Mathematical Modeling of Tumor Growth in Preclinical Mouse Models with Applications in Biomarker Discovery and Drug Mechanism Studies.

Oncology drug efficacy is evaluated in mouse models by continuously monitoring tumor volumes, which can be mathematically described by growth kinetic models. Although past studies have investigated various growth models, their reliance on small datasets raises concerns about whether their findings are truly representative of tumor growth in diverse mouse models under different vehicle or drug treatments. In this study, we systematically evaluated six parametric models (exponential, exponential quadratic, monomolecular, logistic, Gompertz, and von Bertalanffy) and the semiparametric generalized additive model (GAM) on fitting tumor volume data from more than 30,000 mice in 930 experiments conducted in patient-derived xenografts, cell line-derived xenografts, and syngeneic models. We found that the exponential quadratic model is the best parametric model and can adequately model 87% studies, higher than other models including von Bertalanffy (82%) and Gompertz (80%) models; the latter is often considered the standard growth model. At the mouse group level, 7.5% of growth data could not be fit by any parametric model and were fitted by GAM. We show that endpoint gain integrated in time, a GAM-derived efficacy metric, is equivalent to exponential growth rate, a metric we previously proposed and conveniently calculated by simple algebra. Using five studies on paclitaxel, anti-PD1 antibody, cetuximab, irinotecan, and sorafenib, we showed that exponential and exponential quadratic models achieve similar performance in uncovering drug mechanism and biomarkers. We also compared exponential growth rate-based association analysis and exponential modeling approach in biomarker discovery and found that they complement each other. Modeling methods herein are implemented in an open-source R package freely available at https://github.com/hjzhou988/TuGroMix. SIGNIFICANCE: We present a general strategy for mathematically modeling tumor growth in mouse models using data from 30,000 mice and show that modeling and nonmodeling approaches are complementary in biomarker discovery and drug mechanism studies.

An NGS-based assay for accurate detection and quantification of immune gene expression in mouse tumor models.

Tumor microenvironment (TME) is a complex dynamic system with many tumor-interacting components including tumor-infiltrating leukocytes (TILs), cancer associated fibroblasts, blood vessels, and other stromal constituents. It intrinsically affects tumor development and pharmacology of oncology therapeutics, particularly immune-oncology (IO) treatments. Accurate measurement of TME is therefore of great importance for understanding the tumor immunity, identifying IO treatment mechanisms, developing predictive biomarkers, and ultimately, improving the treatment of cancer. Here, we introduce a mouse-IO NGS-based (NGSmIO) assay for accurately detecting and quantifying the mRNA expression of 1080 TME related genes in mouse tumor models. The NGSmIO panel was shown to be superior to the commonly used microarray approach by hosting 300 more relevant genes to better characterize various lineage of immune cells, exhibits improved mRNA and protein expression correlation to flow cytometry, shows stronger correlation with mRNA expression than RNAseq with 10x higher sequencing depth, and demonstrates higher sensitivity in measuring low-expressed genes. We describe two studies; firstly, detecting the pharmacodynamic change of interferon-γ expression levels upon anti-PD-1: anti-CD4 combination treatment in MC38 and Hepa 1-6 tumors; and secondly, benchmarking baseline TILs in 14 syngeneic tumors using transcript level expression of lineage specific genes, which demonstrate effective and robust applications of the NGSmIO panel.

Statistical Assessment of Drug Synergy from In Vivo Combination Studies Using Mouse Tumor Models.

Drug combination therapy is a promising strategy for treating cancer; however, its efficacy and synergy require rigorous evaluation in preclinical studies before going to clinical trials. Existing methods have limited power to detect synergy in animal studies. Here, we introduce a novel approach to assess in vivo drug synergy with high sensitivity and low false discovery rate. It can accurately estimate combination index and synergy score under the Bliss independence model and the highest single agent (HSA) model without any assumption on tumor growth kinetics, study duration, data completeness and balance for tumor volume measurement. We show that our method can effectively validate in vitro drug synergy discovered from cell line assays in in vivo xenograft experiments, and can help to elucidate the mechanism of action for immune checkpoint inhibitors in syngeneic mouse models by combining an anti-PD-1 antibody and several tumor-infiltrating leukocytes depletion treatments. We provide a unified view of in vitro and in vivo synergy by presenting a parallelism between the fixed-dose in vitro and the 4-group in vivo combination studies, so they can be better designed, analyzed, and compared. We emphasize that combination index, when defined here via relative survival of tumor cells, is both dose and time dependent, and give guidelines on designing informative in vivo combination studies. We explain how to interpret and apply Bliss and HSA synergies. Finally, we provide an open-source software package named invivoSyn that enables automated analysis of in vivo synergy using our method and several other existing methods. SIGNIFICANCE: This work presents a general solution to reliably determine in vivo drug synergy in single-dose 4-group animal combination studies.

HX009, a novel BsAb dual targeting PD1 x CD47, demonstrates potent anti-lymphoma activity in preclinical models.

Both PD1/PD-L1 and CD47 blockades have demonstrated limited activity in most subtypes of NHL save NK/T-cell lymphoma. The hemotoxicity with anti-CD47 agents in the clinic has been speculated to account for their limitations. Herein we describe a first-in-class and rationally designed bispecific antibody (BsAb), HX009, targeting PD1 and CD47 but with weakened CD47 binding, which selectively hones the BsAb for tumor microenvironment through PD1 interaction, potentially reducing toxicity. In vitro characterization confirmed: (1) Both receptor binding/ligand blockade, with lowered CD47 affinity; (2) functional PD1/CD47 blockades by reporter assays; (3) T-cell activation in Staphylococcal-enterotoxin-B-pretreated PBMC and mixed-lymphocyte-reaction. In vivo modeling demonstrated antitumor activity in Raji-B and Karpass-229-T xenograft lymphomas. In the humanized mouse syngeneic A20 B-lymphoma (huCD47-A20) HuGEMM model, which has quadruple knocked-in hPD1xhPD-L1xhCD47xhSIRPα genes and an intact autologous immune-system, a contribution of effect is demonstrated for each targeted biologic (HX008 targeting PD1 and SIRPα-Fc targeting CD47), which is clearly augmented by the dual targeting with HX009. Lastly, the expression of the immune-checkpoints PD-L1/L2 and CD47 seemed co-regulated among a panel of lymphoma-derived-xenografts, where HX009 maybe more effective in those with upregulated CD47. Our data warrants HX009's further clinical development for treating NHLs.

Mouse Stromal Cells Confound Proteomic Characterization and Quantification of Xenograft Models.

Xenografts are essential models for studying cancer biology and developing oncology drugs, and are more informative with omics data. Most reported xenograft proteomics projects directly profiled tumors comprising human cancer cells and mouse stromal cells, followed by computational algorithms for assigning peptides to human and mouse proteins. We evaluated the performance of three main algorithms by carrying out benchmark studies on a series of human and mouse cell line mixtures and a set of liver patient-derived xenograft (PDX) models. Our study showed that approximately half of the characterized peptides are common between human and mouse proteins, and their allocations to human or mouse proteins cannot be satisfactorily achieved by any algorithm. As a result, many human proteins are erroneously labeled as differentially expressed proteins (DEP) between samples from the same human cell line mixed with different percentages of mouse cells, and the number of such false DEPs increases superquadratically with the mouse cell percentage. When mouse stromal cells are not removed from PDX tumors, about 30%-40% of DEPs from pairwise comparisons of PDX models are false positives, and about 20% of real DEPs cannot be identified irrespective of the threshold for calling differential expression. In conclusion, our study demonstrated that it is advisable to separate human and mouse cells in xenograft tumors before proteomic profiling to obtain more accurate measurement of species-specific protein expression. Significance: This study advocates the separate-then-run over the run-then-separate approach as a better strategy for more reliable proteomic profiling of xenografts.

A living biobank of matched pairs of patient-derived xenografts and organoids for cancer pharmacology.

Patient-derived tumor xenograft (PDX)/organoid (PDO), driven by cancer stem cells (CSC), are considered the most predictive models for translational oncology. Large PDX collections reflective of patient populations have been created and used extensively to test various investigational therapies, including population-trials as surrogate subjects in vivo. PDOs are recognized as in vitro surrogates for patients amenable for high-throughput screening (HTS). We have built a biobank of carcinoma PDX-derived organoids (PDXOs) by converting an existing PDX library and confirmed high degree of similarities between PDXOs and parental PDXs in genomics, histopathology and pharmacology, suggesting "biological equivalence or interchangeability" between the two. Here we demonstrate the applications of PDXO biobank for HTS "matrix" screening for both lead compounds and indications, immune cell co-cultures for immune-therapies and engineering enables in vitro/in vivo imaging. This large biobank of >550 matched pairs of PDXs/PDXOs across different cancers could become powerful tools for the future cancer drug discovery.

Distinct genomic landscape of Chinese pediatric acute myeloid leukemia impacts clinical risk classification.

Studies have revealed key genomic aberrations in pediatric acute myeloid leukemia (AML) based on Western populations. It is unknown to what extent the current genomic findings represent populations with different ethnic backgrounds. Here we present the genomic landscape of driver alterations of Chinese pediatric AML and discover previously undescribed genomic aberrations, including the XPO1-TNRC18 fusion. Comprehensively comparing between the Chinese and Western AML cohorts reveal a substantially distinct genomic alteration profile. For example, Chinese AML patients more commonly exhibit mutations in KIT and CSF3R, and less frequently mutated of genes in the RAS signaling pathway. These differences in mutation frequencies lead to the detection of previously uncharacterized co-occurring mutation pairs. Importantly, the distinct driver profile is clinical relevant. We propose a refined prognosis risk classification model which better reflected the adverse event risk for Chinese AML patients. These results emphasize the importance of genetic background in precision medicine.

Different syngeneic tumors show distinctive intrinsic tumor-immunity and mechanisms of actions (MOA) of anti-PD-1 treatment.

Cancers are immunologically heterogeneous. A range of immunotherapies target abnormal tumor immunity via different mechanisms of actions (MOAs), particularly various tumor-infiltrate leukocytes (TILs). We modeled loss of function (LOF) in four common anti-PD-1 antibody-responsive syngeneic tumors, MC38, Hepa1-6, CT-26 and EMT-6, by systematical depleting a series of TIL lineages to explore the mechanisms of tumor immunity and treatment. CD8+-T-cells, CD4+-T-cells, Treg, NK cells and macrophages were individually depleted through either direct administration of anti-marker antibodies/reagents or using DTR (diphtheria toxin receptor) knock-in mice, for some syngeneic tumors, where specific subsets were depleted following diphtheria toxin (DT) administration. These LOF experiments revealed distinctive intrinsic tumor immunity and thus different MOAs in their responses to anti-PD-1 antibody among different syngeneic tumors. Specifically, the intrinsic tumor immunity and the associated anti-PD-1 MOA were predominately driven by CD8+ cytotoxic TILs (CTL) in all syngeneic tumors, excluding Hepa1-6 where CD4+ Teff TILs played a key role. TIL-Treg also played a critical role in supporting tumor growth in all four syngeneic models as well as M2-macrophages. Pathway analysis using pharmacodynamic readouts of immuno-genomics and proteomics on MC38 and Hepa1-6 also revealed defined, but distinctive, immune pathways of activation and suppression between the two, closely associated with the efficacy and consistent with TIL-pharmacodynamic readouts. Understanding tumor immune-pathogenesis and treatment MOAs in the different syngeneic animal models, not only assists the selection of the right model for evaluating new immunotherapy of a given MOA, but also can potentially help to understand the potential disease mechanisms and strategize optimal immune-therapies in patients.

Adenosine-related small molecules show utility of recall antigen assay to screen compounds for off-target effects on memory T cells.

Extracellular adenosine suppresses T cell immunity in the tumor microenvironment and in vitro treatment of memory T cells with adenosine can suppress antigen-mediated memory T cell expansion. We describe utilizing the recall antigen assay platform to screen small molecule drug off-target effects on memory T cell expansion/function using a dosing regimen based on adenosine treatment. As a proof of principle, we show low dose GS-5734, a monophosphoramidate prodrug of an adenosine analog, does not alter memory T cell recall at lower doses whereas toxicity observed at high dose favors antigen-specific memory T cell survival/proliferation over non-specific CD8+ T cells. Conversely, parent nucleoside GS-441524 at high dosage does not result in cellular toxicity and reduces antigen-specific T cell recall in most donors. Despite similar chemical structure, these drugs displayed opposing effects on memory T cell expansion and viability highlighting the sensitivity of this assay setup in screening compounds for off-target effects.

Creating Matched In vivo/In vitro Patient-Derived Model Pairs of PDX and PDX-Derived Organoids for Cancer Pharmacology Research.

Patient-derived tumor xenografts (PDXs) are considered the most predictive preclinical models, largely believed to be driven by cancer stem cells (CSC) for conventional cancer drug evaluation. A large library of PDXs is reflective of the diversity of patient populations and thus enables population based preclinical trials ("Phase II-like mouse clinical trials"); however, PDX have practical limitations of low throughput, high costs and long duration. Tumor organoids, also being patient-derived CSC-driven models, can be considered as the in vitro equivalent of PDX, overcoming certain PDX limitations for dealing with large libraries of organoids or compounds. This study describes a method to create PDX-derived organoids (PDXO), thus resulting in paired models for in vitro and in vivo pharmacology research. Subcutaneously-transplanted PDX-CR2110 tumors were collected from tumor-bearing mice when the tumors reached 200-800 mm3, per an approved autopsy procedure, followed by removal of the adjacent non-tumor tissues and dissociation into small tumor fragments. The small tumor fragments were washed and passed through a 100 µm cell strainer to remove the debris. Cell clusters were collected and suspended in basement membrane extract (BME) solution and plated in a 6-well plate as a solid droplet with surrounding liquid media for growth in a CO2 incubator. Organoid growth was monitored twice weekly under light microscopy and recorded by photography, followed by liquid medium change 2 or 3 times a week. The grown organoids were further passaged (7 days later) at a 1:2 ratio by disrupting the BME embedded organoids using mechanical shearing, aided by addition of trypsin and the addition of 10 µM Y-27632. Organoids were cryopreserved in cryo-tubes for long-term storage, after release from BME by centrifugation, and also sampled (e.g., DNA, RNA and FFPE block) for further characterization.

The design, analysis and application of mouse clinical trials in oncology drug development.

BACKGROUND: Mouse clinical trials (MCTs) are becoming wildly used in pre-clinical oncology drug development, but a statistical framework is yet to be developed. In this study, we establish such as framework and provide general guidelines on the design, analysis and application of MCTs. METHODS: We systematically analyzed tumor growth data from a large collection of PDX, CDX and syngeneic mouse tumor models to evaluate multiple efficacy end points, and to introduce statistical methods for modeling MCTs. RESULTS: We established empirical quantitative relationships between mouse number and measurement accuracy for categorical and continuous efficacy endpoints, and showed that more mice are needed to achieve given accuracy for syngeneic models than for PDXs and CDXs. There is considerable disagreement between methods on calling drug responses as objective response. We then introduced linear mixed models (LMMs) to describe MCTs as clustered longitudinal studies, which explicitly model growth and drug response heterogeneities across mouse models and among mice within a mouse model. Case studies were used to demonstrate the advantages of LMMs in discovering biomarkers and exploring drug's mechanisms of action. We introduced additive frailty models to perform survival analysis on MCTs, which more accurately estimate hazard ratios by modeling the clustered mouse population. We performed computational simulations for LMMs and frailty models to generate statistical power curves, and showed that power is close for designs with similar total number of mice. Finally, we showed that MCTs can explain discrepant results in clinical trials. CONCLUSIONS: Methods proposed in this study can make the design and analysis of MCTs more rational, flexible and powerful, make MCTs a better tool in oncology research and drug development.