 of the 15 outcomes. The required sample size per composite 3- and 4-point MACE endpoint was calculated based on the estimated treatment effects in a population and frequencies of the incorporated events in the control group, trial duration, and uncertainty in model parameters.

Conclusion

A quantitative tool was developed and used to benchmark different compositions of 3- and 4-point MACE for statins and anti-PCSK9 therapies, based on the minimum population size required to achieve statistical significance in relative risk reduction, following meta-regression modeling of the single MACE components. The approach we developed may be applied towards the optimization of the design of future trials in dyslipidemia disorders as well as in other therapeutic areas.

Abstract LB212: Allogeneic, IL-2-independent tumor-infiltrating lymphocytes expressing membrane-bound IL-15 (cytoTIL15࣪) eradicate tumors in a melanoma PDX model through recognition of shared tumor antigens

Standard tumor-infiltrating lymphocyte (TIL) therapy requires IL-2 administration to support TIL expansion and survival, but this cytokine is associated with T cell exhaustion and can result in severe toxicities that limit patient eligibility (1). To this end, we genetically engineered TIL to express membrane-bound IL-15 (mbIL15) under the control of Obsidian’s cytoDRIVE® technology (cytoTIL15࣪), which allows regulation of protein expression via a drug-responsive domain upon acetazolamide (ACZ) administration. IL-15 is a preferred cytokine over IL-2 to mediate TIL activation and expansion, because it does not result in CD8 T cell exhaustion or stimulate regulatory CD4 T cells, and enhances development of a memory T-cell phenotype. We have previously demonstrated IL-2-independent, 3-6-fold increased cytoTIL15 persistence in an antigen-independent setting relative to unengineered TIL therapy with IL-2 (uTIL) (2). Due to the challenge of generating autologous tumor/TIL-matched pairs and most importantly, to assess cytoTIL15 cell’s functional impact on anti-tumor growth across multiple donors, we developed an allogeneic patient-derived xenograft (PDX) model. To establish the model, different melanoma tumor digests were co-incubated in vitro with select HLA-A*02-matched, allogeneic melanoma TIL donors to assess their reactivity. Tumors were screened for expression of shared antigens, such as gp100 and MART1, and TIL donor TCRs were screened with tetramers. Once established, serially passaged tumor fragments were grown, measured, and randomized into groups to receive intravenous transfer of TIL (n=8/cohort). Mice receiving uTIL were treated with four saturating doses of recombinant IL-2, and mice receiving cytoTIL15 cells received either vehicle or oral 200 mg/kg ACZ daily for the entire study, without any IL-2. Three of four cytoTIL15 cell preparations from different donors dosed with ACZ achieved significant tumor growth inhibition compared to uTIL. Four mice developed complete responses as early as 17 days post cytoTIL15 cell transfer. The level of anti-tumor response was associated with increased frequency of MART1-reactive cytoTIL15 cells. On day 20 after TIL transfer, tumors and secondary lymphoid organs were collected (n=4/cohort). Tumors treated with cytoTIL15 cells + ACZ showed an 8-10-fold increased TIL infiltration compared to uTIL or cytoTIL15 cells + vehicle. Moreover, enhanced cytoTIL15 cell infiltration and anti-tumor activity was associated with increases in pro-inflammatory cytokines (e.g., IFNγ). Taken together, these data clearly demonstrate the superiority of cytoTIL15 cells over uTIL for controlling or eradicating melanoma tumor outgrowth and the utility of an allogeneic PDX model for comparative evaluation of tumor-antigen specific TIL reactivity.

References: 1. Yang JC. Toxicities associated with adoptive T-cell transfer for Cancer. Cancer J. 2015. 2. Burga R. et al Genetically engineered tumor-infiltrating lymphocytes (cytoTIL15) exhibit IL-2-independent persistence and anti-tumor efficacy against melanoma in vivo. SITC 36th annual meeting 2021.

Citation Format: Jeremy H. Tchaicha, Scott Lajoie, Rachel Burga, Theresa Ross, Benjamin Primack, Meghan Langley, Violet Young, Alonso Villasmil Ocando, Kyle Pedro, Jack Tremblay, Gauri Kulkarni, Mithun Khattar, Dhruv Sethi, Michelle Ols, Gabriel Helmlinger, Gary Vanasse, Shyam Subramanian, Jan ter Meulen. Allogeneic, IL-2-independent tumor-infiltrating lymphocytes expressing membrane-bound IL-15 (cytoTIL15࣪) eradicate tumors in a melanoma PDX model through recognition of shared tumor antigens [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr LB212.

A workflow for the joint modeling of longitudinal and event data in the development of therapeutics: Tools, statistical methods, and diagnostics

Clinical trials investigate treatment endpoints that usually include measurements of pharmacodynamic and efficacy biomarkers in early‐phase studies and patient‐reported outcomes as well as event risks or rates in late‐phase studies. In recent years, a systematic trend in clinical trial data analytics and modeling has been observed, where retrospective data are integrated into a quantitative framework to prospectively support analyses of interim data and design of ongoing and future studies of novel therapeutics. Joint modeling is an advanced statistical methodology that allows for the investigation of clinical trial outcomes by quantifying the association between baseline and/or longitudinal biomarkers and event risk. Using an exemplar data set from non‐small cell lung cancer studies, we propose and test a workflow for joint modeling. It allows a modeling scientist to comprehensively explore the data, build survival models, investigate goodness‐of‐fit, and subsequently perform outcome predictions using interim biomarker data from an ongoing study. The workflow illustrates a full process, from data exploration to predictive simulations, for selected multivariate linear and nonlinear mixed‐effects models and software tools in an integrative and exhaustive manner.

166 Genetically engineered tumor-infiltrating lymphocytes (cytoTIL15) exhibit IL-2-independent persistence and anti-tumor efficacy against melanoma in vivo

Background

Adoptive cell therapy with tumor-infiltrating lymphocytes (TILs) has demonstrated tremendous promise in clinical trials for patients with solid or metastatic tumors.1 However, current TIL therapy requires systemic administration of IL-2 to promote TIL survival, and IL-2-associated toxicities greatly limit patient eligibility and reduce the long-term clinical benefit of TIL therapy.2 3 Unlike IL-2, which promotes T cell exhaustion, IL-15 maintains antigen-independent TIL persistence through homeostatic proliferation and supports CD8+ T cell anti-tumor activity without stimulating regulatory T cells. We designed genetically engineered TILs to express a regulated form of membrane-bound IL-15 (mbIL15) for tunable long-term persistence, leading to enhanced efficacy and safety for the treatment of patients with solid tumors.

Methods

Obsidian’s cytoDRiVE™ platform includes small human protein sequences called drug responsive domains (DRD)s that enable regulated expression of a fused target protein under control of FDA-approved, bioavailable small molecule ligands. cytoTIL15 contains TILs engineered with mbIL15 under the control of a carbonic-anhydrase-2 DRD, controlled by the ligand acetazolamide (ACZ). After isolation from tumors, TILs were transduced and expanded in vitro through a proprietary TIL expansion process. cytoTIL15 were immunophenotyped and assessed for in vitro antigen-independent survival and co-cultured with tumor cells to assess polyfunctionality and cytotoxicity. In vivo TIL persistence and anti-tumor efficacy was evaluated through adoptive transfer of TILs into immunodeficient NSG mice, either naïve or implanted with subcutaneous patient-derived-xenograft (PDX) tumors.

Results cytoTIL15 and conventional IL2-dependent TILs isolated from melanoma tumor samples expanded to clinically relevant numbers over 14 days. Throughout expansion, cytoTIL15 were enriched for CD8+ T cells and acquired enhanced memory-like characteristics, while maintaining diverse TCRVβ sub-family representation. cytoTIL15 demonstrated enhanced potency over conventional TILs, as measured by increased polyfunctionality and cytotoxicity against tumor and PDX lines in vitro (figure 1A). In a 10-day antigen-independent in vitro assay, cytoTIL15 persisted at greater frequencies than conventional TILs in the absence of IL-2 (figure 1B; *p<0.05). cytoTIL15 adoptively transferred into naïve NSG mice demonstrated ACZ-dependent long-term persistence without antigen or exogenous IL-2, whereas conventional TILs were undetectable >30 days following adoptive cell transfer (figure 1C). Importantly, cytoTIL15 achieved significant tumor control in a human PDX model (figure 1D), which correlated with increased TIL accumulation in secondary lymphoid organs.

Abstract 166 Figure 1

cytoTIL15 demonstrate superior persistence. cytoTIL15 is an engineered TIL product expressing regulatable mbIL15. (A) cytoTIL15 demonstrate enhanced in vitro cytotoxicity after co-culture with melanoma tumor lines (representative data from 3 TIL donors). (B) cytoTIL15 have improved persistence in antigen- and IL2- independent culture conditions in vitro compared to conventional TILs cultured in the absence of IL-2 as well as (C) in vivo compared to conventional TILs supplemented with IL-2, when engrafted into NSG mice (in vitro: representative data from 1 TIL donor, performed in >3 replicate donors, in vivo: n=5/group, representative of 1 TIL donor, performed in >3 replicate donors). (D) cytoTIL15 (with 200mg/kg ACZ PO QD) demonstrate enhanced anti-tumor efficacy in a xenograft melanoma model as compared to conventional TILs (with 50000 IU IL-2 q8h BID, IP for 5 days) (n=8/group, representative of 1 TIL donor, performed in >2 replicate donors; ACT = adoptive cell transfer).

Conclusions

Taken together, the superior persistence and potency of cytoTIL15 in the complete absence of IL-2 highlights the clinical potential of cytoTIL15 as a novel TIL product with enhanced safety and efficacy for patients with melanomas, and other solid tumors.

Acknowledgements

The authors wish to acknowledge the Cooperative Human Tissue Network for the their supply of human tumor tissue, and the MD Anderson Cancer Center for technical support; schematic created with BioRender.com.

References

Chandran SS, Somerville RPT, Yang JC, Sherry RM, Klebanoff CA, Goff SL, Wunderlich JR, Danforth DN, Zlott D, Paria BC, Sabesan AC, Srivastava AK, Xi L, Pham TH, Raffeld M, White DE, Toomey MA, Rosenberg SA, Kammula US. Treatment of metastatic uveal melanoma with adoptive transfer of tumour-infiltrating lymphocytes: a single-centre, two-stage, single-arm, phase 2 study. Lancet Oncol 2017 Jun;18(6):792–802. doi: 10.1016/S1470-2045(17)30251-6. Epub 2017 Apr 7. PMID: 28395880; PMCID: PMC5490083.

Yang JC. Toxicities associated with adoptive T-cell transfer for Cancer. Cancer J 2015;21:506–9.

Schwartz RN, Stover L, Dutcher JP. Managing toxicities of high-dose interleukin-2. Oncology (Williston Park) 2002 Nov;16(11 Suppl 13):11–20. PMID: 12469935.

Evaluation of Combination Strategies for the A2AR Inhibitor AZD4635 Across Tumor Microenvironment Conditions via a Systems Pharmacology Model

Background

Adenosine receptor type 2 (A2AR) inhibitor, AZD4635, has been shown to reduce immunosuppressive adenosine effects within the tumor microenvironment (TME) and to enhance the efficacy of checkpoint inhibitors across various syngeneic models. This study aims at investigating anti-tumor activity of AZD4635 alone and in combination with an anti-PD-L1-specific antibody (anti-PD-L1 mAb) across various TME conditions and at identifying, via mathematical quantitative modeling, a therapeutic combination strategy to further improve treatment efficacy.

Methods

The model is represented by a set of ordinary differential equations capturing: 1) antigen-dependent T cell migration into the tumor, with subsequent proliferation and differentiation into effector T cells (Teff), leading to tumor cell lysis; 2) downregulation of processes mediated by A2AR or PD-L1, as well as other immunosuppressive mechanisms; 3) A2AR and PD-L1 inhibition by, respectively, AZD4635 and anti-PD-L1 mAb. Tumor size dynamics data from CT26, MC38, and MCA205 syngeneic mice treated with vehicle, anti-PD-L1 mAb, AZD4635, or their combination were used to inform model parameters. Between-animal and between-study variabilities (BAV, BSV) in treatment efficacy were quantified using a non-linear mixed-effects methodology.

Results

The model reproduced individual and cohort trends in tumor size dynamics for all considered treatment regimens and experiments. BSV and BAV were explained by variability in T cell-to-immunosuppressive cell (ISC) ratio; BSV was additionally driven by differences in intratumoral adenosine content across the syngeneic models. Model sensitivity analysis and model-based preclinical study simulations revealed therapeutic options enabling a potential increase in AZD4635-driven efficacy; e.g., adoptive cell transfer or treatments affecting adenosine-independent immunosuppressive pathways.

Conclusions

The proposed integrative modeling framework quantitatively characterized the mechanistic activity of AZD4635 and its potential added efficacy in therapy combinations, across various immune conditions prevailing in the TME. Such a model may enable further investigations, via simulations, of mechanisms of tumor resistance to treatment and of AZD4635 combination optimization strategies.

Longitudinal FEV1 and Exacerbation Risk in COPD: Quantifying the Association Using Joint Modelling

BACKGROUND

Lung function, measured as forced expiratory volume in one second (FEV1), and exacerbations are two endpoints evaluated in chronic obstructive pulmonary disease (COPD) clinical trials. Joint analysis of these endpoints could potentially increase statistical power and enable assessment of efficacy in shorter and smaller clinical trials.

OBJECTIVE

To evaluate joint modelling as a tool for analyzing treatment effects in COPD clinical trials by quantifying the association between longitudinal improvements in FEV1 and exacerbation risk reduction.

METHODS

A joint model of longitudinal FEV1 and exacerbation risk was developed based on patient-level data from a Phase III clinical study in moderate-to-severe COPD (1740 patients), evaluating efficacy of fixed-dose combinations of a long-acting bronchodilator, formoterol, and an inhaled corticosteroid, budesonide. Two additional studies (1604 and 1042 patients) were used for external model validation and parameter re-estimation.

RESULTS

A significant (p<0.0001) association between FEV1 and exacerbation risk was estimated, with an approximate 10% reduction in exacerbation risk per 100 mL improvement in FEV1, consistent across trials and treatment arms. The risk reduction associated with improvements in FEV1 was relatively small compared to the overall exacerbation risk reduction for treatment arms including budesonide (10-15% per 160 µg budesonide). High baseline breathlessness score and previous history of exacerbations also influenced the risk of exacerbation.

CONCLUSION

Joint modelling can be used to co-analyze longitudinal FEV1 and exacerbation data in COPD clinical trials. The association between the endpoints was consistent and appeared unrelated to treatment mechanism, suggesting that improved lung function is indicative of an exacerbation risk reduction. The risk reduction associated with improved FEV1 was, however, generally small and no major impact on exacerbation trial design can be expected based on FEV1 alone. Further exploration with other longitudinal endpoints should be considered to further evaluate the use of joint modelling in analyzing COPD clinical trials.

Longitudinal Tumor Size and Neutrophil-to-Lymphocyte Ratio Are Prognostic Biomarkers for Overall Survival in Patients With Advanced Non-Small Cell Lung Cancer Treated With Durvalumab

Therapy optimization remains an important challenge in the treatment of advanced non-small cell lung cancer (NSCLC). We investigated tumor size (sum of the longest diameters (SLD) of target lesions) and neutrophil-to-lymphocyte ratio (NLR) as longitudinal biomarkers for survival prediction. Data sets from 335 patients with NSCLC from study NCT02087423 and 202 patients with NSCLC from study NCT01693562 of durvalumab were used for model qualification and validation, respectively. Nonlinear Bayesian joint models were designed to assess the impact of longitudinal measurements of SLD and NLR on patient subgrouping (by Response Evaluation Criteria in Solid Tumors 1.1 criteria at 3 months after therapy start), long-term survival, and precision of survival predictions. Various validation scenarios were investigated. We determined a more distinct patient subgrouping and a substantial increase in the precision of survival estimates after the incorporation of longitudinal measurements. The highest performance was achieved using a multivariate SLD and NLR model, which enabled predictions of NSCLC clinical outcomes.

Quantification of Scheduling Impact on Safety and Efficacy Outcomes of Brain Metastasis Radio- and Immuno-Therapies: A Systematic Review and Meta-Analysis

Objectives: The goal of this quantitative research was to evaluate the impact of various factors (e.g., scheduling or radiotherapy (RT) type) on outcomes for RT vs. RT in combination with immune checkpoint inhibitors (ICI), in the treatment of brain metastases, via a meta-analysis. Methods: Clinical studies with at least one ICI+RT treatment combination arm with brain metastasis patients were identified via a systematic literature search. Data on 1-year overall survival (OS), 1-year local control (LC) and radionecrosis rate (RNR) were extracted; for combination studies which included an RT monotherapy arm, odds ratios (OR) for the aforementioned endpoints were additionally calculated and analyzed. Mixed-effects meta-analysis models were tested to evaluate impact on outcome, for different factors such as combination treatment scheduling and the type of ICI or RT used. Results: 40 studies representing a total of 4,359 patients were identified. Higher 1-year OS was observed in ICI and RT combination vs. RT alone, with corresponding incidence rates of 59% [95% CI: 54-63%] vs. 32% [95% CI: 25-39%] (P < 0.001). Concurrent ICI and RT treatment was associated with significantly higher 1-year OS vs. sequential combinations: 68% [95% CI: 60-75%] vs. 54% [95% CI: 47-61%]. No statistically significant differences were observed in 1-year LC and RNR, when comparing combinations vs. RT monotherapies, with 1-year LC rates of 68% [95% CI: 40-90%] vs. 72% [95% CI: 63-80%] (P = 0.73) and RNR rates of 6% [95% CI: 2-13%] vs. 9% [95% CI: 5-14%] (P = 0.37). Conclusions: A comprehensive, study-level meta-analysis of brain metastasis disease treatments suggest that combinations of RT and ICI result in higher OS, yet comparable neurotoxicity profiles vs. RT alone, with a superiority of concurrent vs. sequential combination regimens. A similar meta-analysis using patient-level data from past trials, as well as future prospective randomized trials would help confirming these findings.

Renal Effects of Dapagliflozin in People with and without Diabetes with Moderate or Severe Renal Dysfunction: Prospective Modeling of an Ongoing Clinical Trial

Sodium glucose cotransporter 2 inhibitors (SGLT2i) reduce cardiovascular events and onset and progression of renal disease by mechanisms that remain incompletely understood but may include clearance of interstitial congestion and reduced glomerular hydrostatic pressure. The ongoing DAPASALT mechanistic clinical study will evaluate natriuretic, diuretic, plasma/extracellular volume, and blood pressure responses to dapagliflozin in people with type 2 diabetes with normal or impaired renal function (D-PRF and D-IRF, respectively) and in normoglycemic individuals with renal impairment (N-IRF). In this study, a mathematical model of renal physiology, pathophysiology, and pharmacology was used to prospectively predict changes in sodium excretion, blood and interstitial fluid volume (IFV), blood pressure, glomerular filtration rate, and albuminuria in DAPASALT. After validating the model with previous diabetic nephropathy trials, virtual patients were matched to DAPASALT inclusion/exclusion criteria, and the DAPASALT protocol was simulated. Predicted changes in glycosuria, blood pressure, glomerular filtration rate, and albuminuria were consistent with other recent studies in similar populations. Predicted albuminuria reductions were 46% in D-PRF, 34.8% in D-IRF, and 14.2% in N-IRF. The model predicts a similarly large IFV reduction between D-PRF and D-IRF and less, but still substantial, IFV reduction in N-IRF, even though glycosuria is attenuated in groups with impaired renal function. When DAPASALT results become available, comparison with these simulations will provide a basis for evaluating how well we understand the cardiorenal mechanism(s) of SGLT2i. Meanwhile, these simulations link dapagliflozin's renal mechanisms to changes in IFV and renal biomarkers, suggesting that these benefits may extend to those with impaired renal function and individuals without diabetes. SIGNIFICANCE STATEMENT: Mechanisms of SGLT2 inhibitors' cardiorenal benefits remain incompletely understood. We used a mathematical model of renal physiology/pharmacology to prospectively predict responses to dapagliflozin in the ongoing DAPASALT study. Key predictions include similarly large interstitial fluid volume (IFV) reductions between subjects with normal and impaired renal function and less, but still substantial, IFV reduction in those without diabetes, even though glycosuria is attenuated in these groups. Comparing prospective simulations and study results will assess how well we understand the cardiorenal mechanism(s) of SGLT2 inhibitors.

Dose dependence of treatment-related adverse events for immune checkpoint inhibitor therapies: a model-based meta-analysis

Programmed cell death-1 (PD-1) and/or cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) immune checkpoint inhibitor (ICI) treatments are associated with adverse events (AEs), which may be dependent on ICI dose. Applying a model-based meta-analysis to evaluate safety data from published clinical trials from 2005 to 2018, we analyzed the dose/exposure dependence of ICI treatment-related AE (trAE) and immune-mediated AE (imAE) rates. Unlike with PD-1 inhibitor monotherapy, CTLA-4 inhibitor monotherapy exhibited a dose/exposure dependence on most AE types evaluated. Furthermore, combination therapy with PD-1 inhibitor significantly strengthened the dependence of trAE and imAE rates on CTLA-4 inhibitor dose/exposure.

Comparative quantitative safety profiles of PD-1 and PD-L1 checkpoint inhibitor monotherapies: A Bayesian model-based meta-analysis

e15167

Background: Anti-cancer therapies with immune checkpoint inhibitors (ICIs) are associated with immune-mediated adverse events (imAEs). The objective of this study was to use a Bayesian model-based meta-analysis, to compare safety profiles of PD-1 and PD-L1 checkpoint inhibitor monotherapies. Methods: We performed an exhaustive search through PubMed and TrialTrove databases, ASCO and ESMO abstracts to identify relevant ICI safety data. A Bayesian meta-analysis (BMA) was performed for all-grade and high-grade (Grades 3-4) treatment-related AEs (trAEs) and immune-mediated AEs (imAEs). The analysis was performed for total AEs and individual AEs affecting different organs: dermatological (rash, pruritus), gastrointestinal (diarrhea, colitis), hepatic (AST, ALT), respiratory (pneumonitis), and endocrine (hyperthyroidism, hypothyroidism). Software STAN along with the brms package in R software were used for the BMA. Results: A total of 145 articles were identified, covering 30,737 patients in 153 cohorts treated with anti PD-1 and anti PD-L1 monotherapies. For total all-grade and grade 3&4 trAEs, anti PD-L1 demonstrated lower AE rates vs. anti PD-1 (65% vs. 69% and 12% vs. 16%, respectively). Results were similar for total all-grade and high-grade imAEs (19% vs. 26% and 4% vs. 7%, respectively). The analysis of all-grade individual trAEs showed that anti PD-L1 exhibited lower AE rates vs. anti PD-1 for rash (8% vs. 10.5%), pruritus (9% vs. 12%), diarrhea (8.5% vs. 11%), colitis (1.0% vs. 1.6%), AST (3.0% vs. 5.5%), ALT (3.5% vs. 5.6%), pneumonitis (3.0% vs. 3.9%), hyperthyroidism (0.8% vs. 4.0%) and hypothyroidism (6.0% vs. 8.0%). For grade 3&4 trAEs, differences between anti PD-L1 and anti PD-1 were: 0.6% vs. 1.3% for diarrhea, 0.7% vs. 1.4% for colitis, and 0.9% vs 1.6% for pneumonitis. All-grade imAEs were less frequent for anti PD-L1 vs. anti PD-1 for: rash (6.5% vs. 10%), colitis (1.0% vs. 2.0%), AST (3.5% vs. 3.9%), ALT (2.9% vs. 4.0%), pneumonitis (2.0% vs. 4.0%), hyperthyroidism (2.5% vs. 4.0%) and hypothyroidism (5.5% vs. 9.0%). Similar results were obtained for grade 3&4 imAEs incidences: rash (0.5% vs. 0.9%), diarrhea (0.5% vs. 1.5%), colitis (0.5% vs. 1.1%) and pneumonitis (0.9% vs. 1.4%). Conclusions: Significant differences in trAEs and imAEs rates were shown for anti PD-L1 vs. anti PD-1 monotherapies. Anti PD-L1 demonstrated more favorable safety profiles for total and individual AE rates. Notably, AE rates for pneumonitis and colitis were twice as high for anti PD-1 vs. anti PD-L1 therapies.

Differentiating the Sodium-Glucose Cotransporter 1 Inhibition Capacity of Canagliflozin vs. Dapagliflozin and Empagliflozin Using Quantitative Systems Pharmacology Modeling

The aim of this research was to differentiate dapagliflozin, empagliflozin, and canagliflozin based on their capacity to inhibit sodium‐glucose cotransporter (SGLT) 1 and 2 in patients with type 2 diabetes using a previously developed quantitative systems pharmacology model of renal glucose filtration, reabsorption, and excretion. The analysis was based on pooled, mean study‐level data on 24‐hour urinary glucose excretion, average daily plasma glucose, and estimated glomerular filtration rate collected from phase I and II clinical trials of SGLT2 inhibitors. Variations in filtered glucose across clinical studies were shown to drive the apparent differences in the glucosuria dose–response relationships among the gliflozins. A normalized dose–response analysis demonstrated similarity of dapagliflozin and empagliflozin, but not canagliflozin. At approved doses, SGLT1 inhibition by canagliflozin but not dapagliflozin or empagliflozin contributed to ~ 10% of daily urinary glucose excretion.

A Physiology-Based Model of Bile Acid Distribution and Metabolism Under Healthy and Pathologic Conditions in Human Beings

BACKGROUND & AIMS

Disturbances of the enterohepatic circulation of bile acids (BAs) are seen in a number of clinically important conditions, including metabolic disorders, hepatic impairment, diarrhea, and gallstone disease. To facilitate the exploration of underlying pathogenic mechanisms, we developed a mathematical model built on quantitative physiological observations across different organs.

METHODS

The model consists of a set of kinetic equations describing the syntheses of cholic, chenodeoxycholic, and deoxycholic acids, as well as time-related changes of their respective free and conjugated forms in the systemic circulation, the hepatoportal region, and the gastrointestinal tract. The core structure of the model was adapted from previous modeling research and updated based on recent mechanistic insights, including farnesoid X receptor-mediated autoregulation of BA synthesis and selective transport mechanisms. The model was calibrated against existing data on BA distribution and feedback regulation.

RESULTS

According to model-based predictions, changes in intestinal motility, BA absorption, and biotransformation rates affected BA composition and distribution differently, as follows: (1) inhibition of transintestinal BA flux (eg, in patients with BA malabsorption) or acceleration of intestinal motility, followed by farnesoid X receptor down-regulation, was associated with colonic BA accumulation; (2) in contrast, modulation of the colonic absorption process was predicted to not affect the BA pool significantly; and (3) activation of ileal deconjugation (eg, in patents with small intestinal bacterial overgrowth) was associated with an increase in the BA pool, owing to higher ileal permeability of unconjugated BA species.

CONCLUSIONS

This model will be useful in further studying how BA enterohepatic circulation modulation may be exploited for therapeutic benefits.

Quantification of dose dependence and frequency of checkpoint inhibitor immune-mediated adverse events: A Bayesian model-based meta-analysis

83

Background: Immune checkpoint inhibitors (ICIs) are associated with immune-mediated adverse events (imAEs). The objective of this study was to use a Bayesian model-based meta-analysis to quantify dose dependence and compare imAE frequencies for PD-1, PD-L1 and CTLA-4 inhibitor monotherapies and their combinations. Methods: We searched PubMed, TrialTrove, ASCO and ESMO databases and retrieved relevant ICI safety data. In order to quantitatively compare safety across doses and drugs against a given target, we converted the various dose regimens used into drug exposures derived from pharmacokinetic models; we also normalized exposures by the corresponding drug potency. We performed a Bayesian meta-analysis for Grades 3&4 of treatment related (trAE) and immune-mediated (imAE) adverse events. Results: A total of 149 articles were identified, covering 35,559 patients in 197 dosing cohorts treated with ICI therapies. For PD-1 and PD-L1 inhibitor monotherapies, no dose dependence of AEs was found; Grades 3&4 trAE rates for anti PD-L1 vs. anti PD-1 were, respectively, 12% and 15%. The AE rates for the different ICI drugs and organ classes were estimated. Dose dependence was found for anti CTLA-4 monotherapies, for total trAEs of Grades 3/4, gastrointestinal and hepatic imAEs. Dose dependence was found and quantified for anti CTLA-4 in combination with anti PD-1 with respect to total trAEs of Grades 3/4 trAEs and imAEs per gastrointestinal and hepatic organ groups. We found that combination of anti PD-L1 agents with anti CTLA-4 exhibited lower AE rates, as compared to anti PD-1 combined with anti CTLA-4. Conclusions: We introduced a novel meta-analysis methodology and used it to quantify and compare AE rates across ICI agents. Significant AE rate dose dependencies were observed for CTLA-4 inhibitors, either as monotherapy or used in combinations. Patients naive to anti-cancer therapies exhibited higher AE rates vs. previously treated patients. AE rates for CTLA-4 + PD-1 inhibitor combination regimens were supra-additive vs. the respective monotherapies. AE rates for anti PD-L1 agents were lower vs. anti PD-1, both in monotherapy and combinations with CTLA-4.

Comparison of the urinary glucose excretion contributions of SGLT2 and SGLT1: A quantitative systems pharmacology analysis in healthy individuals and patients with type 2 diabetes treated with SGLT2 inhibitors

AIM

To develop a quantitative drug-disease systems model to investigate the paradox that sodium-glucose co-transporter (SGLT)2 is responsible for >80% of proximal tubule glucose reabsorption, yet SGLT2 inhibitor treatment results in only 30% to 50% less reabsorption in patients with type 2 diabetes mellitus (T2DM).

MATERIALS AND METHODS

A physiologically based four-compartment model of renal glucose filtration, reabsorption and excretion via SGLT1 and SGLT2 was developed as a system of ordinary differential equations using R/IQRtools. SGLT2 inhibitor pharmacokinetics and pharmacodynamics were estimated from published concentration-time profiles in plasma and urine and from urinary glucose excretion (UGE) in healthy people and people with T2DM.

RESULTS

The final model showed that higher renal glucose reabsorption in people with T2DM versus healthy people was associated with 54% and 28% greater transporter capacity for SGLT1 and SGLT2, respectively. Additionally, the analysis showed that UGE is highly dependent on mean plasma glucose and estimated glomerular filtration rate (eGFR) and that their consideration is critical for interpreting clinical UGE findings.

CONCLUSIONS

Quantitative drug-disease system modelling revealed mechanistic differences in renal glucose reabsorption and UGE between healthy people and those with T2DM, and clearly showed that SGLT2 inhibition significantly increased glucose available to SGLT1 downstream in the tubule. Importantly, we found that the findings of lower than expected UGE with SGLT2 inhibition are explained by the shift to SGLT1, which recovered additional glucose (~30% of total).

Comparative quantitative systems pharmacology modeling of anti-PCSK9 therapeutic modalities in hypercholesterolemia

Since the discovery of proprotein convertase subtilisin/kexin type 9 (PCSK9) as an attractive target in the treatment of hypercholesterolemia, multiple anti-PCSK9 therapeutic modalities have been pursued in drug development. The objective of this research is to set the stage for the quantitative benchmarking of two anti-PCSK9 pharmacological modality classes, monoclonal antibodies (mAbs) and small interfering RNA (siRNA). To this end, we developed an integrative mathematical model of lipoprotein homeostasis describing the dynamic interplay between PCSK9, LDL-cholesterol (LDL-C), VLDL-cholesterol, HDL-cholesterol (HDL-C), apoB, lipoprotein a [Lp(a)], and triglycerides (TGs). We demonstrate that LDL-C decreased proportionally to PCSK9 reduction for both mAb and siRNA modalities. At marketed doses, however, treatment with mAbs resulted in an additional ∼20% LDL-C reduction compared with siRNA. We further used the model as an evaluation tool and determined that no quantitative differences were observed in HDL-C, Lp(a), TG, or apoB responses, suggesting that the disruption of PCSK9 synthesis would provide no additional effects on lipoprotein-related biomarkers in the patient segment investigated. Predictive model simulations further indicate that siRNA therapies may reach reductions in LDL-C levels comparable to those achieved with mAbs if the current threshold of 80% PCSK9 inhibition via siRNA could be overcome.

Abstract 1082: Linking tumor microenvironment properties in murine syngeneic tumors with resistance to immune checkpoint inhibitors: Insights from a quantitative systems approach

Objectives: Studies in murine syngeneic tumors are critical in the development of immune-based therapies, yet there still are knowledge gaps in the functional meaning (vs. response and resistance to treatments) of baseline molecular and immunological features in these tumors. We developed a quantitative systems model of immuno-oncology (IO), to (i) understand factors within the TME which may underlie anti PD-(L)1 and CTLA-4 efficacy in 6 syngeneic tumors (4T1, LLC, CT-26, MC-38, B16, RENCA); (ii) identify potential baseline factors which relate to treatment resistance.

Methods: Our IO model [1] was used, firstly to incorporate rich datasets from the 6 syngeneic tumor types [2] and to characterize differences in baseline TME conditions. The model was then used to perform mechanistic population simulations of the initiation and development of anti-tumor T cell immune responses, linked to observed individual animal- and cohort-level tumor size dynamics (TSD) under anti PD-(L)1 and CTLA-4 treatments. Variability in individual tumor size dynamics was taken into account using a mixed-effects technique, implemented in the model at the level of tumor-infiltrating T cell influx.

Results: The model adequately described individual- and cohort-level TSD patterns, for all treatment regimens in all 6 tumor types. The model incorporated in one quantitative framework immune cell count data measured in these tumors, by capturing empirical dependencies between TME properties and model parameters. Anti PD-L1 therapy was incorporated into the model via a direct increase in an immune activation rate (IAR) function in TME, validating our previous results [1]. Interestingly, an optimal model incorporating anti CTLA-4 mechanism of action was one considering an indirect effect on IAR through the decrease of immuno-suppressive cell (ISC) function, which supports the hypothesis that the driving force of anti CTLA-4 effects in syngeneic tumors would go through ISC deactivation, e.g., via regulatory T cell (Tregs) depletion. Also, higher counts of Tregs at baseline (e.g., CT26, RENCA) correlated well with responses to anti CTLA-4 treatment. Higher levels of macrophages and/or MDSC infiltration in lesser “immunologically hot” tumors (e.g., 4T1, MC38, LLC) were shown to be the main immuno-suppressive factors limiting tumor responsiveness to checkpoint inhibitor treatments.

Conclusions: This quantitative model may be used as a platform to analyze immune-based treatment data from various tumor types, while providing mechanistic insights on the contributions of baseline TME conditions to response or resistance to treatment. The model may be further used to perform predictive tumor response simulations (monotherapies and combinations), of untested anti CTLA-4, PD-(L)1 dose schedules and of other novel IO agents beyond these two checkpoint inhibitors.

Citation Format: Gabriel Helmlinger, Ivan Azarov, Yuri Kosinsky, Veronika Voronova, Lulu Chu, Suzanne Mosely, Simon Dovedi, Kirill Peskov. Linking tumor microenvironment properties in murine syngeneic tumors with resistance to immune checkpoint inhibitors: Insights from a quantitative systems approach [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1082.

Abstract 104: Mechanistic insights and dose optimization for AZD3458, a novel selective PI3Kg immuno-modulator, using a quantitative systems approach

Objectives: PI3Kγ inhibition re-polarizes macrophages to an immuno-stimulatory phenotype, thereby activating a T-cell mediated tumor immune response. AZD3458 is a highly selective PI3Kγ inhibitor. Administration of AZD3458 in combination with checkpoint inhibitors such as α-PD-(L)1 antibodies had greater anti-tumor effects (TGI 26-86%) than checkpoint inhibitor alone in 4T1, LLC, CT-26 and MC-38 syngeneic mouse models. In these, AZD3458 remodeled the tumor microenvironment (TME), reducing immunosuppressive markers (e.g in 4T1 model there was a 20% decrease in total macrophages and 50% decrease in markers of immune suppression like CD206 by flow cytometry) and promoting cytotoxic T-cell activation (e.g. in CT-26 model there was a 2-fold increase in gzmB mRNA). We developed a predictive quantitative systems pharmacology (QSP) model, to quantitatively simulate TME effects and delineate mechanistic principles underlying AZD3458 and α-PD-(L)1 synergistic effects.

Methods: The QSP model captures mechanistic, molecular and cellular interactions between PI3Kγ inhibition and checkpoint inhibitors, together with the pharmacokinetics acting on the respective targets. Features such as PI3Kγ inhibition-dependent tumor-associated macrophages, protein expression of immunosuppressive markers, reduction of MDSC activation and promotion of cytotoxic T-cell activation were included in the model. These immuno-changes were then linked to tumor cell death, resulting in macroscopic dynamic effects on tumor size. Some model parameters were taken from the literature and internal studies; some were estimated using NLME modeling of tumor size data.

Results: The model adequately described individual and population tumor size patterns. Inter-animal variability was described using a random effect on a parameter related to the ability of T cells to infiltrate the tumor in response to systemic antigen. Additionally, the model incorporated in one quantitative framework data from 4 syngeneic tumors capturing respective changes in TME conditions. Simulations for the various treatments supported the mechanistic interpretation of the observed AZD3458 and α-PD-(L)1 synergistic effects. The model was further used to simulate treatment scenarios, to infer optimal dosing and scheduling for the combination and given underlying TME conditions.

Conclusions: This study provides quantitative mechanistic insights into the links between PI3Kγ inhibition and anti-tumor immune responses, supporting our understanding of how AZD3458 may alleviate brakes in a myeloid immuno-suppressive TME and revert resistance to immunotherapy. This mechanistic understanding is critical when proceeding with dose escalation in an early clinical trial setting, as it allows to contextualize any potential compound-induced immuno-modulation in patients, for given doses and schedules.

Citation Format: Pablo Morentin Gutierrez, Yuri Kosinsky, Kirill Peskov, Ivan Azarov, Lulu Chu, Veronika Voronova, Martin Johnson, Yingxue Chen, Larissa Carnevalli, Danielle Carroll, Michele Moschetta, Teresa Klinowska, Gabriel Helmlinger. Mechanistic insights and dose optimization for AZD3458, a novel selective PI3Kg immuno-modulator, using a quantitative systems approach [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 104.

Role of T Cell-To-Dendritic Cell Chemoattraction in T Cell Priming Initiation in the Lymph Node: An Agent-Based Modeling Study

The adaptive immune response is initiated in lymph nodes by contact between antigen-bearing dendritic cells (DCs) and antigen-specific T cells. A selected number of naïve T cells that recognize a specific antigen may proliferate into expanded clones, differentiate, and acquire an effector phenotype. Despite growing experimental knowledge, certain mechanistic aspects of T cell behavior in lymph nodes remain poorly understood. Computational modeling approaches may help in addressing such gaps. Here we introduce an agent-based model describing T cell movements and their interactions with DCs, leading to activation and expansion of cognate T cell clones, in a two-dimensional representation of the lymph node paracortex. The primary objective was to test the putative role of T cell chemotaxis toward DCs, and quantitatively assess the impact of chemotaxis with respect to T cell priming efficacy. Firstly, we evaluated whether chemotaxis of naïve T cells toward a nearest DC may accelerate the scanning process, by quantifying, through simulations, the number of unique T cell—DC contact events. We demonstrate that, in the presence of naïve T cell-to-DC chemoattraction, a higher total number of contacts occurs, as compared to a T cell random walk scenario. However, the forming swarm of naïve T cells, as these cells get attracted to the neighborhood of a DC, may then physically restrict access of additional T cells to the DC, leading to an actual decrease in the cumulative number of unique contacts between naïve T cells and DCs. Secondly, we investigated the potential role of chemotaxis in maintaining cognate T cell clone expansion. The time course of cognate T cells number in the system was used as a quantitative characteristic of the expansion. Model-based simulations indicate that inclusion of chemotaxis, which is selective for already activated (but not naïve) antigen-specific T cells, may strongly accelerate the time of immune response occurrence, which subsequently increases the overall amplitude of the T cell clone expansion process.

Quantitative Systems Pharmacology: An Exemplar Model-Building Workflow With Applications in Cardiovascular, Metabolic, and Oncology Drug Development

Quantitative systems pharmacology (QSP), a mechanistically oriented form of drug and disease modeling, seeks to address a diverse set of problems in the discovery and development of therapies. These problems bring a considerable amount of variability and uncertainty inherent in the nonclinical and clinical data. Likewise, the available modeling techniques and related software tools are manifold. Appropriately, the development, qualification, application, and impact of QSP models have been similarly varied. In this review, we describe the progressive maturation of a QSP modeling workflow: a necessary step for the efficient, reproducible development and qualification of QSP models, which themselves are highly iterative and evolutive. Furthermore, we describe three applications of QSP to impact drug development; one supporting new indications for an approved antidiabetic clinical asset through mechanistic hypothesis generation, one highlighting efficacy and safety differentiation within the sodium‐glucose cotransporter‐2 inhibitor drug class, and one enabling rational selection of immuno‐oncology drug combinations.

Longitudinal tumor size and NLR as predictive factors of individual survival compared to their baseline values in patients with non-small cell lung cancer treated with durvalumab

e20047

Background: Our ability to accurately predict survival of patients with non-small cell lung cancer (NSCLC) while on treatment is limited. Prognostic markers such as stage and tumor size are well established, while neutrophil-to-lymphocytes ratio (NLR) and other hemogram measurements have recently been studied. Gain in prognostic accuracy of these markers when measured longitudinally has not been established. Methods: There were 679 NSCLC patients (Stage 3 or 4, ECOG PS 0 or 1) from clinical studies of durvalumab 10 mg/kg every two weeks (NCT02087423 and NCT01693562). We developed three models of overall survival (OS) all with ECOG as covariate: a Cox proportional hazards model with baseline tumor sum-of-longest-diameters (SLD) and NLR as covariates (COX); a joint model of OS and longitudinal SLD and baseline NLR (JM SLD); and a joint model of OS and longitudinal SLD and NLR (JM SLD&NLR). We compared prognostic accuracy of these markers measured longitudinally vs. at baseline, using predicted probability of OS at 12 months after start of durvalumab as a prognostic score. We evaluated predictive performance of the models using area under the receiver-operating characteristic curve (ROC AUC) describing trade-off between true and false positives (i.e., survival past 12 months). The AUCs were calculated for patients in the dataset using longitudinal data up to different cut-offs. Results: The AUC for all patients starting durvalumab using baseline ECOG, SLD and NLR was 0.73, while it decreased to 0.64 for patients surviving to 6 months, compared to 0.50 for noninformative models. The AUC using longitudinal information for SLD and NLR was larger the more longitudinal data was used for prediction and was 0.81 using 6 months’ worth of data. Conclusions: Using longitudinal information for SLD and NLR increased individual predictive performance of these markers compared to only baseline information in NSCLC patients. [Table: see text]

Quantitative Mechanistic Modeling in Support of Pharmacological Therapeutics Development in Immuno-Oncology

Following the approval, in recent years, of the first immune checkpoint inhibitor, there has been an explosion in the development of immuno-modulating pharmacological modalities for the treatment of various cancers. From the discovery phase to late-stage clinical testing and regulatory approval, challenges in the development of immuno-oncology (IO) drugs are multi-fold and complex. In the preclinical setting, the multiplicity of potential drug targets around immune checkpoints, the growing list of immuno-modulatory molecular and cellular forces in the tumor microenvironment-with additional opportunities for IO drug targets, the emergence of exploratory biomarkers, and the unleashed potential of modality combinations all have necessitated the development of quantitative, mechanistically-oriented systems models which incorporate key biology and patho-physiology aspects of immuno-oncology and the pharmacokinetics of IO-modulating agents. In the clinical setting, the qualification of surrogate biomarkers predictive of IO treatment efficacy or outcome, and the corresponding optimization of IO trial design have become major challenges. This mini-review focuses on the evolution and state-of-the-art of quantitative systems models describing the tumor vs. immune system interplay, and their merging with quantitative pharmacology models of IO-modulating agents, as companion tools to support the addressing of these challenges.

Multiscale Mathematical Model of Drug-Induced Proximal Tubule Injury: Linking Urinary Biomarkers to Epithelial Cell Injury and Renal Dysfunction

Drug-induced nephrotoxicity is a major cause of acute kidney injury, and thus detecting the potential for nephrotoxicity early in the drug development process is critical. Various urinary biomarkers exhibit different patterns following drug-induced injury, which may provide greater information than traditional biomarkers like serum creatinine. In this study, we developed a multiscale quantitative systems pharmacology model relating drug exposure to proximal tubule (PT) epithelial cell injury and subsequently to expression of multiple urinary biomarkers and organ-level functional changes. We utilized urinary kidney injury molecule-1 (Kim-1), alpha glutathione S-transferase, albumin (αGST), glucose, and urine volume time profiles as well as serum creatinine and histopathology data obtained from rats treated with the nephrotoxicant cisplatin to develop the model. Although the model was developed using single-dose response to cisplatin, the model predicted the serum creatinine response to multidose cisplatin regimens. Further, using only the urinary Kim-1 response to gentamicin (a nephrotoxicant with a distinctly different injury time course than cisplatin), the model detected and predicted mild to moderate PT injury, as confirmed with histopathology, even when serum creatinine was unchanged. Thus, the model is generalizable, and can be used to deconvolute the underlying degree and time course of drug-induced PT injury and renal dysfunction from a small number of urinary biomarkers, and may provide a tool to determine optimal dosing regimens that minimize renal injury.

Exenatide effects on gastric emptying rate and the glucose rate of appearance in plasma: A quantitative assessment using an integrative systems pharmacology model

This study aimed to quantify the effect of the immediate release (IR) of exenatide, a short-acting glucagon-like peptide-1 (GLP-1) receptor agonist (GLP-1RA), on gastric emptying rate (GER) and the glucose rate of appearance (GluRA), and evaluate the influence of drug characteristics and food-related factors on postprandial plasma glucose (PPG) stabilization under GLP-1RA treatment. A quantitative systems pharmacology (QSP) approach was used, and the proposed model was based on data from published sources including: (1) GLP-1 and exenatide plasma concentration-time profiles; (2) GER estimates under placebo, GLP-1 or exenatide IR dosing; and (3) GluRA measurements upon food intake. According to the model's predictions, the recommended twice-daily 5- and 10-μg exenatide IR treatment is associated with GluRA flattening after morning and evening meals (48%-49%), whereas the midday GluRA peak is affected to a lesser degree (5%-30%) due to lower plasma drug concentrations. This effect was dose-dependent and influenced by food carbohydrate content, but not by the lag time between exenatide injection and meal ingestion. Hence, GER inhibition by exenatide IR represents an important additional mechanism of its effect on PPG.

Evaluation of renal and cardiovascular protection mechanisms of SGLT2 inhibitors: model-based analysis of clinical data

The mechanisms of cardiovascular and renal protection observed in clinical trials of sodium-glucose cotransporter 2 (SGLT2) inhibitors (SGLT2i) are incompletely understood and likely multifactorial, including natriuretic, diuretic, and antihypertensive effects, glomerular pressure reduction, and lowering of plasma and interstitial fluid volume. To quantitatively evaluate the contribution of proposed SGLT2i mechanisms of action on changes in renal hemodynamics and volume status, we coupled a mathematical model of renal function and volume homeostasis with clinical data in healthy subjects administered 10 mg of dapagliflozin once daily. The minimum set of mechanisms necessary to reproduce observed clinical responses (urinary sodium and water excretion, serum creatinine and sodium) was determined, and important unobserved physiological variables (glomerular pressure, blood and interstitial fluid volume) were then simulated. We further simulated the response to SGLT2i in diabetic virtual patients with and without renal impairment. Multiple mechanisms were required to explain the observed response: 1) direct inhibition of sodium and glucose reabsorption through SGLT2, 2) SGLT2-driven inhibition of Na⁺/H⁺ exchanger 3 sodium reabsorption, and 3) osmotic diuresis coupled with peripheral sodium storage. The model also showed that the consequences of these mechanisms include lowering of glomerular pressure, reduction of blood and interstitial fluid volume, and mild blood pressure reduction, in agreement with clinical observations. The simulations suggest that these effects are more significant in diabetic patients than healthy subjects and that while glucose excretion may diminish with renal impairment, improvements in glomerular pressure and blood volume are not diminished at lower glomerular filtration rate, suggesting that cardiorenal benefits of SGLT2i may be sustained in renally impaired patients.

Abstract 2098: Quantitative modeling as a systematic approach for drug combination evaluation in immuno-oncology (IO)

Objectives: Multiple strategies for eliciting and enhancing antitumor immunity are currently being evaluated. However, a more systematic approach is needed, to analyze and translate such results into clinic practice, while rationally designing combination therapies based on mechanistic understanding of potential synergistic effects (1). The objective of this study was to provide predictive simulations, via a quantitative systems pharmacology (QSP) model, capable of categorizing the types of synergistic effects that may arise from IO agent combinations, across realistic baseline conditions prevailing in the tumor microenvironment (TME).

Methods: The QSP model was developed and qualified using in vivo mouse data published in the literature and from internal research. The following pharmacologic modalities were calibrated: PD-L1/PD-1, CTLA-4, CXCR2, A2AR inhibition, and OX40 agonism. Various combination scenarios were simulated for these modalities, at four baseline conditions prevailing in different syngeneic murine models.

Results: Simulated efficacy results were highly dependent on the baseline conditions. Several combinations and monotherapies were effective only within a specific baseline TME phenotype. These findings were in agreement with experimental data (2). At baselines with higher levels of MDSC, best results were obtained for a PD-L1 mAb combined with either an OX40 agonist or a CXCR2 inhibitor, with 90% of complete responders. Anti (PD-L1 + CTLA-4) combinations showed high efficacy in Treg prevalence, but only moderate efficacy (22% complete responders), under baseline conditions of a dual (Treg + MDSC) immunosuppressive TME.

Conclusion: This work provides a quantitative modeling framework to comparatively predict responses to IO combinations, based on realistic baseline conditions prevailing in the TME, while revealing mechanistic interactions underlying such responses in IO combinations.

References:

1. Melero I et al. Nat Rev Cancer 2015;15:457-72.

2. Mosely S et al. Cancer Immunol Res 2016;5:29-41.

Citation Format: Gabriel Helmlinger, Yuri Kosinsky, Lulu Chu, Kirill Peskov, Veronika Voronova, Alexandra Borodovsky, Richard Woessner, Kris Sachsenmeier, Nidal Al-huniti. Quantitative modeling as a systematic approach for drug combination evaluation in immuno-oncology (IO) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2098.

Abstract 4760: Joint modeling of longitudinal tumor dynamics and survival in non-small cell lung cancer (NSCLC) patients

Objectives: Tumor size dynamics and survival are traditionally analyzed in a 2-stage approach, without consideration of joint dependencies. The objective of this analysis was to develop a joint model which associates tumor size dynamics and progression-free survival (PFS), to predict time-to-progression. Methods: Phase 2 trial data from selumetinib (AZD6244; ARRY-142886, MEK inhibitor), in NSCLC patients (SELECT-2; NCT01750281) were used to develop a joint model for tumor size dynamics and PFS. The analysis was performed using JM package in R. Treatment arm, KRAS mutation and WHO performance status were evaluated as covariates. Model was evaluated by survival estimation based on early time (e.g., first 3 months) tumor size data. The final joint model based on SELECT-2 data was then used to predict PFS of selumetinib in SELECT-1 (NCT01933932) phase 3 trial. Results: The joint model developed on SELECT-2 data identified a significant association (p value &lt;0.001) between the slope of the longitudinal tumor dynamic and PFS. WHO performance status was identified to be the only significant covariate. Furthermore, the model built on phase 2 data adequately predicted PFS of the SELECT-1 data, using SELECT-1 tumor size data within 4 months of treatment. Prediction confirmed no significant difference in PFS between active treatment arm and chemotherapy arm in SELECT-1 trial. Conclusions: Using selumetinib as an example, we showed that joint modeling of tumor size dynamics and PFS may provide a novel quantitative tool to predict long-term outcome in NSCLC based on early tumor size measurements.

Citation Format: Xiao Tong, James Dunyak, Diansong Zhou, David Carlile, Helen Tomkinson, Gabriel Helmlinger, Nidal Al-Huniti, Hongmei Xu. Joint modeling of longitudinal tumor dynamics and survival in non-small cell lung cancer (NSCLC) patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4760.

Assessing QT/QTc interval prolongation with concentration-QT modeling for Phase I studies: impact of computational platforms, model structures and confidence interval calculation methods

Modeling the relationship between drug concentrations and heart rate corrected QT interval (QTc) change from baseline (C-∆QTc), based on Phase I single ascending dose (SAD) or multiple ascending dose (MAD) studies, has been proposed as an alternative to thorough QT studies (TQT), in assessing drug-induced QT prolongation risk. The present analysis used clinical SAD, MAD and TQT study data of an experimental compound, AZD5672, to evaluate the performance of: (i) three computational platforms (linear mixed-effects modeling implemented via PROC MIXED in SAS, as well as in R using LME4 package and linear quantile mixed models (LQMM) implemented via LQMM package; (ii) different model structures with and without treatment- or time-specific intercepts; and (iii) three methods for calculating the confidence interval (CI) of QTc prolongation (analytical and bootstrap methods with fixed or varied geometric mean concentrations). We show that treatment- and time-specific intercepts may need to be included into C-∆QTc modeling through PROC MIXED or LME4, regardless of their statistical significance. With the intersection union test (IUT) in the TQT study as a reference for comparison, inclusion of these intercepts increased the feasibility for C-∆QTc modelling of SAD or MAD to reach the same conclusion as the IUT analysis based on TQT study. Compared to PROC MIXED or LME4, the LQMM method is less dependent on inclusion of treatment- or time-specific intercepts, and the bootstrap CI calculation methods provided higher likelihood for C-∆QTc modeling of SAD and MAD studies to reach the same conclusion as the IUT based on the TQT study.

Why do SGLT2 inhibitors reduce heart failure hospitalization? A differential volume regulation hypothesis

The effect of a sodium glucose cotransporter 2 inhibitor (SGLT2i) in reducing heart failure hospitalization in the EMPA-REG OUTCOMES trial has raised the possibility of using these agents to treat established heart failure. We hypothesize that osmotic diuresis induced by SGLT2 inhibition, a distinctly different diuretic mechanism than that of other diuretic classes, results in greater electrolyte-free water clearance and, ultimately, in greater fluid clearance from the interstitial fluid (IF) space than from the circulation, potentially resulting in congestion relief with minimal impact on blood volume, arterial filling and organ perfusion. We utilize a mathematical model to illustrate that electrolyte-free water clearance results in a greater reduction in IF volume compared to blood volume, and that this difference may be mediated by peripheral sequestration of osmotically inactive sodium. By coupling the model with data on plasma and urinary sodium and water in healthy subjects who received either the SGLT2i dapagliflozin or loop diuretic bumetanide, we predict that dapagliflozin produces a 2-fold greater reduction in IF volume compared to blood volume, while the reduction in IF volume with bumetanide is only 78% of the reduction in blood volume. Heart failure is characterized by excess fluid accumulation, in both the vascular compartment and interstitial space, yet many heart failure patients have arterial underfilling because of low cardiac output, which may be aggravated by conventional diuretic treatment. Thus, we hypothesize that, by reducing IF volume to a greater extent than blood volume, SGLT2 inhibitors might provide better control of congestion without reducing arterial filling and perfusion.

Radiation and PD-(L)1 treatment combinations: immune response and dose optimization via a predictive systems model

BACKGROUND

Numerous oncology combination therapies involving modulators of the cancer immune cycle are being developed, yet quantitative simulation models predictive of outcome are lacking. We here present a model-based analysis of tumor size dynamics and immune markers, which integrates experimental data from multiple studies and provides a validated simulation framework predictive of biomarkers and anti-tumor response rates, for untested dosing sequences and schedules of combined radiation (RT) and anti PD-(L)1 therapies.

METHODS

A quantitative systems pharmacology model, which includes key elements of the cancer immunity cycle and the tumor microenvironment, tumor growth, as well as dose-exposure-target modulation features, was developed to reproduce experimental data of CT26 tumor size dynamics upon administration of RT and/or a pharmacological IO treatment such as an anti-PD-L1 agent. Variability in individual tumor size dynamics was taken into account using a mixed-effects model at the level of tumor-infiltrating T cell influx.

RESULTS

The model allowed for a detailed quantitative understanding of the synergistic kinetic effects underlying immune cell interactions as linked to tumor size modulation, under these treatments. The model showed that the ability of T cells to infiltrate tumor tissue is a primary determinant of variability in individual tumor size dynamics and tumor response. The model was further used as an in silico evaluation tool to quantitatively predict, prospectively, untested treatment combination schedules and sequences. We demonstrate that anti-PD-L1 administration prior to, or concurrently with RT reveal further synergistic effects, which, according to the model, may materialize due to more favorable dynamics between RT-induced immuno-modulation and reduced immuno-suppression of T cells through anti-PD-L1.

CONCLUSIONS

This study provides quantitative mechanistic explanations of the links between RT and anti-tumor immune responses, and describes how optimized combinations and schedules of immunomodulation and radiation may tip the immune balance in favor of the host, sufficiently to lead to tumor shrinkage or rejection.

Integrating dose estimation into a decision-making framework for model-based drug development

Model-informed drug discovery and development offers the promise of more efficient clinical development, with increased productivity and reduced cost through scientific decision making and risk management. Go/no-go development decisions in the pharmaceutical industry are often driven by effect size estimates, with the goal of meeting commercially generated target profiles. Sufficient efficacy is critical for eventual success, but the decision to advance development phase is also dependent on adequate knowledge of appropriate dose and dose-response. Doses which are too high or low pose risk of clinical or commercial failure. This paper addresses this issue and continues the evolution of formal decision frameworks in drug development. Here, we consider the integration of both efficacy and dose-response estimation accuracy into the go/no-go decision process, using a model-based approach. Using prespecified target and lower reference values associated with both efficacy and dose accuracy, we build a decision framework to more completely characterize development risk. Given the limited knowledge of dose response in early development, our approach incorporates a set of dose-response models and uses model averaging. The approach and its operating characteristics are illustrated through simulation. Finally, we demonstrate the decision approach on a post hoc analysis of the phase 2 data for naloxegol (a drug approved for opioid-induced constipation).

Pharmacometric Modeling of Naloxegol Efficacy and Safety: Impact on Dose and Label

Naloxegol is a peripherally acting μ-opioid receptor antagonist that was developed for the treatment of opioid-induced constipation. Modeling and simulation of naloxegol efficacy and tolerability informed selection of doses for phase III studies and provided comprehensive dosage recommendations for the naloxegol US package insert.

Survival prediction using time-evolving tumor load: An approach to rationally design treatment sequencing, staging, and dosing strategies for oncology combinations

e20040

Background: Longitudinal tumor burden has long been used for the clinical diagnosis, staging, prognosis and treatment of non-small cell lung cancer (NSCLC). Tumor burden and growth rate correlate directly with survival and RECIST (measurement of tumor) is widely used as a surrogate endpoint. RECIST doesn’t take into account the impact of the therapy on the speed of growth over time, and poorly correlates with survival, especially in IO therapies. Surrogate endpoints such as progression-free survival (PFS) or dichotomous RECIST-based overall response rate (ORR) allow early assessment of efficacy based on reduced study population size. Unfortunately, ORR and PFS are poorly predictive of OS in NSCLC. This well-known loss of information during categorization emphasizes the potential benefits of continuous tumor load modeling. Methods: This research provides a statistically valid basis for modeling and interpretation of longitudinal response dynamics, in the context of time-to-event (survival) censoring, through development of a joint longitudinal/event model. Using clinical data from the Phase 3 IPASS study of Iressa (gefitinib), we build a generalizable model relating tumor load dynamics to a time-evolving hazard of progression. Results: Using longitudinal data from other smaller and shorter studies, we identify the time-progression of patient’s risk, including post-treatment time of minimum hazard of progression and the time when hazard rapidly increases. This is applied at both the population strata and individual patient level to assess different treatment paradigms. The model shows that EGFR-mutated patients on gefitinib have stable hazard of progression for about 8 months, as which point hazard increases rapidly. EGFR-mutated patients treated with carboplatin/paclitaxel started with a 72% higher initial progression hazard which then increased rapidly after 4 months. Conclusions: Because of their continuous-in-time nature, joint longitudinal/event models provide a basis for exploring the complex timing issues in oncology treatment sequencing, treatment staging strategies for combinations, and dosing regimes.

Dynamic predictions of patient survival using longitudinal tumor size in non-small cell lung cancer: Approach towards personalized medicine

e20606

Background: Tumor burden has long been used for the clinical diagnosis, staging, prognosis and treatment of non-small cell lung cancer (NSCLC), as described, for example, in the 7th edition of the AJCC/UICC NSCLC staging guidelines. Previous longitudinal tumor size approaches have used fixed tumor kinetic parameters or tumor shrinkage at a given timepoint, to correlate PFS and OS in a stepwise fashion. Here we describe a joint modeling approach which allows for individual, patient-level predictions of survival during NSCLC treatment. Joint modeling simultaneously fits OS and tumor size dynamics, converting full information from individual tumor assessments into a personalized prediction of survival - thereby avoiding dichotomization of response measure in a patient. Methods: Clinical data from IPASS Phase 3 study of Iressa (gefitinib) in NSCLC were used to fit a joint model of OS and tumor size. The data from a follow-up study (IFUM, Phase 4) for the same drug in a narrower population were used to validate the model on an independent set of subjects. This part included simulating clinical trials from the model and comparing the simulated survival with the observed data. The survival estimation method for individual patients followed from a Bayesian formulation and was implemented in R packages JM and JMbayes. Results: A joint model for overall survival and tumor size was developed and validated using clinical trial simulations. Individual survival estimates were obtained for subjects in a subsequent study based on early data cut-off for tumor assessments. Patient-level predictions were shown to be accurate as well as study-level survival estimates. The model was able to update individual survival predictions in real time. Conclusions: Joint tumor size / survival modeling provides a promising area of investigation for prediction of survival in individual patients. It can be used as a quantitative tool for estimating time-evolving risk of death based on early tumor size measurements. A clinically validated version of such a tool may allow physicians to better choose between treatment continuation and change, following tumor size measurements from standard clinical care.

Drug-disease modeling in the pharmaceutical industry - where mechanistic systems pharmacology and statistical pharmacometrics meet

Modeling & simulation (M&S) methodologies are established quantitative tools, which have proven to be useful in supporting the research, development (R&D), regulatory approval, and marketing of novel therapeutics. Applications of M&S help design efficient studies and interpret their results in context of all available data and knowledge to enable effective decision-making during the R&D process. In this mini-review, we focus on two sets of modeling approaches: population-based models, which are well-established within the pharmaceutical industry today, and fall under the discipline of clinical pharmacometrics (PMX); and systems dynamics models, which encompass a range of models of (patho-)physiology amenable to pharmacological intervention, of signaling pathways in biology, and of substance distribution in the body (today known as physiologically-based pharmacokinetic models) - which today may be collectively referred to as quantitative systems pharmacology models (QSP). We next describe the convergence - or rather selected integration - of PMX and QSP approaches into 'middle-out' drug-disease models, which retain selected mechanistic aspects, while remaining parsimonious, fit-for-purpose, and able to address variability and the testing of covariates. We further propose development opportunities for drug-disease systems models, to increase their utility and applicability throughout the preclinical and clinical spectrum of pharmaceutical R&D.

Model-based meta-analysis of safety for immune checkpoint inhibitor combinations and monotherapy

89

Background: Immune checkpoint inhibitors (ICIs) have shown efficacy across multiple cancer types; however, ICIs are also associated with immune-mediated (im) adverse events (AEs) especially when used in combination. Methods: Published safety data from 35 melanoma and non-small cell lung cancer clinical trials was normalized for drug exposure using drug concentrations at 50% inhibition (IC50). Pharmacokinetic models from FDA, EMEA, and our own model were used to compare AEs across 4 ICIs: anti-PD-1 (nivolumab, pembrolizumab) and anti-CTLA-4 (ipilumumab, tremelimumab). Data was analyzed by ICI target and organ class (ie, GI, skin, liver, lung, and endocrine). To test dependence of AE rate (%) on drug dose dependence (DD) we calculated mean AE rates for low and high halves of the dose range, and used Pearson’s χ2-squared test for statistical significance (SS) ( p< 0.05). Results: We found AEs to be DD for anti-CTLA-4 monotherapies and anti-CTLA-4 + anti-PD-1 combinations, whereas DD was not SS for anti-PD-1 monotherapies (Table). For anti-CTLA-4 drugs DD was SS for GI, hepatic, skin, and endocrine AEs. In combination, there was DD upon anti-CTLA-4, but data were insufficient to confirm SS. For PD-1 drugs there was no significant DD found. Conclusions: For anti-PD-1 therapy, no DD for Grade 3/4 AEs was found across organ classes. However, for anti-CTLA-4 therapies, DD was SS for GI, skin, hepatic, and endocrine im AEs. Although not all data were sufficient to confirm SS, dose intensity of anti-CTLA-4 may be an important determinant of AEs, when co-administered with anti-PD-1. [Table: see text]

Interpretation of metabolic memory phenomenon using a physiological systems model: What drives oxidative stress following glucose normalization?

Hyperglycemia is generally associated with oxidative stress, which plays a key role in diabetes-related complications. A complex, quantitative relationship has been established between glucose levels and oxidative stress, both in vitro and in vivo. For example, oxidative stress is known to persist after glucose normalization, a phenomenon described as metabolic memory. Also, uncontrolled glucose levels appear to be more detrimental to patients with diabetes (non-constant glucose levels) vs. patients with high, constant glucose levels. The objective of the current study was to delineate the mechanisms underlying such behaviors, using a mechanistic physiological systems modeling approach that captures and integrates essential underlying pathophysiological processes. The proposed model was based on a system of ordinary differential equations. It describes the interplay between reactive oxygen species production potential (ROS), ROS-induced cell alterations, and subsequent adaptation mechanisms. Model parameters were calibrated using different sources of experimental information, including ROS production in cell cultures exposed to various concentration profiles of constant and oscillating glucose levels. The model adequately reproduced the ROS excess generation after glucose normalization. Such behavior appeared to be driven by positive feedback regulations between ROS and ROS-induced cell alterations. The further oxidative stress-related detrimental effect as induced by unstable glucose levels can be explained by inability of cells to adapt to dynamic environment. Cell adaptation to instable high glucose declines during glucose normalization phases, and further glucose increase promotes similar or higher oxidative stress. In contrast, gradual ROS production potential decrease, driven by adaptation, is observed in cells exposed to constant high glucose.

Primary proximal tubule hyperreabsorption and impaired tubular transport counterregulation determine glomerular hyperfiltration in diabetes: a modeling analysis

Glomerular hypertension and hyperfiltration in early diabetes are associated with development and progression of diabetic kidney disease. The tubular hypothesis of diabetic hyperfiltration proposes that it is initiated by a primary increase in sodium (Na) reabsorption in the proximal tubule (PT) and the resulting tubuloglomerular feedback (TGF) response and lowering of Bowman space pressure (P_Bow). Here we utilized a mathematical model of the human kidney to investigate over acute and chronic timescales the mechanisms responsible for the magnitude of the hyperfiltration response. The model implicates that the primary hyperreabsorption of Na in the PT produces a Na imbalance that is only partially restored by the hyperfiltration induced by TGF and changes in P_Bow Thus secondary adaptations are needed to restore Na balance. This may include neurohumoral transport regulation and/or pressure-natriuresis (i.e., the decrease in Na reabsorption in response to increased renal perfusion pressure). We explored the role of each tubular segment in contributing to this compensation and the consequences of impairment in tubular compensation. The simulations indicate that impaired secondary downregulation of transport potentiated the rise in glomerular hypertension and hyperfiltration needed to restore Na balance at a given level of primary PT hyperreabsorption. Therefore, we propose for the first time that both the extent of primary PT hyperreabsorption and the degree of impairment of the distal tubular responsiveness to regulatory signals determine the level of glomerular hypertension and hyperfiltration in the diabetic kidney, thereby extending the tubule-centric concept of diabetic hyperfiltration and potential therapeutic approaches beyond the proximal tubule.

Implications of Dynamic Occupancy, Binding Kinetics, and Channel Gating Kinetics for hERG Blocker Safety Assessment and Mitigation

Blockade of the hERG potassium channel prolongs the ventricular action potential (AP) and QT interval, and triggers early after depolarizations (EADs) and torsade de pointes (TdP) arrhythmia. Opinions differ as to the causal relationship between hERG blockade and TdP, the relative weighting of other contributing factors, definitive metrics of preclinical proarrhythmicity, and the true safety margin in humans. Here, we have used in silico techniques to characterize the effects of channel gating and binding kinetics on hERG occupancy, and of blockade on the human ventricular AP. Gating effects differ for compounds that are sterically compatible with closed channels (becoming trapped in deactivated channels) versus those that are incompatible with the closed/closing state, and expelled during deactivation. Occupancies of trappable blockers build to equilibrium levels, whereas those of non-trappable blockers build and decay during each AP cycle. Occupancies of ~83% (non-trappable) versus ~63% (trappable) of open/inactive channels caused EADs in our AP simulations. Overall, we conclude that hERG occupancy at therapeutic exposure levels may be tolerated for nontrappable, but not trappable blockers capable of building to the proarrhythmic occupancy level. Furthermore, the widely used Redfern safety index may be biased toward trappable blockers, overestimating the exposure-IC50 separation in nontrappable cases.

Requirements for multi-level systems pharmacology models to reach end-usage: the case of type 2 diabetes

We are currently in the middle of a major shift in biomedical research: unprecedented and rapidly growing amounts of data may be obtained today, from in vitro, in vivo and clinical studies, at molecular, physiological and clinical levels. To make use of these large-scale, multi-level datasets, corresponding multi-level mathematical models are needed, i.e. models that simultaneously capture multiple layers of the biological, physiological and disease-level organization (also referred to as quantitative systems pharmacology-QSP-models). However, today's multi-level models are not yet embedded in end-usage applications, neither in drug research and development nor in the clinic. Given the expectations and claims made historically, this seemingly slow adoption may seem surprising. Therefore, we herein consider a specific example-type 2 diabetes-and critically review the current status and identify key remaining steps for these models to become mainstream in the future. This overview reveals how, today, we may use models to ask scientific questions concerning, e.g., the cellular origin of insulin resistance, and how this translates to the whole-body level and short-term meal responses. However, before these multi-level models can become truly useful, they need to be linked with the capabilities of other important existing models, in order to make them 'personalized' (e.g. specific to certain patient phenotypes) and capable of describing long-term disease progression. To be useful in drug development, it is also critical that the developed models and their underlying data and assumptions are easily accessible. For clinical end-usage, in addition, model links to decision-support systems combined with the engagement of other disciplines are needed to create user-friendly and cost-efficient software packages.

A mechanistic tumor penetration model to guide antibody drug conjugate design

Antibody drug conjugates (ADCs) represent novel anti-cancer modalities engineered to specifically target and kill tumor cells expressing corresponding antigens. Due to their large size and their complex kinetics, these therapeutic agents often face heterogeneous distributions in tumors, leading to large untargeted regions that escape therapy. We present a modeling framework which includes the systemic distribution, vascular permeability, interstitial transport, as well as binding and payload release kinetics of ADC-therapeutic agents in mouse xenografts. We focused, in particular, on receptor dynamics such as endocytic trafficking mechanisms within cancer cells, to simulate their impact on tumor mass shrinkage upon ADC administration. Our model identified undesirable tumor properties that can impair ADC tissue homogeneity, further compromising ADC success, and explored ADC design optimization scenarios to counteract upon such unfavorable intrinsic tumor tissue attributes. We further demonstrated the profound impact of cytotoxic payload release mechanisms and the role of bystander killing effects on tumor shrinkage. This model platform affords a customizable simulation environment which can aid with experimental data interpretation and the design of ADC therapeutic treatments.

Inference of imatinib (IM) effects on leukemic stem cell (SC) compartment via mathematical modeling of IRIS treatment response data

7056

Background: Continuous treatment of chronic phase CML (CML-CP) patients with imatinib (IM) induces durable responses in majority of patients with a decreasing rate of relapse (Hochhaus et al, Blood. 2007;110). We present a mechanistic mathematical model which uses 6-year follow up data from the International Randomized Study of Interferon Versus STI571 (IRIS) trial (O'Brien et al, N Engl J Med. 2003;348:994) to explore IM effects on leukemic stem cells (SCs) across the patient population. Methods: The model approximates hemopoiesis as a 4-stage process in which only the first stage, corresponding to the SC compartment, is capable of self-renewal. Leukemic SCs, early and late progenitors were assumed to have higher self-renewal and expansion/differentiation rates than their normal counterparts. Model parameters describing the patient population distributions of leukemic/total SC ratio at diagnosis and sensitivity of leukemic cells to IM, were then fitted to individual cytogenetic and molecular response (CR/MR) data from 200 randomly selected newly-diagnosed CML-CP patients with 6-year follow-up. Results: Based on our analysis, successful characterization of the wide range of clinically observed treatment responses requires the inhibition of the leukemic SC compartment by IM. The median predicted inhibition of the leukemic SC proliferation rate was 79%. The model further predicted that after 6 years of IM treatment, 45% of patients would achieve a leukemic/total SC ratio below 0.1. Conclusions: We have developed a mathematical model of IM effects on CML-CP based on 6-year CR and MR data from the IRIS trial. In contrast to prior reports (Michor et al, Nature. 2005;435), our modeling predicts that IM reduces the leukemic SC compartment in most CML-CP patients, which could provide a mechanistic rationale for the decreasing rate of relapse observed in the study population.

[Table: see text]

Preclinical cardiac safety assessment of pharmaceutical compounds using an integrated systems-based computer model of the heart

Blockade of the delayed rectifier potassium channel current, I(Kr), has been associated with drug-induced QT prolongation in the electrocardiogram and life-threatening cardiac arrhythmias. However, it is increasingly clear that compound-induced interactions with multiple cardiac ion channels may significantly affect QT prolongation that would result from inhibition of only I(Kr) [Redfern, W.S., Carlsson, L., et al., 2003. Relationships between preclinical cardiac electrophysiology, clinical QT interval prolongation and torsade de pointes for a broad range of drugs: evidence for a provisional safety margin in drug development. Cardiovasc. Res. 58(1), 32-45]. Such an assessment may not be feasible in vitro, due to multi-factorial processes that are also time-dependent and highly non-linear. Limited preclinical data, I(Kr) hERG assay and canine Purkinje fiber (PF) action potentials (APs) [Gintant, G.A., Limberis, J.T., McDermott, J.S., Wegner, C.D., Cox, B.F., 2001. The canine Purkinje fiber: an in vitro model system for acquired long QT syndrome and drug-induced arrhythmogenesis. J. Cardiovasc. Pharmacol. 37(5), 607-618], were used for two test compounds in a systems-based modeling platform of cardiac electrophysiology [Muzikant, A.L., Penland, R.C., 2002. Models for profiling the potential QT prolongation risk of drugs. Curr. Opin. Drug. Discov. Dev. 5(1), 127-35] to: (i) convert a canine myocyte model to a PF model by training functional current parameters to the AP data; (ii) reverse engineer the compounds' effects on five channel currents other than I(Kr), predicting significant IC(50) values for I(Na+), sustained and I(Ca2+), L-type , which were subsequently experimentally validated; (iii) use the predicted (I(Na+), sustained and I(Ca2+), L-type) and measured (I(Kr)) IC(50) values to simulate dose-dependent effects of the compounds on APs in endocardial, mid-myocardial, and epicardiac ventricular cells; and (iv) integrate the three types of cellular responses into a tissue-level spatial model, which quantifiably predicted no potential for the test compounds to induce either QT prolongation or increased transmural dispersion of repolarization in a dose-dependent and reverse rate-dependent fashion, despite their inhibition of I(Kr) in vitro.

An integrated approach for inference and mechanistic modeling for advancing drug development

An important challenge facing researchers in drug development is how to translate multi-omic measurements into biological insights that will help advance drugs through the clinic. Computational biology strategies are a promising approach for systematically capturing the effect of a given drug on complex molecular networks and on human physiology. This article discusses a two-pronged strategy for inferring biological interactions from large-scale multi-omic measurements and accounting for known biology via mechanistic dynamical simulations of pathways, cells, and organ- and tissue level models. These approaches are already playing a role in driving drug development by providing a rational and systematic computational framework.